Citation
Dynamics of farmer adoption, adaptation, and management of soil conservation hedgerows in Haiti

Material Information

Title:
Dynamics of farmer adoption, adaptation, and management of soil conservation hedgerows in Haiti
Creator:
Bannister, Michael E. ( Dissertant )
Nair, P. K. Ramachandran ( Thesis advisor )
Place of Publication:
Gainesville, Fla.
Publisher:
University of Florida
Publication Date:
Copyright Date:
2001
Language:
English
Physical Description:
1 online resource (xiv, 235 leaves) : ill.

Subjects

Subjects / Keywords:
Agroforestry ( jstor )
Corn ( jstor )
Crops ( jstor )
Farms ( jstor )
Hedgerows ( jstor )
Soil conservation ( jstor )
Soil fertility ( jstor )
Soil water ( jstor )
Soils ( jstor )
Tillage ( jstor )
Hill farming -- Haiti ( lcsh )
Soil conservation -- Haiti ( lcsh )
Windbreaks, shelterbelts, etc -- Haiti ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
federal government publication ( marcgt )
non-fiction ( marcgt )

Notes

Abstract:
Understanding the conditions under which hillside farmers in Haiti adopt soil conservation practices helps programs to develop technologies that increase farmer revenue and stabilize or improve soil and water resources. Three studies were done in Haiti that examined biophysical and socioeconomic aspects of contour hedgerows, an agroforestry practice with the potential of stabilizing soil, conserving water for crops, increasing soil fertility, and producing wood and fodder. An on-station study of soil water competition between Leucaena leucocephala (Lan.) de Wit hedgerows and adjacent maize {Zea mays L.) over three cropping cycles found that substantial maize yield reduction was caused by the hedgerow trees because of soil water competition. Polyethylene barriers between the hedges and the first row of maize improved maize yield in two subsequent seasons by 18% and 77% over plots without barriers. The percent increase in maize yield was highest during a season of very poor rainfall; but it was lowest in a season of adequate rainfall. Barrier installation reduced Leucaena biomass production by about 1,500 kg/ha over seven months, but this effect was temporary. An examination of the distribution oi Leucaena roots after the final maize harvest showed that the hedgerow roots in the plots with barriers had grown under the hedgerows and developed more fine roots at 200 cm from the hedges than plots without barriers. An on-farm study comparing maize development in two kinds of hedgerows, rock wall terraces, and an untreated control was unable to detect differences in the rate of maize growth with respect to position within the alleys or with respect to position on the overall slope. However, the maize in the hedgerows developed more slowly than in the rock wall terraces or the control plot, indicating competitive interactions between maize and hedgerow plants. There were no differences in maize grain yield among treatments. An on-farm questionnaire-based survey of 1 ,540 Haitian farmers showed that they considered plot characteristics, including mode of tenure, soil fertility, distance from the residence, and slope in their decisions to install different agroforestry practices. Tenure security and soil fertility appeared to be the most important plot characteristics in the decision to install, although it was not possible to separate them. Hedgerows were more likely to be installed in plots having less secure tenure and less fertile soil; the opposite was true for other agroforestry practices. Farmers' qualitative assessments of soil fertility were positively correlated with management quality of hedgerows.
Thesis:
Thesis (Ph. D.)--University of Florida, 2001.
Bibliography:
Includes bibliographical references (leaves 218-234).
General Note:
Printout.
General Note:
Vita.
General Note:
Description based on print version record.
Statement of Responsibility:
by Michael E. Bannister.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
This item is a work of the U.S. federal government and not protected by copyright pursuant to 17 U.S.C. §105.
Resource Identifier:
004744249 ( AlephBibNum )
465403427 ( OCLC )
22109162 ( ALEPH )

Downloads

This item has the following downloads:


Full Text










DYNAMICS OF FARMER ADOPTION, ADAPTATION. AND MANAGEMENT
OF SOIL CONSERVATION HEDGEROWS IN HAITI















By

MICHAEL E. BANNISTER


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2001


















UNIVERSITY OF FLORIDA
3 1II262 I086 32
3 1262 08666 372 0
i "06.1 g


160













ACKNOWLEDGMENTS

I wish to thank my graduate advisor, Dr. P.K.R. Nair, for his unflagging

encouragement and guidance throughout a long period of research and writing and Dr.

Peter Hildebrand for providing my Farming Systems assistantship from 1987 through

1989. My other committee members, Dr. Mary Duryea, Dr. Jerry Murray, and Dr. Ken

Buhr also provided valuable insights and assistance. The untiring work of the Pan

American Development Foundation (PADF) Haitian technical staff during data collection

and the encouragement and financial support of the PADF home office during my

sabbatical were essential to this dissertation, as was the cooperation of a multitude of

Haitian farmers. Deserving of special mention are PADF Haiti Country Representative

Lee Nelson for his key support of my sabbatical and Agronomist Jn-Charlot Bredy for

data collection on farms near Lascahobas. Data collection for the Leucaena/maize study

could not have been completed without the assistance of technicians Duverger Veris and

Rodini St-Juste. Finally, I thank my beloved wife Mojdeh for her essential care of my

sometimes-fragile psyche during this trying time.














TABLE OF CONTENTS

pages

LIST OF TABLES ...................................................... v

LIST OF FIGURES .................................................... x

ABSTRACT .............. ................... .................... xiii

CHAPTERS

1 INTRODUCTION .................................................. 1

2 HILLSIDE FARMING SYSTEM IN HAITI .......... ..............4

Land .............................................................5
Crops ............................................................ 7
Labor .............................................. 9
Farmer Development Groups and Local NGOs ........................... 10

3 THE HEDGEROW INTERCROPPING SYSTEM ........................ 13

Developm ents in the Tropics ......................................... 13
Developments of Hedgerow Intercropping in Haiti ........................ 18
Hedgerow Intercropping and Soil Conservation .......................... 22
Conclusions ...................................................... 26

4 ORGANIZATION OF THE STUDY ................................. 27

Justification .......................................................27
Hypotheses ...................................................... 33
General Description of Hypothesis-Testing Procedures ..................... 34

5 HEDGEROW/CROP COMPETITION ................................ 36

Introduction .................................................... 36
On-Station Studies: Effect of Root Barriers on the Growth of Maize and
Leucaena ....................................................39
On-Farm Study: Maize Growth at Varying Distances from Hedgerows ....... 100








6 LAND AND HOUSEHOLD CHARACTERISTICS IN RELATION TO
ADOPTION AND MANAGEMENT OF HEDGEROWS ................ 108

Introduction and Objectives ......................................... 108
Materials and Methods ................ ........................... 119
Results and Discussion ............... .......................... 127
Summary and General Discussion .................................... 166
Conclusions ..................................................... 175

7 SYNTHESIS AND CONCLUSIONS ................................ 179

APPENDICES

A POT STUDY OF SOIL WATER DEPLETION VS. MAIZE LEAF WATER
POTENTIAL .................................................. 186

B ROOT DISTRIBUTION OF A LEUCAENA LEUCOCEPHALA
HEDGEROW .................................................. 188

C P-VALUES OF ANOVAS PERFORMED ON STATION ................. 191

D HOUSEHOLD QUESTIONNAIRE ................................. 198

E GARDEN PLOT QUESTIONNAIRE ............................... 201

F SOIL NUTRIENTS, ORGANIC CARBON STATUS, AND pH
OF 175 HILLSIDE GARDENS .................................... 211

REFERENCES ................. ..................................... 218

BIOGRAPHICAL SKETCH ................ ........................... 235













LIST OF TABLES


Table page

5-1: Soil texture and organic carbon at the trial site .......................... 43

5-2: Effect of trenching on the production of maize biomass ................... 60

5-3: Effect of fertilizer on the production of maize biomass .................... 61

5-4: Soil water at three depths below the surface ......................... .. 62

5-5: Maize grain weight, number of plants, and number of ears ................. 65

5-6: Stem and leaf biomass harvests and daily growth increments of Leucaena
hedgerows ..................................................... 66

5-7: Height above a 50 cm stump and daily growth increments of Leucaena
hedgerows prior to the second root trenching, 33 and 65 DAS; Spring
1994 ............................ .................... ..........67

5-8: Maize biomass in trenched and nontrenched plots ....................... 69

5-9: Maize leaf water potential in trenched and nontrenched plots over four
distances from the hedgerows, 42 DAS; fall 1994 ........................ 71

5-10: Soil water in trenched and nontrenched plots by depth and distance ......... 73

5-11: Maize grain weight, number of plants, and number of ears at four
distances from the hedgerows and as a total per hectare, with p-values for
the differences between distances, Fall 1994 ............................ 75

5-12: Stem and leaf biomass harvests and daily growth increments of Leucaena
hedgerows after the second root trenching, Fall 1994 ..................... 76

5-13: Height above a 50 cm stump and daily growth increments of Leucaena
hedgerows after the second root trenching. Fall 1994 ..................... 77

5-14: Maize height in trenched and nontrenched plots at four distances from
hedgerows. 40 DAS; Spring 1995..................................... 79








5-15: Maize biomass in trenched and nontrenched plots at four distances from
the hedgerow, 47 and 54 DAS; Spring 1995. ........................... 82

5-16: Soil water in trenched and nontrenched plots by depth and distance from
the hedgerow, 26 and 61 DAS; Spring 1995 ............................ 86

5-17: Maize grain weight, number of plants, and number of ears ................. 89

5-18: Stem and leaf biomass harvests and daily growth increments of Leucaena
hedgerows .......................................... ............ 90

5-19: Height above a 50 cm stump and daily growth increments of Leucaena
hedgerows ................................................... 90

5-20: Maize yield in three soil conservation practices and control at 116 DAS on
four farm s, spring 1997 ........................................... 105

6-1: Age distribution by gender of members of 1,540 farm households
interviewed in the four PADF/PLUS project areas in Haiti, 1996. .......... 128

6-2: Number of years of school by gender in seven age classes of 1,540
households participating in the PADF/PLUS project, 1996 ............... 129

6-3: Percent of female and male members of 1,540 project households who
participated in various economic activities during 1995-1996 ............. 130

6-4: Percent of female and male members of 1,540 project households who
participated in various economic activities during 1995-1996, by age
class......................................................... .131

6-5: Percent of female and male members of 1.540 households who
participated in PLUS project activities during 1995-1996 ................. 132

6-6: Percent of female and male members of 1,540 project households who
participated in PLUS project activities during 1995-1996, by age class ...... 133

6-7: Average total area of three categories of plots held by 1,540 project
households based on the relative distance from the plots to the residence .... 134

6-8: Mean plot area of three categories of plots held by 1,540 project
households during 1995, based on the relative distance from the plots to
the residence ................................................... 134

6-9: Mean plot area of three categories of plots held by 1,540 project
households during 1995 in four PLUS project regions, based on the
relative distance from the plots to the residence ........................ 135








6-10: Percent of three categories of plots held by 1,540 project households
during 1995 under various modes of access ........................... 136

6-11: Area of 5,660 plots held by 1,540 households during 1995, by mode of
access ........................................................ 137

6-12: Percent of plots held by 1,540 households during 1995 in eight mode of
access categories by slope position .................................. 137

6-13: Percent of all plots held by 1,540 households during 1995 in eight modes
of access categories by slope steepness ............................... 138

6-14: Distance from residence of plots having project agroforestry practices held
by 1,540 households during 1995 in the five most common mode of
access categories ................................. ........... 139

6-15: Area of plots having project agroforestry practices held by 1,540
households during 1995 in the five most common mode of access
categories ................................................... 140

6-16: Elevation of plots having project agroforestry practices held by 1,540
households during 1995 in seven mode of access categories .............. 140

6-17: Slope of plots having project agroforestry practices held by 1,540
households during 1995 in seven mode of access categories .............. 141

6-18: Percent of 2,295 surveyed plots with and without three kinds of
agroforestry practice held under secure (purchased or inherited/divided) or
not secure (all other) tenure ..................................... 147

6-19: Percent of 2,295 surveyed plots with and without three kinds of
agroforestry practice held under secure (purchased or inherited/divided) or
not secure (all other) tenure ..................................... . 147

6-20: Percent of 2,295 surveyed plots with and without three kinds of
agroforestry practice having fertile or not fertile soil .................... 147

6-21: Percent of 2,295 surveyed plots with and without three kinds of
agroforestry practice having fertile or not fertile soil .................... 148

6-22: Percent of 2,295 surveyed plots with and without three kinds of
hedgerows held under secure (purchased or inherited/divided) or not
secure (all other) tenure ............... ......................... .149

6-23: Percent of 2,295 surveyed plots with and without three kinds of
hedgerows having fertile or not fertile soil ............................ 149








6-24: Number of all trees larger than 10 cm diameter on PLUS plots, by plot
tenure ..................................................... 150

6-26: Soil fertility and hedgerow management quality ......................... 154

6-27: PADF/PLUS field region and hedgerow management quality .............. 156

6-28: Soil fertility and crop band management quality ........... ... ......... 157

6-29: PADF/PLUS field region and crop band management quality .............. 158

6-30: Soil fertility and rock wall management quality ......................... 159

6-31: PADF/PLUS field region and rock wall management quality .............. 160

6-32: Percent of 1,362 plots where potential benefits of hedgerows (fuel,
charcoal, construction wood, fodder, soil fertility, soil conservation, and
crop production) were classified according to importance by farmers on a
five-point scale, 1=not important, 5=very important. .................... 161

6-33: Percent of hedgerows classified as well-managed by technicians in 1,362
plots where potential benefits of hedgerows (fuel, charcoal, construction
wood, fodder, soil fertility, soil conservation, and crop production) were
classified according to importance by farmers on a five-point scale, l=not
important, 5=very important. ...................................... 162

6-34: Percent of hedgerows classified as poorly-managed by technicians in
1,362 plots where potential benefits of hedgerows (fuel, charcoal,
construction wood, fodder, soil fertility, soil conservation, and crop
production) were classified according to importance by farmers on a five-
point scale, l=not important, 5=very important. ....................... 163

6-35: Number of breaches per 100 m of hedgerow counted by technicians in
1,362 plots where potential benefits of hedgerows (fuel, charcoal,
construction wood, fodder, soil fertility, soil conservation, and crop
production) were classified according to importance by farmers on a five-
point scale, l=not important, 5=very important. ........................ 164

6-36: Percent of gardens where hedgerow breaches were repaired or not repaired
for three potential benefits of hedgerows. .......................... 165

6-37: Percent of 1,362 plots where potential problems of hedgerows (shade,
reduced space, water competition, reduced ability to picket animals,
weediness, and labor cost) were classified according to importance by
farmers on a five-point scale, l=not important, 5=very important .......... 165








6-38: Percent of gardens where hedgerow breaches were repaired or not repaired
for three potential problems of hedgerows. ............................ 166

6-39: Estimated relative costs, benefits, and risk of agroforestry practices ........ 171














LIST OF FIGURES

Figure page

5-1: Layout of the Leucaena/maize hedgerow trial on Operation Double
Harvest property near Croix des Bouquets, Haiti ........................ 45

5-2: Diagram of one plot showing relative positions of a Leucaena hedgerow
and the adjacent rows of maize at four distances. ........................ 46

5-3: Rainfall, Leucaena and maize growth, and agronomic inputs; spring 1994
maize cropping season ................ ........................... 51

5-4: Rainfall, Leucaena and maize growth, and agronomic inputs; fall 1994
maize cropping season ............................................ 51

5-5: Rainfall. Leucaena and maize growth, and agronomic inputs; spring 1995
maize cropping season ................ ........................... 52

5-6: Maize biomass yields at 50, 100, 150, and 200 cm from the hedgerows at
four times after sowing during the spring 1994 season .................... 59

5-7: Growth stage of maize at four distances from hedgerows .................. 61

5-8: Yield of maize grain per six meters at four distances from the hedgerows ..... 64

5-9: Number of maize plants per eight hills at four distances from the
hedgerows ..................................................... 64

5-10: Number of maize ears per eight hills at four distances from the hedgerows .... 65

5-11: Trenched and nontrenched plots at 50, 100, 150, and 200 cm from the
hedgerows. 78 DAS; fall 1994 ............... ...................... 69

5-12: Maize leaf water potential in trenched and nontrenched plots ............... 70

5-13: Soil water in trenched plots (and difference between T+ and T- plots) at
42 DAS, fall 1994 .............................................. 72

5-14: Maize grain yield in trenched (T+) and nontrenched (T-) plots .............. 74








5-15: Maize height trenched and nontrenched at four distances from the
hedgerow, 40 DAS; Spring 1995 .................................... 79

5-16: Maize biomass in trenched and nontrenched plots at four distances from
the hedgerow, 47 DAS, Spring 1995 .................................. 81

5-17: Maize biomass in trenched and nontrenched maize weight at four
distances from the hedgerow, 54 DAS, Spring .......................... 81

5-18: Trenched and nontrenched maize growth stage at four distances from the
hedgerow, 68 DAS; Spring 1995. .................................... 83

5-19: Trenched and nontrenched maize growth stage at four distances from the
hedgerow, 75 DAS; Spring 1995. ....................................83

5-20: Soil water percent in trenched plots (and difference between T+ and T-
plots) at 26 DAS, Spring 1995 ....................................... 85

5-21: Soil water percent in trenched plots (and difference between T+ and T-
plots) at 61 DA S, Spring 1995 ....................................... 85

5-22: Maize grain weight from trenched and nontrenched plots at four distances
from the hedgerows, 131 DAS; Spring 1995. ........................... 87

5-23: Number of maize plants from trenched and nontrenched plots at four
distances from the hedgerows, 131 DAS; Spring 1995. ................... 88

5-24: Number of ears from trenched and nontrenched plots at four distances
from the hedgerows, 131 DAS; Spring 1995 ........................ 88

5-25: Number of very fine root (<2mm) intersections with a 10 by 100 cm plane
parallel to the trenched hedgerows at five distances and 10 depths, with
the difference between the trenched and nontrenched hedgerows shown in
parentheses ................ ................................... 92

5-26: Number of fine root (2-5 mm) intersections with a 10 by 100 cm plane
parallel to the trenched hedgerows at five distances and 10 depths, with
the difference between the trenched and nontrenched hedgerows shown in
parentheses. ..................................................... 92

5-27: Number of medium root (5-10 mm) intersections with a 10 by 100 cm
plane parallel to the trenched hedgerows at five distances and 10 depths,
with the difference between the trenched and nontrenched hedgerows
shown in parentheses ...................................... ...... 93








5-28: Number of large root (>10 mm) intersections with a 10 by 100 cm plane
parallel to the trenched hedgerows at five distances and 10 depths, with
the difference between the trenched and nontrenched hedgerows shown in
parentheses .................................................... 93

5-29: Leucaena small stem and leaf biomass from trenched and nontrenched
plots, May 1991 through September 1995 .............................. 98

5-30: Maize development in three soil conservation practices and control at four
times during the growing season on four farms during Spring 1997 ......... 104

6-1: Relative importance of four plot characteristics to the decision to install
six agroforestry practices ............... ......................... 168













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

DYNAMICS OF FARMER ADOPTION, ADAPTATION, AND MANAGEMENT
OF SOIL CONSERVATION HEDGEROWS IN HAITI

By

Michael Bannister

May 2001

Chair: P. K. Ramachandran Nair
Major Department: School of Forest Resources and Conservation

Understanding the conditions under which hillside farmers in Haiti adopt soil

conservation practices helps programs to develop technologies that increase farmer

revenue and stabilize or improve soil and water resources. Three studies were done in

Haiti that examined biophysical and socioeconomic aspects of contour hedgerows, an

agroforestry practice with the potential of stabilizing soil, conserving water for crops,

increasing soil fertility, and producing wood and fodder.

An on-station study of soil water competition between Leucaena leucocephala

(Lan.) de Wit hedgerows and adjacent maize (Zea mays L.) over three cropping cycles

found that substantial maize yield reduction was caused by the hedgerow trees because of

soil water competition. Polyethylene barriers between the hedges and the first row of

maize improved maize yield in two subsequent seasons by 18% and 77% over plots

without barriers. The percent increase in maize yield was highest during a season of very

poor rainfall; but it was lowest in a season of adequate rainfall. Barrier installation








reduced Leucaena biomass production by about 1,500 kg/ha over seven months, but this

effect was temporary. An examination of the distribution of Leucaena roots after the

final maize harvest showed that the hedgerow roots in the plots with barriers had grown

under the hedgerows and developed more fine roots at 200 cm from the hedges than plots

without barriers.

An on-farm study comparing maize development in two kinds of hedgerows, rock

wall terraces, and an untreated control was unable to detect differences in the rate of

maize growth with respect to position within the alleys or with respect to position on the

overall slope. However, the maize in the hedgerows developed more slowly than in the

rock wall terraces or the control plot, indicating competitive interactions between maize

and hedgerow plants. There were no differences in maize grain yield among treatments.

An on-farm questionnaire-based survey of 1,540 Haitian farmers showed that

they considered plot characteristics, including mode of tenure, soil fertility, distance from

the residence, and slope in their decisions to install different agroforestry practices.

Tenure security and soil fertility appeared to be the most important plot characteristics in

the decision to install, although it was not possible to separate them. Hedgerows were

more likely to be installed in plots having less secure tenure and less fertile soil; the

opposite was true for other agroforestry practices. Farmers' qualitative assessments of

soil fertility were positively correlated with management quality of hedgerows.













CHAPTER 1
INTRODUCTION

The research presented here focuses on agroforestry technologies, which could

contribute to increasing the revenue of small farms on the hillsides of Haiti in a

sustainable, environmentally friendly way. This is a formidable goal, because Haiti is

primarily a mountainous country long since shorn of its natural vegetative cover. Its

once-rich soils are badly eroded, and their productive capacity severely diminished.

Haiti is densely populated by very poor people who generally lack formal education or

access to a functioning legal system. It has a long history of political instability that

continues to this day. At best, agriculture is difficult in such a setting. But, Haitian

farmers regularly experiment with new techniques.

This dissertation investigates the biological and adoption aspects of hedgerow

intercropping, an agroforestry practice that involves growing food crops in the

interspaces between rows of densely planted trees or shrubs. Two types of studies are

included. One focuses on competition for growth resources between trees and crops

while the other focuses on the factors that influence farmers' adoption of the technology.

However, these different studies are linked together, because both social and biological

aspects of agroforestry practices must be understood for providing efficient extension

services to farmers.

The dissertation is presented in seven chapters. Chapter 1 briefly explains the

background of the research and states its objectives. Chapter 2 describes Haitian








smallholder farms, the population of interest to the research. Following a definition of

hedgerow intercropping, the agroforestry practice that is the subject of the research.

Chapter 3 reviews its development in the tropics and in Haiti and examines how it is used

as a soil conservation technique. Chapter 4 justifies the research, presents the hypotheses

to be tested, and describes the studies used to test the hypotheses. The next chapter

(Chapter 5) presents the two studies examining competition between hedgerows and

adjacent crops. The first is done on station, the second on farm. The questionnaire-based

survey of farmers who adopted soil conservation practices promoted by the PLUS project

is the subject of Chapter 6. Finally, Chapter 7 synthesizes the previous chapters.

and presents the overall conclusions of the dissertation.

The research was done while the author was employed by the Pan American

Development Foundation (PADF) in Haiti. The PADF implements a U.S. Agency for

International Development (USAID) project called Productive Land Use Systems

(PLUS). The PADF/PLUS has worked with about 170,000 hillside farmers in soil

conservation, tree planting and grafting, has improved food crop distribution and

marketing, and has been strengthening local farmer organizations between 1992 and

1999. The on-farm study of maize (Zea mays L.) production between soil conservation

structures discussed in Chapter 5 and the questionnaire-based study discussed in Chapter

6 were done as PLUS monitoring and evaluation exercises by PADF staff. The author

designed the questionnaires and trained staff in their use.

The overall objective of this research is to understand how characteristics of

Haitian farm families, their plots, and agroforestry practices affect the adoption and

management of those practices. The purpose is not to use the data gathered in the course

of this research to define and predict adopters so that nonadopters can be excluded from








extension efforts. It is rather an effort to understand how appropriate practices can

evolve and be extended more rapidly.

In particular, the objectives of the three studies that comprise this dissertation are

to determine, in the relatively controlled environment (for a research station),
whether the competition for soil water between trees in hedgerows and
adjacent maize is significant and can be reduced by the use of plastic root
barriers,

to understand whether tree-crop competition in farmers' steeply sloping fields
is similar to that observed in the flat controlled environment of the research
station, and

to examine the factors (e.g., land tenure, perceived tree/crop competition, plot
slope, soil fertility of the plot, distance from the residence to the plot, previous
experience of the farmer, and the availability of other farm resources) that
influence farmer adoption, adaptation, and management of hedgerow
intercropping.













CHAPTER 2
HILLSIDE FARMING SYSTEM IN HAITI

Haiti is mainly hilly, with 20% of the land area below 185 m elevation and 40%

above 450 m-about 75% highlands, 22% lowlands (Weil et al. 1973). The most

common "life zones," according to Holdridge's (1945) classification, are subtropical

moist forest and subtropical dry forest, but subtropical wet and rain forest zones are

common in the middle and upper elevations. Limestone substrate underlies 80% of the

land area; the rest is basaltic or alluvial (Ehrlich et al. 1985). According to Ehrlich et al.

(1985) estimates, 1.3 million hectares (of 2.7 million total for the country) were under

agricultural production; six times the area classified as "good agricultural land." The

practice of annual agriculture on steeply sloping plots causes severe losses of arable soil

due to erosion, estimated to be about 37 million tons annually in 1994 (UNDP, 1996).

Haitians have been practicing agriculture on small, privately held plots in the hills

since the 1840s. as the large plantations and public domain lands were divided (de Young

1958. Leyburn 1941). Haitian peasants are not subsistence farmers in the strict sense that

they do not live off the land self-sufficiently and independently, but rather produce for,

and are linked closely to, the market (Murray 1981). Most farm families have some

nonagricultural income and are net food buyers (Locher 1996). De Young (1958)

describes the economic organization of Haitian farms as "horticultural" rather than

agricultural, and gives the following characteristics that still hold true today:

They are market directed, not subsistence directed.








They give more importance to perennial rather than annual crops.

The farm units are small parcels, and the land is used intensively.

They are capital saving rather than labor saving. Farmers own only a few
hand tools, have little invested in storage or farm buildings, and rarely use
plows and animal traction.

The farm animals are used mainly for transport to market.

Reducing risk is as important to most farmers as increasing production, and to

this end they manage a network of ties to relatives and other farmers as social capital that

can be leveraged in order to gain access to scarce land, labor, and capital (Smucker et al.

2000). Haitian families also spread out risk by diversifying human and economic

resources via selective migration, emigration, and off-farm economic enterprises (Locher

1996). Therefore, not only is any one particular crop not the sole interest of a farm

family, but agriculture is not the only domain of economic activity. It is quite likely that

other activities, such as nonagricultural marketing or remittances from relatives overseas,

bring in more cash.

Land

Haitian hillside farms are small and complex. Farm families usually manage or

cultivate several parcels of land at any given time, held under a mix of land tenure

arrangements (e.g., owned, tenancy, usufruct) (Bloch et al. 1987, Locher 1996) and often

on sites differing in their physical characteristics (hillside, plain, irrigated, rainfed) and in

distance from the residence. This is done as a purposeful strategy to reduce risk of

harvest loss and to spread out crop harvest over the year (McLain and Stienbarger 1988,

Murray 1981, Zuvekas 1979). Plot sizes are small and fallow periods are short. Plots are

worked and then abandoned when yield decreases below the farmer's economical limits.

Declining soil fertility, soil loss due to erosion, large variation in the timing of rainfall,








and the inability of traditional agricultural practices to harvest and store water in the soil

limit agricultural production. Residences tend to be physically disbursed in the mountain

regions instead of concentrated (Friedman 1955).

Two intertwining land tenure systems operate in Haiti, one official and the other

customary (Smucker et al. 2000). created by disenfranchised peasants to protect

themselves against loss of land via a corrupt justice system and government abuse. The

land market is active. Most farmers own most of their land, but the majority of the

transactions are done without recourse to expensive official processes (Smucker et al.

2000, Zuvekas 1979). The customary system provides farmers relatively secure access to

land, but some tenure classes are more stable over time than others so mixed tenure

configurations that change over time are the norm (Locher 1996). Zuvekas (1979)

reviewed data from 12 studies done in Haiti and found that the percentage of farmers

owning (by purchase or inheritance) at least part of their land ranged from 58 to 100%.

Secondary modes of access to land include rental from private owners, lease, and

sharecropping. Usufruct rights to land are also obtained from relatives or by acting as

caretaker for an absentee landowner

Information about farm size is available from several studies related to

agroforestry projects from various parts of the country (Lea 1994a, Lea 1996, Swanson

et al. 1993a, 1993c, White 1992a). The total size of all farm plots ranges from less than

one to about three hectares. The average farm family works two or three plots (Zuvekas

1979) that vary in size from about one half to just over one hectare each.

Haitian laws of inheritance have French antecedents and require that all siblings

inherit; therefore, land holdings are, in general, getting smaller. In 1950, 39% of

households held less than 1.3 hectares; by 1971 this proportion had increased to 71%








(Murray 1981). Brochet (1993) compared the average size of home gardens in one area

of Haiti over fifty years, and found that the average holding that was one hectare per

family in 1910 had decreased to 500 m2 (0.05 ha) by 1970. This study was done in an

area mined for bauxite by Reynolds Aluminum, so sale of land to the mining company

may have accelerated the rate of decrease, but in general average sizes of holdings grow

smaller as the population increases. One strategy adopted by Haitian families to ensure

that a plot is large enough to harvest a crop worth planting is to bequeath land to children

undivided, but this may discourage investing in improvements to the plot if the siblings

are contentious. Older farmers are likely to own a larger proportion of their holdings

(Lundahl 1997, McLain and Stienbarger 1988).

Other types of land tenure found in Haiti include purchased, inherited but

separated, rented from private owners, and rented from the state. Usufruct rights to land

also are obtained by sharecropping (on land owned by the sharecropper as well as on land

owned by another) and by acting as caretaker for an absentee landowner. A recent

survey of 5000 fields across the country (USAID. 1996) showed the following pattern of

land tenure: purchased 39%, inherited/divided 20%, inherited/undivided 14%, rented

12%, share cropped 5%. gift 1%. and other 9%.

Crops

Since small and scattered plots must provide food, fuel, building material, animal

fodder, and other household necessities, agroforestry has traditionally been practiced.

The farm plots, except the plot containing the residence and surrounding tree and

medicinal crops (sometimes called the home garden), are spatially dominated by cereal

grains and grain legumes (pulses), but income from roots, tubers, tree crops, and animals

is probably more important economically, and is able to shield farmers against variations








in rainfall better than grains and pulses. Home gardens include many tree crops such as

mango (mangifera indica L.), breadfruit (Artocarpus altilis [Parkinson] Fosberg), citrus

(Citrus spp.), royal palm (Roystonia borinquena), and coconut (Cocos nucifera L.).

Timber and fuelwood trees are planted or tended where they sprout around property

perimeters, scattered throughout plots, or as small woodlots. Trees are considered to be a

crop. and are managed as a savings account, to be harvested when cash or building

materials are needed (Murray 1981, Smucker and Timyan 1995). Fuelwood and animal

fodder are in short supply in many parts of the country.

Many different combinations of crops are grown. These vary with the season.

land tenure, distance from the residence, market demand, availability of seed, and the

rainfall and soil characteristics. The most common grain crops are maize (Zea mays L.)

and sorghum (Sorghum bicolor (L.) Moench) on the hillsides, and rice (Oryza sativa L.)

in the few irrigated plains. Pulses include beans (Phaseolus vulgaris L.), pigeon pea

(Cajanus cajan (L.) Millsp.), and peanuts (Arachis hypogaea L.). It is common for the

spring maize crop to fail, because of poor rainfall. Beans are able to generate important

income for farmers, but also fail often--rain is insufficient, or rain falls during flowering

(eliminating pod formation), or rain falls before seeds are dry causing them to germinate

in the pods. Root and tuber crops include yams (Diascora spp.), cassava (Manihot

esculenta Crantz), and cocoyam (Xanthosoma sagittifolium (L.) Schott). Plantains (Musa

spp.) are commonly grown in ravines and near the home. Living fences surround the

property where the house is located, and are commonly built of plants serving protective

functions due to the presence of spines or poisonous sap, such as bayonette (Yucca

aloifolia L.). pencilbrush (Euphorbia tirucalli L.), raquette (Euphorbia lactea Haw.), and

penguin (Bromelia pinguin) (Ashley 1986).








A 1996 USAID survey of 6,900 fields during a single spring planting season

(USAID, 1996) found that 27% of the plots had crops growing in association (two or

more crops present simultaneously in the same field for part or all of the cropping cycle),

and 21% in monoculture. There were 16 main maize associations and 98 minor (<5000

ha. each) ones. In addition, there were 12 important associations of other crops, 134

minor ones, and 14 main kinds of monocrops. This does not include the many different

kinds of tree and medicinal crops, crops found during the fall growing season, or animals.

The most commonly seen grain crop in the fall season is sorghum, usually intercropped

with pigeon pea. The USAID study notes only the physical presence of the crops, not

their relative economic importance.

Labor

Households function as separate economic units, but the members participate in

labor and food exchanges with other farmers (Murray 1981, Smucker and Dathis 1998).

The family itself provides the labor for most agricultural activities. This labor pool is

expanded periodically according to seasonal demands for land clearing, working the soil,

planting, weeding, and harvesting crops. Labor is a scarce resource, and has been more

difficult to supply and afford as the deteriorating social and economic situation in Haiti

has forced many rural inhabitants into political refuge or into taking jobs in the

neighboring Dominican Republic (Balzano 1997, Smucker and Dathis 1998). Farmers

form groups to exchange farm labor, earn cash, or to save money to attain a future goal.

However, this communal pooling is restricted to labor, not to the sharing of land or

produce (Murray 1991). Access to labor is crucial because working steep slopes is labor

intensive, and frequently must be done on short notice because rainfall varies so much

over short distances in the mountains. Crops must be planted as soon as the first rains








arrive. Farmers without access to labor at the right time will suffer. Work groups

generally consist of 3 to 15 members from the same socioeconomic level, and can be

permanent or formed for a single task (Murray 1991, Smucker and Dathis 1998, Swanson

et al. 1993a, 1993b 1993d, 1993e). The names given to labor groups differ from place to

place within Haiti, and even from farmer to farmer (Smucker and Thomson 1999). The

two main kinds of groups are those that are for labor exchange only and those that are

formed to work for cash. Client farmers paying cash for labor are generally not group

members, and are better off economically. A wealthy farmer may join a labor exchange

group just for the purpose of gaining access to the labor. Another kind of work party is

called the koumbit, where the host farmer provides one or two meals and sometimes

alcohol to other farmers invited to clear or prepare a field (Swanson et al. 1993a). It is

not possible to control the number of participants, the length of time they stay, or the

quality of the work done, and for this reason the koumbit is becoming rare (Swanson et

al. 1993e). Finally, individual farmers also are engaged by others to do specific tasks.

This is referred to as a djob (job).

Typically, men form labor exchange groups for agriculture while women form

mutual credit supply groups that permit them to engage in commerce. Women have

always participated in some aspects of agricultural labor, and this appears to be on the

rise because of the out-migration of men during times of political and economic troubles

mentioned above (Smucker and Dathis 1998).

Farmer Development Groups and Local NGOs

Indigenous development groups are of interest because, since the early 1980s,

most donor-financed development work, including agroforestry extension, has been done

through them. This happened because of the difficulty in controlling the use of funds








disbursed through the Haitian government, which has always been a predatory structure

serving the interests of the urban elite to the exclusion of the rural majority (Smucker and

Thomson 1999).

Smucker and Dathis (1998) give a comprehensive description of these groups, on

which the rest of this section is based. These groups vary widely in origin and

composition, ranging from community organizations, cooperatives, youth groups,

women's groups, work exchange groups, traditional groups, and religious groups.

Agricultural cooperatives began in the 1930s and expanded in number in the 1950s. The

community council movement began in the 1950s and was politicized by the Duvalier

regime in the 1970s. The Catholic Church created nonhierarchical farmer development

groups beginning in the 1960s that became known as groupement and eventually stood in

opposition to the community councils. After the fall of the Duvaliers in 1986 the

numbers of groups based on the groupement model increased dramatically; but they were

persecuted, and they went underground after the 1991 coup d'etat. All of the groups

have economic survival objectives, but are not limited to that. They look for more

control over the factors of production, including agricultural inputs, credit, grain storage,

more control over export crops, and investment in soil conservation and trees. They are

now experiencing growing pains as their membership increases and they take on more

objectives, without the legal recognition that would permit them to defend their interests

and open bank accounts. These difficulties increase as their growth pushes them to

depart from the basic characteristics of the traditional local groups, which are direct

participation, small size with common interests, hope of tangible profits, direct

investment of members' resources, and links of reciprocity and rotation. Farmers would

be better able to invest in necessary agroforestry practices if these groups were stronger.






12

For example, a group with juridical personality, basic accounting skills, and a bank

account would be more likely to succeed in negotiating crop sales to large buyers.

Profits from sales would allow them to buy labor to protect the plots that produce the

crops being sold.













CHAPTER 3
THE HEDGEROW INTERCROPPING SYSTEM

Developments in the Tropics

Definition

Hedgerow intercropping, also known as alley cropping and alley farming (Rao

et al. 1998), has been suggested as an approach to address decreasing soil fertility and for

controlling erosion (Kang et al. 1990). It is an agroforestry practice consisting of closely

planted and regularly spaced rows of fast-growing. repeatedly pruned trees, usually N2-

fixing legumes, with short cycle agricultural crops cultivated in the alleys or spaces

between the hedges (Kang et al. 1990). It is classified as a simultaneous (trees and crops

present in the field at the same time) and zonal (trees are concentrated in the field rather

than dispersed) agroforestry practice (Nair 1993). The hedges and crops are managed

together for a common purpose, and there are significant above- and below-ground

ecological interactions between them (Sinclair 1999). The most common objective of

hedgerow intercropping is to increase crop production in the alleys by applying the

pruned tops of the hedgerow trees to the soil as a green manure to improve soil fertility.

This objective is pursued by attempting to balance the facilitative and competitive effects

inherent in the practice (Vandermeer 1998). The more leaf biomass harvested from the

trees, the more green manure available to improve soil conditions and thereby increase

crop yield. However, more leaf biomass is obtained by increasing the number of

hedgerows and this reduces the space available to crops. It also increases the area of








tree-crop interface (Ong and Leakey 1999) and therefore creates more possible

competitive interactions (shading, soil water competition, nutrient competition) with the

crops, potentially depressing yield (Vandermeer 1998). Hedgerows fulfill other purposes

as well, including soil erosion control and production of fodder and fuel.

Origins

Hedgerow intercropping developed independently both in the western humid

lowlands of Africa and in southeast Asia. In Nigeria, traditional farmers have for several

generations planted and managed hedges of a shrub, Dactyladenia barteri (Hook f. ex

Oliver), for the purposes of contributing nutrients to crops. controlling weeds, and

production of fodder and stakes (Kang et al. 1984). Looking for a way to respond to soil

impoverishment due to the shortening cycle of traditional slash and burn agriculture,

researchers began investigating similar ways to use trees. The principles suggesting

intercropping plants with complementary root systems occupying different soil regions

(Weaver 1920) and the possibility of tree roots capturing nutrient resources out of the

reach of crops (Nye and Greenland 1960) were familiar, but had not yet been applied

systematically to agroforestry associations by nonfarmer researchers. The International

Institute for Tropical Agriculture (IITA). based in Ibadan, Nigeria. began trying woody

species to improve the fertility of low-activity clay soils. This led to the development of

alley cropping (also referred to as hedgerow intercropping), avenue cropping, and alley

farming if fodder production was a main objective (Kang and Wilson 1987). The first

alley cropping trial at IITA began in 1976; on-farm trials were established in the 1980s

(Tripathi and Psychas 1992).

An earlier start was made in Indonesia. Development of hedgerow intercropping

in this region appears to be related to the introduction of Leucaena spp. to the Philippines








and Indonesia by the Spanish from Mexico (Dijkman 1950) and its promotion by the

Dutch as a shade tree for coffee (Coffea arabica L.), cacao (Theobroma cacao L.), tea

(Camelia sinensis (L.) Kuntze), and tobacco (Nicotiana tabacum L.) in the 18th century

(Wirjodarmodjo and Wiroatodjo 1983). Dutch journal articles from the 1930s report the

use of Leucaena contour hedges to support bench terraces on slopes (Dijkman 1950).

Tacio (1993) cites a traditional ruling in the 1930s obliging all farmers in a region of

Timor, Indonesia to plant contour hedges of Leucaena 3 m apart in cropped areas.

Metzner (1976) notes Leucaena thickets were promoted by the Indonesian government in

the 1930s to recuperate nonarable lands, but farmers did not do it because they feared the

weedy qualities of Leucaena. By the 1950s, Leucaena was commonly used in the Pacific

as contour hedges for erosion control (Dijkman 1950). Hernandez (1961), cited in Benge

(1980), discusses results of Indonesian hedgerow intercropping research done in 1953.

The hedgerows decreased erosion and improved maize yield by 380% on the treated

plots, compared to traditional cultivation. Extension of hedgerows to farmers, supported

by the Indonesian government and religious groups, expanded greatly during the 1970s

(Metzner 1976, Tacio 1993, Wirjodarmodjo and Wiroatodjo 1983).

The Rise and Fall of Alley Cropping

Research on alley cropping in West Africa in the 1980s generated widespread

interest (Kang and Wilson 1987) and led to trials and extension in many parts of the

world. In trials examining their technical performance, hedgerows had positive effects

on the yield of adjacent food crops through the application of tree prunings as green

manure (Tonye and Titi-Nwel 1995), increased rainwater infiltration into the soil (Kiepe

1995), and reduced soil erosion when installed on sloping land (Paningbatan et al. 1995).

Some economic analyses indicated that alley cropping, in spite of its high labor








requirement, can be profitable under certain environmental conditions (Ehui 1988).

Based on these initial results, project based extension of hedgerow intercropping to

farmers in developing countries grew quickly (Ongprasert and Turkelboom 1995,

Sanchez 1995, Scherr 1994).

However, after a few years many came to believe that alley cropping had been

oversold and had much less applicability than previously thought, and that its potential

for adoption was limited to certain niche areas (Carter 1995, Sanchez 1995). Based on a

review of hedgerow trials covering some of the possible biophysical conditions and

hedgerow configurations, Sanchez (1995) lists the conditions where alley cropping is

most likely to succeed: areas having fertile soils with no major nutrient limits, adequate

rainfall during the cropping season, sloping land with erosion hazard, ample labor and

scarce land, and secure land tenure. In spite of the apparent productive advantages

discovered in controlled studies and the enthusiasm of project extension technicians,

adoption among farmers remained generally low. This has been observed in many

regions, including Indonesia and the Philippines (Fujisaka 1991, Nelson et al. 1997),

Thailand (Ongprasert and Turkelboom 1995). Central America (Kass et al. 1995), West

Africa (Lawry et al. 1994), and East Africa (Mutsotso and Gicheru 1995). Several

reasons for poor adoption of hedgerows are given, including damage by grazing animals

(Fujisaka and Cenas 1993, Mutsotso and Gicheru 1995), land tenure issues (Fujisaka et

al. 1995, Lawry et al. 1994), and competing farm and off-farm activities (Fujisaka 1993,

Ongprasert and Turkelboom 1995), but by far the most common reasons are the high

costs of labor for installation and management (David 1995. Fujisaka et al. 1995, Kass et

al. 1995, Mutsotso and Gicheru 1995, Nelson et al. 1997, Ongprasert and Turkelboom








1995, Stark 1996) and farmers' need for short-term economic returns from their plots

(Fujisaka 1993, Ongprasert and Turkelboom 1995, Steiner and Scheidegger 1994).

Initial positive results from alley cropping trials were eventually modified by later

studies that showed decreased crop production from alley cropping in certain biophysical

settings or prohibitive costs of installation and management to farmers in certain

socioeconomic settings. The claims made for the potential of alley cropping. it was felt.

were exaggerated. But criticisms of alley cropping, summarized in a review written by

Sanchez (1995), have also been exaggerated due in part to its having been tested in

limited and inappropriate situations (Nair 1998). and may have caused an unwarranted

backlash among researchers (Vandermeer 1998).

Rao et al. (1998) reviewed 29 hedgerow intercropping trials that had been

conducted for four or more years, mostly on small plots, over a wide range of soils and

climates, but no studies done on sloping land were included. They found both positive

(15 studies indicated positive results for cereals. 8 trials showed positive results for

noncereal crops) and negative (13 trials showed negative results for cereals, 1 showed

negative results for noncereal crops) effects of hedgerow intercropping on crop yields

across the tropics. Including only the studies showing crop yield increases in hedgerow

systems greater than 15% (assuming this would be a minimum level to interest farmers),

positive effects on crop yield were found in only 2 of 10 studies in semiarid (<1000 mm

rainfall) regions. and 7 of 11 studies in subhumid (1000 to 1600 mm) regions where soils

were either inherently fertile or supplemented. In 4 out of 8 trials in the humid tropics

(>2000 mm) maize and taro (Colocasia esculenta (L.)Schott) did not benefit from

hedgerow intercropping, but bean and cowpea yields did. The authors conclude that the







18

hedgerow intercropping system is location specific and sensitive to management, and that

generalizations are difficult to make.

One can understand the caveats attached to prescribing or proscribing hedgerow

intercropping as a general solution; however, a more balanced approach to both research

and extension of hedgerow intercropping is now emerging. There is a growing

recognition that its success or failure depends on many site-specific factors including the

type of crops grown, the species of plants used as the structural component of the

hedgerows, rainfall amount and distribution, soil type, and a host of socioeconomic

variables (land and tree tenure, labor costs, etc.), including equity issues, extension

approaches, and allowance for adaptation of the practice over time to optimize its

economic returns to farmers (Mercer and Miller 1998, Nair 1998, Rao et al. 1998.

Sanchez 1995, Scherr and Current 1997).

Developments of Hedgerow Intercropping in Haiti

Hedgerow intercropping using Leucaena leucocephala (Lam.) de Wit was tried

on a small scale in Haiti for the first time in the early 1980s, but there are earlier

examples of vegetation-based soil conservation structures. Mintz (1962) notes that

Haitian farmers have been building seasonal crop stover barriers staked roughly on the

contour to harvest water and soil since the 1950s. These structures, known as ranp pay in

Creole, are still being built in some parts of the country. They last only a single planting

season. Some farmers say that stover barriers provide habitat for rats and insects, and so

they prefer to burn the crop residue. Contour structures have been promoted by

development projects since the 1950s. A UNESCO project in the community of Marbial

paid people to plant sisal (Agave sisalana Perrine ex Engelm.) on the contour to control

soil erosion in the early 1950s, and a FAO project near the town of Les Cayes in the late








1960s promoted contour hedges of napier grass (Penisetum purpuremn K. Schumach.).

lemon grass (Cymbopogon citratus (DC.) Stapf). vetiver (Vetiveria zizanioides (Linn.)

Nash), and guinea grass (Panicum maximum Jacq.) (G. Brice, personal communication:

1998). These were done by farmers who were paid stipends based on the number of

hedges installed, and they were not always appropriately designed. The napier grass

hedgerows were planted on shallow, drought soils. Many eventually grew taller than

their shallow roots could support, and toppled over. This also happened to some of the

sisal, a crop that is traditionally planted up and down the slope so that farmers harvesting

the sharply pointed leaves do not injure themselves if they fall.

In contrast to the historically widespread nonadoption of soil conservation

strategies designed by technicians, there is an example of a successful practice designed

by Haitian farmers responding to a particular market opportunity. This involves the

yearly, labor-intensive construction of contour soil bunds (tram) to conserve expensive

chemical fertilizers used to produce high-value vegetables in a small region close to the

capital city (Murray 1980).

Initial interest in tree-based hedgerow intercropping was stimulated by reports of

developments in West Africa, Indonesia, and the Philippines, and by the general interest

in the use of Leucaena leucocephala as a fuelwood tree. There is a variety of Leucaena

native to Haiti, known locally as delen, oriman, ti movye, or i pingi. This variety is not

suited to hedgerow construction because it is extremely invasive and does not produce

sufficient biomass, and is therefore very unpopular among farmers. To promote the use

of Leucaena, USAID introduced seeds of productive varieties to Haiti in 1978 (Benge

1985). Seeds ofLeucaena K8, K28, and K67 from trees grown in the Philippines were

distributed, and later seeds from Flores, Indonesia were introduced. In 1979, a USAID






20

agronomist also introduced the concept of using Leucaena in hedgerows, and re-wrote an

existing paper on Leucaena hedgerows especially for Haiti (M. Benge. personal

communication: 1995).

The number of hedgerows reported by PADF as built by farmers participating in

PLUS and its two predecessor projects probably represents the largest percent of the

hedgerows in Haiti. During the seven-year period from 1985 through 1991. 848 km of

hedgerows were reported (PADF 1990). The number of hedgerows installed has been

much greater under the PLUS project, with over 12,000 km. having been installed from

1993 through mid-1999 (PADF 1999). If the number of hedgerows ever installed by

farmers working with all aid projects in Haiti is estimated at 15,000 lineal km, and if

there is an average of 5 meters between rows, then approximately 7,500 hectares of

hillside gardens have had hedgerows. Although this represents less than one-half of a

percent of the total land area, it is a significant figure for a new technique whose

management is still being developed.

Hedgerows have been recommended as a potentially appropriate agroforestry

technology and are commonly promoted in Haiti as part of project or government-based

soil conservation and natural resource strategies (Bannister and Nair 1990). Garrity and

Van Noordwijk (1995) note three paradigm shifts in hillside conservation management:

(1) an engineering approach using rock and earthworks in contour structures has

increasingly given way to vegetative structures; (2) top-down prescriptive watershed

management has evolved to a bottom-up, participative approach; and (3) tree-based alley

cropping has diversified to include a wider array of economically interesting plant

materials as structural components of contour hedgerows. All three of these shifts have

occurred in Haiti. Early attempts at erosion control promoted by projects in Haiti were








mainly based on rock walls constructed across the contour, but these were poorly

maintained because they did not respond to farmers' economic situation and the

extension approach was inappropriate (Murray 1980). The increasing emphasis on

vegetative barriers happened during the 1980s, as noted above, along with as-yet

incomplete attempts to manage natural resources participatively.

However, the ideotype of contour hedgerows is not static in Haiti. Farmers

continually modify the structural composition and management of hedgerows according

to their household needs, their ability to invest resources, their experience with

hedgerows. and changing marketing opportunities for crops and products derived from

hedgerow trees. Leucaena leucocephala was originally the predominant hedgerow

species, but that is changing. During 1995, PADF reported hedgerows made with 22

different tree, perennial food crop, grasses, and other perennial species planted in 54

different combinations. Leucaena was present in only 46% of the hedgerows. In

addition, hedgerows are evolving into perennial crop contour bands in some areas. These

bands, called bann manje (a play on words-either "band of food" or "a lot of food")

consist of rows of food crops planted across the contour, with a dimension up and down

the slope of more than one meter. Perennial crops such as pineapple (Ananas comosus

(L.) Merr.) or sugar cane (Sacharum officinarum L.) serve as the structural components,

but annuals such as yam and sweet potato (Lpomoea batatas (L.) Lam.) are also planted

within the band. Field crops are planted in the interspaces, as in hedgerows. During

1999. 20 different combinations of nine species of perennial food crops, trees, and other

perennials were used in crop contour bands (PADF 1999). This evolution is based on

farmers' needs for short-term cash return. The structural crops, such as pineapple,

respond directly to a marketing opportunity-a quintessentially Haitian strategy (Murray






22

1991), and one found in indigenous agroforestry associations in other cultures (Khaleque

and Gold 1993). This reinforces the notion that hedgerow intercropping should be

considered as a seed technology (Sumberg and Atta-Krah 1988. Wiersum 1994). to be

adapted and modified by farmers as a matter of course to fit into their unique agricultural

systems as they discover (or not) its as-yet unknown economic possibilities.

The cycle of enthusiastic promotion of hedgerow intercropping followed by

disappointment has a parallel in Haiti, although it is not documented apart from some

protest about the widespread promotion of Leucaena hedgerows in the early 1990s

(Swanson et al. 1993d, 1993e). It mainly takes the form of disappointment, informally

expressed by officials of donor agencies, that hedgerows do not yet cover every garden in

Haiti. More often the criticism of nonadoption has been aimed at the extension

methodology rather than the practice per se (Sieder 1994,Villanueva 1993). This has

been due, in part, to the development-driven agenda mentioned by Nair (1998) resulting

in the promotion of large numbers of an untested practice by donor-funded projects.

Reasons for farmer adoption, adaptation, or rejection of hedgerows are not well studied

in Haiti. The best available studies were done in a small region of the central plateau

near the town of Maissade (White 1992b), which noted farmers' reasons for nonadoption

of hedgerows as lack of time (39%), don't own land (21%). and land not appropriate

(17%). Vaval (1997) studied the same project area later, and documented the

disadvantages of hedgerows as expressed by adopting farmers as the time required for

management and loss of crop land due to the presence of hedgerows.

Hedgerow Intercropping and Soil Conservation

Because of the greater risk of soil erosion, performance of hedgerows on sloping

lands is not identical to their performance on flat land. Three kinds of biophysical








differences to consider include (1) the overall effect of hillside contour hedgerows on

crop production and soil erosion. (2) the gradients of soil accumulation/loss, soil physical

properties, tree/crop competition, and crop yield across the overall slope from the top

hedgerow to the bottom one, and (3) the same gradients within the alleys.

The concerns about the poor performance of alley cropping summarized by

Sanchez (1995) did not extend to hedgerow intercropping on sloping land. In part, this is

due to the proven ability of contour hedgerows to control soil erosion. Sanchez (1995)

stated that the hypothesis that contour hedgerows can substantially reduce soil erosion

has been definitively proven. This has been shown in many areas of the world, for

example a review often years of research in Thailand concluded that hedgerows

controlled erosion in areas of 18 to 55% slopes (Ongprasert and Turkelboom 1995); in

Rwanda on 23% slopes hedgerows reduced erosion to less than 2% of the runoff from

unprotected plots (Roose and Ndayizigiye 1993); in Malawi on 44% slopes hedgerows

reduced soil erosion to 2 t/ha compared to 80 t/ha on maize-only plots (Banda et al.

1994); in India on 9% slopes hedgerows significantly reduced soil erosion compared to

cassava-only plots or bare fallow (Ghosh et al. 1989); and in the Philippines there are

many examples of hedgerows substantially reducing erosion on sloped land (Agus et al.

1999, Comia et al. 1994, Paningbatan et al. 1995. Tacio 1993). The studies cited cover a

range of soil types from sandy loams (Banda et al. 1994) to gravelly sandy clays (Ghosh

et al. 1989) to heavy clays (Comia et al. 1994).

However, in spite of their proven ability to control erosion, in many cases

researchers have not been able to show crop yield improvements in hedgerow plots

compared to traditional systems. The principal reason for this appears to be the same on

slopes as it is in flat land alley cropping systems--the tradeoff between the production of








biomass for mulch or green manure needed to increase crop yield, and the loss of

cropping space caused by the presence of the hedgerows that produce the biomass

(Vandermeer 1998). The inter-row spacing required for the crops does not allow enough

biomass production from the hedgerows to improve soil fertility, so maize yield, for

example. is not improved unless inorganic fertilizer is applied (Ongprasert and

Turkelboom 1995, Roose and Ndayizigiye 1993). On steep slopes the Soil Changes

under Agroforestry (SCUAF) model (Young and Muraya 1990) predicts this might be

explained by high rates of leaching and organic matter decomposition, despite reductions

in the rate of erosion (Nelson et al. 1998). Crop yield declines are especially important in

newly established hedgerows as biomass application will not have yet contributed much

to soil improvement, although cropping space will have been reduced. Data taken from

89 hedgerow plots planted with maize on sloping plots in the Philippines showed a

decrease in maize yield in new hedgerows due to the loss of cropping space. The same

study estimated 8 years would be required for soil improvements caused by hedgerows to

compensate for loss in maize yield at a 5 m inter-row spacing (Shively 1998). In some

cases, however, soil conditions are improved by hedgerows sufficiently to maintain a

stable, if low, level of maize production on slopes compared to traditional cultivation,

where maize yield falls off over time (Banda et al. 1994, Comia et al. 1994). There are

also examples where crop yields in hedgerows on slopes were significantly greater and

more stable than those in unprotected control plots, as in a four-year study in Costa Rica

on farms having relatively fertile soils (Kass et al. 1995).

Gradients of soil water, soil physical properties, and crop yield have been

documented in alley cropping systems, as noted earlier in this chapter. For example,

infiltration rate and number of soil macropores were greater under the hedgerow trees






25

than in the middle of the alleys on a 14% slope in Kenya (Kiepe 1995). Competition for

light, water, and nutrients between hedgerow plants and adjacent crops is also noted on

slopes. Rice yield was depressed in the first two rows away from hedgerows on 15%

slopes in the Philippines, but the depression varied with different hedgerow species; it

was greatest near grass (Penisetum purpurenm) hedgerows (Garrity and Mercado 1994).

It has also been found that food crops in the alleys can restrict the growth of adjacent

hedgerow trees (Ghosh et al. 1989).

But as the slope increases a redistribution of soil fertility across the alley is often

seen that differs from that of flat plots (Garrity 1994, Ongprasert and Turkelboom 1995).

Increases in soil macropores near Calliandra callothrysus hedgerows in Reunion were

greater on the uphill side of the hedges than on the downhill side (Tassin et al. 1995). A

study done in the Philippines on 22 to 30% slopes found that average mid-alley maize

yields were greater than yields near hedgerows of Gliricidia sepium, Paspalum

conjugatum. and Penisetum pwupureum; and that the Ap soil horizon was thicker on the

lower part of the alleys (Agus et al. 1999). Garrity (1996) reviewed this phenomenon,

and noted that it was caused in a large part by the displacement of soil during crop

tillage. He has pointed out that on unprotected slopes there is often no net loss of soil

from the mid-slope area because any soil loss is replaced by soil eroding from the upper

portion of the hillside. The presence of hedgerows prevents a net accumulation of

upslope soil into the alleys, but the same displacement effect occurs within the alleys on

a smaller scale. Crop yield response on sloped hedgerow plots skews after a few years of

cultivation because scouring due to tillage displaces soil from the upper part of the alley

to the lower part, changing the soil profile, slope length, soil depth, and hydrological

properties. These changes combine with competitive interactions between the crop and

the hedgerows. The result is lower crop yields on the upper parts of the alleys and higher






26

yields on the lower middle part, with a slight decrease again as crops approach the lower

alley near the next hedge. The net yield curve across the alley is due to a combination of

soil scouring and competition with hedgerow trees. The effect of soil scouring on yield

is usually more important than that of tree/crop competition. Solera (1993), cited in

Garrity (1996) put in 50 cm plastic barriers between Gliricidia sepium hedgerows and

adjacent upland rice on a 20% slope, and found that the across-alley yield response was

unchanged by the barriers, and so was not due to competition between the hedgerows and

crop. Yield in the upper rows was 50% lower than that in the lower rows within the

alleys. The effects of slope on hedgerows are important because in wide alleys the soil

displaced from the upper position may lead to long-lasting crop yield depression in that

area, and because invalid comparisons between yield in hedgerows are often made

against an adjacent no-hedgerow control plot in the mid-slope area that has thicker soil

(Garrity 1996).

Conclusions

Hedgerow intercropping is an evolving technology meant to address the

increasingly rapid destruction of soil resources in developing countries. The defining

paradox in its history is that researchers and extension agents initially addressed it as a

technical solution, whereas farmers looked at it primarily as a potential economic

solution. Crop yield improvements due to hedgerow intercropping in certain regions of

the world justify continuing research. However, it has not yet been widely adopted even

in those regions, mainly due to high initial costs of installation and adaptation and farmer

need for short-term returns in the face of the probability of a yield loss in hedgerow plots

during the first few years after installation. Both the technical and socioeconomic

aspects of hedgerow intercropping must be resolved in order for adaptation and adoption

to increase.













CHAPTER 4
ORGANIZATION OF THE STUDY

This study is presented in three parts, beginning with an on-station study of

competitive interactions between hedgerows and adjacent maize, followed first by an on-

farm study of the performance of maize in hedgerows on sloping plots, and then by an

extensive field survey of the biophysical characteristics of plots and socioeconomic

characteristics of farmers who have constructed hedgerows and other soil conservation

practices with the PADF/PLUS project. The hypotheses to be tested correspond to some

of the biophysical and socioeconomic hypotheses referred to by Sanchez (1995) as being

fundamental to agroforestry, namely, biophysical hypothesis number 16: competition for

water in agroforestry systems can be reduced by modifying the spatial arrangement of

trees; and socioeconomic hypotheses numbers 1, 3, and 4:

the identification of key driving socioeconomic and ecological processes
within land-use typologies permits the spatial delineation of target and
recommendation domains for agroforestry interventions;

the adoptability of a new agroforestry practice is determined by 5 principal
components-the farmers' natural resource base, their resource endowment,
degree of market integration, cultural preferences and perceived benefits; and

at the farming systems scale, agroforestry adoption has different impacts for
different classes of farmers, such as their gender.

Justification

Declining farm productivity is a major problem confronting Haitian hillside

farmers. It is caused and exacerbated by a number of factors. Important among these are

soil degradation as a consequence of erosion. and farmers' difficulty in making capital








investments to support improvement of lost soil fertility and farm productivity.

Traditional agroforestry systems such as home gardens, shade coffee, and living fences

have not been able to cope with the degradation of soil resources on hill slopes as

population increased. Although rural migration to the cities has steadily increased

(Locher 1996), there is no reason to suppose that this will result in a significant decrease

in the population density in the mountains. Haiti's population might already have

exceeded the ecological carrying capacity of the land (Ewel 1977). Two decades ago, the

Haitian landscape was described as the victim of mismanagement: "the ruthless

application by the rural population of ecologically inappropriate and catastrophic lumber

extraction and gardening techniques which have transformed the once lush landscape of

entire regions into grim, denuded savannas" (Murray 1980, page 3). In addition to

decreasing productivity of individual hillside plots, inappropriate farming practices also

led to other local and regional environmental problems, including transport of rock

overburden that covers productive soil of farms on lower slopes and plains, reduction in

the water supply from springs, road washouts, and destruction of houses by water and

debris flows.

Hedgerow intercropping has gone through a cycle of research, widespread

promotion based on expected positive performance. extension by governments and

funded development projects, mounting criticism when expected performance was not

achieved, a resulting waning interest on the part of researchers and projects, and a

concern that alley cropping might be rejected due to inadequate identification of the

biophysical and socioeconomic conditions where it might succeed. This point was

discussed in more depth in Chapter 3.








Consequent to the reported promise and potential of hedgerow intercropping

technology in other parts of the tropics in the 1980s (Kang et al. 1984). it was introduced

in Haiti in the middle 1980's through externally funded development projects (Murray

1997. PADF 1986). As with most project-driven interventions, hedgerow intercropping

has had varying degrees of success. Several reasons could be attributed to this, both

biophysical and socioeconomic. They might include inappropriate extension approaches.

mismatch between farmer objectives and project objectives, inappropriate choice of

hedgerow species, competition between hedgerows and adjacent crops, high cost of

hedgerow establishment and management relative to the resulting effect on crop yield,

and farmers' and technicians' ignorance of the costs and benefits of hedgerows. But the

list of relevant factors contributing to hedgerow success or failure is poorly understood.

Indeed, farmers, extension agents, technicians, and project administrators have varying

definitions and perceptions of success and failure. The same cycle of widespread interest

followed by disappointment seen on the international level has been repeated on a

smaller scale in Haiti, where hedgerow intercropping has sometimes been labeled a

failure by consultants and donors who tend to observe the practice at an early stage of

development, in a small number of locations, or without a complete understanding of

important nontechnical parameters of the practice.

During the period 1984 to 1999, hedgerow intercropping technology has been

installed on approximately 10,000 hectares by farmers working with various funded

projects in Haiti (PADF 1990. PADF 1999). This is a significant coverage for a

development activity involving a relatively new and evolving technology. The rate of

hedgerow construction appears to have surpassed the level of understanding of how and

why they are installed and managed.








A number of studies have been published regarding various aspects of

hedgerow intercropping in Haiti. These include the effect of hedgerows on soil erosion

(Cunard and Lorcy 1990, Regis 1990), financial analysis of some of the benefits of

hedgerows (Bellerive 1991, Lea 1993. White and Quinn 1992). choice of hedgerow

species (Isaac and Shannon 1994, Shannon and Isaac 1993. Toussaint 1989. Treadwell

and Cunard 1992). management of hedgerow tree biomass (Isaac et al. 1995). farmer

adoption (Chary 1990, Emmanuel 1990, Lea 1994a. 1994b. Villanueva 1993. White

1992a), and farmer adaptation and management (Lea 1995, Pierre et al. 1995. Swanson et

al. 1993a, 1993b, 1993c, 1993d, 1993e. White 1992b).

Of the nine papers cited above reporting adaptation and management of hedgerow

intercropping in Haiti, only the one by Pierre et al. (1995) was specifically designed to

study how farmers manage hedgerows. They studied at 105 hedgerow gardens installed

before March 1994 in one region of southwestern Haiti. Although the study reported a

number of valuable observations on the management and performance of hedgerows, it

did not correlate the management quality indicator (the number of unrepaired breaches)

with other factors. Furthermore, the observations were based only on hedgerows

constructed of Leucaena leucocephala in a limited area of project intervention.

Another major study on hedgerows in Haiti was the series of qualitative,

questionnaire-based diagnostic surveys (Swanson 1993, Swanson et al. 1993a. 1993b.

1993c. 1993d. 1993e,) done just after the first field season of the PLUS project at the

request of PADF and Cooperative for American Relief Everywhere (CARE), the

implementing organizations. These studies were targeted at the whole farming

environment, not particularly at hedgerows. The diagnostic team visited over 200

farmers in fifteen small watersheds, interviewing individual farmers in their plots and






31

speaking with groups of farmers. The majority of the hedgerow gardens seen by the team

were installed under the Agroforestry II (AFII) project (1990-1991). The number of

hedgerow gardens visited was not stated, and the survey did not attempt to link hedgerow

management to specific conditions on the farms. Several useful observations were made

regarding hedgerows, however both the composition and management of hedgerows have

evolved since that time. and the number of hedgerows installed by farmers has increased

dramatically. In addition, an evaluation of the PLUS project monitoring and evaluation

system done in 1995 (Romanoff et al. 1995) suggested that the sampling areas used were

not representative of the whole project. They recommended that project impact be

measured over the whole area of intervention.

Competition between hedgerow trees and adjacent crop plants for soil water has

been the subject of an increasing number of studies in recent years (Chirwa et al. 1994a,

Govindarajan et al. 1996, Korwar and Radder 1994, 1997, Ong et al. 1991, Rane et al.

1995, Schroth et al. 1995a, Schroth and Zech 1993, Ssekabembe et al. 1994). This is

potentially a problem in Haitian hedgerow gardens because of the highly variable timing

of seasonal rainfall and the inability of eroded soils on steeply sloping land to absorb and

store water. Ewel (1977) referred to this as pseudo-drought, a condition that results from

a decrease in the water-holding capacity of the soil rather than from decreased rainfall.

The literature on hedgerow-crop soil water competition shows that, under some

conditions, hedgerows compete to the disadvantage of adjacent crops (Fernandes et al.

1993), (Govindarajan et al. 1996. Kamasho 1994. Korwar and Radder 1994, Ong et al.

1991. Salazari et al. 1993, Singh et al. 1986) and under other conditions, they do not

(Bohringer and Leihner 1997, Chirwa et al. 1994a. Haggar and Beer 1993, Jeanes et al.

1996, Schroth and Zech 1993).








One possible influence on farmer adaptation and management of hedgerows is

farmers' perception of the importance of soil water competition between the hedgerow

species and the interplanted crops. Experience in Haiti shows that maize (Zea mays) and

bean (Phaseolus vulgaris) crops, even without hedgerows, very often fail because of

drought conditions. Therefore, hedgerow trees are likely to compete with interplanted

crops for soil water at some time during the cropping cycle. The degree of competition

may also vary according to the position of the crop on the slope relative to the hedgerows

(Garrity 1996). However, it is not clear if farmers consider this competition as more or

less important than other perceived problems caused by hedgerows. Do farmers modify

hedgerow structure and management to compensate for perceived soil water

competition? Neither competition for soil water between crops and hedgerows. nor crop

yield relative to slope position and /or relative to hedgerows have been studied in Haiti.

Since intensive agriculture is practiced on small steeply sloping plots, it is

necessary to look for economically rational ways to maintain the soil's ability to sustain

crop production and to protect the local environment. Development of new agroforestry

systems that build upon traditional practices is needed. Hedgerow intercropping has been

shown to produce desirable results under certain circumstances. However, adoption or

lack of adoption may depend on factors extraneous to the characteristics of the

technology itself, including the socioeconomic status of farm families and the kinds of

land they access for agriculture. The extremely degraded condition of Haitian natural

resources makes it necessary to try any practice that could improve the economic

condition of Haitian farmers. Because of experience gained from agroforestry projects

operating in Haiti since the 1980s, there is reason to believe that adaptations of hedgerow

intercropping may increase farm stability and productivity in some areas of the country.






33
This study will help inform decision makers regarding the potential for use of hedgerow

intercropping in Haiti, and possibly avoid premature decisions to discontinue the effort to

adapt the technology.

Thus, hedgerow intercropping, a rapidly evolving and yet unproven technology.

has been extended in Haiti without the necessary supporting studies and documentation.

Development and extension resources could be used more efficiently by responding to

two documented shortcomings of agroforestry research that seem particularly appropriate

to the Haitian situation: that there is an inadequate understanding of below-ground

biophysical interactions in hedgerow systems; and that socioeconomic and biophysical

research must be linked, as adoption of the appropriate agroforestry practice depends

upon both (Rao et al. 1998).

Hypotheses

On-station Study

Leucaena hedgerows reduce the amount of soil water available to adjacent
maize, thereby causing a reduction in maize growth and yield.

Root barriers installed next to Leucaena hedgerows increase the amount of
soil water available to adjacent maize, thereby causing an increase in maize
growth and yield.

Cutting the roots of Leucaena hedgerows affects the production of tree
biomass.

On-Farm Study

*The pattern of competition between hedgerow trees and adjacent crops is
different on slopes vs. flat land, and maize growth varies with regard to the
position of the crop within the alleys and on the slope.

On-Farm Survey

SFamily characteristics influence the adoption and management of hedgerow
intercropping.







Plot characteristics influence the adoption and management of hedgerow
intercropping.

Problems and benefits with hedgerow intercropping as perceived by the
adopting farmer influence management quality.

General Description of Hypothesis-Testing Procedures

The study of soil water competition between Leucaena hedgerows and adjacent

maize was done on an NGO-operated research station located near Port au Prince. Four

treatment combinations were tested in nine replications: hedgerows with and without root

barriers between the trees and the adjacent row of maize, and maize fertilized or not

fertilized. Data for three maize seasons are reported, including gravimetric soil water:

maize height, leaf width, growth stage, leaf water potential, and final harvest: and

Leucaena height and biomass production. A pre-study examination of the distribution of

Leucaena hedgerow roots at a different location was done, and a post study count of the

distribution of Leucaena roots in hedgerows with and without barriers was done on-

station. A pot study of the correspondence between soil water percent and maize leaf

water potential was done as well.

The small on-farm study was an attempt to relate the on-station findings on flat

land under "controlled" conditions to sloping land managed by farmers. The study

collected maize growth and yield data for one season from four farms having previously

established trials of Leucaena hedgerows, crop contour bands, rock wall terraces, and

traditional cultivation. Maize growth data were collected by position in the alleys and on

the slopes. The four farm plots had been a part of an abandoned PLUS project research

effort previously carried out by the Southeast Consortium for International Development

(SECID)/Auburn University in conjunction with PADF. The author designed the data

collection protocol and performed the analysis.






35
A large list-frame questionnaire-based study was conducted, which provided on-

farm data about households adopting agroforestry soil conservation practices (including

hedgerows), how they were managed, and the characteristics of the plots where they were

installed. The study was undertaken by the author with the help of PADF PLUS

technicians as part of project monitoring during the early part of 1996. We visited 1.540

PLUS project agroforestry adopters and collected information regarding all plots

controlled by the family, and information regarding family members. We then visited

2.295 plots where farmers had installed at least one PLUS project agroforestry practice.

Data were collected on the biophysical parameters of the plots, the amount and condition

of the agroforestry practices installed on the plot, and the opinions of the farmers

regarding problems and benefits of the practices. Because farmers' assessment of soil

fertility appeared to influence where hedgerows were installed, soil samples were

collected from 175 farms. These were analyzed for percent organic carbon and soil

nutrients, and compared to farmers' classification of soil fertility.

Details of data collection and analysis for the on-station and on-farm

hedgerow/crop competition studies are reported in Chapter 5. Those for the on-farm

survey are reported in Chapter 6.













CHAPTER 5
HEDGEROW/CROP COMPETITION

Introduction

This chapter looks at competitive interactions between hedgerows and adjacent

maize on station and on farm. Competition is important because it is a key consideration

in designing agroforestry practices, it is difficult to measure, and farmers make adoption

and adaptation decisions based on its perceived severity. Competition is a central issue

in agroforestry because of the potential advantage in combining plants in such a way that

they capture more growth resources in combination than they would growing separately,

thereby producing greater physical yields (Cannell et al. 1996). The cost of the positive

effects of facilitation gained by growing plants in close proximity is the negative effects

of competition. In hedgerow intercropping in particular, the competitive characteristics

of the trees (they take up space, light, soil water, nutrients) must be balanced with their

facilitative characteristics (they impede soil erosion, capture soil water and nutrients out

of reach of the crops, improve soil fertility through biomass additions, improve the

microclimate, improve soil physical properties) (Vandermeer 1998). It is hard to develop

a hedgerow system that attains that balance because so many combinations of plants.

physical environments, and socioeconomic situations are possible. It is also difficult to

measure competition and to identify the processes that contribute to it because

competitive interactions change as the environment changes (Anderson and Sinclair

1993), and the hedgerows themselves change the environment-indeed, that is their






37

function. The way hedgerows and adjacent crops interact varies according to the age of

the hedgerows, relative size of the crops and hedgerows, population density, and the

supply of and access to limiting growth resources (Ong and Leakey 1999).

Farmers make adoption decisions based on competition between hedgerows and

crops. For example, farmers in the Philippines eliminated napier grass as a hedgerow

component because it was too competitive with crops grown in the alleys (Fujisaka et al.

1995), where it decreased maize yield by 86% by the second year after establishment

(Garrity and Mercado 1994). Therefore. competitive characteristics are important

considerations in the design of hedgerow systems.

Leucaena has characteristics that make it an aggressive competitor for growth

resources, including rapid growth and roots that occupy approximately the same soil

depth as that of maize (Jonsson et al. 1988), the crop of interest in this research. It was

noted several decades ago in Southeast Asia that Leucaena used as shade trees in

plantations rarely competed with the crop plants; but did so in times of drought, when

signs of water competition with coffee and rubber were reported (Dijkman 1950). In

more recent alley cropping studies, competition is a frequent subject of investigation, but

results vary with the specific site and plant components. Leucaena has been shown to

compete for soil nutrients to the detriment of rice (Salazari et al. 1993) and grasses

(George et al. 1996). In the semiarid tropics especially it has been shown to reduce the

amount of soil water available to pigeon pea, but less so for sorghum and millet

(Pennisetum glaucum L.) (Singh et al. 1986). In other areas of the semiarid tropics,

however, yields of both sorghum (Korwar and Radder 1997) and millet (Ong et al. 1991)

were significantly reduced in Leucaena hedgerows.








The on-station and on-farm studies reported on here both have maize planted in

the interspaces between Leucaena hedgerows. This combination has produced both

positive (Akonde et al. 1996, Banda et al. 1994, Hauser and Kang 1993, Hernandez 1961,

Mugendi et al. 1999, Mureithi et al. 1994) and negative (Chirwa et al. 1994b.

Govindarajan et al. 1996, Jama et al. 1995, Kamasho 1994) results. The positive results

in the cited studies ranged from maintaining the existing maize yield (versus a decrease

without hedgerows) to increases of 26%. 44%, 50%, and 400% in grain yield compared

to traditional cultivation. Positive results tended to be from subhumid or humid areas

with soils of moderate or better fertility, and from sloping areas. The negative results

ranged from 16% to 20% decreases in maize yield. These tended to be from semiarid,

moisture limiting, low fertility sites.

Effects of competition between maize and Leucaena are noted in several studies,

even where the overall effects of hedgerows are positive. In an early report of alley

cropping research in Ghana, Balasubramanian (1983) mentioned that soil water

competition with Leucaena might be a partial explanation of yield depression in adjacent

maize. Later hedgerow studies have attributed yield depression in adjacent crops to

competition for soil water (Govindarajan et al. 1996, Korwar and Radder 1997, Ong et al.

1991. Ssekabembe et al. 1994), nutrients (Chirwa et al. 1994b, Livesley et al. 1997,

Mureithi et al. 1994). or light (Kang et al. 1981. Lal 1989, Tilander et al. 1995, Welke et

al. 1995. Zoschke et al. 1990). Some studies claim all three resources are competed for

(Gaddanakeri et al. 1994). It is difficult to use these location-specific results to make

predictions about the probable success of a hedgerow intercropping system in another

location. However. Rao et al. (1998) have drawn some general conclusions: (1) In water

limiting areas, low yield of hedgerow prunings and competition for water between

hedgerows and adjacent crops were the major reasons for the negative results.








Inadequate water limited the response of crops even though hedgerows improved soil

fertility in certain sites of the semiarid tropics. (2) In poor soils, low yield of hedgerow

prunings and competition for nutrients were responsible for negative results. (3) The

advantages of hedgerow intercropping are more common on relatively fertile. N-deficient

soils in subhumid and humid environments, in areas where hedgerows produce a large

quantity of prunings, and where there is adequate water for both hedge and crop growth.

They also note that if competition is substantially reduced, even in semiarid climates.

hedgerows may be able to increase crop yields even where relatively small fertility

contributions are made by the hedges-but the cost of managing the hedges could be too

high to be practical.

Much of Haiti is subhumid, and there are still soils having a little remaining

fertility, on sloping land where hedgerows may be appropriate. The discussion to be

presented in this chapter will focus on the biophysical aspects of hedgerows and so will

necessarily be incomplete, lacking the socioeconomic elements, which will be discussed

in Chapter 6.

On-Station Studies: Effect of Root Barriers on the Growth of Maize and Leucaena

The objectives of this experiment were to test whether or not permanent root

barriers placed on either side of Leucaena leucocephala hedgerows affect the production

of maize grown in the alleys and the production of Leucaena leucocephala biomass. The

hypotheses were that the barriers will reduce competition for soil water between

Leucaena and adjacent maize, thereby increasing maize yield; and that production of

Leucaena will not be adversely affected by the barriers. If maize yield were to be

increased significantly an appropriate modification of the barrier could then be devised

for use on-farm, or a less competitive tree could be sought to construct hedgerows.






40

Crop yield depression close to trees is common. It has been reported for mustard

growing under Acacia nilotica in semiarid central India, where a yield depression of 54%

extending up to 8 m from the tree was correlated with increasing shallowness of the tree

roots (Yadav et al. 1993). Yield depression of maize and soybean near Siberian elm

(Ulnmus pumila) and eastern cottonwood (Populus deltoides) windbreaks in Nebraska was

significantly relieved by plowing parallel to the trees (Rasmussen and Shapiro 1989). A

study done in the savanna region of Kenya reported that Acacia tortilis trees competed

more intensely with understory grasses and herbs in wetter sites where the tree roots

terminated near the crown region, than in drier sites where the roots extended farther into

the grassland (Belsky 1994). Yield depression in millet extended significantly beyond

the area shaded by eucalyptus shelterbelts in Nigeria, corresponding to the area of tree

root growth (Ojyewotu et al. 1994). Nadagouda et al. (1994) found that severing the

roots of eucalyptus trees in India improved yield of adjacent maize.

Unless effective experimental controls are put into place, it is difficult to separate

the effects of above- from below-ground competition, and the effects of competition for

water from those of nutrients. Corlett et al. (1992b) describe a study done in the semiarid

tropics of India in which millet was grown alone and in Leucaena hedgerows. either with

or without barriers between the tree and crop roots. Millet yield was the same in the sole

crop and in the hedgerows with root barriers, but was 36% less in the hedgerows without

barriers. The authors proposed that, although one might have concluded that the yield

difference was due solely to below-ground competition. a better explanation was that the

millet adjacent to hedges with barriers grew taller because it had access to more soil

water than millet in hedgerows without barriers, and avoided light competition with the

Leucaena. There was an interaction between above- and below-ground processes.








Common methods used for studying the effects of below-ground competition in

alley cropping research include root pruning (Fernandes et al. 1993, Korwar and Radder

1994, Sitompul et al. 1992), trenching (Schroth et al. 1995b) or the installation of

permanent plastic barriers between the hedgerow trees and the crops (Corlett et al. 1992b.

Hauser and Gichuru 1994, Ong et al. 1991, Schroth et al. 1995a, Schroth and Lehmann

1995, Schroth and Zech 1993, Solera 1993c, Ssekabembe et al. 1994). The usual practice

is to open a narrow trench to a depth of 50 to 100 cm between the hedgerow trees and the

first crop row, install plastic awning or polyethylene sheeting, and then fill in the trench.

Studies using plastic barriers have produced results specific to the characteristics

of the site, species of hedgerow tree and crop. alley width, planting density and distance

between the hedgerow and crop, and the top-pruning regime (height and timing of cut)

used on the hedgerow. Ong et al. (1991) showed that yield of millet in hedgerows

without barriers was 40% less than yield in sole-cropped millet in semiarid India, but

where barriers were parallel to hedgerows the yield was the same as that in sole cropping.

They also showed that the barriers did not affect the production of the Leucaena

hedgerows. but the barriers did cause a redistribution of the Leucaena roots. Schroth and

Lehman (1995) found in subhumid Togo that barriers increased the yield of maize in

Calliandra callothyrsus hedgerows, but when the hedgerow species were Senna siamea

or Gliricidia sepium the presence of barriers was associated with a decrease in maize

yield. They concluded that Calliandra was relatively competitive compared to the other

trees, and that the barriers reduced the volume of exploitable soil available to the roots of

the other two species. The results obtained by studies using plastic barriers between trees

and crops do not always show competition for soil water. Two studies done in the West

African humid tropics put plastic barriers to 90 cm between Gliricidia hedgerows and








adjacent crops of rice, maize, and peanut. They showed crop yield improvement in

hedgerows compared to the sole cropping controls, but crop yield in hedgerows was not

affected by the presence of barriers (Schroth et al. 1995a, Schroth and Zech 1993). A

study done in the United States was able to isolate the effect of Robinia pseudoacacia

hedgerows on soil water at two sites having different soil characteristics (Ssekabembe et

al. 1994). The plots were irrigated, no crops were grown in the alleys between the

hedges, and the soil was covered with plastic to reduce evaporation. At the site having a

higher soil organic carbon content, hedgerows without barriers reduced soil water content

of the alleys by about 8%. At the second site, where higher gravel content impeded the

growth of tree roots, hedgerows without barriers reduced soil water content by 32%. and

there was significantly less water in alleys without barriers.

Materials and Methods

The experiment was carried out over a period of six years, from the spring of

1990 through the spring of 1995. During this period data collection was interrupted

several times by civil unrest in Haiti,' so although six maize crops and one sorghum crop

were planted only three successive maize crops from spring 1994 though spring 1995 are


'There was a coup d'etat in September 1991, which resulted in several missed
observations due to curfews. A night guard at the Operation Double Harvest (ODH) site,
where the experiment was installed, was killed by soldiers during that period, preventing
pre-dawn leaf moisture potential measurements. I was evacuated to the U.S. in October
1991 and returned to Haiti in September of the following year. By that time. the
Direction of ODH had changed hands, requiring that I obtain permission to continue my
work on the plot. This took several months, and the removal of a leaking irrigation line
installed next to the plots was necessary. During the regime of the defacto military
government (September 1991 to September 1994). fuel shortages were common because
of an international embargo, again interrupting data collection. Farmers with friends in
the local police station let donkeys into my plots, ruining the Fall 1993 harvest. Farm
workers bathing in the plots behind the screen provided by the Leucaena also prevented
soil moisture observations from being taken at least twice. The U.S military intervention
in September 1994 made adequate data collection difficult during that period as well.







discussed. Data available from the three maize crops were not identical, again due to

civil unrest and equipment problems. The spring 1991 data are used only for the

discussion of the effect of trenching on Leucaena height and biomass.

Trial site

The trial site was located at 18033' north latitude 721l 1' west longitude, with an

elevation of 80 m above sea level on property owned by Operation Double Harvest

(ODH), a nongovernmental organization (NGO) operating a farm and tree nursery in the

Cul de Sac plain, 30 minutes northeast of Port au Prince near the town of Croix des

Bouquets, Haiti. The soil was a vertisol of zero slope. From observations made in a soil

pit near the trial plots, heavy texture predominated to a depth of 1.1 m, where there was

an abrupt change to sandy, unconsolidated limestone to at least 1.7 m, the depth of the

pit. Soil texture from the surface to 30 cm depth was silty clay, changing to silty clay

loam in the 30-45 cm depth. Table 5-1 shows texture and organic carbon content for the

trial site.


Table 5-1: Soil texture and organic carbon at the trial site.

% Organic
Depth % Sand % Silt % Clay Carbon Texture
0-15 cm 8.4 49 42.6 1.71 silty clay
15-30 cm 7.2 41 51.8 1.10 silty clay
30-45 cm 7.2 53 39.8 1.0 silty clay loam


A pot study was established in January 1994 to determine how soil water

depletion at the ODH site correlated with plant water stress (Appendix A). When soil

water percent in the root zone decreased below about 25%, leaf water pressure passed






44

below 10 bars. Maize plants began to exhibit signs of drought stress at this point. Field

capacity at the site was reached at 41% soil water content.

Average annual rainfall based on data collected by ODH staff since 1977 was 968

mm per year. Distribution is bimodal, with most of the spring rains falling from April

through June. and the fall rains from August through November. The total annual rainfall

since 1977 varied from a low of 670 mm to a high of 1.301 mm. but in most years

rainfall was at least 900 mm. The trial site was located to the north of a small-container

tree nursery, and was previously used for tractor cultivation of hybrid sorghum. The area

to the east was used for composting sugarcane bagasse. A low earth barrier separated the

trial area from the composting area, to prevent water used for composting from entering

the trial.

Trial design

The trial design was a 2 by 2 factorial randomized complete block having 11

hedgerows of Leucaena leucocephala variety K636 planted at a within-row density of 10

trees per meter and a between row spacing of 5 meters. Each block consisted of a

hedgerow 22 meters long running through the center of four maize plots, to which were

randomly assigned the four treatment combinations. The hedgerows were laid out

exactly on an east-west axis. with hedgerow (block) 1 on the north end of the site and

hedgerow (block) 11 on the south side (Figure 5-1). The symbol T+ denotes a trenched

plot. T--a nontrenched plot: the symbol F+ denotes a fertilized plot. F--a nonfertilized

plot. All plots received the designated treatments throughout the three seasons of the

study. Figure 5-2 shows the layout of one of the plots within a block.

Each plot was centered on the hedgerow, and included four rows of maize on the

north side and four on the south side at 50, 100, 150, and 200 cm distances from the












2 m buffers


BLOCK

1



2



3



4



5



6



7



8



9



10



11


-- 3m --


5m

T-F+ T+ F-


T-F+ T+F- T+F+ T-F-


T-F-


T-F+


T+F+ I FT-F- T+F- T-F+



T+F- T-F+ __ T+F+ I T-F-I



T+F+I T-F+ I T-F- T+F-



T+F- T-F+ T+F+ T-F-



T+F- T-F+ T-F- T+F+



T-F+ T+F+ T+F- T-F-




ST+F- T-F- T+F+ T-F+


w


Hedgerow


IT+F+~


Layout of the Leucaena/maize hedgerow trial on Operation
Double Harvest property near Croix des Bouquets, Haiti


lI-.


T+F-


T-F-


I


Figure 5-1:


I i


a

















200 cm crop row



200 cm crop row

150 cm crop row

100 cm crop row

50 cm crop row

Leucaena Hedgerow

50 cm crop row

100 cm crop row

150 cm crop row

200 cm crop row

B

200 cm crop row


|4 3m --___ _|


Buffer crop row

Net plot A


uffer crop row


Diagram of one plot showing relative positions of a
Leucaena hedgerow and the adjacent rows of maize at four
distances.


Figure 5-2:







hedgerow, respectively. Buffer rows of maize were planted at 250 cm from the

hedgerows, marking the middle of the alley between hedgerows. Two-meter buffers

separated the plots within the blocks. Rows 1 and 11 were borders from which no data

were collected. Hedgerow experiments have been criticized for producing invalid results

because the no-hedgerow controls (sole crop plot) have been exploited by hedgerow roots

growing into adjacent plots, thereby over-estimating the benefits of hedgerow

intercropping (Hauser and Gichuru 1994). Some more recent studies have compensated

for this by digging trenches between hedgerow plots and control plots (Chamshama et al.

1998. Mugendi et al. 1999). Although the study reported here does not have sole crop

plots, and a two-meter buffer was left between plots, it is possible that Leucaena roots in

plots both with and without barriers exploited the soil in adjacent plots. This could not

be confirmed in this experiment.

Treatment installation

The hedgerow trees, grown in the ODH nursery, were planted out in April 1990

as one-year-old seedlings. The trees were cut to 50 cm stumps on 23 March 1991. In

April 1991, two randomly selected plots in each block were trenched to a depth of 30 cm

at a distance of 20 cm from each side of the hedgerows. The trenching was done with

picks, severing all roots. Plastic sheeting was put into the trenches to inhibit root re-

growth; the small canals created during root cutting were backfilled with soil. Trenching

extended for one meter beyond each side of the net plot boundaries, halfway into the two-

meter buffer between each plot. The decision to cut the roots to a depth of 30 cm was

based on a preliminary study of Leucaena roots done at a location off-station

(Appendix B). That study indicated 70% to 96% of the very fine roots (< I mm

diameter) were found in the top 30 cm of the soil, the lowest percent being found 150 cm








and 200 cm uphill from the hedgerow. A lesser percent of the fine roots (1-5 mm

diameter) were found in the top 30 cm (20 to 83%), with the least (20%) at the 200 cm

uphill position. Very few medium and large roots were found, but of those that were.

50% or more were deeper than 30 cm. Studies in semiarid India also found the majority

of Leucaena hedgerow root biomass in the top 30 cm of soil (Ong et al. 1991, Singh et al.

1996).

Due to my forced absence after the spring 1991 maize season the Leucaena

hedgerows in the main experiment were not top pruned between October 1991 and

September 1993. By that time, the hedgerow trees had grown to basal diameters of up to

10 cm. They were then cut back to a 50 cm stump height. Because the Spring 1994

maize harvest indicated the effect of trenching had almost disappeared, the plots were re-

trenched to a depth of 50 cm in August 1994 before the fall maize planting. Plastic

barriers were buried as before. The plastic barriers from the first root pruning were still

in place, but were missing or stiff and degraded in spots. In some places, soil had

covered the top of the barriers during cultivation, and Leucaena roots had grown over the

barriers into the alleys.

A selected local maize variety, known as chicken corn, was planted between the

hedgerows. The within-row spacing for maize hills was 75 cm, with 50 cm between

rows. Hill positions were laid out precisely using nylon cord with flagging tied at 75 cm

intervals. Maize was planted by hand three seeds per hill, and later thinned to two plants

per hill. The maize density of the interspaces at full stocking was then 53,333 plants per

hectare. If the space occupied by the hedgerows is included, maize plant density was

48,000 plants per hectare. Observations of 28,000 and 40,000 maize plants per hectare

were made on two farms in the south of Haiti at harvest time (after substantial mortality








during the season) in plots without hedgerows. Farmer interviews in one region of

northern Haiti indicated a higher density at planting, 62,500 plants per hectare (Swanson

et al. 1993a).

The fertilizer was to have been applied according to recommendations of the

Ministry of Agriculture, Haiti (MARNDR 1990): an initial broadcast application of 15-

15-15 (NPK) of 175 kg/ha at seeding followed by two applications of banded urea at 50

kg per hectare. In practice, however, due to the timing of rainfall and political instability

delaying travel to the site, only the spring 1995 maize crop received the complete

fertilizer treatment. The spring 1994 crop received only the initial application of 15-15-

15, and the fall 1994 crop received the 15-15-15 and the first urea application. Once a

given plot had been designated as fertilized, it remained a fertilized plot throughout the

three seasons reported here.

Operations by maize season

The spring 1994 maize crop was planted on 30 April. There had been 51 mm of

rain during the previous 7 days, 93 mm during the previous 14 days. Most of the rain

during the growing season fell between 11 and 21 DAS, with only one substantial rain

event after that, 23 mm at 67 DAS. A total of 206 mm of rain fell between sowing and

final harvest. The stand was thinned to two plants per hill at 28 DAS, and was weeded

once at 16 DAS. Insecticide was applied twice during the season. Fertilizer was applied

at a rate of 175 kg/ha at planting only; no urea was applied. Leucaeana top pruning was

done 35 days before sowing, then at 33 DAS (when trees were about 2.3 m high,

including the stump), and 65 DAS (when trees were about 1.8 m high). Figure 5-3 shows

the amount and timing of the rainfall in relation to the growth of Leucaena and maize,

and the timing of the agronomic interventions made to the plots during the spring 1994








season. The fall 1994 maize crop was planted on 10 September; two weeks after the

hedgerows were re-trenched to a depth of 50 cm (Figure 5-4). No measurable rain fell

during the 7 days preceding sowing, but 49 mm fell during the previous 14 days.

Tropical Storm Gordon passed over Haiti on 14 November (65 DAS) and dropped over

150 mm of rain in a 24-hour period on the trial site. This surpassed the capacity of the

rain gauge, so an accurate measurement was not recorded. This was the rainiest season

of the three, with at least (because of the inaccurate measurement during Tropical Storm

Gordon) 374 mm measured between sowing and final harvest. Most of the seeds had

germinated and had attained a height of 10 cm by 21 DAS, when they were weeded and

thinned to two plants per hill. Insecticide and urea (50 kg/ha) were applied at 26 DAS,

when the maize was about 25 cm tall. Because the effect of trenching was weak during

the previous season, the hedgerows were re-trenched to a 50 cm depth on 27 August

1994, 14 days before sowing the maize. Fewer measurements were made this season

than in others due to the U.S. military intervention in Haiti on 19 September, and the

events preceding and subsequent to it. Figure 5-4 shows the amount and timing of the

rainfall in relation to the growth of Leucaena and maize, and the timing of the agronomic

interventions made to the plots during the fall 1994 season.

The spring 1995 maize was planted initially on 26 March, but this planting failed

due to lack of rain. A new planting was done on 22 May. During the seven days before

sowing. 20 mm of rain had fallen, for a total of 39 mm during the 14 days before sowing.

After planting no rain was recorded until the 34th day, then 171 mm fell between 34 and

73 DAS. However, because the technician responsible for recording rainfall was on

vacation between the planting date and 36 DAS, two or three rainfall events were not

recorded. No additional rain fell before final harvest. I applied 15-15-15 at sowing and










Rain (mm) 49 0.8

(-4


67 83 49

rCIr? r"7


174+

r-S-


Total rain 0-114 DAS: 374+ mm Height (cm)
-250
200

-150

1 100
rrn


applied wedei urea 0
15-15-15 /inned insecticide

Avg. maize Avg. maize
height 10 cm height 25 cm


+ Leucaena topped, 28 & 126 DAS

* Maize biomass harvest, 78 DAS


Figure 5-3:


St Maize har


Most maize in Maize height
anthesis, silk up to 300 cm


+ Maize leaf w ater stress, 42 DAS

* Soilw ater content, 42 DAS


Rainfall. Leucaena and maize growth, and agronomic
inputs; spring 1994 maize cropping season


Rain (mm) 93 27 154

(-1- FJ--f


Leucaena height



DAS -20


20 1 40
I + 4


60


23 4 Total rain 0-91 DAS: 206 mm Height (cm)
-250
-200

-150

-100

50
80 100 120 140
,I I I o


ed
15-15-15

SAvg. maize
heiht 20 cm


weeded thinned,\ M' I Maize harvest, 91 DAS
insecticide insectici

Avg. maize _J Some maize \ Most maize in Maize leaf ma
height 100 cm in tassel anthesis, silk fired, soil has


+ Leucaena topped, 33 & 65 DAS

* Maize biomass harvest, 42, 49, 63, & 70 DAS


Figure 5-4:


rgins
1" cracks


+ Maize growth stage, 56 DAS

* Soil water content, 56 & 70 DAS


Rainfall, Leucaena and maize growth, and agronomic
inputs; fall 1994 maize cropping season


i


I ~ ~ [ -- d I





' "'






52

banded urea at 12 and 37 DAS. The maize was thinned to two plants per hill at 12 DAS.

The plots were weeded four times during the growing season; insecticide was applied

twice. Figure 5-5 shows the amount and timing of the rainfall in relation to the growth of

Leucaena and maize, and the timing of the agronomic interventions made to the plots

during the spring 1995 season.




Rain (mm) 83 1 40 41 91 Total rain 0-131 DAS: 171+ mm Height (cm)
rL- -250
200
Leucaena height -150

-- 100

DAS -20 0 20 40 60 80 100 120 140 50
+ + + +
applied thinned needed ulea weeded Maize harvest, 131 DAS
15-15-15 ure insecticide

Maize height 10-30 Maize height up to 95 Some maize
cm, leaf curl crm poor close to in tassel
hedgerow s in T-

+ Leucaena topped, 19, 57 & 130 DAS A Maize height, 40 DAS + Maize growth stage, 68 & 75 DAS
Maize biomass harvest, 47 & 54 DAS % Soil water content, 26 & 61 DAS


Figure 5-5: Rainfall. Leucaena and maize growth, and agronomic
inputs; spring 1995 maize cropping season


Measurements recorded

The following observations and measurements were made during the three maize

cropping seasons:

Maize height. This was done only for the spring 1995 season. Height to the top

of the highest fully open leaf of the tallest plant in each hill in blocks 2, 4, 6, and 8 was

measured at 40 days after sowing. Analysis was done on the mean height of eight plants

in each of the four hills to the north and four hills to the south of the hedgerow at a given






53
distance in each plot. If all plants in a hill had died. a height of zero was assigned for that

hill.

Maize growth stage. Growth stages of the plants were noted at 56 DAS in the

spring 1994 season, and at 68 and 75 DAS in the spring 1995 season. In the spring 1994

season, the scale was: 0=no maize plant present. 1= vegetative growth. 2=tassel.

3=anthesis, 4=silk. For the spring 1995 season, the scale was changed to include the

presence of ears: 0=no maize plant present, l=vegetative growth, 2=tassel/anthesis,

3=silk. 4=ear. Observations were made in blocks 2, 3. 5, 6, 8, and 9 on every maize plant

(2 plants per hill) in the spring 1994 season, and on only the most developed plant per

hill in the spring 1995 season.

Maize biomass during the growing season. For each of the four distances and in

each of the four plots in blocks 4, 7, and 10, the combined weight of one hill of maize on

the north side and one hill on the south side of the hedgerow was recorded. The plants

were cut at ground level, and then dried in a Watlow drying oven at 70 degrees C until a

constant weight was obtained.

Maize leaf water pressure. Predawn leaf water pressure was taken using a PMS

pressure chamber for one leaf at each distance in blocks 3, 6, and 9. A 2 cm by 10 cm

section was cut from the highest fully expanded leaf, one end was placed in a zip-lock

bag, and the sample was inserted into the pressure chamber. Compressed nitrogen gas is

normally used in the chamber, but as this was not available in Haiti compressed air was

used instead. This measurement was to have been done periodically throughout each

maize season, but was done only once because of the security problems discussed in the

footnote on page 44.








Maize yield. Grain yield. number of plants, and number of ears were measured

from blocks 2. 3, 5, 6, 8, and 9 at 91 DAS for the spring 1994 harvest. 114 DAS for the

fall 1995 harvest, and 131 DAS for the spring 1995 harvest. At each of the four distances

(50, 100. 150, and 200 cm), plants were harvested from a 3-meter length on the north and

south sides of the hedgerows, for a total of 6 meters (8 hills, or 16 plants maximum)

harvested in each plot at each distance. Because the grain moisture meter ceased to

function correctly, field dry grain weights are shown in the tables and figures. Harvests

for all treatments were made at the same time. and all samples were weighed at the same

time. Final harvest measurements are converted to per hectare equivalents by dividing

the sample weight or count for a given distance from the hedgerows by six (six meters

per plot were sampled) and multiplying by 4,500 (each distance represented 4,500 meters

per hectare given a 50 cm row spacing). The per-hectare contributions of the rows at

each distance were then summed to give a total per hectare figure, which included the

2,000 m occupied by the hedgerows.

Soil water percent. Soil cores were taken using a JMC Back-saver corer. Each

core was divided into three sections. 0-15 cm, 15-30 cm. and 30-45 cm. Four cores for

each depth and distance were combined for each measurement. The samples were

weighed wet in the field and dried in an oven at 70 degrees C until a constant weight was

obtained. Soil water content was determined gravimetrically by dividing the weight of

water lost in drying by the dry weight of the sample. The range of soil water content that

could be sampled at this site was narrow due to the heavy texture of the vertisol. When

the soil was very dry, the 0-15 cm depth often was not retained in the sampling tube, and

when the soil was wet, the end of the tube plugged and did not allow soil to enter. For

the calculation of field capacity, a large-diameter hammer-driven corer was used. This is








a slow procedure and the large-diameter corer could not be used under normal

circumstances due to the necessity of completing 192 cores in a single day each time

sampling was done. Soil water content was sampled twice during the spring 1994

growing season, at 56 and 70 DAS. The first, at 56 DAS. was done on blocks 3 and 6

only because the corer shaft broke. All subsequent samplings were done on blocks 3, 6.

and 9. A single sample was taken in fall 1994 at 42 DAS, and two were taken in spring

1995 at 26 and 61 DAS. An additional attempt was made to measure soil water content

at 54 DAS, but this was not successful because the soil was too wet and the corer

jammed.

Leucaena small stem and leaf biomass and height. Leucaena tops were cut back

to a 50 cm stump height using a machete. Leaves were not separated from stems; they

were weighed together in the field with a spring scale. Sub-samples were taken and dried

in a Watlow oven at 70 degrees C until a constant weight was obtained. Height above the

stump of the tallest stem in each plot was measured with a fiberglass tape. All leaves and

stems were removed from the site after harvest.

Leucaena root intersections in trenched and nontrenched hedgerows. After the

spring 1995 maize harvest was completed, numbers of Leucaena roots were counted in

trenched and nontrenched hedgerows to discover what influence plastic barriers had on

root distribution. Transects perpendicular to the hedgerows were excavated through rows

4, 5, and 6 (Figure 5-1). where there happened to be three trenched plots adjacent to three

nontrenched plots. Work began on 13 January 1996, 105 days after the maize harvest.

The plots were weeded three times during that period to keep out roots of grasses and

weeds. Thirty sections 1.2 m wide by 1 m deep parallel to the trenched hedgerows and

an equal number parallel to the nontrenched hedgerows were mapped, at 20, 50, 100,








150, and 200 cm distances. A Plexiglas plate inscribed with a 10 by 10 cm grid was

pinned to each cut surface. The positions of the roots were marked on plastic sheets

clipped over the Plexiglas plate, a separate symbol was used for each of four diameter

classes: very fine (10 mm).

Statistical analysis

Statistical analysis was done using the SPSS version 9.0 univariate GLM

ANOVA procedure. The model used for the dependent variables Leucaena leaf and

small stem biomass and height above the stump was as follows:

y,jk, = p + block, + t + f + t*fj + error,,,
where: y=dependent variable, t=trenching, f=fertilizer

i=2, 3, 4, 5, 6, 7, 8, 9, 10

j=0, 1 where 0=not trenched and =trenched

k=0, 1 where 0=no fertilizer applied andl=fertilizer applied

For each kind of field measurement taken at the four fixed distances and/or the

three soil depths, three levels of analysis were done-an overall ANOVA (referred to as

the complete model throughout the text) that included all the main effects and their

interactions, ANOVAs done separately for trenched and nontrenched plots that included

all other effects, and ANOVAS done separately for each distance from the hedgerows

that included only trenching and fertilization factors and their interaction. For percent

soil water, separate ANOVAs were done for each combination of distance from the

hedgerows with each of the three depths (0-15 cm, 15-30 cm. 30-45 cm). If there were

significant trenching by fertilization interactions produced by the ANOVAs done

separately for each distance and depth, contrast statements were run to determine at

which levels the interactions were significantly different.








The following example shows the complete ANOVA model used for the

dependent variables: maize biomass taken during the growing season, maize height.

maize leaf water pressure, maize grain at final harvest, number of maize plants at final

harvest, and number of maize ears at final harvest.

yji.k, = p + block, + t ,+ f, + t*f j + d, + t*d + f*d ,j + t*f*d 4, + error ,l,,,

where: y=dependent variable, t-trenching, f=fertilizer. d=distance

i=2, 3, 5, 6, 8. 9 (blocks used varied for each dependent variable)

j=0, 1 where 0=not trenched and 1-trenched

k=-, 1 where 0=no fertilizer applied andl=fertilizer applied

1=50, 100, 150, 200 cm distance from hedgerows

The overall mean square error was used to test the effect of distance and its

interactions. However, because measurements were taken at fixed distances and depths,

giving a split-plot characteristic to the experimental design. an "error a" term was

computed by adding sums of squares for the interactions of blocks with trenching and

fertilizer. This was used to test the effects of trenching and fertilizer and their

interaction.

The complete model for the percent soil water ANOVAs was:

ykl,,,, = p + block, + t + f + t *fjk + d, + t *d + f*dj/ + t *f*djk
+ dp,,, + t dp ,,, + f dpk,,, + d dp,,, + t f dpk,,, + error,k,,,,,

where: y=dependent variable, t=trenching, f=fertilizer, d=distance

i=3. 6. 9

j=0. 1 where O=not trenched and l=trenched

k=-, 1 where 0=no fertilizer applied andl=fertilizer applied

1=50. 100, 150, 200 cm distance from hedgerows

m=0-15, 15-30, 30-45 cm depth from soil surface






58

In addition to the "error a" term computed and applied as above, for the soil water

ANOVAs an "error b" term was computed to test the effects of distance and its

interactions with trenching and fertilizer. Sums of squares for the interactions of blocks

with distance and its interactions with trenching and fertilizer were added to compute this

term. The overall error term was used to test the effects of depth and its interactions.

"Error a" and "error b" and the corresponding F values were computed by hand;

SAS version 7.0 was used to compute the appropriate p-values. Soil water was expressed

as a percent, however the range of values was narrow (13% to 39%) and not extremely

high or low, so an arcsin transformation to stabilize the variances was not applied before

running the ANOVAs (Ott 1992).

Not all of the interactions that resulted in significant tests were meaningful due to

the very low numerical values of the differences. P-values for all factors and their

interactions are shown in Appendix C for all of the ANOVAs and contrasts.

Paired t-tests were used to detect significant differences in the numbers of root

intersections between trenched and nontrenched plots. This was possible because each

pair of measurements at each distance and depth were taken from physically adjacent

positions.

Results and Discussion

Results for each maize season are presented separately in this section, followed

by the results of the root distribution study and a general discussion.

Spring 1994 season

Two factors made this season differ from the two subsequent ones-the 30 cm

root barriers installed in the trenched plots in 1991 apparently no longer presented an

effective obstacle to the passage of Leucaena roots from the hedgerow into the alleys,

and the rainfall was poorly distributed.








Rainfall was 206 mm for the growing season and there was good soil water at

planting (51 mm fell in the 7 days prior to planting). However, most of the rain during

the growing season fell between 11 and 21 DAS (154 mm), and then no rain fell until 67

DAS (23 mm), which was the last effective rainfall event before final harvest. The plants

had reached about 20 cm height by 14 DAS. At 35 DAS the height was about 1 m, but

the soil was dry and cracking. Maize was always shorter than Leucaena until the 33

DAS hedgerow pruning; at that time the maize was about one meter high and the

hedgerows 50 cm. Maize tasseling began about 42 DAS, with most of the plants in

anthesis or silk by 50 DAS. The plants showed increasing signs of stress as the season

progressed. Many of the plants had fired leaf margins, and soil cracks opened to about

2.5 cm after 60 DAS.

Maize biomass production. Aboveground biomass production during the growing

season was not affected significantly by trenching or by fertilizer application, but the

amount of biomass did differ with distance from the hedgerows. The four

measurements of maize biomass are shown together in Figure 5-6.



a 300
N b
c
E 250 ---- c
4 b DC
8 200 50 cm
I, c C
E b E 100 cm
.2 o 150 b
"0 4 [ 150 cm
c 100 a a a 200 cm

0, w
5, 0 -- a

42 49 63 70
Days after sowing


Figure 5-6: Maize biomass yields at 50, 100, 150, and 200 cm from the hedgerows at
four times after sowing during the spring 1994 season






60

For the ANOVA model that included distance, there were no significant effects of

fertilization (o=.05 or =. 10) on the weight of maize biomass, and only one case where

trenching was significant (p = .085) at 63 DAS. P-values for the effect of distance on

maize biomass were <.001 at all four observations. The weight of maize biomass

produced was always significantly lower at 50 cm from the hedgerows. It appeared to

peak at 150 cm for the first three observation periods, although that distance was

significantly greater than the others only at 49 DAS (Figure 5-6). At 70 DAS, the weight

of biomass peaked at 200 cm. Competition between the maize and the trees clearly

extended to 100 cm at 42 DAS, and to 50 cm thereafter.

When the analyses were done separately at each distance, the effect of trenching

was significant only at 50 cm from the hedgerows at 42 DAS and at 63 DAS (Table 5-2).

In these two cases, the amount of maize biomass produced in trenched plots was at least

double that produced in nontrenched plots.


Table 5-2: Effect of trenching on the production of maize biomass 50
cm from the hedgerows at 42 and 63 DAS; spring 1994.

42 DAS 63 DAS
Trenched biomass (g) 50 93
Nontrenched biomass (g) 25 37
p-value .039 .039


There were also four instances when the effect of fertilizer was significant. More

maize biomass was produced in the nonfertilized (F-) plots at 49 DAS at both the 50 cm

and 100 cm distance from the hedgerows (Table 5-3). At both 42 and 49 DAS there was a

significant interaction of fertilizer and trenching at the 150 cm distance from the

hedgerows, in both cases more maize biomass was produced in trenched (T+) plots








(Table 5-3). However, at 42 DAS this effect was seen only between trenched and

nontrenched plots that were also fertilized, while at 49 DAS the effect was seen only

between plots that were not fertilized.


Table 5-3: Effect of fertilizer on the production of maize biomass 50. 100, and 150 cm
from the hedgerows at 42 and 49 DAS; spring 1994.

49 DAS 42 DAS 49DAS
50 cm 100 cm 150 cm F+ 150 cm F-
Maize (g) F+ 50 168 Maize (g) T+ 196 238
Maize (g) F- 62 209 Maize (g) T- 163 174
p-value .063 .002 p-value .055 .010


Maize growth stage. Numerical values corresponding to maize growth stages

were assigned at 56 DAS using the scale 0=absent, 1= vegetative growth, 2-tassel,

3=anthesis, 4=silk. On average, the plants at 50 cm were in the vegetative stage, all

others were in tassel or anthesis (Figure 5-7).



anthesis
3


tassel
2 ___





0



50 100 150 200
Distance from the hedgerow (cm)

Figure 5-7: Growth stage of maize at four distances from hedgerows, 56 DAS; spring
1994.








Soil water content. This was measured twice during this season, at 56 and 70

DAS. The ANOVAs that included both distance and depth in the model did not reveal

any differences in soil water due to trenching, fertilization, or distance from the

hedgerows. There were differences among depths below the surface, as shown in Table

5-4. At 70 DAS the surface 15 cm had more soil water than the deeper layers, probably

because 23 mm of rain fell just three days before the measurement. This was the first

rain since 21 DAS. At 56 DAS the water content of the soil increased steadily with

depth. The water content of the 30-45 cm layer was the same for both observations.


Table 5-4: Soil water at three depths below the surface at 56 and 70
DAS; spring 1994.

Soil Water %-Gravimetric
56 DAS 70 DAS
0-15 cm 15.0 a 24.0 c
15-30 cm 21.6 b 22.2 a
30-45 cm 23.5 c 23.4 b
p-value <.001 <.001


The complete model also indicated significant soil water interactions between

trenching and fertilizer (p = .081) at 56 DAS. and between trenching and distance (p =

.020) and fertilizer and distance (p = .007) at 70 DAS. When ANOVAs were done

separately for each depth and distance, with trenching and fertilizer as the only

independent variables, there were no significant differences in soil water percent between

the two trenching treatments at any distance or depth. In only two cases were the

differences between the two fertilizer treatments significant: at 56 DAS 100 cm distance

in the 30-45 cm depth the fertilized plots had 23.1% soil water vs. 24.2% for the

unfertilized plots (p = .004), and at 70 DAS 200 cm distance in the 15-30 cm depth






63

fertilized plots had 21.1% soil water vs. 23.1% for the unfertilized plots (p = .027). At 56

DAS there was a significant interaction at the 100 cm distance from the hedgerows: at the

0-15 cm depth trenched plots had less soil water than nontrenched plots where no

fertilizer was applied (p = .067); the reverse was true in the fertilized plots (p = .034).

The meaning of these interactions is not clear, since fertilizer was not significant as a

main effect. P-values for all interactions are listed in Appendix C, but not all will be

discussed as they are too numerous and add little to the understanding of the treatment

effects.

Maize yield. Grain yield for this season was very low, with no significant

differences due to trenching or fertilizer. A per-hectare equivalent corresponding to the

average values per six-meters as shown in Figure 5-8 is 182 kg/ha maize grain, with only

5.2 kg/ha of the total contributed by the 50 cm rows. As with maize biomass and sexual

development, grain yield was affected only by distance from the hedgerows. In addition,

the peak production in grain yield and number of ears appeared at 150 cm from the

hedgerows, then declined slightly at 200 cm. This is also similar to the pattern observed

in biomass production and sexual development, with a decline between 150 cm and 200

cm, although it was only statistically significant at 49 and 56 DAS.

Apparently, competition between maize and hedgerows affected the survival of

maize plants differently than grain yield or number of ears. There were significantly

fewer surviving plants at harvest time in the row closest to the Leucaena, but in the other

three more distant rows maize plant survival was the same, and did not decrease between

150 and 200 cm (Figure 5-9). Full stocking for each distance depicted in Figure 5-9

would have been 16 plants (per 6 m, or per 8 hills), equal to 12,000 plants per hectare at

each distance, totaling 48,000 plants per hectare if all had survived. Maize plant survival






64

with distance from the hedgerow was 48%, 84%, 87%, and 80% at 50 cm. 100 cm. 150

cm, and 200 cm from the hedgerows, respectively.

Figure 5-8: Yield of maize grain per six meters at four distances from the hedgerows;


b





a

50cm 100lcm


S150cm


Distance from hedgerow


spring 1994.


50cm


100 cm 150 cm
Distance from the hedgerow


200 cm


Figure 5-9: Number of maize plants per eight hills at four distances from the hedgerows;
spring 1994.


200 cm






65

Since the number of ears did not differ significantly between 100 and 150 cm

(Figure 5-10) and the grain yield did differ, the implication is that grain yield per ear was

lower at 100 cm than at 150 cm. This was the case, with the grain weight (g) per ear for

the four distances being, respectively, 3.5, 10.8, 13.0. and 9.7.





bc
o5 -- -
6-


E
3 .
z a
2
01-1

50 100 150 200
Distance from the hedgerow

Figure 5-10: Number of maize ears per eight hills at four distances from the hedgerows;
spring 1994.


Table 5-5 shows the data presented in Figures 5-8, 5-9, and 5-10 on a per hectare

basis.


Table 5-5: Maize grain weight, number of plants, and number of ears at
four distances from the hedgerows and as a total per hectare,
with p-values for the differences between distances.

Distance (cm) Grain (kg) No. of plants No. of ears
50 5.2 a 5,813 a 1,500 a
100 56.9 b 10,031 b 5,281 bc
150 75.3 c 10,406 b 5.813 c
200 44.6 b 9,563 b 4,594 b
Total/ha 182.0 35,813 17,188
p-value <.001 <.001 <.001






66

Leucaena biomass-production and height. The four Leucaena harvests discussed

in conjunction with the spring 1994 maize season are those made after my post-coup

d'etat-retur to Haiti, but before the hedgerows were re-trenched prior to the fall 1994

maize season. The 20 November 1993 harvest was the first re-growth after the

overgrown trees were cut back to 50 cm stumps. but this was before the first application

of fertilizer to a maize crop so the F+ and F- designations are not meaningful for this

harvest. Table 5-6 shows the weight of biomass harvested and Table 5-7 shows the

height above the stump for the four treatment combinations. The 33 and 65 DAS

harvests were taken during the spring 1994 maize season.


Table 5-6: Stem and leaf biomass harvests and daily growth increments of Leucaena
hedgerows 33 and 65 DAS; Spring 1994.

20 Nov 93 2 Jun 94 4 Jul 94 20 Aug 94
Biomass Increment Biomass Increment Biomass Increment Biomass Increment
(kg/m) (g/m/d) (kg/m) (g/m/d) (kg/m) (g/m/d) (kg/m) (g/m/d)
T+F+ 1.1 17.8 0.9 12.4 0.4 11.5 0.5 10.0
T+F- 1.2 18.3 0.9 12.6 0.3 10.4 0.4 9.1
T-F+ 1.4 21.7 1.0 14.1 0.4 12.2 0.5 10.2
T-F- 1.2 18.9 0.9 12.5 0.4 11.3 0.5 10.9
p-value T .074 .139 .300 .283
p-value F .358 .224 .187 .941
p-value TxF .198 .091 .994 .400
DSPHa 63 69 32 47
RSPH (mm)b 162 312 0 53
DAS 42c 33 65 between crops

a: Days since previous hedgerow harvest
b: Rainfall since previous hedgerow harvest
c: DAS (days after sowing) is reported because a fall 1993 maize crop was growing in the
plots at this time. This maize season is not reported here, as it was lost due to animal
damage.


It appears that the effect of the initial trenching on Leucaena biomass production

did not last beyond 20 November 1993, when there was a small reduction in biomass








production in the trenched plots (p = .074). Trenched plots produced 1.15 kg/m. while

nontrenched plots produced 1.3 kg/m. A five-meter spacing means there are 2.000

meters of hedgerow per hectare, so this amounts to a reduction of 300 kg/ha due to

trenching. There was no significant trenching effect in the subsequent three harvests, nor

did fertilizer application to the maize influence Leucaena production. The reduction in

Leucaena daily growth rate beginning in July and continuing in August was probably a

response to less rainfall during that period.


Table 5-7: Height above a 50 cm stump and daily growth increments of Leucaena
hedgerows prior to the second root trenching. 33 and 65 DAS; Spring 1994.

20 Nov 93 2 Jun 94 4 Jul 94 20 Aug 94
Height Increment Height Increment Height Increment Height Increment
(m) (cm/d) (m) (cm/d) (m) (cm/d) (m) (cm/d)
T+F+ 2.1 3.3 1.8 2.6 1.2 3.8 1.4 3.0
T+F- 2.1 3.3 1.8 2.6 1.3 4.1 1.4 3.0
T-F+ 2.0 3.2 1.8 2.6 1.2 3.8 1.4 3.0
T-F- 2.2 3.5 1.7 2.5 1.2 3.8 1.5 3.2
p-value T .725 .462 .095 .403
p-value F .076 .840 .393 .640
p-value TxF .129 .329 .160 .640
DSPHa 63 69 32 47
RSPH (mm)b 162 312 0 53
DAS 42c 33 65 between crops

a: Days since previous hedgerow harvest
b: Rainfall since previous hedgerow harvest
c: DAS (days after sowing) is reported because a fall 1993 maize crop was growing in the
plots at this time. This maize season is not reported here, as it was lost due to animal
damage.


Leucaena height growth is a less precise measure than the biomass yield, and was

taken to the nearest 10 cm. Neither trenching nor fertilizer application appeared to affect

height growth during the four harvests shown in Table 5-7. The p-value of .076 for the

effect of fertilizer on 20 November 1993 is difficult to explain, because this was before








the first fertilizer application to the maize crop. The significant (p = .095) effect of

trenching on 4 July is associated with a very small numerical height advantage in the

trenched, nonfertilized plots. It is interesting to note that while daily Leucaena biomass

increment decreased after 2 June (Table 5-7), daily height growth rate increased. The

July and August cutting intervals were shorter than the previous two, so this may mean

that Leucaena puts on stem height first, then adds stem diameter, leaf biomass, or new

stems later in the cycle.

Fall 1994 season

The fall 1994 season was the best of the three in terms of maize production,

because it had the highest rainfall and the best rainfall distribution through the growing

season. After the Leucaena was pruned at 28 DAS, the maize quickly gained a height

advantage over the hedgerow trees and maintained it throughout the rest of the season in

both the trenched and nontrenched plots, including the rows closest to the trees. Most of

the plants were in flower (tassel and silk) by the time tropical storm Gordon passed at 65

DAS. By 70 DAS the maize was about 3 m tall, taller by about 1 m than the Leucaena

hedgerows.

Maize biomass production. Maize biomass production was measured only once

during the season, at 78 DAS. The ANOVA for the complete model (including distance)

was significant for trenching (p = .018) and distance from the hedgerow (p = .023).

ANOVAs separate by distance showed that only at the 50 cm distance from the

hedgerows was trenching a statistically significant factor (p = .037). At all other

distances trenched plots produced more maize biomass than nontrenched plots, but the

difference was not significant (Figure 5-11). The complete model also indicates an

interaction between trenching and fertilizer at the 150 cm distance (p = .093). There was






69

significantly more biomass produced in trenched (509 g) than nontrenched plots (307 g)

where they were not fertilized, but the difference was not significant in fertilized plots.



600 -
/ ~ns
500 -ns-
N a ns
M 400 -
0 OT+
m 300 T
E b IT-
'j 200 -
N
'c 100 -

0
50 100 150 200
Distance from the hedgerow

Figure 5-11: Trenched and nontrenched plots at 50, 100, 150, and 200 cm from the
hedgerows, 78 DAS; fall 1994.


ANOVAs run separately for trenched and nontrenched hedgerows showed that

distance was a significant factor only for nontrenched hedgerows, with the 50 cm

distance producing less biomass than the other distances. This was not the case for

trenched hedgerows, where no significant difference in biomass production was detected

over the four distances (Table 5-8).


Table 5-8: Maize biomass in trenched and nontrenched plots over four
distances from the hedgerows, 78 DAS; fall 1994.

Trenched Nontrenched
Distance (cm) (g/2 hills) (g/2 hills)
50 412.6 a 249.2 a
100 501.9 a 389.1 b
150 475.5 a 416.4 b
200 400.6 a 356.2 b
p-value .485 0.001








Maize leaf water potential. The complete ANOVA model did not show any

significant leaf water potential differences among factor levels for any factor. However.

when the ANOVAs were run separately for each distance, maize plants in trenched plots

indicated less leaf water pressure than did those in nontrenched plots (p = .081) at the 50

cm distance from the hedgerows (Figure 5-12). Differences were not significant at the

other distances.





S3.0
r ns
2.5 -
b ns ns
2.0
P i E T+
S1.5 E T-
a
1.0

N 0.5

0.0
50 100 150 200
Distance from the hedgerow (cm)



Figure 5-12: Maize leaf water potential in trenched and nontrenched plots at four
distances from the hedgerow. 42 DAS; fall 1994.


Similarly, when ANOVAs were run separately for trenched and nontrenched

plots, there was no significant difference in maize leaf water potential among the four

distances for the nontrenched plot. For the trenched plots, however, the leaf water

potential in the 50 cm plots was significantly lower (p = .008) than in the other three

distances (Table 5-9).








Table 5-9: Maize leaf water potential in trenched and nontrenched plots
over four distances from the hedgerows, 42 DAS; fall 1994.

Distance (cm) Trenched (bars) Nontrenched (bars)
50 1.1 a 2.0 a
100 2.5 b 1.7 a
150 1.8 b 2.0 a
200 2.1 b 2.1 a
p-value .008 .917


Soil water content. Soil water was measured on the same day as maize leaf water

potential. Figure 5-13 shows the soil water percent for trenched plots at all

combinations of distance and depth. The differences in soil water percent between

trenched and nontrenched plots are shown in parentheses. The overall model indicated

that trenching (p = .042), distance from the hedgerow (p < .001). and depth below the

surface (p < .001) were significantly affecting soil water content. The effect of fertilizer

was not significant (p = .352). Separate ANOVAs for each combination of distance and

depth were not significant for either trenching or fertilizer in the 0-15 cm depth. There

were significant differences due to trenching in the lower depths. In the 15-30 cm layer,

soil water percent was higher in trenched plots at the 50 cm (p = .033) and 150 cm (p =

.095) distances from the hedgerow. In the 30-45 cm layer, soil water percent was higher

in trenched plots at the 50 cm (p = .002) and 100 cm (p = .014) distance from the

hedgerow.

The effect of distance from the hedgerows was significant in trenched plots at the

15-30 cm (p = .004) and the 30-45 cm depths (p < .001), with more soil water present

close to the hedgerows. In the surface layer distance was significant for trenched plots at

the a=. 10 level, with more soil water at the 50 cm distance than the 150 distance (p =

.078). In Figure 5-13, cells having the same letter, in a given row, are not significantly








different. Soil water percent in nontrenched plots did not differ significantly with

distance from the hedgerow in any of the three depths.





Distance from Hedgerow (cm))

50 100 150 200
A I I I I
Soil Depth (cm) Soil Water (%, Gravimetric)
0-15 27.8 (1.4) 24.6 (1.5) 23.6 (1.4) 24.5 (3.6)

r a ab bc c
15-30 129.9 (3.5)i 27.2 (1.6) 25.1 (1.2) 23.5 (-3.2)
r a b c c
30-45 132.5 (4.9) 29.7 (3.4)1 25.6 (0.1) 26.7 (-0.6)




Figure 5-13: Soil water in trenched plots (and difference between T+ and T- plots) at
42 DAS, fall 1994. Cells with solid border (a=.05) or dashed border
(a=. 10) indicate significantly greater soil water in trenched plots than in
the corresponding nontrenched plots at that position. Letters indicate
differences between distances within a depth for trenched plots (a=.05).


Soil water percent increased with depth differently in trenched plots than in

nontrenched plots. In the trenched plots (Table 5-10), soil water increased with depth at

the 50 cm distance (p = .066) and the 100 cm distance (p = .020), with the deepest layer

having significantly more water than the surface layer. However, there were no

significant differences in soil water with depth at the 150 and 200 cm distances. In

contrast, the nontrenched plots had significantly more soil water with depth at the 100

cm. 150 cm, and 200 cm distances, but did not show significantly more soil water with

depth at the 50 cm distance.






73
Table 5-10: Soil water in trenched and nontrenched plots by depth and distance from the
hedgerow, 42 DAS; Fall 1994.

Distance (cm) Depth (cm) Trenched (%) Nontrenched (%)
0-15 27.8 a 26.4
50 15-30 29.9 ab 26.3
30-45 32.5 b 27.6
p-value .066 .820
0-15 24.6 a 23.1 a
100 15-30 27.2 ab 25.6 ab
30-45 29.7 b 26.3 b
p-value .020 .075
0-15 23.6 22.2 a
150 15-30 25.0 23.9 ab
30-45 23.6 25.5 b
p-value .368 .012
0-15 24.4 20.8 a
200 15-30 23.5 26.7 b
30-45 26.7 27.3 b
p-value .502 .048


Maize yield. The complete model for maize grain was significant for the effect of

trenching (p = .009) and distance from the hedgerows (p = .076), but not for the effect of

fertilizer (p = .385). The complete models for number of plants and for number of ears

were not significant for any of those factors. ANOVAs done separately at each distance

showed that maize grain from trenched plots was significantly greater (ao=. 10) than that

from nontrenched plots at the 50 cm (p = .071) and 100 cm (p = .068) distances from the

hedgerows (Figure 5-14). There was also a statistically significant difference between

the numbers of plants in trenched plots vs. nontrenched plots at the 100 cm distance (p =






74

.020), with the nontrenched plots having about one more plant per six-meter sample at

that distance.



1000
900 a
E a ns ns
C5 800 b
700
S-I b
S600 -
"l OT+
500
C ET-
400 --
a 300
S200 --
100
0
50 100 150 200
Distance from the hedgerow (cm)


Figure 5-14: Maize grain yield in trenched (T+) and nontrenched (T-) plots at four
distances from the hedgerow, Fall 1994.


Final harvest is presented on a per-hectare basis in Table 5-11. Although there

were differences in grain yield between trenched and nontrenched plots at 50 and 100 cm

from the hedgerows (cr=.10, Figure 5-14 above), there was no significant difference in

grain, number of plants or number of ears over distance from the hedgerows, for either

trenched or nontrenched plots.

In general, there were no strong differences in yield parameters this season,

probably due to the adequate, well-distributed rainfall partially negating the potential

advantages of the root barriers. If only the differences in grain weight at 50 cm and 100

cm are accepted as significant, then trenching made a difference of 252 kg maize








grain/ha, an increase of 12% over the nontrenched plots. However, since the effect of

trenching for the complete model was significant (p = .009), then the advantage of

trenching increased to 360 kg/ha (2,382 kg-2,022 kg), an increase of 18% over the

nontrenched plots. Since there were no differences in the numbers of plants or ears, the

advantage of trenching was manifested in more grain weight per ear in trenched plots.

especially at the 50 and 100 cm distances.


Table 5-11: Maize grain weight, number of plants, and number of ears at four
distances from the hedgerows and as a total per hectare, with p-values for
the differences between distances, Fall 1994.

Grain (kg) No. plants No. ears
Distance (cm) T+ T- T+ T- T+ T-
50 563.1 423.1 10,563 10,625 11,250 10,313
100 648.1 535.6 10,188 11,125 10.750 11,313
150 601.3 561.3 11,188 10,813 11,313 11,688
200 569.4 501.9 10,938 10,875 10,688 10,750
Total/ha 2,382 2,022 42,875 43,438 44,000 44,063
P-value .488 .103 .152 .637 .869 .314


Leucaena biomas-production and height. Four Leucaena harvests are presented

in Tables 5-12 and 5-13-the first was done during the fall 1994 maize season at 28

DAS, the others were done after the fall maize harvest but before the spring 1995 maize

was sown. The reaction of Leucaena to the re-trenching to 50 cm depth done in August

1994 is clear. In every case, the trenched plots produced less biomass than the

nontrenched plots. The loss of small stem and leaf biomass due to the effect of trenching

was 532, 452. 422. and 154 kg per hectare for the four pruning dates in Table 5-12.

respectively, for a total loss over this seven-month period of 1,561 kg per hectare. The






76

relatively slow growth rate between the 8 October and 14 January harvests in spite of the

plentiful rainfall was probably due to insect damage (species not identified) in the

southeast quarter of the trial site, in blocks 6-10. The damage affected trenched and

nontrenched plots equally. Fertilizer application on the adjacent maize did not affect

Leucaena biomass or height growth. The significant trenching by fertilization interaction

at the 6 May harvest does not appear to be important in magnitude.


Table 5-12: Stem and leaf biomass harvests and daily growth increments of Leucaena
hedgerows after the second root trenching, Fall 1994.

8 Oct 94 14 Jan 95 16 Mar 95 6 May 95
Biomass Increment Biomass Increment Biomass Increment Biomass Increment
(kg/m) (g/m/d) (kg/m) (g/m/d) (kg/m) (g/m/d) (kg/m) (g/m/d)
T+F+ 0.3 6.8 0.4 3.8 0.5 8.5 0.4 7.1
T+F- 0.3 6.5 0.4 3.6 0.5 8.6 0.3 5.9
T-F+ 0.6 12.0 0.5 5.3 0.7 11.3 0.4 7.3
T-F- 0.6 12.2 0.7 6.8 0.8 13.0 0.4 8.7
p-value T <.001 .005 <.001 .013
p-value F .985 .401 .301 .919
p-value TxF .745 .255 .358 .033
DSPHa 49 98 61 51
RSPHb (mm) 174 320+ 56 76
DAS 28 between crops between crops between crops

a: Days since previous hedgerow harvest
b: Rainfall since previous hedgerow harvest


Trenching significantly reduced Leucaena height growth only for the first two

harvests following re-trenching (Table 5-13). Height difference between trenched and

nontrenched plots was not significant during the third harvest, even though the difference

in small stem and leaf biomass was significant.








Table 5-13: Height above a 50 cm stump and daily growth increments of Leucaena
hedgerows after the second root trenching, Fall 1994.

8 Oct. 94 14 Jan. 95 16 Mar. 95 6 May 95
Height Increment Height Increment Height Increment Height Increment
(m) (cm/d) (m) (cm/d) (m) (cm/d) (m) (cm/d)
T+F+ 1.3 2.7 1.9 1.9 1.5 2.5 1.4 2.7
T+F- 1.4 2.9 1.8 1.8 1.5 2.5 1.3 2.5
T-F+ 1.5 3.1 2.2 2.2 1.6 2.6 1.5 2.9
T-F- 1.5 3.1 2.2 2.2 1.6 2.6 1.5 2.9
p-value T .005 .038 .113 .311
p-value F 1.000 .945 .633 .311
p-value TxF .471 .331 .447 .030
DSPH 49 98 61 51
RSPH (mm) 174 320+ 56 76
DAS 28 between crops between crops between crops
a: Days since previous hedgerow harvest
b: Rainfall since previous hedgerow harvest


Spring 1995 season

This season had less rainfall during the growing season than fall 1994, but the

distribution was better than that during the spring 1994 season. Germination initially

appeared to be good, with the tallest plants 20 cm tall at 19 DAS. However, at 26 DAS it

was evident that germination at the 50 cm distance was poor. Maize height was up to 30

cm at that time, but some plants were showing leaf curl. Flowering began at 47 DAS; by

54 DAS most of the plants were flowering. At this time there appeared to have been an

underground leak in the irrigation system (used in the nearby tree nursery) that might

have increased soil water content in the east side of blocks 9 and 10. The only

measurement that might have been influenced was the maize biomass harvest done at 54

DAS. Since maize was cut in blocks 4, 7, and 10 and the eastern-most plot in block 10








was a nontrenched plot. an inadvertent increase in soil water content in that plot only

strengthens the result of the ANOVA that showed a significant difference in maize

biomass due to trenching.

The effect of trenching on maize development and yield was more pronounced

than during the other two seasons. However, even though this was the only season for

which the complete amount of fertilizer recommended by the Ministry of Agriculture

(MARNDR 1990)was applied, there were no significant effects of fertilizer on any of the

parameters measured for the complete ANOVA models.

All the ANOVAs run using the complete model for maize growth and yield

during this season produced very significant p-values for the effects of trenching, with

trenched plots producing more than nontrenched plots, especially close to the hedgerows.

All but the two maize biomass harvests were significant for distance from the hedgerow

and the trenching by distance interaction (Appendix C). Much of the difference was due

to the poor survival of maize plants in the nontrenched plots, especially in the two rows

closest to the trees.

Maize height. Figure 5-15 shows a significant maize height advantage at 40 DAS

in trenched plots at all four distances from the hedgerows (p < .001). Missing plants

were assigned values of zero because the mortality was attributed to competition with the

hedgerows. There were more missing plants in nontrenched plots than in trenched plots:

at the 50 cm distance 84% of the hills in nontrenched plots had no surviving plants vs.

20% for trenched plots; at the 100 cm distance the corresponding numbers were 25% in

the nontrenched plots and 4% in the trenched plots.









45 a
40 -a .. a
40 5 ------- ---------a-- -----
E a
5030 100 150 200
Ds 25 m te hw T+
2 hedg0 -40 DAS; Spg 19-b T-
N 15
Z 10
5 t

50 100 150 200
Distance from the hedgerow (cm)

Figure 5-15: Maize height trenched and nontrenched at four distances from the
hedgerow, 40 DAS; Spring 1995.


Distance from hedgerows was significant for both trenched and nontrenched

plots, but not in the same way. Maize height in trenched plots reached a maximum at

100 cm, and then decreased (although not significantly) at 150 and 200 cm. Maize height

in nontrenched plots was minimum at 50 cm, and then increased progressively with

distance but never attained the height as maize in trenched plots (Table 5-14).


Table 5-14: Maize height in trenched and nontrenched plots at four distances from
hedgerows, 40 DAS; Spring 1995.

Distance (cm) Trenched (cm) Nontrenched (cm)
50 30.0 a 2.2 a
100 40.6 b 13.5 b
150 36.1 ab 16.8 b
200 36.5 b 23.9 c
p-value .009 <.001


Maize height was less than Leucaena height until the hedgerows were pruned at

60 DAS (Figure 5-5). This raises the question of whether the maize on the north side of








the hedgerows experienced a shade regime different from that on the south side. No

direct measurements of light were done, but maize height at 40 DAS was analyzed for the

effect of aspect. Neither the complete model (including distance as a factor) nor the

separate models for each distance showed orientation with respect to the hedgerows to be

a significant influence on maize height (data not shown).

The effect of fertilizer on maize height was significant only at the 50 cm distance

from the hedgerows and only in the trenched plots (p = .035), where fertilized maize was

10 cm shorter than nonfertilized maize.

Maize biomass production. Differences between maize biomass harvested in

trenched and nontrenched plots at 47 and 54 DAS were similar, and continued the pattern

of differences in maize height discussed above. However, the statistical tests produced

higher p-values than did those for height, probably due to smaller samples sizes and high

variability caused by three outliers in block 10 at the 50 and 100 cm distances. The

irrigation system leak causing water contamination in blocks 9 and 10, discussed above,

was probably responsible. The complete model ANOVA at 47 DAS was significant for

trenching (p = .035), but not for fertilizer or distance from the hedgerow. The 54 DAS

complete model was significant only for trenching at a=.10 (p = .085).

At 47 DAS, ANOVAs done separately for each distance showed that trenched

plots produce significantly more biomass at all four distances from the hedgerows

(Figure 5-16). P-values for trenching are = .030, .096, .016, and .017 for the four

distances. At 54 DAS, the difference due to trenching is significant only at the 50 cm

distance (p = .072). The difference between trenched and nontrenched plots at the 100

cm distance shown in Figure 5-17 was numerically large, but because of the outliers

discussed above, did not result in a significant difference.
















OT+
ST-


50 100 150 200


Distance from the hedgerow (cm)


Figure 5-16:


Maize biomass in trenched and nontrenched plots at four distances from
the hedgerow, 47 DAS, Spring 1995.


ST+
mT-


100 150
Distance from the hedgerow


Figure 5-17:


Maize biomass in trenched and nontrenched maize weight at four
distances from the hedgerow, 54 DAS, Spring.






82

The changes in biomass over distance were similar for both observation dates. In

neither case were there significant differences over distance for trenched plots. In both

cases, maize biomass in nontrenched plots was less in the two rows closest to the

hedgerows than in the more distant rows (Table 5-15).


Table 5-15: Maize biomass in trenched and nontrenched plots at four distances from
the hedgerow, 47 and 54 DAS; Spring 1995.

47 DAS 54 DAS
Distance (cm) T+ (g) T- (g) T+ (g) T- (g)
50 25.9 1.6 a 61.9 1.3 a
100 43.7 5.3 ab 77.9 6.0 a
150 30.4 9.2 b 36.1 18.8 b
200 25.0 6.5 b 21.3 18.3 b

p-value .232 .010 .118 .007



Maize growth stage. Maize in trenched plots had attained a more advanced stage

of development than maize in nontrenched plots at both 68 and 75 DAS. Competition

with the trees retarded development of maize plants growing in nontrenched plots

increasingly as they neared the hedgerows. This was not the case for maize plants

growing in trenched plots-development was most advanced at 100 cm from the

hedgerows.

Figures 5-18 and 5-19 use different scales to indicate maize growth stage (see

Materials and Methods). A four-number scale was used at 68 DAS: 0=missing.

l=vegetative growth, 2=tassel/anthesis, 3=silk, and 4=ear. At 75 DAS a five-number

scale was used: 0=missing, l=vegetative, 2=tassel, 3=anthesis, 4=silk, and 5=ear.

















3
Silk


^2

o T





0 -


Figure 5-18:


Trenched and nontrenched maize growth stage at four distances from the
hedgerow, 68 DAS; Spring 1995.


Ear

4.0
Silk


3.0 -
A

o 2.0

1.0
Ve

0.0 -







Figure 5-19:


iithesis

assel

(etative


I I


ET+
MT-


100 150
Distance from the hedgerow (cm)


200


Trenched and nontrenched maize growth stage at four distances from the
hedgerow, 75 DAS; Spring 1995.


100 150 200
Distance from the hedgerow (cm)


O T+
mT-








As was the case with maize biomass production. growth stage changed with

distance from the hedgerows differently for plants in trenched plots than for those in

nontrenched plots. Plants in trenched plots were most advanced at 100 cm from the

hedgerows, and identical at the other three distances. There was no retardation in the 50

cm row, and conditions for maize development appear to have been enhanced at 100 cm

due to the presence of hedgerows. Maize plants in the nontrenched plots were least

developed in the 50 cm row, and progressively more developed as distance increased.

Soil water content. The complete ANOVA models for soil water at 26 and 61

DAS did not produce significant p-values for trenching, fertilizer, or distance from the

hedgerow. Only depth was significant (p < .001) on both dates. ANOVAs run separately

for each distance and depth were also not significant for trenching or fertilizer, with the

exception of a significant p-value (.016) for fertilizer at 150 cm distance, 15-30 cm depth

at 26 DAS. The nonfertilized plots in that position had only 0.5% more soil water than

did fertilized plots, so this difference is probably meaningless. Figures 5-20 and 5-21

show the percent soil water for the various positions in the trenched plots, with the

difference between trenched and nontrenched plots in parentheses. None of these

differences were significant.

When ANOVAs were run separately for trenched and nontrenched plots for each

depth, there were no significant differences in soil water between distances from the

hedgerow for either trenched or nontrenched plots, with one exception. At 61 DAS in the

nontrenched plots in the 0-15 cm depth, the 50 cm distance had about 1.5% more soil

water than the other three more distant positions (p = .002). Soil water increased with

depth for both trenched and nontrenched plots at all distances from the hedgerows, at

both observation periods (Table 5-16).












t i Distance from Hedgerow (cm)

50 100 150 200
_k__ I


Soil depth (cm)

0-15 cm


15-30 cm


30-45 cm <


II


Figure 5-20:


26.4 (0.7)


28.5 (1.0)


28.8 (-0.3)


Soil Water (%, Gravimetric)

24.7 (0.5) 24.4 (1.4)


27.6 (0.6) 27.2 (=)


29.2 (0.4) 29.2 (-0.2)


24.1 (0.6)


27.0 (0.1)


29.6 (0.7)


Soil water percent in trenched plots (and difference between T+ and T-
plots) at 26 DAS, Spring 1995.


Figure 5-21: Soil water percent in trenched plots (and difference between T+ and T-
plots) at 61 DAS, Spring 1995.




Full Text
DYNAMICS OF FARMER ADOPTION, ADAPTATION. AND MANAGEMENT
OF SOIL CONSERVATION HEDGEROWS IN HAITI
By
MICHAEL E. BANNISTER
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2001

lililí
LD
1730
2QJL1
. 62-i
ACKNOWLEDGMENTS
I wish to thank my graduate advisor, Dr. P.K.R. Nair, for his unflagging
encouragement and guidance throughout a long period of research and writing and Dr.
Peter Hildebrand for providing my Farming Systems assistantship from 1987 through
1989. My other committee members, Dr. Mary Duryea, Dr. Jerry Murray, and Dr. Ken
Buhr also provided valuable insights and assistance. The untiring work of the Pan
American Development Foundation (PADF) Haitian technical staff during data collection
and the encouragement and financial support of the PADF home office during my
sabbatical were essential to this dissertation, as was the cooperation of a multitude of
Haitian farmers. Deserving of special mention are PADF Haiti Country Representative
Lee Nelson for his key support of my sabbatical and Agronomist Jn-Charlot Bredy for
data collection on farms near Lascahobas. Data collection for the Leucaena/maize study
could not have been completed without the assistance of technicians Duverger Vemis and
Rodini St-Juste. Finally, I thank my beloved wife Mojdeh for her essential care of my
sometimes-fragile psyche during this trying time.

TABLE OF CONTENTS
pages
LIST OF TABLES v
LIST OF FIGURES x
ABSTRACT xiii
CHAPTERS
1 INTRODUCTION 1
2 HILLSIDE FARMING SYSTEM IN HAITI 4
Land 5
Crops 7
Labor 9
Farmer Development Groups and Local NGOs 10
3 THE HEDGEROW INTERCROPPING SYSTEM 13
Developments in the Tropics 13
Developments of Hedgerow Intercropping in Haiti 18
Hedgerow Intercropping and Soil Conservation 22
Conclusions 26
4 ORGANIZATION OF THE STUDY 27
Justification 27
Hypotheses 33
General Description of Hypothesis-Testing Procedures 34
5 HEDGEROW/CROP COMPETITION 36
Introduction 36
On-Station Studies: Effect of Root Barriers on the Growth of Maize and
Leucaena 39
On-Farm Study: Maize Growth at Varying Distances from Hedgerows 100
iii

6 LAND AND HOUSEHOLD CHARACTERISTICS IN RELATION TO
ADOPTION AND MANAGEMENT OF HEDGEROWS 108
Introduction and Objectives 108
Materials and Methods 119
Results and Discussion 127
Summary and General Discussion 166
Conclusions 175
7 SYNTHESIS AND CONCLUSIONS 179
APPENDICES
A POT STUDY OF SOIL WATER DEPLETION VS. MAIZE LEAF WATER
POTENTIAL 186
B ROOT DISTRIBUTION OF A LEUCAENA LEUCOCEPHALA
HEDGEROW 188
C P-VALUES OF ANOVAS PERFORMED ON STATION 191
D HOUSEHOLD QUESTIONNAIRE 198
E GARDEN PLOT QUESTIONNAIRE 201
F SOIL NUTRIENTS, ORGANIC CARBON STATUS, AND pH
OF 175 HILLSIDE GARDENS 211
REFERENCES 218
BIOGRAPHICAL SKETCH 235
iv

LIST OF TABLES
Table pane
5-1: Soil texture and organic carbon at the trial site 43
5-2: Effect of trenching on the production of maize biomass 60
5-3: Effect of fertilizer on the production of maize biomass 61
5-4: Soil water at three depths below the surface 62
5-5: Maize grain weight, number of plants, and number of ears 65
5-6: Stem and leaf biomass harvests and daily growth increments of Leucaena
hedgerows 66
5-7: Height above a 50 cm stump and daily growth increments of Leucaena
hedgerows prior to the second root trenching, 33 and 65 DAS; Spring
1994 T 67
5-8: Maize biomass in trenched and nontrenched plots 69
5-9: Maize leaf water potential in trenched and nontrenched plots over four
distances from the hedgerows, 42 DAS; fall 1994 71
5-10: Soil water in trenched and nontrenched plots by depth and distance 73
5-11: Maize grain weight, number of plants, and number of ears at four
distances from the hedgerows and as a total per hectare, with p-values for
the differences between distances, Fall 1994 75
5-12: Stem and leaf biomass harvests and daily growth increments of Leucaena
hedgerows after the second root trenching, Fall 1994 76
5-13: Height above a 50 cm stump and daily growth increments of Leucaena
hedgerows after the second root trenching. Fall 1994 77
5-14: Maize height in trenched and nontrenched plots at four distances from
hedgerows. 40 DAS; Spring 1995
v
79

5-15: Maize biomass in trenched and nontrenched plots at four distances from
the hedgerow, 47 and 54 DAS; Spring 1995 82
5-16: Soil water in trenched and nontrenched plots by depth and distance from
the hedgerow, 26 and 61 DAS; Spring 1995 86
5-17: Maize grain weight, number of plants, and number of ears 89
5-18: Stem and leaf biomass harvests and daily growth increments of Leucaena
hedgerows 90
5-19: Height above a 50 cm stump and daily growth increments of Leucaena
hedgerows 90
5-20: Maize yield in three soil conservation practices and control at 116 DAS on
four farms, spring 1997 105
6-1: Age distribution by gender of members of 1,540 farm households
interviewed in the four PADF/PLUS project areas in Haiti, 1996 128
6-2: Number of years of school by gender in seven age classes of 1,540
households participating in the PADF/PLUS project, 1996 129
6-3: Percent of female and male members of 1,540 project households who
participated in various economic activities during 1995-1996 130
6-4: Percent of female and male members of 1,540 project households who
participated in various economic activities during 1995-1996, by age
class 131
6-5: Percent of female and male members of 1.540 households who
participated in PLUS project activities during 1995-1996 132
6-6: Percent of female and male members of 1,540 project households who
participated in PLUS project activities during 1995-1996, by age class 133
6-7: Average total area of three categories of plots held by 1,540 project
households based on the relative distance from the plots to the residence .... 134
6-8: Mean plot area of three categories of plots held by 1,540 project
households during 1995, based on the relative distance from the plots to
the residence 134
6-9: Mean plot area of three categories of plots held by 1,540 project
households during 1995 in four PLUS project regions, based on the
relative distance from the plots to the residence 135
vi

6-10: Percent of three categories of plots held by 1,540 project households
during 1995 under various modes of access 136
6-11: Area of 5,660 plots held by 1,540 households during 1995, by mode of
access 137
6-12: Percent of plots held by 1,540 households during 1995 in eight mode of
access categories by slope position 137
6-13: Percent of all plots held by 1,540 households during 1995 in eight modes
of access categories by slope steepness 138
6-14: Distance from residence of plots having project agro forestry practices held
by 1,540 households during 1995 in the five most common mode of
access categories 139
6-15: Area of plots having project agroforestry practices held by 1,540
households during 1995 in the five most common mode of access
categories 140
6-16: Elevation of plots having project agro forestry practices held by 1,540
households during 1995 in seven mode of access categories 140
6-17: Slope of plots having project agroforestry practices held by 1,540
households during 1995 in seven mode of access categories 141
6-18: Percent of 2,295 surveyed plots with and without three kinds of
agroforestry practice held under secure (purchased or inherited/divided) or
not secure (all other) tenure 147
6-19: Percent of 2,295 surveyed plots with and without three kinds of
agroforestry practice held under secure (purchased or inherited/divided) or
not secure (all other) tenure 147
6-20: Percent of 2,295 surveyed plots with and without three kinds of
agroforestry practice having fertile or not fertile soil 147
6-21: Percent of 2,295 surveyed plots with and without three kinds of
agro forestry practice having fertile or not fertile soil 148
6-22: Percent of 2,295 surveyed plots with and without three kinds of
hedgerows held under secure (purchased or inherited/divided) or not
secure (all other) tenure 149
6-23: Percent of 2,295 surveyed plots with and without three kinds of
hedgerows having fertile or not fertile soil 149
vii

6-24: Number of all trees larger than 10 cm diameter on PLUS plots, by plot
tenure 150
6-26: Soil fertility and hedgerow management quality 154
6-27: PADF/PLUS field region and hedgerow management quality 156
6-28: Soil fertility and crop band management quality 157
6-29: PADF/PLUS field region and crop band management quality 158
6-30: Soil fertility and rock wall management quality 159
6-31: PADF/PLUS field region and rock wall management quality 160
6-32: Percent of 1,362 plots where potential benefits of hedgerows (fuel,
charcoal, construction wood, fodder, soil fertility, soil conservation, and
crop production) were classified according to importance by farmers on a
five-point scale, l=not important, 5=very important 161
6-33: Percent of hedgerows classified as well-managed by technicians in 1,362
plots where potential benefits of hedgerows (fuel, charcoal, construction
wood, fodder, soil fertility, soil conservation, and crop production) were
classified according to importance by farmers on a five-point scale, l=not
important, 5=very important 162
6-34: Percent of hedgerows classified as poorly-managed by technicians in
1,362 plots where potential benefits of hedgerows (fuel, charcoal,
construction wood, fodder, soil fertility, soil conservation, and crop
production) were classified according to importance by farmers on a five-
point scale, l=not important, 5=very important 163
6-35: Number of breaches per 100 m of hedgerow counted by technicians in
1,362 plots where potential benefits of hedgerows (fuel, charcoal,
construction wood, fodder, soil fertility, soil conservation, and crop
production) were classified according to importance by farmers on a five-
point scale, l=not important, 5=very important 164
6-36: Percent of gardens where hedgerow breaches were repaired or not repaired
for three potential benefits of hedgerows 165
6-37: Percent of 1,362 plots where potential problems of hedgerows (shade,
reduced space, water competition, reduced ability to picket animals,
weediness, and labor cost) were classified according to importance by
farmers on a five-point scale, l=not important, 5=very important 165
viii

6-38: Percent of gardens where hedgerow' breaches were repaired or not repaired
for three potential problems of hedgerows 166
6-39: Estimated relative costs, benefits, and risk of agroforestry practices 171

LIST OF FIGURES
5-1: Layout of the Leucaena!maize hedgerow trial on Operation Double
Harvest property near Croix des Bouquets, Haiti 45
5-2: Diagram of one plot showing relative positions of a Leucaena hedgerow
and the adjacent rows of maize at four distances 46
5-3: Rainfall, Leucaena and maize growth, and agronomic inputs; spring 1994
maize cropping season 51
5-4: Rainfall, Leucaena and maize growth, and agronomic inputs; fall 1994
maize cropping season 51
5-5: Rainfall, Leucaena and maize growth, and agronomic inputs; spring 1995
maize cropping season 52
5-6: Maize biomass yields at 50, 100, 150, and 200 cm from the hedgerows at
four times after sowing during the spring 1994 season 59
5-7: Growth stage of maize at four distances from hedgerows 61
5-8: Yield of maize grain per six meters at four distances from the hedgerows 64
5-9: Number of maize plants per eight hills at four distances from the
hedgerows 64
5-10: Number of maize ears per eight hills at four distances from the hedgerows .... 65
5-11: Trenched and nontrenched plots at 50, 100, 150, and 200 cm from the
hedgerows, 78 DAS; fall 1994 69
5-12: Maize leaf water potential in trenched and nontrenched plots 70
5-13: Soil water in trenched plots (and difference between T+ and T- plots) at
42 DAS, fall 1994 72
5-14: Maize grain yield in trenched (T+) and nontrenched (T-) plots
74

5-15: Maize height trenched and nontrenched at four distances from the
hedgerow, 40 DAS; Spring 1995 79
5-16: Maize biomass in trenched and nontrenched plots at four distances from
the hedgerow, 47 DAS, Spring 1995 81
5-17: Maize biomass in trenched and nontrenched maize weight at four
distances from the hedgerow, 54 DAS, Spring 81
5-18: Trenched and nontrenched maize growth stage at four distances from the
hedgerow, 68 DAS; Spring 1995 83
5-19: Trenched and nontrenched maize growth stage at four distances from the
hedgerow, 75 DAS; Spring 1995 83
5-20: Soil water percent in trenched plots (and difference between T+ and T-
plots) at 26 DAS, Spring 1995 85
5-21: Soil water percent in trenched plots (and difference between T+ and T-
plots) at 61 DAS, Spring 1995 85
5-22: Maize grain weight from trenched and nontrenched plots at four distances
from the hedgerows, 131 DAS; Spring 1995 87
5-23: Number of maize plants from trenched and nontrenched plots at four
distances from the hedgerows, 131 DAS; Spring 1995 88
5-24: Number of ears from trenched and nontrenched plots at four distances
from the hedgerows, 131 DAS; Spring 1995 88
5-25: Number of very fine root (<2mm) intersections with a 10 by 100 cm plane
parallel to the trenched hedgerows at five distances and 10 depths, with
the difference between the trenched and nontrenched hedgerows shown in
parentheses 92
5-26: Number of fine root (2-5 mm) intersections with a 10 by 100 cm plane
parallel to the trenched hedgerows at five distances and 10 depths, with
the difference between the trenched and nontrenched hedgerows shown in
parentheses 92
5-27: Number of medium root (5-10 mm) intersections with a 10 by 100 cm
plane parallel to the trenched hedgerows at five distances and 10 depths,
with the difference between the trenched and nontrenched hedgerows
shown in parentheses 93
xi

5-28: Number of large root (>10 mm) intersections with a 10 by 100 cm plane
parallel to the trenched hedgerows at five distances and 10 depths, with
the difference between the trenched and nontrenched hedgerows shown in
parentheses 93
5-29: Leucaena small stem and leaf biomass from trenched and nontrenched
plots, May 1991 through September 1995 98
5-30: Maize development in three soil conservation practices and control at four
times during the growing season on four farms during Spring 1997 104
6-1: Relative importance of four plot characteristics to the decision to install
six agro forestry practices 168
xii

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
DYNAMICS OF FARMER ADOPTION, ADAPTATION. AND MANAGEMENT
OF SOIL CONSERVATION HEDGEROWS IN HAITI
By
Michael Bannister
May 2001
Chair: P. K. Ramachandran Nair
Major Department: School of Forest Resources and Conservation
Understanding the conditions under which hillside farmers in Haiti adopt soil
conservation practices helps programs to develop technologies that increase farmer
revenue and stabilize or improve soil and water resources. Three studies were done in
Haiti that examined biophysical and socioeconomic aspects of contour hedgerows, an
agroforestry practice with the potential of stabilizing soil, conserving water for crops,
increasing soil fertility, and producing wood and fodder.
An on-station study of soil water competition between Leucaena leucocephala
(Lan.) de Wit hedgerows and adjacent maize (Zea mays L.) over three cropping cycles
found that substantial maize yield reduction was caused by the hedgerow trees because of
soil water competition. Polyethylene barriers between the hedges and the first row of
maize improved maize yield in two subsequent seasons by 18% and 77% over plots
without barriers. The percent increase in maize yield was highest during a season of very
poor rainfall; but it was lowest in a season of adequate rainfall. Barrier installation
xiii

reduced Leucaena biomass production by about 1,500 kg/ha over seven months, but this
effect was temporary. An examination of the distribution of Leucaena roots after the
final maize harvest showed that the hedgerow roots in the plots with barriers had grown
under the hedgerows and developed more fine roots at 200 cm from the hedges than plots
without barriers.
An on-farm study comparing maize development in two kinds of hedgerows, rock
wall terraces, and an untreated control was unable to detect differences in the rate of
maize growth with respect to position within the alleys or with respect to position on the
overall slope. However, the maize in the hedgerows developed more slowly than in the
rock wall terraces or the control plot, indicating competitive interactions between maize
and hedgerow plants. There were no differences in maize grain yield among treatments.
An on-farm questionnaire-based survey of 1,540 Haitian farmers showed that
they considered plot characteristics, including mode of tenure, soil fertility, distance from
the residence, and slope in their decisions to install different agroforestry practices.
Tenure security and soil fertility appeared to be the most important plot characteristics in
the decision to install, although it was not possible to separate them. Hedgerows were
more likely to be installed in plots having less secure tenure and less fertile soil; the
opposite was true for other agroforestry practices. Farmers’ qualitative assessments of
soil fertility were positively correlated with management quality of hedgerows.
xiv

CHAPTER 1
INTRODUCTION
The research presented here focuses on agroforestry technologies, which could
contribute to increasing the revenue of small farms on the hillsides of Haiti in a
sustainable, environmentally friendly way. This is a formidable goal, because Haiti is
primarily a mountainous country long since shorn of its natural vegetative cover. Its
once-rich soils are badly eroded, and their productive capacity severely diminished.
Haiti is densely populated by very poor people who generally lack formal education or
access to a functioning legal system. It has a long history of political instability that
continues to this day. At best, agriculture is difficult in such a setting. But, Haitian
farmers regularly experiment with new techniques.
This dissertation investigates the biological and adoption aspects of hedgerow
intercropping, an agroforestry practice that involves growing food crops in the
interspaces between rows of densely planted trees or shrubs. Two types of studies are
included. One focuses on competition for growth resources between trees and crops
while the other focuses on the factors that influence farmers’ adoption of the technology.
However, these different studies are linked together, because both social and biological
aspects of agroforestry practices must be understood for providing efficient extension
services to farmers.
The dissertation is presented in seven chapters. Chapter 1 briefly explains the
background of the research and states its objectives. Chapter 2 describes Haitian
I

smallholder farms, the population of interest to the research. Following a definition of
hedgerow intercropping, the agroforestry practice that is the subject of the research.
Chapter 3 reviews its development in the tropics and in Haiti and examines how it is used
as a soil conservation technique. Chapter 4 justifies the research, presents the hypotheses
to be tested, and describes the studies used to test the hypotheses. The next chapter
(Chapter 5) presents the two studies examining competition between hedgerows and
adjacent crops. The first is done on station, the second on farm. The questionnaire-based
survey of farmers who adopted soil conservation practices promoted by the PLUS project
is the subject of Chapter 6. Finally, Chapter 7 synthesizes the previous chapters,
and presents the overall conclusions of the dissertation.
The research was done while the author was employed by the Pan American
Development Foundation (PADF) in Haiti. The PADF implements a U.S. Agency for
International Development (USAID) project called Productive Land Use Systems
(PLUS). The PADF/PLUS has worked with about 170,000 hillside farmers in soil
conservation, tree planting and grafting, has improved food crop distribution and
marketing, and has been strengthening local farmer organizations between 1992 and
1999. The on-farm study of maize (Zea mays L.) production between soil conservation
structures discussed in Chapter 5 and the questionnaire-based study discussed in Chapter
6 were done as PLUS monitoring and evaluation exercises by PADF staff. The author
designed the questionnaires and trained staff in their use.
The overall objective of this research is to understand how characteristics of
Haitian farm families, their plots, and agroforestry practices affect the adoption and
management of those practices. The purpose is not to use the data gathered in the course
of this research to define and predict adopters so that nonadopters can be excluded from

extension efforts. It is rather an effort to understand how appropriate practices can
evolve and be extended more rapidly.
In particular, the objectives of the three studies that comprise this dissertation are
• to determine, in the relatively controlled environment (for a research station),
whether the competition for soil water between trees in hedgerows and
adjacent maize is significant and can be reduced by the use of plastic root
barriers,
• to understand whether tree-crop competition in farmers' steeply sloping fields
is similar to that observed in the flat controlled environment of the research
station, and
• to examine the factors (e.g., land tenure, perceived tree/crop competition, plot
slope, soil fertility of the plot, distance from the residence to the plot, previous
experience of the farmer, and the availability of other farm resources) that
influence farmer adoption, adaptation, and management of hedgerow
intercropping.

CHAPTER 2
HILLSIDE FARMING SYSTEM IN HAITI
Haiti is mainly hilly, with 20% of the land area below 185 m elevation and 40%
above 450 m—about 75% highlands, 22% lowlands (Weil et al. 1973). The most
common “life zones,” according to Holdridge's (1945) classification, are subtropical
moist forest and subtropical dry forest, but subtropical wet and rain forest zones are
common in the middle and upper elevations. Limestone substrate underlies 80% of the
land area; the rest is basaltic or alluvial (Ehrlich et al. 1985). According to Ehrlich et al.
(1985) estimates, 1.3 million hectares (of 2.7 million total for the country) were under
agricultural production; six times the area classified as “good agricultural land.” The
practice of annual agriculture on steeply sloping plots causes severe losses of arable soil
due to erosion, estimated to be about 37 million tons annually in 1994 (UNDP, 1996).
Haitians have been practicing agriculture on small, privately held plots in the hills
since the 1840s. as the large plantations and public domain lands were divided (de Young
1958. Levbum 1941). Haitian peasants are not subsistence farmers in the strict sense that
they do not live off the land self-sufficiently and independently, but rather produce for,
and are linked closely to, the market (Murray 1981). Most farm families have some
nonagricultural income and are net food buyers (Locher 1996). De Young (1958)
describes the economic organization of Haitian farms as “horticultural” rather than
agricultural, and gives the following characteristics that still hold true today:
• They are market directed, not subsistence directed.
4

5
• They give more importance to perennial rather than annual crops.
• The farm units are small parcels, and the land is used intensively.
• They are capital saving rather than labor saving. Farmers own only a few
hand tools, have little invested in storage or farm buildings, and rarely use
plows and animal traction.
• The farm animals are used mainly for transport to market.
Reducing risk is as important to most farmers as increasing production, and to
this end they manage a network of ties to relatives and other farmers as social capital that
can be leveraged in order to gain access to scarce land, labor, and capital (Smucker et al.
2000). Haitian families also spread out risk by diversifying human and economic
resources via selective migration, emigration, and off-farm economic enterprises (Locher
1996). Therefore, not only is any one particular crop not the sole interest of a farm
family, but agriculture is not the only domain of economic activity. It is quite likely that
other activities, such as nonagricultural marketing or remittances from relatives overseas,
bring in more cash.
Land
Haitian hillside farms are small and complex. Farm families usually manage or
cultivate several parcels of land at any given time, held under a mix of land tenure
arrangements (e.g., owned, tenancy, usufruct) (Bloch et al. 1987, Locher 1996) and often
on sites differing in their physical characteristics (hillside, plain, irrigated, rainfed) and in
distance from the residence. This is done as a purposeful strategy to reduce risk of
harvest loss and to spread out crop harvest over the year (McLain and Stienbarger 1988,
Murray 1981, Zuvekas 1979). Plot sizes are small and fallow periods are short. Plots are
worked and then abandoned when yield decreases below the farmer’s economical limits.
Declining soil fertility, soil loss due to erosion, large variation in the timing of rainfall,

6
and the inability of traditional agricultural practices to harvest and store water in the soil
limit agricultural production. Residences tend to be physically disbursed in the mountain
regions instead of concentrated (Friedman 1955).
Two intertwining land tenure systems operate in Haiti, one official and the other
customary (Smucker et al. 2000). created by disenfranchised peasants to protect
themselves against loss of land via a corrupt justice system and government abuse. The
land market is active. Most farmers own most of their land, but the majority of the
transactions are done without recourse to expensive official processes (Smucker et al.
2000, Zuvekas 1979). The customary system provides farmers relatively secure access to
land, but some tenure classes are more stable over time than others so mixed tenure
configurations that change over time are the norm (Locher 1996). Zuvekas (1979)
reviewed data from 12 studies done in Haiti and found that the percentage of farmers
owning (by purchase or inheritance) at least part of their land ranged from 58 to 100%.
Secondary modes of access to land include rental from private owners, lease, and
sharecropping. Usufruct rights to land are also obtained from relatives or by acting as
caretaker for an absentee landowner
Information about farm size is available from several studies related to
agroforestry projects from various parts of the country (Lea 1994a, Lea 1996, Swanson
et al. 1993a, 1993c, White 1992a). The total size of all farm plots ranges from less than
one to about three hectares. The average farm family works two or three plots (Zuvekas
1979) that vary in size from about one half to just over one hectare each.
Haitian laws of inheritance have French antecedents and require that all siblings
inherit; therefore, land holdings are, in general, getting smaller. In 1950, 39% of
households held less than 1.3 hectares; by 1971 this proportion had increased to 71%

7
(Murray 1981). Brochet (1993) compared the average size of home gardens in one area
of Haiti over fifty years, and found that the average holding that was one hectare per
family in 1910 had decreased to 500 nf (0.05 ha) by 1970. This study was done in an
area mined for bauxite by Reynolds Aluminum, so sale of land to the mining company
may have accelerated the rate of decrease, but in general average sizes of holdings grow
smaller as the population increases. One strategy adopted by Haitian families to ensure
that a plot is large enough to harvest a crop worth planting is to bequeath land to children
undivided, but this may discourage investing in improvements to the plot if the siblings
are contentious. Older farmers are likely to own a larger proportion of their holdings
(Lundahl 1997, McLain and Stienbarger 1988).
Other types of land tenure found in Haiti include purchased, inherited but
separated, rented from private owners, and rented from the state. Usufruct rights to land
also are obtained by sharecropping (on land owned by the sharecropper as well as on land
owned by another) and by acting as caretaker for an absentee landowner. A recent
survey of 5000 fields across the country (USAID. 1996) showed the following pattern of
land tenure: purchased 39%, inherited/divided 20%, inherited/undivided 14%, rented
12%, share cropped 5%. gift 1%. and other 9%.
Crops
Since small and scattered plots must provide food, fuel, building material, animal
fodder, and other household necessities, agroforestry has traditionally been practiced.
The farm plots, except the plot containing the residence and surrounding tree and
medicinal crops (sometimes called the home garden), are spatially dominated by cereal
grains and grain legumes (pulses), but income from roots, tubers, tree crops, and animals
is probably more important economically, and is able to shield farmers against variations

in rainfall better than grains and pulses. Home gardens include many tree crops such as
mango (mangifera indica L.), breadfruit (Arto carpus altilis [Parkinson] Fosberg), citrus
(Citrus spp.). royal palm (Roystonia borinqueño), and coconut (Cocos nucífera L.).
Timber and fuelwood trees are planted or tended where they sprout around property
perimeters, scattered throughout plots, or as small woodlots. Trees are considered to be a
crop, and are managed as a savings account, to be harvested when cash or building
materials are needed (Murray 1981, Smucker and Timyan 1995). Fuelwood and animal
fodder are in short supply in many parts of the country.
Many different combinations of crops are grown. These vary with the season,
land tenure, distance from the residence, market demand, availability of seed, and the
rainfall and soil characteristics. The most common grain crops are maize (Zea mays L.)
and sorghum (Sorghum bicolor (L.) Moench) on the hillsides, and rice (Oryza sativa L.)
in the few irrigated plains. Pulses include beans (Phaseolus vulgaris L.), pigeon pea
(Cajanus cajan (L.) Millsp.), and peanuts (Arachis hypogaea L.). It is common for the
spring maize crop to fail, because of poor rainfall. Beans are able to generate important
income for farmers, but also fail often-rain is insufficient, or rain falls during flowering
(eliminating pod formation), or rain falls before seeds are dry causing them to germinate
in the pods. Root and tuber crops include yams (Diascora spp.), cassava (Manihot
esculenta Crantz), and cocoyam (Xcmthosoma sagittifolium (L.) Schott). Plantains (Musa
spp.) are commonly grown in ravines and near the home. Living fences surround the
property where the house is located, and are commonly built of plants serving protective
functions due to the presence of spines or poisonous sap, such as bayonette (Yucca
aloifolia L.). pencilbrush (Euphorbia tirucalli L.), raquette (Euphorbia lactea Haw.), and
penguin (Bromelia pinguin) (Ashley 1986).

9
A 1996 USAID survey of 6,900 fields during a single spring planting season
(USAID, 1996) found that 27% of the plots had crops growing in association (two or
more crops present simultaneously in the same field for part or all of the cropping cycle),
and 21% in monoculture. There were 16 main maize associations and 98 minor (<5000
ha. each) ones. In addition, there were 12 important associations of other crops, 134
minor ones, and 14 main kinds of monocrops. This does not include the many different
kinds of tree and medicinal crops, crops found during the fall growing season, or animals.
The most commonly seen grain crop in the fall season is sorghum, usually intercropped
with pigeon pea. The USAID study notes only the physical presence of the crops, not
their relative economic importance.
Labor
Households function as separate economic units, but the members participate in
labor and food exchanges with other farmers (Murray 1981, Smucker and Dathis 1998).
The family itself provides the labor for most agricultural activities. This labor pool is
expanded periodically according to seasonal demands for land clearing, working the soil,
planting, weeding, and harvesting crops. Labor is a scarce resource, and has been more
difficult to supply and afford as the deteriorating social and economic situation in Haiti
has forced many rural inhabitants into political refuge or into taking jobs in the
neighboring Dominican Republic (Balzano 1997, Smucker and Dathis 1998). Farmers
form groups to exchange farm labor, earn cash, or to save money to attain a future goal.
However, this communal pooling is restricted to labor, not to the sharing of land or
produce (Murray 1991). Access to labor is crucial because working steep slopes is labor
intensive, and frequently must be done on short notice because rainfall varies so much
over short distances in the mountains. Crops must be planted as soon as the first rains

10
arrive. Farmers without access to labor at the right time will suffer. Work groups
generally consist of 3 to 15 members from the same socioeconomic level, and can be
permanent or formed for a single task (Murray 1991, Smucker and Dathis 1998, Swanson
et al. 1993a, 1993b 1993d, 1993e). The names given to labor groups differ from place to
place within Haiti, and even from farmer to farmer (Smucker and Thomson 1999). The
two main kinds of groups are those that are for labor exchange only and those that are
formed to work for cash. Client farmers paying cash for labor are generally not group
members, and are better off economically. A wealthy farmer may join a labor exchange
group just for the purpose of gaining access to the labor. Another kind of work party is
called the koumbit, where the host farmer provides one or two meals and sometimes
alcohol to other farmers invited to clear or prepare a field (Swanson et al. 1993a). It is
not possible to control the number of participants, the length of time they stay, or the
quality of the work done, and for this reason the koumbit is becoming rare (Swanson et
al. 1993e). Finally, individual farmers also are engaged by others to do specific tasks.
This is referred to as a djob (job).
Typically, men form labor exchange groups for agriculture while women form
mutual credit supply groups that permit them to engage in commerce. Women have
always participated in some aspects of agricultural labor, and this appears to be on the
rise because of the out-migration of men during times of political and economic troubles
mentioned above (Smucker and Dathis 1998).
Farmer Development Groups and Local NGOs
Indigenous development groups are of interest because, since the early 1980s,
most donor-financed development work, including agroforestry extension, has been done
through them. This happened because of the difficulty in controlling the use of funds

11
disbursed through the Haitian government, which has always been a predatory structure
serving the interests of the urban elite to the exclusion of the rural majority (Smucker and
Thomson 1999).
Smucker and Dathis (1998) give a comprehensive description of these groups, on
which the rest of this section is based. These groups vary widely in origin and
composition, ranging from community organizations, cooperatives, youth groups,
women’s groups, work exchange groups, traditional groups, and religious groups.
Agricultural cooperatives began in the 1930s and expanded in number in the 1950s. The
community council movement began in the 1950s and was politicized by the Duvalier
regime in the 1970s. The Catholic Church created nonhierarchical farmer development
groups beginning in the 1960s that became known as groupement and eventually stood in
opposition to the community councils. After the fall of the Duvaliers in 1986 the
numbers of groups based on the groupement model increased dramatically; but they were
persecuted, and they went underground after the 1991 coup d’etat. All of the groups
have economic survival objectives, but are not limited to that. They look for more
control over the factors of production, including agricultural inputs, credit, grain storage,
more control over export crops, and investment in soil conservation and trees. They are
now experiencing growing pains as their membership increases and they take on more
objectives, without the legal recognition that would permit them to defend their interests
and open bank accounts. These difficulties increase as their growth pushes them to
depart from the basic characteristics of the traditional local groups, which are direct
participation, small size with common interests, hope of tangible profits, direct
investment of members’ resources, and links of reciprocity and rotation. Farmers would
be better able to invest in necessary agroforestry practices if these groups were stronger.

12
For example, a group with juridical personality, basic accounting skills, and a bank
account would be more likely to succeed in negotiating crop sales to large buyers.
Profits from sales would allow them to buy labor to protect the plots that produce the
crops being sold.

CHAPTER 3
THE HEDGEROW INTERCROPPING SYSTEM
Developments in the Tropics
Definition
Hedgerow intercropping, also known as alley cropping and alley farming (Rao
et al. 1998), has been suggested as an approach to address decreasing soil fertility and for
controlling erosion (Kang et al. 1990). It is an agroforestry practice consisting of closely
planted and regularly spaced rows of fast-growing, repeatedly pruned trees, usually N2-
fixing legumes, with short cycle agricultural crops cultivated in the alleys or spaces
between the hedges (Kang et al. 1990). It is classified as a simultaneous (trees and crops
present in the field at the same time) and zonal (trees are concentrated in the field rather
than dispersed) agroforestry practice (Nair 1993). The hedges and crops are managed
together for a common purpose, and there are significant above- and below-ground
ecological interactions between them (Sinclair 1999). The most common objective of
hedgerow intercropping is to increase crop production in the alleys by applying the
pruned tops of the hedgerow trees to the soil as a green manure to improve soil fertility.
This objective is pursued by attempting to balance the facilitative and competitive effects
inherent in the practice (Vandermeer 1998). The more leaf biomass harvested from the
trees, the more green manure available to improve soil conditions and thereby increase
crop yield. However, more leaf biomass is obtained by increasing the number of
hedgerows and this reduces the space available to crops. It also increases the area of
13

14
tree-crop interface (Ong and Leakey 1999) and therefore creates more possible
competitive interactions (shading, soil water competition, nutrient competition) with the
crops, potentially depressing yield (Vandermeer 1998). Hedgerows fulfill other purposes
as well, including soil erosion control and production of fodder and fuel.
Origins
Hedgerow intercropping developed independently both in the western humid
lowlands of Africa and in southeast Asia. In Nigeria, traditional farmers have for several
generations planted and managed hedges of a shrub, Dactyladenia barteri (Hook f. ex
Oliver), for the purposes of contributing nutrients to crops, controlling weeds, and
production of fodder and stakes (Kang et al. 1984). Looking for a way to respond to soil
impoverishment due to the shortening cycle of traditional slash and bum agriculture,
researchers began investigating similar ways to use trees. The principles suggesting
intercropping plants with complementary root systems occupying different soil regions
(Weaver 1920) and the possibility of tree roots capturing nutrient resources out of the
reach of crops (Nye and Greenland 1960) were familiar, but had not yet been applied
systematically to agroforestry associations by nonfarmer researchers. The International
Institute for Tropical Agriculture (UTA), based in Ibadan, Nigeria, began trying woody
species to improve the fertility of low-activity clay soils. This led to the development of
alley cropping (also referred to as hedgerow intercropping), avenue cropping, and alley
farming if fodder production was a main objective (Kang and Wilson 1987). The first
alley cropping trial at IITA began in 1976; on-farm trials were established in the 1980s
(Tripathi and Psychas 1992).
An earlier start was made in Indonesia. Development of hedgerow intercropping
in this region appears to be related to the introduction of Leucaena spp. to the Philippines

15
and Indonesia by the Spanish from Mexico (Dijkman 1950) and its promotion by the
Dutch as a shade tree for coffee (Coffea arabica L.), cacao (Theobroma cacao L.), tea
(Camelia sinensis (L.) Kuntze), and tobacco (Nicotiana tabacum L.) in the 18th century
(Wirjodarmodjo and Wiroatodjo 1983). Dutch journal articles from the 1930s report the
use of Leucaena contour hedges to support bench terraces on slopes (Dijkman 1950).
Tacio (1993) cites a traditional ruling in the 1930s obliging all farmers in a region of
Timor, Indonesia to plant contour hedges of Leucaena 3 m apart in cropped areas.
Metzner (1976) notes Leucaena thickets were promoted by the Indonesian government in
the 1930s to recuperate nonarable lands, but farmers did not do it because they feared the
weedy qualities of Leucaena. By the 1950s, Leucaena was commonly used in the Pacific
as contour hedges for erosion control (Dijkman 1950). Hernandez (1961), cited in Benge
(1980), discusses results of Indonesian hedgerow intercropping research done in 1953.
The hedgerows decreased erosion and improved maize yield by 380% on the treated
plots, compared to traditional cultivation. Extension of hedgerows to farmers, supported
by the Indonesian government and religious groups, expanded greatly during the 1970s
(Metzner 1976. Tacio 1993, Wirjodarmodjo and Wiroatodjo 1983).
The Rise and Fall of Aliev Cropping
Research on alley cropping in West Africa in the 1980s generated widespread
interest (Kang and Wilson 1987) and led to trials and extension in many parts of the
world. In trials examining their technical performance, hedgerows had positive effects
on the yield of adjacent food crops through the application of tree prunings as green
manure (Tonye and Titi-Nwel 1995), increased rainwater infiltration into the soil (Kiepe
1995), and reduced soil erosion when installed on sloping land (Paningbatan et al. 1995).
Some economic analyses indicated that alley cropping, in spite of its high labor

16
requirement, can be profitable under certain environmental conditions (Ehui 1988).
Based on these initial results, project based extension of hedgerow intercropping to
farmers in developing countries grew quickly (Ongprasert and Turkelboom 1995,
Sanchez 1995, Scherr 1994).
However, after a few years many came to believe that alley cropping had been
oversold and had much less applicability than previously thought, and that its potential
for adoption was limited to certain niche areas (Carter 1995, Sanchez 1995). Based on a
review of hedgerow trials covering some of the possible biophysical conditions and
hedgerow configurations, Sanchez (1995) lists the conditions where alley cropping is
most likely to succeed: areas having fertile soils with no major nutrient limits, adequate
rainfall during the cropping season, sloping land with erosion hazard, ample labor and
scarce land, and secure land tenure. In spite of the apparent productive advantages
discovered in controlled studies and the enthusiasm of project extension technicians,
adoption among farmers remained generally low. This has been observed in many
regions, including Indonesia and the Philippines (Fujisaka 1991, Nelson et al. 1997),
Thailand (Ongprasert and Turkelboom 1995). Central America (Kass et al. 1995). West
Africa (Lawry et al. 1994). and East Africa (Mutsotso and Gicheru 1995). Several
reasons for poor adoption of hedgerows are given, including damage by grazing animals
(Fujisaka and Cenas 1993, Mutsotso and Gicheru 1995), land tenure issues (Fujisaka et
al. 1995, Lawry et al. 1994), and competing farm and off-farm activities (Fujisaka 1993,
Ongprasert and Turkelboom 1995), but by far the most common reasons are the high
costs of labor for installation and management (David 1995. Fujisaka et al. 1995, Kass et
al. 1995, Mutsotso and Gicheru 1995, Nelson et al. 1997, Ongprasert and Turkelboom

17
1995, Stark 1996) and farmers’ need for short-term economic returns from their plots
(Fujisaka 1993, Ongprasert and Turkelboom 1995, Steiner and Scheidegger 1994).
Initial positive results from alley cropping trials were eventually modified by later
studies that showed decreased crop production from alley cropping in certain biophysical
settings or prohibitive costs of installation and management to farmers in certain
socioeconomic settings. The claims made for the potential of alley cropping, it was felt,
were exaggerated. But criticisms of alley cropping, summarized in a review written by
Sanchez (1995), have also been exaggerated due in part to its having been tested in
limited and inappropriate situations (Nair 1998). and may have caused an unwarranted
backlash among researchers (Vandermeer 1998).
Rao et al. (1998) reviewed 29 hedgerow intercropping trials that had been
conducted for four or more years, mostly on small plots, over a wide range of soils and
climates, but no studies done on sloping land were included. They found both positive
(15 studies indicated positive results for cereals, 8 trials showed positive results for
noncereal crops) and negative (13 trials showed negative results for cereals, 1 showed
negative results for noncereal crops) effects of hedgerow intercropping on crop yields
across the tropics. Including only the studies showing crop yield increases in hedgerow
systems greater than 15% (assuming this would be a minimum level to interest farmers),
positive effects on crop yield were found in only 2 of 10 studies in semiarid (<1000 mm
rainfall) regions, and 7 of 11 studies in subhumid (1000 to 1600 mm) regions where soils
were either inherently fertile or supplemented. In 4 out of 8 trials in the humid tropics
(>2000 mm) maize and taro (Colocasia escalenta (L.)Schott) did not benefit from
hedgerow intercropping, but bean and cowpea yields did. The authors conclude that the

18
hedgerow intercropping system is location specific and sensitive to management, and that
generalizations are difficult to make.
One can understand the caveats attached to prescribing or proscribing hedgerow
intercropping as a general solution; however, a more balanced approach to both research
and extension of hedgerow intercropping is now emerging. There is a growing
recognition that its success or failure depends on many site-specific factors including the
type of crops grown, the species of plants used as the structural component of the
hedgerows, rainfall amount and distribution, soil type, and a host of socioeconomic
variables (land and tree tenure, labor costs, etc.), including equity issues, extension
approaches, and allowance for adaptation of the practice over time to optimize its
economic returns to farmers (Mercer and Miller 1998, Nair 1998, Rao et al. 1998.
Sanchez 1995, Scherr and Current 1997).
Developments of Hedgerow Intercropping in Haiti
Hedgerow intercropping using Lencaena leucocephala (Lam.) de Wit was tried
on a small scale in Haiti for the first time in the early 1980s, but there are earlier
examples of vegetation-based soil conservation structures. Mintz (1962) notes that
Haitian farmers have been building seasonal crop stover barriers staked roughly on the
contour to harvest water and soil since the 1950s. These structures, known as ranp pay in
Creole, are still being built in some parts of the country. They last only a single planting
season. Some farmers say that stover barriers provide habitat for rats and insects, and so
they prefer to bum the crop residue. Contour structures have been promoted by
development projects since the 1950s. A UNESCO project in the community of Marbial
paid people to plant sisal (Agave sisalana Perrine ex Engelm.) on the contour to control
soil erosion in the early 1950s. and a FAO project near the town of Les Cayes in the late

19
1960s promoted contour hedges of napier grass (Penisetum purpure am K. Schumach.).
lemon grass (Cymbopogon citratus (DC.) Stapf), vetiver (Vetiveria zizanioides (Linn.)
Nash), and guinea grass (Panicum maximum Jacq.) (G. Brice, personal communication:
1998). These were done by farmers who were paid stipends based on the number of
hedges installed, and they were not always appropriately designed. The napier grass
hedgerows were planted on shallow, droughty soils. Many eventually grew taller than
their shallow roots could support, and toppled over. This also happened to some of the
sisal, a crop that is traditionally planted up and down the slope so that farmers harvesting
the sharply pointed leaves do not injure themselves if they fall.
In contrast to the historically widespread nonadoption of soil conservation
strategies designed by technicians, there is an example of a successful practice designed
by Haitian farmers responding to a particular market opportunity. This involves the
yearly, labor-intensive construction of contour soil bunds (tram) to conserve expensive
chemical fertilizers used to produce high-value vegetables in a small region close to the
capital city (Murray 1980).
Initial interest in tree-based hedgerow intercropping was stimulated by reports of
developments in West Africa, Indonesia, and the Philippines, and by the general interest
in the use of Leucaena leucocephala as a fuelwood tree. There is a variety of Leucaena
native to Haiti, known locally as delen, oriman, ti movye, or ti pingi. This variety is not
suited to hedgerow construction because it is extremely invasive and does not produce
sufficient biomass, and is therefore very unpopular among farmers. To promote the use
of Leucaena, USAID introduced seeds of productive varieties to Haiti in 1978 (Benge
1985). Seeds of Leucaena K8, K28, and K67 from trees grown in the Philippines were
distributed, and later seeds from Flores, Indonesia were introduced. In 1979, a USAID

20
agronomist also introduced the concept of using Leucciena in hedgerows, and re-wrote an
existing paper on Leucaena hedgerows especially for Haiti (M. Benge, personal
communication: 1995).
The number of hedgerows reported by PADF as built by farmers participating in
PLUS and its two predecessor projects probably represents the largest percent of the
hedgerows in Haiti. During the seven-year period from 1985 through 1991. 848 km of
hedgerows were reported (PADF 1990). The number of hedgerows installed has been
much greater under the PLUS project, with over 12,000 km. having been installed from
1993 through mid-1999 (PADF 1999). If the number of hedgerows ever installed by
farmers working with all aid projects in Haiti is estimated at 15,000 lineal km, and if
there is an average of 5 meters between rows, then approximately 7,500 hectares of
hillside gardens have had hedgerows. Although this represents less than one-half of a
percent of the total land area, it is a significant figure for a new technique whose
management is still being developed.
Hedgerows have been recommended as a potentially appropriate agroforestry
technology and are commonly promoted in Haiti as part of project or government-based
soil conservation and natural resource strategies (Bannister and Nair 1990). Garrity and
Van Noordwijk (1995) note three paradigm shifts in hillside conservation management:
(1) an engineering approach using rock and earthworks in contour structures has
increasingly given way to vegetative structures; (2) top-down prescriptive watershed
management has evolved to a bottom-up, participative approach; and (3) tree-based alley
cropping has diversified to include a wider array of economically interesting plant
materials as structural components of contour hedgerows. All three of these shifts have
occurred in Haiti. Early attempts at erosion control promoted by projects in Haiti were

21
mainly based on rock walls constructed across the contour, but these were poorly
maintained because they did not respond to farmers’ economic situation and the
extension approach was inappropriate (Murray 1980). The increasing emphasis on
vegetative barriers happened during the 1980s, as noted above, along with as-yet
incomplete attempts to manage natural resources participatively.
However, the ideotype of contour hedgerows is not static in Haiti. Fanners
continually modify the structural composition and management of hedgerows according
to their household needs, their ability to invest resources, their experience with
hedgerows, and changing marketing opportunities for crops and products derived from
hedgerow trees. Leucaena lencocephala was originally the predominant hedgerow
species, but that is changing. During 1995, PADF reported hedgerows made with 22
different tree, perennial food crop, grasses, and other perennial species planted in 54
different combinations. Leucaena was present in only 46% of the hedgerows. In
addition, hedgerows are evolving into perennial crop contour bands in some areas. These
bands, called banrt manje (a play on words—either “band of food” or “a lot of food”)
consist of rows of food crops planted across the contour, with a dimension up and down
the slope of more than one meter. Perennial crops such as pineapple {Ananas comosus
(L.) Merr.) or sugar cane {Sacharían officinarum L.) serve as the structural components,
but annuals such as yam and sweet potato {Lpomoea batatas (L.) Lam.) are also planted
within the band. Field crops are planted in the interspaces, as in hedgerows. During
1999. 20 different combinations of nine species of perennial food crops, trees, and other
perennials were used in crop contour bands (PADF 1999). This evolution is based on
farmers' needs for short-term cash return. The structural crops, such as pineapple,
respond directly to a marketing opportunity—a quintessential^ Haitian strategy (Murray

22
1991), and one found in indigenous agro forestry associations in other cultures (Khaleque
and Gold 1993). This reinforces the notion that hedgerow intercropping should be
considered as a seed technology (Sumberg and Atta-Krah 1988. Wiersum 1994). to be
adapted and modified by farmers as a matter of course to fit into their unique agricultural
systems as they discover (or not) its as-yet unknown economic possibilities.
The cycle of enthusiastic promotion of hedgerow intercropping followed by
disappointment has a parallel in Haiti, although it is not documented apart from some
protest about the widespread promotion of Leucaena hedgerows in the early 1990s
(Swanson et al. 1993d. 1993e). It mainly takes the form of disappointment, informally
expressed by officials of donor agencies, that hedgerows do not yet cover every garden in
Haiti. More often the criticism of nonadoption has been aimed at the extension
methodology rather than the practice per se (Sieder 1994,Villanueva 1993). This has
been due, in part, to the development-driven agenda mentioned by Nair (1998) resulting
in the promotion of large numbers of an untested practice by donor-funded projects.
Reasons for farmer adoption, adaptation, or rejection of hedgerows are not well studied
in Haiti. The best available studies were done in a small region of the central plateau
near the town of Maissade (White 1992b), which noted farmers’ reasons for nonadoption
of hedgerows as lack of time (39%), don’t own land (21%). and land not appropriate
(17%). Vaval (1997) studied the same project area later, and documented the
disadvantages of hedgerows as expressed by adopting farmers as the time required for
management and loss of crop land due to the presence of hedgerows.
Hedgerow Intercropping and Soil Conservation
Because of the greater risk of soil erosion, performance of hedgerows on sloping
lands is not identical to their performance on flat land. Three kinds of biophysical

23
differences to consider include (1) the overall effect of hillside contour hedgerows on
crop production and soil erosion. (2) the gradients of soil accumulation/loss, soil physical
properties, tree/crop competition, and crop yield across the overall slope from the top
hedgerow to the bottom one, and (3) the same gradients within the alleys.
The concerns about the poor performance of alley cropping summarized by
Sanchez (1995) did not extend to hedgerow intercropping on sloping land. In part, this is
due to the proven ability of contour hedgerows to control soil erosion. Sanchez (1995)
stated that the hypothesis that contour hedgerows can substantially reduce soil erosion
has been definitively proven. This has been shown in many areas of the world, for
example a review of ten years of research in Thailand concluded that hedgerows
controlled erosion in areas of 18 to 55% slopes (Ongprasert and Turkelboom 1995); in
Rwanda on 23% slopes hedgerows reduced erosion to less than 2% of the runoff from
unprotected plots (Roose and Ndayizigiye 1993); in Malawi on 44% slopes hedgerows
reduced soil erosion to 2 t/ha compared to 80 t/ha on maize-only plots (Banda et al.
1994); in India on 9% slopes hedgerows significantly reduced soil erosion compared to
cassava-only plots or bare fallow (Ghosh et al. 1989); and in the Philippines there are
many examples of hedgerows substantially reducing erosion on sloped land (Agus et al.
1999, Comia et al. 1994, Paningbatan et al. 1995. Tacio 1993). The studies cited cover a
range of soil types from sandy loams (Banda et al. 1994) to gravelly sandy clays (Ghosh
et al. 1989) to heavy clays (Comia et al. 1994).
However, in spite of their proven ability to control erosion, in many cases
researchers have not been able to show crop yield improvements in hedgerow plots
compared to traditional systems. The principal reason for this appears to be the same on
slopes as it is in flat land alley cropping systems-the tradeoff between the production of

24
biomass for mulch or green manure needed to increase crop yield, and the loss of
cropping space caused by the presence of the hedgerows that produce the biomass
(Vandermeer 1998). The inter-row spacing required for the crops does not allow enough
biomass production from the hedgerows to improve soil fertility, so maize yield, for
example, is not improved unless inorganic fertilizer is applied (Ongprasert and
Turkelboom 1995. Roose and Ndayizigive 1993). On steep slopes the Soil Changes
under Agroforestry (SCUAF) model (Young and Muraya 1990) predicts this might be
explained by high rates of leaching and organic matter decomposition, despite reductions
in the rate of erosion (Nelson et al. 1998). Crop yield declines are especially important in
newly established hedgerows as biomass application will not have yet contributed much
to soil improvement, although cropping space will have been reduced. Data taken from
89 hedgerow plots planted with maize on sloping plots in the Philippines showed a
decrease in maize yield in new hedgerows due to the loss of cropping space. The same
study estimated 8 years would be required for soil improvements caused by hedgerows to
compensate for loss in maize yield at a 5 m inter-row spacing (Shively 1998). In some
cases, however, soil conditions are improved by hedgerows sufficiently to maintain a
stable, if low, level of maize production on slopes compared to traditional cultivation,
where maize yield falls off over time (Banda et al. 1994, Comia et al. 1994). There are
also examples where crop yields in hedgerows on slopes were significantly greater and
more stable than those in unprotected control plots, as in a four-year study in Costa Rica
on farms having relatively fertile soils (Kass et al. 1995).
Gradients of soil water, soil physical properties, and crop yield have been
documented in alley cropping systems, as noted earlier in this chapter. For example,
infiltration rate and number of soil macropores were greater under the hedgerow trees

25
than in the middle of the alleys on a 14% slope in Kenya (Kiepe 1995). Competition for
light, water, and nutrients between hedgerow plants and adjacent crops is also noted on
slopes. Rice yield was depressed in the first two rows away from hedgerows on 15%
slopes in the Philippines, but the depression varied with different hedgerow species; it
was greatest near grass (Penisetum purpurenm) hedgerows (Garrity and Mercado 1994).
It has also been found that food crops in the alleys can restrict the growth of adjacent
hedgerow trees (Ghosh et al. 1989).
But as the slope increases a redistribution of soil fertility across the alley is often
seen that differs from that of flat plots (Garrity 1994, Ongprasert and Turkelboom 1995).
Increases in soil macropores near Calliandra callothrysus hedgerows in Reunion were
greater on the uphill side of the hedges than on the downhill side (Tassin et al. 1995). A
study done in the Philippines on 22 to 30% slopes found that average mid-alley maize
yields were greater than yields near hedgerows of Gliricidia sepium, Paspalum
conjugatum. and Penisetum purpurenm; and that the Ap soil horizon was thicker on the
lower part of the alleys (Agus et al. 1999). Garrity (1996) reviewed this phenomenon,
and noted that it was caused in a large part by the displacement of soil during crop
tillage. He has pointed out that on unprotected slopes there is often no net loss of soil
from the mid-slope area because any soil loss is replaced by soil eroding from the upper
portion of the hillside. The presence of hedgerows prevents a net accumulation of
upslope soil into the alleys, but the same displacement effect occurs within the alleys on
a smaller scale. Crop yield response on sloped hedgerow plots skews after a few years of
cultivation because scouring due to tillage displaces soil from the upper part of the alley
to the lower part, changing the soil profile, slope length, soil depth, and hydrological
properties. These changes combine with competitive interactions between the crop and
the hedgerows. The result is lower crop yields on the upper parts of the alleys and higher

26
yields on the lower middle part, with a slight decrease again as crops approach the low'er
alley near the next hedge. The net yield curve across the alley is due to a combination of
soil scouring and competition w'ith hedgerow' trees. The effect of soil scouring on yield
is usually more important than that of tree/crop competition. Solera (1993), cited in
Garrity (1996) put in 50 cm plastic barriers between Gliricidia sepiunt hedgerows and
adjacent upland rice on a 20% slope, and found that the across-alley yield response w'as
unchanged by the barriers, and so was not due to competition between the hedgerow's and
crop. Yield in the upper rows was 50% lower than that in the low'er rows within the
alleys. The effects of slope on hedgerows are important because in wide alleys the soil
displaced from the upper position may lead to long-lasting crop yield depression in that
area, and because invalid comparisons between yield in hedgerows are often made
against an adjacent no-hedgerow control plot in the mid-slope area that has thicker soil
(Garrity 1996).
Conclusions
Hedgerow intercropping is an evolving technology meant to address the
increasingly rapid destruction of soil resources in developing countries. The defining
paradox in its history is that researchers and extension agents initially addressed it as a
technical solution, whereas farmers looked at it primarily as a potential economic
solution. Crop yield improvements due to hedgerow intercropping in certain regions of
the world justify continuing research. However, it has not yet been widely adopted even
in those regions, mainly due to high initial costs of installation and adaptation and farmer
need for short-term returns in the face of the probability of a yield loss in hedgerow plots
during the first few years after installation. Both the technical and socioeconomic
aspects of hedgerow intercropping must be resolved in order for adaptation and adoption
to increase.

CHAPTER 4
ORGANIZATION OF THE STUDY
This study is presented in three parts, beginning with an on-station study of
competitive interactions between hedgerows and adjacent maize, followed first by an on-
farm study of the performance of maize in hedgerows on sloping plots, and then by an
extensive field survey of the biophysical characteristics of plots and socioeconomic
characteristics of farmers who have constructed hedgerows and other soil conservation
practices with the PADF/PLUS project. The hypotheses to be tested correspond to some
of the biophysical and socioeconomic hypotheses referred to by Sanchez (1995) as being
fundamental to agroforestry, namely, biophysical hypothesis number 16: competition for
water in agroforestry systems can be reduced by modifying the spatial arrangement of
trees; and socioeconomic hypotheses numbers 1, 3, and 4:
• the identification of key driving socioeconomic and ecological processes
within land-use typologies permits the spatial delineation of target and
recommendation domains for agroforestry interventions;
• the adoptability of a new agroforestry practice is determined by 5 principal
components-the farmers' natural resource base, their resource endowment,
degree of market integration, cultural preferences and perceived benefits; and
• at the farming systems scale, agroforestry adoption has different impacts for
different classes of farmers, such as their gender.
Justification
Declining farm productivity is a major problem confronting Haitian hillside
farmers. It is caused and exacerbated by a number of factors. Important among these are
soil degradation as a consequence of erosion, and farmers’ difficulty in making capital
27

28
investments to support improvement of lost soil fertility and farm productivity.
Traditional agroforestry systems such as home gardens, shade coffee, and living fences
have not been able to cope with the degradation of soil resources on hill slopes as
population increased. Although rural migration to the cities has steadily increased
(Locher 1996), there is no reason to suppose that this will result in a significant decrease
in the population density in the mountains. Haiti's population might already have
exceeded the ecological carrying capacity of the land (Ewel 1977). Two decades ago, the
Haitian landscape was described as the victim of mismanagement: "the ruthless
application by the rural population of ecologically inappropriate and catastrophic lumber
extraction and gardening techniques which have transformed the once lush landscape of
entire regions into grim, denuded savannas" (Murray 1980, page 3). In addition to
decreasing productivity of individual hillside plots, inappropriate farming practices also
led to other local and regional environmental problems, including transport of rock
overburden that covers productive soil of farms on lower slopes and plains, reduction in
the water supply from springs, road washouts, and destruction of houses by water and
debris flows.
Hedgerow intercropping has gone through a cycle of research, widespread
promotion based on expected positive performance, extension by governments and
funded development projects, mounting criticism when expected performance was not
achieved, a resulting waning interest on the part of researchers and projects, and a
concern that alley cropping might be rejected due to inadequate identification of the
biophysical and socioeconomic conditions where it might succeed. This point was
discussed in more depth in Chapter 3.

29
Consequent to the reported promise and potential of hedgerow intercropping
technology in other parts of the tropics in the 1980s (Kang et al. 1984). it was introduced
in Haiti in the middle 1980’s through externally funded development projects (Murray
1997. PADF 1986). As with most project-driven interventions, hedgerow intercropping
has had varying degrees of success. Several reasons could be attributed to this, both
biophysical and socioeconomic. They might include inappropriate extension approaches,
mismatch between farmer objectives and project objectives, inappropriate choice of
hedgerow species, competition between hedgerows and adjacent crops, high cost of
hedgerow establishment and management relative to the resulting effect on crop yield,
and farmers’ and technicians’ ignorance of the costs and benefits of hedgerows. But the
list of relevant factors contributing to hedgerow success or failure is poorly understood.
Indeed, farmers, extension agents, technicians, and project administrators have varying
definitions and perceptions of success and failure. The same cycle of widespread interest
followed by disappointment seen on the international level has been repeated on a
smaller scale in Haiti, where hedgerow intercropping has sometimes been labeled a
failure by consultants and donors who tend to observe the practice at an early stage of
development, in a small number of locations, or without a complete understanding of
important nontechnical parameters of the practice.
During the period 1984 to 1999, hedgerow intercropping technology has been
installed on approximately 10,000 hectares by farmers working with various funded
projects in Haiti (PADF 1990. PADF 1999). This is a significant coverage for a
development activity involving a relatively new and evolving technology. The rate of
hedgerow construction appears to have surpassed the level of understanding of how and
why they are installed and managed.

30
A number of studies have been published regarding various aspects of
hedgerow intercropping in Haiti. These include the effect of hedgerows on soil erosion
(Cunard and Lorcy 1990, Regis 1990), financial analysis of some of the benefits of
hedgerows (Bellerive 1991, Lea 1993. White and Quinn 1992). choice of hedgerow
species (Isaac and Shannon 1994, Shannon and Isaac 1993. Toussaint 1989. Treadwell
and Cunard 1992). management of hedgerow tree biomass (Isaac et al. 1995). farmer
adoption (Chéry 1990, Emmanuel 1990, Lea 1994a. 1994b. Villanueva 1993. White
1992a), and farmer adaptation and management (Lea 1995, Pierre et al. 1995. Swanson et
al. 1993a, 1993b, 1993c, 1993d, 1993e. White 1992b).
Of the nine papers cited above reporting adaptation and management of hedgerow
intercropping in Haiti, only the one by Pierre et al. (1995) was specifically designed to
study how farmers manage hedgerows. They studied at 105 hedgerow gardens installed
before March 1994 in one region of southwestern Haiti. Although the study reported a
number of valuable observations on the management and performance of hedgerows, it
did not correlate the management quality indicator (the number of unrepaired breaches)
with other factors. Furthermore, the observations were based only on hedgerows
constructed of Leucaena leucocephala in a limited area of project intervention.
Another major study on hedgerows in Haiti was the series of qualitative,
questionnaire-based diagnostic surveys (Swanson 1993, Swanson et al. 1993a. 1993b.
1993c, 1993d. 1993e,) done just after the first field season of the PLUS project at the
request of PADF and Cooperative for American Relief Everywhere (CARE), the
implementing organizations. These studies were targeted at the whole farming
environment, not particularly at hedgerows. The diagnostic team visited over 200
farmers in fifteen small watersheds, interviewing individual farmers in their plots and

31
speaking with groups of farmers. The majority of the hedgerow gardens seen by the team
were installed under the Agroforestry II (AFII) project (1990-1991). The number of
hedgerow gardens visited was not stated, and the survey did not attempt to link hedgerow
management to specific conditions on the farms. Several useful observations were made
regarding hedgerows, however both the composition and management of hedgerows have
evolved since that time, and the number of hedgerows installed by farmers has increased
dramatically. In addition, an evaluation of the PLUS project monitoring and evaluation
system done in 1995 (Romanoff et al. 1995) suggested that the sampling areas used were
not representative of the whole project. They recommended that project impact be
measured over the whole area of intervention.
Competition between hedgerow trees and adjacent crop plants for soil water has
been the subject of an increasing number of studies in recent years (Chirwa et al. 1994a,
Govindarajan et al. 1996, Korwar and Radder 1994, 1997, Ong et al. 1991, Rane et al.
1995, Schroth et al. 1995a, Schroth and Zech 1993, Ssekabembe et al. 1994). This is
potentially a problem in Haitian hedgerow gardens because of the highly variable timing
of seasonal rainfall and the inability of eroded soils on steeply sloping land to absorb and
store water. Ewel (1977) referred to this as pseudo-drought, a condition that results from
a decrease in the water-holding capacity of the soil rather than from decreased rainfall.
The literature on hedgerow-crop soil water competition shows that, under some
conditions, hedgerows compete to the disadvantage of adjacent crops (Fernandes et al.
1993), (Govindarajan et al. 1996. Kamasho 1994. Korwar and Radder 1994, Ong et al.
1991. Salazari et al. 1993, Singh et al. 1986) and under other conditions, they do not
(Bohringer and Leihner 1997, Chirwa et al. 1994a. Haggar and Beer 1993, Jeanes et al.
1996, Schroth and Zech 1993).

32
One possible influence on farmer adaptation and management of hedgerows is
farmers’ perception of the importance of soil water competition between the hedgerow
species and the interplanted crops. Experience in Haiti shows that maize (Zea mays) and
bean {Phaseolus vulgaris) crops, even without hedgerows, very often fail because of
drought conditions. Therefore, hedgerow trees are likely to compete with interplanted
crops for soil water at some time during the cropping cycle. The degree of competition
may also vary according to the position of the crop on the slope relative to the hedgerows
(Garrity 1996). However, it is not clear if farmers consider this competition as more or
less important than other perceived problems caused by hedgerows. Do farmers modify
hedgerow structure and management to compensate for perceived soil water
competition? Neither competition for soil water between crops and hedgerows, nor crop
yield relative to slope position and /or relative to hedgerows have been studied in Haiti.
Since intensive agriculture is practiced on small steeply sloping plots, it is
necessary to look for economically rational ways to maintain the soil’s ability to sustain
crop production and to protect the local environment. Development of new agroforestry
systems that build upon traditional practices is needed. Hedgerow intercropping has been
shown to produce desirable results under certain circumstances. However, adoption or
lack of adoption may depend on factors extraneous to the characteristics of the
technology itself, including the socioeconomic status of farm families and the kinds of
land they access for agriculture. The extremely degraded condition of Haitian natural
resources makes it necessary to try any practice that could improve the economic
condition of Haitian farmers. Because of experience gained from agroforestry projects
operating in Haiti since the 1980s, there is reason to believe that adaptations of hedgerow
intercropping may increase farm stability and productivity in some areas of the country.

This study will help inform decision makers regarding the potential for use of hedgerow
intercropping in Haiti, and possibly avoid premature decisions to discontinue the effort to
adapt the technology.
Thus, hedgerow intercropping, a rapidly evolving and yet unproven technology,
has been extended in Haiti without the necessary supporting studies and documentation.
Development and extension resources could be used more efficiently by responding to
tw'o documented shortcomings of agroforestry research that seem particularly appropriate
to the Haitian situation: that there is an inadequate understanding of below-ground
biophysical interactions in hedgerow systems; and that socioeconomic and biophysical
research must be linked, as adoption of the appropriate agroforestry practice depends
upon both (Rao et al. 1998).
Hypotheses
On-station Study
• Leucaena hedgerows reduce the amount of soil water available to adjacent
maize, thereby causing a reduction in maize growth and yield.
• Root barriers installed next to Leucaena hedgerows increase the amount of
soil water available to adjacent maize, thereby causing an increase in maize
growth and yield.
• Cutting the roots of Leucaena hedgerows affects the production of tree
biomass.
On-Farm Study
• The pattern of competition between hedgerow trees and adjacent crops is
different on slopes vs. flat land, and maize growth varies with regard to the
position of the crop w'ithin the alleys and on the slope.
On-Farm Survey
• Family characteristics influence the adoption and management of hedgerow
intercropping.

34
• Plot characteristics influence the adoption and management of hedgerow
intercropping.
• Problems and benefits with hedgerow intercropping as perceived by the
adopting farmer influence management quality.
General Description of Hvpothesis-Testine Procedures
The study of soil water competition between Leucaena hedgerows and adjacent
maize was done on an NGO-operated research station located near Port au Prince. Four
treatment combinations were tested in nine replications: hedgerows with and without root
barriers between the trees and the adjacent row of maize, and maize fertilized or not
fertilized. Data for three maize seasons are reported, including gravimetric soil water:
maize height, leaf width, growth stage, leaf water potential, and final harvest: and
Leucaena height and biomass production. A pre-study examination of the distribution of
Leucaena hedgerow roots at a different location was done, and a post study count of the
distribution of Leucaena roots in hedgerows with and without barriers was done on-
station. A pot study of the correspondence between soil water percent and maize leaf
water potential was done as well.
The small on-farm study was an attempt to relate the on-station findings on flat
land under “controlled” conditions to sloping land managed by farmers. The study
collected maize growth and yield data for one season from four farms having previously
established trials of Leucaena hedgerows, crop contour bands, rock wall terraces, and
traditional cultivation. Maize growth data were collected by position in the alleys and on
the slopes. The four farm plots had been a part of an abandoned PLUS project research
effort previously carried out by the Southeast Consortium for International Development
(SECID)/Auburn University in conjunction with PADF. The author designed the data
collection protocol and performed the analysis.

35
A large list-frame questionnaire-based study was conducted, which provided on-
farm data about households adopting agroforestry soil conservation practices (including
hedgerows), how they were managed, and the characteristics of the plots where they were
installed. The study was undertaken by the author with the help of PADF PLUS
technicians as part of project monitoring during the early part of 1996. We visited 1.540
PLUS project agroforestry adopters and collected information regarding all plots
controlled by the family, and information regarding family members. We then visited
2.295 plots where farmers had installed at least one PLUS project agroforestry practice.
Data were collected on the biophysical parameters of the plots, the amount and condition
of the agroforestry practices installed on the plot, and the opinions of the farmers
regarding problems and benefits of the practices. Because farmers" assessment of soil
fertility appeared to influence where hedgerows were installed, soil samples were
collected from 175 farms. These were analyzed for percent organic carbon and soil
nutrients, and compared to farmers' classification of soil fertility.
Details of data collection and analysis for the on-station and on-farm
hedgerow/crop competition studies are reported in Chapter 5. Those for the on-farm
survey are reported in Chapter 6.

CHAPTER 5
HEDGEROW/CROP COMPETITION
Introduction
This chapter looks at competitive interactions between hedgerows and adjacent
maize on station and on farm. Competition is important because it is a key consideration
in designing agroforestry practices, it is difficult to measure, and farmers make adoption
and adaptation decisions based on its perceived severity. Competition is a central issue
in agroforestry because of the potential advantage in combining plants in such a way that
they capture more growth resources in combination than they would growing separately,
thereby producing greater physical yields (Cannell et al. 1996). The cost of the positive
effects of facilitation gained by growing plants in close proximity is the negative effects
of competition. In hedgerow intercropping in particular, the competitive characteristics
of the trees (they take up space, light, soil water, nutrients) must be balanced with their
facilitative characteristics (they impede soil erosion, capture soil water and nutrients out
of reach of the crops, improve soil fertility through biomass additions, improve the
microclimate, improve soil physical properties) (Vandermeer 1998). It is hard to develop
a hedgerow system that attains that balance because so many combinations of plants,
physical environments, and socioeconomic situations are possible. It is also difficult to
measure competition and to identify the processes that contribute to it because
competitive interactions change as the environment changes (Anderson and Sinclair
1993), and the hedgerows themselves change the environment—indeed, that is their
36

37
function. The way hedgerows and adjacent crops interact varies according to the age of
the hedgerows, relative size of the crops and hedgerows, population density, and the
supply of and access to limiting growth resources (Ong and Leakey 1999).
Farmers make adoption decisions based on competition between hedgerows and
crops. For example, farmers in the Philippines eliminated napier grass as a hedgerow
component because it was too competitive with crops grown in the alleys (Fujisaka et al.
1995), where it decreased maize yield by 86% by the second year after establishment
(Garrity and Mercado 1994). Therefore, competitive characteristics are important
considerations in the design of hedgerow systems.
Leucaena has characteristics that make it an aggressive competitor for growth
resources, including rapid growth and roots that occupy approximately the same soil
depth as that of maize (Jonsson et al. 1988), the crop of interest in this research. It was
noted several decades ago in Southeast Asia that Leucaena used as shade trees in
plantations rarely competed with the crop plants; but did so in times of drought, when
signs of water competition with coffee and rubber were reported (Dijkman 1950). In
more recent alley cropping studies, competition is a frequent subject of investigation, but
results vary with the specific site and plant components. Leucaena has been shown to
compete for soil nutrients to the detriment of rice (Salazari et al. 1993) and grasses
(George et al. 1996). In the semiarid tropics especially it has been shown to reduce the
amount of soil water available to pigeon pea, but less so for sorghum and millet
(Pennisetum glaucum L.) (Singh et al. 1986). In other areas of the semiarid tropics,
however, yields of both sorghum (Korwar and Radder 1997) and millet (Ong et al. 1991)
were significantly reduced in Leucaena hedgerows.

The on-station and on-farm studies reported on here both have maize planted in
the interspaces between Leucaena hedgerows. This combination has produced both
positive (Akonde et al. 1996, Banda et al. 1994, Hauser and Kang 1993, Hernandez 1961,
Mugendi et al. 1999, Mureithi et al. 1994) and negative (Chirwa et al. 1994b.
Govindarajan et al. 1996, Jama et al. 1995, Kamasho 1994) results. The positive results
in the cited studies ranged from maintaining the existing maize yield (versus a decrease
without hedgerows) to increases of 26%. 44%, 50%, and 400% in grain yield compared
to traditional cultivation. Positive results tended to be from subhumid or humid areas
with soils of moderate or better fertility, and from sloping areas. The negative results
ranged from 16% to 20% decreases in maize yield. These tended to be from semiarid,
moisture limiting, low fertility sites.
Effects of competition between maize and Leucaena are noted in several studies,
even where the overall effects of hedgerows are positive. In an early report of alley
cropping research in Ghana, Balasubramanian (1983) mentioned that soil water
competition with Leucaena might be a partial explanation of yield depression in adjacent
maize. Later hedgerow studies have attributed yield depression in adjacent crops to
competition for soil water (Govindarajan et al. 1996, Korwar and Radder 1997, Ong et al.
1991. Ssekabembe et al. 1994), nutrients (Chirwa et al. 1994b, Livesley et al. 1997,
Mureithi et al. 1994), or light (Kang et al. 1981, Lai 1989, Tilander et al. 1995, Welke et
al. 1995. Zoschke et al. 1990). Some studies claim all three resources are competed for
(Gaddanakeri et al. 1994). It is difficult to use these location-specific results to make
predictions about the probable success of a hedgerow intercropping system in another
location. However. Rao et al. (1998) have drawn some general conclusions: (1) In water
limiting areas, low yield of hedgerow prunings and competition for water between
hedgerows and adjacent crops were the major reasons for the negative results.

39
Inadequate water limited the response of crops even though hedgerows improved soil
fertility in certain sites of the semiarid tropics. (2) In poor soils, low yield of hedgerow
prunings and competition for nutrients were responsible for negative results. (3) The
advantages of hedgerow intercropping are more common on relatively fertile. N-deficient
soils in subhumid and humid environments, in areas where hedgerows produce a large
quantity of prunings, and where there is adequate water for both hedge and crop growth.
They also note that if competition is substantially reduced, even in semiarid climates,
hedgerows may be able to increase crop yields even where relatively small fertility
contributions are made by the hedges—but the cost of managing the hedges could be too
high to be practical.
Much of Haiti is subhumid, and there are still soils having a little remaining
fertility, on sloping land where hedgerows may be appropriate. The discussion to be
presented in this chapter will focus on the biophysical aspects of hedgerows and so will
necessarily be incomplete, lacking the socioeconomic elements, which will be discussed
in Chapter 6.
On-Station Studies: Effect of Root Barriers on the Growth of Maize and Leucaena
The objectives of this experiment were to test whether or not permanent root
barriers placed on either side of Leucaena leucocephala hedgerows affect the production
of maize grown in the alleys and the production of Leucaena leucocephala biomass. The
hypotheses were that the barriers will reduce competition for soil water between
Leucaena and adjacent maize, thereby increasing maize yield; and that production of
Leucaena will not be adversely affected by the barriers. If maize yield were to be
increased significantly an appropriate modification of the barrier could then be devised
for use on-farm, or a less competitive tree could be sought to construct hedgerows.

40
Crop yield depression close to trees is common. It has been reported for mustard
growing under Acacia nilotica in semiarid central India, where a yield depression of 54%
extending up to 8 m from the tree was correlated with increasing shallowness of the tree
roots (Yadav et al. 1993). Yield depression of maize and soybean near Siberian elm
(Ulnius pumila) and eastern cottonwood (Populus deltoides) windbreaks in Nebraska was
significantly relieved by plowing parallel to the trees (Rasmussen and Shapiro 1989). A
study done in the savanna region of Kenya reported that Acacia tortilis trees competed
more intensely with understory grasses and herbs in wetter sites where the tree roots
terminated near the crown region, than in drier sites where the roots extended farther into
the grassland (Belsky 1994). Yield depression in millet extended significantly beyond
the area shaded by eucalyptus shelterbelts in Nigeria, corresponding to the area of tree
root growth (Ojyewotu et al. 1994). Nadagouda et al. (1994) found that severing the
roots of eucalyptus trees in India improved yield of adjacent maize.
Unless effective experimental controls are put into place, it is difficult to separate
the effects of above- from below-ground competition, and the effects of competition for
water from those of nutrients. Corlett et al. (1992b) describe a study done in the semiarid
tropics of India in which millet was grown alone and in Leucaena hedgerows, either with
or without barriers between the tree and crop roots. Millet yield was the same in the sole
crop and in the hedgerows with root barriers, but was 36% less in the hedgerows without
barriers. The authors proposed that, although one might have concluded that the yield
difference was due solely to below-ground competition, a better explanation was that the
millet adjacent to hedges with barriers grew taller because it had access to more soil
water than millet in hedgerows without barriers, and avoided light competition with the
Leucaena. There was an interaction between above- and below-ground processes.

41
Common methods used for studying the effects of below-ground competition in
alley cropping research include root pruning (Fernandes et al. 1993, Korwar and Radder
1994, Sitompul et al. 1992), trenching (Schroth et al. 1995b) or the installation of
permanent plastic barriers between the hedgerow trees and the crops (Corlett et al. 1992b.
Hauser and Gichuru 1994, Ong et al. 1991, Schroth et al. 1995a, Schroth and Lehmann
1995, Schroth and Zech 1993, Solera 1993c, Ssekabembe et al. 1994). The usual practice
is to open a narrow trench to a depth of 50 to 100 cm between the hedgerow trees and the
first crop row, install plastic awning or polyethylene sheeting, and then fill in the trench.
Studies using plastic barriers have produced results specific to the characteristics
of the site, species of hedgerow tree and crop, alley width, planting density and distance
between the hedgerow and crop, and the top-pruning regime (height and timing of cut)
used on the hedgerow. Ong et al. (1991) showed that yield of millet in hedgerows
without barriers was 40% less than yield in sole-cropped millet in semiarid India, but
where barriers were parallel to hedgerows the yield was the same as that in sole cropping.
They also showed that the barriers did not affect the production of the Leucaena
hedgerows, but the barriers did cause a redistribution of the Lencaena roots. Schroth and
Lehman (1995) found in subhumid Togo that barriers increased the yield of maize in
Calliandra callothyrsus hedgerows, but when the hedgerow species were Senna siamea
or Gliricidia sepium the presence of barriers was associated with a decrease in maize
yield. They concluded that Calliandra was relatively competitive compared to the other
trees, and that the barriers reduced the volume of exploitable soil available to the roots of
the other two species. The results obtained by studies using plastic barriers between trees
and crops do not always show competition for soil water. Two studies done in the West
African humid tropics put plastic barriers to 90 cm between Gliricidia hedgerows and

42
adjacent crops of rice, maize, and peanut. They showed crop yield improvement in
hedgerows compared to the sole cropping controls, but crop yield in hedgerows was not
affected by the presence of barriers (Schroth et al. 1995a, Schroth and Zech 1993). A
study done in the United States was able to isolate the effect of Robinia pseudoacacia
hedgerows on soil water at two sites having different soil characteristics (Ssekabembe et
al. 1994). The plots were irrigated, no crops were grown in the alleys between the
hedges, and the soil was covered with plastic to reduce evaporation. At the site having a
higher soil organic carbon content, hedgerows without barriers reduced soil water content
of the alleys by about 8%. At the second site, where higher gravel content impeded the
growth of tree roots, hedgerows without barriers reduced soil water content by 32%. and
there was significantly less water in alleys without barriers.
Materials and Methods
The experiment was carried out over a period of six years, from the spring of
1990 through the spring of 1995. During this period data collection was interrupted
several times by civil unrest in Haiti,1 so although six maize crops and one sorghum crop
were planted only three successive maize crops from spring 1994 though spring 1995 are
1 There was a coup d'etat in September 1991, which resulted in several missed
observations due to curfews. A night guard at the Operation Double Harvest (ODH) site,
where the experiment was installed, was killed by soldiers during that period, preventing
pre-dawn leaf moisture potential measurements. I was evacuated to the U.S. in October
1991 and returned to Haiti in September of the following year. By that time, the
Direction of ODH had changed hands, requiring that I obtain permission to continue my
work on the plot. This took several months, and the removal of a leaking irrigation line
installed next to the plots was necessary. During the regime of the de facto military
government (September 1991 to September 1994). fuel shortages were common because
of an international embargo, again interrupting data collection. Farmers with friends in
the local police station let donkeys into my plots, ruining the Fall 1993 harvest. Farm
workers bathing in the plots behind the screen provided by the Leucaena also prevented
soil moisture observations from being taken at least twice. The U.S military intervention
in September 1994 made adequate data collection difficult during that period as well.

43
discussed. Data available from the three maize crops were not identical, again due to
civil unrest and equipment problems. The spring 1991 data are used only for the
discussion of the effect of trenching on Leucaena height and biomass.
Trial site
The trial site was located at 18°33' north latitude 72°11 ’ west longitude, with an
elevation of 80 m above sea level on property owned by Operation Double Harvest
(ODH), a nongovernmental organization (NGO) operating a farm and tree nursery in the
Cul de Sac plain, 30 minutes northeast of Port au Prince near the town of Croix des
Bouquets, Haiti. The soil was a vertisol of zero slope. From observations made in a soil
pit near the trial plots, heavy texture predominated to a depth of 1.1 m, where there was
an abrupt change to sandy, unconsolidated limestone to at least 1.7 m, the depth of the
pit. Soil texture from the surface to 30 cm depth was silty clay, changing to silty clay
loam in the 30-45 cm depth. Table 5-1 shows texture and organic carbon content for the
trial site.
Table 5-1: Soil texture and organic carbon at the trial site.
Depth
% Sand
% Silt
% Clav
% Organic
Carbon
Texture
0-15 cm
8.4
49
42.6
1.71
silty clay
15-30 cm
7.2
41
51.8
1.10
silty clay
30-45 cm
7.2
53
39.8
1.0
silty clay loam
A pot study was established in January 1994 to determine how soil water
depletion at the ODH site correlated with plant water stress (Appendix A). When soil
water percent in the root zone decreased below about 25%, leaf water pressure passed

44
below 10 bars. Maize plants began to exhibit signs of drought stress at this point. Field
capacity at the site was reached at 41% soil water content.
Average annual rainfall based on data collected by ODH staff since 1977 was 968
mm per year. Distribution is bimodal, with most of the spring rains falling from April
through June, and the fall rains from August through November. The total annual rainfall
since 1977 varied from a low of 670 mm to a high of 1.301 mm. but in most years
rainfall was at least 900 mm. The trial site was located to the north of a small-container
tree nursery, and was previously used for tractor cultivation of hybrid sorghum. The area
to the east was used for composting sugarcane bagasse. A low earth barrier separated the
trial area from the composting area, to prevent water used for composting from entering
the trial.
Trial design
The trial design was a 2 by 2 factorial randomized complete block having 11
hedgerows of Leucaena leucocephala variety K636 planted at a within-row density of 10
trees per meter and a between row spacing of 5 meters. Each block consisted of a
hedgerow 22 meters long running through the center of four maize plots, to which were
randomly assigned the four treatment combinations. The hedgerows were laid out
exactly on an east-west axis, with hedgerow (block) 1 on the north end of the site and
hedgerow (block) 11 on the south side (Figure 5-1). The symbol T+ denotes a trenched
plot. T—a nontrenched plot; the symbol F+ denotes a fertilized plot. F—a nonfertilized
plot. All plots received the designated treatments throughout the three seasons of the
study. Figure 5-2 shows the layout of one of the plots within a block.
Each plot was centered on the hedgerow, and included four rows of maize on the
north side and four on the south side at 50, 100, 150, and 200 cm distances from the

45
H
BLOCK
1
T-F-
2 m buffer
I
T+F+
22 m
T+F-
â–º
\ 3^
T-F+
â–º 1
t
4.5 m
2
T+F+
T-F-
T-F+
1
5
J
f
T+F-
3
T-F+
T+F-
T+F+
T-F-
T+F-
T-F-
T-F+
T+F+
Hedgerow
5
T+F+
T-F-
T+F-
T-F+
6
T+F-
T-F+
T+F+
T-F-
7
T+F+
T-F+
T-F-
T+F-
8
T+F-
T-F+
T+F+
T-F-
9
T+F-
T-F+
T-F-
T+F+
10
T-F+
T+F+
T+F-
T-F-
11
T+F-
T-F-
T+F+
T-F+
Figure 5-1: Layout of the Leucaena/maize hedgerow trial on Operation
Double Harvest property near Croix des Bouquets, Haiti

46
h 3 m â–º!
Figure 5-2: Diagram of one plot showing relative positions of a
Lencaena hedgerow and the adjacent rows of maize at four
distances.

47
hedgerow, respectively. Buffer rows of maize were planted at 250 cm from the
hedgerows, marking the middle of the alley between hedgerows. Two-meter buffers
separated the plots within the blocks. Rows 1 and 11 were borders from which no data
were collected. Hedgerow experiments have been criticized for producing invalid results
because the no-hedgerow controls (sole crop plot) have been exploited by hedgerow roots
growing into adjacent plots, thereby over-estimating the benefits of hedgerow
intercropping (Hauser and Gichuru 1994). Some more recent studies have compensated
for this by digging trenches between hedgerow plots and control plots (Chamshama et al.
1998. Mugendi et al. 1999). Although the study reported here does not have sole crop
plots, and a two-meter buffer was left between plots, it is possible that Leucaena roots in
plots both with and without barriers exploited the soil in adjacent plots. This could not
be confirmed in this experiment.
Treatment installation
The hedgerow trees, grown in the ODH nursery, were planted out in April 1990
as one-year-old seedlings. The trees were cut to 50 cm stumps on 23 March 1991. In
April 1991, two randomly selected plots in each block were trenched to a depth of 30 cm
at a distance of 20 cm from each side of the hedgerows. The trenching was done with
picks, severing all roots. Plastic sheeting was put into the trenches to inhibit root re¬
growth; the small canals created during root cutting were backfilled with soil. Trenching
extended for one meter beyond each side of the net plot boundaries, halfway into the two-
meter buffer between each plot. The decision to cut the roots to a depth of 30 cm was
based on a preliminary study of Leucaenct roots done at a location off-station
(Appendix B). That study indicated 70% to 96% of the very fine roots (< 1 mm
diameter) were found in the top 30 cm of the soil, the lowest percent being found 150 cm

48
and 200 cm uphill from the hedgerow. A lesser percent of the fine roots (1-5 mm
diameter) were found in the top 30 cm (20 to 83%), with the least (20%) at the 200 cm
uphill position. Very few medium and large roots were found, but of those that were.
50% or more were deeper than 30 cm. Studies in semiarid India also found the majority
of Leucaena hedgerow root biomass in the top 30 cm of soil (Ong et al. 1991. Singh et al.
1996).
Due to my forced absence after the spring 1991 maize season the Leucaena
hedgerows in the main experiment were not top pruned between October 1991 and
September 1993. By that time, the hedgerow trees had grown to basal diameters of up to
10 cm. They were then cut back to a 50 cm stump height. Because the Spring 1994
maize harvest indicated the effect of trenching had almost disappeared, the plots were re¬
trenched to a depth of 50 cm in August 1994 before the fall maize planting. Plastic
barriers were buried as before. The plastic barriers from the first root pruning were still
in place, but were missing or stiff and degraded in spots. In some places, soil had
covered the top of the barriers during cultivation, and Leucaena roots had grown over the
barriers into the alleys.
A selected local maize variety, known as chicken com, was planted between the
hedgerows. The within-row spacing for maize hills was 75 cm, with 50 cm between
rows. Hill positions were laid out precisely using nylon cord with flagging tied at 75 cm
intervals. Maize was planted by hand three seeds per hill, and later thinned to two plants
per hill. The maize density of the interspaces at full stocking was then 53,333 plants per
hectare. If the space occupied by the hedgerows is included, maize plant density was
48,000 plants per hectare. Observations of 28.000 and 40,000 maize plants per hectare
were made on two farms in the south of Haiti at harvest time (after substantial mortality

49
during the season) in plots without hedgerows. Farmer interviews in one region of
northern Haiti indicated a higher density at planting, 62,500 plants per hectare (Swanson
et al. 1993a).
The fertilizer was to have been applied according to recommendations of the
Ministry of Agriculture, Haiti (MARNDR 1990): an initial broadcast application of 15-
15-15 (NPK) of 175 kg/ha at seeding followed by two applications of banded urea at 50
kg per hectare. In practice, however, due to the timing of rainfall and political instability
delaying travel to the site, only the spring 1995 maize crop received the complete
fertilizer treatment. The spring 1994 crop received only the initial application of 15-15-
15, and the fall 1994 crop received the 15-15-15 and the first urea application. Once a
given plot had been designated as fertilized, it remained a fertilized plot throughout the
three seasons reported here.
Operations bv maize season
The spring 1994 maize crop was planted on 30 April. There had been 51 mm of
rain during the previous 7 days, 93 mm during the previous 14 days. Most of the rain
during the growing season fell between 11 and 21 DAS, with only one substantial rain
event after that, 23 mm at 67 DAS. A total of 206 mm of rain fell between sowing and
final harvest. The stand was thinned to two plants per hill at 28 DAS, and was weeded
once at 16 DAS. Insecticide was applied twice during the season. Fertilizer was applied
at a rate of 175 kg/ha at planting only; no urea was applied. Leucaeana top pruning was
done 35 days before sowing, then at 33 DAS (when trees were about 2.3 m high,
including the stump), and 65 DAS (when trees were about 1.8 m high). Figure 5-3 shows
the amount and timing of the rainfall in relation to the growth of Leucaena and maize,
and the timing of the agronomic interventions made to the plots during the spring 1994

50
season. The fall 1994 maize crop was planted on 10 September; two weeks after the
hedgerows were re-trenched to a depth of 50 cm (Figure 5-4). No measurable rain fell
during the 7 days preceding sowing, but 49 mm fell during the previous 14 days.
Tropical Storm Gordon passed over Haiti on 14 November (65 DAS) and dropped over
150 mm of rain in a 24-hour period on the trial site. This surpassed the capacity of the
rain gauge, so an accurate measurement was not recorded. This was the rainiest season
of the three, with at least (because of the inaccurate measurement during Tropical Storm
Gordon) 374 mm measured between sowing and final harvest. Most of the seeds had
germinated and had attained a height of 10 cm by 21 DAS, when they were weeded and
thinned to two plants per hill. Insecticide and urea (50 kg/ha) were applied at 26 DAS,
when the maize was about 25 cm tall. Because the effect of trenching was weak during
the previous season, the hedgerows were re-trenched to a 50 cm depth on 27 August
1994, 14 days before sowing the maize. Fewer measurements were made this season
than in others due to the U.S. military intervention in Haiti on 19 September, and the
events preceding and subsequent to it. Figure 5-4 shows the amount and timing of the
rainfall in relation to the growth of Leucaena and maize, and the timing of the agronomic
interventions made to the plots during the fall 1994 season.
The spring 1995 maize was planted initially on 26 March, but this planting failed
due to lack of rain. A new planting was done on 22 May. During the seven days before
sowing. 20 mm of rain had fallen, for a total of 39 mm during the 14 days before sowing.
After planting no rain was recorded until the 34,h day, then 171 mm fell between 34 and
73 DAS. However, because the technician responsible for recording rainfall was on
vacation between the planting date and 36 DAS, two or three rainfall events were not
recorded. No additional rain fell before final harvest. I applied 15-15-15 at sowing and

51
Figure 5-3: Rainfall. Leucaena and maize growth, and agronomic
inputs; spring 1994 maize cropping season
Figure 5-4: Rainfall, Leucaena and maize growth, and agronomic
inputs; fall 1994 maize cropping season

52
banded urea at 12 and 37 DAS. The maize was thinned to two plants per hill at 12 DAS.
The plots were weeded four times during the growing season; insecticide was applied
twice. Figure 5-5 shows the amount and timing of the rainfall in relation to the growth of
Leucaena and maize, and the timing of the agronomic interventions made to the plots
during the spring 1995 season.
Figure 5-5: Rainfall. Leucaena and maize growth, and agronomic
inputs; spring 1995 maize cropping season
Measurements recorded
The following observations and measurements were made during the three maize
cropping seasons:
Maize height. This was done only for the spring 1995 season. Height to the top
of the highest fully open leaf of the tallest plant in each hill in blocks 2, 4, 6, and 8 was
measured at 40 days after sowing. Analysis was done on the mean height of eight plants
in each of the four hills to the north and four hills to the south of the hedgerow at a given

53
distance in each plot. If all plants in a hill had died, a height of zero was assigned for that
hill.
Maize growth stage. Growth stages of the plants were noted at 56 DAS in the
spring 1994 season, and at 68 and 75 DAS in the spring 1995 season. In the spring 1994
season, the scale was: 0=no maize plant present. 1= vegetative growth. 2=tassel.
3=anthesis, 4=silk. For the spring 1995 season, the scale was changed to include the
presence of ears: 0=no maize plant present, l=vegetative growth, 2=tassel/anthesis,
3=silk. 4=ear. Observations were made in blocks 2, 3. 5, 6, 8, and 9 on every maize plant
(2 plants per hill) in the spring 1994 season, and on only the most developed plant per
hill in the spring 1995 season.
Maize biomass during the growing season. For each of the four distances and in
each of the four plots in blocks 4, 7, and 10, the combined weight of one hill of maize on
the north side and one hill on the south side of the hedgerow was recorded. The plants
were cut at ground level, and then dried in a Watlow drying oven at 70 degrees C until a
constant weight was obtained.
Maize leaf water pressure. Predawn leaf water pressure was taken using a PMS
pressure chamber for one leaf at each distance in blocks 3, 6, and 9. A 2 cm by 10 cm
section was cut from the highest fully expanded leaf, one end was placed in a zip-lock
bag, and the sample was inserted into the pressure chamber. Compressed nitrogen gas is
normally used in the chamber, but as this was not available in Haiti compressed air was
used instead. This measurement was to have been done periodically throughout each
maize season, but was done only once because of the security problems discussed in the
footnote on page 44.

Maize yield. Grain yield, number of plants, and number of ears were measured
from blocks 2. 3, 5, 6, 8, and 9 at 91 DAS for the spring 1994 harvest. 114 DAS for the
54
fall 1995 harvest, and 131 DAS for the spring 1995 harvest. At each of the four distances
(50, 100. 150, and 200 cm), plants were harvested from a 3-meter length on the north and
south sides of the hedgerows, for a total of 6 meters (8 hills, or 16 plants maximum)
harvested in each plot at each distance. Because the grain moisture meter ceased to
function correctly, field dry grain weights are shown in the tables and figures. Harvests
for all treatments were made at the same time, and all samples were weighed at the same
time. Final harvest measurements are converted to per hectare equivalents by dividing
the sample weight or count for a given distance from the hedgerows by six (six meters
per plot were sampled) and multiplying by 4,500 (each distance represented 4,500 meters
per hectare given a 50 cm row spacing). The per-hectare contributions of the rows at
each distance were then summed to give a total per hectare figure, which included the
2,000 m occupied by the hedgerows.
Soil water percent. Soil cores were taken using a JMC Back-saver corer. Each
core was divided into three sections. 0-15 cm, 15-30 cm. and 30-45 cm. Four cores for
each depth and distance were combined for each measurement. The samples were
weighed wet in the field and dried in an oven at 70 degrees C until a constant weight was
obtained. Soil water content was determined gravimetrically by dividing the weight of
water lost in drying by the dry weight of the sample. The range of soil water content that
could be sampled at this site was narrow due to the heavy texture of the vertisol. When
the soil was very dry, the 0-15 cm depth often was not retained in the sampling tube, and
when the soil was wet, the end of the tube plugged and did not allow soil to enter. For
the calculation of field capacity, a large-diameter hammer-driven corer was used. This is

55
a slow procedure and the large-diameter corer could not be used under normal
circumstances due to the necessity of completing 192 cores in a single day each time
sampling was done. Soil water content was sampled twice during the spring 1994
growing season, at 56 and 70 DAS. The first, at 56 DAS. was done on blocks 3 and 6
only because the corer shaft broke. All subsequent samplings were done on blocks 3, 6.
and 9. A single sample was taken in fall 1994 at 42 DAS. and two were taken in spring
1995 at 26 and 61 DAS. An additional attempt was made to measure soil water content
at 54 DAS, but this was not successful because the soil was too wet and the corer
jammed.
Leucaena small stem and leaf biomass and height. Leucaena tops were cut back
to a 50 cm stump height using a machete. Leaves were not separated from stems; they
were weighed together in the field with a spring scale. Sub-samples were taken and dried
in a Watlow oven at 70 degrees C until a constant weight was obtained. Height above the
stump of the tallest stem in each plot was measured with a fiberglass tape. All leaves and
stems were removed from the site after harvest.
Leucaena root intersections in trenched and nontrenched hedgerows. After the
spring 1995 maize harvest was completed, numbers of Leucaena roots were counted in
trenched and nontrenched hedgerows to discover what influence plastic barriers had on
root distribution. Transects perpendicular to the hedgerows were excavated through rows
4. 5, and 6 (Figure 5-1). where there happened to be three trenched plots adjacent to three
nontrenched plots. Work began on 13 January 1996, 105 days after the maize harvest.
The plots were weeded three times during that period to keep out roots of grasses and
weeds. Thirty sections 1.2 m wide by 1 m deep parallel to the trenched hedgerows and
an equal number parallel to the nontrenched hedgerows were mapped, at 20, 50, 100,

56
150. and 200 cm distances. A Plexiglas plate inscribed with a 10 by 10 cm grid was
pinned to each cut surface. The positions of the roots were marked on plastic sheets
clipped over the Plexiglas plate, a separate symbol was used for each of four diameter
classes: very fine (10 mm).
Statistical analysis
Statistical analysis was done using the SPSS version 9.0 univariate GLM
ANOVA procedure. The model used for the dependent variables Leucaena leaf and
small stem biomass and height above the stump was as follows:
yljkl = ft + block ,+tj + fk + t*fjk + error ljkl
where: y=dependent variable, t=trenching, f=fertilizer
/=2, 3,4, 5, 6, 7, 8,9, 10
j-0, 1 where 0=not trenched and 1 =trenched
k= 0, 1 where 0=no fertilizer applied andl=fertilizer applied
For each kind of field measurement taken at the four fixed distances and/or the
three soil depths, three levels of analysis were done—an overall ANOVA (referred to as
the complete model throughout the text) that included all the main effects and their
interactions, ANOVAs done separately for trenched and nontrenched plots that included
all other effects, and ANOVAS done separately for each distance from the hedgerows
that included only trenching and fertilization factors and their interaction. For percent
soil water, separate ANOVAs were done for each combination of distance from the
hedgerows with each of the three depths (0-15 cm, 15-30 cm. 30-45 cm). If there were
significant trenching by fertilization interactions produced by the ANOVAs done
separately for each distance and depth, contrast statements were run to determine at
which levels the interactions were significantly different.

57
The following example shows the complete ANOVA model used for the
dependent variables: maize biomass taken during the growing season, maize height,
maize leaf water pressure, maize grain at final harvest, number of maize plants at final
harvest, and number of maize ears at final harvest.
yljUm = n + block ,+tj + fk + t*fjk + d,+ t*d Jt + f*d ,, + t*f*d /kl + error yklm
where: y=dependent variable, t=trenching, f=fertilizer. d=distance
/=2, 3, 5, 6, 8. 9 (blocks used varied for each dependent variable)
_/'=0, 1 where 0=not trenched and l=trenched
k=0, 1 where 0=no fertilizer applied andl=fertilizer applied
7=50, 100, 150, 200 cm distance from hedgerows
The overall mean square error was used to test the effect of distance and its
interactions. However, because measurements were taken at fixed distances and depths,
giving a split-plot characteristic to the experimental design, an “error a” term was
computed by adding sums of squares for the interactions of blocks with trenching and
fertilizer. This was used to test the effects of trenching and fertilizer and their
interaction.
The complete model for the percent soil water ANOVAs was:
y.jkh,,,, =M + block, +tj+fk +t*fjk + d, +t*djt + f*d Jt + t*f*djk,
+ dp m +1 * dp Jm + / * dp km +d* dplm + t* / * dp jkm + errortjklmn
where: y=dependent variable. t=trenching, f=fertilizer, d=distance
/=3. 6. 9
7=0. 1 where 0=not trenched and l=trenched
k= 0, 1 where 0=no fertilizer applied andl=fertilizer applied
7=50, 100, 150, 200 cm distance from hedgerows
w=0-15, 15-30, 30-45 cm depth from soil surface

58
In addition to the “’error a” term computed and applied as above, for the soil water
ANOVAs an “error b” term was computed to test the effects of distance and its
interactions with trenching and fertilizer. Sums of squares for the interactions of blocks
with distance and its interactions with trenching and fertilizer were added to compute this
term. The overall error term was used to test the effects of depth and its interactions.
“Error a” and ‘“error b” and the corresponding F values were computed by hand;
SAS version 7.0 was used to compute the appropriate p-values. Soil water was expressed
as a percent, however the range of values was narrow (13% to 39%) and not extremely
high or low, so an arcsin transformation to stabilize the variances was not applied before
running the ANOVAs (Ott 1992).
Not all of the interactions that resulted in significant tests were meaningful due to
the very low numerical values of the differences. P-values for all factors and their
interactions are shown in Appendix C for all of the ANOVAs and contrasts.
Paired t-tests were used to detect significant differences in the numbers of root
intersections between trenched and nontrenched plots. This was possible because each
pair of measurements at each distance and depth were taken from physically adjacent
positions.
Results and Discussion
Results for each maize season are presented separately in this section, followed
by the results of the root distribution study and a general discussion.
Spring 1994 season
Two factors made this season differ from the two subsequent ones—the 30 cm
root barriers installed in the trenched plots in 1991 apparently no longer presented an
effective obstacle to the passage of Leucaena roots from the hedgerow into the alleys,
and the rainfall was poorly distributed.

59
Rainfall was 206 mm for the growing season and there was good soil water at
planting (51 mm fell in the 7 days prior to planting). However, most of the rain during
the growing season fell between 11 and 21 DAS (154 mm), and then no rain fell until 67
DAS (23 mm), which was the last effective rainfall event before final harvest. The plants
had reached about 20 cm height by 14 DAS. At 35 DAS the height was about 1 m. but
the soil was dry and cracking. Maize was always shorter than Leucaena until the 33
DAS hedgerow pruning; at that time the maize was about one meter high and the
hedgerows 50 cm. Maize tasseling began about 42 DAS, with most of the plants in
anthesis or silk by 50 DAS. The plants showed increasing signs of stress as the season
progressed. Many of the plants had fired leaf margins, and soil cracks opened to about
2.5 cm after 60 DAS.
Maize biomass production. Aboveground biomass production during the growing
season was not affected significantly by trenching or by fertilizer application, but the
amount of biomass did differ with distance from the hedgerows. The four
measurements of maize biomass are shown together in Figure 5-6.
Figure 5-6:
Maize biomass yields at 50, 100, 150, and 200 cm from the hedgerows at
four times after sowing during the spring 1994 season

60
For the ANOVA model that included distance, there were no significant effects of
fertilization (a=.05 or a=. 10) on the weight of maize biomass, and only one case where
trenching was significant (p = .085) at 63 DAS. P-values for the effect of distance on
maize biomass were <.001 at all four observations. The weight of maize biomass
produced was always significantly lower at 50 cm from the hedgerows. It appeared to
peak at 150 cm for the first three observation periods, although that distance was
significantly greater than the others only at 49 DAS (Figure 5-6). At 70 DAS, the weight
of biomass peaked at 200 cm. Competition between the maize and the trees clearly
extended to 100 cm at 42 DAS, and to 50 cm thereafter.
When the analyses were done separately at each distance, the effect of trenching
was significant only at 50 cm from the hedgerows at 42 DAS and at 63 DAS (Table 5-2).
In these two cases, the amount of maize biomass produced in trenched plots was at least
double that produced in nontrenched plots.
Table 5-2: Effect of trenching on the production of maize biomass 50
cm from the hedgerows at 42 and 63 DAS; spring 1994.
42 DAS
63 DAS
Trenched biomass (g)
50
93
Nontrenched biomass (g)
25
37
p-value
.039
.039
There were also four instances when the effect of fertilizer was significant. More
maize biomass was produced in the nonfertilized (F-) plots at 49 DAS at both the 50 cm
and 100 cm distance from the hedgerows (Table 5-3). At both 42 and 49 DAS there was a
significant interaction of fertilizer and trenching at the 150 cm distance from the
hedgerows, in both cases more maize biomass was produced in trenched (T+) plots

(Table 5-3). However, at 42 DAS this effect was seen only between trenched and
nontrenched plots that were also fertilized, while at 49 DAS the effect was seen onl\
between plots that were not fertilized.
61
Table 5-3: Effect of fertilizer on the production of maize biomass 50. 100, and 150 cm
from the hedgerows at 42 and 49 DAS; spring 1994.
49 DAS
50 cm 100 cm
42 DAS
150 cm F+
49DAS
150 cm F-
Maize (g) F+
50
168
Maize (g) T+
196
238
Maize (g) F-
62
209
Maize (g) T-
163
174
p-value
.063
.002
p-value
.055
.010
Maize growth stage. Numerical values corresponding to maize growth stages
were assigned at 56 DAS using the scale 0=absent, 1= vegetative growth, 2=tassel,
3=anthesis, 4=silk. On average, the plants at 50 cm were in the vegetative stage, all
others were in tassel or anthesis (Figure 5-7).
s
o
<5
1
o
Distance from the hedgerow (cm)
Figure 5-7: Growth stage of maize at four distances from hedgerows, 56 DAS; spring
1994.

Soil water content. This was measured twice during this season, at 56 and 70
DAS. The ANOVAs that included both distance and depth in the model did not reveal
62
any differences in soil water due to trenching, fertilization, or distance from the
hedgerows. There were differences among depths below the surface, as shown in Table
5-4. At 70 DAS the surface 15 cm had more soil water than the deeper layers, probably
because 23 mm of rain fell just three days before the measurement. This was the first
rain since 21 DAS. At 56 DAS the water content of the soil increased steadily with
depth. The water content of the 30-45 cm layer was the same for both observations.
Table 5-4: Soil water at three depths below the surface at 56 and 70
DAS; spring 1994.
Soil Water %-
—Gravimetric
56 DAS
70 DAS
0-15 cm
15.0 a
24.0 c
15-30 cm
21.6b
22.2 a
30-45 cm
23.5 c
23.4 b
p-value
<.001
<.001
The complete model also indicated significant soil water interactions between
trenching and fertilizer (p = .081) at 56 DAS. and between trenching and distance (p =
.020) and fertilizer and distance (p = .007) at 70 DAS. When ANOVAs were done
separately for each depth and distance, with trenching and fertilizer as the only
independent variables, there were no significant differences in soil water percent between
the two trenching treatments at any distance or depth. In only two cases were the
differences between the two fertilizer treatments significant: at 56 DAS 100 cm distance
in the 30-45 cm depth the fertilized plots had 23.1% soil water vs. 24.2% for the
unfertilized plots (p = .004), and at 70 DAS 200 cm distance in the 15-30 cm depth

63
fertilized plots had 21.1% soil water vs. 23.1% for the unfertilized plots (p = .027). At 56
DAS there was a significant interaction at the 100 cm distance from the hedgerows: at the
0-15 cm depth trenched plots had less soil water than nontrenched plots where no
fertilizer was applied (p = .067); the reverse was true in the fertilized plots (p = .034).
The meaning of these interactions is not clear, since fertilizer was not significant as a
main effect. P-values for all interactions are listed in Appendix C, but not all will be
discussed as they are too numerous and add little to the understanding of the treatment
effects.
Maize yield. Grain yield for this season was very low, with no significant
differences due to trenching or fertilizer. A per-hectare equivalent corresponding to the
average values per six-meters as shown in Figure 5-8 is 182 kg/ha maize grain, with only
5.2 kg/ha of the total contributed by the 50 cm rows. As with maize biomass and sexual
development, grain yield was affected only by distance from the hedgerows. In addition,
the peak production in grain yield and number of ears appeared at 150 cm from the
hedgerows, then declined slightly at 200 cm. This is also similar to the pattern observed
in biomass production and sexual development, with a decline between 150 cm and 200
cm, although it was only statistically significant at 49 and 56 DAS.
Apparently, competition between maize and hedgerows affected the survival of
maize plants differently than grain yield or number of ears. There were significantly
fewer surviving plants at harvest time in the row closest to the Leucaena, but in the other
three more distant rows maize plant survival was the same, and did not decrease between
150 and 200 cm (Figure 5-9). Full stocking for each distance depicted in Figure 5-9
would have been 16 plants (per 6 m, or per 8 hills), equal to 12,000 plants per hectare at
each distance, totaling 48,000 plants per hectare if all had survived. Maize plant survival

64
with distance from the hedgerow was 48%, 84%, 87%, and 80% at 50 cm. 100 cm. 150
cm, and 200 cm from the hedgerows, respectively.
Figure 5-8: Yield of maize grain per six meters at four distances from the hedgerows;
50 cm 100 cm 150 cm 200 cm
Distance from hedgerow
spring 1994.
50 cm 100 cm 150 cm 200 cm
Distance from the hedgerow
Figure 5-9: Number of maize plants per eight hills at four distances from the hedgerows;
spring 1994.

65
Since the number of ears did not differ significantly between 100 and 150 cm
(Figure 5-10) and the grain yield did differ, the implication is that grain yield per ear was
lower at 100 cm than at 150 cm. This was the case, with the grain weight (g) per ear for
the four distances being, respectively, 3.5, 10.8, 13.0. and 9.7.
9 t
Distance from the hedgerow
Figure 5-10: Number of maize ears per eight hills at four distances from the hedgerows;
spring 1994.
Table 5-5 shows the data presented in Figures 5-8, 5-9, and 5-10 on a per hectare
basis.
Table 5-5: Maize grain weight, number of plants, and number of ears at
four distances from the hedgerows and as a total per hectare,
with p-values for the differences between distances.
Distance (cm)
Grain (ka)
No. of plants
No. of ears
50
5.2 a
5,813 a
1,500 a
100
56.9 b
10,031 b
5,281 be
150
75.3 c
10,406 b
5.813 c
200
44.6 b
9,563 b
4.594 b
Total/ha
182.0
35,813
17,188
p-value
<.001
<.001
<.001

66
Leucaena biomass-production and height. The four Leucaena harvests discussed
in conjunction with the spring 1994 maize season are those made after my post-coup
d’etat-return to Haiti, but before the hedgerows were re-trenched prior to the fall 1994
maize season. The 20 November 1993 harvest was the first re-growth after the
overgrown trees were cut back to 50 cm stumps, but this was before the first application
of fertilizer to a maize crop so the F+ and F- designations are not meaningful for this
harvest. Table 5-6 shows the weight of biomass harvested and Table 5-7 shows the
height above the stump for the four treatment combinations. The 33 and 65 DAS
harvests were taken during the spring 1994 maize season.
Table 5-6: Stem and leaf biomass harvests and daily growth increments of Leucaena
hedgerows 33 and 65 DAS; Spring 1994.
20 Nov 93
2 Jun 94
4 Jul94
20 Aug 94
Biomass
Increment
Biomass
Increment
Biomass
Increment
Biomass
Increment
(kg/m)
(g/m/d)
(kg/m)
(g/m/d)
(kg/m)
(g/m/d)
(kg/m)
(g/m/d)
T+F+
1.1
17.8
0.9
12.4
0.4
11.5
0.5
10.0
T+F-
1.2
18.3
0.9
12.6
0.3
10.4
0.4
9.1
T-F+
1.4
21.7
1.0
14.1
0.4
12.2
0.5
10.2
T-F-
1.2
18.9
0.9
12.5
0.4
11.3
0.5
10.9
p-value T
.074
.139
.300
.283
p-value F
.358
.224
.187
.941
p-value TxF
.198
.091
.994
.400
DSPHa
63
69
32
47
RSPH (mm)b
162
312
0
53
DAS
42°
33
65
between crops
a: Days since previous hedgerow harvest
b: Rainfall since previous hedgerow harvest
c: DAS (days after sowing) is reported because a fall 1993 maize crop was growing in the
plots at this time. This maize season is not reported here, as it was lost due to animal
damage.
It appears that the effect of the initial trenching on Leucaena biomass production
did not last beyond 20 November 1993, when there was a small reduction in biomass

67
production in the trenched plots (p = .074). Trenched plots produced 1.15 kg/m. while
nontrenched plots produced 1.3 kg/m. A five-meter spacing means there are 2.000
meters of hedgerow per hectare, so this amounts to a reduction of 300 kg/ha due to
trenching. There was no significant trenching effect in the subsequent three harvests, nor
did fertilizer application to the maize influence Leucaena production. The reduction in
Leucaena daily growth rate beginning in July and continuing in August was probably a
response to less rainfall during that period.
Table 5-7: Height above a 50 cm stump and daily growth increments of Leucaena
hedgerows prior to the second root trenching. 33 and 65 DAS; Spring 1994.
20 Nov 93
2 Jun 94
4 Jul94
20 Aug 94
Height
(m)
Increment
(cm/d)
Height
(m)
Increment
(cm/d)
Height
(m)
Increment
(cm/d)
Height
(m)
Increment
(cm/d)
T+F+
2.1
3.3
1.8
2.6
1.2
3.8
1.4
3.0
T+F-
2.1
3.3
1.8
2.6
1.3
4.1
1.4
3.0
T-F+
2.0
3.2
1.8
2.6
1.2
3.8
1.4
3.0
T-F-
2.2
3.5
1.7
2.5
1.2
3.8
1.5
3.2
p-value T
.725
.462
.095
.403
p-value F
.076
.840
.393
.640
p-value TxF
.129
.329
.160
.640
DSPHa
63
69
32
47
RSPH (mm)b
162
312
0
53
DAS
42c
33
65
between crops
a: Days since previous hedgerow harvest
b: Rainfall since previous hedgerow harvest
c: DAS (days after sowing) is reported because a fall 1993 maize crop was growing in the
plots at this time. This maize season is not reported here, as it was lost due to animal
damage.
Leucaena height growth is a less precise measure than the biomass yield, and was
taken to the nearest 10 cm. Neither trenching nor fertilizer application appeared to affect
height growth during the four harvests shown in Table 5-7. The p-value of .076 for the
effect of fertilizer on 20 November 1993 is difficult to explain, because this was before

68
the first fertilizer application to the maize crop. The significant (p = .095) effect of
trenching on 4 July is associated with a very small numerical height advantage in the
trenched, nonfertilized plots. It is interesting to note that while daily Leucaena biomass
increment decreased after 2 June (Table 5-7), daily height growth rate increased. The
July and August cutting intervals were shorter than the previous two, so this may mean
that Leucaena puts on stem height first, then adds stem diameter, leaf biomass, or new
stems later in the cycle.
Fall 1994 season
The fall 1994 season was the best of the three in terms of maize production,
because it had the highest rainfall and the best rainfall distribution through the growing
season. After the Leucaena was pruned at 28 DAS, the maize quickly gained a height
advantage over the hedgerow trees and maintained it throughout the rest of the season in
both the trenched and nontrenched plots, including the rows closest to the trees. Most of
the plants were in flower (tassel and silk) by the time tropical storm Gordon passed at 65
DAS. By 70 DAS the maize was about 3 m tall, taller by about 1 m than the Leucaena
hedgerows.
Maize biomass production. Maize biomass production was measured only once
during the season, at 78 DAS. The ANOVA for the complete model (including distance)
was significant for trenching (p = .018) and distance from the hedgerow (p = .023).
ANOVAs separate by distance showed that only at the 50 cm distance from the
hedgerows was trenching a statistically significant factor (p = .037). At all other
distances trenched plots produced more maize biomass than nontrenched plots, but the
difference was not significant (Figure 5-11). The complete model also indicates an
interaction between trenching and fertilizer at the 150 cm distance (p = .093). There was

69
significantly more biomass produced in trenched (509 g) than nontrenched plots (307 g)
where they were not fertilized, but the difference was not significant in fertilized plots.
100 150
Distance from the hedgerow
Figure 5-11: Trenched and nontrenched plots at 50, 100, 150, and 200 cm from the
hedgerows, 78 DAS; fall 1994.
ANOVAs run separately for trenched and nontrenched hedgerows showed that
distance was a significant factor only for nontrenched hedgerows, with the 50 cm
distance producing less biomass than the other distances. This was not the case for
trenched hedgerows, where no significant difference in biomass production was detected
over the four distances (Table 5-8).
Table 5-8: Maize biomass in trenched and nontrenched plots over four
distances from the hedgerows, 78 DAS; fall 1994.
Distance (cm)
Trenched
(g/2 hills)
Nontrenched
(g/2 hills)
50
412.6 a
249.2 a
100
501.9 a
389.1 b
150
475.5 a
416.4 b
200
400.6 a
356.2 b
p-value
.485
0.001

70
Maize leaf water potential. The complete ANOVA model did not show an\
significant leaf water potential differences among factor levels for any factor. However,
when the ANOVAs were run separately for each distance, maize plants in trenched plots
indicated less leaf water pressure than did those in nontrenched plots (p = .081) at the 50
cm distance from the hedgerows (Figure 5-12). Differences were not significant at the
other distances.
50 100 150 200
Distance from the hedgerow (cm)
Figure 5-12: Maize leaf water potential in trenched and nontrenched plots at four
distances from the hedgerow. 42 DAS; fall 1994.
Similarly, when ANOVAs were run separately for trenched and nontrenched
plots, there was no significant difference in maize leaf water potential among the four
distances for the nontrenched plot. For the trenched plots, however, the leaf water
potential in the 50 cm plots was significantly lower (p = .008) than in the other three
distances (Table 5-9).

71
Table 5-9: Maize leaf water potential in trenched and nontrenched plots
over four distances from the hedgerows, 42 DAS; fall 1994.
Distance (cm)
Trenched (bars)
Nontrenched (bars)
50
1.1 a
2.0 a
100
2.5 b
1.7a
150
1.8b
2.0 a
200
2.1 b
2.1 a
p-value
.008
.917
Soil water content. Soil water was measured on the same day as maize leaf water
potential. Figure 5-13 shows the soil water percents for trenched plots at all
combinations of distance and depth. The differences in soil water percents between
trenched and nontrenched plots are shown in parentheses. The overall model indicated
that trenching (p = .042), distance from the hedgerow (p < .001). and depth below the
surface (p < .001) were significantly affecting soil water content. The effect of fertilizer
was not significant (p = .352). Separate ANOVAs for each combination of distance and
depth were not significant for either trenching or fertilizer in the 0-15 cm depth. There
were significant differences due to trenching in the lower depths. In the 15-30 cm layer,
soil water percent was higher in trenched plots at the 50 cm (p = .033) and 150 cm (p =
.095) distances from the hedgerow. In the 30-45 cm layer, soil water percent was higher
in trenched plots at the 50 cm (p = .002) and 100 cm (p = .014) distance from the
hedgerow'.
The effect of distance from the hedgerows was significant in trenched plots at the
15-30 cm (p = .004) and the 30-45 cm depths (p < .001), with more soil water present
close to the hedgerows. In the surface layer distance was significant for trenched plots at
the a=.10 level, with more soil water at the 50 cm distance than the 150 distance (p =
.078). In Figure 5-13, cells having the same letter, in a given row, are not significantly

72
different. Soil water percent in nontrenched plots did not differ significantly with
distance from the hedgerow in any of the three depths.
Figure 5-13: Soil water in trenched plots (and difference between T+ and T- plots) at
42 DAS, fall 1994. Cells with solid border (a=.05) or dashed border
(a=.10) indicate significantly greater soil water in trenched plots than in
the corresponding nontrenched plots at that position. Letters indicate
differences between distances within a depth for trenched plots (a=.05).
Soil water percent increased with depth differently in trenched plots than in
nontrenched plots. In the trenched plots (Table 5-10), soil water increased with depth at
the 50 cm distance (p = .066) and the 100 cm distance (p = .020), with the deepest layer
having significantly more water than the surface layer. However, there were no
significant differences in soil water with depth at the 150 and 200 cm distances. In
contrast, the nontrenched plots had significantly more soil water with depth at the 100
cm. 150 cm, and 200 cm distances, but did not show significantly more soil water with
depth at the 50 cm distance.

73
Table 5-10: Soil water in trenched and nontrenched plots by depth and distance from the
hedgerow, 42 DAS; Fall 1994.
Distance (cm)
Depth (cm)
Trenched (%)
Nontrenched (%)
0-15
27.8 a
26.4
50
15-30
29.9 ab
26.3
30-45
32.5 b
27.6
p-value
.066
.820
0-15
24.6 a
23.1 a
100
15-30
27.2 ab
25.6 ab
30-45
29.7 b
26.3 b
p-value
.020
.075
0-15
23.6
22.2 a
150
15-30
25.0
23.9 ab
30-45
23.6
25.5 b
p-value
.368
.012
0-15
24.4
20.8 a
200
15-30
23.5
26.7 b
30-45
26.7
27.3 b
p-value
.502
.048
Maize yield. The complete model for maize grain was significant for the effect of
trenching (p = .009) and distance from the hedgerows (p = .076), but not for the effect of
fertilizer (p = .385). The complete models for number of plants and for number of ears
were not significant for any of those factors. ANOVAs done separately at each distance
showed that maize grain from trenched plots was significantly greater (a=.10) than that
from nontrenched plots at the 50 cm (p = .071) and 100 cm (p = .068) distances from the
hedgerows (Figure 5-14). There was also a statistically significant difference between
the numbers of plants in trenched plots vs. nontrenched plots at the 100 cm distance (p =

.020), with the nontrenched plots having about one more plant per six-meter sample at
that distance.
74
Figure 5-14: Maize grain yield in trenched (T+) and nontrenched (T-) plots at four
distances from the hedgerow. Fall 1994.
Final harvest is presented on a per-hectare basis in Table 5-11. Although there
were differences in grain yield between trenched and nontrenched plots at 50 and 100 cm
from the hedgerows (a=.10, Figure 5-14 above), there was no significant difference in
grain, number of plants or number of ears over distance from the hedgerows, for either
trenched or nontrenched plots.
In general, there were no strong differences in yield parameters this season,
probably due to the adequate, well-distributed rainfall partially negating the potential
advantages of the root barriers. If only the differences in grain weight at 50 cm and 100
cm are accepted as significant, then trenching made a difference of 252 kg maize

75
grain/ha, an increase of 12% over the nontrenched plots. However, since the effect of
trenching for the complete model was significant (p = .009), then the advantage of
trenching increased to 360 kg/ha (2,382 kg-2,022 kg), an increase of 18% over the
nontrenched plots. Since there were no differences in the numbers of plants or ears, the
advantage of trenching was manifested in more grain weight per ear in trenched plots,
especially at the 50 and 100 cm distances.
Table 5-11: Maize grain weight, number of plants, and number of ears at four
distances from the hedgerows and as a total per hectare, with p-values for
the differences between distances. Fall 1994.
Distance (cm)
Grain (kg)
T+ T-
No. plants
T+ T-
No. ears
T+ T-
50
563.1
423.1
10,563
10,625
11,250
10,313
100
648.1
535.6
10,188
11,125
10.750
11,313
150
601.3
561.3
11,188
10,813
11,313
11,688
200
569.4
501.9
10,938
10,875
10,688
10,750
Total/ha
2,382
2,022
42,875
43.438
44,000
44,063
P-value
.488
.103
.152
.637
.869
.314
Leticaena biomas—production and height. Four Leucaena harvests are presented
in Tables 5-12 and 5-13—the first was done during the fall 1994 maize season at 28
DAS, the others were done after the fall maize harvest but before the spring 1995 maize
was sown. The reaction of Leucaena to the re-trenching to 50 cm depth done in August
1994 is clear. In every case, the trenched plots produced less biomass than the
nontrenched plots. The loss of small stem and leaf biomass due to the effect of trenching
was 532,452.422. and 154 kg per hectare for the four pruning dates in Table 5-12.
respectively, for a total loss over this seven-month period of 1,561 kg per hectare. The

76
relatively slow growth rate between the 8 October and 14 January harvests in spite of the
plentiful rainfall was probably due to insect damage (species not identified) in the
southeast quarter of the trial site, in blocks 6-10. The damage affected trenched and
nontrenched plots equally. Fertilizer application on the adjacent maize did not affect
Leucaena biomass or height growth. The significant trenching by fertilization interaction
at the 6 May harvest does not appear to be important in magnitude.
Table 5-12: Stem and leaf biomass harvests and daily growth increments of Leucaena
hedgerows after the second root trenching, Fall 1994.
8 Oct 94
14 Jan 95
16 Mar 95
6 May 95
Biomass
Increment
Biomass
Increment
Biomass
Increment
Biomass
Increment
(kg/m)
(g/m/d)
(kg/m)
(g/m/d)
(kg/m)
(g/m/d)
(kg/m)
(g/m/d)
T+F+
0.3
6.8
0.4
3.8
0.5
8.5
0.4
7.1
T+F-
0.3
6.5
0.4
3.6
0.5
8.6
0.3
5.9
T-F+
0.6
12.0
0.5
5.3
0.7
11.3
0.4
7.3
T-F-
0.6
12.2
0.7
6.8
0.8
13.0
0.4
8.7
p-value T
<.001
.005
<.001
.013
p-value F
.985
.401
.301
.919
p-value TxF
.745
.255
.358
.033
DSPHa
49
98
61
51
RSPHb (mm)
174
320+
56
76
DAS
28
between crops
between crops
between crops
a: Days since previous hedgerow harvest
b: Rainfall since previous hedgerow harvest
Trenching significantly reduced Leucaena height growth only for the first two
harvests following re-trenching (Table 5-13). Height difference between trenched and
nontrenched plots wras not significant during the third harvest, even though the difference
in small stem and leaf biomass was significant.

Table 5-13: Height above a 50 cm stump and daily growth increments of Leucaena
hedgerows after the second root trenching. Fall 1994.
77
8 Oct. 94
14 Jan. 95
16 Mar. 95
6 May 95
Height
Increment
Height
Increment
Height
Increment
Height
Increment
(m)
(cm/d)
(m)
(cm/d)
(m)
(cm/d)
(m)
(cm/d)
T+F+
1.3
2.7
1.9
1.9
1.5
2.5
1.4
2.7
T+F-
1.4
2.9
1.8
1.8
1.5
2.5
1.3
2.5
T-F+
1.5
3.1
2.2
2.2
1.6
2.6
1.5
2.9
T-F-
1.5
3.1
2.2
2.2
1.6
2.6
1.5
2.9
p-value T
.005
.038
.113
.311
p-value F
1.000
.945
.633
.311
p-value TxF
.471
.331
.447
.030
DSPH
49
98
61
51
RSPH (mm)
174
320+
56
76
DAS
28
between crops
between crops
between crops
a: Days since previous hedgerow harvest
b: Rainfall since previous hedgerow harvest
Spring 1995 season
This season had less rainfall during the growing season than fall 1994, but the
distribution was better than that during the spring 1994 season. Germination initially
appeared to be good, with the tallest plants 20 cm tall at 19 DAS. However, at 26 DAS it
was evident that germination at the 50 cm distance was poor. Maize height was up to 30
cm at that time, but some plants were showing leaf curl. Flowering began at 47 DAS; by
54 DAS most of the plants were flowering. At this time there appeared to have been an
underground leak in the irrigation system (used in the nearby tree nursery) that might
have increased soil water content in the east side of blocks 9 and 10. The only
measurement that might have been influenced was the maize biomass harvest done at 54
DAS. Since maize was cut in blocks 4, 7, and 10 and the eastern-most plot in block 10

78
was a nontrenched plot, an inadvertent increase in soil water content in that plot only
strengthens the result of the ANOVA that showed a significant difference in maize
biomass due to trenching.
The effect of trenching on maize development and yield was more pronounced
than during the other two seasons. However, even though this was the only season for
which the complete amount of fertilizer recommended by the Ministry of Agriculture
(MARNDR 1990)was applied, there were no significant effects of fertilizer on any of the
parameters measured for the complete ANOVA models.
All the ANOVAs run using the complete model for maize growth and yield
during this season produced very significant p-values for the effects of trenching, with
trenched plots producing more than nontrenched plots, especially close to the hedgerows.
All but the two maize biomass harvests were significant for distance from the hedgerow
and the trenching by distance interaction (Appendix C). Much of the difference was due
to the poor survival of maize plants in the nontrenched plots, especially in the two rows
closest to the trees.
Maize height. Figure 5-15 shows a significant maize height advantage at 40 DAS
in trenched plots at all four distances from the hedgerows (p < .001). Missing plants
were assigned values of zero because the mortality was attributed to competition with the
hedgerows. There were more missing plants in nontrenched plots than in trenched plots:
at the 50 cm distance 84% of the hills in nontrenched plots had no surviving plants vs.
20% for trenched plots; at the 100 cm distance the corresponding numbers were 25% in
the nontrenched plots and 4% in the trenched plots.

79
50 100 150 200
Distance from the hedgerow (cm)
Figure 5-15: Maize height trenched and nontrenched at four distances from the
hedgerow. 40 DAS; Spring 1995.
Distance from hedgerows was significant for both trenched and nontrenched
plots, but not in the same way. Maize height in trenched plots reached a maximum at
100 cm, and then decreased (although not significantly) at 150 and 200 cm. Maize height
in nontrenched plots was minimum at 50 cm, and then increased progressively with
distance but never attained the height as maize in trenched plots (Table 5-14).
Table 5-14: Maize height in trenched and nontrenched plots at four distances from
hedgerows, 40 DAS; Spring 1995.
Distance (cm)
Trenched (cm)
Nontrenched (cm)
50
30.0 a
2.2 a
100
40.6 b
13.5 b
150
36.1 ab
16.8 b
200
36.5 b
23.9 c
p-value
.009
<.001
Maize height was less than Leucaena height until the hedgerows were pruned at
60 DAS (Figure 5-5). This raises the question of whether the maize on the north side of

80
the hedgerows experienced a shade regime different from that on the south side. No
direct measurements of light were done, but maize height at 40 DAS was analyzed for the
effect of aspect. Neither the complete model (including distance as a factor) nor the
separate models for each distance showed orientation with respect to the hedgerows to be
a significant influence on maize height (data not shown).
The effect of fertilizer on maize height was significant only at the 50 cm distance
from the hedgerows and only in the trenched plots (p = .035), where fertilized maize was
10 cm shorter than nonfertilized maize.
Maize biomass production. Differences between maize biomass harvested in
trenched and nontrenched plots at 47 and 54 DAS were similar, and continued the pattern
of differences in maize height discussed above. However, the statistical tests produced
higher p-values than did those for height, probably due to smaller samples sizes and high
variability caused by three outliers in block 10 at the 50 and 100 cm distances. The
irrigation system leak causing water contamination in blocks 9 and 10, discussed above,
was probably responsible. The complete model ANOVA at 47 DAS was significant for
trenching (p = .035), but not for fertilizer or distance from the hedgerow. The 54 DAS
complete model was significant only for trenching at a=.10 (p = .085).
At 47 DAS, ANOVAs done separately for each distance showed that trenched
plots produce significantly more biomass at all four distances from the hedgerows
(Figure 5-16). P-values for trenching are = .030, .096, .016, and .017 for the four
distances. At 54 DAS, the difference due to trenching is significant only at the 50 cm
distance (p = .072). The difference between trenched and nontrenched plots at the 100
cm distance shown in Figure 5-17 was numerically large, but because of the outliers
discussed above, did not result in a significant difference.

81
50 100 150 200
Distance from the hedgerow (cm)
Figure 5-16: Maize biomass in trenched and nontrenched plots at four distances from
the hedgerow, 47 DAS, Spring 1995.
50 100 150 200
Distance from the hedgerow
Figure 5-17: Maize biomass in trenched and nontrenched maize weight at four
distances from the hedgerow, 54 DAS, Spring.

82
The changes in biomass over distance were similar for both observation dates. In
neither case were there significant differences over distance for trenched plots. In both
cases, maize biomass in nontrenched plots was less in the two rows closest to the
hedgerows than in the more distant rows (Table 5-15).
Table 5-15: Maize biomass in trenched and nontrenched plots at four distances from
the hedgerow, 47 and 54 DAS; Spring 1995.
Distance (cm)
47 DAS
T+ (g) T- (g)
54 DAS
T+(g) T- (g)
50
25.9
1.6 a
61.9
1.3 a
100
43.7
5.3 ab
77.9
6.0 a
150
30.4
9.2 b
36.1
18.8 b
200
25.0
6.5 b
21.3
18.3 b
p-value
.232
.010
.118
.007
Maize growth stage. Maize in trenched plots had attained a more advanced stage
of development than maize in nontrenched plots at both 68 and 75 DAS. Competition
with the trees retarded development of maize plants growing in nontrenched plots
increasingly as they neared the hedgerows. This was not the case for maize plants
growing in trenched plots—development was most advanced at 100 cm from the
hedgerows.
Figures 5-18 and 5-19 use different scales to indicate maize growth stage (see
Materials and Methods). A four-number scale was used at 68 DAS: 0=missing.
l=vegetative growth, 2=tassel/anthesis, 3=silk, and 4=ear. At 75 DAS a five-number
scale was used: 0=missing, l=vegetative, 2=tassel, 3=anthesis, 4=silk, and 5=ear.

83
50 100 150 200
Distance from the hedgerow (cm)
Figure 5-18: Trenched and nontrenched maize growth stage at four distances from the
hedgerow, 68 DAS; Spring 1995.
; 3.0 â– 
; 2.0 j
Anthesis
Ta ssel
MJ
100 150
Distance from the hedgerow (cm)
Figure 5-19: Trenched and nontrenched maize growth stage at four distances from the
hedgerow, 75 DAS; Spring 1995.

84
As was the case with maize biomass production, growth stage changed with
distance from the hedgerows differently for plants in trenched plots than for those in
nontrenched plots. Plants in trenched plots were most advanced at 100 cm from the
hedgerows, and identical at the other three distances. There was no retardation in the 50
cm row. and conditions for maize development appear to have been enhanced at 100 cm
due to the presence of hedgerows. Maize plants in the nontrenched plots were least
developed in the 50 cm row. and progressively more developed as distance increased.
Soil water content. The complete ANOVA models for soil water at 26 and 61
DAS did not produce significant p-values for trenching, fertilizer, or distance from the
hedgerow. Only depth was significant (p < .001) on both dates. ANOVAs run separately
for each distance and depth were also not significant for trenching or fertilizer, with the
exception of a significant p-value (.016) for fertilizer at 150 cm distance, 15-30 cm depth
at 26 DAS. The nonfertilized plots in that position had only 0.5% more soil water than
did fertilized plots, so this difference is probably meaningless. Figures 5-20 and 5-21
show the percent soil water for the various positions in the trenched plots, with the
difference between trenched and nontrenched plots in parentheses. None of these
differences were significant.
When ANOVAs were run separately for trenched and nontrenched plots for each
depth, there were no significant differences in soil water between distances from the
hedgerow for either trenched or nontrenched plots, with one exception. At 61 DAS in the
nontrenched plots in the 0-15 cm depth, the 50 cm distance had about 1.5% more soil
water than the other three more distant positions (p = .002). Soil water increased with
depth for both trenched and nontrenched plots at all distances from the hedgerows, at
both observation periods (Table 5-16).

85
Figure 5-20: Soil water percent in trenched plots (and difference between T+ and T-
plots) at 26 DAS. Spring 1995.
Figure 5-21: Soil water percent in trenched plots (and difference between T+ and T-
plots) at 61 DAS, Spring 1995.

Table 5-16: Soil water in trenched and nontrenched plots by depth and distance from
the hedgerow. 26 and 61 DAS; Spring 1995.
26 DAS 61 DAS
Distance (cm)
Depth (cm)
T+ (%)
T- (%)
T+ (%)
T- (%)
0-15
24.1 a
21.4 a
26.4
25.7 a
50
15-30
28.6 b
27.1 b
28.5
27.5 b
30-45
29.2 b
28.5 c
28.8
29.1 c
p-value
.044
<.001
.371
<.001
0-15
21.6 a
22.1 a
24.7 a
24.2 a
100
15-30
29.2 b
27.5 b
27.6 b
27.0 b
30-45
30.1 b
29.0 b
29.2 c
28.8 c
p-value
<.001
<.001
<.001
<.001
0-15
22.3 a
20.9 a
24.4 a
23.0 a
150
15-30
27.6 b
27.5 b
27.2 b
27.2 b
30-45
29.3 c
28.8 b
29.2 c
29.4 c
p-value
<.001
<.001
<.001
.001
0-15
20.9 a
22.4 a
24.1 a
23.5 a
200
15-30
27.4 b
27.4 b
27.0 b
26.9 b
30-45
29.2 c
28.9 b
29.6 c
28.9 c
p-value
<.001
<.001
<.001
<.001
Maize yield. Trenching during this season showed a clear advantage in maize
production over nontrenched hedgerow plots, and reflected the relationships observed
earlier in maize height, biomass production, and growth stage. This advantage extended
through 100 cm from the hedgerows, and for number of maize plants and ears as well as
grain weight. ANOVAs using the complete model produced very significant p-values
(.002 or better) for trenching, distance from the hedgerow, and the trenching by distance
interaction (Appendix C). The effect of fertilizer was not significant.
Figures 5-22, 5-23. and 5-24 show the results of the ANOVAs done separately for
grain weight, number of surviving plants, and number of ears, respectively, at each
distance. P-values for the effect of trenching for all three were <.001 at the 50 cm
distance, and .045 or better at the 100 cm distance. Fertilizer made no significant
difference at any distance for grain weight, but there were suggestive p-values for the

87
effect of fertilizer on number of plants (p = .074) and number of ears (p = .068) at the 50
cm distance only. In both cases, the fertilized plots had lower numbers, which would
indicate a depressing effect of fertilizer of 1.750 plants per hectare and of 2.250 ears per
hectare due to the reduction in the 50 cm rows.
Distance from the hedgerow (cm)
Figure 5-22: Maize grain weight from trenched and nontrenched plots at four distances
from the hedgerows. 131 DAS; Spring 1995.
ANOVAs run separately for trenched and nontrenched plots showed that distance
from the hedgerow significantly influenced yield in both cases (Table 5-17). In the
nontrenched plots, yield in the 50 cm rows was less than in the more distant rows, but
there were no significant differences in yield among the rows at 100, 150, and 200 cm
distances. In both nontrenched and trenched plots, the numerical maxima occurred at
100 cm, then decreased with distance, but this pattern was significant only in trenched
plots. Grain weight, number of plants, and number of ears produced in the trenched plot
100 cm rows were always significantly greater than those produced in the 200 cm rows.

Number of maize plants/6m
12
a
Distance from the hedgerow (cm)
Figure 5-23: Number of maize plants from trenched and nontrenched plots at four
distances from the hedgerows. 131 DAS; Spring 1995.
16
50 100 150 200
Distance from the hedgerow (cm)
Figure 5-24: Number of ears from trenched and nontrenched plots at four distances
from the hedgerows, 131 DAS; Spring 1995.

89
Table 5-17: Maize grain weight, number of plants, and number of ears at four
distances from the hedgerows and as a total per hectare, with p-values for
the differences between distances; Spring 1995.
Distance (cm)
Grain (kg/ha)
T+ T-
No. plants/ha
T+ T-
No. ears/ha
T+ T-
50
249.0 ab
25.3 a
6,750 ab
1,375 a
8,188 a
1,375 a
100
284.3 a
180.1 b
7,875 b
5,938 b
11,000 b
7,625 b
150
187.7 be
146.7 b
6,625 ab
5,250 b
8,500 a
7,313 b
200
158.1 c
144.0 b
5,125 a
5,313 b
6,688 a
6,438 b
Total/ha
879
496
26,375
17,875
34,375
22,750
P-value
.019
<.001
.040
<.001
.007
<.001
The quantity of maize produced during this season was intermediate between the
spring 1994 and fall 1994 seasons, however the influence of trenching on maize
production was stronger. If the increased grain weight is considered only in the 50 and
100 cm rows, where there were significant differences between trenching treatments,
then the increased maize grain due to trenching was 328 kg/ha, or a 66% increase over
nontrenched plots. If the differences at all four distances are considered the advantage is
383 kg/ha, a 77% increase over nontrenched plots. The difference was due to higher
maize mortality, lower grain weight per plant, and lower grain weight per ear in
nontrenched plots, especially in the 50 cm row but extending through 100 cm.
Leucaena biomass—production and height
While the effect of trenching on maize production was very strong during the last
season, its effects on Leucaena growth appears to have been mitigated by root re-growth
for the final three harvests. There were no significant effects of trenching or fertilizer on
Leucaena small stem and leaf biomass (Table 5-18). Leucaena height does show a

90
significant p-value for the effect of trenching at the 57 DAS harvest, with hedgerow trees
in trenched plots being about 25 cm taller than those in nontrenched plots (Table 5-19).
Table 5-18: Stem and leaf biomass harvests and daily growth increments of Leucaena
hedgerows at 19. 57, and 130 DAS; Spring 1995.
10 Jun. 95
18 Jul. 95
29 Sep. 95
Biomass
Increment
Biomass
Increment
Biomass
Increment
(kg/m)
(g/m/d)
(kg/m)
(g/m/d)
(kg/m)
(g/m/d)
T+F+
0.4
10.9
0.3
6.8
0.4
5.0
T+F-
0.4
10.3
0.2
6.4
0.4
5.3
T-F+
0.4
11.0
0.2
5.8
0.4
6.2
T-F-
0.4
11.3
0.2
6.2
0.4
6.1
p-value T
.246
.213
.164
p-value F
.833
.984
.924
p-value TxF
.478
.310
.840
DSPH
35
38
72
RSPH (mm)
39
81
90
DAS
19
57
130
Table 5-19: Height above a 50 cm stump and daily growth increments of Leucaena
hedgerows at 19, 57, and 130 DAS; Spring 1995.
10 Jun.95
18 Jul. 95
29 Sep. 95
Height
Increment
Height
Increment
Height
Increment
(m)
(cm/d)
(m)
(em/d)
(m)
(cm/d)
T+F+
1.1
3.1
1.3
3.4
1.4
1.9
T+F-
1.2
3.4
1.4
3.7
1.3
1.8
T-F+
1.2
3.4
1.1
2.9
1.4
1.9
T-F-
1.2
3.4
1.1
2.9
1.7
2.4
p-value T
.418
.041
.092
p-value F
.156
.553
.664
p-value TxF
.945
.520
.112
DSPH
35
38
72
RSPH (mm)
39
81
90
DAS
19
57
130
These results indicate that, 10 months after trenching, the trees in the hedgerows
no longer suffered a production penalty, while the effect of trenching on maize
production in adjacent plots remained strong.

91
Leucaena root distribution in trenched and nontrenched plots
Figures 5-25, 5-26, 5-27, and 5-28 show distinct differences in the number of root
intersections found in trenched plots and nontrenched plots 105 days after the final maize
harvest. Where the numbers in parentheses are positive, trenched plots had that many
more intersections than nontrenched plots; where the numbers are negative, they had that
many less intersections than nontrenched plots. Taking only the positions where the
differences were statistically different (indicated in the figures by solid or dashed lines
around the cells), a pattern can be observed. Trenched plots have more large (>10 mm)
roots 20 cm from the hedgerows in the 90-100 cm layer; more medium (5-10 mm) roots
20 cm from the hedgerows in the 40-60 cm layer; more fine (2-5 mm) roots 50 cm from
the hedgerows in the 40-50 cm layer; and more very fine (<2 mm) roots 200 cm from the
hedgerows in the 30-60 cm layer. It appears as though, because of the barriers, roots in
trenched plots have developed below the barriers and produced more very fine roots at
the 200 cm distance.
Trenched plots have fewer large and medium roots 20 to 50 cm from the
hedgerows in the 0-30 cm layer, fewer medium roots 100 to 200 cm from the hedgerows
in the 20-30 cm layer, fewer fine roots 20 to 50 cm from the hedgerows in 0-10 cm layer,
100 cm from the hedgerows in the 20-70 cm layer, and 150 cm from the hedgerows in the
30-40 cm layer; and fewer very fine roots 100 cm from the hedgerows in the 30-40 cm
layer. Root barriers have clearly reduced the numbers of root intersections of all
diameter classes, especially near the hedgerows and in the surface layers of soil through
150 cm from the hedgerows, but particularly at 100 cm from the hedgerows.

92
Figure 5-25: Number of very fine root (<2mm) intersections with a 10 by 100 cm plane
parallel to the trenched hedgerows at five distances and 10 depths, with the
difference between the trenched and nontrenched hedgerows shown in
parentheses. The double vertical line represents the plastic root barrier.
Numbers enclosed by dashed lines indicate statistical difference between
trenched and nontrenched plots at a=. 10.
Figure 5-26: Number of fine root (2-5 mm) intersections with a 10 by 100 cm plane
parallel to the trenched hedgerows at five distances and 10 depths, with the
difference between the trenched and nontrenched hedgerows shown in
parentheses. The double vertical line represents the plastic root barrier.
Numbers enclosed by solid lines indicate statistical difference at a=.05,
dashed lines at a=.10.

93
Figure 5-27: Number of medium root (5-10 mm) intersections with a 10 by 100 cm
plane parallel to the trenched hedgerows at five distances and 10 depths,
with the difference between the trenched and nontrenched hedgerows
shown in parentheses. The double vertical line represents the plastic root,
barrier. Numbers enclosed by solid lines indicate statistical difference at
a=.05, dashed lines at a=.10.
Figure 5-28: Number of large root (>10 mm) intersections with a 10 by 100 cm plane
parallel to the trenched hedgerows at five distances and 10 depths, with the
difference between the trenched and nontrenched hedgerows shown in
parentheses. The double vertical line represents the plastic root barrier.
Numbers enclosed by solid lines indicate statistical difference at a=.05.

94
General Discussion
Summary bv season
Spring 1994. This was a season of poor rainfall and poorly functioning root
barriers. The root barriers were three years old at the time the maize was planted. In
addition, the root systems of the Leucaena hedgerows were allowed to develop
unhindered during a two-year period from 1991 to 1993 when the tops were not pruned,
allowing Leucaena roots to re-grow into the alleys over and under the old 30 cm plastic
barriers. Based on field observations of soil cracks and leaf wilt, maize plants were
under water stress during the flowering period. There was no rainfall from 21 DAS until
67 DAS. Drought stress was severe during flowering, which began about 42 DAS. The
soil water content of 22% to 24% from 15 cm to 45 cm depth taken at 56 and 70 DAS
correspond to leaf water potentials of 20 bars to 12 bars in this soil (Appendix A). Maize
is severely stressed at 16 to 18 bars leaf water potential (Herrero and Johnson 1981).
Drought during flowering causes severe decrease in maize yield (de Geus 1973). There
were no significant effects of trenching or fertilization on maize growth and yield, or on
soil water content. A residual effect of the old root barriers was seen only in the maize
biomass harvests at 42 DAS and 63 DAS, when more maize biomass was produced in
trenched plots at 50 cm from the hedgerows. Distance from the hedgerow was
significant for all maize measurements and soil water content. Maize growth (biomass
taken during the season, growth stage, grain weight, number of plants, number of ears)
was always lowest at the 50 cm distance and most commonly highest at the 150 cm
distance. In three cases (the 49 DAS biomass, grain weight, and number of ears), there
was a decline from the 150 cm distance to the 200 cm distance. Overall maize grain
yield was only 182 kg/ha. There may have been some competition for light between

95
hedgerows and trees since the Leucaena were taller than the maize until 33 DAS when
the hedgerows were pruned, but the principal limitation on maize growth this season was
water. Neither trenching nor fertilizer application in the maize plots had any effect on
Leucaena biomass or height.
Fall 1994. This season saw very good rainfall and newly re-established root
barriers to 50 cm depth. Maize was taller than the Leucaena after the 28 DAS hedgerow
pruning, most of the maize was in tassel and silk at 65 DAS. At 42 DAS, maize leaf
water potential was less in trenched than in nontrenched plots at the 50 cm distance, and
there was more soil water close to hedgerows in trenched plots. The maize was able to
take advantage of the extra soil water, because at 78 DAS more maize biomass was
produced in trenched plots than in nontrenched plots at the 50 cm distance. In the
nontrenched plots, maize biomass at the 50 cm distance was significantly less than that
produced farther from the trees, but there was no difference by distance in the trenched
plots. Since both trenched and nontrenched plots experienced the same shade conditions,
this difference in biomass production was apparently due to water competition from the
hedgerows. The increased grain yields at the 50 and 100 cm distance in trenched plots
correspond to the other measurements made earlier in the growing season. Differences in
maize yield over distance were muted by the abundant rainfall. Although the maize
biomass in nontrenched plots at 78 DAS did produce the lowest value at 50 cm from the
hedgerows, there was no difference in grain weight, number of plants, or number of ears
between distances from the hedgerow for either trenched or nontrenched plots. Grain
yield in trenched plots was 2,382 kg/ha, an increase of 18% over the nontrenched plots.
Some of these findings are consistent with reduced numbers of roots in trenched plots
close to the hedgerows. Soil water percent was highest at 50 cm from the hedgerows and

96
lowest 150-200 cm from the hedgerows in the trenched plots, corresponding to lowest
maize leaf water pressure at the 50 cm distance (both measurements taken at 42 DAS).
Maize biomass at 78 DAS and grain weight at final harvest were both greater in trenched
plots than in nontrenched plots at the 50 cm distance. The recent re-trenching reduced
Leucaena biomass production by 1,561 kg/ha over the four pruning operations carried
out during this period compared to nontrenched hedgerows.
Spring 1995. This season’s rainfall was about the same as that of spring 1994,
but the distribution was much better. Maize was taller than Leucaena only after the 57
DAS pruning, and there was poor maize survival in the 50 cm row of the nontrenched
plots. The root barriers were apparently effective. All maize growth and yield
parameters were significantly greater in trenched plots than in nontrenched plots at the 50
cm distance from the hedgerows, and all but one at the 100 cm distance. In addition, all
growth and yield measurements in nontrenched plots were lowest at the 50 cm distance
and peaked at the 150-200 cm distance. In the trenched plots, the highest values were
most often found at the 100 cm distance, with a decline at the 150 or 200 cm distance.
Grain yield was 879 kg/ha in the trenched plots, an increase of 77% over the nontrenched
plots. These findings are consistent with the decreased number of root intersections at
the 20-100 cm distance and the increased number of very fine root intersections at the
200 cm distance in trenched plots. Soil water content measured at 21 and 61 DAS was
not significantly different between trenching treatments or by distance from the
hedgerows, but only by depth. Trenched and nontrenched plots both increased in soil
water content with depth, but the amount of soil water present did not represent a
condition of severe water stress for the maize plants, according to the relationship
established in the pot study (Appendix A). A soil water percent above 25% would be
approximately equivalent to 10 bars of leaf water pressure, above which maize begins to

97
experience water stress (Herrero and Johnson 1981). At soil depths below 15 cm, all
positions had more than that amount of water at both observations. The effects of
trenching on Leucaena biomass production were no longer apparent during this season.
Impact of trenching on Leucaena
Trenching has a short-term impact on Leucaena growth, primarily on small stem
and leaf biomass and to a much lesser extent on height. Figure 5-29 shows the results of
15 harvests from 1991 through 1995. The initial trenching to 30 cm was done in April
1991, just before the first harvest shown in Figure 5-29. Trenched plots always produced
less biomass (a=.05) than nontrenched plots during the four subsequent harvests in 1991,
for a total loss of 940 kg during seven months. Since no Leucaena harvests were taken
for the next two years, it is not possible to know how long the effect of the first trenching
lasted. In September 1993, after a prolonged absence from the site, the overgrown
hedgerow trees (many with a diameter >10 cm at 10 cm from ground level) were cut
back to 50 cm stumps. The next four harvests produced no significant differences in
Leucaena biomass between trenched and nontrenched plots, presumably because the tree
roots in the trenched plots had re-grown into a new configuration that exploited a
sufficient soil volume. The plots were re-trenched to 50 cm in late August 1994. The
loss of small stem and leaf biomass due to the effect of trenching over the next four
harvests during a seven-month period was 1,561 kg per hectare (a=.05). The only time
trenching affected height growth of Leucaena was during the first two harvests after the
second trenching, when stems cut from trenched plots were about 10 cm shorter than
those cut from nontrenched plots. By June 1995, the trenched trees had again recovered
and there were no significant differences during the final three harvests.

98
Overgrown hedgerows cut back
Harvest Date
Figure 5-29: Leucaena small stem and leaf biomass from trenched and nontrenched
plots, May 1991 through September 1995.
Conclusions
Installation of root barriers appeared to mitigate soil water competition between
Leucaena hedgerows and adjacent maize under certain circumstances, thereby increasing
grain yield compared with nontrenched hedgerows. Installation of the barriers also
decreased Leucaena biomass production. However, the effects of barriers on hedgerow
biomass production and on maize grain yield were temporary.
Amount and distribution of rainfall and the time elapsed since the installation of
root barriers influenced maize growth and yield and the relative impact of trenching
during the three seasons discussed here. Root barriers were not effective during the fall
1994 season, and there were no differences in maize or Leucaena yield. Adequate
rainfall, as seen in the fall 1994 season, tended to dampen the advantage gained by
trenching. Trenched plots yielded an extra 360 kg/ha maize grain over nontrenched plots
that season, an increase of 18%; Leucaena biomass yield decreased by 1651 kg/ha for

99
that period. During the final season having lower rainfall, but still effective barriers,
maize yield in trenched plots increased by 383 kg/ha (77%) over nontrenched plots.
There were no longer any Leucaena yield losses due to trenching. At Machakos. Kenya,
Ong and Leakey (1999) noted that when rainfall was below 250 mm, maize and
hedgerow trees (Gliricidia sepium and Senna spectabilis) competed with maize for soil
water, but when rainfall exceeded 650 mm the trees and crop used water from the same
profile without decreasing crop yield.
Maize plants in nontrenched plots were fewer in number, shorter, produced fewer
ears, and produced less grain weight per ear than maize in trenched plots. This is
consistent with the effect of water stress during flowering (Olson 1988). Corlett et al.
(1992a) found that flowering of millet next to nontrenched Leucaena hedgerow was
delayed compared to sole cropped millet.
Maize yield was usually depressed closest to the hedgerows, a phenomenon noted
in several studies (Govindarajan et al. 1996, Kamasho 1994, Mugendi et al. 1999). The
causes of maize yield depression, or lack thereof, cannot be determined with certainty,
but were probably due mainly to increased soil water available near trenched hedgerows.
The pruning height of Leucaena was the same for both trenched and nontrenched plots,
hedge prunings were removed from both trenching treatments, and fertilizer application
was the same for both. However, increased soil water availability also means increased
nutrient supply, which promotes increased height growth near hedgerows resulting in
improved access to light. Yield depression still occurred in the 50 cm row in the
trenched hedgerows, so shading probably influenced yield to some degree at that
distance.
Maize yield increased steadily with distance from hedgerows in nontrenched
plots, but in the trenched plots it increased with distance to 100 or 150 cm, then

100
decreased at the 200 cm distance. This could be partially explained by the re-distribution
of Leucaena roots in the alleys in trenched plots, which were fewer in number than those
in nontrenched plots through 150 cm, then increased at the 200 cm distance. Ong et al.
(1991) found that barriers next to hedgerows changed Leucaena root distribution in a
similar way.
The results of this experiment confirm that Leucaena hedgerows compete for soil
water to the detriment of adjacent maize and cast doubt on the utility of root barriers as a
management practice. Their beneficial effect is temporary, and the additional maize
grain realized would probably not pay for the cost of installing the barriers. If farmers
perceive soil water competition to be a major problem it would be worthwhile to look for
a cost efficient replacement for barriers or a less competitive hedgerow species. These
issues are discussed in Chapter 6.
On-Farm Study: Maize Growth at Varying Distances from Hedgerows
Introduction and Objectives
The conclusions drawn from the on-station studies in the previous section of this
chapter are that Leucaena trees in hedgerows reduce grain yield in maize growing in the
adjacent one meter area, that permanent root barriers can temporarily mitigate the loss of
maize grain yield by reducing soil water competition between the hedges and the maize,
but that within a couple of years Leucaena roots will grow under the barriers and
eliminate their favorable effect on maize yield.
An appropriate next question would be are these results of any use to Haitian
farmers, whose plots are mainly on steeply sloping hillsides? This is both a technical and
a socioeconomic question; the socioeconomic issues will be discussed in Chapter 6.
Before that, the present section of this chapter will address mainly the technical ones and
serve as a bridge between the on-station and on-farm work.

101
The objective of the research supporting this section is limited to showing the
development of maize growth and yield on some sloping hedgerow gardens built by
Haitian farmers participating in the PLUS project. The logical design to complement the
on-station work should have been to install barriers on hillside farms between hedgerows
and adjacent maize to investigate whether this practice would economically increase
maize yield to farmers. However, such an attempt was not successful.2 Instead, data
were collected from an on-farm trial comparing three different types of soil conservation
structures with traditional cultivation to examine how competitive interactions between
trees and crops differ between vegetative and nonvegetative soil conservation practices
on sloping land. The objective was to determine if plant development was affected in
maize rows adjacent to soil conservation structures on sloping land, in plots where
farmers were doing the cultivation.
Study Site
The four farm plots, referred to as gardens (jaden), from which the measurements
were taken are located in the lower central plateau near the town of Lascahobas, about 55
km north and east of Port au Prince and 23 km west of the border between Haiti and the
Dominican Republic. These gardens had been part of a larger national study being
carried out by SECID and PADF under the PLUS project to compare yields of crops
grown in various kinds of soil conservation structures. The national study was
abandoned in 1996 because of funding problems, but the four gardens in Lascahobas still
contained the treatments intact in spring 1997 when this research was done. The soils are
2 The research as originally planned included a series of Leucaena hedgerow/maize
water competition experiments on 30 farms in the south of Haiti as a complement to the
on-station trials. This proved infeasible due to political and logistic problems, and was
terminated. However, a group of eleven farms in the south of Haiti participated in
preliminary work in the spring of 1991 and some of the information gathered from it is
used in this section.

102
calcareous, with slopes ranging between 20 to 45 percent. Three of the plots were one
hectare in size; the fourth was 1.3 hectares. The treated portion of the gardens occupied
between 12 to 25 percent of the total garden areas. Rainfall in this area is bimodal,
falling mainly in spring and fall. In 1997, rainfall during May through August at the
nearest rain gauge was 723 mm, with rain falling on 48 days during that period.
Methods
The study was a complete randomized block design with three kinds of soil
conservation structures as treatments, plus a control without soil conservation. The
subplot treatments were Leucaena hedgerows; crop bands constructed of sugarcane,
pineapple, and cassava; and rock wall barriers. The control plot had no soil conservation
structures, and represented traditional maize cultivation practice for this area. The three
soil conservation structures and control plot occupied adjacent subplots, assigned
randomly. Each of the four gardens was considered a block. The gardens were rented
from the owners, and PADF technicians supervised installation of the structures and
collected all data. The owners planted and managed crops in the alleys and kept the
harvest. Treated subplots contained four rows built on the contour in April 1993 at a
distance of 10 m between rows, each row being 10 m in length. In two of the gardens the
contour rows were oriented in north-south direction, and in the other two in east-west
direction. A local maize variety was planted in mid-May 1997. The Leucaena
hedgerows were pruned to a height of 50 cm on 28 April, two weeks before the maize
was planted. It was not pruned again before final maize harvest, when it had attained a
height of about 1.5 to 2 m. The sugarcane was the tallest component of the crop bands,
and remained at a height of about 1.5 m during the maize growing season.
Observations were made on the growth stage of the maize on four dates during
the growing season: 43, 49, 56, and 67 DAS. Each stage of growth was assigned a

103
numerical value as follows: vegetative = 1, tassel = 2, anthesis = 3, silk = 4, and ear = 5.
A line transect of observations was taken through the three alleys, perpendicular to the
contour rows, from top to bottom at the midpoint of each subplot. One maize plant was
observed at specific positions along each transect: 1 m below the uphill structures, mid¬
way between structures, and 1 m above the downhill structures. Although the control
subplots had no soil conservation structures, maize growth stage observations were taken
at corresponding positions to those in the treated subplots.
Final harvest was taken in mid-September, about 116 DAS. Whole maize plants
were harvested in transects 5 m wide by 25, 26, or 28 m long through the middle of each
subplot passing through, and perpendicular to, three alleys and the corresponding
position in the control plot. Technicians did not record final harvest data by position on
the slope or within the alleys, as was done for the growth stage observations. Weights
were taken in the field on suspended spring scales. All weights were field dry and maize
grain was not corrected to a standard moisture percent, as no drying ovens or grain
moisture meters were available. Analysis of variance was done using SPSS version 9.0.
Results
Maize development was different among soil conservation treatments, but not
among slope positions or within-alley positions. The maize growing in the rock wall and
control plots developed more rapidly than that in the hedgerow and crop band plots at 43
and 49 DAS (Figure 5-30). Maize grown in hedgerows developed less rapidly than that
grown in all other plots at 56 and 67 DAS. The difference in growth stage between
maize in hedgerows and maize in the control plot increased through 56 DAS, and then
decreased between 56 and 67 DAS.

104
Figure 5-30: Maize development in three soil conservation practices and control at four
times during the growing season on four farms during Spring 1997.
The differences in the pattern of maize development among soil conservation
structures did not result in statistically significant differences in yield (Table 5-22).
There were, however, some relatively large numerical differences. There was a wide
variation in maize plant density, which was highest in crop band plots and lowest in
control plots. Stover weight was greatest in rock wall plots, as was yield of maize grain.
Grain yield was least in the control plot. Harvest indexes calculated with field dry
weights were: hedgerows, 0.25; crop bands, 0.38; rock walls, 0.30, and control, 0.15.
Discussion
The pattern of maize development in the alleys between hedgerows and crop
bands did not reflect the yield curve caused by soil scouring and hedgerow/crop
competition as described by Garrity (1996), even though soil scouring due to hoe tillage
is commonly observed in Haiti in the upper parts of hedgerows. Three row-by-row
sorghum and maize harvests on sloped farms done as preliminary work by the author in
1991 also produced no clear pattern of yield variations with respect to alley position in
Leucaena hedgerows. However, the overall development of maize in the soil

105
conservation structures built of plants, i.e., hedgerows and crop bands, was slower than
that in rock walls and in the control plots. This could be due to competition between
hedgerow plants and adjacent maize. The differences in maize development between the
four treatments diminished over the growing season.
Table 5-20: Maize yield in three soil conservation practices and control at 116 DAS on
four farms, spring 1997
Hills
(no/ha)
Plants
(no/ha)
Stover
(kg/ha)
Ears
(no/ha)
Ears
(kg/ha)
Grain
(kg/ha)
Hedgerow
6,745
13,700
714
9,944
287
200
Crop Band
7,446
15,893
546
9,318
255
222
Rock Wall
7,774
13,399
902
9,893
325
286
Control
6,572
12,426
783
6,994
148
121
P-value
.913
.714
.591
.694
.110
.206
Grain yield was overall poor by the standards of Haitian farmers, but not entirely
out of range. Farmers in two regions of the country estimated they should harvest 40 to
80 marmites (roughly a gallon measure weighing about 2.6 kg) of field dry maize grain
for every marmite planted (Swanson et al. 1993a, 1993e). With one marmite seeding
about 0.32 hectares (at an initial density of 62,500 plants/ha) (Swanson et al. 1993e), this
is equivalent to 325 to 650 kg/ha expected yield. This would be the standard for the
control plot, but since the presence of soil conservation structures every 10 m removes
10% of the available cropping space, farmers’ expectations would be 293 to 585 kg/ha of
maize on plots having hedgerows or rock walls. Taking the lower expected return of 325
kg/ha for a plot without soil conservation (Swanson et al. 1993e), the control plot reached
37%, hedgerows 68%, crop bands 76%, and rock walls 98% of expectations. Although
the differences among treatments were not significant (p = .206), it is tempting to
speculate that maize yield in the control was least because it had the least amount of soil

106
water. Soil water stress reduces harvest index (HI) in maize (Shaw 1988), and the
control plots had the lowest HI. Rock walls produced the highest yield because that
treatment harvested water but did not compete for light, nutrients or water with maize.
Crop bands and hedgerows were in the middle because they harvested water, but did
compete for resources with maize.
Better yield from rock wall terraces compared to hedgerows has precedent in
Haiti, but gradually the biomass contributed by trees changes the yield ranking in favor
of hedgerows. An on-station experiment on 17 to 21% slopes (Isaac et al. 1995) reported
that rock wall plots produced greater maize yields than hedgerows in the first year after
establishment, equaled the hedgerow treatments during the second year, then yields
steadily declined in rock walls but remained steady in hedgerows. Isaac et al. (1996b)
reported that during two years and four maize harvests there was a slight increase in yield
in all Leucaena hedgerows, and a steady decrease in all other treatments, including rock
walls. Increased yield in farmer-managed sloping hedgerow plots has also been shown.
Lea (1995e) reported 40% greater sorghum yields in hedgerow plots compared to
untreated control plots in experiments done in two regions of Haiti, including farms close
to those reported on here. In other regions having different soils and rainfall distribution,
hedgerow plots did not show improved yield over traditional cultivation (Lea 1995e).
However, on-farm experimental results can be complicated by management
factors outside the intended experimental design. An on-farm study in the same region as
the plots in this experiment that reported 70% higher sorghum yields from rock wall
terraces compared to control plots without soil conservation (J.D. Lea, personal
communication. Sept. 1995), was followed up by this author. The follow up showed that
the sorghum harvest had been measured correctly by technicians, but the farmers who
owned the plots had managed the protected plots differently than the control plots, which

107
were not physically adjacent or always physically the same as the treated plots. It was
found that the rock walls served to stop sorghum seed from washing out of the plots
during heavy rainfall. These seeds sprouted above the structures and the farmers later
transplanted the seedlings back into the alleys, resulting in a higher plant density
compared to the control plots. In addition, farmers practiced better, deeper, tillage on rock
wall plots in some cases, tended to weed them earlier than control plots and were more
likely to thin hills (planted 10 to 15 seeds/hill) after germination in the rock wall plots,
resulting in larger panicles. This points out a major difference between farmer-managed on-
farm trials and researcher-managed on-farm trials, at least for treatments involving
relatively complex inputs such as soil conservation structures. Farmers adjust their
management intensity according to their investment in the gardens and the economic
potential of the gardens, whereas in researcher-managed trials this difference is eliminated.
Conclusions
Conclusions based on these data are limited to differences in the pattern of maize
plant development among soil conservation treatments. There was a greater variation in the
pattern of maize development between type of soil conservation structure than in relative
nearness to the structure or to position of the alley on the slope. Lack of competition
between plants serving as the structural components of hedgerows (Leucaena) and crop
bands (sugarcane and pineapple) is the probable cause of faster maize development in rock
wall terraces and untreated control plots. The numerically smaller yield in the control plots
compared to the plots having soil conservation structures, although not statistically
different, is probably due to the water harvesting effect of the contour structures. A much
larger sample size and row by row harvest over several years would be required to draw
clearer conclusions regarding the effect of slope position within alleys in plots with such
great variability in soil type, soil fertility, soil depth, and rainfall regime.

CHAPTER 6
LAND AND HOUSEHOLD CHARACTERISTICS IN RELATION TO
ADOPTION AND MANAGEMENT OF HEDGEROWS
Introduction and Objectives
The two previous chapters described research regarding competition between trees
in hedgerows and adjacent crops, with a focus on soil water. If soil water competition were
an important factor in reducing crop yield in hedgerows, then farmers might consider it in
their adoption and management decisions on farm. Results from the on-station research
described in Chapter 4 showed that soil water competition between hedgerow trees and
maize can reduce maize yield under certain rainfall conditions. The on-farm research in
Chapter 5 was not able to show the same phenomenon at work under the highly variable
environmental conditions found in hillside farm plots, but did show differences in maize
growth between different types of soil conservation structures. Those differences in maize
development may have been due to differences in the way soil conservation structures
exerted competitive pressures on the crop.
The research described in this chapter focuses on how farmers adjust their adoption
and management decisions based on the technical aspects of soil conservation structures, the
land and human resources available to the adopting household, physical and tenure
characteristics of the plot, and farmers perceptions of potential benefits and problems
associated with the soil conservation practices. The links between the work described in
this chapter and the on-station and on-farm studies are: (1) hedgerows remain the principal
focus of the research, and (2) farmers were asked their opinion regarding the severity of
108

109
tree/crop water competition in hedgerows to explore whether the degree of that
perception was correlated with management practices. However, the adoption and
management questions explored here included farmers’ responses to other soil
conservation structures as well as hedgerows, and this proved to be useful in revealing
differences in farmers’ decision-making strategies. The hypotheses tested in this chapter
are
• Farmer household characteristics and farm resources influence the adoption
and management of hedgerow intercropping.
• Plot characteristics influence the adoption and management of hedgerow
intercropping.
• Problems and benefits with hedgerow intercropping as perceived by the
adopting farmer influence management quality.
This study seeks to discover relationships between farmers’ decision to install an
agro forestry intervention and three groups of variables: (1) characteristics of the
technology, represented by the decision to install either hedgerows or other agroforestry
practices (crop bands, rock walls, gully plugs, or trees) with different characteristics; (2)
farm household characteristics representing wealth, education, and household size; and
(3) physical characteristics of the plot and the mode of access through which the farmer
worked it. Having installed a particular practice on a plot, some measures of
management quality of the soil conservation practices are compared to the same three
groups of variables. Finally, hedgerow management quality is compared to farmers’
opinions regarding the benefits and problems associated with the practice.
Studies on the adoption of soil conservation technologies began in the 1950s
(Ervin and Ervin 1982). Study results predict whether farmers adopt, the rate at which
they adopt, or identify the characteristics of adopters compared to non-adopters. This

110
information is used to design interventions (Bonnard and Scherr 1994), evaluate policy or
adjust extension messages (Earle et al. 1979, Murray 1980), or to report on and evaluate
extension projects (Pierre et al. 1995, Scherr 1994, Shultz et al. 1997, Smucker 1988,
Sunderlin 1997, Swanson 1993, White 1992a).
Adoption studies generally followed the model described by Rogers ( 1995), that
diffusion of adoptions is a process having four components: (1) the innovation. (2) the
social system in which the innovation is being adopted, (3) channels of communication
(the ways that adopters pass on information about an innovation), and (4) the time over
which a social system adopts an innovation. Initially adoption models were static, with
the underlying assumptions that the adoption decision is binary; there is a fixed, finite
ceiling on adoption; the rate and degree of diffusion through a population is fixed; the
innovation is not modified once it is introduced and its diffusion is not affected by other
innovations; there is one adoption per adopting unit; and the geographical boundaries of
the adopting social system stays constant over the diffusion process (Knudson 1991).
However, some of these assumptions are not useful in understanding the adoption of a
complex technology such as hedgerows. Adoption researchers began to recommend
dynamic models that separated determinations of diffusion over time as the technology
being investigated was affected by other technologies and as the adopting population
changed in various ways (Knudson 1991), and to integrate physical, economic, and other
factors (Femandez-Comejo et al. 1994).
Feder et al. (1985) defined final adoption of agricultural technology, at the level
of the individual farmer, as the degree of use of a new technology in long-run
equilibrium when the farmer has full information about the new technology and its
potential. They recommended studying both the extent and intensity of use of the new

Ill
technology throughout the adoption process to allow for changes in parameters affecting
farmers’ decisions. Specifically for hedgerow adoption studies, Dvorak (1991) suggested
dividing hedgerow adoption into three phases (establishment, maintenance, and
productive) to avoid pronouncements of success or failure based only on the initial
establishment.
Some agroforestry adoption studies are regional overviews based on cost/benefit
analysis (Current and Scherr 1995), but most are based on case studies using
combinations of farmer interviews and plot monitoring (David 1995, Dvorak 1991,
Fujisaka et al. 1995, Murray 1980, Neef et al. 1996, Smucker 1988, Wiersum 1994) or
questionnaire surveys (Kessey and O'Kting'ati 1994, Lea 1994, Pierre et al. 1995, Singhai
and Kumar 1997, Swanson 1993, Vaval et al. 1997). Usually more than one method of
data collection is required to interpret the meaning of adoption (Dvorak 1991, Sunderlin
1997, White 1992a). Once the data is collected, analysis methods vary from multiple
regression (Ervin and Ervin 1982), logistic regression (McNamara et al. 1991, Norris and
Batie 1987, Sureshwaran et al. 1996), ethnographic interpretation (Murray 1980,
Smucker 1988, Swanson 1993), or cost/benefit analysis (Nelson et al. 1997).
Adaptability analysis is another methodology specifically developed to incorporate the
wide diversity of biophysical and socioeconomic conditions found on developing country
farms. It uses regression techniques, but has a focus on finding technology solutions for
multiple environments rather than simply defining and describing adopters and non¬
adopters (Hildebrand and Russell 1996).
Variables Used to Predict Adoption
The dependent variables used to represent adoption of soil conservation practices
are commonly the binary decision to adopt or not adopt, the number of practices adopted,

112
and estimated rates of erosion (Ervin and Ervin 1982). The author did not encounter any
studies that used the quality of management of soil conservation structures as a
dependent variable. The independent variables used to make the predictions/descriptions
can be economic characteristics of the technology (Byerlee and Hesse de Polanco 1986,
Jarvis 1981, Stark 1996), socioeconomic characteristics of the farm or farm household
(Earle et al. 1979, Harper et al. 1990, Kessey and O'Kting'ati 1994, Shultz et al. 1997,
Singhai and Kumar 1997, Stark 1996, Sunderlin 1997, Sureshwaran et al. 1996, Vaval et
al. 1997, White 1992a), physical or tenure characteristics of the plot (Smucker 1988.
Sureshwaran et al. 1996, Vaval et al. 1997, White 1992a), or farmer attitudes regarding
actual or potential benefits or problems associated with a technology (David 1995, Earle
et al. 1979, Fujisaka et al. 1995, McNamara et al. 1991, Norris and Batie 1987, Wiersum
1994). A combination of these is the norm. Findings from studies using these categories
of variables are discussed below.
Technology Characteristics
Farming technologies differ in several ways, including complexity, adaptability,
cost of installation and management, potential return from the investment, timing of the
return, and the inputs (land, farmer knowledge, labor, chemicals, seed) involved. Some
technologies are simple, relatively low in initial cost, and have a short return on
investment (e.g., a new variety of crop). Others are complex, involve relatively high
investments in installation and management, and can take several years before benefits
are realized (e.g., hedgerows). The success of some technologies depends on the
adoption of others.
A study of an improved barley variety and attendant inputs in Mexico found that
farmers rarely adopted complete technology packages, but adopted them in sequence

113
based on predictions of profitability and riskiness. Most adopted one component at a
time, those giving the highest return on capital were adopted earliest (Byerlee and Hesse
de Polanco 1986).
In Kenya a study of 3,000 farmers found they decreased risk associated with new
agroforestry practices through incremental adoption and adaptation, and cost and risk-
reducing modifications to the technology design (Scherr 1995). The authors noted that
farmers adopt agroforestry practices only when they bring economic gain. This is a
widely noted, and hardly surprising, observation. Soil conservation programs in
developing countries have not had much success because they are often too costly to
implement and maintain, so only the better-off farmers are be able to adopt the best
practices, poor farmers have to modify them to reduce the cost (Napier 1991). In
Rwanda, where a high population of farmers intensely crop marginal soils on steep
slopes, soil conservation practices proposed to them were not attractive because of the
time lag between initial investment and return (Steiner and Scheidegger 1994).
Successful alley cropping adoption in Sri Lanka depended on profitability, time to profit,
labor requirement and seasonal availability (Nuberg and Evans 1993).
Farmer Household Characteristics
Household characteristics are labor, land, and knowledge resources that influence
adoption of technologies. These characteristics are not consistent in the way they
influence adoption, but vary according to the technology. Problems occasionally arise
because the correlations between some household variables can be influenced by other
variables so that the correlation includes the spurious effect of the others (Feder et al
1985). In Africa, where larger farm size and greater extension contact were important to
adoption of alley cropping, both were highly correlated to level of education (Tripathi

114
and Psychas 1992). Ervin and Ervin (1982) rejected farmer age as a household variable
in a study of adoption of soil conservation in Missouri, USA, because of a high
correlation with other independent variables.
A study of an integrated pest management (IPM) practice in Florida, Michigan,
and Texas in the US found a significant and positive relationship between adoption and
farm size and availability of family labor; a significant and negative correlation between
adoption and the importance of livestock (Femandez-Comejo et al. 1994). Farm size and
income had a significant and positive impact on soil conservation expenditures in
Virginia; off-farm employment, debt level, and tenure had significant and negative
impacts (Norris and Batie 1987). In a soil conservation project in El Salvador, adoption
of soil conservation and agroforestry practices was greater among younger than older
farmers, among land owners than tenants, and increased with more extension visits
(Shultz et al. 1997). A model based on a survey predicted intention toward soil
conservation to be stronger as farm size, family income, and education level increased
(Earle et al. 1979). A study of adoption of sweep nets (IPM) in the US found that higher
education was correlated with lower adoption, but farm size was not a significant factor
(Harper et al. 1990).
A study of agroforestry adoption (tree and grass associations) in Tanzania found
farm size, household size, number of animals, and distance traveled in fuelwood
collection, were significantly correlated with adoption. Number of days used for fodder
collection was not significant (Kessey and O'Kting'ati 1994). Farm size, household size,
and total number of animals were positively correlated with adoption of silvopastoral
associations in the Garhwal Himalaya region of India. Age of head of household and

115
distance traveled for fuelwood collection were negatively correlated with adoption
(Singhai and Kumar 1997).
A survey in the Philippines found the main constraint to hedgerow adoption was
high labor demand for installation and maintenance. As the number of persons per
household decreased and the percentage of female members increased, adoption
decreased; women did not usually help in hedgerow establishment (Stark 1996). A
model of hedgerow-based soil conservation adoption in the Philippines suggested that
government assistance, land size, farmer age, land intensity, and tenure impacted
adoption, but income and education did not (Sureshwaran et al 1996).
Plot Characteristics
A commonly held opinion is that farmers will invest more resources on a plot of
land that is held under a secure form of tenure. It has been recommended in some areas
that alley cropping should be targeted first to farmers who own inherited or purchased
land (Tonye et al. 1994). Other researchers note that tenure insecurity, but not lack of
title itself, could be a disincentive to adoption of agroforestry practices (Current et al.
1995). Apparent effects of tenancy, however, could be caused by indirect relationships
between tenure and access to credit or other inputs (Feder et al. 1985). In West Africa
tenure plays a significant role in adoption of alley cropping, land tenure security is
favorable to it (Lawry et al. 1994). In Benin, tenants, the landless, and most women are
worse off than landowners in their ability to adopt agroforestry practices because of
tenure insecurity among other reasons (Neef and Heidhues 1994). In the US, land tenure
was found to be not significant to IPM adoption because it was an investment in human
capital, not in the land (Femandez-Comejo et al. 1994)—so the impact of tenure depends
on the kind of technology.

116
Physical characteristics of the plot sometimes impact adoption by themselves, but
they usually interact with other characteristics. Size of the farm holding could be a
surrogate for other factors, such as access to credit, information or wealth (Feder et al.
1985). In Benin, there was a significant difference between owned and leased fields with
regard to distance from the compound (Neef et al. 1996). An adoption study in Indonesia
found that non-adopters had significantly less area in sloping land than adopters, but the
slope percent of the sloping land was not significant (Fujisaka et al. 1995). Another
study in Benin found the key parameters determining adoption of farming technology
included the function and history of the field, tenure, and the field’s position with regard
to fertility flows in the farming system (Koudokpon et al. 1994).
Farmer Attitudes Regarding Problems and Benefits of the Technology
Perceived problems and benefits are sometimes reported as observations and
sometimes used as independent variables in adoption models. However, it can be
difficult to reconcile expressed attitudes with farmer actions. Dvorak (1991) reported in
her survey farmer responses about alley cropping were generally favorable, but in no case
had a farmer, household member or neighbors extended an alley or planted a new alley
farm. She noted that two apparently enthusiastic farmers even uprooted the alley trees,
and concluded this reflected the difficulty in using a one-time survey to evaluate a
complex system (Dvorak 1991).
Nevertheless, farmer attitudes can help understand adoption. Farmers’ expressed
benefits of hedgerows in Indonesia were decreased erosion, flattening alleys for
cropping, and ability to use fertilizers without loss; drawbacks were animal damage.
Most important reasons for non-adoption were work demands on lowland plots, high
labor in the hedgerows, off-farm work opportunities, lack of draft animals, and lack of

117
capital for labor and inputs (Fujisaka et al. 1995). A model based on survey data
predicted intention toward soil conservation was stronger as perception of soil erosion as
a problem increased (Earle et al. 1979). Perception of erosion did have a significant and
positive impact on soil conservation expenditures in Virginia (Norris and Batie 1987).
Farmer attitudes may be contrary to the expectations of researchers. After 12
years of extension education in Georgia with peanut farmers showing increased income
and environmental benefits, only 31% of farmers adopted IPM. The study found
promoting the environmental impact of IPM did not affect adoption (McNamara et al.
1991). In Rwanda, small farmers rarely mentioned erosion as an important production
constraint (Steiner and Scheidegger 1994). In a risky production environment in Kenya,
where crop yields fluctuate constantly with the amount of rainfall, soil fertility was not an
urgent concern to farmers (David 1995). Study of hedgerow adoption in Indonesia noted
that some farmers adopted not because of its productive benefits, but as a means to gain
access to land or credit, or to demonstrate allegiances to social networks (Wiersum
1994).
Findings from Adoption Studies in Haiti
The independent variables used in this study, and others that were not included,
have been used by other researchers in Haiti. Smucker (1988) found that farmers
planting tree seedlings distributed by a project were older, more protestant, more often
married, and better schooled; had larger households and more children; were more
involved in wood-related occupations and more likely to own large animals; hired more
labor and sold their own labor less; and had more land and more securely-held land, and
had more land in fallow than did nonplanters. White (1992a) found that religion, age.

118
and wealth were not significant to participation in group soil conservation practices in
Haiti, but that group membership and cooperative labor tendencies were very important.
Both White (1992a) and Vaval et al. (1997) studied soil conservation and
hedgerow adoption in the Maissade region of Haiti and found that land tenure of plots
was not important to participation in cooperative watershed management or soil
conservation adoption. There were, however, differences of plot tenure among farmers
of different economic levels within adopters and non-adopters (Vaval et al. 1997).
Haitian farmers’ perceived problems and benefits associated with hedgerows have
been surveyed in several studies. The most frequently cited problem was animals
destroying the hedgerows, especially during the period after crop harvest when goats are
let free to forage (Pierre et al. 1995, Swanson et al. 1993a, Swanson et al. 1993b,
Villanueva 1993). Animal problems and mode of access to the plot can be related.
Farmers in the northwest of Haiti reported to Swanson et al. (1993a) that, on inherited-
undivided plots, they often tie animals nearby their hedgerows to permit direct grazing
because other family members will do so in any case. Plot characteristics and labor
required to install hedgerows are also cited as problems. White (1992b) reported that
39% of farmers not adopting hedgerows and other soil conservation practices cited lack
of time as the reason, 21% said it was due to not owning land, 17% said they owned
inappropriate land. A study following up a terminated soil conservation project in the
south of Haiti cited farmers as saying hedgerows were too difficult to manage, and that
they could not extend them because it was too expensive and took too much time and
labor to repair them (Villanueva 1993). Only one study reported farmers’ perceived
benefits. Over 80% of 105 farmers interviewed in the south of Haiti said that hedgerows
increased crop production by 25 to 50%, and some said they felt more secure in their

119
access to the plots because of the positive reaction of the landlords to hedgerow
construction (Pierre et al. 1995).
This literature review has shown that variables representing characteristics of the
technology, farm household resources, plot characteristics, and farmer attitudes toward
the technology are sometimes useful in predicting adoption. However, the predictive
utility of each variable is case specific. A variable might be useful in predicting adoption
in one case but not another, or the influence of the variable might be positive to adoption
in one case and negative in another. The most consistent variables influencing adoption
of hedgerows appear to be economic gain and lag to profit, and threshold cost of
installation, especially labor cost. Farmers are more likely to adopt when it makes
economic sense to do so.
The studies done on adoption of soil conservation in Haiti included mode of
access to the land as a variable. In some cases the conclusion appeared to be that land
tenure was important, but that it was not in other cases. This apparent contradiction
probably means that it was not important to a farmer’s participation, in some form, in a
project, because at least one of the farmer’s plots is likely to be suitable for some activity
that counts as participation. However, tenure could be important in deciding what
technology to adopt on a particular plot. It is noted that no other studies were discovered
that used management quality of installed agroforestry practices as a dependent variable.
Materials and Methods
A survey of farmers participating in the PADF PLUS project was carried out in
spring 1996. The objectives of the survey beyond those of interest to this research were
explained in Chapter 4. Sampled farmers were those who had first participated in the
project before 1 January 1995, and who, therefore, had been working with the project for

120
at least 16 months at the time of the survey. In each of the four project regions (Camp
Perrin in the southwest, Marigot/Palmiste á Vin in the southeast. Cap Haitien in the
north, Mirebalais in the lower central plateau), we selected 35 participating farmers from
the geographic areas of responsibility of eleven PADF agronomist technicians by using a
random number list to pull farmer dossiers from the regional files. The technicians
visited 1,540 farmers, 5.6% of all eligible farmers. During the whole period of data
collection, other project staff carried out a continuous check of the technicians’ work,
rotating among the four PADF/PLUS field regions. Data cleaning was performed by
sorting variables in the database, and checking suspect data with original questionnaires.
The Household Questionnaire
The survey data were recorded on two questionnaires, one completed in the home
of the sampled farmer and one in the garden plots. Technicians completed a two-page
household questionnaire recording a description of all plots owned and cultivated by the
household during 1995, and a description of all the members of the household (Appendix
D). Plot information recorded included distance from the residence (home garden,
nearby, or distant), topographic position (top of slope, mid-slope, foot of slope, or plain),
and slope steepness (flat, sloping, very steep). The size of the plot and the mode of
access (purchased, inherited and separated among siblings, inherited and unseparated,
share-cropped, rented from a private party, rented from the state) under which the plot
was held were also recorded. For each individual member of the household, age, sex,
highest level of education, kinds of PLUS project participation, and kinds of economic
activities participated in were recorded. Finally, the head of household was asked
whether he or she was originally from the locality, and if not, when did he arrive there.
The farmer being interviewed defined the composition of the household and identified

121
who was the head of the household. The information recorded on the household
questionnaire was collected during an in-home interview.
The Plot Questionnaire
During the home interview, the farmer was asked to indicate in which of the plots
listed on the household questionnaire he or she had installed PLUS-inspired agroforestry
practices. The farmer and the technician then visited those plots. A total of 2.295 plots
were visited, an average of 1.5 plots per farm. The technician recorded information on
the physical properties of the plots, the yield of crops during the previous twelve months
and the average yield before soil conservation structures were built, and management and
attitude information for each of the PLUS agroforestry interventions in the plots
(Appendix E). Data collected about the physical characteristics of the plot included the
fallow status of the plot, slope (taken by the technician with a plumb line clinometer
attached to a clipboard), slope aspect, an estimate of the elevation of the plot to within
100 m, the number of erosion gullies greater than 20 cm depth, parent material of the soil
(basaltic or calcareous), and the local name for the soil type on the plot. The number of
trees per hectare (of any species from any source) whose trunk diameter was greater than
10 cm at 1.3 meters from ground level (DBH = diameter at breast height) was recorded.
These trees were counted by the technician conducting the interview, using a plywood
caliper with a 10 cm opening to define countable trees. The distance in minutes from the
interviewed farmer’s house to the plot was estimated. The farmer was then asked four
qualitative questions about the soil on the plot; the answers were recorded as a number
between one and five. These were regarding soil fertility (1 = infertile, 5 = fertile), soil

122
depth (1 = shallow, 5 = deep), the “hotness” or “coldness” of the soil1 (1 = hot. 5 = cold),
and the degree of erosion on the plot (1 = no erosion, 5 = severe erosion).
The farmers’ qualitative analyses of soil fertility were followed up with a
laboratory analysis. Soil samples were taken from 175 gardens selected at random from
all gardens visited, 35 each from gardens having fertility classes 1 through 5. The
samples were taken by technicians using a machete from the 0-20 cm layer; a composite
of four positions for each sample. They were put into ziplock bags and labeled, then
transported to the University of Florida for laboratory analysis. Complete methodology
and results for the soil analysis are discussed in Appendix F.
Farmers were asked about the yield of the principal crops they harvested during
1995 and an average yield of the same crops before soil conservation structures were
installed on the plot. This information was analyzed by SECID for reporting on PLUS
project impact to USAID, and so was not used by the author in this study. Questions
about the number of family members working on the plot, number of person days of
labor from outside the family used on the plot, and the kinds of agroforestry and soil
conservation structures installed on the plot completed the general questions. Then
followed a separate page of questions for each of seven possible kinds of PLUS project
activities installed on the plot: hedgerows, crop bands, rock walls, gully plugs, trees
planted on the plot during 1995, top-grafted fruit trees, and vegetable gardens. The
appropriate page was used according to the types of structures or techniques used on the
plot. The questions asked on these pages varied with the technique. For the soil
'Most writers agree that “hot" soils (CHO) are dry, well-drained soils and “cold” soils
{FWET) have more soil water available for plants. Other concepts are integrated into this
system as well, including soil parent material, slope, orientation, and vegetative cover
(McClain and Stienbarger 1988, Murray 1981). Some crops grow better in hot soils,
some in cold, but in general cold soils are preferred (Smucker 1981).

conservation practices they included: the amount present as measured by the technician
(number of rows and total length of each row), qualitative assessments (1 to 5 scale) by
the farmer of the importance of a series of potential benefits and problems associated
with the practice, who of the household participated in the construction and repair of the
practice, the number of breaches in the structures over a specified size, whether or not the
breaches were repaired, and a row-by-row rating by the technician of well-managed,
adequately-managed, or poorly-managed. The management ratings integrated the
technicians’ opinion about hedgerow plant vigor, pruning height, density, and number of
breaches.
Agroforestrv Practices Compared on the Plots
Farmers participating in the PLUS project are asked to protect at least one plot of
land with an agroforestry practice before they gain access to crop seed banks operated by
their community-based organization (CBO) with PLUS assistance. The practices are
discussed during training sessions with PADF technicians and CBO extension agents..
The farmer decides what practice to install, and where to install it. It is noted that this
project requirement is probably complied with, in many cases, only to gain access to crop
seeds and tools. Some hedgerows, therefore, have been installed only to satisfy this rule,
by farmers who judged the value of the seed and tool credit to be greater than the cost of
trying the new technology.
Although the central focus of this study was the hedgerow, several other
agroforestry practices were included in the analysis to contrast how farmers change their
installation and management strategies to accommodate characteristics of the technology.
The agroforestry practices vary in their general types. Hedgerows, crop bands, and rock
wall terraces are built across the slope in contour rows. The area between rows, the

124
interspaces or alleys, is planted to field crops. Gully plugs are confined, obviously, to
ravines or gullies. Trees may be planted anywhere. The usual planting configurations
are perimeter plantings, small woodlots, or in widely spaced rows in farmed plots. Top
grafting is done on existing fruit trees, planted before the PLUS project began.
Hedgerows
Hedgerows were defined in Chapter 3. They consist of a single or double row of
densely planted trees (e.g., Leucaena leucocephala, Gliricidia sepium), perennial food
crops (e.g., sugar cane, pineapple), or other perennial (e.g., perennial cotton, castor bean)
planted on the contour as a physical barrier to soil erosion, as a soil fertility enrichment
structure, and a source of diverse products of economic interest to farmers. Annual crops
are planted in the alleys between hedgerows.
Crop bands (barm manie')
These evolved from hedgerows built from sugar cane and pineapple. They are
similar to hedgerows in that they are installed as contour rows of perennial vegetation in
a farmed plot, leaving room for field crops in the alleys between rows, with the goal of
producing useful vegetation and holding soil and water on the plot. The differences are:
1) crop bands have a larger dimension—they are a band of one or two meters width as
opposed to a single or double row of plants found in hedgerows, 2) the perennial plants
that serve as structural components are food crops, such as sugar cane, pineapple, and
plantain, and 3) annual crops are also planted in the band, such as yams and sweet
potatoes. While differences between hedgerows and crop bands are not always
clear—hedgerows can be made with sugar cane or pineapple, for instance, and often have
an annual crop planted on the uphill side of the hedgerow that differs from the field crops
planted in the alleys—crop bands are wider, contain a greater number of crop species,

125
and usually do not contain a woody perennial. The number of crop bands is still small
(4.4% of PLUS plots had them in 1995), but farmers are interested and the practice is
growing.
Rock walls
Rock walls are constructed of fieldstones place in rows across the slope contour.
Two general kinds are built. The most common has a footing dug into the soil, and the
stones are stacked carefully with flat sides and top (mi sek). The other kind requires less
skill; the stones are simply piled in rows along the contour without preparing a footing
(kodon pyé).
Gully plugs
These are constructed across narrow, actively eroding ravines. They are built
from rocks where available (séy woch), but also from stakes cut from trees, interwoven
with smaller branches ( a considerable amount of eroded soil in a short time. The soil trapped by the gully plug
is quite valuable, because it is deep, well drained, and collects water from adjacent slopes
during rainfall. The usual practice is to plant high-value crops uphill from the structures,
often plantains and taro, as soon as enough soil accumulates. They can be costly to
install and maintain, but the income from the crops can be high and a well-protected
ravine does less damage to the farmer’s property.
Trees
Two kinds of trees are referred to in the analysis: those planted on the plot as
seedlings by a farmer during 1995 as a project activity, or mature trees of any species or
source over 10 cm in diameter at breast height.

126
Top-grafted fruit trees
These are mature low-value mango or citrus that were grafted, either by the
farmer or by a project extension agent, with high-value buds.
Statistical Analysis
Data transformation and statistical analyses were done using SPSS version 9.
Statistical tests were considered significant at the 95% level of probability. Because the
distribution of most of the variables recorded on the questionnaires was not normal and
because variables having several factor levels usually had great differences in the
numbers of observations among the factor levels, nonparametric analysis was performed
on the questionnaire data. The Kruskal Wallis test of mean ranks was used instead of
analysis of variance. Because SPSS version 9 is not able to do mean separation after a
significant Kruskal Wallis result, Mann-Whitney comparisons were done on the ranked
mean pairs, followed by a Bonferroni correction (the p-value of the Mann-Whitney tests
were multiplied by the number of factor levels, R. Littel, personal communication,
September 1999) (Ott 1992, Zar 1984). Cross tabulations were tested with Pearson chi-
square; bivariate correlations were tested with Spearman's rho.
Limitations of the Analyses
The regions where the PLUS project was working were selected because they are
smallholder hillside farming areas with above-average agricultural potential. Because
only farmers participating in the PLUS project were used as the sample population,
conclusions based on the data produced by the survey may not apply to Haiti generally,
and they may not apply to farmers not participating in PLUS. It is also noted that the
information regarding participation of household members in PLUS agroforestry
activities and other economic activities was given by one person, usually a male head of

127
household, and may therefore be biased. The analysis presented here is limited in that no
plots were sampled that 1) were controlled by PLUS participants but had no agroforestry
practices installed, 2) were controlled by farmers not participating in the project, or 3)
had PLUS-inspired practices installed by secondary adopters (farmers not receiving
PLUS extension visits or subsidies).
Results and Discussion
Results are presented in three sections. The first section describes the household
members and all the garden plots under household control during 1995 based on the in-
home interview questionnaires, and then the 2,295 plots visited by PADF technicians
during the survey. The plots visited by the technicians were those where interviewed
farmers had installed at least one project-sponsored agroforestry practice.
The second section addresses the decision to install hedgerows and other related
agroforestry practices. Several types of household resources bearing on land, education,
and labor are compared between households that installed and households that did not
install particular agroforestry practices. Physical characteristics of plots having the
practice are then compared to those of the plots where that practice was not installed.
The third section discusses the relationships between the quality of management
applied to agroforestry practices on the plots, and three categories of independent
variables: 1) household resources, 2) plot physical characteristics, and 3) problems and
benefits associated with hedgerows as perceived by the farmer.
Characteristics of the Household and Household Plots
Numbers of family members bv gender
The 1,540 households interviewed reported a total of 8,584 members, of which 52
percent were male and 48 percent female. The mean number of people per household

128
was 5.6. Table 6-1 shows the percent of household members by gender in seven age
classes. Males and females have about the same distributions among the age classes.
Heads of household
Eighty-five percent of heads of household were male; fifteen percent were female.
The mean age of male heads of household was 46 years, and 50 years for female heads of
household. Eighty-four percent of heads of household reported that they were from the
area where they now lived, and sixteen percent moved into the locality from outside.
There was a small but significant age difference between local (46 years) and immigrant
(49 years) heads of household. A higher percent of females (18.6%) than males (15.6%)
were immigrants, but the difference was not statistically significant. The number of
years of schooling for heads of household was 2.7, but it was not distributed normally.
About half of them had no schooling whatsoever; the median number of years was zero.
There was a significant difference in schooling between female and male heads of
household, the mean number of years for females was 1.2, and for males 3.0.
Table 6-1: Age distribution by gender of members of 1,540 farm households
interviewed in the four PADF/PLUS project areas in Haiti, 1996.
Age class (vrs.)
N
Female
Percent
N
Male
Percent
36529
561
13.7
587
13.1
36686
631
15.4
613
13.7
36844
510
12.4
627
14.0
16-20
471
11.5
558
12.5
21-60
1646
40.2
1763
39.4
61-70
197
4.8
232
5.2
>70
81
2.0
99
2.2
Total
4,097
48
4.479
52

129
Years of school for household members
Increased education could facilitate access to extension material, or increase the
likelihood that a farmer might be selected as a project extension agent, and thereby
increase his probability of adopting an agroforestry technology. The mean total
combined number of years of school for all members of the household was 16
(median = 12). The average male household member had gone to school for 3.3 years,
significantly more than the average female household member, who had 2.6 years of
school. The distribution of years of education is better understood when broken down by
age, as shown in Table 6-2.
Table 6-2: Number of years of school by gender in seven age classes of 1,540
households participating in the PADF/PLUS project, 1996
Age class (yrs.)
N
Female
Years
N
Male
Years
36529
509
0.3 ea
524
0.2 e
36686
610
2.0 c
593
1.9 c
36844
502
3.9 b
615
4.0 b
16-20
455
6.1 a
547
5.9 a
21-60
1534
2.5 d
1652
4.0 b
61-70
178
0.7 e
213
1.8 d
>70
73
0.0 f
89
1.8 d
a Numbers followed by the same letter are not significantly different (a=0.05)
For both females and males, there were significant differences among age classes.
For both sexes, the number of years of school increased as age increased until it peaked
in the 16-20 year age class, and then decreased as age increased. When comparisons
were done by gender within each age class, it was shown that females and males through
20 years received the same amount of education. Above 20 years, however, males had
attended school longer than females. The data seem to indicate that, in the past, men

130
received more education than women did, but that has apparently changed for the
sampled population and now there is no difference.
Participation of household members in economic activities
The interviewers asked PLUS farmers if each household member participated in
economic activities other than PLUS soil conservation, tree culture, and crop
improvement practices. A list of categories of activity was produced from the results, but
this list does not include normal agricultural work, childcare, cooking, or home
maintenance. The mean number of economic activities was three per household. Table
6-3 shows the percent of female and male household members engaged in various
categories of economic activity. Small businesses include furniture construction,
tailoring, plowing fields, and other activities requiring specialized training or equipment.
Local official positions include persons elected to government posts and schoolteachers.
Two categories, fishing and factory work, are not shown because they included less than
0.5 percent of household members.
Table 6-3: Percent of female and male members of 1,540 project households who
participated in various economic activities during 1995-1996.
% of Members Participating
Activity
Female
Male
Animal raising
6
25
Marketing agr. products
6
0
Marketing other products
20
3
Small businesses
2
9
Handicrafts
1
1
Agricultural day labor
0.5
1
Charcoal or lime production
0
2
Local official position
1
1
Leaf doctor
0
1
Sawyer
0
2
Other day labor
0
1
Sample size
4097
4479

131
Table 6-3 confirms the commonly made observation that women are more
involved in marketing (Murray 1981) and men in animal raising. The percent of women
involved in marketing, men selling charcoal, and men selling agricultural day labor
appears to be very low. This might be an indication that sampled fanners, being project
participants, were wealthier than average. Table 6-4 presents the same information by
age class. The two youngest age classes are not included because of very low
participation. Only 2 percent of the 6-10 year old boys and 1 percent of the 6-10 year old
girls were active in raising animals.
Table 6-4: Percent of female and male members of 1,540 project households who
participated in various economic activities during 1995-1996, by age class.
Age classes (years)
319
16-20
21-60
61-70
>70
Activity
F
M
F
M
F
M
F
M
F
M
Animal raising
3
9
3
21
12
43
17
56
7
40
Marketing agr. products
0
0
4
0
13
1
7
0
1
1
Marketing other products
2
0
11
0
42
6
34
5
17
1
Nonagr. professions
0
1
0
4
5
18
2
12
0
12
Handicrafts
0
1
1
1
2
2
2
3
0
5
Agricultural day labor
0
1
0
1
1
2
1
3
1
3
Charcoal or lime production
0
0
0
1
0
4
0
3
0
2
Local official position
0
0
0
1
1
3
1
1
0
1
Leaf doctor
0
0
0
0
0
2
2
3
0
5
Sawyer
0
0
0
0
0
5
0
3
0
1
Other day labor
0
0
0
0
1
l
1
2
0
1
Sample size
510
627
471
558
1646
1763
197
232
81
99
Table 6-4 shows percent participation within each age class (i.e., the column
percents). When percent participation within each activity was considered (i.e., the row
percents, which are not shown), it was noted that most of the activities (usually more than
80%) are done by persons in the 21-60 years age class. Animal raising, however, spreads

132
more into the younger and older age classes, especially for males. Table 6-4 confirms
that more boys than girls participate in animal raising, and likewise more old men
participate than do older women. The counterpart activity for women is marketing. They
begin this activity at an earlier age and continue it longer into older age than do men. and
more women than men work in marketing. Men are more active in small businesses,
charcoal and lime production, and sawing wood than are women.
Participation of household members in PLUS agroforestrv practices
A greater percent of males than females participate in the installation and
management of agroforestry practices. The PADF field staff commonly report that
mainly females participate in vegetable gardens, but Table 6-5 shows that males are
equally involved.
As in the case of other economic activities, persons in the 21 to 60 year age class
(usually over 70% of those participating) do most of the agroforestry work. However,
the highest percent participation within an age class is seen in the 61 to 70 year olds,
where 79% of the males and 36% of the females participate in soil conservation
structures on slopes (Table 6-6).
Table 6-5: Percent of female and male members of 1,540 households who participated
in PLUS project activities during 1995-1996.
Activity
Female
Male
Soil conservation on slopes
15
39
Gully plugs
2
7
Vegetable gardens
4
5
Fruit tree grafting
1
3
Tree planting
7
20
Sample size
4097
4480

133
Table 6-6: Percent of female and male members of 1,540 project households who
participated in PLUS project activities during 1995-1996, by age class.
Age classes (years)
11-15 16-20 21-60 61-70 >70
Activity
F
M
F
M
F
M
F
M
F
M
Soil conservation on
slopes
4
9
3
25
28
72
36
79
25
68
Gully plugs
0
1
0
5
3
13
3
16
6
9
Vegetable gardens
4
3
3
3
6
9
4
7
3
8
Fruit tree grafting
0
0
0
0
2
7
5
8
1
7
Tree planting
1
3
2
10
13
37
20
48
7
32
Sample size
510
627
471
558
1646
1763
197
232
81
99
Number and area of total household plots
The mean number of plots that the household worked or owned during 1995 was
3.7, normally distributed. The minimum number of plots was one; the maximum was
twelve. The mean total area of all family plots was 1.7 hectares. Plot area data are
skewed, with a small number of farmers having a large number of hectares. The
minimum total size of all plots taken together was .08 hectares, the maximum was 20.7
hectares per family.
Farmers were asked to describe each plot, with respect to distance from the house,
as a home plot, a near plot, or a distant plot. These are qualitative categories as defined
by the farmers. The total number of hectares per household in each type of plot is shown
in Table 6-7. The areas are summed for all plots of a particular type controlled by the
household.
The home plot area was smallest, with more land being held in nearby plots, and
the most in plots further from the residence. The sum of the means in Table 6-7 is larger
than the mean value of 1.7 hectares of total household plot area because the distributions

134
of the sizes of each plot type, considered separately, are skewed, so the means are not
good predictors of the average value. Each distribution has a number of extreme values
(plots much larger than average). The sum of the median values (1.5 ha) is close to the
1.3-hectare median value for the total household plot area. The average size of each type
of plot (as opposed to the total amount of land area in all plots of each type as shown in
Table 6-7) is shown in Table 6-8.
Table 6-7: Average total area of three categories of plots held by 1,540 project
households based on the relative distance from the plots to the residence.
Mean (ha)
Median (ha)
Home plots, total area
.45
.32
Near plots, total area
.78
.49
Distant plots, total area
1.12
.71
Table 6-8: Mean plot area of three categories of plots held by 1,540 project households
during 1995, based on the relative distance from the plots to the residence.
Mean (ha)
Median (ha)
N
Home plot
.42a
.32
1173
Near plots
,40a
.32
2134
Distant plots
,53b
.32
2349
Distant plots were significantly larger than home plot or near plots. A previous
study (McLain and Stienbarger 1988) had shown that farmer estimates of plot areas in
Haiti were inaccurate and showed a poor relationship between the estimates and actual
plot size. Therefore, size of each plot type was computed for each PADF region
separately (Table 6-9). The same relationship between plot size and distance from the
residence was found in the subsets of plot size data for three of the four regions.
Therefore, the hypothesis that farmers are able to estimate plot size relatively, if not

135
absolutely, is strengthened. The distant plots were significantly larger than the other two
plot types in Regions 1, 3, and 4. In Region 2, the mean area of the near plots was
smaller than that of the other plot types.
Table 6-9: Mean plot area of three categories of plots held by 1,540 project households
during 1995 in four PLUS project regions, based on the relative distance
from the plots to the residence.
Region 1 Region 2 Region 3 Region 4
ha.
N
ha.
N
ha.
N
ha.
N
Home
,34a
298
.37b
238
,41a
310
.55a
327
Near
,36a
467
,31a
595
,41a
629
.58a
443
Distant
,50b
509
.36b
364
,47b
861
,74b
615
Tenure of plots
The particular aspect of tenure of interest to this analysis is the level of
confidence farmers feel regarding their ability to work and eventually harvest agricultural
produce from the plot over time. This is referred to as tenure “security,” although there
is no consistent relationship between different tenure conditions and levels of security.
Tenure categories used in the survey are: purchased by the farmer, inherited and divided
(legally or otherwise) among siblings, inherited and undivided, sharecropped by the
farmer surveyed, rented by the farmer from a private landowner, rented by the farmer
from the state, and plots on which the farmer acts as a caretaker for the landowner. This
study hypothesizes that purchased and inherited-divided tenure categories are generally
more secure than other types.
There are differences in tenure condition among the three types of plots according
to distance from the residence, as shown in Table 6-10. Cell values are percents of the
row totals. There are statistically significant differences among the cell values.

136
Table 6-10: Percent of three categories of plots held by 1,540 project households during
1995 under various modes of access.
Purch
Divid
Undiv
Share
Rent
State
Caretak.
Other
N
Home
48.9
17.2
23.7
2.0
6.5
0.1
1.1
0.6
1174
Near
35.0
13.9
20.9
12.4
15.0
0.3
1.5
1.0
2134
Dist.
36.6
14.1
19.8
11.8
13.4
2.0
0.9
1.4
2350
All
38.5
14.7
21.0
10.0
12.6
1.0
1.2
1.1
5658
The most common mode of access for all three types of plot is purchased,
followed by inherited undivided. The home plot, however, is more likely than the others
to be on purchased or inherited land. It is much less likely than the other two types to be
sharecropped or rented. A plot rented from the state is most likely to be a distant plot.
Overall, 74% of plots are held in primary access mode (purchased or inherited), with
home plots having the highest proportion in primary access (90%) and near plots having
the lowest (70%). The finding that fewer nearby plots than distant plots are held under
primary access may be due to farmers’ having property located too far from the
homestead, and therefore adopting a strategy of leasing distant plots to someone and
renting land closer for their own production (Bloch et al. 1987). Tenure relates to plot
area, topographic position, and slope, as shown in Tables 6-11,6-12, and 6-13.
Tenure and plot area. Table 6-11 shows land area by land tenure category taken
from the household database of all plots worked during 1995. Numbers followed by the
same letter are not statistically different. Numbers of gardens in each tenure category
appear in the “N” column. The mean area of purchased plots is greater than the mean
area of inherited, separated plots. The area of plots rented from the state is greater than
all other plot types except plots where the farmer is acting as a caretaker, which have the
largest area.

137
Table 6-11: Area of 5,660 plots held by 1,540 households during 1995. by mode of
access.
Mean plot area (ha)
N
Rented, state
0.90 a
55
Caretaker
1.22 ab
66
Purchased
0.53 b
2,183
Rented, private
0.42 be
712
Inherited, unseparated
0.41 c
1,189
Sharecropped
0.39 c
565
Other
0.38 c
61
Inherited, separated
0.36 c
829
All
5.660
Tenure and topographic position. A cross tabulation of land tenure by
topographic position of the plot shows similar frequencies for all tenure types except for
plots on land rented from the state and plots where the planter acts as caretaker (Table 6-
12). State land is much more likely to be in the top slope position than all other tenure
types, and much less likely to be on the foot slope or plain position. The opposite is true
for caretaker plots, that are much less likely to be on the top of slope position, and more
likely to be on a plain. Units are percents of row totals.
Table 6-12: Percent of plots held by 1,540 households during 1995 in eight mode of
access categories by slope position.
Top slope
Mid slope
Foot slope
Plain
N
Purchased
12.7
43.8
25.5
18.0
2180
Inh,sep
14.3
46.6
24.4
14.8
827
Inh, unsep
14.0
46.3
22.9
16.7
1190
Sharecrp
13.1
42.7
26.7
17.5
565
Rent,priv
13.1
42.7
24.3
19.9
712
Rent,state
40.0
45.5
12.7
1.8
55
Caretaker
9.1
40.9
25.8
24.2
66
Other
11.5
36.1
36.1
16.4
61
Total
13.5
44.4
24.8
17.4
5656

138
Tenure and slope class. A cross tabulation of land tenure by slope class shows
results similar to those for topographic position, as it should since topographic position
and slope are related (Table 6-13). Land rented from the state is much less likely to be
flat, and much more likely to be very steep than other tenure types. Again, the opposite
is true for caretaker plots, which are less likely to be on very steep land than are other
tenure types.
Table 6-13: Percent of all plots held by 1,540 households during 1995 in eight modes of
access categories by slope steepness .
Flat
Steep
Very Steep
N
Purchased
18.8a
56.1
25.2
2178
lnh, sep
16.2
58.4
25.4
826
Inh, unsep
17.1
57.7
25.3
1188
Sharecrp
19.3
59.5
21.2
565
Rent,priv
22.2
54.9
22.9
712
Rent,state
3.6
38.2
58.2
55
Caretaker
23.7
57.6
15.2
66
Other
21.3
49.2
29.5
61
Total
18.5
56.7
24.8
5651
a Units are percent of row totals.
Characteristics of the plots having agroforestrv practices
All plots having PLUS practices were visited by PADF technicians, quantitative
data were noted from them. Of the 5,651 plots described as being in their control during
1995, farmers had installed PLUS-inspired agroforestry practices on 2,295 (41%) of
them. On the average, out of the 3.7 total plots controlled by a household during 1995,
1.5 had PLUS interventions. Furthermore, among an average of three PLUS
interventions that a household had on its total plot area, two were soil conservation

139
practices. If we consider only the plots having PLUS practices, each plot had two
agroforestry interventions, one of which was a soil conservation structure.
Distance from the residence. There were small but statistically significant
differences in distance between the residence and the plot, measured in minutes, among
tenure classes2 (Table 6-14).
Table 6-14: Distance from residence of plots having project agroforestry practices held
by 1,540 households during 1995 in the five most common mode of access
categories
Mean distance (minutes)
N
Sharecropped
16 a
189
Rented, private
14 ab
284
Inherited, unseparated
13 b
481
Purchased
13 b
948
Inherited, separated
12 b
324
From the data in Table 6-10, we might have expected purchased plots to be
significantly nearer the residence than plots of other tenure types. This was the case;
rented and sharecropped plots were more distant than purchased or inherited plots.
However, the range of distances is so small that the result is not important. The seven
plots in the sample that were rented from the state averaged 25 minutes away from the
residence.
Tenure and size of plot. Comparison of tenure category by plot area done only on
the plots having PLUS interventions (Table 6-15) agreed with the analysis done on all
plots controlled by the household (Table 6-11). Purchased plots were again significantly
larger than the other tenure types. Inherited, separated plots were the smallest.
2 Three tenure classes having very few observations (State land, 7; Caretaker, 32;
Other, 29) are eliminated from statistical analyses of PLUS plots.

140
Table 6-15: Area of plots having project agroforestry practices held by 1,540 households
during 1995 in the five most common mode of access categories
Mean plot area (ha)
N
Inherited, separated
0.36a
324
Rented, private
0.41a
284
Inherited, unseparated
0.41a
482
Sharecropped
0.42a
189
Purchased
0.54b
948
Tenure and elevation. We estimated elevation within plus or minus 100 meters
above sea level for the PLUS subset of plots (Table 6-16). Elevation did not correspond
to relative topographic position as described by farmers in Table 6-12 (for example, it is
possible to have a high elevation plain), but the results were similar. There were no
significant differences in elevation among tenure categories. This might indicate that the
technician’s estimates were not accurate enough, that farmers’ plots tend to be found
within 100 meters of each other, or a combination of both. Although elevation of plots
rented from the state was significantly higher, there were only seven plots in that
category. State land in PLUS areas was more commonly found in higher elevations.
Table 6-16: Elevation of plots having project agroforestry practices held by 1,540
households during 1995 in seven mode of access categories
Elevation (m)
N
Purchased
339
948
Inherited, separated
340
324
Inherited, unseparated
341
481
Sharecropped
348
189
Caretaker
355
32
Rented, private
363
284
Other
439
29
Rented, state
743
7

141
Tenure and slope. Slope percents measured by technicians on the subset of plots
with PLUS interventions (Table 6-17) shows caretaker plots as having the gentlest
slopes. State land, however, is also shown as being on gentler slopes, which is contrary
to the finding in the cross tabulation done for all household plots (Table 6-13).
Table 6-17: Slope of plots having project agroforestry practices held by 1,540
households during 1995 in seven mode of access categories
Slope (%)
N
Caretaker
22a
31
Rented, state
26b
7
Inherited, separated
31b
324
Purchased
32b
947
Rented, private
33b
284
Inherited, unseparated
33b
482
Sharecropped
34b
189
Other
34b
29
The Decision to Install Agroforestrv Practices
This section considers the 2,295 plots where the 1,540 surveyed households had
installed at least one project agroforestry practice. These plots are referred to as “PLUS”
plots, meaning the practices were installed under the auspices of the PLUS project. The
characteristics of all PLUS plots having a given agroforestry practice were compared to
those of all PLUS plots not having that practice. The plots excluded from consideration
are the 3,356 plots accessed by the households during 1995 that did not have at least one
agroforestry practice installed. Using all household plots in the comparison would have
been preferable, as it would have given a better idea of how farmers make decisions
based on plot criteria, but technicians visited and collected data only from PLUS plots.

142
However, it is still possible to see differences in plot characteristics using only the PLUS
plot data.
Household resources and the decision to install
The independent variables of interest in this section pertain to the amount and
security of access of the land controlled by the household (number of plots, total plot
area, number of purchased or inherited-separated plots, total plot area of purchased or
inherited-separated plots); the number of family members available as labor (total
number of household members, total number of household members of ages of 21 to 60,
number of females of ages 21 to 60, number of males of ages 21 to 60); level of
education (total years of education for all household members, years of education of the
head of household); and the amount of labor used on the plot of interest during 1995
(person-days of agricultural labor purchased in the plot, number of household members
working in the plot).
The reasons for including these variables were that the probability of adoption
might increase as the amount of total land and land under secure tenure increased and as
the number of family members available to work the land increased. Farmers with more
land and labor might be more able to risk land for new technologies and with a lower
cash outlay for labor if family members were available. Level of education was of
interest because more education could allow a farmer easier access to extension material
or increase the likelihood that he or she be selected as an officer in the farmer group or as
an extension agent. Either of these outcomes could increase the probability of adoption.
Households that had installed a particular practice (hedgerow, crop band, rock
wall, gully plug, project trees planted during 1995, top-grafted fruit trees) on at least one
of the visited plots were compared to those who had not installed that practice. Very few

143
of the comparisons between adopting and nonadopting households were significant, and
those that were did not have very large numerical differences. In spite of the small
differences, interesting trends were revealed:
Hedgerows. Households installing hedgerows had 0.3 ha less total plot area than
those not installing (p = .012), and had 0.2 fewer family members working on the plot
during 1995 (p <.001). Both of these findings might appear to be counter intuitive, since
less land and labor available to a household should decrease the likelihood of adoption.
However, the comparisons were made only within the population of project participants,
all of whom installed at least one soil conservation practice. Since hedgerows were the
least expensive of the structures to install, it is possible the poorest participants installed
hedgerows.
Crop bands. Households installing crop bands had 0.7 more plots (p < .001), 0.6
more secure plots (p = .005), 0.27 ha more area in secure plots (p = .011), but heads of
household having 0.9 fewer years of school (p = .011) than did households not installing
crop bands. This indicates that households adopting crop bands were better off compared
to nonadopters. Crop bands are more expensive to install that hedgerows because they
require large quantities of perennial crop cultures (e.g., sugar can, pineapple) and they
take up more space in the plot than hedgerows.
Rock walls. Households installing rock walls had 0.3 more household members
(p = .001), 2 years less of total schooling (p = .006), and 0.2 more family members
working on the plot in 1995 (p < .001) than did households not installing rock walls.
Rock wall adopters apparently have more labor available than non-adopters, but no
difference in amount of land. Rock walls are costly to install, but the cost is exclusively

144
in labor as no crop germplasm is used. Returns to labor are seen more quickly than for
hedgerows, since rock walls function immediately to slow erosion and collect water.
Gullv plugs. Households installing gully plugs had 0.5 more plots (p < .001), 0.5
more total hectares (p < .001), and purchased 4 more person-days of labor to work in the
plot during 1995 (p = .001) than did households not installing gully plugs. This appears
to indicate that better-off farmers built gully plugs. Gully plugs built of stone can be
costly to install, but they often create new agricultural land (because they collect
sediment on the uphill side) where water accumulates from the side slopes. This can
happen quickly (within one rainy season), especially in sandy, basaltic soils. Farmers
typically plant plantains and taro in these microsites. As for the installation, getting the
crop germplasm can be expensive but the profit can be substantial.
PLUS project tree seedlings. Households planting PLUS project tree seedlings on
at least one plot during 1995 had 0.2 more plots (p = .012), 2.2 more total years of school
(p = .005), and 0.2 more family members working on the plot during 1995 (p = .004) than
did households not planting project tree seedlings.
Top-grafting. Households having top-grafted fruit trees on at least one plot had
0.5 more plots (p = .001), 0.4 more hectares (p = .012), and 3.4 more total years of school
(p = .006) than did households not having top-grafted fruit trees. It is interesting that
households planting project tree seedlings and those grafting fruit trees had more total
years of school than households not engaging in those activities. This was not true for
any of the other agroforestry practices. These findings are similar to those found by
Smucker (1988), but the reasons for it are not clear. Tree seedlings are either provided
free or grown by the farmer, and the first four grafts are also either done by the fanner or
the extension agent without charge, so costs of adoption are minimal.

145
The small number of significant results and the small numerical differences, even
when significant, between households having and not having the indicated practices
means there were small household resource differences in the sampled population. This
was probably because nonproject farmers were not included in the sample. However
small the differences, though, it appears that
• households installing hedgerows had less land than households installing
crop bands and gully plugs or top grafting;
• households installing crop bands had more land in secure tenure than other
households;
• households installing gully plugs invested more in purchased labor on the
plot; and
• households planting project trees or top-grafting attended more years of
school than other households.
These associations appear to make sense based on the costs of installation of the
agroforestry technologies, as explained above.
Plot characteristics and the decision to install
The plot characteristics of interest were tenure security, soil fertility, slope, and
distance from the plot to the residence. The assumption was that these influence the
probability that farmers adopt agroforestry practices in various ways. Land held under
secure modes of access would be more likely to be used when the practice is expensive to
install and potential returns are high and extend over time. Plots having fertile soil would
be used for practices that could provide a quick return to that resource. In contrast, plots
having infertile soil might be used for practices that were previously unknown to the
farmer and looked risky. Slope might influence adoption in a similar way to soil

146
fertility—the best (less sloped) plots would be used for practices requiring larger
investment, steeper slopes for untried practices. Distance from the residence might be an
issue if theft of potential products (e.g., trees, fruit) was likely.
Direct measurements or observations were used to determine tenure, slope, and
distance. Purchased plots and inherited divided plots are referred to as “secure".
Farmers assessed soil fertility on a qualitative scale, with the soils rated as 4 and 5 on that
scale referred to as “fertile.” Analyses showed there was a correspondence between
farmers’ ratings and laboratory tests: soils rated as fertile had a lower pH, more
potassium, and more organic carbon (Appendix F). Plot characteristics for six
agroforestry practices were considered in the 2,295 gardens visited during the survey:
hedgerows, found in 60% of the gardens; crop bands, found in 4%, rock walls, found in
30%; gully plugs, found in 17%; tree seedlings, found in 46%; and grafted trees, found in
9%.
The importance of tenure security and soil fertility on the decision to install a
practice is shown in Tables 6-18 to 6-21. These tables present two-by-two cross
tabulations comparing tenure security (Tables 6-18 and 6-19) and soil fertility
(Tables 6-21 and 6-21) of plots having the indicated practice to those not having the
practice. Three kinds of results were seen. A lower percentage of plots with hedgerows
were in secure tenure (52%) compared to plots without hedgerows (61%). However, a
higher percentage of plots with crop bands, tree seedlings, and tree grafting were in
secure tenure compared to plots without those practices. Tenure status of plots did not
differ for rock walls or gully plugs.

147
Table 6-18: Percent of 2,295 surveyed plots with and without three kinds of agroforestry
practice held under secure (purchased or inherited/divided) or not secure (all
other) tenure
Hedgerows Crop bands Rock walls
with without with without with without
% of plots having secure tenure
52
61
66
55
54
56
% of plots not having secure tenure
48
39
34
45
46
44
Number of plots
1,365
930
100
2,195
686
1,609
2-bv-2 chi-square P-value
<.001
.030
.349
Table 6-19: Percent of 2,295 surveyed plots with and without three kinds of agroforestry
practice held under secure (purchased or inherited/divided) or not secure (all
other) tenure
Gully plugs Tree seedlings Tree grafting
with without with without with without
% of plots having secure tenure
57
55
64
48
69
54
% of plots not having secure tenure
43
45
36
52
31
46
Number of plots
385
1,910
1,050
1,245
207
2,088
2-by-2 chi-square P-value
.604
<.001
<.001
All cross tabulations for the effect of soil fertility on adoption showed significant
differences from the expected cell values. A lower percent of plots with hedgerows had
fertile soil (41%) compared to plots without hedgerows (51%). The opposite was true for
all other practices, where a higher percent of the plots with them were rated as having
fertile soil compared to plots without them.
Table 6-20: Percent of 2,295 surveyed plots with and without three kinds of agroforestry
practice having fertile or not fertile soil
Hedgerows Crop bands Rock walls
with without with without with without
% of plots having fertile soil
41
51
58
44
50
43
% of plots not having fertile soil
59
49
42
56
50
57
Number of plots
1,365
930
100
2,195
686
1,609
2-by-2 chi-square P-value<.001.007.001

148
Table 6-21: Percent of 2,295 surveyed plots with and without three kinds of agro forestry
practice having fertile or not fertile soil
Gully plugs
Tree seedlings
Tree grafting
with
without
with
without
with without
% of plots having fertile soil
50
44
47
43
58 44
% of plots not having fertile soil
50
56
53
57
42 56
Number of plots
385
1,910
1,050
1,245
207 2,088
2-by-2 chi-square P-value .026 .036 <.001
Two observations can be drawn from these results: 1) assignment of agroforestry
practices to plots is not random and 2) hedgerows seem to be assigned to plots of less
secure tenure and lower fertility. However, the hedgerows found in the sampled plots
were not uniform. They were composed of various tree, crop, and grass species. There
were sixteen different tree species represented in the sample, the most common being
Leucaena leucocephala. There were also hedgerows whose components included
pineapple, sugar cane, plantain, pigeon pea, and cassava, as well as five kinds of forage
grasses. If each type of hedgerow is analyzed separately, quite different sets of results
are obtained that strengthen the findings in Tables 6-18 to 6-21. These results are shown
in Tables 6-22 and 6-23. Tree-based hedgerow plots contained only tree species; crop-
based hedgerow plots contained at least one food crop species, but may also have
contained trees and grasses; and grass-based hedgerow plots were made exclusively of
forage grasses. The characteristics of each kind of hedgerow plot were compared to the
same characteristics of all other plots in the sample of 2,295 PLUS plots, as above.

149
Table 6-22: Percent of 2,295 surveyed plots with and without three kinds of hedgerows
held under secure (purchased or inherited/divided) or not secure (all other)
tenure
Tree-based Crop-based Grass-based
with without with without with without
% of plots having secure tenure
49
59
54
56
48
56
% of plots not having secure tenure
51
41
46
44
52
44
Number of plots
701
1,594
507
1,788
44
2.251
2-bv-2 chi-square P-value<,001.543,300
Table 6-23: Percent of 2.295 surveyed plots with and without three kinds of hedgerows
having fertile or not fertile soil
Tree-based Crop-based Grass-based
with without with without with without
% of plots having fertile soil
34
50
51
43
43
45
% of plots not having fertile soil
66
50
49
57
57
55
Number of plots
701
1,549
507
1,788
44
2,251
2-by-2 chi-square P-value<.001.001.810
The results for tree-based hedgerows were the same as for hedgerows in
general—a lower percent of plots where they were installed were in secure tenure and
fertile soil compared to plots where they were not installed. However, when perennial
crops were added as structural components of hedgerows the results were similar to those
for crop bands, with a higher percent of plots having crop-based hedgerows in fertile soil
compared to plots without them. Tenure security was not significant for crop-based
hedgerows; neither tenure security nor soil fertility were significant for grass-based
hedgerows.
There were statistically significant differences (oc = 0.05) in the slope of plots
having agroforestry practices compared with those not without them, but the differences
were numerically small. Hedgerows (34% vs. 30%) and crop bands (37% vs. 32%) were

150
on plots having steeper slopes compared to plots without those practices. Rock walls
(31% vs. 33%) and grafted trees (26% vs. 33%) were on plots having less steep slopes
compared to plots without those practices. There were no differences in slope for gully
plugs or tree seedlings. The differences in distance from the residence to the plots having
agroforestry practices and those without them were also statistically different (a = 0.05),
but the numerical differences were only from one to three minutes, except for grafted
trees. Grafted trees were found on plots averaging seven minutes walk from the
residence, while plots without grafted trees were fourteen minutes distant.
It is possible to gain additional insights on the relationship between plot
characteristics and the decision to invest in agroforestry practices by considering all trees
on the plot whose trunk diameter is greater than 10 cm DBH. These trees may or may
not have been planted in association with PADF projects, and may be either fruit or
hardwood species. There were a significantly greater number of mature trees on
purchased plots, and on inherited but separated plots, than on inherited unseparated,
sharecropped, or on rented land (Table 6-24). More (but not statistically different) trees
were found on inherited separated land than on purchased land, probably because the
inherited land had been held longer by the household than the purchased land.
Table 6-24: Number of all trees larger than 10 cm diameter on PLUS plots, by plot
tenure
Mean no. of trees/haa
No. of plots
Inherited, separated
103a
324
Purchased
88 a
946
Inherited, unseparated
69 b
481
Sharecropped
61 b
189
Rented, private
56 b
283
aKruskall-Wallis p-value <.001

A similar analysis showed that more mature trees were found on plots having
151
fertile soil (Table 6-25).
Table 6-25: Number of all trees larger than 10 cm diameter on PLUS plots, by soil
fertility
Qualitative fertility classes
Mean no. of trees/haa
No. of Plots
Very fertile
98 a
274
Above average fertility
90 a
758
Moderate fertility
73 b
893
Less than average fertility
60 b
308
Infertile
58 b
58
a: Kruskall-Wallis p-value <.001
Management Quality of Agroforestrv Practices
The hypotheses for this section are that the quality of management a farmer gives
to an agroforestry practice is better in households with more labor resources are
available, in plots having physical resources (slope, soil fertility) that potentially would
provide a better chance of return to management investment, and in households that have
a greater expectation (based on experience or demonstration) of benefits from the
technology.
Management quality variables were recorded by PADF technicians during the site
visit. Four variables are used as indicators of management quality: percent of rows/gully
plugs judged to be well managed, percent poorly managed, number of breaches larger
than 25 cm in the structures, and whether the farmer repairs the breaches. Each row of
hedges, crop bands, or rock wall, and each gully plug, was judged separately by the
technician as being well managed, medium managed, or poorly managed based on plant
vigor, pruning practice, plant density, and number of breaches (holes) in the structure.
The numbers of breaches were counted by the technician. The technician determined

152
whether the farmer repaired breaches by observation and by asking the farmer, since this
kind of repair is normally done just before crops are planted and might not have been
done yet at the time of the visit. As for the decision to install tests, the four management
quality indicators were tested against plot slope, distance from the residence in minutes,
mode of tenure/access to the plot, and farmers’ qualitative estimates of soil fertility.
Household resources and management quality
The statistical tests used in this section were bivariate correlations (percent well
managed, percent poorly managed, number of breaches) and t-tests without the
assumption of uniform variances (whether breaches were repaired). Similar to the
section on household resources and the decision to install, there were statistically
significant correlations and t-tests, but the correlation coefficients and differences
between means were small.
Hedgerows. The percent of well-managed hedgerows decreased as the number of
plots held by the household (p = .004) and the number of plots in secure tenure (p = .054)
increased. The percent of well-managed hedgerows increased as the number of males of
age 21-60 (p = .003), the number of household members of age 21-60 (p = .016), and the
number of person-days of labor purchased in the plot during 1995 (p = .050) increased.
The number of breaches in the hedgerows decreased as the number of plots in secure
tenure (p < .001), number of hectares in secure tenure (p < .001), number of person-days
of labor purchased in the plot during 1995 (p = .012), and the number of family members
working in the plot (p = .002) increased. Households repairing breaches in hedgerows
had 0.3 more family members working in the plot than did households not repairing
breaches (p = .002).

153
Crop bands. The percent of well-managed crop bands increased as the number of
years of school attended by the head of household increased (p = .043). The number of
breaches in crop bands decreased as the number of secure plots (p = .036), the number of
person-days of labor purchased in the plot during 1995 (p = .017), and the number of
years of school attended by the head of household increased (p = .027). Households
repairing breaches in crop bands purchased 19 person-days of labor in the plot,
households not repairing breaches bought 10 person-days (p = .012).
Rock walls. The percent of poorly managed rock walls decreased as the number
of females of age 21-60 increased (p = .021). The number of breaches in rock walls
increased as the number of plots per household increased (p = .034). Households
repairing breaches bought four more person-days of labor in the plot (p = .049) and had
0.3 hectares less total plot area (p = .037) than households not repairing breaches.
Gullv plugs. The percent of well managed gully plugs increased as the number of
secure plots (p = .025), the number of hectares in secure tenure (p = .041), the number of
household members (p = .004), the number of females age 21-60 (p = .002), and the
number of household members age 21-60 (p = .006) decreased. The number of poorly-
managed gully plugs increased as the number of females age 21-60 (p = .012) and the
number of household members age 21-60 (p = .039) increased.
Plot characteristics and management quality
Although plot tenure security appears to have been an important factor in
farmers’ decisions to install agroforestry practices on a plot, it was not related to the
quality of management of the practices. None of the tests of land tenure status against
management quality (percent well-managed, percent poorly-managed, number of
breaches, repairs breaches) were significant for any agroforestry practice.

154
Hedgerows. Hedgerows were installed on plots having less-secure tenure
compared to other practices (Table 6-18), but, once installed, the best management of the
hedgerow was practiced on plots having the most fertile soil (Table 6-26). Note that the
very fertile category appears to have poorer management (fewer percent well-managed
rows and more breaches) than the above average category. This kind of anomaly in the
order of soil fertility categories is found in many of the analyses, and is probably due to
the inexact and qualitative nature of this variable. Table 6-26 indicates that the best-
managed hedgerows were found on the most fertile soil, the worst managed hedgerows
were found on the most infertile soils, and that farmers were more likely to repair
breaches on more fertile soils.
Table 6-26: Soil fertility and hedgerow management quality.
Qualitative soil
fertility classes
Percent well
managed rows
Percent poorly
managed rows
No. of breaches
per 100 m
Repairing (%)
Yes No
infertile
24 bb
33a
19a
37
63
less than average
24 b
32a
18a
34
66
moderate fertility
37ab
22ab
18a
46
54
above average
43a
20b
17a
52
48
very fertile
40ab
29ab
13b
44
56
P-valuea
<.001
.003
<.001
.001
a P-values in first three columns are for Kruskal-Wallis tests; last column for 2 by 5
cross-tabulation
b numbers within columns followed by the same letter are not significantly different
(a=.05).
Percent slope of the garden was not an important factor in management quality
except in the case of number of breaches. There was a significant negative correlation
between slope and the number of breaches (p = .002), that is, there were fewer breaches
on steeper slopes, more on gentler slopes. A negative correlation was also found for

155
distance between the house and the plot; there are more breaches in hedgerow plots
closer to the house (p < .001). These results seem contradictory. It seems reasonable to
suppose steeper slopes would have more breaches because it is harder to establish
hedgerow trees on steeper slopes, and more difficult to work at repairing them. The case
of more breaches in closer more gently sloping plots might be explained by more foot
traffic closer to the house cutting paths through the hedgerows, and the possibility that
animals might be picketed in plots close to the house.
When the three kinds of hedgerows are tested separately, as they were in the
previous section discussing farmers’ decisions to install practices, we find a similar but
weaker result (as compared to the aggregated hedgerow results in Table 6-26) regarding
management quality and soil fertility. For tree-based hedgerows, the soil fertility vs.
management quality Kruskal-Wallis tests (percent well managed rows, percent poorly
managed rows, number of breaches, and whether or not the breaches were repaired)
showed significant and similar differences in management quality as all hedgerows
combined—better management on more fertile soil. However, the mean separation tests
were either not significant or did not produce a rank order that could be interpreted.
Crop-based hedgerows were better managed in plots having fertile soils. The percent
well- and percent-poorly-managed variables increased and decreased in the same order as
soil fertility ratings increased (p = .006) and decreased (p < .001), respectively. Again,
however, mean separation was not significant in the case of percent well managed. Slope
was not an important factor in management quality in tree-based hedgerows, but
significant correlations were found for crop-based hedgerows. The percent of well-

156
managed rows increased as slope became steeper (p = .009), and the number of breaches
decreased as slope became steeper (p < .001). The mean slope of repaired crop-based
hedgerows was slightly, but significantly (p = .007), steeper (36%) than the slope of
unrepaired ones (32%). In only two cases was distance from plot to residence
significant. Number of breaches decreased with increasing distance for both tree-based
(p = .003) and crop-based (p = .009) hedgerows. No differences were found in any of the
tests of management quality for grass-based hedgerows.
Differences in hedgerow management quality among PADF field regions were at
least as important as differences among plot characteristics (Table 6-27).
Table 6-27: PADF/PLUS field region and hedgerow management quality.
PADF/PLUS
Field Region
Percent well
managed rows
Percent poorly
managed rows
No. of breaches
per 100 m
Repairing (%)
Yes No
1: Les Cayes
41abb
21bc
19b
48
52
2: Jacmel
15c
31a
24a
33
67
3: Cap Haitien
45a
28ab
9c
54
46
4: Mirebalais
35b
19c
22ab
40
60
P-valuea
<.001
<.001
<.001
<.001
a P-values in first three columns are for Kruskal-Wallis tests; last column for 2 by 5
cross-tabulation
b Numbers within columns followed by the same letter are not significantly different
(«=•05).
Overall, there were more well managed hedgerows than poorly managed ones,
but there were significant differences among regions. Region 2 had a much lower
percentage of well-managed hedgerows, a higher number of breaches, and a lower
percent of farmers repairing breaches than do the other three teams. Most of the
hedgerows in region 2 were built by farmers in areas where the soils are poor and steep.

157
Crop bands. The small number of crop band gardens in the survey (about 100)
made it necessary to recode the soil fertility ratings into three categories instead of five.
Even so, only five crop band plots were in the infertile category. Farmers apparently
eliminated infertile plots during the installation, leaving few to be analyzed for
differences in management quality. Table 6-28 indicates that very fertile plots had fewer
breaches and a larger percent of farmers repairing breaches. The test for percent of rows
well managed and soil fertility was significant, but mean separation tests were not.
Table 6-28: Soil fertility and crop band management quality.
Qualitative soil
fertility classes
Percent well
managed rows
Percent poorly No. of breaches Repairing (%)
managed rows per 100 m Yes No
infertile
13
40
48ab
20
80
moderate fertility
30
11
15a
50
50
very fertile
68
7
6b
87
13
P-valuea
.005
.250
.001
<.001
a P-values in first three columns are for Kruskal-Wallis tests; last column for 2 by 3
cross-tabulation
b Numbers within columns followed by the same letter are not significantly different
(a=.05).
Slope percent of the plot was important to crop band management quality only for
the percent of well-managed rows, where there was a significant negative correlation (p =
.013). The percent of well-managed rows increased as the slope became gentler.
Distance from the house to the plot, and the tenure category of the plot were not
important to management quality.
Table 6-29 shows the differences in crop band management quality among PADF
field regions. There were only six crop band plots in the Team 4 region, so it was not

158
included in the table. All other teams had more than 30 crop band plots each, but
because of missing data for some of the variables, not all of the plots were included in the
analyses.
Table 6-29: PADF/PLUS field region and crop band management quality.
PADF/PLUS
Field Region
Percent well
managed rows
Percent poorly No. of breaches Repairing (%)
managed rows per 100 m Yes No
1: Les Cayes
46bb
15
22a
44
56
2: Jacmel
9ab
0
3ab
60
40
3: Cap Haitien
73a
8
4b
89
11
P-valuea
.011
.165
<.001
<.001
a P-values in first three columns are for Kruskal-Wallis tests; last column for 2 by 3
cross-tabulation
b Numbers within columns followed by the same letter are not significantly different
(ot=.05).
Crop bands in region 3 were the best managed. Crop bands were first installed in
the Cap Haitien area, so the region 3 team has more of them and a longer history of
working with them.
Rock walls. Soil fertility of rock wall plots is not an obvious characteristic to test
for differences in management quality, yet Table 6-30 shows that differences did exist.
Better management was found in plots of higher soil fertility and poorer management
was found on less fertile plots, even though the differences were not great and the order
of the fertility classes (in this case the unfertile class has better management than the less
than average class) was inconsistent. A more marked difference was seen in the percent
of farmers repairing breaches in rock walls. In plots of moderate to high fertility,
repairing farmers outnumbered nonrepairing farmers by a large margin. In the less fertile
plots, the reverse was true.

159
Table 6-30: Soil fertility and rock wall management quality.
Qualitative soil
fertility classes
Percent well
managed rows
Percent poorly No. of breaches Repairing (%)
managed rows per 100 m Yes No
infertile
51abb
15ab
8ab
29
71
less than average
44b
17a
9a
41
59
moderate fertility
59a
6b
6b
66
34
above average
63a
7b
6b
66
34
very fertile
61a
8ab
5b
55
45
P-valuea
.004
.002
.002
<.001
a P-values in first three columns are for Kruskal-Wallis tests; last column for 2 by 5
cross-tabulation
b Numbers within columns followed by the same letter are not significantly different
(«=■05).
Slope of the plot was important to management quality of rock walls. A slight but
significant effect of slope was found for each of the four management quality indicators,
with rock walls found in plots of gentler slope being better managed and having fewer
breaches. There was a significant negative correlation between slope and the percent of
well managed rows (p = .023), and significant positive correlations between slope and the
percent of poorly managed rows (p = .001), and between slope and the number of
breaches per 100 meters (p = .016). The average slope of plots where farmers repaired
rock walls (30%) was slightly, but significantly (p < .001), less than the average slope of
plots where farmers did not repair (33%). Distance between the plot and the house was a
factor for only one management indicator. There was a slight but significant negative
correlation between the percent of poorly managed rows and distance from the house (p
= .020), that is, rock walls closer to the house were less well managed. The reason for
this is not obvious, but may have to do with the presence of large animals closer to the
house.

160
There is less difference in rock wall management quality among PADF regions
than there is for plant-based structures. However, rock walls in regions 1 and 4 appear to
be better managed (Table 6-31).
Table 6-31: PADF/PLUS field region and rock wall management quality.
PADF/PLUS
Field Region
Percent well
managed rows
Percent poorly
managed rows
No. of breaches
per 100 m
Repairing
Yes
:(%)
No
1: Les Cayes
72ab
7ab
4bc
74
26
2: Jacmel
53b
9a
7a
56
44
3: Cap Haitien
52b
11a
5c
58
42
4: Mirebalais
66a
5b
6ab
67
33
P-valuea
<.001
.010
<001
.007
a P-values in first three columns are for Kruskal-Wallis tests; last column for 2 by 4
cross-tabulation
b Numbers within columns followed by the same letter are not significantly different
(«=•05).
Gully plugs. Overall, there were few significant differences in management
quality due to plot characteristics for gully plugs. Soil fertility, distance from the house,
and tenure category of the plot were not significant factors. There was a slight but
significant positive correlation (p = .043) between the percent of poorly managed gully
plugs on the plot and the slope of the plot. There were not great management quality
differences among PADF regions.
Fanners’ perceived benefits and problems and hedgerow management quality
Farmers expressed their perception of the importance of several potential benefits
and problems associated with hedgerows during the plot interview. A potential benefit
thought to be not important would be assigned a value of one or two on a qualitative five-
point scale; a potential benefit thought to be very important by the farmer would be
assigned a value of four or five. These observations pertain to the 1,362 hedgerow

161
gardens in which the interviews took place, and were not observations about hedgerows
in general. The percentages of ratings falling in each qualitative category for six
potential benefits are shown in Table 6-32.
Table 6-32: Percent of 1,362 plots where potential benefits of hedgerows (fuel,
charcoal, construction wood, fodder, soil fertility, soil conservation, and
crop production) were classified according to importance by farmers on a
five-point scale, l=not important, 5=very important.
Importance
Fuel
Charcoal
Construction
Fodder
Soil
Fertility
Soil
Conservation
Crop
Production
1
65
79
86
22
13
4
40
2
12
10
6
13
15
8
10
3
9
5
3
22
22
24
14
4
7
4
3
22
24
31
17
5
7
2
2
21
27
35
18
Most farmers did not consider fuelwood, charcoal, or construction wood to be
important benefits of hedgerows. These three benefits were assigned to the two lowest
categories of importance (categories 1 and 2) on 77%, 89%, and 94% of all hedgerow
plots, respectively. The importance of other potential benefits was less skewed. The use
of hedgerows as a source of animal fodder was highly rated (categories 4 and 5) on 43%
of all hedgerow plots. However, the benefits of hedgerows in increasing crop yield were
held to be low on 50% of the plots and high in only 35%. A surprising finding was that,
on over half of the plots, farmers said that hedgerows’ effect on improving soil fertility
and soil conservation were very important—in 51% and 66% of the plots these two
benefits were placed in the top two categories of importance, respectively. One would
have thought a more concrete and immediate benefit (e.g., crop production or fodder)
would be given a higher rating, unless farmers were just parroting back the extension
message to the technicians conducting the interviews. One way to check this is to

162
compare stated importance of benefits to how well farmers manage hedgerows, on the
theory that hedgerows on a plot where a potential benefit is highly-rated would be better
cared for. The results of Kruskal-Wallis tests comparing the mean ranks of the
percentage of well-managed hedgerows in gardens where the potential benefits were
rated are shown in Table 6-33.
Table 6-33: Percent of hedgerows classified as well-managed by technicians in 1,362
plots where potential benefits of hedgerows (fuel, charcoal, construction
wood, fodder, soil fertility, soil conservation, and crop production) were
classified according to importance by farmers on a five-point scale, l=not
important, 5=very important.
Importance
Fuel
Charcoal
Construction
Fodder
Soil
Fertility
Soil
Conservation
Crop
Production
1
33ab
35a
34a
37b
26ab
19ab
29a
2
37ab
35a
45ab
25a
21a
17a
24a
3
46bc
55b
46ab
33abc
32b
27b
42b
4
41b
51b
56bc
41c
41c
37c
46b
5
55c
41 ab
70c
44c
51d
49d
50b
P-valuea
<.001
<.001
<.001
<.001
<.001
<.001
<.001
a P-values are for Kruskal-Wallis tests
b Numbers within columns followed by the same letter are not significantly different
(a=.05).
All of the tests done on potential hedgerow benefits vs. the percent of well-
managed rows in the garden produced significant results (Table 6-33). A greater
proportion of hedgerows were well managed in gardens where the farmers said any of the
potential benefits were very important. This was true for benefits given top ratings in
very few gardens (fuel, charcoal, construction wood) as well as for benefits given top
ratings in large numbers of gardens (fodder, crop production). The 69 hedgerow plots
where the farmers rated construction wood as being of highest importance (in categories
4 and 5) had the highest percentages of well-managed hedgerows, but these plots were

163
only 5% of the total. There was apparently some substance to farmers categorizing soil
fertility and soil conservation as of highest importance in a large number of gardens,
since in those gardens were found substantially greater proportions of well managed
hedgerows compared to gardens rated lower for those benefits.
The opposite trend should show up when the same analyses are done for
proportion of poorly managed hedgerows and for the number of breaches per 100 m of
hedgerow. This is the case, as shown in Tables 6-34 and 6-35. All Kruskal-Wallis tests
were significant for percent poorly managed vs. benefit rating, however mean separation
for the charcoal benefit was not significant at a=0.05.
Table 6-34: Percent of hedgerows classified as poorly-managed by technicians in 1,362
plots where potential benefits of hedgerows (fuel, charcoal, construction
wood, fodder, soil fertility, soil conservation, and crop production) were
classified according to importance by farmers on a five-point scale, l=not
important, 5=very important.
Importance Fuel
Charcoal
Construction
Soil
Fodder Fertility
Soil
Conservation
Crop
Production
1
27bb
26
26b
31b
44b
60e
28b
2
2 lab
21
14a
30b
36b
41 d
30b
3
19ab
13
16ab
22ab
22a
29c
21ab
4
17ab
16
16ab
19a
16a
21b
19a
5
13a
10
6a
20a
16a
16a
19a
P-valuea
.022
.013
<.001
<.001
<.001
<.001
<.001
a P-values are for Kruskal-Wallis tests
b Numbers within columns followed by the same letter are not significantly different
(a=.05).
When the number of breaches per 100 m of hedgerow is used as the dependent
variable, only four of the benefits showed significant difference among importance
ratings (Table 6-35). In general there were fewer breaches in plots where the benefits

164
were highly rated but the differences are not great and the trend, according to the results
of the mean separations, are not always clear.
Table 6-35: Number of breaches per 100 m of hedgerow counted by technicians in
1,362 plots where potential benefits of hedgerows (fuel, charcoal,
construction wood, fodder, soil fertility, soil conservation, and crop
production) were classified according to importance by farmers on a five-
point scale, l=not important, 5=very important.
Importance
Fuel
Charcoal
Construction
Fodder
Soil
Fertility
Soil
Conse nation
Crop
Production
1
17
18
17
17ab
22abc
22ab
19c
2
15
16
16
23b
21cd
25b
24c
3
18
17
17
17b
19d
23b
18bc
4
20
20
19
17b
17bcd
16ab
17b
5
18
17
20
15a
13a
13a
11a
P-valuea
.065
.225
.364
<.001
<.001
<.001
<.001
a P-values are for Kruskal-Wallis tests
b Numbers within columns followed by the same letter are not significantly different
(«=•05).
The final variable tested against the benefit categories was whether the farmer
repaired breaches in the hedgerows. Chi-square tests for cross tabulations of all benefits
except charcoal were significant (a=0.05), and in general showed that as the benefit was
rated higher in importance, farmers repaired breaches in a greater percentage of those
hedgerow plots. This trend was shown most strongly for soil fertility, soil conservation,
and crop production (Table 6-36).
Very few farmers interviewed said there were problems associated with
hedgerows. In more than 90% of hedgerow plots farmers rated shade competition, loss
of cropping space, hedgerow/crop water competition, and weediness in the two lowest
categories of importance; reduced space for picketing animals and labor cost were not
important in more than 70% of hedgerow gardens (Table 6-37).

165
Table 6-36: Percent of gardens where hedgerow breaches were repaired or not repaired
for three potential benefits of hedgerows.
Soil Fertility
Soil Conservation
Crop Production
Importance
Yes
No
Yes
No
Yes
No
1
36
64
30
70
37
63
2
34
66
29
71
31
69
3
42
58
31
69
54
46
4
53
47
52
48
55
45
5
52
48
55
45
58
42
Chi2 P-value
<.001
<.001
<.001
Table 6-37:
Percent of 1,362 plots where potential problems of hedgerows (shade,
reduced space, water competition, reduced ability to picket animals,
weediness, and labor cost) were classified according to importance by
farmers on a five-point scale, l=not important, 5=very important.
Importance
Shade
Loss of
Space
Water
Competition
Space for
Animals
Weediness
Labor
Cost
1
84
78
81
51
90
48
2
12
14
13
25
7
27
3
3
5
4
14
2
15
4
1
3
1
6
1
7
5
0
1
0
4
0
3
Only seven of the possible 18 Kruskal-Wallis tests (percent well managed,
percent poorly managed, and number of breaches per 100m for each of the six problems)
of hedgerow problems against importance categories were significant (a=0.05).
However, for four of the seven no significant differences could be detected between
importance categories during the mean separation procedure. For the remaining three
tests where mean separation did show significant differences the results were very small
or they did not make any sense, and they are not shown here. This was probably due to
the skewed distribution of the data—very few of the observations were in the higher
importance classes. To reduce this effect in the cross tabulations of hedgerow problems

166
vs. whether the breaches were repaired, the problem rating categories were reduced from
5 to 3, with the old categories 1 and 2 (the problem is not important) grouped together
and old categories 4 and 5 (the problem is very important) grouped together. Only two
cross tabulations produced significant results, as shown in Table 6-38. As reduced space
for picketing animals in hedgerows and labor cost of maintaining hedgerows became
more important, fewer farmers repaired breaches in the hedgerows.
Table 6-38: Percent of gardens where hedgerow breaches were repaired or not repaired
for three potential problems of hedgerows.
Space for Animals
Labor Cost
Importance
Yes
No
Yes
No
1
48
52
48
52
2
40
60
43
57
3
36
64
36
64
Chi2 P-value
.007
.023
Summary and General Discussion
Household and Plots
On average, the households sampled in the survey were composed of five to six
members, about half of whom had no formal education. Their three or four farm plots
totaled about 3.7 hectares and were on slopes ranging from 22 to 34%; just over half of
the plots were accessed via purchase or inheritance divided among siblings. Households
engaged in an average of three economic activities other than raising food crops. The
men mainly kept animals and engaged in various small businesses; the women worked in
marketing of both agricultural and non-agricultural products, but also kept animals. Each
household had installed one or two soil conservation structures, over half of the
households had built hedgerows. About 40% of the men and 15% of the women

167
participated in some phase of the installation or maintenance of soil conservation
structures.
The survey found no significant differences between tenure of plots in terms of
elevation, topographic position, or severity of erosion. There were small differences in
slope (caretaker plots were less steep, all others were about 33%) and distance from the
residence (sharecropped and rented plots were one or two minutes more distant) by
tenure category. Purchased plots averaged 0.53 hectares, significantly larger than
divided, undivided, and sharecropped plots. There were statistically significant
differences in soil fertility (data not shown). A higher percent of purchased plots (49%)
were in the high fertility category compared to other plots (42-44%), with rented plots
having the lowest percent in high fertility.
Decision to Install Soil Conservation Technologies
There were indications that household resources influenced decisions about what
kind of soil conservation practice to install. Households installing hedgerows had
smaller farms and fewer family members working on plots than did households not
installing hedgerows. Households installing crop bands accessed more plots, had more
plots in secure tenure status, and had more land area in secure tenure status than
households not installing crop bands. Households installing rock walls had more
household members and more family members working on the plots than households not
installing rock walls.
There were strong indications that farmers considered plot characteristics in their
decision to install hedgerows and other agroforestry practices. Figure 6-1 shows the
relative influence of plot characteristics. The numbers for tenure security and soil

168
fertility are the percent of plots having the indicated practice that are held in secure
tenure or that have fertile soil, respectively.
Tenure security
Soil fertility
Slope (%)
Distance from
residence (min)
Less —
52 54 57 64 66 69
HR RW GP TS CB GT
_4_1 47 50 58
HR TS RW,GP CB,GT
_26 31 32 33 34 37_
GT RW GP TS HR CB
_1_ 12 14 16_
GT TS, HR, RW GP
CB
More
Figure 6-1: Relative importance of four plot characteristics to the decision to install six
agroforestry practices (HR = hedgerow, CB = crop band, RW = rock wall,
GP = gully plugs, TS = tree seedlings, GT = grafted trees)
In summary,
• Hedgerow were installed on less secure plots having lower soil fertility,
steeper slopes, very slightly farther from the residence.
• Plots having hedgerows constructed with at least one perennial crop plant
were found on more fertile plots, and there were no differences in tenure
status of plots with or without crop-based hedgerows.
• Crop bands were installed on more fertile and more secure plots having
steeper slopes, but there was no difference in distance from the residence.
• Rock walls were installed on more fertile, slightly less steep plots slightly
more distant from the residence, but there was no difference in tenure
security.
• Gully plugs were installed on more fertile plots slightly farther from the
residence, but there was no difference in tenure security or slope.
• PLUS project tree seedlings were planted on more secure, more fertile
plots slightly closer to the residence, but there was no difference in slope.

169
• Fruit trees were top-grafted on more secure, more fertile plots of gentler
slope, much closer to the residence.
• More mature trees of any species were found on more fertile plots. There
were a significantly greater number of mature trees on purchased plots, and
on inherited but separated plots, than on inherited-unseparated,
sharecropped, or on rented land.
Management Quality
Management quality of agroforestry practices, as defined by the condition of the
structures at the time of the survey, was not correlated with level of household resources.
The strongest trend noted was that quality of management for hedgerows, crop bands,
and rock walls increased as the level of labor employed on the plots increased.
Although tenure security was an important factor in the decision to install an
agroforestry practice, it did not influence the quality of management of those practices
once they were installed. The management quality of hedgerows, crop bands, and rock
walls did not differ according to the tenure of the plots. The most important variable
appeared to be soil quality as perceived by the farmer. All soil conservation structures
were better managed on plots having better soil fertility. Management quality increased
as slope decreased. The effect of distance between the plot and the residence was not
strong. These results could not be compared to those from other studies, because no
other studies used management quality as a dependent variable.
Most farmers did not consider fuelwood, charcoal, or construction wood to be
important benefits of hedgerows. Soil conservation and soil fertility improvement were
considered important by substantial numbers of farmers, as were, to a lesser extent, crop
yield improvement and fodder. Management quality of hedgerows was significantly
better on plots where any of the potential benefits was highly rated. Very few farmers
said any of the listed potential drawbacks of hedgerow was important—94% said water

170
competition between hedgerows and adjacent food crops was not a serious problem.
There was little correspondence between opinions regarding severity of hedgerow
problems and management quality.
Differences Among Technologies
Differences among technology adoption revealed in the survey are better
understood if discussed in the context of investment, risk, and timing of benefits to the
farmer. Agro forestry practices have a cost of installation and a cost of management that
can be measured in labor and materials. The farmer must compare these costs to the
resources commanded by the household, and then to the amount and timing of the
anticipated benefits, which must in turn be compared to the needs of the household. All
this must be balanced against the risk of failure. Risk might be higher if the farmer or his
peers have no previous experience with the practice, or if the practice is sensitive to
environmental factors, such as free ranging animals or drought. Risk is diminished when
some of the costs are subsidized by projects, or when there is a continuing outreach
program. The survey did not provide cost and benefit data, but some general
comparisons among practices can be made based on experience of the project staff. In
the matrix shown in Table 6-39, the relative costs, benefits, and risks of agroforestry
techniques are compared.
The ratings shown in Table 6-39 cannot be fixed exactly, because they change
with the length of the farmer’s experience, with the available resources and needs of the
household, with the stage of development of the practice, and with the degree of project
subsidy. For example, tree-based hedgerows are at greater risk of animal damage in the
early stages of growth, and are less prone to animal damage if there are perennial crops
associated with the trees. In that case, neighboring farmers would be more likely to

171
contain their animals. Tree seedlings are much more vulnerable to loss due to accidental
weeding by paid laborers and by animal damage in the early stages of growth. Crop
bands containing sugar cane, pineapple, or plantains could be expensive to install in some
areas, unless the plant material is subsidized by a project.
Table 6-39: Estimated relative costs, benefits, and risk of agroforestry practices
AF practice
Cost of
installation
Cost of
management
Amount of
benefits
Timing of
benefits
Risk of
loss
Hedgerows
low-med.
med.-high
med.-high
med.-long
med.-high
Crop bands
med.-high
med.-high
high
short
med.
Rock walls
high
low
low-med.
short
low
Gully plugs
med.-high
low-med.
med.-high
short-med.
low-med.
PLUS trees
low
low
med.-high
med.-long
low-med.
Grafted
trees
low-med.
low-med.
med.-high
med.-long
low-med.
Hedgerows
It is apparent that hedgerows were installed on less-desirable plots. There are
several possible reasons for this. Hedgerows are a relatively new agroforestry practice in
Haiti, even though there are similar indigenous practices such as trash barriers and living
fences, and farmers tend to install new practices on their worst plots (Murray 1979).
They are not difficult to install, and the PLUS project sometimes provides the seeds.
Hedgerows are a relatively easy way to comply with the PLUS project requirement that
participants install soil conservation structures on some part of their holdings in order to
get access to crop seed banks operated by their group-about 60% of the sampled PLUS
plots had hedgerows. Although the cost of installation is reasonable, hedgerows require
regular maintenance in the form of repairing breaches, top-pruning the hedgerow trees,
and eliminating invasive hedgerow seedlings from the area between the rows. The

172
timing and amount of returns to hedgerows depend on the management objective of the
farmer. If they are being managed for soil fertility, three to five years may be required to
make a noticeable contribution. If they are being managed for animal production, useful
amounts of fodder can be taken after the first year.
Tree-based hedgerows exhibited the same plot characteristics (Tables 6-22, 6-23)
as all types of hedgerows were considered together (Tables 6-18, 6-20). They were on
plots of less security, lower fertility, and steeper slope. Tree-based hedgerows showed no
difference in distance from the residence compared with all other plots. The most
striking result displayed in Tables 6-22 and 6-23 is that when even one food crop species
entered the hedgerow, plot characteristics changed, and became similar to those of the
crop band plots: they were installed in plots having soils of higher fertility and slightly
steeper slopes. Unlike crop bands, however, crop-based hedgerow plots did not differ in
tenure security from other plots, and there was a slight, but significant tendency for crop-
based hedgerows to be farther from the residence. The key plot parameter affecting the
farmers’ decision in this case seems to be soil fertility, since 51% of the plots having
crop-based hedgerows had fertile soils, while only 34% of the plots having tree-based
hedgerows were rated as fertile.
Crop bands
Because of their larger dimension and greater number of food crops, crop bands
are more expensive to install than hedgerows, and need more fertile soils and regular
rainfall to flourish. Quite often in the PLUS project, the food crop plants are distributed
to new participants in a crop multiplication arrangement. The farmer must return the
same amount of plant material to his or her group as was given, and it is then spread to
other farmers under the same arrangement. This underwrites the cost of installation, and

173
reduces risk to the farmer. Crop bands were planted on steeper plots of higher fertility
compared to plots having other interventions. Land tenure appears to be a factor in the
choice of plots, as it is in the case of hedgerows, except that crop bands are installed on
more secure plots. High soil fertility is required to sustain crop growth, and to reduce the
risk of losing the investment in valuable crop germplasm. The choice of steeper slopes
possibly reflects the extension message that crop bands are a soil conservation technique.
Rock walls
Rock wall terraces are well known to most farmers because they have been
promoted for many decades by various projects. They require the most skill and labor to
install of all the soil conservation practices promoted by the PLUS project. Once
installed, however, they are relatively undemanding in their upkeep. They produce no
vegetation to take care of, and they are stable if large animals are kept out of the garden
and the slope is not too steep. About 30% of the sampled gardens had rock walls
installed. The availability of rocks prevents farmers from constructing rock walls in
some areas, especially where the soils are derived from basaltic parent material. Analysis
of plots having rock walls show them to be similar in most characteristics compared to
others, except they are located on soils of higher fertility. This may reflect a choice of
the farmer—since rock walls are an expensive investment they might want to install them
on the best soils—or it might be that soils having enough rocks for constructing the
terraces are of higher fertility than normal. Although slope of rock wall plots was less
steep and distance from the residence was greater compared to other plots, the
differences were small. Land tenure was not a factor in choice of plots for rock walls.

174
Gullv plugs
Tenure and slope were the same on plots with and without gully plugs. Soils on
gully plug plots were more fertile, but it is not clear whether the farmer was judging the
fertility of the soil collected by the structure, or the soil on the plot in general. Gully
plug plots were slightly farther from the house than are other plots. It could be that
houses are located away from gullies for safety.
Number of mature trees on the plot
There was clearly an association between the number of trees on a plot and the
security of tenure. More securely-held plots had greater numbers of trees. There was
also a very slight but statistically significant negative correlation between the numbers of
trees on a plot and the distance of the plot from the house. There tended to be more trees
on plots closer to the house. There was not a significant correlation between slope of the
plot and numbers of trees. There was also a significant difference in the numbers of trees
found on plots of different soil fertility, as shown in Table 6-25. The number of trees
increased with increasing soil fertility. From observation, most trees nurtured by Haitian
farmers are fruit trees or trees whose timber is valuable. Valuable trees are planted near
the house to avoid theft. Home garden plots were more likely to be in secure tenure
categories (Table 6-10). Greater numbers of trees could be found on soil of higher
fertility because they survive and grow better there. In addition, animals are often tied
under trees near the home, and this, along with the accumulation of tree litter, enriches
soil fertility.
Project tree seedlings planted during 1995
Trees planted with project assistance during 1995 exhibited similar plot
characteristics as existing large trees (Tables 6-19 and 6-21). They were located on plots

175
of more secure tenure, slightly greater soil fertility, closer to the house, but with no
difference in slope. Overall, more trees were planted and maintained on more fertile,
more secure plots close to the residence. Establishment and maintenance cost little, and
potential returns are high. These plot characteristics reinforce the idea of trees as a store
of value.
Top-grafted fruit trees
The decision to graft fruit trees appears to have been made according to specific
plot characteristics. Alternatively, the original decision of the farmer to plant or nurture
the low value fruit trees could be reflected in these plot characteristics. All four plot
characteristics are significantly different from those of plots not having grafted trees.
Grafted trees are found on plots with greater tenure security, more fertile soils, gentler
slopes, and much closer to the house. This clearly indicates that grafted trees are mainly
found in or near the home garden, the plot where the house is found.
Conclusions
Previous studies found differences in household resources between adopters and
non-adopters. These differences were weak in this study, but appeared to indicate that
hedgerow adopters were less well off (less land and household members), while crop
band adopters were better off (more land and more securely held land). This could have
been due to regional differences (most of the crop bands were in the north of the
country), or because they are more demanding in their requirements for space, soil
fertility, and rainfall. Possibly only better off farmers can make crop bands succeed.
Farmers apparently based their decisions regarding installation of agroforestry
practices on characteristics of the plot and the technology. There were differences in the
selection of plots between the different agroforestry practices. Overall, the survey

176
indicated that farmer decisions to adopt new technologies are correlated with several plot
characteristics. Tenure appears to influence adoption in five of six technologies
surveyed. Trees and grafted fruit trees were more common on purchased and divided
inheritance plots. The value of tree products increases over time, so farmers need to
protect their rights to harvest. Crop bands and gully plugs are also more common on
purchased or divided inheritance plots. This is likely attributable to the high value of
perennial food crops in crop contour bands (pineapple, plantain, sugar cane) and the
economically important crops planted in soil collected by gully plugs (plantains, taro).
Hedgerows are more commonly found on plots with other modes of access. Hedgerows
are relatively easy to install, so this may reflect a strategy of risk minimization when
trying a new practice or fulfilling project requirements to install soil conservation
measures.
Correlation between technology adoption and soil fertility was as important as the
relation between adoption and tenure status. This is perhaps to be expected since tenure
status and soil fertility were also related (more purchased plots were classified as very
fertile); however, farmer assessments of fertility also appear to integrate other productive
factors not measured by laboratory analysis of soil nutrient levels.
Tenure and soil fertility were both associated with adoption in parallel fashion.
Technologies (crop contour bands, gully plugs, trees, top-grafted fruit trees) more
common on purchased and divided inheritance plots were also more common on fertile
plots, and conversely (hedgerows). Although overall analysis of the data indicated that
mode of access to land was an important variable, the data showed no definitive
relationship between tenure status and adoption. The evidence does not allow clear
separation of the relative influence of tenure and fertility on adoption; therefore, it is not

177
possible to determine which is more important in a particular decision to adopt new
technology.
The analysis found no association between tenure status and differences in
management; but soil fertility was important to management quality, with the best
management found on the best soils. Based on these results, an extension program would
not need to target technologies towards or away from any particular tenure type. It
might, however, want to work with farmers on cost-effective ways to improve soil
fertility while simultaneously ensuring a supply of dry-season fodder.
Farmers did not consider soil water competition between hedgerows and adjacent
crops to be a problem. Only 16 of the 1,540 farmers considered it to be a serious
problem. This could be because there was adequate rainfall in 1995, the success or
failure of a crop is dominated by other factors that effect the whole field, such as the
timing of the rainfall compared to crop demands, or because very few farmers rated any
of the problems on the questionnaire as severe.
This study revealed relationships between plot characteristics and farmers’
decision to install various soil conservation structures, and how those structures were
managed. However, the findings should be considered as indicative, and not definitive,
until follow-up interviews clarify the relationships, for several reasons:
1. Decisions to install various technologies appeared to correlate with both tenure
and soil fertility in the same way, the relative contribution cannot be determined from the
data in this study.
2. Management quality of soil conservation practices varied significantly
between regions of the country at least as much as differences due to soil fertility, but it

178
is not clear if this was a reflection of differences in regional soil fertility, extension
methods, or the length of time since a technology had been introduced into the region.
3. The analysis mixed together farmers just recently installing a practice with
those having several years experience. Hedgerows can require several years to show
benefits, and they evolve as farmers get more experience with them. Analysis should be
separated by age of the practice.
4. The correlations between farmers’ expressed problems and benefits regarding
hedgerows and the management quality of hedgerows should have been compared as
well to their pattern of building or removing hedges after the initial installation. In
general, getting good responses to these kind of questions might require a more intensive
type of data collection from a much smaller sample of farmers as a complement to the
larger survey.

CHAPTER 7
CONCLUSIONS AND RECOMMENDATIONS
The over-arching objective of this series of studies was to produce information
that could improve extension and adoption of agroforestry technologies in Haiti,
particularly of hedgerow intercropping. The experimental designs employed reflected
the range of information needed to understand agroforestry systems, and included on-
station and on-farm agronomic studies as well as a large-scale questionnaire-based
socioeconomic survey. The objectives of the three studies comprising the dissertation
were to determine the importance of soil water competition between hedgerows and
adjacent maize on station and on farm, and to examine the factors that influence farmer
adoption, adaptation, and management of hedgerow intercropping. These objectives
were attained, but with some limitations as indicated in the following discussion based on
the hypotheses tested in each study.
The on-station trial showed that soil water competition caused by Leucaena
hedgerows could reduce maize yield substantially, however the percent reduction in yield
depended on how limiting rainfall was during the season. The mitigating effect of
installing root barriers between the maize and the hedgerows was temporary, as Leucaena
roots re-grew under the barriers and into the alleys. Severing Leucaena roots during the
installation of the barriers resulted in a loss of 1,600 kg/ha Leucaena biomass over a
seven-month period. Thereafter, the hedgerow trees recovered and there was no
179

180
subsequent difference in biomass yield between plots having root barriers and those
without barriers.
The on-farm trial of maize development did not detect any differences at various
distances from three types of soil conservation structures and at various positions on the
slope. There were differences in maize development between types of soil conservation
structures and the untreated control, with the maize growing in plant-based structures
(hedgerows and crop bands) developing more slowly than maize in rock wall terraces and
plots without soil conservation structures. No statistical differences in maize yield were
detected, even though the untreated control plots produced numerically less than the
treated plots. The plots having plant-based structures yielded less maize than the rock
wall plots, indicating competition between maize and the plants in the hedges, but again
the difference was not significant. It is the opinion of the author that differences in maize
growth and yield do exist between flat land and sloping land. The failure to show
differences was probably due to the small sample size and short duration of the on-farm
trials.
No direct recommendation for the use of hedgerow root barriers on farms can be
made based on these results. Assuming the increased maize grain yield due to root
barriers on station could be realized on farm by a similar technique, say trenching on the
uphill side only of the hedgerows before each maize crop, it still might not make
economic sense to the farmer. On the station trial, increased maize grain yield due to
barriers was about 350 kg/ha. If a typical hillside half-hectare plot would realize an extra
175 kg of maize grain, the cost of installation would be about halfway between the range
of prices the extra maize would bring at the farm gate. In addition, the maize crop might
fail completely due to poor rain up to 30% of the time, so the risk to the farmer might be

181
too great for the expected benefit, unless another crop of higher value were planted or the
trenches bring some other value to the garden plot, such as increased water harvesting.
The on-farm survey hypotheses were supported by the results, but within a
limited range because the sampled farmers were all project participants and therefore
similar. Level of household land and labor resources in some cases influenced the choice
of technologies and the management quality applied to agroforestry practices.
Associations were detected that appeared to confirm that more available labor and land
improved management quality, and that farmers with better resources invested in more
expensive agroforestry practices. Farmers based their decisions about where to install
different soil conservation structures on characteristics of the plots. Mode of access and
soil fertility were important in the decision to install contour soil conservation structures
(hedgerows, crop bands, rock walls) but their relative influence could not be separated.
Farmers’ decisions to install hedgerows on plots having relatively poorer soil fertility and
less secure tenure could indicate they have not yet seen that practice as one returning
sufficient benefits to risk better plots, whereas the short-term benefits taken from the
crops bands apparently justify installing them on better plots. Soil fertility appeared to be
the most important determinant of management quality of the structures, with the best
management occurring on the most fertile plots for all practices. Tenure status of the plot
did not affect how well farmers managed hedgerows or any other agroforestry practice.
Hedgerow plots where farmers cited any of several potential benefits (wood products,
fodder, soil fertility and conservation, improved crop production) as being very important
were better managed. However, very few farmers rated wood products as important
hedgerow benefits. Over half of the interviewed hedgerow farmers cited soil fertility and
soil conservation as being very important hedgerow benefits. Because plots where that

182
opinion was expressed were significantly better managed, the farmers were probably
revealing their true attitudes toward hedgerows, as opposed to parroting an extension
message.
The conceptual thread of tree/crop soil water competition was not successfully
traced through the three studies described here. This was because the study designs
recorded different scales of data from environments of increasing variability. The small-
scale phenomenon of soil water competition close to hedgerows seen on the station was
not encountered on the four hillside farms comparing different soil conservation
techniques in the second study. A larger sample of farmers over several seasons, with
yield measured row by row might have picked it up. Farmers claimed that hedgerow/crop
soil water competition was a serious problem on less than one percent of hedgerow plots
surveyed. From observation, drought over the whole garden causing crop failure
probably covers up tree/crop competition for soil water on hillside farms.
These three studies represent components of what should be a broader program of
multidisciplinary trials and field exercises to understand the dynamics of hedgerow
performance and adoption in Haiti. The results are indicative of some aspects of
competition and adoption, and the links between them, but they are by no means
complete. It is evident that by focusing on a specific biophysical phenomenon on station,
such as soil water competition between hedgerow trees and adjacent maize, one can not
necessarily usefully apply those results to a system as complex as Haitian hillside farms.
There are three overriding, and interrelated, characteristics of hillside agriculture in Haiti
that make that so: complexity, variability, and need for cash.
Complexity is manifested in the number of plots of varying quality worked, the
several modes of access through which the plots are worked, numbers of crops

18
grown/animals raised, numbers of extra-agricultural enterprises undertaken by family
members, and the social networks and dependencies nurtured by the family. Great
variability is seen in soil quality and other physical properties within and between plots,
timing of rainfall compared to the needs of the crops, the quantity and timing of
agricultural labor available to the family from year to year, sale price of crops, and the
variability of crop yield. The need for cash for both cyclic and cataclysmic expenses in a
cash poor economy guides farmer decisions in a complex and highly variable farming
system. Complexity can be a response to soil and rainfall variability and the need for
cash, as a risk management strategy, such as holding several plots in different microsites
under different modes of access. But complexity in a new technology, such as
hedgerows, can make it difficult for a farmer to plan investment strategies that allow him
to manage it optimally, given his level of family resources and cash constraints. This is
because farmers’ price expectations from a plot contain both a mean and a cyclical
component, and farmers are unlikely to invest in a practice, like hedgerows, unless they
perceive it as affecting the mean component. Time of adoption for a given farmer occurs
when his or her beliefs about the expected value of profits from the new technology is at
least as great as the return from the old technology. But it takes time for farmers know
the variance of profits with the new technology. Because of the initial delay in realizing
benefits from hedgerows, due to loss of cropping space and time required to establish the
hedges, and because this takes place in a highly heterogeneous and variable cropping
system, the farmer has difficulty in discerning the mean component from the seasonal
variations. Hedgerows are an evolving, complex, relatively expensive agroforestry
practice being experimented with by very poor farmers in a highly variable and risky
social and agricultural environment. When new technologies are constantly being

184
modified, adoption equilibrium may never be attained because the parameters affecting
farmers’ decisions are changing as well.
Methods for studying adoption under these conditions were not completely met
by the research undertaken in this study. Additional farmer focus groups to follow up soil
water competition on farm and to separate the influence of mode of access on adoption
from that of soil fertility would have been useful. Most socioeconomic adoption models
have not incorporated the distinctive features of agroforestry, such as multiple outputs,
production variability, the economic role of trees, impact of off-farm employment, and
the sociocultural context of farmer decision-making. On farm methods of testing
dynamic agroforestry technologies have not been adequately tested. However, in spite of
the shortcomings of the methodology used in these studies, useful information was
developed that might guide future investigations. Some recommendations based on these
studies are:
It is more important to focus on stabilizing the soil and water flow on the
garden plot as a whole than to reduce competition at the tree/crop interface.
Tree/crop soil water competition was not observed in on-farm trials, nor
did farmers consider it to be a problem. It is unlikely that farmers would
make large investments to prevent soil water competition at the tree/crop
interface. Farmers should be occasionally surveyed to see if tree/crop soil
water competition becomes important, and under what conditions.
Analysis of survey data should compare farmers according to how long
they have been using particular agroforestry practices and whether or not
they are changing the structure and management of the practice. Non¬
adopters should be included in surveys to better understand how household
resources limit adoption.
Survey data should be supplemented with other data collection and
interpretation methods, such as farmer focus groups or a small number of
ethnographic studies of household economics.
Farmers’ assessment of soil fertility appears to be important in the decision
to adopt a practice and in how well it is managed. Since improving soil
fertility makes agronomic sense, a substantial proportion of farmers with

185
hedgerows say that soil fertility is a very important benefit of hedgerows,
and since farmers expressing that opinion manage their hedgerows better
than those who do not, extension programs should put more focus on
developing appropriate ways of increasing soil fertility.

APPENDIX A
POT STUDY OF SOIL WATER DEPLETION VS.
MAIZE LEAF WATER POTENTIAL
I established a pot study in January 1994 to determine how soil water depletion at
the ODH site correlated with plant water stress. I planted 12 pots of maize on 2 January
1994 with soil taken from the top 25 cm at the trial site, one plant per pot. Field capacity
of the soil measured at the trial site was about 41%. I stopped watering the maize in the
pots on 4 March, 60 DAS. Pre-dawn leaf water stress was measured with a pressure
chamber. Figure A-1 shows the result of four sets of measurements taken from 3 to 12
days after watering stopped, 63 to 72 DAS. After 72 DAS, the maize leaves were too dry
to give pressure chamber readings. Most of the plants were wilted by that time, and
some had fired leaves. After the final measurement at 74 DAS, I re-watered six of the
pots. Two pots did not recover from the wilting, the other four did. Figure A-1 shows
the mean values of the 12 pots as they dry from field capacity to just under 25% soil
water over the four observation dates.
Figure A-2 shows the regression relationship between leaf water stress and
percent soil water. When soil water percent in the root zone decreases below about 25%,
leaf water pressure passes below 10 bars. Maize plants begin to suffer drought stress at
this point.
186

187
Figure A-l: Changes in maize leaf water pressure as soil water percent decreases, 63 to
72 DAS (n = 12)
Figure A-2: Pot study regression of soil water percent against maize leaf water potential
in soil 25 cm deep taken from the study site.

APPENDIX B
ROOT DISTRIBUTION OF A LEUCAENA LEUCOCEPHALA HEDGEROW
This study was done in November 1990 on a nursery/demonstration site owned by a
community church group in the town of Mirebalais, located in Haiti’s lower central
plateau. There were no repetitions (and therefore no statistical analyses), just one
transect 3 m long through one Leaucaena hedgerow planted on a deep, sandy soil of
gentle slope (about 5%) on a river terrace. I mapped eight sections 1.2 m wide by 1 m
deep parallel to the hedgerow, beginning 200 cm uphill of the row, cutting through the
row of trees, and ending 100 cm downhill. A plexiglass plate inscribed with a 10 by 10
cm grid was pinned to the cut surface. The positions of the roots were drawn on plastic
sheets clipped over the Plexiglas plate, a separate symbol was used for each of four
diameter classes: very fine (10
mm). From 70 to 96% of the very fine roots were found in the top 30 cm of the soil, the
lowest percents being found 150 cm and 200 cm uphill from the hedgerow (Figure B-l).
A lesser percent of the fine roots were found in the top 30 cm (20 to 83%), with the least
(20%) at the 200 cm uphill position. Very few medium and large roots were found, but
of those that were, 50% or more were deeper than 30 cm.
188

Percent
189
Figure B-l. Percent of very fine ( (>10 mm) Leucaena root intersections in the 0-30 cm soil layer,
Mirebalais.

APPENDIX C
P-VALUES OF ANOVAS PERFORMED ON STATION

Table C-l. P-values for the analyses of variance done on the on-station trial of maize/leueaena soil water eompetion
p-values
Spring 1994
test
block
trench
fert
trench*fer
distance
trench
*dist
fert*dist
trench
*fert*dlst
depth
trench*
depth
fert*
depth
dist*
depth
trench
*fert*
depth
R2
42 DAS maize biomass
.001
.260
.142
.316
<.001
.780
.577
.172
.904
49 DAS maize biomass
<.001
.189
.156
.340
<.001
.857
<001
.022
.961
63 DAS maize biomass
.392
.085
.525
.380
<.001
.497
.414
.358
.810
70 DAS maize biomass
.995
.907
.463
.381
<.001
.831
.998
.618
.809
56 DAS soil water
.009
.837
.603
.081
.656
.905
.733
.531
<.001
.663
.456
.940
.183
.935
70 DAS soil water
.013
.504
.362
.105
.178
.020
.007
.477
<001
.593
.617
.529
.720
.602
Final grain
<.001
.805
.710
.015
<.001
.826
.454
.016
.765
Final # plants
.104
.354
.172
.485
<.001
.721
.898
.088
.691
Final # ears
<.001
.792
.724
.299
<.001
.264
.807
.114
.762
20 Nov. 93 LELE
biomass
.002
.074
.358
.198
.659
20 Nov. 93 LELE height
.001
.725
.076
.196
.684
33 DAS LELE biomass
.089
.139
.224
.091
.490
33 DAS LELE height
.506
.462
.840
.329
.274
65 DAS LELE biomass
.124
.300
.187
.994
.421
65 DAS LELE height
.059
.095
.393
.160
.500
20 Aug. 94 LELE
biomass
.508
.283
.941
.400
.282
20 Aug. 94 LELE height
.131
.403
.640
.640
.391

Table C-l--continued.
p-values
trench
trench* trench fert trench trench fert dist *fert*
block trench fert fert distance *dist *dist *fert*dist depth 'depth ‘depth 'depth depth R2
<001
.018
.223
.093
.023
.458
.386
.415
.748
.497
.839
.717
.298
.439
.155
.274
.656
.487
.146
.042
.345
1.000
<001
.071
.855
.487
<.001
.549
.376
.943
.022
.608
.031
.009
.385
.798
.076
.627
.441
.051
.497
.953
.488
.939
.067
.464
.157
.647
.256
.349
.494
.975
.187
.638
.571
.619
.074
.017
.477
.168
<001
.985
.745
.804
.234
.005
1.000
.471
.471
.006
.005
.401
.255
.629
.007
.038
.945
.331
.592
.509
<.001
.301
.358
.558
.744
.113
.633
.447
.263
.627
.013
.919
.033
.436
.152
.311
.311
.030
.466
Fall 1994
test
77 DAS maize biomass
42 DAS pms
42 DAS soil water
Final grain
Final # plants
Final # ears
28 DAS LELE biomass
28 DAS LELE height
14 Jan. 95 LELE
biomass
14 Jan. 95 LELE height
16 Mar. 95 LELE
biomass
16 Mar. 95 LELE height
May 95 LELE
biomass
6 May 95 LELE height

Table C-l -continued.
Spring 1995
test
block
trench
fert
trench
*fert
p-values
trench
trench fert trench trench* fert*d dist* *fert*
distance *dist *dist *fert* dist depth depth epth depth depth R2
40 DAS maize height
47 DAS maize biomass
54 DAS maize biomass
26 DAS soil water
61 DAS soil water
Final grain
Final # plants
Final # ears
19 DAS LELE biomass
19 DAS LELE height
57 DAS LELE biomass
57 DAS LELE height
130 DAS LELE
biomass
130 DAS LELE height
<.001
<.001
.449
.433
<.001
<.001
.652
.847
.512
.005
.035
.445
.423
.152
.202
.569
.676
.853
.001
.085
.287
.388
.324
.020
.654
.496
.809
<001
.373
.884
.095
.594
.310
.804
.927
<001
.859
.663
.514
.871
.914
.080
.344
.152
.386
.134
.996
.459
.994
<.001
.665
.075
.018
.398
.842
<.001
.001
.329
.346
.001
<.001
.248
.459
.714
.004
.002
.783
.180
<.001
<.001
.101
.447
.694
.003
.002
.840
.402
<.001
<.001
.066
.513
.715
.003
.246
.833
.478
.594
.061
.418
.156
.945
.463
<001
.213
.984
.310
.714
.334
.041
.553
.520
.387
.376
.164
.924
.840
.318
.921
.092
.664
.112
.273

Table C-2. P-values for the analyses of variance done on the on-station trial of maize/leucaena soil water competion
50 cm distance
100 cm distance
150 cm distance
200 cm distance
trench
fert
trench*
fert
trench
fert
trench*
fert
trench
fert
trench*
fert
trench
fert
trench*
fert
Spring 1994
42 DAS maize biomass
.039
.508
.126
.913
.231
.765
.229
.987
.090
.870
.307
.155
49 DAS maize biomass
.162
.387
.520
.427
.063
.575
.160
.002
.010
.799
.199
.648
56 DAS growth stage
.439
' .491
1.000
.423
.390
.531
.604
.324
.604
.064
.407
.250
63 DAS maize biomass
1 .039
I .384
.805
.311
.371
.200
.514
.215
.331
.767
.625
.547
70 DAS maize biomass
! .328
.427
.607
.694
.661
.254
.510
.745
.508
.716
.618
.980
56 DAS soil water, 0-15 cm
.580
.310
.877
.568
.797
.019
.850
.523
.585
.280
.778
.038
56 DAS soil water, 15-30 cm
.884
.703
.713 |
.293
.177
.218
.386
.910
.890
.704
.522
.394
56 DAS soil water, 30-45 cm
.427
.445
.923 j
.170
.004
.279
.379
.855
.641
.763
.457
.728
70 DAS soil water, 0-15 cm
.323
.177
.694
.186
.143
I .730
.451
.762
.325
.388
.352
.688
70 DAS soil water, 15-30 cm
.590
.643
.058
.804
.428
.646
.995
.731
.451
.231
.027
.176
70 DAS soil water, 30-45 cm
.159
.519
.161
.441
.949
.151
.544
.584
.859
.281
.164
.033
Final grain
.180
.141
.057 1
.512
.463
.219
.728
.550
.008
.623
.230
.043
Final # plants
.370
.572
.370 1
.709
.077
.184
.654
.191
.115
.523
.872
.069
Final # ears
.722
' .479
7023 j
.139 '
.918
.918
.361
.276
.276
.224
.942
.062

Table C-2-continued.
50 cm distance
100 cm distance
150 cm distance
200 cm distance
trench
fert
trench*
fert
trench
fert
trench*
fert
trench
fert
trench*
fert
trench
fert
trench*
fert
Fall 1994
77 DAS maize biomass
.037
.191
.635
.105
.508
.747
.285
.185
.030
.531
.613
.570
42 DAS pms
.081
.242
.775
.103
.471
.175
.829
.524
.829
1.000
.126
.232
42 DAS soil water, 0-15 cm
.666
.514
.306
.259
.110
.912
.402
.630
.981
.240
.465
.384
42 DAS soil water, 15-30 cm
.033
.546
.787
.405
.567
.966
.095
.120
.523
.264
.312
.467
42 DAS soil water, 30-45 cm
.002
.091
.093
.014
.617
.055
.886
.932
.178
.675
.617
.101
Final grain
.071
.467
.330
.068
.209
.043
.398
.272
.318
.177
.522
.163
Final # plants
.895
.362
.037
.020
.865
.865
.436
.602
.793
.883
.659
.310
Final # ears
.389
.160
.070
.617
.065
.176
.643
.360
.443
.939
.939
.159

Table C-2-continued.
50 cm distance
100 cm distance
150 cm distance
200 cm distance
trench
fert
trench*
fert
trench
fert
trench*
fert
trench
fert
trench*
fert
trench
fert
trench*
fert
Spring 1995
40 DAS maize height
<001
.035
.074
<001
.536
.141
<.001
.626
.517
<001
.234
.475
47 DAS maize biomass
.030
.201
.202
.096
.530
.566
.016
.750
.787
.017
.557
.285
54 DAS maize biomass
.072
.241
.260
.103
.484
.538
.111
.302
.950
.464
.019
.049
26 DAS soil water, 0-15 cm
.376
.926
.602
.641
.823
.136
.140
.780
.077
.242
.277
.588
26 DAS soil water, 15-30 cm
.140
.278
.065
.172
.720
.182
.718
.016
.010
.964
.615
.054
26 DAS soil water, 30-45 cm
.370
.562
.196
.429
.985
.309
.298
.112
.122
.412
.611
.006
61 DAS soil water, 0-15 cm
.743
.149
.859
.566
.222
.806
.201
.736
.361
.393
.555
.694
61 DAS soil water, 15-30 cm
.338
.674
.857
.358
.145
.329
.900
.419
.145
.836
.691
.336
61 DAS soil water, 30-45 crn
.724
.411 J
.233
.219
.571
.049
.678
.973
.720
.289
.884
.365
Final grain
<.001
.102
.211
.039
.286
.346
.333
.475
.594
.659
.543
.370
Final # plants
<.001
.074
.290
.045
.199
.199
.158
.894
.691
.779
.779
.063
Final # ears
<001
.068
.330
.011
.256
.752
.365
.596
.664
.825
.912
0.197
so
o\

APPENDIX D
HOUSEHOLD QUESTIONNAIRE

3. Nom chef kay-la:
Nimewo kocL_
Sinyati
b. Si genyen lét planté PLUS ak pwép kod pa-yo, ki n;
Prenom
m menm kay, ekri nom ak kod yo ¡sit:
/
Nimewo kod:
Sinyati
Prenom
/
Nimewo kod:
Sinyati
Prenom
c. Lokalite: d. Seksyon: e. Kominn:
f. Ann al fé lis chak móso té moun nan kay sa te genyen oubyen te travay nan 12 mwa k¡ sót pase:
Jaden
Lokalite
Tip iaden /I
Kantite té
(1/100 ex)
Sistem
fonsyé /2
Pozisyon
topografik /3
Klas pant /4
Aktivite PLUS ki
fet nan ¡aden /5
1
2
3
4
5
6
7
8
9
10
11
12
1 Tip jaden:
l=lakou, 2=pre kay, 3=lwen kay
2 Sistém fonsyé: l=té achte, 2=éritaj deja separe, 3=éritaj poko separe
4=demwatye, 5=anfeme nan men moun, 6=femyé leta
3 Pozisyon topo: l=tet mónn, 2=flan mónn, 3=pyé mónn, 4=laplenn 7=jeran, 8= Lot
4 Klas pant: l=plat, 2=panche, 3=tré panche “ ~
5 Aktivite PLUS: RV=ranp vivan; BM=bann manje; MS=misek; RP=ranp pay; BR=baraj ravinn; PB=pyebwa; JL=jaden legim; DR=danre; KO=konp

g. Anfómasyon sou moun ki abite nan kay-la:
h. Eske chef kay la moun bó isit? I jWi I I Non i. Depi konbyen ane chef kay la abite ¡sit? ane
j. Eske planté PLUS la te patisipe nan AFII? | |w¡ I I Non k. Nom ankété:
I. Depi konbyen ane planté-a konmanse patisipe ak PLUS? ane
Kod: 6 Seks moun: F=fi G=gason
7 Aktivite PLUS: RV=ranp vivan; BM=bann manje; MS=misek; RP=ranp pay; BR=baraj ravinn; PB=pyebwa; JL=jaden legim; DR=danre; KO=konpos; GR=gr
8 Lot Aktivite: 1=atelye, moulin kann; 2=atizana; 3=dokte fey; 4=elvaj, gadinaj; 5=fe chabon, lacho; 6=fonksyone; 7=jounalye jaden, vann jou
8=jounalye, lot; 9=komes pwodwi agrikol; 10=komes, lot; 11=peche; 12=siye bwa; 13=timetye; 14=lot aktivite

APPENDIX E
GARDEN PLOT QUESTIONNAIRE

PADF/PLUS ETID EKSTANSIF 1995
Data Collection Form for annual large-sample impact survey
I. IDANTIFIKASYON JADEN
a. Norn patisipan PLUS ki travay jaden sa:
Nimewo kod:
d. Depi konbyen ane planté s,
f. Lokalite:
~~ Prenom
in jaden PLUS pou fwaye sa. c. Jaden sa se nimewo nan lis jaden sou Dosye Fwaye.
n pwoje PLUS? ane e.Konbyen tan sa pran sóti lakay li rive nan jaden sa? rnjr
g. Seksyon: h. Kominn:
II. DESKRIPSYON JADEN
a. Valé té nan jaden sa:
□ 1=té achte, 2=éritaj deja separe, 3=éritaj poko separe
4=demwatye. 5=anfeme nan men moun, 6=femyé teta
7=jeran, 8= Lot
c. Eske jaden sa te an jaché pandan 1995?
e. Pant mwayén:
I lwi I. _ln(
f. Ekspozisyon pant:
d. Si wi, depi konbyen ane?
h. Konbyén pyebwa (dyamét >10 cm) genyen nan jaden sa?_
_ i. Konbyén rigól (>20 cm pwofondé) nan jaden?
Klasifikasyon lokal:
j- ¡
Dapre mét jaden an, éske té nan jaden sa fétil: (1=póv, 5=fét¡l)
Dapre mét jaden an, éske té nan jaden sa fon: (1=mens, 5=fon)
Eske té ki ant strikti KSDL montre siy ewozyon: (1=okenn, 5=anpil)
Tip wóch mé: Kalké | l Bazaltik | | Lot | l
PADF 11/97

IDNO:
PADF/PLUS ETID EKSTANSIF 1995
III. DANRE NAN JADEN PANDAN 199S
a. Dénezén danre, pri nan epók rekolt. ak estimasyon planté sou gwose rekolt danre ak KSDL vésis sistém tradisyonél
c. Konbyén jounen (óm jou) yo te peye (an nati oubyen nan kach) pou travay nan jaden sa pandan 1995? óm/jou
d. Konbyen moun kl nan lwaye-a tap travay nan jaden pandan 1995 moun
IV. AKTIVITE PLUS NAN JADEN SA
Kwaze chak ti bwat ki reprezante aktivite PLUS planté a té nan jaden sa. Pou chak aktivite. rampli ley pa l
I iRamp vivan I jMisék I IRamppav I iPvebwa I ISemans danre amélvore | iKompos
I iBannmanie I IKanal kontou I IBarairavinn I | Greta i sou gwo bwa 1 I Jaden legim
* NB: Nan ka plante-a pa kapab ba-w kantite rekolte, sévi ak kod sa yo: a=rekolt la rate okoz move tan b=rekolHa rate okoz bet c^rekoltla poko fet
Nom Asistan
PAOF 11/97
IO
O

IDNO: PADF/PLUS ETID EKSTANS1F 1995 EKIP:_
V. RANP VIVAN

It

IDNO:
PADF/PLUS ETID EKSTANSIF 1995
VII. MISEK
EKIP:_
a. Depi konbyen ane premyé misék nan jaden sa te fét? b. Konbyen ranje ki pa sou bon koub ñivo?,
c. Ranie/mét fét pandan 1995: ranje mét d. Ranie/mét (ét avan 1995: ranje mét
e. Distans ant misek yo? mét f. Akimilasyon té dévan misek yo? cm_
g. Konbyen ranje ou we ki tre byen jere? ranje h. Konbyen konsi konsa? ranje i. Konbyen move net? ranje
j. Konbyen bréch genyen nan misék yo? k. Eske mét jaden an konn repare bréch yo? I |wi | jnon
I. Ensékle nimewo ki endike anpétans mét jaden an bay bénéfis ki séti nan m
Misék pémét té a vjnn pi fasil pou travay 1
Misék kreye plis espas pou fé jaden 1
Misék sévi kom baraj pou krém té a réte nan jaden 1
Misék fé li jwenn plis rekolt danre nan jaden 1
l=pa enpótan ditou, 5=trezanpótan
2 3 4 5
2 3 4 5
2 3 4 5
2 3 4 5
m. Ensékle nimewo ki endike anpétans mét jaden an bay pwoblem nan mise
Eske misek yo okipe twop espas nan jaden?
Eske misek nan jaden bay difikilte , ou mare bét ?
Eske misek yo pran twop tan pou okipe yo?
l=pwoblem sa pa anpotan ditou, 5=pwoblem
2 3
2 3
2 3
n. Konbyen moun ki te ede konstwi misek yo?
I l Mesyé
l [Medanm
l ¡Gran ti mou
. O
o. Konbyen moun ki te ede repare misek yo?
I ¡Mesyé
l [Medanm
I [Gran ti mou
. O
p. Konbyen moun, ki pai patisipe nan PLUS, aplike teknik sa li te aprann nan men-ou? moun
DAT:,
PADF 11/97
â–¡ â–¡

IDNO:
PADF/PLUS ETID EKSTANSIF 1995
mi. BARAJ RAVINN
i. Konbyén baraj jaden sa genyen ki te konstwi pandan 1995, ak avan 1995?
Sak té
a¡ pandan 1995? _
b Ki danre met laden an te plante nan bara¡ yo pandan 1995?
c. Ki danre li rekolte
d. Ki valé latan li jwei
e. Konbyen rnoun nai
I. Konbyen moun nar
g. Konbyen baraj ou '
j. Ensékle
renn nan danre ki sfiti nan baraj yo pandan 1995?
lan fwayea ki te ede repare baraj yo?
u we ki tre byen jere? barai
ki endike repons pou kesyon swivan:
i. Konbyen me
Eske konstriksyon baraj yo teknikman korék?
Eske kondisyon aktyél baraj yo bon?
Eske mét baraj konn fé reparasyon?
Eske baraj yo de|a kenbe kek té ewode?
Eske baraj yo reziste gwo dlo lapli?
Eske baraj yo pwoteje kay ak té ki anba monn sa?
i. Konbyen moun. ki pal patisipe nan PLUS, aplike teknik sa
l [klesyé
1 liytedanm
1 l^ran ti mi
- â–¡
-
1 It/tesvé
1 1 Medanm
1 iGranti mi
- O—,
Pili O
i te aprann nan men or
PADF 11/97
DAT:.
O
Os

IDNO:
PADF/PLUS ETID EKSTANSIF 1995
IX. KANTITE PYEBWA NAN JADEN
a Konbyen pyebwa total ou te plante nan jaden s<
te plante nan jaden sa pandan 1995?
Ki jan ou te pwodwi\plante pyebwa 1995 yo? I ISemi direk
I IRootramer | |L6t véso | |L6t(ekri):
I iRasinnni I IRache plante I ISache plastik
11 pyebwa sa yo? I IPADF I IZanmi planté I I Pepmyé endividyél I iPepmyé an rwoup I I Lot kc
g. Konbyen moun travay nan jesyon ak rekolt pyebwa yo;
h. Ensékle mmewo kl endike |esyon ak devlopman pyebwa ki te plante nan 1995
Eske pyebwa yo soulri anba sechres?
Eske pyebwa yo soulri anba maladi ak ensek?
Eske pyebwa yo soutrl anba zanimo?
Eske pyebwa yo soulri doma| ki te tét nan saklaj jaden?
Eske move zeb domine pyebwa yo?
Eske lot danre domine pyebwa yo?
□ Mesyé □ Medanm □<— Ek
â–¡u
Ki espes pyebwa ki pi byen devlope?
Ki espes pyebwa ki pa byen devlope ditou’
. Nan tout pyebwa nan laden, ki pwodwi met ianden rekolte pandan 1997? | |Fw¡ I |Mame bét I [Planch I |Poto I |Gol
I IChabon | IBwadité I iBwaklisai | iPyepran I iPikel
Nan tout pyebwa nan jaden, ki valé lanjan met jaden jwenn nan pwodwi li rekolte pandan 1997?
i Konbyen moun, ki pat patisipe nan PLUS, aprann teknik plante pyebwa li te aprann nan men-ou?
PADF 11/97

IDNO:
PADF/PLUS ETID EKSTANSIF 1995
X. KANTITE BWA GREFE NAN JADEN
a. Konbyen pyebwa k¡ te grefe nan jaden sa avan 1995?
c. Konbyen grefon ki te grefe sou pyebwa sa yo?
e. Konbyen pyebwa ki te grefe nan jaden sa pandan 1995?
g. Konbyen grefon ki te grefe sou pyebwa sa yo?
1 Konbyen grefon te pran?
l. Konbyen grefon te pi
Ki tés moun tefe grefaj pandan 1995? 1 Ueknisyen PADF I I Bós grefé 1 lEkstansyonis I 1 Planté a I I Lot mou
l. Konbyen moun te travay nan |esyon fwi grefe yo? □ Mesyé □ Medanm Ti moun piti n.
<. Konbyen moun te travay nan rekolt fwi grefe yo? □ Mesyé □ Medanm □ Gran ti moun □ Ti moun piti □ Lé
I. Pou pyebwa ki te grefe pandan 1995, ki bó ou te jwenn grefon yo? I I PADF I I Bós grefé I iLakav mwen I I Lot kote
m. Eske met jaden te elimine gouman yo? | I Li te fe li korek n<
n. Eske met jaden rekolte twi grefe pandan 1995?
o. Si wi, konbyen ak ki kalite fwi li rekolte?
I I Li te koupe kek landan
d]w¡ CZHn.
I I Li pa te fe li ditou
alite Fwi Li Rekolte
Kantite Fwi Li Rekolte
p. Ki sa met jaden an te fe ak fwi yo? I iFanmi li manje yo I I Li bay lot moun manie yo I Hivannyu
q. Si li te vann fwi yo, konbyen lajan li iwenn nan vant la? gdcs.
â– . Konbyen moun, ki pat patisipe nan PLUS, aplike teknik grefe pyebwa li te aprann nan men ou?
PADF 11/97
DAT..
208

(I. JADEN LEGIM
s Konbyen fwa met jaden an plante legim pandan 1995? _
PADF/PLUS ETID EKSTANSIF 1995
3. Ki gwosé jaden legim nan?
c. Ki kalite legim li te pi;
Konbyen moun te travay nan jaden legim sa pandan 1995?
Esékle mmewo ki dekri pwoblem nan jaden sa pandan 1995
Varyete ak kalite semans
â–¡ Medanm â–¡ (
i Gran ti moun I l Ti moun ptti
(1-pa gen pwoblem sa ditou. 5=pwoblem sa te grav anpil)
o.
Lót pwoblem
f Esékle mmewo ki dekri anpótans benefis planté a te jwenn nan jaden pand
Vann legim
Vann plantil
Manje legim lakay
Bay zanmi legim kom kado
Lót benefis
g Ki sa met jaden an te mete nan jaden an pou ogmante fetilite? 1 |mi
h. Ki sa met jaden an aplike nan jaden an pou kontwole ensek ak maid jPestisid ógamk 1 I pestisid chimik
i Eske met jaden te resevwa fómasyon nan pestisid ógamk? (Z=]w, anon
j Kwaze ti bwat ki endike kalite depans an kach mét jaden an te fe pandan 1997
I Iachte semans | lachte plantil I 1 Mendév nan jaden I [Pestisid
I [konpos
j Ipqupou bét I lanRre chimik
I lli kontwole yo a men
k Ki depans total h te fe pou jaden sa pandan 1997?
I lAngre I ~ iTranspó nan mache
K» valé total mét jadan an reyisi nan jaden sa pandan 1997? Gdes
in, ki paj patisipe nan PLUS, aprann tekmk jaden legim li te aprann nan men ou?
PADF 11/97

IDNO:
PADF/PLUS ETID EKSTANSIF 1995
XII. REMAK JENERAL SOU JADEN SA
EKIP:_
DAT:
PADF 11/97

APPENDIX F
SOIL NUTRIENTS, ORGANIC CARBON STATUS. AND pH
OF 175 HILLSIDE GARDENS
As a part of the interview that took place during the garden visits (n=2,295),
technicians asked the farmers to evaluate several soil parameters of the plot. A
qualitative scale of 1 to 5 was used. The parameters were soil fertility (l=poor, 5=very
fertile), soil depth (l=shallow, 5=deep), an evaluation of the hot/cold scale1 (l=hot,
5=cold), and signs of soil erosion between the rows of soil conservation structures on the
plot (l=none, 5=many). In addition, the farmer was asked to give the local Creole name
for the soil type, and the technicians noted the type of parent material (calcareous or
basaltic).
The relationship of Haitian indigenous soil classification systems to the western
systems is not well understood (McLain and Stienbarger 1988). Since these qualitative
opinions about soil were used as independent variables in analyses of farmers’ decision
to install and manage various agroforestry practices, it is necessary to understand if they
relate to measurable fertility parameters.
Method
Cross tabulations of farmers’ soil names and ratings were done using information
from all 2,295 gardens visited. Soil samples were taken for laboratory analysis from 175
'Most writers agree that “hot” soils (cho) are dry, well-drained soils and “cold” soils
(fwet) have more soil water available for plants. Other concepts are integrated into this
system as well, including soil parent material, slope, orientation, and vegetative cover
(McLain and Stienbarger 1988, Murray 1981). Some crops grow better in hot soils, some
in cold, but in general cold soils are preferred (Smucker 1981).
211

212
gardens selected at random from all gardens visited, 35 each from gardens having
fertility classes 1 through 5. Two samples were mislabeled and could not be used. The
samples were taken by technicians using a machete from the 0-20 cm layer; a composite
of four positions for each sample. They were put into ziplock bags and labeled, then
transported to the University of Florida for laboratory analysis. Inductively coupled
argon plasma spectroscopy (ICAP) was used with the Mehlich 3 procedure for Ca, Mg,
K, and P. ICAP was also used to determine water extractable pH. Percent organic matter
was determined using the Walkley-Black procedure. Statistical analyses were done using
SPSS; nonparametric tests (chi-square, Mann-Whitney) because distributions were not
normal and numbers of observations were not equal for all factor levels. The 1 to 5
scales were recoded by combining ratings 4 and 5 into one category (the “best” end of the
scale) and ratings 1, 2, and 3 into another category (the “worst” end of the scale). The
comparisons were therefore fertile vs. not fertile, deep vs. not deep, not eroded vs.
eroded, and cold vs. not cold soil.
Results
Farmers used 21 different names for their soils in the 2.279 gardens for which
they gave that information. Of these, four are based on color, six refer to parent material
or texture, six refer to the fertility, water-holding ability, or other capacity of the soil to
produce crops, and five refer to miscellaneous or unknown properties. Table F-l
summarizes these, and gives the percent of each named soil given the top two fertility
ratings (categories 4 and 5).

213
Table F-l: Haitian farmer soil classifications
Group
Creole name
Translation
No. of cases
% fertile
Color
Nwa
Black
578
51
Wouj
Red
228
38
Jonn
Yellow
69
52
Gri
Grev
46
33
Parent material
Grennen
Grainy?
256
45
or texture
Sab
Sand
246
40
Tif
Lime
174
36
Ajil
Clay
111
65
Woch
Rock
92
30
Fen
Fine
13
15
Fertility, water,
Gra
Fat
167
83
productive capacity
Cho
Hot
117
14
Meg
Thin
77
8
Pat
Paste
55
73
Dio
Water
9
89
Fwet
Cold
5
80
Miscellaneous
Gro
Big
17
29
Ravet
Roach
15
0
Men
Unknown
2
50
Si
Sour
1
0
Grison
Unknown
1
0
It appears that black, yellow, clay, fat, paste, water, and cold soils tend to be
fertile, while thin, fine, and hot soils tend towards the infertile end of the scale.
However, since farmers were sampled over four large regions of the country from the
north to the south, several of these soil names probably refer to the same underlying
characteristics. For example, Figure F-l shows that the distribution of fertility ratings
from 1 (very infertile) to 5 (very fertile) is skewed to the left—farmers tend to give their
soils high ratings. If hot always meant infertile and cold fertile a corresponding pattern
would be expected in the hot/cold scale, but this is not the case. Figure F-2 shows that
the distribution of hot/cold soils is skewed to the right, farmers tend to rate their soils as
hot more frequently than cold. A cross tabulation of the fertility ratings with the hot/cold

214
ratings shows that 63% of the 431 gardens given the top two cold ratings are also given
the top two fertility ratings. However, 41% of the not cold gardens (n= 1,864) are also
rated as fertile, so the distinction is not clear-cut.
Farmer fertility rating
Farmer "hot/cold" rating
Figure F-l: Distribution of fertility ratings Figure F-2: Distribution of “hot/cold”
ratings
However, if cross tabulations are done using the soil names hot, cold, fat, and thin
against soil fertility ratings, stronger correlations are shown: 80% of cold soils (n=5) are
rated as fertile, and 86% of hot soils (n=l 17) are rated as not fertile; 83% of fat soils
(n=167) are rated as fertile, and 92% of thin soils (n=77) are rated as not fertile. Because
the cold name was used in only five gardens, doubtless the same concept is included in
other soil names.
The mean and median pH. nutrient, and organic carbon values for all 173 plots
where soil samples were taken are shown in Table F-2. Based on interpretation tables for
the Mehlich 3 procedure, the phosphorus level is low for maize, potassium is high,
calcium and magnesium are not limiting. Maize planted in soils having these values

would probably respond to phosphorus and nitrogen, and possibly micronutrients (R.
Tucker, pers. comm.).2
215
Table F-2: pH, nutrient, and percent organic carbon for 173 gardens
pH
P (mg/kg)
K
(mg/kg)
Mg
(mg/kg)
Ca
(mg/kg)
%OC
Mean
7.7
5.6
143.6
435.5
14,783
1.7
Median
8.0
3.6
117.0
261.0
13.500
1.5
There were statistically significant differences between soils rated as fertile and
those rated as not fertile (Table F-3). Fertile soils had a lower pH, more potassium, and
more organic carbon. However, since the difference in pH is very small and potassium is
abundant in both classes of soils, the difference in organic carbon is the most interesting
result. These differences are indicative, but doubtless do not reveal everything farmers
integrate into the concept of fertility.
Table F-3: Nutrients, pH, and % organic carbon for soils rated as fertile and not fertile
PH
P
(mg/kg)
K
(mg/kg)
Mg
(mg/kg)
Ca
(mg/kg)
%
OC
n
Fertile
Mean
7.6
6.1
161.2
439.6
13,287
1.9
69
Median
7.9
3.6
132.0
287.0
12,100
1.6
Not fertile
Mean
7.8
5.3
131.8
432.7
15,776
1.5
104
Median
8.1
3.6
104.5
243.5
13.850
1.4
p-value
.004
.268
.020
.171
.117
.018
Similar comparisons were made between deep and not deep soils, eroded and not
eroded soils, and cold and not cold soils. Each had only one significant difference: deep
soils had 21 mg/kg more potassium than not deep soils (p = .023), eroded soils had 143
2Based on interpretation tables for maize in North Carolina, the southernmost state in
the U.S. that uses Mehlich 3.

216
mg/kg less magnesium than not eroded soils (p = .030), and cold soils had 30.5 mg/kg
more potassium than did not cold soils (p = .081). These differences do not appear to be
important.
Limestone substrate underlies 80% of the land area in Haiti; the rest is basaltic or
alluvial (Ehrlich et al. 1985). The soils found in basaltic areas are visually different than
those found in limestone areas, they are of a coarser texture and they erode faster. Only
limestone (68%) and basaltic (32%) soils were recorded in this survey. Do farmers'
fertility ratings differ between these two categories of soils, and are the differences in
fertility shown in Table F-3 reflecting the differences in parent material? A cross
tabulation revealed that 72% of fertile soils (n=1032) and 65% of not fertile soils
(n=1262) were of limestone origin. This appears to indicate that limestone soils are more
likely to be considered fertile. Table F-4 shows the differences as found in laboratory
analyses.
Table F-4: Nutrients, pH, and % organic carbon for basaltic and limestone soils
pH
P
K
Mg
Ca
°/o
n
(mg/kg)
(mg/kg)
(mg/kg)
(mg/kg)
oc
Basaltic
Mean
7.5
6.2
112.6
595.9
10,166
1.2
59
Median
8.0
3.7
90.5
326.0
8,980
0.9
Limestone
Mean
7.8
5.3
159.6
352.4
17,172
2.0
114
Median
8.0
3.6
131.0
256.5
15.300
1.7
p-value
.097
.422
.001
.218
<.001
<.00
1
The expected differences in pH and calcium are found, limestone soils having
higher values for both. Limestone soils also have significantly more potassium and a
higher percentage of organic carbon. The difference in organic carbon between basaltic

217
and limestone soils is somewhat greater than the difference between soils rated as fertile
and not fertile (Table F-3).
Conclusions
Soil fertility ratings as expressed by farmers in the survey appear to be in
agreement with opinions expressed in the literature (Murray 1981, McLain and
Stienbarger 1988, Smucker 1981), for example the hot/cold fat/thin scales. They also
generally conform to laboratory results, but the soil analysis does not completely explain
them. The ratings are relative, subjective, and relate to other nearby plots, not to plots
over the whole country. Although farmers rate limestone soils as fertile more often, the
basaltic soils in the north of the country often produce better crops because rainfall is
higher and more regular. Farmers integrate rainfall, regularity of rainfall, slope, water
holding capacity and other undiscovered factors into their fertility ratings. However,
their ratings as expressed in the survey appear to be well founded enough to be used in
analysis of adoption and management choices.

REFERENCES
Agus F, Garrity DP, Cassel DK and Mercado A (1999) Grain crop response to contour
hedgerow systems on sloping oxisols. Agroforestry Systems 42: 107-120.
Akonde TP, Leihner DE and Steinmuller N (1996) Alley cropping on an Ultisol in
subhumid Benin. Part 1: long-term effect on maize, cassava and tree
productivity. Agroforestry Systems 34: 1-12.
Anderson LS and Sinclair FL (1993) Ecological interactions in agroforestry systems.
Agroforestry Abstracts 6: 57-91.
Ashley MD (1986) Illustrations of gardens and land use representing agroforestry
systems in Haiti. UMO/AFORP. Port au Prince, Haiti: USAID.
Balasubramanian V (1983) Alley cropping: can it be an alternative to chemical
fertilizers in Ghana? Third national maize workshop. Kumasi, Ghana: Kwadeso
Agricultural College.
Balzano A (1997) A Haitian community: ten years after. (Unpublished)
Banda AZ, Maghembe JA, Ngugi DN and Chome VA (1994) Effect in intercropping
maize and closely spaced Leucaena hedgerows on soil coservation and maize
yield on a steep slope at Ntcheu, Malawi. Agroforestry Systems 27: 17-22.
Bannister ME and Nair PKR (1990) Alley cropping as a sustainable technology for the
hillsides of Haiti: experience of an agroforestry outreach project. American
Journal of Alternative Agriculture 5: 51-57.
Bellerive PA (1991) A financial analysis of selected hedgerow operations in Haiti's
southern and northwestern regions, pp. 27-31. SECID/Aubum Agroforestry
Report.
Belsky AG (1994) Influences of trees on savanna productivity: tests of shade, nutrients,
and tree-grass competition. Ecology 75: 922-932.
Benge MD (1980) Agro-forestation systems using Leucaena as erosion control barriers.
1980. Washington, DC, Office of Agriculture, Agro-forestation, Development
Support Bureau, USAID.
218

219
Benge MD (1985) Memorandum to USAID/Haiti regarding Leucaena seed dated 25
September 1985.
Bloch P, Lambert V and Singer N (1987) Land tenure issues in rural Haiti. Review of
the evidence and research and policy agendas. Draft for comments—May 1987.
DAN-5301-1-00-5025-00. 1987. Port au Prince, Haiti: USAID
Bohringer A and Leihner DE (1997) A comparison of alley cropping and block planting
systems in sub-humid Benin. Agroforestry Systems 35: 117-130.
Bonnard P and Scherr S (1994) Withing gender differences in tree management: is
gender distinction a reliable concept? Agroforestry Systems 25: 71 -93.
Brochet M (1993) Les strategies de lutte contre l’erosion et l'amenagement des bassins-
versants en Haiti. Tiers Monde 34: 423-436.
Byerlee D and Hesse de Polanco E (1986) Farmers' stepwise adoption of technological
packages: evidence from the Mexican altiplano. American Journal of
Agricultural Economics 68: 519-527.
Cannell MGR, Van Noordwijk M and Ong CK (1996) The central agroforestry
hypothesis: the trees must acquire resources that the crop would not otherwise
acquire. Agroforestry Systems 34: 27-31.
Carter J (1995) Alley farming: have resource-poor farmers benefitted? ODI Natural
Resource Perspectives.
Chamshama SAO, Mugasha AG, Klovstad A, Haveraaen O and Maliondo SMS (1998)
Growth and yield of maize alley cropped with Leucaena leucocephala and
Faidherbia albida in Morogoro, Tanzania. Agroforestry Systems 40: 215-225.
Chéry Y (1990) Les haies vives et leur place dans la production agrocole: Expérience du
Projet Ouest Jacmel, pp. 18-22. Port au Prince, Haiti, Groupe de Recherches,
d'Etudes et de Formation en Environment, Secretariat Technique á
l'Amenagement des Bassens Versants (STABV/MARNDR, USAID). Actes du
Colloque sur les Haies Vives pour une Agriculture Continue et Soutenue, 12
December 1989.
Chirwa PW, Nair PKR and Kamara CS (1994a) Soil moisture changes and maize
productivity under alley cropping with Leucaena and Flemingia hedgerows at
Chalimbana near Lusaka, Zambia. Forest Ecology and Management 64: 231 -243.
Chirwa PW, Nair PKR and Nkedi-Kizza P (1994b) Pattern of soil moisture depletion in
alley cropping under semiarid conditions in Zambia. Agroforestry Systems 26:
89-99.

220
Comía RA, Paningbatan EP and Hakansson I (1994) Erosion and crop yield response to
soil conditions under alley cropping systems in the Philippines. Soil & Tillage
Research 31: 249-261.
Corlen JE, Black CR, Ong CK and Monteith JL (1992a) Above- and below-ground
interactions in a leucaena/millet alley cropping system. II. Light interception and
dry matter prodution. Agricultural and Forest Meteorology 60: 73-91.
Corlen JE, Ong CK, Black CR and Monteith JL (1992b) Above- and below-ground
interactions in a leucaena/millet alley cropping system. I. Experimental design,
instrumentation and diurnal trends. Agricultural and Forest Meteorology 60:
53-72
Cunard AC and Lorcy MR (1990) Soil erosion estimations from the cropping systems
trial at the Brutus site, 1990. Associates in Rural Development. Port au Prince,
Haiti: USAID.
Current D, Lutz E and Scherr SJ (1995) The costs and benefits of agro forestry to
farmers. World Bank Research Observer 10: 151-180.
Current D and Scherr SJ (1995) Farmer costs and benefits from agroforestry and farm
forestry projects in Central America and the Caribbean: implications for policy.
Agroforestry Systems 30: 87-103.
DavidS (1995) What do farmers think? Farmer evaluations of hedgerow intercropping
under semi-arid conditions. Agro forestry Systems 32: 15-28.
de Geus JG (1973) Fertilizer Guide for the Tropics and Subtropics. Centre d'Etude de
l'Azote, Zurich.
de Young M (1958) Man and Land in the Haitian Economy. Gainesville, FL,
University of Florida, University of Florida Latin American Monographs No. 3.
Dijkman MJ (1950) Leucaena-a promising soil-erosion-control plant. Economic
Botany 4: 337-349.
Dvorak KA (1991) Methods of on-farm, diagnostic research on adoption potential of
alley cropping. Agroforestry Systems 15: 167-181.
Earle TR, Rose CW and Brownlea AA (1979) Socio-economic predictors of intention
towards soil conservation and their implication in environmental management.
Journal of Environmental Management 9: 225-236.
Ehrlich M, Conway F, Adrien N, LeBeau F, Lewis L, Lauwerysen H, Lowenthal 1,
Mayda Y, Paryski P, Smucker G, Talbot J and Wilcox E (1985) Haiti Country
Environmental Profile: a field study. Port au Prince, Haiti: USAID.

221
Ehui SK (1988) Evaluating the impact of soil erosion and productivity loss on the
relative profitability of alternative land use systems in the tropics. Evidence from
Western Nigeria. Lima, Peru, CIP. S econd biennial meeting of the Rockefeller
Foundation social sciences research program.
Emmanuel PC (1990) Une strategic en matiere de defense et restauration des sols—le
cas de la zone Perches/Acul Samedi (Nord-est d’Haiti). Port au Prince. Haiti.
Ervin CA and Ervin DE (1982) Factors affecting the use of soil conservation practices:
hypotheses, evidence, and policy implications. Land Economics 58: 277-292.
Ewel J (1977) A report on soil erosion and prospects for land restoration in Haiti. Port
au Prince, Haiti: USAID.
Feder G, Just RE and Zilberman D (1985) Adoption of agricultural innovations in
developing countries: a survey. Economic Development and Cultural Change 33:
255-298.
Feder G and O'Mara GT (1982) On information and innovation diffusion: a Bayesian
approach. American Journal of Agricultural Economics 64: 145-147.
Fernandes EC, Davey CB and Nelson LA (1993) Alley cropping on an acid soil in the
Upper Amazon: mulch, fertilizer, and hedgerow root pruning effects. In:
Ragland, J and Lai, RJ (eds) Technologies for sustainable agriculture in the
tropics: Proceedings of two international symposia held in San Antonio TX and
Denver, CO, 1990 and 1991 respectively, pp. 77-96. American Society of
Agronomy.
Femandez-Comejo J, Beach ED and Huang W-Y (1994) The adoption of IPM
techniques by vegetable growers in Florida, Michigan and Texas. Journal of
Agronomy and Applied Economics 26: 158-172.
Friedman JRP (1955) Developmental planning in Haiti: a critique of the U.N. report.
Economic Development and Cultural Change 4: 39-54
Fujisaka S (1991) Thirteen reasons why farmers do not adopt innovations intended to
improve the sustainability of upland agriculture. In: Dumanski, J, Pushparajah, E,
Latham, M, and Myers, R (eds) IBSRAM proceedings No. 12 (2). Evaluation for
sustainable land management in the developing world. Volume 2: technical
papers.
Fujisaka S (1993) A case of farmer adaptation and adoption of contour hedgerows for
soil conservation. Experimental Agriculture 29: 97-105.
Fujisaka S and Cenas PA (1993) Contour hedgerow technology in the Philippines: not
yet sustainable. Indigenous Knowledge & Development Monitor 1:14

222
Fujisaka S, Mercado AR, Jr and Garrity DP (1995) Farmer adaptation and adoption of
contour hedgerows for soil coservation. In: Kang, BT, Osiname, AO, and Larbi,
A (eds) Ibadan, Nigeria, Alley Farming Network for Tropical Africa. Alley
farming research and development. Proceedings of the International Conference
on Alley Farming, 14-18 September 1992, Ibadan, Nigeria, pp. 547-555.
Gaddanakeri SA, Gumaste SK, Hunshal CS and Angadi SS (1994) Effects of Leucaena
live bund on growth and yield of companion crop of rabi sorghum. Journal of
Maharashtra Agricultural Universities 19: 244-247.
Garrity, DP (1994) The fate of organic matter and nutrients in agroforestry systems. In:
Lefroy, RDB, Blair, GJ, and Craswell, ET (eds) Soil organic matter management
for sustainable agriculture: a workshop held in Ubon, Thailand, 24-26 August
1994, ACIAR,pp. 69-77.
Garrity DP (1996) Tree-soil-crop interactions on slopes. In: Ong CK and Huxley PA
(eds) Tree-Crop Interactions, a Physiological Approach, p 299-318. Wallingford,
UK: CABI.
Garrity, DP and Mercado, AR, Jr (1994) Nitrogen fixation capacity in the component
species of contour hedgerows: how important? Agroforestry Systems 27: 241 -
258.
Garrity DP and Van Noordwijk M (1995) Research imperatives in conservation farming
and environmental management of sloping lands: an ICRAF perspective. Bogor,
Indonesia: ICRAF
George SJ, Kumar BM, Wahid. PA and Kamalam NV (1996) Root competition for
phosphorus between the tree and herbaceous components of silvopastoral systems
in Kerala, India. Plant and Soil 179: 189-196.
Ghosh SP, Mohan Kumar B, Kabeerathumma S and Nair GM (1989) Productivity, soil
fertility, and soil erosion under cassava based agroforestry systems. Agroforestry
Systems 8: 67-82.
Govindarajan M, Rao MR, Mathuva MN and Nair PKR (1996) Soil-water and root
dynamics under hedgerow intercropping in semiarid Kenya. Agronomy Journal
88:513-520.
Haggar JP and Beer JW (1993) Effect of maize growth of [on] the interaction between
increased nitrogen availability and competition with trees in alley cropping.
Agro forestry Systems 21: 239-249.
Harper JK, Rister ME, Mjelde JW, Drees BM and Way MO (1990) Factors influencing
the adoption of insect management technology. American Journal of Agricultural
Economics 72: 997-1005.

223
Hauser S and Gichuru MP (1994) Root distribution of Dactyladenia (Acioa) barteri and
Senna (Cassia) siamea in alley cropping on Ultisol. II. Impact on water regime
and consequences for experimental design. Agroforestry Systems 26: 9-21.
Hauser S and Kang BT (1993) Nutrient dynamics, maize yield and soil organic matter in
alley cropping with Leucaena leucocephala. In: Mulongoy, K and Merckx, R
(eds) Soil organic matter dynamics and the sustainability of tropical agriculture:
proceedings of an International Symposium, Leuven, Belgium.
Hernandez CC (1961) A new method of farming hillsides. Coffee and Cacao Journal
14: 5.
Herrero MP and Johnson RR (1981) Drought stress and its effects on maize
reproductive systems. Crop Science 21: 105-110.
Hildebrand PE and Russell JT (1996) Adaptability Analysis: A Method for the Design,
Analysis and Interpretation of On-Farm Research-Extension. Ames: Iowa State
University Press.
Holdridge LR (1947) Determination of world plant formations from simple climatic
data. Science 105: 367-368.
Isaac L and Shannon DA. (1994) Evaluation of tree species adaptation for alley
cropping in four environments in Haiti. B. First year of pruning. SECID/Aubum
PLUS Report No. 15. Port au Prince, Haiti.
Isaac L, Shannon DA, Brockman FE and Bernard CR. (1995) The effects of Leucaena
hedgerow management on maize and hedgerow biomass yields over two years of
cropping. SECID/Aubum PLUS Report No. 27. Port au Prince, Haiti.
Isaac L, Shannon DA, Brockman FE and Bernard CR (1996) The effects of alley
cropping and other soil conservation practices on maize (Zea mays) yields over
two years of cropping. SECID/Aubum PLUS Report No. 30. Port au Prince,
Haiti.
Jama B, Nair PKR and Rao MR (1995) Productivity of hedgerow shrubs and maize
under alley cropping and block planting systems in semiarid Kenya. Agroforestry
Systems 31: 257-274.
Jarvis LS (1981) Predicting the diffusion of improved pastures in Uruguay. American
Journal of Agricultural Economics 63: 495-502.
Jeanes KW, Gutteridge RC and Shelton HM (1996) Competition effects between
leucaena and maize grown simultaneously in an alley cropping system in sub¬
tropical Australia. Experimental Agriculture 32: 49-56.

224
Jonsson K. Fidjeland L, Maghembe JA and Hogberg P (1988) The vertical distribution
of fine roots of five tree species and maize in Morogoro, Tanzania. Agroforestry
Systems 6: 63-69.
Kamasho JA (1994) Agroforestry in the southern highlands of Tanzania. Norwegian
Journal of Agricultural Sciences 21, Supplement: 87-92.
Kang BT, Reynolds L and Attah-Krah AN (1990) Alley farming. Advances in
Agronomy 43: 315-359.
Kang BT and Wilson GF (1987) The development of alley cropping as a promising
agroforestry technology. In: Steppler HA and Nair PKR (eds) Agroforestry: a
decade of development, pp 227-243. Nairobi, Kenya: 1CRAF.
Kang BT, Wilson GF and Lawson TL. (1984) Alley cropping: a stable alternative to
shifting cultivation. Ibadan, Nigeria: UTA.
Kang BT, Wilson GF and Sipkens L (1981) Alley cropping maize {Zea mays L.) and
leucaena (Leucaena leucocephala Lam.) in southern Nigeria. Plant and Soil 63:
165-179.
Kass DL, Araya JS, Sanchez JO, Pinto LS and Ferreira P (1995) Ten years of
experience with alley farming in Central America. In: Kang BT, Osiname AO
and Larbi A (eds) Alley Farming Network for Tropical Africa. Alley farming
research and development. Proceedings of the International Conference on Alley
Farming, 14-18 September 1992, pp. 393-402. Ibadan, Nigeria.
Kessey JF and O'Kting'ati A (1994) An analysis of some socioeconomic factors
affecting farmers involvement in agroforestry extension projects in Tanzania.
Annals of Forestry 2: 26-32.
Khaleque K and Gold MA (1993) Pineapple agro forestry: an indigenous system among
the Garo community of Bangladesh. Society & Natural Resources 6: 71-78.
Kiepe P (1995) Effect of Cassia siamea hedgerow barriers on soil physical properties.
Geoderma 66: 113-120.
Knudson MK (1991) Incorporating technological change in diffusion models.
American Journal of Agricultural Economics 73: 724-733.
Korwar GB and Radder GD (1994) Influence of root pruning and cutting interval of
Leucaena hedgerows on performance of alley croped rabi sorghum. Agroforestry
Systems 25: 95-109.
Korwar GB and Radder GD (1997) Alley cropping of sorghum with Leucaena during
the post-rainy season on vertisols in semi-arid India. Agroforestry Systems 37:
265-277.

225
Koudokpon V, Brouwers J, Versteeg MN and Budelman A (1994) Priority setting in
research for sustainable land use: the case of the Adja Plateau, Benin.
Agroforestry Systems 26: 101-122.
Lai R (1989) Agroforestry systems and soil surface management of a tropical alfísol: I:
Soil moisture and crop yields. Agroforestry Systems 8: 7-29.
Lawry S, Stienbarger D and Jabbar MA (1994) Land tenure and the potential for the
adoption of alley farming in West Africa. Outlook on Agriculture 23: 183-187
Lea JD (1993) Initial financial evaluation of hedgerows. SECID/Aubum PLUS Report
No. 17. Port au Prince, Haiti.
Lea JD (1994a) Project PLUS baseline information (revised). SECID/Aubum PLUS
Report No. 18. Port au Prince, Haiti,.
Lea JD (1994b) Rates of adoption of PLUS project interventions, Northwest Haiti.
SECID/Aubum PLUS Report No. 22. Port au Prince, Haiti.
Lea JD (1995) PLUS project hedgerow case study report. PADF zones: Gayita, Banat,
and Mirebalais; care zone: Bombardopolis. Harvests of December 1994 -
February 1995. SECID/Aubum PLUS Report. Port au Prince, Haiti.
Lea JD (1996) Analysis of CARE 1995 Grande Anse Baseline Data (draft).
SECID/Aubum PLUS Report. Port au Prince, Haiti.
Leybum JG (1941) The Haitian People. New Haven, Connecticut: Yale University
Press.
Livesley SJ, Gregory PJ and Buresh RJ (1997) Approaches to modeling root growth and
the uptake of water and nutrients. Agroforestry Forum 8: 24-27.
LocherU (1996) USAID Food Security Team Report (draft). Port au Prince, Haiti:
USAID.
Lundahl M (1997) Income and land distribution in Haiti: some remarks on available
statistics. Journal of Inter-American Studies and World Affairs 38: 109-126.
MARNDR (1990) Gid plantasyon mayi [Maize planting guide]. Port au Prince, Haiti:
Ministry of Agriculture (MARNDR), Pwojé Entansifikasyon Kilit
Vivriye/P.l.C.V.
McLain RJ and Stienbarger DM (1988) Land tenure and land use in southern Haiti: case
studies of the Les Anglais and Grande Ravine du Sud Watersheds. LTC Research
Paper No. 95. Madison: University of Wisconsin-Madison.

226
McNamara KT, Wetzstein ME and Douce GK (1991) Factors affecting peanut producer
adoption of integrated pest management. Review of Agricultural Economics 13:
129-139.
Mercer DE and Miller RP (1998) Socioeconomic research in agroforestry: progress,
prospects, priorities. Agroforestry Systems 38: 177-193.
MetznerJK (1976) Lamtoronisasi: an experiment in soil conservation. Bulletin of
Indonesian Studies (Canberra, Australia) 2: 103-109.
Mintz SW (1962) Living fences in the Fond-des-Negres region, Haiti. Economic
Botany 16: 101-105.
Mugendi DN, Nair PKR, Mugwe JN, O'Neill MK and Woomer PL (1999) Alley
cropping of maize with calliandra and leucaena in the subhumid highlands of
Kenya. Part 1: soil-fertility changes and maize yield. Agroforestry Systems 46:
39-50.
Mureithi JG, Tayler RS and Thorpe W (1994) The effects of alley cropping with
Leucaena leucocephala and of different management practices on the
productivity of maize and soil chemical properties in lowland coastal Kenya.
Agroforestry Systems 27: 31 -51.
Murray GF (1979) Terraces, trees, and the Haitian peasant: an assessment of twenty-five
years of erosion control. Port au Prince, Haiti: USAID.
Murray GF (1980) Haitian peasant contour ridges: The evolution of indigenous erosion
control technology. HIID Development Discussion Paper No. 86. Cambridge,
Massachusetts: Harvard University.
Murray GF (1981) Peasant tree planting in Haiti: a social soundness analysis. Portau
Prince, Haiti: USAID.
Murray GF (1991) The tree gardens of Haiti: agroforestry among Caribbean peasants—
from extraction to domestication. In: Challinor D and Frondorf MH (eds) Johns
Hopkins University. Social forestry: communal and private management
strategies compared. Case studies presented at a conference, February 14, 1991,
Johns Hopkins University.
Murray GF (1997) A Haitian peasant tree chronicle: adaptive evolution and institutional
intrusion. In: Krishna A, Uphoff N and Esman MJ (eds) Reasons for hope:
instructive experiences in rural development, pp 241-254. West Hartford,
Connecticut: Kumarian Press.

227
Mutsotso D and Gicheru L (1995) Development/extension in alley farming. In: Kang B
T, Osiname, AO and Larbi A (eds) Ibadan, Nigeria, Alley Farming Network for
Tropical Africa. Alley farming research and development. Proceedings of the
International Conference on Alley Farming, 14-18 September 1992, Ibadan.
Nigeria, pp. 507-510.
Nadagouda VB, Manjappa K, Desai BK, Palled Y and Kalaghatagi SB (1994)
Management of Eucalyptus tree rows for sustained agroforestry system. Journal
of Maharashtra Agricultural Universities 19: 308-309
NairPKR (1993) An Introduction to Agro forestry. The Netherlands: Kluwer Academic
Publishers.
Nair PKR (1998) Directions in tropical agroforestry research: past, present, and future.
Agroforestry Systems 38: 223-245.
Napier TL (1991) Factors affecting acceptance and continued use of soil conservation
practices in developing societies: a diffusion perspective. Agriculture,
Ecosystems & Environment 36: 127-140.
Neef A, Haigis J and Heidhues F (1996) L'impact du pluralisme institutionnel et
juridique sur l'introduction de mesures anti-erosive-Le cas du Niger [Impact of
institutional and legal pluralism on the introduction of erosion control measures-
the case of Niger]. Buerkert B, Allison BE and von Oppen, M (eds.) Wind
erosion in West Africa: the problem and its control. Margraf Verlag.
Proceedings, University Hohenheim, Germany, 5-7 December 1994., pp. 283-
296.
Neef A and Heidhues F (1994) The role of land tenure in agroforestry: lessons from
Benin. Agro forestry Systems 27: 145-161.
Nelson RA, Cramb RA, Menz KM and Mamicpic MA (1998) Cost-benefit analysis of
alternative forms of hedgerow intercropping in the Philippine uplands.
Agro forestry Systems 39: 241-262.
Nelson RA, Grist PG, Menz KM, Cramb RA, Paningbatan EP and Mamicpic MA (1997)
A cost-benefit analysis of hedgerow intercropping in the Philippine uplands using
the SCUAF model. Agroforestry Systems 35: 203-220.
Norris PE and Batie SS (1987) Virginia farmers' soil conservation decisions: an
application of Tobit analysis. Southern Journal of Agricultural Economics 19:
79-90.
Nuberg IK and Evans DG (1993) Alley cropping and analog forests for soil
conservation in the dry uplands of Sri Lanka. Agroforestry Systems 24: 247-269.

Nye PH and Greenland DJ. (1960) The Soil under Shifting Cultivation. Technical
Communication No. 51. Harpenden, UK: Commonwealth Bureau of Soils.
228
Ojyewotu LOZ, Ogigirigi MA and Stigter CJ (1994) A study of competitive effects
between a Eucalyptus camaldulensis shelterbelt and an adjacent millet
(Pennisetum typhoides) crop. Agriculture, Ecosystems & Environment 51:
281-286.
Ong CK, Corlett JE, Singh RP and Black CR (1991) Above and below ground
interactions in agroforestry systems, Forest Ecology and Management 45: 45-57
Ong CK and Leakey RRB (1999) Why tree-crop interactions in agro forestry appear at
odds with tree-grass interactions in tropical savannahs. Agro forestry Systems 45:
109-129.
Ongprasert S and Turkelboom F (1995) 20 years of alley cropping research and
extension in the slopes of northern Thailand. In: Turkelboom F, van Look-
Rothschild K, and van Keer K (eds) Highland farming: soil and the future?
Proceedings of a conference, Chaing Mai, Thailand, 21-22 December 1995.
OttRL (1992) An Introduction to Statistical Methods and Data Analysis. Belmont,
California: Duxbury Press.
PADF (1986) Pwoje Pyebwa 1986 Annual Report. Agroforestry Outreach Project.
Report for U.S. Agency for International Development (USAID). Port au Prince,
Haiti: PADF (Pan American Development Foundation).
PADF (1990) Pwoje Pyebwa 1990 Annual Report. Agroforestry II Project. Report for
U.S. Agency for International Development (USAID). Port au Prince, Haiti:
PADF (Pan American Development Foundation).
PADF (1999) PLUS Project 1999 Annual Report. Report for U.S. Agency for
International Development (USAID). Port au Prince, Haiti: PADF (Pan
American Development Foundation).
Paningbatan EP, Ceisiolka CA, Coughlan KJ and Rose CW (1995) Alley cropping for
managing soil erosion of hilly lands in the Philippines. Soil Technology 8:
193-204.
Pierre F, Lea JD and St-Dic R (1995) Gestión et impacts des haies vives PADF/Camp
Perrin. SECID/Aubum PLUS Report No. 24. Port au Prince, Haiti.
Rane PV, Sagare BN and Rewatkar SS (1995) Available soil moisture storage and
nutrient uptake by cotton as influenced by vegetative barriers. Annals of Plant
Physiology 9: 139-141.

229
Rao MR, Nair PKR and Ong CK (1998) Biophysical interactions in tropical
agroforestry systems. Agroforestry Systems 38: 3-50.
Rasmussen SD and Shapiro CA (1989) Crop responses to root plowing along multi-row
broadleaf windbreak borders. Forty First Annual Meeting, Great Plains
Agricultural Council Forestry Committee, June 12-15, 1989, pp. 123-125. Great
Plains Agricultural Council.
Regis G (1990) Resultáis de deux ans de recherches sur les pertes de sols effectuees au
Centre de Formation en Amenagement de Limbe. Groupe de Recherches,
d'Etudes et de Formation en Environment, Secretariat Technique á
1'Amenagement des Bassens Versants (STABV/MARNDR, USAID). Actes du
Colloque sur les Haies Vives pour une Agriculture Continue et Soutenue.
Proceedings of a conference held at Cap Haitien, 12-15 December, 1989, Port au
Prince, Haiti, pp. 63-76.
Rogers EM (1995) Diffusion of Innovations. New York: The Free Press.
Romanoff S, Voth D and Douglas M (1995) Further assessment and refinement of the
PLUS M&E system. SECID/Aubum PLUS Report No. 25. Port au Prince, Haiti.
Roose E and Ndayizigiye F (1993) Agroforestry, water and soil fertility management to
fight erosion in tropical mountains of Rwanda. Soil Technology 11: 109-119.
Salazari A, Szott LT and Palm CA (1993) Crop-tree interactions in alley cropping
systems on alluvial soils of the Upper Amazon Basin. Agroforestry Systems 22:
67-82.
Sanchez P A (1995) Science in agroforestry. Agroforestry Systems 30: 5-55.
Scherr SJ (1994) Agroforestry in the East African highlands: a sustainable strategy for
land-use intensification. Research & Exploration 10: 145-157.
Scherr SJ (1995) Economic factors in farmer adoption of agroforestry: patterns
observed in western Kenya. World Development (Oxford) 23: 787-804.
Scherr SJ and Current D (1997) What makes agroforestry profitable to farmers?
Evidence from Central America and the Caribbean. Agroforestry Today 9: 10-15.
Schroth G, Baile P and Peltier R (1995a), Alley cropping groundnut with Gliricidia
sepium in Cote d'Ivoire: effects on yields, microclimate and crop diseases.
Agroforestry Systems 29: 147-163.
Schroth G and Lehmann J (1995) Contrasting effects of roots and mulch from three
agroforestry tree species on yields of alley cropped maize. Agriculture,
Ecosystems & Environment 54: 89-101.

230
Schroth G, Poidy N, Morshauser T and Zech W (1995b) Effects of different methods of
soil tillage and biomass application on crop yields and soil properties in
agroforestry with high tree competition. Agriculture, Ecosystems & Environment
52: 129-140.
Schroth G and Zech W (1993) Root length dynamics in agroforestry with Gliricidia
sepium as compared to sole cropping in the semi-deciduous rainforest zone of
West Africa. Plant and Soil 170: 297-306.
Shannon DA and Isaac L (1993) Evaluation of tree species adaptation for alley cropping
in four environments in Haiti. A. Establishment phase. SECID/Aubum. PLUS
Report No. 6. Port au Prince, Haiti.
ShawRH (1988) Climate Requirement. In: Sprague GF and Dudley JW (eds) Com and
Com Improvement. Madison, Wisconsin: ASA/CSSA/SSSA Publishers.
Shively GE (1998) Impact of contour hedgerows on upland maize yields in the
Philippines. Agroforestry Systems 39: 59-71.
Shultz S, Faustino J and Melgar D (1997) Agroforestry and soil conservation: adoption
and profitability in El Salvador. Agroforestry Today 9: 16-17.
SiederP (1994) Haiti-traum und wirklichkeit. Ein tragisches beispiel aus der
forstlichen enwicklungshilfe [Haiti-dream and reality. A tragic example of
forestry development aid]. Forst und Holz 49: 40-46.
Sinclair FL (1999) A general classification of agroforestry practice. Agroforestry
Systems 46: 161-180.
Singh RP, Van den Beldt RJ and Hocking D (1986) Korwar GR. Alley farming in the
semi-arid regions of India. In: Kang BT and Reynolds L (eds) Alley farming in
the humid and sub-humid tropics. Proceedings of an international workshop held
in Ibadan, Nigeria, IDRC.
Singh V, Rajbahadur, Hooda MS and Balkrishan (1996) Pattern of root distribution in
30-month old five tree and two shrub species of an arid region of northwestern
India. Indian Forester 122: 323-330.
Singhai RM and Kumar V (1997) An analysis of some socio-economic implications
influencing people's participation in silvi-pastoral systems of a micro watershed
in Garhwal Himalaya. Indian Forester 123: 136-142.
Sitompul SM, Syekhfani MS and Van der Heide J (1992) Yield of maize and soybean in
a hedgerow intercropping system. Agrivita 15: 69-75.

231
Smucker GR (1988) Decisions and motivations in peasant tree farming: Mome Franck
and the PADF cycle of village studies. Port au Prince, Haiti: PADF/Haiti
Agroforestry Outreach Project.
Smucker GR and Dathis N (1998) Organisations Paysannes en Haiti. Tendences et
Implications. Port au Prince, Haiti: Editions La Ruche, MES Impressions et
Publicité.
Smucker GR and Thomson J (1999) Social capital and development in Haiti. Report to
ADF-Asosye. Port au Prince, Haiti: ARD-PACTE, ASSET.
Smucker GR and Timyan J (1995) Impact of tree planting in Haiti: 1982-1995.
SECID/Aubum PLUS Report No. 23. Port au Prince, Haiti.
Smucker GR, White TA and Bannister ME (2000) Land tenure and the adoption of
agricultural technology in Haiti. International Food Policy Research Institute
(IFPRI).
Solera CR (1993) Determinants of competition between hedgerow and alley species in a
contour hedgerow intercropping system. Doctoral dissertation, University of the
Philippines at Los Baños.
Ssekabembe CK, Henderlong PR and Larson M (1994) Soil moisture relations at the
tree/crop interface in black locust alleys. Agroforestry Systems 25: 135-140.
Stark M (1996) Adoption and adaption of contour hedgerow farming in the Philippine
uplands: results of an early case study. Tropenlandwirt 97: 3-16.
Steiner KG and Scheidegger U (1994) Participatory technology development for soil
fertility management in tropical highlands, Montpellier, France. In: Recherches-
systeme en agriculture et developpement rural: Symposium international, 21-25
Novembre 1994, pp. 310-315.
Sumberg JE and Atta-Krah AN (1988) The potential of alley farming in humid West
Africa-a re-evaluation. Agroforestry Systems 6: 163-168.
Sunderlin WD (1997) An ex-post methodology for measuring poor people's
participation in social forestry: an example from Java, Indonesia. Agroforestry
Systems 37: 297-310.
Sureshwaran S, Londhe SR and Frazier P (1996) A logit model for evaluating farmer
participation in soil conservation programs: sloping agricultural land technology
on upland farms in the Philippines. Journal of Sustainable Agriculture 7: 57-69.
Swanson RA (1993) Farmer needs assessment exploratory surveys. Field information
acquisition guide and methodology. SECID/Aubum PLUS Report No. 8. Port au
Prince, Haiti.

232
Swanson RA, Gustave W, Jean Y and Conde G (1993a) Farmer needs assessment
exploratory surveys. PADF Cap Haitien Region 3. SECID/Auburn PLUS Report
No. 9. Port au Prince, Haiti.
Swanson RA, Gustave W, Jean Y, Saint-Die R (1993b) Farmer needs assessment
exploratory surveys. CARE Northwest Regions 2, 3, & 4. SECID/Auburn PLUS
Report No. 10. Port au Prince, Haiti.
Swanson RA, Gustave W, Jean Y, Saint-Die R (1993c) Farmer needs assessment
exploratory surveys. PADF Jacmel Region 2. SECID/Auburn PLUS Report No.
11. Port au Prince, Haiti.
Swanson RA, Gustave W, Jean Y, Saint-Die R (1993d) Farmer needs assessment
exploratory surveys. PADF Les Cayes Region 1. SECID/Auburn PLUS Report
No. 13. Port au Prince, Haiti.
Swanson RA, Gustave W, Jean Y, and Saint-Die R (1993e) Farmer needs assessment
exploratory surveys. PADF Mirebalais Region 4. SECID/Auburn PLUS Report
No. 12. Port au Prince, Haiti.
Tacio HD (1993) Sloping agricultural land technology (SALT): a sustainable
agro forestry scheme for the uplands. Agro forestry Systems 22: 145-152.
Tassin J, Perret S and Cattet R (1995) Alteration in the porosity of an andisol in
Reunion under a hedge of Calliandra calothyrsus. Bois et Forets des Trapiques
245: 91-99.
Tilander Y, Ouedraogo G and Yougma F (1995) Impact of tree coppicing on tree-crop
competition in parkland and alley farming systems in semiarid Burkina Faso.
Agroforestry Systems 30: 363-378.
Tonye J, Duguma B and Tiki-Manga T (1994) Stepwise approach to alley cropping
technology development and transfer in the forest zone of Cameroon.
Agroforestry Systems 28: 269-278.
Tonye J and Titi-Nwel P (1995) Agronomic and economic evaluation of methods of
establishing alley cropping under a maize/groundnut intercrop system.
Agriculture, Ecosystems & Environment 56: 29-36.
Toussaint RJ (1989) Les haies vives arbustives: etablissement, gestión et production-
contraintes et limitations techniques dans Taire du Projet LORD 1 a Maissade.
Port au Prince, Haiti, Groupe de Recherches, d'Etudes et de Formation en
Environment, Secretariat Technique á TAmenagement des Bassens Versants
(STABV/MARNDR, USAID). Actes du Colloque sur les Haies Vives pour une
Agriculture Continue et Soutenue. Proceedings of a conference held at Cap
Haitien, 12-15 December, 1989, pp. 101-113.

233
Treadwell BD and Cunard AC (1992) Perennial herbaceous leguminous plants as
permanent contour land improvements for Haitian hillside farms: results &
observations from a 2-year field study and current recommendations, pp. 14-24.
Port au Prince, Haiti, Associates in Rural Development. Pwoje Sove Te Livestock
Working Document Series.
Tripathi BR and Psychas P (1992) The AFNETA alley farming training manual.
Volume 1. Core course in alley farming. Addis Ababa: ILCA.
United Nations Development Programme (1996) PNUD Haiti-Econet [newsletter]
January 1996, vol. 2, no. 1.
USAID/Haiti (1996) USAID Haiti Monitoring Report, Vol. 4, No. 6. Port au Prince,
Haiti: USAID/Haiti.
VandermeerJ (1998) Maximizing crop yield in alley crops. Agroforestry Systems 40:
199-206.
Vaval F, Soto Pinto L, Nelson K and Serrano Gálvez E (1997) La percepción de los
productores sobre el impacto de la practica agroforestal "barrera viva" en
Maissade, Plateau Central, Haiti.
Villanueva MR (1993) A review of PDAI and ADS1I project technologies.
SECID/Aubum PLUS Report No. 2. Port au Prince, Haiti.
Weaver J (1920) Root development in the grassland formation. A correlation of the
root systems of native vegetation and crop plants. Washington, DC: Carnegie
Institution of Washington.
Weil TE, Black JK, Blutstein HI, Johnston KT, McMorris DS and Munson FP (1973)
Area Handbook for Haiti. Washington, DC: U.S. Government Printing Office.
Welke SE, Tanyi E and Kang BT (1995) Effect of Leucaena leucocephala (Lam.) de
Wit pruning frequency on alley cropped maize/cassava. In: Kang BT, Osiname
AO and Larbi A (eds) Alley Farming Network for Tropical Africa. Alley farming
research and development. Proceedings of the International Conference on Alley
Farming, 14-18 September 1992, Ibadan, Nigeria, pp. 211-225.
White TA (1992a) Peasant cooperation for watershed management in Maissade, Haiti:
factors associated with participation. EPAT/MUCIA Working Paper No. 4.
White TA (1992b) Peasant initiative for soil conservation: case studies of recent
technical and social innovations from Maissade, Haiti. EPAT/MUCIA Working
Paper No. 3.

234
White TA and Quinn RM (1992) An economic analysis of the Maissade. Haiti.
Integrated Watershed Management Project. [Working Paper No. 2, July 1992], -
24. 1992. EPAT/MUCIA.
Wiersum KF (1994) Farmer adoption of contour hedgerow intercropping, a case study
from east Indonesia. Agroforestry Systems 27: 163-182.
Wirjodarmodjo H and Wiroatodjo P (1983) Leucaena leucocephala: the Indonesian
experience. Bangkok: FAO Regional Office for Asia and the Pacific.
Yadav JP, Sharma KK and Khanna P (1993) The effect of Acacia nilotica on mustard
crop. Agroforestry Systems 21: 91-98.
Young A and Muraya P (1990) Soil Changes Under Agroforestry. Computer program
with User's Handbook. Version 2. Nairobi, Kenya: 1CRAF.
ZarJH (1984) Biostatistical Analysis. Englewood Cliffs, New Jersey: Prentice-Hall,
Inc.
Zoschke A, Ebner L and Quadranti M (1990) Conservation farming in Malaysia-first
experiences with alley cropping. Malaysian Plant Protection Society 6: 272-283.
Zuvekas C (1979) Land tenure in Haiti and its policy implications: a summary of the
literature. Social and Economic Studies 28: 1-30.

BIOGRAPHICAL SKETCH
After serving four years in the U.S. Navy, Michael Bannister earned a B.S. and
Master of Forestry (M.F.) in forest management from Oregon State University in 1976
and 1981, respectively. Between the two degrees he served as a Peace Corps volunteer in
Guatemala, establishing two community pine/oak nurseries and promoting soil
conservation. In 1981 he accepted a position working in agroforestry extension in Haiti
with the Pan American Development Foundation (PADF). PADF has implemented three
USAID-funded agroforestry projects in Haiti from 1981 through 2000. Mr. Bannister
worked on all three projects as regional team leader, research and documentation
coordinator, and assistant director for agroforestry; the latter position he still holds.
Since 1981 he has also done short consultancies for PADF in the Dominican Republic,
Honduras, El Salvador, and Panama. Mr. Bannister began his Ph.D. at the University of
Florida in 1987, taking three years off from his professional work for required classes.
He then returned to Haiti as a full-time employee of PADF while simultaneously doing
his field research, analysis, and much of the writing. This process, aggravated by
political and civil turmoil in Haiti, has taken 14 years.
235

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy
P. K. Ramachandran Nair, Chair
Professor of Forest Resources and
Conservation
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy
Kenneth L. Buhr
Assistant Professor of Agronomy
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy /
-u9u
Mai
Profes'
L. Duryea
f Forest Resources and
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fullyatiequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy
L Hildebrand
Professor of Food and Resource
Economics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy ,
Serald Murray
Associate Professor of Anth/opology

This dissertation was submitted to the Graduate Faculty of the School of Forest
Resources and Conservation in the College of Agricultural and Life Sciences and to the
Graduate School and was accepted as partial fulfillment of the requirements for the
degree of Doctor of Philosophy.
May 2001
Conservation
Dean, Graduate School

07/30/0]




PAGE 1

DYNAMICS OF FARMER ADOPTION, ADAPTATION. AND MANAGEMENT OF SOIL CONSERVATION HEDGEROWS IN HAITI By MICHAEL E. BANNISTER A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2001

PAGE 2

UNWERS.TY OF aOgOA *S 1262 08666 372 .b^\^

PAGE 3

ACKNOWLEDGMENTS I wish to thank my graduate advisor. Dr. P.K.R. Nair, for his unflagging encouragement and guidance throughout a long period of research and writing and Dr. Peter Hildebrand for providing my Farming Systems assistantship from 1 987 through 1989. My other committee members, Dr. Mary Duryea, Dr. Jerry Murray, and Dr. Ken Buhr also provided valuable insights and assistance. The untiring work of the Pan American Development Foundation (PADF) Haitian technical staff during data collection and the encouragement and financial support of the PADF home office during my sabbatical were essential to this dissertation, as was the cooperation of a multitude of Haitian farmers. Deserving of special mention are PADF Haiti Country Representative Lee Nelson for his key support of my sabbatical and Agronomist Jn-Charlot Bredy for data collection on farms near Lascahobas. Data collection for the Leucaena/maize study could not have been completed without the assistance of technicians Duverger Vemis and Rodini St-Juste. Finally. 1 thank my beloved wife Mojdeh for her essential care of my sometimes-fragile psyche during this trying time. 11

PAGE 4

TABLE OF CONTENTS pages LIST OF TABLES v LIST OF FIGURES x ABSTRACT xiii CHAPTERS 1 INTRODUCTION 1 2 HILLSIDE FARMING SYSTEM IN HAITI 4 Land 5 Crops 7 Labor 9 Farmer Development Groups and Local NGOs 10 3 THE HEDGEROW INTERCROPPING SYSTEM 13 Developments in the Tropics 13 Developments of Hedgerow Intercropping in Haiti 18 Hedgerow Intercropping and Soil Conservation 22 Conclusions 26 4 ORGANIZATION OF THE STUDY 27 Justification 27 Hypotheses 33 General Description of Hypothesis-Testing Procedures 34 5 HEDGEROW/CROP COMPETITION 36 Introduction 36 On-Station Studies: Effect of Root Barriers on the Growth of Maize and Leucaena 39 On-Farm Study: Maize Growth at Varying Distances from Hedgerows 100 111

PAGE 5

6 LAND AND HOUSEHOLD CHARACTERISTICS IN RELATION TO ADOPTION AND MANAGEMENT OF HEDGEROWS 108 Introduction and Objectives 1 08 Materials and Methods 119 Results and Discussion 127 Summar>' and General Discussion 166 Conclusions 1 75 7 SYNTHESIS AND CONCLUSIONS 179 APPENDICES A POT STUDY OF SOIL WATER DEPLETION VS. MAIZE LEAF WATER POTENTIAL 186 B ROOT DISTRIBUTION OF A LEUCAENA LEUCOCEPHALA HEDGEROW 188 C P-VALUES OF ANOVAS PERFORMED ON STATION 191 D HOUSEHOLD QUESTIONNAIRE 198 E GARDEN PLOT QUESTIONNAIRE 201 F SOIL NUTRIENTS, ORGANIC CARBON STATUS, AND pH OF 175 HILLSIDE GARDENS 211 REFERENCES 218 BIOGRAPHICAL SKETCH 235 IV

PAGE 6

LIST OF TABLES Table page 5-1 : Soil texture and organic carbon at the trial site 43 5-2: Effect of trenching on the production of maize biomass 60 5-3: Effect of fertilizer on the production of maize biomass 61 5-4: Soil water at three depths below the surface 62 5-5: Maize grain weight, number of plants, and number of ears 65 5-6: Stem and leaf biomass harvests and daily growth increments of Leucaena hedgerows 66 5-7: Height above a 50 cm stump and daily growth increments of Leucaena hedgerows prior to the second root trenching. 33 and 65 DAS; Spring 1994 T 67 5-8: Maize biomass in trenched and nontrenched plots 69 5-9: Maize leaf water potential in trenched and nontrenched plots over four distances from the hedgerows. 42 DAS; fall 1994 71 5-10: Soil water in trenched and nontrenched plots by depth and distance 73 5-11: Maize grain weight, number of plants, and number of ears at four distances from the hedgerows and as a total per hectare, with p-values for the differences between distances. Fall 1994 75 5-12: Stem and leaf biomass harvests and daily growth increments of Leucaena hedgerows after the second root trenching. Fall 1994 76 5-13: Height above a 50 cm stump and daily growth increments of Leucaena hedgerows after the second root trenching, Fall 1 994 77 5-14: Maize height in trenched and nontrenched plots at four distances from hedgerows. 40 DAS; Spring 1995 79

PAGE 7

5-15: Maize biomass in trenched and nontrenched plots at four distances from the hedgerow. 47 and 54 DAS; Spring 1995 82 5-16: Soil water in trenched and nontrenched plots by depth and distance from the hedgerow, 26 and 61 DAS; Spring 1995 86 5-17: Maize grain weight, number of plants, and number of ears 89 5-18: Stem and leaf biomass harvests and daily growlh increments of Leucaena hedgerows 90 5-19: Height above a 50 cm stump and daily growth increments of Leucaena hedgerows 90 5-20: Maize yield in three soil conservation practices and control at 1 16 DAS on four farms, spring 1 997 105 61 : Age distribution by gender of members of 1 ,540 farm households interviewed in the four PADF/PLUS project areas in Haiti, 1996 128 6-2: Number of years of school by gender in seven age classes of 1,540 households participating in the PADF/PLUS project, 1996 129 6-3: Percent of female and male members of 1,540 project households who participated in various economic activities during 1995-1996 130 6-4: Percent of female and male members of 1 ,540 project households who participated in various economic activities during 1995-1996, by age class 131 6-5 : Percent of female and male members of 1 .540 households who participated in PLUS project activities during 1995-1996 132 6-6: Percent of female and male members of 1 .540 project households who participated in PLUS project activities during 1995-1996. by age class 133 6-7: Average total area of three categories of plots held by 1 ,540 project households based on the relative distance from the plots to the residence .... 134 6-8: Mean plot area of three categories of plots held by 1.540 project households during 1995. based on the relative distance from the plots to the residence 134 6-9: Mean plot area of three categories of plots held by 1,540 project households during 1995 in four PLUS project regions, based on the relative distance from the plots to the residence 135 VI

PAGE 8

6-10: Percent of three categories of plots held by 1.540 project households during 1995 under various modes of access 136 6-11: Area of 5,660 plots held by L540 households during 1995, by mode of access 137 6-12: Percent of plots held by 1.540 households during 1995 in eight mode of access categories by slope position 137 6-13: Percent of all plots held by 1,540 households during 1995 in eight modes of access categories by slope steepness 138 6-14: Distance from residence of plots having project agro forestry practices held by 1 ,540 households during 1 995 in the five most common mode of access categories 139 6-15: Area of plots having project agro forestry practices held by 1,540 households during 1995 in the five most common mode of access categories 140 6-16: Elevation of plots having project agro forestry practices held by 1,540 households during 1995 in seven mode of access categories 140 6-17: Slope of plots having project agroforestry practices held by 1,540 households during 1995 in seven mode of access categories 141 6-18: Percent of 2,295 surveyed plots with and without three kinds of agroforestry practice held under secure (purchased or inherited/divided) or not secure (all other) tenure 147 6-19: Percent of 2.295 surveyed plots with and without three kinds of agroforestry practice held under secure (purchased or inherited/divided) or not secure (all other) tenure 147 6-20: Percent of 2.295 surveyed plots with and without three kinds of agroforestry practice having fertile or not fertile soil 147 6-21 : Percent of 2,295 surveyed plots with and without three kinds of agroforestry practice having fertile or not fertile soil 148 6-22: Percent of 2,295 surveyed plots with and without three kinds of hedgerows held under secure (purchased or inherited/divided) or not secure (all other) tenure 149 6-23: Percent of 2.295 surveyed plots with and without three kinds of hedgerows having fertile or not fertile soil 149 Vll

PAGE 9

6-24: Number of all trees larger than 10 cm diameter on PLUS plots, by plot tenure 150 6-26: Soil fertility and hedgerow management quality 154 6-27: PADF/PLUS field region and hedgerow management quality 156 6-28: Soil fertility and crop band management quality 157 6-29: PADF/PLUS field region and crop band management quality 158 6-30: Soil fertility and rock wall management quality 159 6-3 1 : PADF/PLUS field region and rock wall management quality 1 60 6-32: Percent of 1,362 plots where potential benefits of hedgerows (fuel, charcoal, construction wood, fodder, soil fertility, soil conservation, and crop production) were classified according to importance by farmers on a five-point scale, l=not important, 5=very important 161 6-33: Percent of hedgerows classified as well-managed by technicians in 1,362 plots where potential benefits of hedgerows (fuel, charcoal, construction wood, fodder, soil fertility, soil conservation, and crop production) were classified according to importance by farmers on a five-point scale. l=not important, 5=very important 1 62 6-34: Percent of hedgerows classified as poorly-managed by technicians in 1,362 plots where potential benefits of hedgerows (fuel, charcoal, construction wood, fodder, soil fertility, soil conservation, and crop production) were classified according to importance by farmers on a fivepoint scale, l=not important, 5=very important 163 6-35: Number of breaches per 100 m of hedgerow counted by technicians in 1,362 plots where potential benefits of hedgerows (fuel, charcoal, construction wood, fodder, soil fertility, soil conservation, and crop production) were classified according to importance by farmers on a fivepoint scale, l=not important. 5=very important 164 6-36: Percent of gardens where hedgerow breaches were repaired or not repaired for three potential benefits of hedgerows 165 6-37: Percent of 1.362 plots where potential problems of hedgerows (shade, reduced space, water competition, reduced ability to picket animals, weediness, and labor cost) were classified according to importance by farmers on a five-point scale, l=not important, 5=very important 165 vin

PAGE 10

6-38: Percent of gardens where hedgerow breaches were repaired or not repaired for three potential problems of hedgerows 166 6-39; Estimated relative costs, benefits, and risk of agro forestry' practices 171 IX

PAGE 11

LIST OF FIGURES Figure page 51 : Layout of the Leucaena/maize hedgerow trial on Operation Double Harvest property near Croix des Bouquets. Haiti 45 5-2: Diagram of one plot showing relative positions of a Leucaena hedgerow and the adjacent rows of maize at four distances 46 5-3: Rainfall, Leucaena and maize growth, and agronomic inputs; spring 1994 maize cropping season 51 5-4: Rainfall, Leucaena and maize growth, and agronomic inputs; fall 1994 maize cropping season 51 5-5: Rainfall. Leucaena and maize growth, and agronomic inputs; spring 1995 maize cropping season 52 5-6: Maize biomass yields at 50, 100, 150, and 200 cm from the hedgerows at four times after sowing during the spring 1994 season 59 5-7: Growth stage of maize at four distances from hedgerows 61 5-8: Yield of maize grain per six meters at four distances from the hedgerows 64 5-9: Number of maize plants per eight hills at four distances from the hedgerows 64 5-10: Number of maize ears per eight hills at four distances from the hedgerows .... 65 5-11: Trenched and nontrenched plots at 50, 100, 150. and 200 cm from the hedgerows. 78 DAS; fall 1994 69 5-12: Maize leaf water potential in trenched and nontrenched plots 70 5-13: Soil water in trenched plots (and difference between T+ and Tplots) at 42 DAS, fall 1994 72 5-14: Maize grain yield in trenched (T+) and nontrenched (T-) plots 74

PAGE 12

5-15: Maize height trenched and nontrenched at four distances from the hedgerow, 40 DAS; Spring 1995 79 5-16: Maize biomass in trenched and nontrenched plots at four distances from the hedgerow, 47 DAS, Spring 1995 81 5-17: Maize biomass in trenched and nontrenched maize weight at four distances from the hedgerow, 54 DAS, Spring 81 5-18: Trenched and nontrenched maize growth stage at four distances from the hedgerow, 68 DAS; Spring 1995 83 5-19: Trenched and nontrenched maize growth stage at four distances from the hedgerow, 75 DAS; Spring 1995 83 5-20: Soil water percent in trenched plots (and difference between T+ and Tplots) at 26 DAS, Spring 1995 85 5-21 : Soil water percent in trenched plots (and difference between T+ and Tplots) at 61 DAS, Spring 1995 85 5-22: Maize grain weight from trenched and nontrenched plots at four distances from the hedgerows, 131 DAS; Spring 1995 87 5-23: Number of maize plants from trenched and nontrenched plots at four distances from the hedgerows, 131 DAS; Spring 1995 88 5-24: Number of ears from trenched and nontrenched plots at four distances from the hedgerows, 131 DAS; Spring 1995 88 5-25: Number of very fine root (<2mm) intersections with a 10 by 100 cm plane parallel to the trenched hedgerows at five distances and 10 depths, with the difference between the trenched and nontrenched hedgerows shown in parentheses 92 5-26: Number of fine root (2-5 mm) intersections with a 10 by 100 cm plane parallel to the trenched hedgerows at five distances and 10 depths, with the difference between the trenched and nontrenched hedgerows shown in parentheses 92 5-27: Number of medium root (5-10 mm) intersections with a 10 by 100 cm plane parallel to the trenched hedgerows at five distances and 10 depths, with the difference between the trenched and nontrenched hedgerows shown in parentheses 93 XI

PAGE 13

5-28: Number of large root (>10 mm) intersections with a 10 by 100 cm plane parallel to the trenched hedgerows at five distances and 1 depths, with the difference between the trenched and nontrenched hedgerows shown in parentheses 93 5-29: Leucaena small stem and leaf biomass from trenched and nontrenched plots. May 1991 through September 1995 98 5-30: Maize development in three soil conservation practices and control at four times during the growing season on four farms during Spring 1 997 1 04 6-1 : Relative importance of four plot characteristics to the decision to install six agro forestry practices 168 xn

PAGE 14

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy DYNAMICS OF FARMER ADOPTION, ADAPTATION. AND MANAGEMENT OF SOIL CONSERVATION HEDGEROWS IN HAITI By Michael Bannister May 2001 Chair: P. K. Ramachandran Nair Major Department: School of Forest Resources and Conservation Understanding the conditions under which hillside farmers in Haiti adopt soil conservation practices helps programs to develop technologies that increase farmer revenue and stabilize or improve soil and water resources. Three studies were done in Haiti that examined biophysical and socioeconomic aspects of contour hedgerows, an agro forestry practice with the potential of stabilizing soil, conserving water for crops, increasing soil fertility, and producing wood and fodder. An on-station study of soil water competition between Leucaena leucocephala (Lan.) de Wit hedgerows and adjacent maize {Zea mays L.) over three cropping cycles found that substantial maize yield reduction was caused by the hedgerow trees because of soil water competition. Polyethylene barriers between the hedges and the first row of maize improved maize yield in two subsequent seasons by 18% and 77% over plots without barriers. The percent increase in maize yield was highest during a season of very poor rainfall; but it was lowest in a season of adequate rainfall. Barrier installation xiii

PAGE 15

reduced Leucaena biomass production by about 1,500 kg/ha over seven months, but this effect was temporary. An examination of the distribution oi Leucaena roots after the final maize harvest showed that the hedgerow roots in the plots with barriers had grown under the hedgerows and developed more fine roots at 200 cm from the hedges than plots without barriers. An on-farm study comparing maize development in two kinds of hedgerows, rock wall terraces, and an untreated control was unable to detect differences in the rate of maize growth with respect to position within the alleys or with respect to position on the overall slope. However, the maize in the hedgerows developed more slowly than in the rock wall terraces or the control plot, indicating competitive interactions between maize and hedgerow plants. There were no differences in maize grain yield among treatments. An on-farm questionnaire-based survey of 1 ,540 Haitian farmers showed that they considered plot characteristics, including mode of tenure, soil fertility, distance from the residence, and slope in their decisions to install different agroforestry practices. Tenure security and soil fertility appeared to be the most important plot characteristics in the decision to install, although it was not possible to separate them. Hedgerows were more likely to be installed in plots having less secure tenure and less fertile soil; the opposite was true for other agroforestry practices. Farmers' qualitative assessments of soil fertility were positively correlated with management quality of hedgerows. XIV

PAGE 16

CHAPTER 1 INTRODUCTION The research presented here focuses on agroforestr\' technologies, which could contribute to increasing the revenue of small farms on the hillsides of Haiti in a sustainable, environmentally friendly way. This is a formidable goal, because Haiti is primarily a mountainous country long since shorn of its natural vegetative cover. Its once-rich soils are badly eroded, and their productive capacity severely diminished. Haiti is densely populated by very poor people who generally lack formal education or access to a functioning legal system. It has a long history of political instability that continues to this day. At best, agriculture is difficult in such a setting. But, Haitian farmers regularly experiment with new techniques. This dissertation investigates the biological and adoption aspects of hedgerow intercropping, an agroforestry practice that involves growing food crops in the interspaces between rows of densely planted trees or shrubs. Two types of studies are included. One focuses on competition for growth resources between trees and crops while the other focuses on the factors that influence farmers' adoption of the technology. However, these different studies are linked together, because both social and biological aspects of agroforestry practices must be understood for providing efficient extension services to farmers. The dissertation is presented in seven chapters. Chapter 1 briefly explains the background of the research and states its objectives. Chapter 2 describes Haitian 1

PAGE 17

2 smallholder farms, the population of interest to the research. Following a definition of hedgerow intercropping, the agroforestr\' practice that is the subject of the research. Chapter 3 reviews its development in the tropics and in Haiti and examines how it is used as a soil conservation technique. Chapter 4 justifies the research, presents the hypotheses to be tested, and describes the studies used to test the hypotheses. The next chapter (Chapter 5) presents the two studies examining competition between hedgerows and adjacent crops. The first is done on station, the second on farm. The questionnaire-based survey of farmers who adopted soil conservation practices promoted by the PLUS project is the subject of Chapter 6. Finally. Chapter 7 synthesizes the previous chapters, and presents the overall conclusions of the dissertation. The research was done while the author was employed by the Pan American Development Foundation (PADF) in Haiti. The PADF implements a U.S. Agency for International Development (USAID) project called Productive Land Use Systems (PLUS). The PADF/PLUS has worked with about 170,000 hillside farmers in soil conservation, tree planting and grafting, has improved food crop distribution and marketing, and has been strengthening local farmer organizations between 1 992 and 1999. The on-farm study of maize {Zea mays L.) production between soil conservation structures discussed in Chapter 5 and the questionnaire-based study discussed in Chapter 6 were done as PLUS monitoring and evaluation exercises by PADF staff. The author designed the questionnaires and trained staff in their use. The overall objective of this research is to understand how characteristics of Haitian farm families, their plots, and agroforestry practices affect the adoption and management of those practices. The purpose is not to use the data gathered in the course of this research to define and predict adopters so that nonadopters can be excluded from

PAGE 18

3 extension efforts. It is rather an effort to understand how appropriate practices can evolve and be extended more rapidly. In particular, the objectives of the three studies that comprise this dissertation are • to determine, in the relatively controlled environment (for a research station), whether the competition for soil water between trees in hedgerows and adjacent maize is significant and can be reduced by the use of plastic root barriers, • to understand whether tree-crop competition in farmers' steeply sloping fields is similar to that observed in the flat controlled environment of the research station, and • to examine the factors (e.g., land tenure, perceived tree/crop competition, plot slope, soil fertility of the plot, distance from the residence to the plot, previous experience of the farmer, and the availability of other farm resources) that influence farmer adoption, adaptation, and management of hedgerow intercropping.

PAGE 19

CHAPTER 2 HILLSIDE FARMING SYSTEM IN HAITI Haiti is mainly hilly, with 20% of the land area below 185 m elevation and 40% above 450 m — about 75% highlands, 22% lowlands (Weil et al. 1973). The most common "life zones," according to Holdridge's (1945) classification, are subtropical moist forest and subtropical dry forest, but subtropical wet and rain forest zones are common in the middle and upper elevations. Limestone substrate underlies 80% of the land area; the rest is basaltic or alluvial (Ehrlich et al. 1985). According to Ehrlich et al. (1985) estimates, 1.3 million hectares (of 2.7 million total for the country) were under agricultural production; six times the area classified as "good agricultural land." The practice of annual agriculture on steeply sloping plots causes severe losses of arable soil due to erosion, estimated to be about 37 million tons annually in 1994 (UNDP, 1996). Haitians have been practicing agriculture on small, privately held plots in the hills since the 1840s. as the large plantations and public domain lands were divided (de Young 1958. Leybum 1941). Haitian peasants are not subsistence farmers in the strict sense that they do not live off the land self-sufficiently and independently, but rather produce for, and are linked closely to, the market (Murray 1981). Most farm families have some nonagricultural income and are net food buyers (Locher 1996). De Young (1958) describes the economic organization of Haitian farms as "horticultural" rather than agricultural, and gives the following characteristics that still hold true today: • They are market directed, not subsistence directed.

PAGE 20

• They give more importance to perennial rather than annual crops. • The farm units are small parcels, and the land is used intensively. • They are capital saving rather than labor saving. Farmers own only a fewhand tools, have little invested in storage or farm buildings, and rarely use plows and animal traction. • The farm animals are used mainly for transport to market. Reducing risk is as important to most farmers as increasing production, and to this end they manage a network of ties to relatives and other farmers as social capital that can be leveraged in order to gain access to scarce land, labor, and capital (Smucker et al. 2000). Haitian families also spread out risk by diversifying human and economic resources via selective migration, emigration, and off-farm economic enterprises (Locher 1996). Therefore, not only is any one particular crop not the sole interest of a farm family, but agriculture is not the only domain of economic activity. It is quite likely that other activities, such as nonagricultural marketing or remittances from relatives overseas, bring in more cash. Land Haitian hillside farms are small and complex. Farm families usually manage or cultivate several parcels of land at any given time, held under a mix of land tenure arrangements (e.g., owned, tenancy, usufruct) (Bloch et al. 1987, Locher 1996) and often on sites differing in their physical characteristics (hillside, plain, irrigated, rainfed) and in distance from the residence. This is done as a purposeful strategy to reduce risk of harvest loss and to spread out crop harvest over the year (McLain and Stienbarger 1988, Murray 1981, Zuvekas 1979). Plot sizes are small and fallow periods are short. Plots are worked and then abandoned when yield decreases below the farmer's economical limits. Declining soil fertility, soil loss due to erosion, large variation in the timing of rainfall.

PAGE 21

6 and the inability of traditional agricultural practices to harvest and store water in the soil limit agricultural production. Residences tend to be physically disbursed in the mountain regions instead of concentrated (Friedman 1955). Two intertwining land tenure systems operate in Haiti, one official and the other customary (Smucker et al. 2000). created by disenfranchised peasants to protect themselves against loss of land via a corrupt justice system and government abuse. The land market is active. Most farmers own most of their land, but the majority of the transactions are done without recourse to expensive official processes (Smucker et al. 2000, Zuvekas 1979). The customary system provides farmers relatively secure access to land, but some tenure classes are more stable over time than others so mixed tenure configurations that change over time are the norm (Locher 1996). Zuvekas (1979) reviewed data from 12 studies done in Haiti and found that the percentage of farmers owning (by purchase or inheritance) at least part of their land ranged from 58 to 100%. Secondary modes of access to land include rental from private owners, lease, and sharecropping. Usufruct rights to land are also obtained from relatives or by acting as caretaker for an absentee landowner Information about farm size is available from several studies related to agroforestry projects from various parts of the country (Lea 1994a. Lea 1996, Swanson et al. 1993a, 1993c, White 1992a). The total size of all farm plots ranges from less than one to about three hectares. The average farm family works two or three plots (Zuvekas 1979) that vary in size from about one half to just over one hectare each. Haitian laws of inheritance have French antecedents and require that all siblings inherit; therefore, land holdings are, in general, getting smaller. In 1950, 39% of households held less than 1.3 hectares; by 1971 this proportion had increased to 71%

PAGE 22

7 (Murray 1981). Brochet (1993) compared the average size of home gardens in one area of Haiti over fifty years, and found that the average holding that was one hectare per family in 1910 had decreased to 500 m(0.05 ha) by 1970. This study was done in an area mined for bauxite by Reynolds Aluminum, so sale of land to the mining company may have accelerated the rate of decrease, but in general average sizes of holdings grow smaller as the population increases. One strategy adopted by Haitian families to ensure that a plot is large enough to harvest a crop worth planting is to bequeath land to children undivided, but this may discourage investing in improvements to the plot if the siblings are contentious. Older farmers are likely to own a larger proportion of their holdings (Lundahl 1 997, McLain and Stienbarger 1 988). Other types of land tenure found in Haiti include purchased, inherited but separated, rented from private owners, and rented from the state. Usufruct rights to land also are obtained by sharecropping (on land owned by the sharecropper as well as on land owned by another) and by acting as caretaker for an absentee landowner. A recent survey of 5000 fields across the country (US AID. 1 996) showed the following pattern of land tenure: purchased 39%. inherited/divided 20%, inherited/undivided 14%. rented 12%, share cropped 5%, gift 1%, and other 9%. Crops Since small and scattered plots must provide food, fuel, building material, animal fodder, and other household necessities, agroforestry has traditionally been practiced. The farm plots, except the plot containing the residence and surrounding tree and medicinal crops (sometimes called the home garden), are spatially dominated by cereal grains and grain legumes (pulses), but income from roots, tubers, tree crops, and animals is probably more important economically, and is able to shield farmers against variations

PAGE 23

8 in rainfall better than grains and pulses. Home gardens include many tree crops such as mango {mangifera indica L.), breadfruit {Artocarpus altilis [Parkinson] Fosberg). citrus {Citrus spp.). royal palm {Roystonia borinquena), and coconut (Cocos nucifera L.). Timber and fuelwood trees are planted or tended where they sprout around property perimeters, scattered throughout plots, or as small woodlots. Trees are considered to be a crop, and are managed as a savings account, to be harvested when cash or building materials are needed (Murray 1981, Smucker and Timyan 1995). Fuelwood and animal fodder are in short supply in many parts of the country. Many different combinations of crops are grown. These vary with the season, land tenure, distance from the residence, market demand, availability of seed, and the rainfall and soil characteristics. The most common grain crops are maize {Zea mays L.) and sorghum {Sorghum bicolor (L.) Moench) on the hillsides, and rice {Oryza saliva L.) in the few irrigated plains. Pulses include beans {Phaseolus vulgaris L.), pigeon pea {Cajanus cajan (L.) Millsp.). and peanuts {Arachis hypogaea L.). It is common for the spring maize crop to fail, because of poor rainfall. Beans are able to generate important income for farmers, but also fail often— rain is insufficient, or rain falls during flowering (eliminating pod formation), or rain falls before seeds are dry causing them to germinate in the pods. Root and tuber crops include yams {Diascora spp.), cassava {Manihot esculenta Crantz), and cocoyam {Xanthosoma sagittifolium (L.) Schott). Plantains {Musa spp.) are commonly grown in ravines and near the home. Living fences surround the property where the house is located, and are commonly built of plants serving protective functions due to the presence of spines or poisonous sap, such as bayonette {Yucca aloifolia L.). pencilbrush {Euphorbia tirucalli L.), raquette {Euphorbia lactea Haw.), and penguin {Bromelia pinguin) (Ashley 1986).

PAGE 24

9 A 1996 US AID survey of 6,900 fields during a single spring planting season (USAID, 1996) found that 27% of the plots had crops growing in association (two or more crops present simultaneously in the same field for part or all of the cropping cycle), and 21% in monoculture. There were 16 main maize associations and 98 minor (<5000 ha. each) ones. In addition, there were 12 important associations of other crops. 134 minor ones, and 14 main kinds of monocrops. This does not include the many different kinds of tree and medicinal crops, crops found during the fall growing season, or animals. The most commonly seen grain crop in the fall season is sorghum, usually intercropped with pigeon pea. The USAID study notes only the physical presence of the crops, not their relative economic importance. Labor Households function as separate economic units, but the members participate in labor and food exchanges with other farmers (Murray 1981, Smucker and Dathis 1998). The family itself provides the labor for most agricultural activities. This labor pool is expanded periodically according to seasonal demands for land clearing, working the soil, planting, weeding, and harvesting crops. Labor is a scarce resource, and has been more difficult to supply and afford as the deteriorating social and economic situation in Haiti has forced many rural inhabitants into political refuge or into taking jobs in the neighboring Dominican Republic (Balzano 1997, Smucker and Dathis 1998). Farmers form groups to exchange farm labor, earn cash, or to save money to attain a future goal. However, this communal pooling is restricted to labor, not to the sharing of land or produce (Murray 1991). Access to labor is crucial because working steep slopes is labor intensive, and frequently must be done on short notice because rainfall varies so much over short distances in the mountains. Crops must be planted as soon as the first rains

PAGE 25

10 arrive. Farmers without access to labor at the right time will suffer. Work groups generally consist of 3 to 15 members from the same socioeconomic level, and can be permanent or formed for a single task (Murray 1991, Smucker and Dathis 1998. Swanson et al. 1993a, 1993b 1993d, 1993e). The names given to labor groups differ from place to place within Haiti, and even from farmer to farmer (Smucker and Thomson 1999). The two main kinds of groups are those that are for labor exchange only and those that are formed to work for cash. Client farmers paying cash for labor are generally not group members, and are better off economically. A wealthy farmer may join a labor exchange group just for the purpose of gaining access to the labor. Another kind of work party is called the koumbit, where the host farmer provides one or two meals and sometimes alcohol to other farmers invited to clear or prepare a field (Swanson et al. 1 993a). It is not possible to control the number of participants, the length of time they stay, or the quality of the work done, and for this reason the koumbit is becoming rare (Swanson et al. 1993e). Finally, individual farmers also are engaged by others to do specific tasks. This is referred to as a djob (job). Typically, men form labor exchange groups for agriculture while women form mutual credit supply groups that permit them to engage in commerce. Women have always participated in some aspects of agricultural labor, and this appears to be on the rise because of the out-migration of men during times of political and economic troubles mentioned above (Smucker and Dathis 1998). Farmer Development Groups and Local NGOs Indigenous development groups are of interest because, since the early 1 980s, most donor-financed development work, including agroforestry extension, has been done through them. This happened because of the difficulty in controlling the use of funds

PAGE 26

11 disbursed through the Haitian government, which has always been a predatory structure serving the interests of the urban elite to the exclusion of the rural majority (Smucker and Thomson 1999). Smucker and Dathis (1998) give a comprehensive description of these groups, on which the rest of this section is based. These groups vary widely in origin and composition, ranging from community organizations, cooperatives, youth groups, women's groups, work exchange groups, traditional groups, and religious groups. Agricultural cooperatives began in the 1930s and expanded in number in the 1950s. The community council movement began in the 1950s and was politicized by the Duvalier regime in the 1970s. The Catholic Church created nonhierarchical farmer development groups beginning in the 1960s that became known as groupement and eventually stood in opposition to the community councils. After the fall of the Duvaliers in 1986 the numbers of groups based on the groupement model increased dramatically; but they were persecuted, and they went underground after the 1991 coup d'etat. All of the groups have economic survival objectives, but are not limited to that. They look for more control over the factors of production, including agricultural inputs, credit, grain storage, more control over export crops, and investment in soil conservation and trees. They are now experiencing growing pains as their membership increases and they take on more objectives, without the legal recognition that would permit them to defend their interests and open bank accounts. These difficulties increase as their growth pushes them to depart from the basic characteristics of the traditional local groups, which are direct participation, small size with common interests, hope of tangible profits, direct investment of members' resources, and links of reciprocity and rotation. Farmers would be better able to invest in necessary agro forestry practices if these groups were stronger.

PAGE 27

12 For example, a group with juridical personality, basic accounting skills, and a bank account would be more likely to succeed in negotiating crop sales to large buyers. Profits from sales would allow them to buy labor to protect the plots that produce the crops being sold.

PAGE 28

CHAPTER 3 THE HEDGEROW INTERCROPPING SYSTEM Developments in the Tropics Definition Hedgerow intercropping, also known as alley cropping and alley farming (Rao et al. 1998), has been suggested as an approach to address decreasing soil fertility and for controlling erosion (Kang et al. 1990). It is an agroforestry practice consisting of closely planted and regularly spaced rows of fast-growing, repeatedly pruned trees, usually N,fixing legumes, with short cycle agricultural crops cultivated in the alleys or spaces between the hedges (Kang et al. 1990). It is classified as a simultaneous (trees and crops present in the field at the same time) and zonal (trees are concentrated in the field rather than dispersed) agroforestry practice (Nair 1993). The hedges and crops are managed together for a common purpose, and there are significant aboveand below-ground ecological interactions between them (Sinclair 1999). The most common objective of hedgerow intercropping is to increase crop production in the alleys by applying the pruned tops of the hedgerow trees to the soil as a green manure to improve soil fertility. This objective is pursued by attempting to balance the facilitative and competitive effects inherent in the practice (Vandermeer 1998). The more leaf biomass harvested from the trees, the more green manure available to improve soil conditions and thereby increase crop yield. However, more leaf biomass is obtained by increasing the number of hedgerows and this reduces the space available to crops. It also increases the area of 13

PAGE 29

14 tree-crop interface (Ong and Leakey 1 999) and therefore creates more possible competitive interactions (shading, soil water competition, nutrient competition) with the crops, potentially depressing yield (Vandermeer 1998). Hedgerows fulfill other purposes as well, including soil erosion control and production of fodder and fuel. Origins Hedgerow intercropping developed independently both in the western humid lowlands of Africa and in southeast Asia. In Nigeria, traditional farmers have for several generations planted and managed hedges of a shrub. Dactyladenia barteri (Hook f ex Oliver), for the purposes of contributing nutrients to crops, controlling weeds, and production of fodder and stakes (Kang et al. 1984). Looking for a way to respond to soil impoverishment due to the shortening cycle of traditional slash and burn agriculture, researchers began investigating similar ways to use trees. The principles suggesting intercropping plants with complementary root systems occupying different soil regions (Weaver 1920) and the possibility of tree roots capturing nutrient resources out of the reach of crops (Nye and Greenland 1960) were familiar, but had not yet been applied systematically to agroforestry associations by nonfarmer researchers. The International Institute for Tropical Agriculture (IITA). based in Ibadan. Nigeria, began trying woody species to improve the fertility of low-activity clay soils. This led to the development of alley cropping (also referred to as hedgerow intercropping), avenue cropping, and alley farming if fodder production was a main objective (Kang and Wilson 1987). The first alley cropping trial at IITA began in 1976; on-farm trials were established in the 1980s (Tripathi and Psychas 1992). An earlier start was made in Indonesia. Development of hedgerow intercropping in this region appears to be related to the introduction of Leucaena spp. to the Philippines

PAGE 30

15 and Indonesia by the Spanish from Mexico (Dijkman 1950) and its promotion by the Dutch as a shade tree for coffee {Coffea arabica L.), cacao {Theobroma cacao L.). tea {Camelia sinensis (L.) Kuntze). and tobacco (Nicotiana tabacum L.) in the 18"" century (Wirjodarmodjo and Wiroatodjo 1983). Dutch journal articles from the 1930s report the use ofLeiicaena contour hedges to support bench terraces on slopes (Dijkman 1950). Tacio (1993) cites a traditional ruling in the 1930s obliging all farmers in a region of Timor, Indonesia to plant contour hedges of Leucaena 3 m apart in cropped areas. Metzner ( 1 976) notes Leucaena thickets were promoted by the Indonesian government in the 1930s to recuperate nonarable lands, but farmers did not do it because they feared the weedy qualities of Leucaena. By the 1950s, Leucaena was commonly used in the Pacific as contour hedges for erosion control (Dijkman 1950). Hernandez (1961), cited in Benge (1980), discusses results of Indonesian hedgerow intercropping research done in 1953. The hedgerows decreased erosion and improved maize yield by 380% on the treated plots, compared to traditional cultivation. Extension of hedgerows to farmers, supported by the Indonesian government and religious groups, expanded greatly during the 1 970s (Metzner 1976. Tacio 1993, Wirjodarmodjo and Wiroatodjo 1983). The Rise and Fall of Allev Cropping Research on alley cropping in West Africa in the 1980s generated widespread interest (Kang and Wilson 1987) and led to trials and extension in many parts of the world. In trials examining their technical performance, hedgerows had positive effects on the yield of adjacent food crops through the application of tree prunings as green manure (Tonye and Titi-Nwel 1995), increased rainwater infiltration into the soil (Kiepe 1995), and reduced soil erosion when installed on sloping land (Paningbatan et al. 1995). Some economic analyses indicated that alley cropping, in spite of its high labor

PAGE 31

16 requirement, can be profitable under certain environmental conditions (Ehui 1988). Based on these initial results, project based extension of hedgerow intercropping to farmers in developing countries grew quickly (Ongprasert and Turkelboom 1995, Sanchez 1995, Scherr 1994). However, after a few years many came to believe that alley cropping had been oversold and had much less applicability than previously thought, and that its potential for adoption was limited to certain niche areas (Carter 1995. Sanchez 1995). Based on a review of hedgerow trials covering some of the possible biophysical conditions and hedgerow configurations. Sanchez (1995) lists the conditions where alley cropping is most likely to succeed: areas having fertile soils with no major nutrient limits, adequate rainfall during the cropping season, sloping land with erosion hazard, ample labor and scarce land, and secure land tenure. In spite of the apparent productive advantages discovered in controlled studies and the enthusiasm of project extension technicians, adoption among farmers remained generally low. This has been observed in many regions, including Indonesia and the Philippines (Fujisaka 1991, Nelson et al. 1997), Thailand (Ongprasert and Turkelboom 1995), Central America (Kass et al. 1995), West Africa (Lawry et al. 1994), and East Africa (Mutsotso and Gicheru 1995). Several reasons for poor adoption of hedgerows are given, including damage by grazing animals (Fujisaka and Cenas 1993, Mutsotso and Gicheru 1995), land tenure issues (Fujisaka et al. 1995, Lawry et al. 1994), and competing farm and off-farm activities (Fujisaka 1993, Ongprasert and Turkelboom 1995), but by far the most common reasons are the high costs of labor for installation and management (David 1995, Fujisaka et al. 1995, Kass et al. 1995. Mutsotso and Gicheru 1995, Nelson et al. 1997, Ongprasert and Turkelboom

PAGE 32

17 1995, Stark 1996) and farmers' need for short-term economic returns from their plots (Fujisaka 1993, Ongprasert and Turkelboom 1995. Steiner and Scheidegger 1994). Initial positive results from alley cropping trials were eventually modified by later studies that showed decreased crop production from alley cropping in certain biophysical settings or prohibitive costs of installation and management to farmers in certain socioeconomic settings. The claims made for the potential of alley cropping, it was felt, were exaggerated. But criticisms of alley cropping, summarized in a review written by Sanchez (1995), have also been exaggerated due in part to its having been tested in limited and inappropriate situations (Nair 1 998). and may have caused an unwarranted backlash among researchers (Vandermeer 1998). Rao et al. (1998) reviewed 29 hedgerow intercropping trials that had been conducted for four or more years, mostly on small plots, over a wide range of soils and climates, but no studies done on sloping land were included. They found both positive (15 studies indicated positive results for cereals. 8 trials showed positive results for noncereal crops) and negative (13 trials showed negative results for cereals, 1 showed negative results for noncereal crops) effects of hedgerow intercropping on crop yields across the tropics. Including only the studies showing crop yield increases in hedgerow systems greater than 15% (assuming this would be a minimum level to interest farmers), positive effects on crop yield were found in only 2 of 10 studies in semiarid (<1000 mm rainfall) regions, and 7 of 1 1 studies in subhumid (1000 to 1600 mm) regions where soils were either inherently fertile or supplemented. In 4 out of 8 trials in the humid tropics (>2000 mm) maize and taro {Colocasia esculenta (L.)Schott) did not benefit from hedgerow intercropping, but bean and cowpea yields did. The authors conclude that the

PAGE 33

18 hedgerow intercropping system is location specific and sensitive to management, and that generalizations are difficult to make. One can understand the caveats attached to prescribing or proscribing hedgerowintercropping as a general solution; however, a more balanced approach to both research and extension of hedgerow intercropping is now emerging. There is a growing recognition that its success or failure depends on many site-specific factors including the type of crops grown, the species of plants used as the structural component of the hedgerows, rainfall amount and distribution, soil type, and a host of socioeconomic variables (land and tree tenure, labor costs, etc.). including equity issues, extension approaches, and allowance for adaptation of the practice over time to optimize its economic returns to farmers (Mercer and Miller 1998, Nair 1998, Rao et al. 1998, Sanchez 1995, Scherr and Current 1997). Developments of Hedgerow Intercropping in Haiti Hedgerow intercropping using Leucaena leucocephala (Lam.) de Wit was tried on a small scale in Haiti for the first time in the early 1980s, but there are earlier examples of vegetation-based soil conservation structures. Mintz ( 1 962) notes that Haitian farmers have been building seasonal crop stover barriers staked roughly on the contour to harvest water and soil since the 1950s. These structures, known as ranp pay in Creole, are still being built in some parts of the country. They last only a single planting season. Some farmers say that stover barriers provide habitat for rats and insects, and so they prefer to bum the crop residue. Contour structures have been promoted by development projects since the 1950s. A UNESCO project in the community of Marbial paid people to plant sisal {Agave sisalana Perrine ex Engelm.) on the contour to control soil erosion in the early 1950s, and a FAO project near the town of Les Cayes in the late

PAGE 34

19 1960s promoted contour hedges of napier grass {Penisetum purpweum K. Schumach.), lemon grass {Cymbopogon citratus (DC.) Stapf). vetiver {Vetiveria zizanioides (Linn.) Nash), and guinea grass {Panicum maximum Jacq.) (G. Brice. personal communication: 1 998). These were done by farmers who were paid stipends based on the number of hedges installed, and they were not always appropriately designed. The napier grass hedgerows were planted on shallow, droughty soils. Many eventually grew taller than their shallow roots could support, and toppled over. This also happened to some of the sisal, a crop that is traditionally planted up and down the slope so that farmers harvesting the sharply pointed leaves do not injure themselves if they fall. In contrast to the historically widespread nonadoption of soil conservation strategies designed by technicians, there is an example of a successful practice designed by Haitian farmers responding to a particular market opportunity. This involves the yearly, labor-intensive construction of contour soil bunds (tram) to conserve expensive chemical fertilizers used to produce high-value vegetables in a small region close to the capital city (Murray 1 980). Initial interest in tree-based hedgerow intercropping was stimulated by reports of developments in West Africa, Indonesia, and the Philippines, and by the general interest in the use of Leucaena leucocephala as a fuel wood tree. There is a variety of Leucaena native to Haiti, known locally as cielen, oriman. ti movye. or li pingi. This variety is not suited to hedgerow construction because it is extremely invasive and does not produce sufficient biomass. and is therefore very unpopular among farmers. To promote the use of Leucaena, US AID introduced seeds of productive varieties to Haiti in 1978 (Benge 1985). Seeds of Leucaena K8, K28, and K67 from trees grown in the Philippines were distributed, and later seeds from Flores, Indonesia were introduced. In 1979, a USAID

PAGE 35

20 agronomist also introduced the concept of using Leucaena in hedgerows, and re-wrote an existing paper on Leucaena hedgerows especially for Haiti (M. Benge. personal communication: 1995). The number of hedgerows reported by PADF as built by farmers participating in PLUS and its two predecessor projects probably represents the largest percent of the hedgerows in Haiti. During the seven-year period from 1985 through 1991. 848 km of hedgerows were reported (PADF 1990). The number of hedgerows installed has been much greater under the PLUS project, with over 12,000 km. having been installed from 1993 through mid1999 (PADF 1999). If the number of hedgerows ever installed by farmers working with all aid projects in Haiti is estimated at 15,000 lineal km. and if there is an average of 5 meters between rows, then approximately 7.500 hectares of hillside gardens have had hedgerows. Although this represents less than one-half of a percent of the total land area, it is a significant figure for a new technique whose management is still being developed. Hedgerows have been recommended as a potentially appropriate agroforestry technology and are commonly promoted in Haiti as part of project or government-based soil conservation and natural resource strategies (Bannister and Nair 1990). Garrity and Van Noordwijk (1995) note three paradigm shifts in hillside conservation management: (1) an engineering approach using rock and earthworks in contour structures has increasingly given way to vegetative structures; (2) top-down prescriptive watershed management has evolved to a bottom-up. participative approach; and (3) tree-based alley cropping has diversified to include a wider array of economically interesting plant materials as structural components of contour hedgerows. All three of these shifts have occurred in Haiti. Early attempts at erosion control promoted by projects in Haiti were

PAGE 36

21 mainly based on rock walls constructed across the contour, but these were poorly maintained because they did not respond to farmers' economic situation and the extension approach was inappropriate (Murray 1980). The increasing emphasis on vegetative barriers happened during the 1 980s, as noted above, along with as-yet incomplete attempts to manage natural resources participatively. However, the ideot>'pe of contour hedgerows is not static in Haiti. Farmers continually modify the structural composition and management of hedgerows according to their household needs, their ability to invest resources, their experience with hedgerows, and changing marketing opportunities for crops and products derived from hedgerow trees. Leucaena leucocephala was originally the predominant hedgerow species, but that is changing. During 1995, PADF reported hedgerows made with 22 different tree, perennial food crop, grasses, and other perennial species planted in 54 different combinations. Leucaena was present in only 46% of the hedgerows. In addition, hedgerows are evolving into perennial crop contour bands in some areas. These bands, called bann manje (a play on words — either "'band of food" or "a lot of food") consist of rows of food crops planted across the contour, with a dimension up and down the slope of more than one meter. Perennial crops such as pineapple {Ananas comosus (L.) Merr.) or sugar cane {Sacharum officinarum L.) serve as the structural components, but annuals such as yam and sweet potato {Lpomoea batatas (L.) Lam.) are also planted within the band. Field crops are planted in the interspaces, as in hedgerows. During 1999. 20 different combinations of nine species of perennial food crops, trees, and other perennials were used in crop contour bands (PADF 1999). This evolution is based on farmers' needs for short-term cash return. The structural crops, such as pineapple, respond directly to a marketing opportunity — a quintessentially Haitian strategy (Murray

PAGE 37

22 1991), and one found in indigenous agroforestry associations in other cultures (Khaleque and Gold 1993). This reinforces the notion that hedgerow intercropping should be considered as a seed technology (Sumberg and Atta-Krah 1988. Wiersum 1994). to be adapted and modified by farmers as a matter of course to fit into their unique agricultural systems as they discover (or not) its as-yet unknown economic possibilities. The cycle of enthusiastic promotion of hedgerow intercropping followed by disappointment has a parallel in Haiti, although it is not documented apart from some protest about the widespread promotion of Leucaena hedgerows in the early 1990s (Swanson et al. 1993d. 1993e). It mainly takes the form of disappointment, informally expressed by officials of donor agencies, that hedgerows do not yet cover every garden in Haiti. More often the criticism of nonadoption has been aimed at the extension methodology rather than the practice per se (Sieder 1994.Villanueva 1993). This has been due, in part, to the development-driven agenda mentioned by Nair (1998) resulting in the promotion of large numbers of an untested practice by donorfunded projects. Reasons for farmer adoption, adaptation, or rejection of hedgerows are not well studied in Haiti. The best available studies were done in a small region of the central plateau near the town of Maissade (White 1992b), which noted farmers' reasons for nonadoption of hedgerows as lack of time (39%). don't own land (21%). and land not appropriate (17%). Vaval (1997) studied the same project area later, and documented the disadvantages of hedgerows as expressed by adopting farmers as the time required for management and loss of crop land due to the presence of hedgerows. Hedgerow Intercropping and Soil Conservation Because of the greater risk of soil erosion, performance of hedgerows on sloping lands is not identical to their performance on flat land. Three kinds of biophysical

PAGE 38

23 differences to consider include (1) the overall effect of hillside contour hedgerows on crop production and soil erosion, (2) the gradients of soil accumulation/loss, soil physical properties, tree/crop competition, and crop yield across the overall slope from the top hedgerow to the bottom one, and (3) the same gradients within the alleys. The concerns about the poor performance of alley cropping summarized by Sanchez (1995) did not extend to hedgerow intercropping on sloping land. In part, this is due to the proven ability of contour hedgerows to control soil erosion. Sanchez (1995) stated that the hypothesis that contour hedgerows can substantially reduce soil erosion has been definitively proven. This has been shown in many areas of the world, for example a review often years of research in Thailand concluded that hedgerows controlled erosion in areas of 18 to 55% slopes (Ongprasert and Turkelboom 1995); in Rwanda on 23% slopes hedgerows reduced erosion to less than 2% of the runoff from unprotected plots (Roose and Ndayizigiye 1993); in Malawi on 44%) slopes hedgerows reduced soil erosion to 2 t/ha compared to 80 t/ha on maize-only plots (Banda et al. 1994); in India on 9% slopes hedgerows significantly reduced soil erosion compared to cassava-only plots or bare fallow (Ghosh et al. 1989); and in the Philippines there are many examples of hedgerows substantially reducing erosion on sloped land (Agus et al. 1999, Comia et al. 1994. Paningbatan et al. 1995. Tacio 1993). The studies cited cover a range of soil types from sandy loams (Banda et al. 1994) to gravelly sandy clays (Ghosh et al. 1989) to heavy clays (Comia et al. 1994). However, in spite of their proven ability to control erosion, in many cases researchers have not been able to show crop yield improvements in hedgerow plots compared to traditional systems. The principal reason for this appears to be the same on slopes as it is in flat land alley cropping systems—the tradeoff between the production of

PAGE 39

24 biomass for mulch or green manure needed to increase crop yield, and the loss of cropping space caused by the presence of the hedgerows that produce the biomass (Vandermeer 1998). The inter-row spacing required for the crops does not allow enough biomass production from the hedgerows to improve soil fertility, so maize >'ield. for example, is not improved unless inorganic fertilizer is applied (Ongprasert and Turkelboom 1995. Roose and Ndayizigiye 1993). On steep slopes the Soil Changes under Agro forestry (SCUAF) model (Young and Muraya 1 990) predicts this might be explained by high rates of leaching and organic matter decomposition, despite reductions in the rate of erosion (Nelson et al. 1998). Crop yield declines are especially important in newly established hedgerows as biomass application will not have yet contributed much to soil improvement, although cropping space will have been reduced. Data taken from 89 hedgerow plots planted with maize on sloping plots in the Philippines showed a decrease in maize yield in new hedgerows due to the loss of cropping space. The same study estimated 8 years would be required for soil improvements caused by hedgerows to compensate for loss in maize yield at a 5 m inter-row spacing (Shively 1998). In some cases, however, soil conditions are improved by hedgerows sufficiently to maintain a stable, if low, level of maize production on slopes compared to traditional cultivation, where maize yield falls off over time (Banda et al. 1994, Comia et al. 1994). There are also examples where crop yields in hedgerows on slopes were significantly greater and more stable than those in unprotected control plots, as in a four-year study in Costa Rica on farms having relatively fertile soils (Kass et al. 1995). Gradients of soil water, soil physical properties, and crop yield have been documented in alley cropping systems, as noted earlier in this chapter. For example, infiltration rate and number of soil macropores were greater under the hedgerow trees

PAGE 40

25 than in the middle of the alleys on a 14% slope in Kenya (Kiepe 1995). Competition for light, water, and nutrients between hedgerow plants and adjacent crops is also noted on slopes. Rice yield was depressed in the first two rows away from hedgerows on 1 5% slopes in the Philippines, but the depression varied with different hedgerow species: it was greatest near grass {Penisetum purpureum) hedgerows (Garrity and Mercado 1994). It has also been found that food crops in the alleys can restrict the growth of adjacent hedgerow trees (Ghosh et al. 1989). But as the slope increases a redistribution of soil fertility across the alley is often seen that differs from that of flat plots (Garrity 1994, Ongprasert and Turkelboom 1995). Increases in soil macropores near CaUiandra callothrysus hedgerows in Reunion were greater on the uphill side of the hedges than on the downhill side (Tassin et al. 1995). A study done in the Philippines on 22 to 30% slopes found that average mid-alley maize yields were greater than yields near hedgerows of Gliricidia sepium, Paspalum conjugatum. and Penisetum purpureum; and that the Ap soil horizon was thicker on the lower part of the alleys (Agus et al. 1999). Garrity (1996) reviewed this phenomenon, and noted that it was caused in a large part by the displacement of soil during crop tillage. He has pointed out that on unprotected slopes there is often no net loss of soil from the mid-slope area because any soil loss is replaced by soil eroding from the upper portion of the hillside. The presence of hedgerows prevents a net accumulation of upslope soil into the alleys, but the same displacement effect occurs within the alleys on a smaller scale. Crop yield response on sloped hedgerow plots skews after a few years of cultivation because scouring due to tillage displaces soil from the upper part of the alley to the lower part, changing the soil profile, slope length, soil depth, and hydrological properties. These changes combine with competitive interactions between the crop and the hedgerows. The result is lower crop yields on the upper parts of the alleys and higher

PAGE 41

26 yields on the lower middle part, with a slight decrease again as crops approach the lower alley near the next hedge. The net yield curve across the alley is due to a combination of soil scouring and competition with hedgerow trees. The effect of soil scouring on yield is usually more important than that of tree/crop competition. Solera (1993). cited in Garrity ( 1 996) put in 50 cm plastic barriers between Gliricidia sepium hedgerows and adjacent upland rice on a 20% slope, and found that the across-alley yield response was unchanged by the barriers, and so was not due to competition between the hedgerows and crop. Yield in the upper rows was 50% lower than that in the lower rows within the alleys. The effects of slope on hedgerows are important because in wide alleys the soil displaced from the upper position may lead to long-lasting crop yield depression in that area, and because invalid comparisons between yield in hedgerows are often made against an adjacent no-hedgerow control plot in the mid-slope area that has thicker soil (Garrity 1996). Conclusions Hedgerow intercropping is an evolving technology meant to address the increasingly rapid destruction of soil resources in developing countries. The defining paradox in its history is that researchers and extension agents initially addressed it as a technical solution, whereas farmers looked at it primarily as a potential economic solution. Crop yield improvements due to hedgerow intercropping in certain regions of the world justify continuing research. However, it has not yet been widely adopted even in those regions, mainly due to high initial costs of installation and adaptation and farmer need for short-term returns in the face of the probability of a yield loss in hedgerow plots during the first few years after installation. Both the technical and socioeconomic aspects of hedgerow intercropping must be resolved in order for adaptation and adoption to increase.

PAGE 42

CHAPTER 4 ORGANIZATION OF THE STUDY This study is presented in three parts, beginning with an on-station study of competitive interactions between hedgerows and adjacent maize, followed first by an onfarm study of the performance of maize in hedgerows on sloping plots, and then by an extensive field survey of the biophysical characteristics of plots and socioeconomic characteristics of farmers who have constructed hedgerows and other soil conservation practices with the PADF/PLUS project. The hypotheses to be tested correspond to some of the biophysical and socioeconomic hypotheses referred to by Sanchez (1995) as being fundamental to agro forestry, namely, biophysical hypothesis number 16: competition for water in agroforestry systems can be reduced by modifying the spatial arrangement of trees; and socioeconomic hypotheses numbers 1. 3. and 4: • the identification of key driving socioeconomic and ecological processes within land-use typologies permits the spatial delineation of target and recommendation domains for agroforestry interventions; • the adoptability of a new agroforestry practice is determined by 5 principal components-the farmers' natural resource base, their resource endowment, degree of market integration, cultural preferences and perceived benefits; and • at the farming systems scale, agroforestry adoption has different impacts for different classes of farmers, such as their gender. Justification Declining farm productivity is a major problem confronting Haitian hillside farmers. It is caused and exacerbated by a number of factors. Important among these are soil degradation as a consequence of erosion, and farmers" difficulty in making capital 27

PAGE 43

28 investments to support improvement of lost soil fertility and farm productivity. Traditional agroforestrj' systems such as home gardens, shade coffee, and living fences have not been able to cope with the degradation of soil resources on hill slopes as population increased. Although rural migration to the cities has steadily increased (Locher 1 996). there is no reason to suppose that this will result in a significant decrease in the population density in the mountains. Haiti's population might already have exceeded the ecological carr>'ing capacity of the land (Ewel 1977). Two decades ago, the Haitian landscape was described as the victim of mismanagement: "the ruthless application by the rural population of ecologically inappropriate and catastrophic lumber extraction and gardening techniques which have transformed the once lush landscape of entire regions into grim, denuded savannas" (Murray 1980, page 3). In addition to decreasing productivity of individual hillside plots, inappropriate farming practices also led to other local and regional environmental problems, including transport of rock overburden that covers productive soil of farms on lower slopes and plains, reduction in the water supply from springs, road washouts, and destruction of houses by water and debris flows. Hedgerow intercropping has gone through a cycle of research, widespread promotion based on expected positive performance, extension by governments and funded development projects, mounting criticism when expected performance was not achieved, a resulting waning interest on the part of researchers and projects, and a concern that alley cropping might be rejected due to inadequate identification of the biophysical and socioeconomic conditions where it might succeed. This point was discussed in more depth in Chapter 3.

PAGE 44

29 Consequent to the reported promise and potential of hedgerow intercropping technology in other parts of the tropics in the 1980s (Kang et al. 1984). it was introduced in Haiti in the middle 1980"s through externally funded development projects (Murray 1997. PADF 1986). As with most project-driven interventions, hedgerow intercropping has had var\'ing degrees of success. Several reasons could be attributed to this, both biophysical and socioeconomic. They might include inappropriate extension approaches, mismatch between farmer objectives and project objectives, inappropriate choice of hedgerow species, competition between hedgerows and adjacent crops, high cost of hedgerow establishment and management relative to the resulting effect on crop yield, and farmers' and technicians" ignorance of the costs and benefits of hedgerows. But the list of relevant factors contributing to hedgerow success or failure is poorly understood. Indeed, farmers, extension agents, technicians, and project administrators have varying definitions and perceptions of success and failure. The same cycle of widespread interest followed by disappointment seen on the international level has been repeated on a smaller scale in Haiti, where hedgerow intercropping has sometimes been labeled a failure by consultants and donors who tend to observe the practice at an early stage of development, in a small number of locations, or without a complete understanding of important nontechnical parameters of the practice. During the period 1 984 to 1 999, hedgerow intercropping technology has been installed on approximately 10.000 hectares by farmers working with various funded projects in Haiti (PADF 1990. PADF 1999). This is a significant coverage for a development activity involving a relatively new and evolving technology. The rate of hedgerow construction appears to have surpassed the level of understanding of how and why they are installed and managed.

PAGE 45

30 A number of studies have been published regarding various aspects of hedgerow intercropping in Haiti. These include the effect of hedgerows on soil erosion (Cunard and Lorcy 1990, Regis 1990). financial analysis of some of the benefits of hedgerows (Bellerive 1991, Lea 1993. White and Quinn 1992). choice of hedgerow species (Isaac and Shannon 1994. Shannon and Isaac 1993. Toussaint 1989. Treadwell and Cunard 1992). management of hedgerow tree biomass (Isaac et al. 1995). farmer adoption (Chery 1990. Emmanuel 1990. Lea 1994a. 1994b. Villanueva 1993. White 1992a). and farmer adaptation and management (Lea 1995, Pierre et al. 1995, Swanson et al. 1993a. 1993b. 1993c, 1993d, 1993e. White 1992b). Of the nine papers cited above reporting adaptation and management of hedgerow intercropping in Haiti, only the one by Pierre et al. (1995) was specifically designed to study how farmers manage hedgerows. They studied at 105 hedgerow gardens installed before March 1994 in one region of southwestern Haiti. Although the study reported a number of valuable observations on the management and performance of hedgerows, it did not correlate the management quality indicator (the number of unrepaired breaches) with other factors. Furthermore, the observations were based only on hedgerows constructed of Leucaena leucocephala in a limited area of project intervention. Another major study on hedgerows in Haiti was the series of qualitative, questionnaire-based diagnostic surveys (Swanson 1993. Swanson et al. 1993a, 1993b, 1993c. 1993d. 1993e,) done just after the first field season of the PLUS project at the request of PADF and Cooperative for American Relief Everywhere (CARE), the implementing organizations. These studies were targeted at the whole farming environment, not particularly at hedgerows. The diagnostic team visited over 200 farmers in fifteen small watersheds, interviewing individual farmers in their plots and

PAGE 46

31 speaking with groups of farmers. The majority of the hedgerow gardens seen by the team were installed under the Agroforestr\' II (AFIl) project (1990-1991). The number of hedgerow gardens visited was not stated, and the survey did not attempt to link hedgerow management to specific conditions on the farms. Several useful obser\'ations were made regarding hedgerows, however both the composition and management of hedgerows have evolved since that time, and the number of hedgerows installed by farmers has increased dramatically. In addition, an evaluation of the PLUS project monitoring and evaluation system done in 1995 (Romanoff et al. 1995) suggested that the sampling areas used were not representative of the whole project. They recommended that project impact be measured over the whole area of intervention. Competition between hedgerow trees and adjacent crop plants for soil water has been the subject of an increasing number of studies in recent years (Chirwa et al. 1994a, Govindarajan et al. 1996. Korwar and Radder 1994. 1997, Ong et al. 1991, Rane et al. 1995, Schroth et al. 1995a, Schroth and Zech 1993, Ssekabembe et al. 1994). This is potentially a problem in Haitian hedgerow gardens because of the highly variable timing of seasonal rainfall and the inability of eroded soils on steeply sloping land to absorb and store water. Ewel (1977) referred to this as pseudo-drought, a condition that results from a decrease in the water-holding capacity of the soil rather than from decreased rainfall. The literature on hedgerow-crop soil water competition shows that, under some conditions, hedgerows compete to the disadvantage of adjacent crops (Femandes et al. 1993), (Govindarajan et al. 1996. Kamasho 1994, Korwar and Radder 1994, Ong et al. 1991. Salazari et al. 1993. Singh et al. 1986) and under other conditions, they do not (Bohringer and Leihner 1997, Chirwa et al. 1994a, Haggar and Beer 1993, Jeanes et al. 1996, Schroth and Zech 1993).

PAGE 47

32 One possible influence on farmer adaptation and management of hedgerows is farmers' perception of the importance of soil water competition between the hedgerowspecies and the interplanted crops. Experience in Haiti shows that maize (Zea mays) and bean {Phaseolus vulgaris) crops, even without hedgerows, ver\' often fail because of drought conditions. Therefore, hedgerow trees are likely to compete with interplanted crops for soil water at some time during the cropping cycle. The degree of competition may also vary according to the position of the crop on the slope relative to the hedgerows (Garrity 1996). However, it is not clear if farmers consider this competition as more or less important than other perceived problems caused by hedgerows. Do farmers modify hedgerow structure and management to compensate for perceived soil water competition? Neither competition for soil water between crops and hedgerows, nor crop yield relative to slope position and /or relative to hedgerows have been studied in Haiti. Since intensive agriculture is practiced on small steeply sloping plots, it is necessary to look for economically rational ways to maintain the soil's ability to sustain crop production and to protect the local environment. Development of new agro forestry systems that build upon traditional practices is needed. Hedgerow intercropping has been shown to produce desirable results under certain circumstances. However, adoption or lack of adoption may depend on factors extraneous to the characteristics of the technology itself, including the socioeconomic status of farm families and the kinds of land they access for agriculture. The extremely degraded condition of Haitian natural resources makes it necessary to try any practice that could improve the economic condition of Haitian farmers. Because of experience gained from agroforestry projects operating in Haiti since the 1980s, there is reason to believe that adaptations of hedgerow intercropping may increase farm stability and productivity in some areas of the country.

PAGE 48

33 This study will help inform decision makers regarding the potential for use of hedgerowintercropping in Haiti, and possibly avoid premature decisions to discontinue the effort to adapt the technology. Thus, hedgerow intercropping, a rapidly evolving and yet unproven technology, has been extended in Haiti without the necessar}' supporting studies and documentation. Development and extension resources could be used more efficiently by responding to two documented shortcomings of agro forestry research that seem particularly appropriate to the Haitian situation: that there is an inadequate understanding of below-ground biophysical interactions in hedgerow systems: and that socioeconomic and biophysical research must be linked, as adoption of the appropriate agro forestry practice depends upon both (Rao et al. 1998). Hypotheses On-station Study • Leucaena hedgerows reduce the amount of soil water available to adjacent maize, thereby causing a reduction in maize growth and yield. • Root barriers installed next to Leucaena hedgerows increase the amount of soil water available to adjacent maize, thereby causing an increase in maize growth and yield. • Cutting the roots oi Leucaena hedgerows affects the production of tree biomass. On-Farm Study • The pattern of competition between hedgerow trees and adjacent crops is different on slopes vs. flat land, and maize growth varies with regard to the position of the crop within the alleys and on the slope. On-Farm Sur\ev • Family characteristics influence the adoption and management of hedgerow intercropping.

PAGE 49

34 • Plot characteristics influence the adoption and management of hedgerow intercropping. • Problems and benefits with hedgerow intercropping as perceived by the adopting farmer influence management quality. General Description of Hvpothesis-Testing Procedures The study of soil water competition between Leucaena hedgerows and adjacent maize was done on an NGO-operated research station located near Port au Prince. Four treatment combinations were tested in nine replications: hedgerows with and without root barriers between the trees and the adjacent row of maize, and maize fertilized or not fertilized. Data for three maize seasons are reported, including gravimetric soil water; maize height, leaf width, growth stage, leaf water potential, and final harvest; and Leucaena height and biomass production. A pre-study examination of the distribution of Leucaena hedgerow roots at a different location was done, and a post study count of the distribution of Leucaena roots in hedgerows with and without barriers was done onstation. A pot study of the correspondence between soil water percent and maize leaf water potential was done as well. The small on-farm study was an attempt to relate the on-station findings on flat land under "controlled" conditions to sloping land managed by farmers. The study collected maize growth and yield data for one season from four farms having previously established trials of Leucaena hedgerows, crop contour bands, rock wall terraces, and traditional cultivation. Maize growth data were collected by position in the alleys and on the slopes. The four farm plots had been a part of an abandoned PLUS project research effort previously carried out by the Southeast Consortium for International Development (SECID)/Auburn University in conjunction with PADF. The author designed the data collection protocol and performed the analysis.

PAGE 50

35 A large list-frame questionnaire-based study was conducted, which provided onfarm data about households adopting agroforestry soil conservation practices (including hedgerows), how they were managed, and the characteristics of the plots where they were installed. The study was undertaken by the author with the help of PADF PLUS technicians as part of project monitoring during the early part of 1996. We visited 1.540 PLUS project agroforestrvadopters and collected information regarding all plots controlled by the family, and information regarding family members. We then visited 2.295 plots where farmers had installed at least one PLUS project agroforestry practice. Data were collected on the biophysical parameters of the plots, the amount and condition of the agroforestry practices installed on the plot, and the opinions of the farmers regarding problems and benefits of the practices. Because farmers' assessment of soil fertility appeared to influence where hedgerows were installed, soil samples were collected from 175 farms. These were analyzed for percent organic carbon and soil nutrients, and compared to farmers" classification of soil fertility. Details of data collection and analysis for the on-station and on-farm hedgerow/crop competition studies are reported in Chapter 5. Those for the on-farm survey are reported in Chapter 6.

PAGE 51

CHAPTER 5 HEDGEROW/CROP COMPETITION Introduction This chapter looks at competitive interactions between hedgerows and adjacent maize on station and on farm. Competition is important because it is a key consideration in designing agroforestr>' practices, it is difficult to measure, and farmers make adoption and adaptation decisions based on its perceived severity. Competition is a central issue in agro forestry because of the potential advantage in combining plants in such a way that they capture more growth resources in combination than they would growing separately, thereby producing greater physical yields (Cannell et al. 1996). The cost of the positive effects of facilitation gained by growing plants in close proximity is the negative effects of competition. In hedgerow intercropping in particular, the competitive characteristics of the trees (they take up space, light, soil water, nutrients) must be balanced with their facilitative characteristics (they impede soil erosion, capture soil water and nutrients out of reach of the crops, improve soil fertility through biomass additions, improve the microclimate, improve soil physical properties) (Vandermeer 1998). It is hard to develop a hedgerow system that attains that balance because so many combinations of plants, physical environments, and socioeconomic situations are possible. It is also difficult to measure competition and to identify the processes that contribute to it because competitive interactions change as the environment changes (Anderson and Sinclair 1993). and the hedgerows themselves change the environment — indeed, that is their 36

PAGE 52

37 function. The way hedgerows and adjacent crops interact varies according to the age of the hedgerows, relative size of the crops and hedgerows, population density, and the supply of and access to limiting growth resources (Ong and Leakey 1999). Farmers make adoption decisions based on competition between hedgerows and crops. For example, farmers in the Philippines eliminated napier grass as a hedgerowcomponent because it was too competitive with crops grown in the alleys (Fujisaka et al. 1995), where it decreased maize yield by 86% by the second year after establishment (Garrity and Mercado 1994). Therefore, competitive characteristics are important considerations in the design of hedgerow systems. Leucaena has characteristics that make it an aggressive competitor for growth resources, including rapid growth and roots that occupy approximately the same soil depth as that of maize (Jonsson et al. 1988), the crop of interest in this research. It was noted several decades ago in Southeast Asia that Leucaena used as shade trees in plantations rarely competed with the crop plants; but did so in times of drought, when signs of water competition with coffee and rubber were reported (Dijkman 1950). In more recent alley cropping studies, competition is a frequent subject of investigation, but results vary with the specific site and plant components. Leucaena has been shown to compete for soil nutrients to the detriment of rice (Salazari et al. 1993) and grasses (George et al. 1996). In the semiarid tropics especially it has been shown to reduce the amount of soil water available to pigeon pea. but less so for sorghum and millet {Pennisetum glaucum L.) (Singh et al. 1986). In other areas of the semiarid tropics, however, yields of both sorghum (Korwar and Radder 1997) and millet (Ong et al. 1991) were significantly reduced in Leucaena hedgerows.

PAGE 53

38 The on-station and on-farm studies reported on here both have maize planted in the interspaces between Leucaena hedgerows. This combination has produced both positive (Akonde et al. 1996. Banda et al. 1994, Hauser and Kang 1993, Hernandez 1961, Mugendi et al. 1999. Mureithi et al. 1994) and negative (Chirwa et al. 1994b, Govindarajan et al. 1996, Jama et al. 1995. Kamasho 1994) results. The positive results in the cited studies ranged from maintaining the existing maize yield (versus a decrease without hedgerows) to increases of 26%. 44%, 50%, and 400% in grain yield compared to traditional cultivation. Positive results tended to be from subhumid or humid areas with soils of moderate or better fertility, and from sloping areas. The negative results ranged from 1 6% to 20% decreases in maize yield. These tended to be from semiarid, moisture limiting, low fertility sites. Effects of competition between maize and Leucaena are noted in several studies, even where the overall effects of hedgerows are positive. In an early report of alley cropping research in Ghana, Balasubramanian (1983) mentioned that soil water competition with Leucaena might be a partial explanation of yield depression in adjacent maize. Later hedgerow studies have attributed yield depression in adjacent crops to competition for soil water (Govindarajan et al. 1996, Korwar and Radder 1997. Ong et al. 1991. Ssekabembe et al. 1994), nutrients (Chirwa et al. 1994b, Livesley et al. 1997, Mureithi et al. 1994). or light (Kang et al. 1981. Lai 1989. Tilander et al. 1995. Welke et al. 1995, Zoschke et al. 1990). Some studies claim all three resources are competed for (Gaddanakeri et al. 1994). It is difficult to use these location-specific results to make predictions about the probable success of a hedgerow intercropping system in another location. However, Rao et al. (1998) have drawn some general conclusions: (1) In water limiting areas, low yield of hedgerow prunings and competition for water between hedgerows and adjacent crops were the major reasons for the negative results.

PAGE 54

39 Inadequate water limited the response of crops even though hedgerows improved soil fertility in certain sites of the semiarid tropics. (2) In poor soils, low yield of hedgerow prunings and competition for nutrients were responsible for negative results. (3) The advantages of hedgerow intercropping are more common on relatively fertile. N-deficient soils in subhumid and humid environments, in areas where hedgerows produce a large quantity of prunings. and where there is adequate water for both hedge and crop growth. They also note that if competition is substantially reduced, even in semiarid climates, hedgerows may be able to increase crop yields even where relatively small fertility contributions are made by the hedges — but the cost of managing the hedges could be too high to be practical. Much of Haiti is subhumid. and there are still soils having a little remaining fertility, on sloping land where hedgerows may be appropriate. The discussion to be presented in this chapter will focus on the biophysical aspects of hedgerows and so will necessarily be incomplete, lacking the socioeconomic elements, which will be discussed in Chapter 6. Qn-Station Studies: Effect of Root Barriers on the Growth of Maize and Leucaena The objectives of this experiment were to test whether or not permanent root barriers placed on either side of Leucaena leucocephala hedgerows affect the production of maize grown in the alleys and the production of Leucaena leucocephala biomass. The hypotheses were that the barriers will reduce competition for soil water between Leucaena and adjacent maize, thereby increasing maize yield; and that production of Leucaena will not be adversely affected by the barriers. If maize yield were to be increased significantly an appropriate modification of the barrier could then be devised for use on-farm, or a less competitive tree could be sought to construct hedgerows.

PAGE 55

40 Crop yield depression close to trees is common. It has been reported for mustard growing under Acacia nilotica in semiarid central India, where a yield depression of 54% extending up to 8 m from the tree was correlated with increasing shallowness of the tree roots (Yadav et al. 1993). Yield depression of maize and soybean near Siberian elm {Ulmus pumila) and eastern cottonwood (Populus delloides) windbreaks in Nebraska was significantly relieved by plowing parallel to the trees (Rasmussen and Shapiro 1989). A study done in the savanna region of Kenya reported that Acacia tortilis trees competed more intensely with understory grasses and herbs in wetter sites where the tree roots terminated near the crown region, than in drier sites where the roots extended farther into the grassland (Belsky 1994). Yield depression in millet extended significantly beyond the area shaded by eucalyptus shelterbelts in Nigeria, corresponding to the area of tree root growth (Ojyewotu et al. 1994). Nadagouda et al. (1994) found that severing the roots of eucalyptus trees in India improved yield of adjacent maize. Unless effective experimental controls are put into place, it is difficult to separate the effects of abovefrom below-ground competition, and the effects of competition for water from those of nutrients. Corlett et al. (1992b) describe a study done in the semiarid tropics of India in which millet was grown alone and in Leucaena hedgerows, either with or without barriers between the tree and crop roots. Millet yield was the same in the sole crop and in the hedgerows with root barriers, but was 36% less in the hedgerows without barriers. The authors proposed that, although one might have concluded that the yield difference was due solely to below-ground competition, a better explanation was that the millet adjacent to hedges with barriers grew taller because it had access to more soil water than millet in hedgerows without barriers, and avoided light competition with the Leucaena. There was an interaction between aboveand below-ground processes.

PAGE 56

41 Common methods used for studying the effects of below-ground competition in alley cropping research include root pruning (Fernandes et al. 1993. Korwar and Radder 1994, Sitompul et al. 1992), trenching (Schroth et al. 1995b) or the installation of permanent plastic barriers between the hedgerow trees and the crops (Corlett et al. 1992b, Hauser and Gichuru 1994, Ong et al. 1991. Schroth et al. 1995a, Schroth and Lehmann 1995. Schroth and Zech 1993. Solera 1993c. Ssekabembe et al. 1994). The usual practice is to open a narrow trench to a depth of 50 to 100 cm between the hedgerow trees and the first crop row, install plastic awning or polyethylene sheeting, and then fill in the trench. Studies using plastic barriers have produced results specific to the characteristics of the site, species of hedgerow tree and crop, alley width, planting density and distance between the hedgerow and crop, and the top-pruning regime (height and timing of cut) used on the hedgerow. Ong et al. (1991) showed that yield of millet in hedgerows without barriers was 40% less than yield in sole-cropped millet in semiarid India, but where barriers were parallel to hedgerows the yield was the same as that in sole cropping. They also showed that the barriers did not affect the production of the Leucaena hedgerows, but the barriers did cause a redistribution of the Leucaena roots. Schroth and Lehman (1995) found in subhumid Togo that barriers increased the yield of maize in Calliandra callothyrsus hedgerows, but when the hedgerow species were Senna siamea or Gliricidia sepium the presence of barriers was associated with a decrease in maize yield. They concluded that Calliandra was relatively competitive compared to the other trees, and that the barriers reduced the volume of exploitable soil available to the roots of the other two species. The results obtained by studies using plastic barriers between trees and crops do not always show competition for soil water. Two studies done in the West African humid tropics put plastic barriers to 90 cm between Gliricidia hedgerows and

PAGE 57

42 adjacent crops of rice, maize, and peanut. They showed crop yield improvement in hedgerows compared to the sole cropping controls, but crop yield in hedgerows was not affected by the presence of barriers (Schroth et al. 1995a. Schroth and Zech 1993). A study done in the United States was able to isolate the effect oi Robinia pseiidoacacia hedgerows on soil water at two sites having different soil characteristics (Ssekabembe et al. 1994). The plots were irrigated, no crops were grown in the alleys between the hedges, and the soil was covered with plastic to reduce evaporation. At the site having a higher soil organic carbon content, hedgerows without barriers reduced soil water content of the alleys by about 8%. At the second site, where higher gravel content impeded the growlh of tree roots, hedgerows without barriers reduced soil water content by 32%, and there was significantly less water in alleys without barriers. Materials and Methods The experiment was carried out over a period of six years, from the spring of 1990 through the spring of 1995. During this period data collection was interrupted several times by civil unrest in Haiti,' so although six maize crops and one sorghum crop were planted only three successive maize crops from spring 1 994 though spring 1 995 are ' There was a coup d'etat in September 1991, which resulted in several missed observations due to curfews. A night guard at the Operation Double Harvest (ODH) site, where the experiment was installed, was killed by soldiers during that period, preventing pre-dawn leaf moisture potential measurements. 1 was evacuated to the U.S. in October 1991 and returned to Haiti in September of the following year. By that time, the Direction of ODH had changed hands, requiring that I obtain permission to continue my work on the plot. This took several months, and the removal of a leaking irrigation line installed next to the plots was necessary. During the regime of the de facio military government (September 1991 to September 1994). fuel shortages were common because of an international embargo, again interrupting data collection. Farmers with friends in the local police station let donkeys into my plots, ruining the Fall 1993 harvest. Farm workers bathing in the plots behind the screen provided by the Leucaena also prevented soil moisture observations from being taken at least twice. The U.S military intervention in September 1994 made adequate data collection difficult during that period as well.

PAGE 58

43 discussed. Data available from the three maize crops were not identical, again due to civil unrest and equipment problems. The spring 1991 data are used only for the discussion of the effect of trenching on Leucaena height and biomass. Trial site The trial site was located at 18°33" nonh latitude 72° 1 T west longitude, with an elevation of 80 m above sea level on property owned by Operation Double Harvest (ODH), a nongovernmental organization (NGO) operating a farm and tree nursery in the Cul de Sac plain, 30 minutes northeast of Port au Prince near the town of Croix des Bouquets. Haiti. The soil was a vertisol of zero slope. From observations made in a soil pit near the trial plots, heavy texture predominated to a depth of 1 . 1 m, where there was an abrupt change to sandy, unconsolidated limestone to at least 1 .7 m, the depth of the pit. Soil texture from the surface to 30 cm depth was silty clay, changing to silty clay loam in the 30-45 cm depth. Table 5-1 shows texture and organic carbon content for the trial site. Table 5-1 : Soil texture and organic carbon at the trial site. "/o Organic Depth % Sand % Silt % Clav Carbon Texture 49 42.6 1.71 silty clay 41 51.8 1.10 silty clay 53 39^8 LO silty clay loam A pot study was established in January 1994 to determine how soil water depletion at the ODH site correlated with plant water stress (Appendix A). When soil water percent in the root zone decreased below about 25%, leaf water pressure passed 0-15 cm

PAGE 59

44 below 10 bars. Maize plants began to exhibit signs of drought stress at this point. Field capacity at the site was reached at 41% soil water content. Average annual rainfall based on data collected by ODH staff since 1977 was 968 mm per year. Distribution is bimodal. with most of the spring rains falling from April through June, and the fall rains from August through November. The total annual rainfall since 1977 varied from a low of 670 mm to a high of 1.301 mm. but in most years rainfall was at least 900 mm. The trial site was located to the north of a small-container tree nursery, and was previously used for tractor cultivation of hybrid sorghum. The area to the east was used for composting sugarcane bagasse. A low earth barrier separated the trial area from the composting area, to prevent water used for composting from entering the trial. Trial design The trial design was a 2 by 2 factorial randomized complete block having 1 1 hedgerows of Leucaena leucocephala variety K636 planted at a within-row density of 10 trees per meter and a between row spacing of 5 meters. Each block consisted of a hedgerow 22 meters long running through the center of four maize plots, to which were randomly assigned the four treatment combinations. The hedgerows were laid out exactly on an east-west axis, with hedgerow (block) 1 on the north end of the site and hedgerow (block) 1 1 on the south side (Figure 5-1 ). The symbol T+ denotes a trenched plot. T-a nontrenched plot: the symbol F+ denotes a fertilized plot. F~a nonfertilized plot. All plots received the designated treatments throughout the three seasons of the study. Figure 5-2 shows the layout of one of the plots within a block. Each plot was centered on the hedgerow, and included four rows of maize on the north side and four on the south side at 50, 100, 150, and 200 cm distances from the

PAGE 60

45 BLOCK 1 2 3 4 5 6 7 8 9 10 11

PAGE 61

46

PAGE 62

47 hedgerow, respectively. Buffer rows of maize were planted at 250 cm from the hedgerows, marking the middle of the alley between hedgerows. Two-meter buffers separated the plots within the blocks. Rows 1 and 11 were borders from which no data were collected. Hedgerow experiments have been criticized for producing invalid results because the no-hedgerow controls (sole crop plot) have been exploited by hedgerow roots growing into adjacent plots, thereby over-estimating the benefits of hedgerow intercropping (Hauser and Gichuru 1994). Some more recent studies have compensated for this by digging trenches between hedgerow plots and control plots (Chamshama et al. 1998. Mugendi et al. 1999). Although the study reported here does not have sole crop plots, and a two-meter buffer was left between plots, it is possible that Leucaena roots in plots both with and without barriers exploited the soil in adjacent plots. This could not be confirmed in this experiment. Treatment installation The hedgerow trees, grown in the ODH nursery', were planted out in April 1990 as one-year-old seedlings. The trees were cut to 50 cm stumps on 23 March 1991 . In April 1991. two randomly selected plots in each block were trenched to a depth of 30 cm at a distance of 20 cm from each side of the hedgerows. The trenching was done with picks, severing all roots. Plastic sheeting was put into the trenches to inhibit root regrowth; the small canals created during root cutting were backfilled with soil. Trenching extended for one meter beyond each side of the net plot boundaries, halfway into the twometer buffer between each plot. The decision to cut the roots to a depth of 30 cm was based on a preliminary study oi Leucaena roots done at a location off-station (Appendix B). That study indicated 70% to 96% of the very fine roots (< 1 mm diameter) were found in the top 30 cm of the soil, the lowest percent being found 150 cm

PAGE 63

48 and 200 cm uphill from the hedgerow. A lesser percent of the fine roots (1-5 mm diameter) were found in the top 30 cm (20 to 83%). with the least (20%) at the 200 cm uphill position. Very few medium and large roots were found, but of those that were. 50% or more were deeper than 30 cm. Studies in semiarid India also found the majority of Leucaena hedgerow root biomass in the top 30 cm of soil (Ong et al. 1991. Singh et al. 1996). Due to my forced absence after the spring 1991 maize season the Leiicaena hedgerows in the main experiment were not top pruned between October 1 99 1 and September 1993. By that time, the hedgerow trees had grown to basal diameters of up to 10 cm. They were then cut back to a 50 cm stump height. Because the Spring 1994 maize harvest indicated the effect of trenching had almost disappeared, the plots were retrenched to a depth of 50 cm in August 1994 before the fall maize planting. Plastic barriers were buried as before. The plastic barriers from the first root pruning were still in place, but were missing or stiff and degraded in spots. In some places, soil had covered the top of the barriers during cultivation, and Leucaena roots had grown over the barriers into the alleys. A selected local maize variety, known as chicken com, was planted between the hedgerows. The within-row spacing for maize hills was 75 cm. with 50 cm between rows. Hill positions were laid out precisely using nylon cord with flagging tied at 75 cm intervals. Maize was planted by hand three seeds per hill, and later thinned to two plants per hill. The maize density of the interspaces at full stocking was then 53.333 plants per hectare. If the space occupied by the hedgerows is included, maize plant density was 48,000 plants per hectare. Observations of 28.000 and 40,000 maize plants per hectare were made on two farms in the south of Haiti at harvest time (after substantial mortality

PAGE 64

49 during the season) in plots without hedgerows. Farmer interviews in one region of northern Haiti indicated a higher density at planting. 62,500 plants per hectare (Swanson etal. 1993a). The fertilizer was to have been applied according to recommendations of the Ministry of Agriculture. Haiti (MARNDR 1990): an initial broadcast application of 1515-15 (NPK) of 175 kg/ha at seeding followed by two applications of banded urea at 50 kg per hectare. In practice, however, due to the timing of rainfall and political instability delaying travel to the site, only the spring 1995 maize crop received the complete fertilizer treatment. The spring 1994 crop received only the initial application of 15-1 515, and the fall 1994 crop received the 15-15-15 and the first urea application. Once a given plot had been designated as fertilized, it remained a fertilized plot throughout the three seasons reported here. Operations bv maize season The spring 1994 maize crop was planted on 30 April. There had been 5 1 mm of rain during the previous 7 days, 93 mm during the previous 14 days. Most of the rain during the growing season fell between 1 1 and 21 DAS, with only one substantial rain event after that, 23 mm at 67 DAS. A total of 206 mm of rain fell between sowing and final harvest. The stand was thinned to two plants per hill at 28 DAS, and was weeded once at 16 DAS. Insecticide was applied twice during the season. Fertilizer was applied at a rate of 1 75 kg/ha at planting only; no urea was applied. Leucaeana top pruning was done 35 days before sowing, then at 33 DAS (when trees were about 2.3 m high, including the stump), and 65 DAS (when trees were about 1.8 m high). Figure 5-3 shows the amount and timing of the rainfall in relation to the growth ofLeucaena and maize, and the timing of the agronomic interventions made to the plots during the spring 1994

PAGE 65

50 season. The fall 1994 maize crop was planted on 10 September; two weeks after the hedgerows were re-trenched to a depth of 50 cm (Figure 5-4). No measurable rain fell during the 7 days preceding sowing, but 49 mm fell during the previous 1 4 days. Tropical Storm Gordon passed over Haiti on 14 November (65 DAS) and dropped over 150 mm of rain in a 24-hour period on the trial site. This surpassed the capacity of the rain gauge, so an accurate measurement was not recorded. This was the rainiest season of the three, with at least (because of the inaccurate measurement during Tropical Storm Gordon) 374 mm measured between sowing and final harvest. Most of the seeds had germinated and had attained a height of 10 cm by 21 DAS, when they were weeded and thinned to two plants per hill. Insecticide and urea (50 kg/ha) were applied at 26 DAS, when the maize was about 25 cm tall. Because the effect of trenching was weak during the previous season, the hedgerows were re-trenched to a 50 cm depth on 27 August 1994, 14 days before sowing the maize. Fewer measurements were made this season than in others due to the U.S. military intervention in Haiti on 19 September, and the events preceding and subsequent to it. Figure 5-4 shows the amount and timing of the rainfall in relation to the growth of Leucaena and maize, and the timing of the agronomic interventions made to the plots during the fall 1994 season. The spring 1995 maize was planted initially on 26 March, but this planting failed due to lack of rain. A new planting was done on 22 May. During the seven days before sowing, 20 mm of rain had fallen, for a total of 39 mm during the 14 days before sowing. After planting no rain was recorded until the 34'^ day, then 171 mm fell between 34 and 73 DAS. However, because the technician responsible for recording rainfall was on vacation between the planting date and 36 DAS, two or three rainfall events were not recorded. No additional rain fell before final harvest. I applied 15-15-15 at sowing and

PAGE 66

51 Rain (mm) 49 0.8 67 83 49 174+ Total rain 0-1 14 DAS: 374+ nrn Height (cm) Leucaena height DAS -20 applied we^eb urea • 15-15-15 /(ninned insecticide Avg. maize height 10 cm Avg, maize height 25 cm Most maize in anthesis, silk 140' 200 150 100 50 Maize harvest, 114 DAS Maize height up to 300 cm ^ Leucaena topped, 28 & 126 DAS ^ Maize biomass harvest, 78 DAS + Maize leaf w ater stress, 42 DAS I Soil w ater content, 42 DAS Figure 5-3: Rainfall. Leucaena and maize growth, and agronomic inputs: spring 1994 maize cropping season Rain (mm) 93 27 154 rM r^ Leucaena height DAS -20 23 4 Total rain 0-91 DAS: 206 mm Height (cm) I 1 —.250 200 20 ± 40 led weeded thinned,\^ 15-15-15 insecticide, \insecticii Avg, maize height 20 cm Avg. maize height 100 cm — 150 100 120 140 —10 — 100 50 Maize harvest, 91 DAS Some maize in tassel Most maize in anthesis, silk Maize leaf margins fired, soil has 1" cracks X. Leucaena topped, 33 & 65 DAS ^ Maize biomass harvest, 42, 49, 63, & 70 DAS + Maize grow^th stage, 56 DAS • Soilwater content, 56 & 70 DAS Figure 5-4: Rainfall, Leucaena and maize growth, and agronomic inputs; fall 1 994 maize cropping season

PAGE 67

52 banded urea at 12 and 37 DAS. The maize was thinned to two plants per hill at 12 DAS. The plots were weeded four times during the growing season; insecticide was applied twice. Figure 5-5 shows the amount and timing of the rainfall in relation to the growth of Leucaena and maize, and the timing of the agronomic interventions made to the plots during the spring 1995 season. Rain (mm) 83 1 40 41 91 h r^ r^ Total rain 0-131 DAS: 171+ mm Heiglit (cm) —,250 Leucaena height DAS -20 applied thinned/weeded ui 15-15-15 urea/ insecticide Maize height 10-30 cm. leaf curl 200 150 100 50 Maize harvest, 131 DAS 140" Maize height up to 95 cm. poor close to hedgerow s in TSome maize in tassel .X. Leucaena topped, 19, 57 & 130 DAS ^ Maize height, 40 DAS + Maize growth stage, 68 & 75 DAS jf Maize biomass harvest, 47 & 54 DAS % Soil w ater content, 26 & 61 DAS Figure 5-5: Rainfall. Leucaena and maize growth, and agronomic inputs; spring 1995 maize cropping season Measurements recorded The following observations and measurements were made during the three maize cropping seasons: Maize height . This was done only for the spring 1995 season. Height to the top of the highest fully open leaf of the tallest plant in each hill in blocks 2, 4. 6, and 8 was measured at 40 days after sowing. Analysis was done on the mean height of eight plants in each of the four hills to the north and four hills to the south of the hedgerow at a given

PAGE 68

53 distance in each plot. If all plants in a hill had died, a height of zero was assigned for that hill. Maize growth stage . Growth stages of the plants were noted at 56 DAS in the spring 1994 season, and at 68 and 75 DAS in the spring 1995 season. In the spring 1994 season, the scale was: 0=no maize plant present. 1= vegetative growth. 2=tassel. 3=anthesis. 4=silk. For the spring 1995 season, the scale was changed to include the presence of ears: 0=no maize plant present, l=vegetative growth. 2=tassel/anthesis, 3=silk. 4=ear. Observations were made in blocks 2, 3. 5, 6, 8, and 9 on every maize plant (2 plants per hill) in the spring 1994 season, and on only the most developed plant per hill in the spring 1995 season. Maize biomass during the growing season . For each of the four distances and in each of the four plots in blocks 4, 7, and 10, the combined weight of one hill of maize on the north side and one hill on the south side of the hedgerow was recorded. The plants were cut at ground level, and then dried in a Watlow drying oven at 70 degrees C until a constant weight was obtained. Maize leaf water pressure . Predawn leaf water pressure was taken using a PMS pressure chamber for one leaf at each distance in blocks 3, 6, and 9. A 2 cm by 10 cm section was cut from the highest fully expanded leaf, one end was placed in a zip-lock bag, and the sample was inserted into the pressure chamber. Compressed nitrogen gas is normally used in the chamber, but as this was not available in Haiti compressed air was used instead. This measurement was to have been done periodically throughout each maize season, but was done only once because of the security problems discussed in the footnote on page 44.

PAGE 69

54 Maize vield . Grain yield, number of plants, and number of ears were measured from blocks 2. 3, 5. 6, 8, and 9 at 91 DAS for the spring 1994 harvest. 1 14 DAS for the fall 1995 harvest, and 131 DAS for the spring 1995 harvest. At each of the four distances (50, 100. 150. and 200 cm), plants were harvested from a 3-meter length on the north and south sides of the hedgerows, for a total of 6 meters (8 hills, or 16 plants maximum) harvested in each plot at each distance. Because the grain moisture meter ceased to function correctly, field dry grain weights are shown in the tables and figures. Harvests for all treatments were made at the same time, and all samples were weighed at the same time. Final harvest measurements are converted to per hectare equivalents by dividing the sample weight or count for a given distance from the hedgerows by six (six meters per plot were sampled) and multiplying by 4,500 (each distance represented 4,500 meters per hectare given a 50 cm row spacing). The per-hectare contributions of the rows at each distance were then summed to give a total per hectare figure, which included the 2,000 m occupied by the hedgerows. Soil water percent . Soil cores were taken using a JMC Back-saver corer. Each core was divided into three sections. 0-15 cm. 15-30 cm. and 30-45 cm. Four cores for each depth and distance were combined for each measurement. The samples were weighed wet in the field and dried in an oven at 70 degrees C until a constant weight was obtained. Soil water content was determined gravimetrically by dividing the weight of water lost in drying by the dry weight of the sample. The range of soil water content that could be sampled at this site was narrow due to the heavy texture of the vertisol. When the soil was very dry, the 0-15 cm depth often was not retained in the sampling tube, and when the soil was wet. the end of the tube plugged and did not allow soil to enter. For the calculation of field capacity, a large-diameter hammer-driven corer was used. This is

PAGE 70

55 a slow procedure and the large-diameter corer could not be used under normal circumstances due to the necessity of completing 192 cores in a single day each time sampling was done. Soil water content was sampled twice during the spring 1 994 growing season, at 56 and 70 DAS. The first, at 56 DAS. was done on blocks 3 and 6 only because the corer shaft broke. All subsequent samplings were done on blocks 3. 6. and 9. A single sample was taken in fall 1994 at 42 DAS. and two were taken in spring 1995 at 26 and 61 DAS. An additional attempt was made to measure soil water content at 54 DAS, but this was not successful because the soil was too wet and the corer jammed. Leucaena small stem and leaf biomass and height . Leucaena tops were cut back to a 50 cm stump height using a machete. Leaves were not separated from stems: they were weighed together in the field with a spring scale. Sub-samples were taken and dried in a Watlow oven at 70 degrees C until a constant weight was obtained. Height above the stump of the tallest stem in each plot was measured with a fiberglass tape. All leaves and stems were removed from the site after harvest. Leucaena root intersections in trenched and nontrenched hedgerows . After the spring 1995 maize harvest was completed, numbers of Leucaena roots were counted in trenched and nontrenched hedgerows to discover what influence plastic barriers had on root distribution. Transects perpendicular to the hedgerows were excavated through rows 4. 5, and 6 (Figure 5-1), where there happened to be three trenched plots adjacent to three nontrenched plots. Work began on 1 3 January 1 996. 1 05 days after the maize harvest. The plots were weeded three times during that period to keep out roots of grasses and weeds. Thirty sections 1 .2 m wide by 1 m deep parallel to the trenched hedgerows and an equal number parallel to the nontrenched hedgerows were mapped, at 20, 50, 100,

PAGE 71

56 150. and 200 cm distances. A Plexiglas plate inscribed with a 10 by 10 cm grid was pinned to each cut surface. The positions of the roots were marked on plastic sheets clipped over the Plexiglas plate, a separate symbol was used for each of four diameter classes: very fine (10 mm). Statistical analysis Statistical analysis was done using the SPSS version 9.0 univariate GLM ANOVA procedure. The model used for the dependent variables Leucaena leaf and small stem biomass and height above the stump was as follows: y,jki = /« + block ,+tj + /, + t*f^, + error, ^„ where: y=dependent variable. t=trenching, f=fertilizer /=2, 3, 4. 5, 6, 7, 8. 9. 10 7=0, 1 where O^not trenched and l=trenched ^0, 1 where 0=no fertilizer applied andl=fertilizer applied For each kind of field measurement taken at the four fixed distances and/or the three soil depths, three levels of analysis were done — an overall ANOVA (referred to as the complete model throughout the text) that included all the main effects and their interactions, ANOVAs done separately for trenched and nontrenched plots that included all other effects, and ANOVAS done separately for each distance from the hedgerows that included only trenching and fertilization factors and their interaction. For percent soil water, separate ANOVAs were done for each combination of distance from the hedgerows with each of the three depths (0-15 cm, 15-30 cm, 30-45 cm). If there were significant trenching by fertilization interactions produced by the ANOVAs done separately for each distance and depth, contrast statements were run to determine at which levels the interactions were significantly different.

PAGE 72

57 The following example shows the complete ANOVA model used for the dependent variables: maize biomass taken during the growing season, maize height. maize leaf water pressure, maize grain at final harvest, number of maize plants at final harvest, and number of maize ears at final harvest. y.jkin. =M + block ,+t, + J\+ t*f,, +d,+ t*d ,, + fd ,,+ t*fd ,,, + e/ro/-,,,,,, where: >'=dependent variable. t=trenching. f=fertilizer. d=distance /=2, 3. 5, 6. 8. 9 (blocks used varied for each dependent variable) 7=0. 1 where O=not trenched and 1 =trenched k=0. 1 where 0=no fertilizer applied andl=fertilizer applied /=50. 100. 150. 200 cm distance from hedgerows The overall mean square error was used to test the effect of distance and its interactions. However, because measurements were taken at fixed distances and depths, giving a split-plot characteristic to the experimental design, an "error a" term was computed by adding sums of squares for the interactions of blocks with trenching and fertilizer. This was used to test the effects of trenching and fertilizer and their interaction. The complete model for the percent soil water ANOVAs was: y „,,„.„ =f^ + block, + tj + /, + /%, + d, +t*d^, + rd„ +tTd,„ + dp,,, + 1 * dp^„, + f * dp,,,, + d * dp,,,, +t*f* dp^„„ + error,^,,,,,,, where: y=dependent variable, t^trenching, f=fertilizer, d^distance /=3. 6. 9 7=0, 1 where O=not trenched and 1 =trenched ^0. 1 where 0=no fertilizer applied andl=fertilizer applied /=50, 100, 150, 200 cm distance from hedgerows A77=0-15. 15-30, 30-45 cm depth from soil surface

PAGE 73

58 In addition to the "error a" term computed and applied as above, for the soil water ANOVAs an "error b" term was computed to test the effects of distance and its interactions with trenching and fertilizer. Sums of squares for the interactions of blocks with distance and its interactions with trenching and fertilizer were added to compute this term. The overall error term was used to test the effects of depth and its interactions. "Error a" and "error b" and the corresponding F values were computed by hand; SAS version 7.0 was used to compute the appropriate p-values. Soil water was expressed as a percent, however the range of values was narrow (13% to 39%) and not extremely high or low, so an arcsin transformation to stabilize the variances was not applied before running the ANOVAs (Ott 1992). Not all of the interactions that resulted in significant tests were meaningful due to the very low numerical values of the differences. P-values for all factors and their interactions are shown in Appendix C for all of the ANOVAs and contrasts. Paired t-tests were used to detect significant differences in the numbers of root intersections between trenched and nontrenched plots. This was possible because each pair of measurements at each distance and depth were taken from physically adjacent positions. Results and Discussion Results for each maize season are presented separately in this section, followed by the results of the root distribution study and a general discussion. Spring 1994 season Two factors made this season differ from the two subsequent ones — the 30 cm root barriers installed in the trenched plots in 1991 apparently no longer presented an effective obstacle to the passage of Leucaena roots from the hedgerow into the alleys, and the rainfall was poorly distributed.

PAGE 74

59 Rainfall was 206 mm for the growing season and there was good soil water at planting (51 mm fell in the 7 days prior to planting). However, most of the rain during the growing season fell between 11 and 21 DAS (154 mm), and then no rain fell until 67 DAS (23 mm), which was the last effective rainfall event before final harvest. The plants had reached about 20 cm height by 14 DAS. At 35 DAS the height was about 1 m. but the soil was dry and cracking. Maize was always shorter than Leucaena until the 33 DAS hedgerow pruning; at that time the maize was about one meter high and the hedgerows 50 cm. Maize tasseling began about 42 DAS, with most of the plants in anthesis or silk by 50 DAS. The plants showed increasing signs of stress as the season progressed. Many of the plants had fired leaf margins, and soil cracks opened to about 2.5 cm after 60 DAS. Maize biomass production . Aboveground biomass production during the growing season was not affected significantly by trenching or by fertilizer application, but the amount of biomass did differ with distance from the hedgerows. The four measurements of maize biomass are shown together in Figure 5-6. o .N 'is E 300 250 o It) — > « (/) E = .2 -^ "a u> c 100 £ r o < 200 150 50 42 49 63 Days after sowing 70 ^50 cm 100 cm B 150 cm D 200 cm Figure 5-6: Maize biomass yields at 50, 100, 1 50. and 200 cm from the hedgerows at four times after sowing during the spring 1994 season

PAGE 75

60 For the ANOVA model that included distance, there were no significant effects of fertilization (a=.05 or a=.10) on the weight of maize biomass, and only one case where trenching was significant (p = .085) at 63 DAS. P-values for the effect of distance on maize biomass were <.001 at all four observations. The weight of maize biomass produced was always significantly lower at 50 cm from the hedgerows. It appeared to peak at 150 cm for the first three observation periods, although that distance was significantly greater than the others only at 49 DAS (Figure 5-6). At 70 DAS. the weight of biomass peaked at 200 cm. Competition between the maize and the trees clearly extended to 1 00 cm at 42 DAS, and to 50 cm thereafter. When the analyses were done separately at each distance, the effect of trenching was significant only at 50 cm from the hedgerows at 42 DAS and at 63 DAS (Table 5-2). In these two cases, the amount of maize biomass produced in trenched plots was at least double that produced in nontrenched plots. Table 5-2: Effect of trenching on the production of maize biomass 50 cm from the hedgerows at 42 and 63 DAS; spring 1994. 42 DAS 63 DAS Trenched biomass (g) 50 93 Nontrenched biomass (g) 25 37 p-value ^39 .039 There were also four instances when the effect of fertilizer was significant. More maize biomass was produced in the nonfertilized (F-) plots at 49 DAS at both the 50 cm and 100 cm distance from the hedgerows (Table 5-3). At both 42 and 49 DAS there was a significant interaction of fertilizer and trenching at the 150 cm distance from the hedgerows, in both cases more maize biomass was produced in trenched (T+) plots

PAGE 76

61 (Table 5-3). However, at 42 DAS this effect was seen only between trenched and nontrenched plots that were also fertilized, while at 49 DAS the effect was seen only between plots that were not fertilized. Table 5-3: Effect of fertilizer on the production of maize biomass 50. 100. and 150 cm from the hedgerows at 42 and 49 DAS: spring 1994.

PAGE 77

62 Soil water content . This was measured twice during this season, at 56 and 70 DAS. The ANOVAs that included both distance and depth in the model did not reveal any differences in soil water due to trenching, fertilization, or distance from the hedgerows. There were differences among depths below the surface, as shown in Table 5-4. At 70 DAS the surface 15 cm had more soil water than the deeper layers, probably because 23 mm of rain fell just three days before the measurement. This was the first rain since 21 DAS. At 56 DAS the water content of the soil increased steadily with depth. The water content of the 30-45 cm layer was the same for both observations. Table 5-4: Soil water at three depths below the surface at 56 and 70 DAS; spring 1994. Soil Water % — Gravimetric

PAGE 78

63 fertilized plots had 21.1% soil water vs. 23.1% for the unfertilized plots (p = .027). At 56 DAS there was a significant interaction at the 100 cm distance from the hedgerows: at the 0-15 cm depth trenched plots had less soil water than nontrenched plots where no fertilizer was applied (p = .067): the reverse was true in the fertilized plots (p = .034). The meaning of these interactions is not clear, since fertilizer was not significant as a main effect. P-values for all interactions are listed in Appendix C. but not all will be discussed as they are too numerous and add little to the understanding of the treatment effects. Maize vield . Grain yield for this season was very low. with no significant differences due to trenching or fertilizer. A per-hectare equivalent corresponding to the average values per six-meters as shown in Figure 5-8 is 182 kg/ha maize grain, with only 5.2 kg/ha of the total contributed by the 50 cm rows. As with maize biomass and sexual development, grain yield was affected only by distance from the hedgerows. In addition, the peak production in grain yield and number of ears appeared at 150 cm from the hedgerows, then declined slightly at 200 cm. This is also similar to the pattern observed in biomass production and sexual development, with a decline between 150 cm and 200 cm. although it was only statistically significant at 49 and 56 DAS. Apparently, competition between maize and hedgerows affected the survival of maize plants differently than grain yield or number of ears. There were significantly fewer surviving plants at harvest time in the row closest to the Leucaena. but in the other three more distant rows maize plant survival was the same, and did not decrease between 150 and 200 cm (Figure 5-9). Full stocking for each distance depicted in Figure 5-9 would have been 16 plants (per 6 m, or per 8 hills), equal to 12,000 plants per hectare at each distance, totaling 48,000 plants per hectare if all had survived. Maize plant survival

PAGE 79

64 with distance from the hedgerow was 48%. 84%. 87%. and 80% at 50 cm. 100 cm. 150 cm. and 200 cm from the hedgerows, respectively. Figure 5-8: Yield of maize grain per six meters at four distances from the hedgerows;

PAGE 80

65 Since the number of ears did not differ significantly between 100 and 150 cm (Figure 5-10) and the grain yield did differ, the implication is that grain yield per ear was lower at 100 cm than at 150 cm. This was the case, with the grain weight (g) per ear for the four distances being, respectively, 3.5, 10.8, 13.0. and 9.7. S 6 5 1 4 I 3 ^ 2 1 be

PAGE 81

66 Leucaena biomass-production and height . The four Leucaena harvests discussed in conjunction with the spring 1994 maize season are those made after my post-coz/p d'etat-XQinm to Haiti, but before the hedgerows were re-trenched prior to the fall 1994 maize season. The 20 November 1 993 harvest was the first re-growlh after the overgrown trees were cut back to 50 cm stumps, but this was before the first application of fertilizer to a maize crop so the F+ and Fdesignations are not meaningful for this harvest. Table 5-6 shows the weight of biomass harvested and Table 5-7 shows the height above the stump for the four treatment combinations. The 33 and 65 DAS harvests were taken during the spring 1 994 maize season. Table 5-6: Stem and leaf biomass harvests and daily growth increments of^ Leucaena hedgerows 33 and 65 DAS; Spring 1994. 20 Nov 93 2Jun94 4 Jul 94 20 Aug 94 Biomass Increment Biomass Increment Biomass Increment Biomass Increment (kg/m) (g/m/d) (kg/m) (g/m/d) (kg/m) (g/m/d) (kg/m) (g/m/d) T+F+ T+FT-F+ T-F1.2 1.4 1.2 17.8 18.3 21.7 18.9 0.9 0.9 1.0 0.9 12.4 12.6 14.1 12.5 0.4 0.3 0.4 0.4 11.5 10.4 12.2 11.3 0.5 0.4 0.5 0.5 10.0 9.1 10.2 10.9 p-value T

PAGE 82

67 production in the trenched plots (p = .074). Trenched plots produced 1.15 kg/m. while nontrenched plots produced 1.3 kg/m. A five-meter spacing means there are 2.000 meters of hedgerow per hectare, so this amounts to a reduction of 300 kg/ha due to trenching. There was no significant trenching effect in the subsequent three harvests, nor did fertilizer application to the maize influence Leucaena production. The reduction in Leucaena daily growth rate beginning in July and continuing in August was probably a response to less rainfall during that period. Table 5-7: Height above a 50 cm stump and daily growth increments oi Leucaena hedgerows prior to the second root trenching. 33 and 65 DAS; Spring 1994. 20 Nov 93 2 Jun 94 4 Jul 94 20 Aug 94 Height Increment Height Increment Height Increment Height Increment (m) (cm/d) (m) (cm/d) (m) (cm/d) (m) (cm/d) 1.8 1.7 T+F+

PAGE 83

68 the first fertilizer application to the maize crop. The significant (p = .095) effect of trenching on 4 July is associated with a verv' small numerical height advantage in the trenched, nonfertilized plots. It is interesting to note that while daily Leucaena biomass increment decreased after 2 June (Table 5-7). daily height growth rate increased. The July and August cutting intervals were shorter than the previous two. so this may mean that Leucaena puts on stem height first, then adds stem diameter, leaf biomass. or newstems later in the cycle. Fall 1 994 season The fall 1994 season was the best of the three in terms of maize production, because it had the highest rainfall and the best rainfall distribution through the growing season. After the Leucaena was pruned at 28 DAS, the maize quickly gained a height advantage over the hedgerow trees and maintained it throughout the rest of the season in both the trenched and nontrenched plots, including the rows closest to the trees. Most of the plants were in flower (tassel and silk) by the time tropical storm Gordon passed at 65 DAS. By 70 DAS the maize was about 3 m tall, taller by about 1 m than the Leucaena hedgerows. Maize biomass production . Maize biomass production was measured only once during the season, at 78 DAS. The ANOVA for the complete model (including distance) was significant for trenching (p = .018) and distance from the hedgerow (p = .023). ANOVAs separate by distance showed that only at the 50 cm distance from the hedgerows was trenching a statistically significant factor (p = .037). At all other distances trenched plots produced more maize biomass than nontrenched plots, but the difference was not significant (Figure 5-11). The complete model also indicates an interaction between trenching and fertilizer at the 150 cm distance (p = .093). There was

PAGE 84

69 significantly more biomass produced in trenched (509 g) than nontrenched plots (307 g) where they were not fertilized, but the difference was not significant in fertilized plots. 600 vt = 500 50 100 150 Distance from the hedgerow 200 Figure 5-11: Trenched and nontrenched plots at 50. 1 00. 1 50, and 200 cm from the hedgerows. 78 DAS: fall 1994. ANOVAs run separately for trenched and nontrenched hedgerows showed that distance was a significant factor only for nontrenched hedgerows, with the 50 cm distance producing less biomass than the other distances. This was not the case for trenched hedgerows, where no significant difference in biomass production was detected over the four distances (Table 5-8). Table 5-8: Maize biomass in trenched and nontrenched plots over four distances from the hedgerows. 78 DAS; fall 1994.

PAGE 85

70 Maize leaf water potential . The complete ANOVA model did not show any significant leaf water potential differences among factor levels for any factor. However, when the ANOVAs were run separately for each distance, maize plants in trenched plots indicated less leaf water pressure than did those in nontrenched plots (p = .081) at the 50 cm distance from the hedgerows (Figure 5-12). Differences were not significant at the other distances. nT+ 50 100 150 200 Distance from the hedgerow (cm) Figure 5-12: Maize leaf water potential in trenched and nontrenched plots at four distances from the hedgerow, 42 DAS; fall 1994. Similarly, when ANOVAs were run separately for trenched and nontrenched plots, there was no significant difference in maize leaf water potential among the four distances for the nontrenched plot. For the trenched plots, however, the leaf water potential in the 50 cm plots was significantly lower (p = .008) than in the other three distances (Table 5-9).

PAGE 86

71 Table 5-9: Maize leaf water potential in trenched and nontrenched plots over four distances from the hedgerows, 42 DAS; fall 1994. Distance (cm) Trenched (bars) Nontrenched (bars) 50 1.1a 2.0 a 100 2.5 b 1.7 a 150 1.8 b 2.0 a 200 2J_b 2A_a p-value ^08 ^17 Soil water content . Soil water was measured on the same day as maize leaf water potential. Figure 5-13 shows the soil water percents for trenched plots at all combinations of distance and depth. The differences in soil water percents between trenched and nontrenched plots are shown in parentheses. The overall model indicated that trenching (p = .042), distance from the hedgerow (p < .001). and depth below the surface (p < .001) were significantly affecting soil water content. The effect of fertilizer was not significant (p = .352). Separate ANOVAs for each combination of distance and depth were not significant for either trenching or fertilizer in the 0-15 cm depth. There were significant differences due to trenching in the lower depths. In the 15-30 cm layer, soil water percent was higher in trenched plots at the 50 cm (p = .033) and 1 50 cm (p = .095) distances from the hedgerow. In the 30-45 cm layer, soil water percent was higher in trenched plots at the 50 cm (p = .002) and 100 cm (p = .014) distance from the hedgerow. The effect of distance from the hedgerows was significant in trenched plots at the 15-30 cm (p = .004) and the 30-45 cm depths (p < .001). with more soil water present close to the hedgerows. In the surface layer distance was significant for trenched plots at the a=.10 level, with more soil water at the 50 cm distance than the 150 distance (p = .078). In Figure 5-13, cells having the same letter, in a given row, are not significantly

PAGE 87

72 different. Soil water percent in nontrenched plots did not differ significantly with distance from the hedgerow in any of the three depths. MSoil Depth (cm) 0-15 J 15-30 4 30-45 A Distance from Hedgerow (cm)) 50 100 150 27.8 (1.4) a

PAGE 88

73 Table 5-10: Soil water in trenched and nontrenched plots by depth and distance from the hedgerow, 42 DAS; Fall 1994. Distance (cm)

PAGE 89

74 .020). with the nontrenched plots having about one more plant per six-meter sample at that distance. 1000 50 100 150 200 Distance from the hedgerow (cm) Figure 5-14: Maize grain yield in trenched (T+) and nontrenched (T-) plots at four distances from the hedgerow. Fall 1994. Final harvest is presented on a per-hectare basis in Table 5-11. Although there were differences in grain yield between trenched and nontrenched plots at 50 and 1 00 cm from the hedgerows (a=.10, Figure 5-14 above), there was no significant difference in grain, number of plants or number of ears over distance from the hedgerows, for either trenched or nontrenched plots. In general, there were no strong differences in yield parameters this season, probably due to the adequate, well-distributed rainfall partially negating the potential advantages of the root barriers. If only the differences in grain weight at 50 cm and 100 cm are accepted as significant, then trenching made a difference of 252 kg maize

PAGE 90

75 grain/ha, an increase of 12% over the nontrenched plots. However, since the effect of trenching for the complete model was significant (p = .009). then the advantage of trenching increased to 360 kg/ha (2.382 kg-2.022 kg), an increase of 18% over the nontrenched plots. Since there were no differences in the numbers of plants or ears, the advantage of trenching was manifested in more grain weight per ear in trenched plots, especially at the 50 and 100 cm distances. Table 5-11: Maize grain weight, number of plants, and number of ears at four distances from the hedgerows and as a total per hectare, with p-values for the differences between distances. Fall 1 994. Grain (kg) No. plants No. ears Distance (cm) T+ TT+ TT+ T50

PAGE 91

76 relatively slow growth rate between the 8 October and 14 January harvests in spite of the plentiful rainfall was probably due to insect damage (species not identified) in the southeast quarter of the trial site, in blocks 6-10. The damage affected trenched and nontrenched plots equally. Fertilizer application on the adjacent maize did not affect Leucaena biomass or height growth. The significant trenching by fertilization interaction at the 6 May harvest does not appear to be important in magnitude. Table 5-12: Stem and leaf biomass harvests and daily growth increments o^ Leucaena hedgerows after the second root trenching. Fall 1994. 8 Oct 94 14 Jan 95 16 Mar 95 6 May 95 Biomass Increment Biomass Increment Biomass Increment Biomass Increment (kg/m) (g/m/d) (kg/m) (g/m/d) (kg/m) (g/m/d) (kg/m) (g/m/d) f+F+ OJ 6^8 0^4 Is 05 8^5 04 tT! T+F0.3 6.5 0.4 3.6 0.5 8.6 0.3 5.9 T-F+ 0.6 12.0 0.5 5.3 0.7 11.3 0.4 7.3 T-F0.6 12.2 0.7 6.8 0.8 13.0 0.4 8.7 p-value T

PAGE 92

77 Table 5-13: Height above a 50 cm stump and daily growth increments of Leucaena hedgerows after the second root trenching, Fall 1994. 8 Oct. 94

PAGE 93

78 was a nontrenched plot, an inadvertent increase in soil water content in that plot only strengthens the result of the ANOVA that showed a significant difference in maize biomass due to trenching. The effect of trenching on maize development and yield was more pronounced than during the other two seasons. However, even though this was the only season for which the complete amount of fertilizer recommended by the Ministrv' of Agriculture (MARNDR 1990)was applied, there were no significant effects of fertilizer on any of the parameters measured for the complete ANOVA models. All the ANOVAs run using the complete model for maize growth and yield during this season produced very significant pvalues for the effects of trenching, with trenched plots producing more than nontrenched plots, especially close to the hedgerows. All but the two maize biomass harvests were significant for distance from the hedgerow and the trenching by distance interaction (Appendix C). Much of the difference was due to the poor survival of maize plants in the nontrenched plots, especially in the two rows closest to the trees. Maize height . Figure 5-15 shows a significant maize height advantage at 40 DAS in trenched plots at all four distances from the hedgerows (p < .001). Missing plants were assigned values of zero because the mortality was attributed to competition with the hedgerows. There were more missing plants in nontrenched plots than in trenched plots: at the 50 cm distance 84% of the hills in nontrenched plots had no surviving plants vs. 20% for trenched plots; at the 100 cm distance the corresponding numbers were 25% in the nontrenched plots and 4% in the trenched plots.

PAGE 94

79 4 5 40

PAGE 95

80 the hedgerows experienced a shade regime different from that on the south side. No direct measurements of light were done, but maize height at 40 DAS was analyzed for the effect of aspect. Neither the complete model (including distance as a factor) nor the separate models for each distance showed orientation with respect to the hedgerows to be a significant influence on maize height (data not shown). The effect of fertilizer on maize height was significant only at the 50 cm distance from the hedgerows and only in the trenched plots (p = .035), where fertilized maize was 10 cm shorter than nonfertilized maize. Maize biomass production . Differences between maize biomass harvested in trenched and nontrenched plots at 47 and 54 DAS were similar, and continued the pattern of differences in maize height discussed above. However, the statistical tests produced higher p-values than did those for height, probably due to smaller samples sizes and high variability caused by three outliers in block 10 at the 50 and 100 cm distances. The irrigation system leak causing water contamination in blocks 9 and 10, discussed above, was probably responsible. The complete model ANOVA at 47 DAS was significant for trenching (p = .035), but not for fertilizer or distance from the hedgerow. The 54 DAS complete model was significant only for trenching at a=.10 (p = .085). At 47 DAS, ANOVAs done separately for each distance showed that trenched plots produce significantly more biomass at all four distances from the hedgerows (Figure 5-16). P-values for trenching are = .030, .096, .016, and .017 for the four distances. At 54 DAS. the difference due to trenching is significant only at the 50 cm distance (p = .072). The difference between trenched and nontrenched plots at the 100 cm distance shown in Figure 5-17 was numerically large, but because of the outliers discussed above, did not result in a significant difference.

PAGE 96

81 0)

PAGE 97

82 The changes in biomass over distance were similar for both observation dates. In neither case were there significant differences over distance for trenched plots, hi both cases, maize biomass in nontrenched plots was less in the two rows closest to the hedgerows than in the more distant rows (Table 5-15). Table 5-15: Maize biomass in trenched and nontrenched plots at four distances from the hedgerow, 47 and 54 DAS; Spring 1995. 47 DAS 54 DAS Distance (cm) T+ (g) T(g) T+ (g) T(g) 50

PAGE 98

83 3

PAGE 99

84 As was the case with maize biomass production, growth stage changed w ith distance from the hedgerows differently for plants in trenched plots than for those in nontrenched plots. Plants in trenched plots were most advanced at 1 00 cm from the hedgerows, and identical at the other three distances. There was no retardation in the 50 cm row. and conditions for maize development appear to have been enhanced at 1 00 cm due to the presence of hedgerows. Maize plants in the nontrenched plots were least developed in the 50 cm row. and progressively more developed as distance increased. Soil water content . The complete ANOVA models for soil water at 26 and 61 DAS did not produce significant p-values for trenching, fertilizer, or distance from the hedgerow. Only depth was significant (p < .001) on both dates. ANOVAs run separately for each distance and depth were also not significant for trenching or fertilizer, with the exception of a significant p-value (.016) for fertilizer at 150 cm distance. 15-30 cm depth at 26 DAS. The nonfertilized plots in that position had only 0.5% more soil water than did fertilized plots, so this difference is probably meaningless. Figures 5-20 and 5-21 show the percent soil water for the various positions in the trenched plots, with the difference between trenched and nontrenched plots in parentheses. None of these differences were significant. When ANOVAs were run separately for trenched and nontrenched plots for each depth, there were no significant differences in soil water between distances from the hedgerow for either trenched or nontrenched plots, with one exception. At 61 DAS in the nontrenched plots in the 0-1 5 cm depth, the 50 cm distance had about 1 .5% more soil water than the other three more distant positions (p = .002). Soil water increased with depth for both trenched and nontrenched plots at all distances from the hedgerows, at both observation periods (Table 5-16).

PAGE 100

85

PAGE 101

86 Table 5-16: Soil water in trenched and nontrenched plots by depth and distance from the hedgerow. 26 and 61 DAS: Spring 1995. Distance (cm)

PAGE 102

87 effect of fertilizer on number of plants (p = .074) and number of ears (p = .068) at the 50 cm distance only. In both cases, the fertilized plots had lower numbers, which would indicate a depressing effect of fertilizer of 1 ,750 plants per hectare and of 2,250 ears per hectare due to the reduction in the 50 cm rows. 50 100 150 Distance from the hedgerow (cm) DT+ 200 Figure 5-22: Maize grain weight from trenched and nontrenched plots at four distances from the hedgerows. 131 DAS; Spring 1995. ANOVAs run separately for trenched and nontrenched plots showed that distance from the hedgerow significantly influenced yield in both cases (Table 5-17). In the nontrenched plots, yield in the 50 cm rows was less than in the more distant rows, but there were no significant differences in yield among the rows at 100, 150, and 200 cm distances. In both nontrenched and trenched plots, the numerical maxima occurred at 100 cm. then decreased with distance, but this pattern was significant only in trenched plots. Grain weight, number of plants, and number of ears produced in the trenched plot 100 cm rows were always significantly greater than those produced in the 200 cm rows.

PAGE 103

88 nT+ 50 100 150 200 Distance from the hedgerow (cm) Figure 5-23: Number of maize plants from trenched and nontrenclied plots at four distances from the hedgerows, 131 DAS; Spring 1995. 16 14 I 12 s. n 10 o •o 8 I 4 50 a a " gnT+ 100 150 Distance from the hedgerow (cm) 200 Figure 5-24: Number of ears from trenched and nontrenched plots at four distances from the hedgerows. 131 DAS; Spring 1995.

PAGE 104

89 Table 5-17: Maize grain weight, number of plants, and number of ears at four distances from the hedgerows and as a total per hectare, with p-values for the differences between distances; Spring 1995. Grain (kg/ha) No. plants/ha No. ears/ha Distance (cm) T+ T^^ T+ TT+ T50 249.0 ab 25.3 a 6.750 ab 1.375 a 8.188 a 1.375 a 100 284.3 a 180.1b 7.875 b 5,938 b 11,000 b 7.625 b 150 187.7 be 146.7 b 6.625 ab 5.250 b 8.500 a 7.313 b 200 158.1c 144.0 b 5.125 a 5.313 b 6.688 a 6.43 8 b Total/ha 879 496 26.375 17.875 34,375 22.750 P-value .019 <001 .040 <.001 .007 <.001 The quantity of maize produced during this season was intermediate between the spring 1994 and fall 1994 seasons, however the influence of trenching on maize production was stronger. If the increased grain weight is considered only in the 50 and 100 cm rows, where there were significant differences between trenching treatments, then the increased maize grain due to trenching was 328 kg/ha. or a 66% increase over nontrenched plots. If the differences at all four distances are considered the advantage is 383 kg/ha, a 77% increase over nontrenched plots. The difference was due to higher maize mortality, lower grain weight per plant, and lower grain weight per ear in nontrenched plots, especially in the 50 cm row but extending through 100 cm. Leucaena biomass — production and height While the effect of trenching on maize production was very strong during the last season, its effects on Leucaena growth appears to have been mitigated by root re-growth for the final three harvests. There were no significant effects of trenching or fertilizer on Leucaena small stem and leaf biomass (Table 5-18). Leucaena height does show a

PAGE 105

90 significant pvalue for the effect of trenching at the 57 DAS harvest, with hedgerow trees in trenched plots being about 25 cm taller than those in nontrenched plots (Table 5-19). Table 5-18: Stem and leaf biomass harvests and daily growth increments of Leucaena hedgerows at 19, 57, and 130 DAS; Spring 1995. lOJun.95

PAGE 106

91 Leucaena root distribution in trenched and nontrenched plots Figures 5-25, 5-26, 5-27, and 5-28 siiow distinct differences in the number of root intersections found in trenched plots and nontrenched plots 1 05 days after the final maize harvest. Where the numbers in parentheses are positive, trenched plots had that many more intersections than nontrenched plots; where the numbers are negative, they had that many less intersections than nontrenched plots. Taking only the positions where the differences were statistically different (indicated in the figures by solid or dashed lines around the cells), a pattern can be observed. Trenched plots have more large (>10 mm) roots 20 cm from the hedgerows in the 90-100 cm layer; more medium (5-10 mm) roots 20 cm from the hedgerows in the 40-60 cm layer; more fine (2-5 mm) roots 50 cm from the hedgerows in the 40-50 cm layer; and more very fine (<2 mm) roots 200 cm from the hedgerows in the 30-60 cm layer. It appears as though, because of the barriers, roots in trenched plots have developed below the barriers and produced more very fine roots at the 200 cm distance. Trenched plots have fewer large and medium roots 20 to 50 cm from the hedgerows in the 0-30 cm layer, fewer medium roots 100 to 200 cm from the hedgerows in the 20-30 cm layer, fewer fine roots 20 to 50 cm from the hedgerows in 0-10 cm layer, 100 cm from the hedgerows in the 20-70 cm layer, and 150 cm from the hedgerows in the 30-40 cm layer; and fewer very fine roots 100 cm from the hedgerows in the 30-40 cm layer. Root barriers have clearly reduced the numbers of root intersections of all diameter classes, especially near the hedgerows and in the surface layers of soil through 150 cm from the hedgerows, but particularly at 100 cm from the hedgerows.

PAGE 107

92 ^

PAGE 108

93 &&!^

PAGE 109

94 General Discussion Summary bv season Spring 1994. This was a season of poor rainfall and poorly functioning root barriers. The root barriers were three years old at the time the maize was planted. In addition, the root systems of the Leucaena hedgerows were allowed to develop unhindered during a two-year period from 1991 to 1 993 when the tops were not pruned, allowing Leucaena roots to re-grow into the alleys over and under the old 30 cm plastic barriers. Based on field observations of soil cracks and leaf wilt, maize plants were under water stress during the flowering period. There was no rainfall from 21 DAS until 67 DAS. Drought stress was severe during flowering, which began about 42 DAS. The soil water content of 22% to 24% from 1 5 cm to 45 cm depth taken at 56 and 70 DAS correspond to leaf water potentials of 20 bars to 12 bars in this soil (Appendix A). Maize is severely stressed at 16 to 18 bars leaf water potential (Herrero and Johnson 1981). Drought during flowering causes severe decrease in maize yield (de Geus 1973). There were no significant effects of trenching or fertilization on maize growth and yield, or on soil water content. A residual effect of the old root barriers was seen only in the maize biomass harvests at 42 DAS and 63 DAS, when more maize biomass was produced in trenched plots at 50 cm from the hedgerows. Distance from the hedgerow was significant for all maize measurements and soil water content. Maize growth (biomass taken during the season, growth stage, grain weight, number of plants, number of ears) was always lowest at the 50 cm distance and most commonly highest at the 150 cm distance. In three cases (the 49 DAS biomass, grain weight, and number of ears), there was a decline from the 150 cm distance to the 200 cm distance. Overall maize grain yield was only 1 82 kg/ha. There may have been some competition for light between

PAGE 110

95 hedgerows and trees since the Leucaena were taller than the maize until 33 DAS when the hedgerows were pruned, but the principal limitation on maize growth this season was water. Neither trenching nor fertilizer application in the maize plots had any effect on Leucaena biomass or height. Fall 1994 . This season saw very good rainfall and newly re-established root barriers to 50 cm depth. Maize was taller than the Leucaena after the 28 DAS hedgerow pruning, most of the maize was in tassel and silk at 65 DAS. At 42 DAS. maize leaf water potential was less in trenched than in nontrenched plots at the 50 cm distance, and there was more soil water close to hedgerows in trenched plots. The maize was able to take advantage of the extra soil water, because at 78 DAS more maize biomass was produced in trenched plots than in nontrenched plots at the 50 cm distance. In the nontrenched plots, maize biomass at the 50 cm distance was significantly less than that produced farther from the trees, but there was no difference by distance in the trenched plots. Since both trenched and nontrenched plots experienced the same shade conditions, this difference in biomass production was apparently due to water competition from the hedgerows. The increased grain yields at the 50 and 100 cm distance in trenched plots correspond to the other measurements made earlier in the growing season. Differences in maize yield over distance were muted by the abundant rainfall. Although the maize biomass in nontrenched plots at 78 DAS did produce the lowest value at 50 cm from the hedgerows, there was no difference in grain weight, number of plants, or number of ears between distances from the hedgerow for either trenched or nontrenched plots. Grain yield in trenched plots was 2,382 kg/ha. an increase of 18% over the nontrenched plots. Some of these findings are consistent with reduced numbers of roots in trenched plots close to the hedgerows. Soil water percent was highest at 50 cm from the hedgerows and

PAGE 111

96 lowest 1 50-200 cm from the hedgerows in the trenched plots, corresponding to lowest maize leaf water pressure at the 50 cm distance (both measurements taken at 42 DAS). Maize biomass at 78 DAS and grain weight at final harvest were both greater in trenched plots than in nontrenched plots at the 50 cm distance. The recent re-trenching reduced Lencaena biomass production by 1,561 kg/ha over the four pruning operations carried out during this period compared to nontrenched hedgerows. Spring 1995 . This season's rainfall was about the same as that of spring 1 994, but the distribution was much better. Maize was taller than Leucaena only after the 57 DAS pruning, and there was poor maize survival in the 50 cm row of the nontrenched plots. The root barriers were apparently effective. All maize growth and yield parameters were significantly greater in trenched plots than in nontrenched plots at the 50 cm distance from the hedgerows, and all but one at the 1 00 cm distance. In addition, all growth and yield measurements in nontrenched plots were lowest at the 50 cm distance and peaked at the 150-200 cm distance. In the trenched plots, the highest values were most often found at the 100 cm distance, with a decline at the 150 or 200 cm distance. Grain yield was 879 kg/ha in the trenched plots, an increase of 77% over the nontrenched plots. These findings are consistent with the decreased number of root intersections at the 20-100 cm distance and the increased number of very fine root intersections at the 200 cm distance in trenched plots. Soil water content measured at 21 and 61 DAS was not significantly different between trenching treatments or by distance from the hedgerows, but only by depth. Trenched and nontrenched plots both increased in soil water content with depth, but the amount of soil water present did not represent a condition of severe water stress for the maize plants, according to the relationship established in the pot study (Appendix A). A soil water percent above 25% would be approximately equivalent to 10 bars of leaf water pressure, above which maize begins to

PAGE 112

97 experience water stress (Herrero and Johnson 1981). At soil depths below 15 cm, all positions had more than that amount of water at both observations. The effects of trenching on Leucaena biomass production were no longer apparent during this season. Impact of trenching on Lencaena Trenching has a short-term impact on Leucaena growth, primarily on small stem and leaf biomass and to a much lesser extent on height. Figure 5-29 shows the results of 15 harvests from 1991 through 1995. The initial trenching to 30 cm was done in April 1991, just before the first harvest shown in Figure 5-29. Trenched plots always produced less biomass (a=.05) than nontrenched plots during the four subsequent harvests in 1991, for a total loss of 940 kg during seven months. Since no Leucaena harvests were taken for the next two years, it is not possible to know how long the effect of the first trenching lasted. In September 1993, after a prolonged absence from the site, the overgrown hedgerow trees (many with a diameter >10 cm at 10 cm from ground level) were cut back to 50 cm stumps. The next four harvests produced no significant differences in Leucaena biomass between trenched and nontrenched plots, presumably because the tree roots in the trenched plots had re-grown into a new configuration that exploited a sufficient soil volume. The plots were re-trenched to 50 cm in late August 1994. The loss of small stem and leaf biomass due to the effect of trenching over the next four harvests during a seven-month period was 1,561 kg per hectare (a=.05). The only time trenching affected height growth oi Leucaena was during the first two harvests after the second trenching, when stems cut from trenched plots were about 1 cm shorter than those cut from nontrenched plots. By June 1995, the trenched trees had again recovered and there were no significant differences during the final three harvests.

PAGE 113

98 Overgrown hedgerows cut back >• ^ Harvest Date Figure 5-29: Leucaena small stem and leaf biomass from trenched and nontrenched plots, May 1991 through September 1995. Conclusions Installation of root barriers appeared to mitigate soil water competition between Leucaena hedgerows and adjacent maize under certain circumstances, thereby increasing grain yield compared with nontrenched hedgerows. Installation of the barriers also decreased Leucaena biomass production. However, the effects of barriers on hedgerow biomass production and on maize grain yield were temporary. Amount and distribution of rainfall and the time elapsed since the installation of root barriers influenced maize growth and yield and the relative impact of trenching during the three seasons discussed here. Root barriers were not effective during the fall 1994 season, and there were no differences in maize or Leucaena yield. Adequate rainfall, as seen in the fall 1994 season, tended to dampen the advantage gained by trenching. Trenched plots yielded an extra 360 kg/ha maize grain over nontrenched plots that season, an increase of 1 8%; Leucaena biomass yield decreased by 1 65 1 kg/ha for

PAGE 114

99 that period. During the final season having lower rainfall, but still effective barriers, maize yield in trenched plots increased by 383 kg/ha (77%) over nontrenched plots. There were no longer any Leucaena yield losses due to trenching. At Machakos. Kenya, Ong and Leakey (1999) noted that when rainfall was below 250 mm. maize and hedgerow trees {Gliricidia sepium and Senna spectabilis) competed with maize for soil water, but when rainfall exceeded 650 mm the trees and crop used water from the same profile without decreasing crop yield. Maize plants in nontrenched plots were fewer in number, shorter, produced fewer ears, and produced less grain weight per ear than maize in trenched plots. This is consistent with the effect of water stress during flowering (Olson 1988). Corlett et al. (1992a) found that flowering of millet next to nontrenched Leucaena hedgerow was delayed compared to sole cropped millet. Maize yield was usually depressed closest to the hedgerows, a phenomenon noted in several studies (Govindarajan et al. 1996, Kamasho 1994, Mugendi et al. 1999). The causes of maize yield depression, or lack thereof, cannot be determined with certainty, but were probably due mainly to increased soil water available near trenched hedgerows. The pruning height of Leucaena was the same for both trenched and nontrenched plots, hedge prunings were removed from both trenching treatments, and fertilizer application was the same for both. However, increased soil water availability also means increased nutrient supply, which promotes increased height growth near hedgerows resulting in improved access to light. Yield depression still occurred in the 50 cm row in the trenched hedgerows, so shading probably influenced yield to some degree at that distance. Maize yield increased steadily with distance from hedgerows in nontrenched plots, but in the trenched plots it increased with distance to 100 or 150 cm. then

PAGE 115

100 decreased at the 200 cm distance. This could be partially explained by the re-distribution of Leucaena roots in the alleys in trenched plots, which were fewer in number than those in nontrenched plots through 150 cm, then increased at the 200 cm distance. Ong et al. (1991) found that barriers next to hedgerows changed Leucaena root distribution in a similar way. The results of this experiment confirm that Leucaena hedgerows compete for soil water to the detriment of adjacent maize and cast doubt on the utility of root barriers as a management practice. Their beneficial effect is temporary, and the additional maize grain realized would probably not pay for the cost of installing the barriers. If farmers perceive soil water competition to be a major problem it would be worthwhile to look for a cost efficient replacement for barriers or a less competitive hedgerow species. These issues are discussed in Chapter 6. On-Farm Studv: Maize Growth at Varying Distances from Hedgerows Introduction and Objectives The conclusions drawn from the on-station studies in the previous section of this chapter are that Leucaena trees in hedgerows reduce grain yield in maize growing in the adjacent one meter area, that permanent root barriers can temporarily mitigate the loss of maize grain yield by reducing soil water competition between the hedges and the maize, but that within a couple of years Leucaena roots will grow under the barriers and eliminate their favorable effect on maize yield. An appropriate next question would be are these results of any use to Haitian farmers, whose plots are mainly on steeply sloping hillsides? This is both a technical and a socioeconomic question; the socioeconomic issues will be discussed in Chapter 6. Before that, the present section of this chapter will address mainly the technical ones and serve as a bridge between the on-station and on-farm work.

PAGE 116

101 The objective of the research supporting this section is limited to showing the development of maize growth and yield on some sloping hedgerow gardens built by Haitian farmers participating in the PLUS project. The logical design to complement the on-station work should have been to install barriers on hillside farms between hedgerows and adjacent maize to investigate whether this practice would economically increase maize yield to farmers. However, such an attempt was not successful.' Instead, data were collected from an onfarm trial comparing three different types of soil conservation structures with traditional cultivation to examine how competitive interactions between trees and crops differ between vegetative and nonvegetative soil conservation practices on sloping land. The objective was to determine if plant development was affected in maize rows adjacent to soil conservation structures on sloping land, in plots where farmers were doing the cultivation. Studv Site The four farm plots, referred to as gardens (jaden), from which the measurements were taken are located in the lower central plateau near the town of Lascahobas, about 55 km north and east of Port au Prince and 23 km west of the border between Haiti and the Dominican Republic. These gardens had been part of a larger national study being carried out by SECID and PADF under the PLUS project to compare yields of crops grown in various kinds of soil conservation structures. The national study was abandoned in 1 996 because of funding problems, but the four gardens in Lascahobas still contained the treatments intact in spring 1997 when this research was done. The soils are The research as originally planned included a series ofLeucaena hedgerow/maize water competition experiments on 30 farms in the south of Haiti as a complement to the on-station trials. This proved infeasible due to political and logistic problems, and was terminated. However, a group of eleven farms in the south of Haiti participated in preliminary work in the spring of 1991 and some of the information gathered from it is used in this section.

PAGE 117

102 calcareous, with slopes ranging between 20 to 45 percent. Three of the plots were one hectare in size; the fourth was 1.3 hectares. The treated portion of the gardens occupied between 12 to 25 percent of the total garden areas. Rainfall in this area is bimodal, falling mainly in spring and fall. In 1997, rainfall during May through August at the nearest rain gauge was 723 mm, with rain falling on 48 days during that period. Methods The study was a complete randomized block design with three kinds of soil conservation structures as treatments, plus a control without soil conservation. The subplot treatments were Leucaena hedgerows: crop bands constructed of sugarcane, pineapple, and cassava; and rock wall barriers. The control plot had no soil conservation structures, and represented traditional maize cultivation practice for this area. The three soil conservation structures and control plot occupied adjacent subplots, assigned randomly. Each of the four gardens was considered a block. The gardens were rented from the owners, and PADF technicians supervised installation of the structures and collected all data. The owners planted and managed crops in the alleys and kept the harvest. Treated subplots contained four rows built on the contour in April 1 993 at a distance of 10 m between rows, each row being 10 m in length. In two of the gardens the contour rows were oriented in north-south direction, and in the other two in east-west direction. A local maize variety was planted in mid-May 1997. The Leucaena hedgerows were pruned to a height of 50 cm on 28 April, two weeks before the maize was planted. It was not pruned again before final maize harvest, when it had attained a height of about 1 .5 to 2 m. The sugarcane was the tallest component of the crop bands, and remained at a height of about 1 .5 m during the maize growing season. Observations were made on the growth stage of the maize on four dates during the growing season: 43, 49, 56, and 67 DAS. Each stage of growth was assigned a

PAGE 118

103 numerical value as follows: vegetative = 1, tassel = 2, anthesis = 3, silk = 4, and ear = 5. A line transect of observations was taken through the three alleys, perpendicular to the contour rows, from top to bottom at the midpoint of each subplot. One maize plant was observed at specific positions along each transect: 1 m below the uphill structures, midway between structures, and 1 m above the downhill structures. Although the control subplots had no soil conservation structures, maize growth stage observations were taken at corresponding positions to those in the treated subplots. Final harvest was taken in mid-September, about 1 16 DAS. Whole maize plants were harvested in transects 5 m wide by 25, 26. or 28 m long through the middle of each subplot passing through, and perpendicular to, three alleys and the corresponding position in the control plot. Technicians did not record final harvest data by position on the slope or within the alleys, as was done for the growth stage observations. Weights were taken in the field on suspended spring scales. All weights were field dry and maize grain was not corrected to a standard moisture percent, as no drying ovens or grain moisture meters were available. Analysis of variance was done using SPSS version 9.0. Results Maize development was different among soil conservation treatments, but not among slope positions or within-alley positions. The maize growing in the rock wall and control plots developed more rapidly than that in the hedgerow and crop band plots at 43 and 49 DAS (Figure 5-30). Maize grown in hedgerows developed less rapidly than that grown in all other plots at 56 and 67 DAS. The difference in growth stage between maize in hedgerows and maize in the control plot increased through 56 DAS, and then decreased between 56 and 67 DAS.

PAGE 119

104 49 56 Days after sowing D Hedgerow DCrop band Rock wall m Control Figure 5-30: Maize development in three soil conservation practices and control at four times during the growing season on four farms during Spring 1997. The differences in the pattern of maize development among soil conservation structures did not result in statistically significant differences in yield (Table 5-22). There were, however, some relatively large numerical differences. There was a wide variation in maize plant density, which was highest in crop band plots and lowest in control plots. Stover weight was greatest in rock wall plots, as was yield of maize grain. Grain yield was least in the control plot. Harvest indexes calculated with field dry weights were: hedgerows, 0.25; crop bands. 0.38; rock walls, 0.30, and control, 0.15. Discussion The pattern of maize development in the alleys between hedgerows and crop bands did not reflect the yield curve caused by soil scouring and hedgerow/crop competition as described by Garrity (1996), even though soil scouring due to hoe tillage is commonly observed in Haiti in the upper parts of hedgerows. Three row-by-row sorghum and maize harvests on sloped farms done as preliminary work by the author in 1991 also produced no clear pattern of yield variations with respect to alley position in Leucaena hedgerows. However, the overall development of maize in the soil

PAGE 120

105 conservation structures built of plants, i.e., hedgerows and crop bands, was slower than that in rock walls and in the control plots. This could be due to competition between hedgerow plants and adjacent maize. The differences in maize development between the four treatments diminished over the growing season. Table 5-20: Maize yield in three soil conservation practices and control at 1 1 6 DAS on four farms, spring 1 997

PAGE 121

106 water. Soil water stress reduces harvest index (HI) in maize (Shaw 1988), and the control plots had the lowest HI. Rock walls produced the highest yield because that treatment harvested water but did not compete for light, nutrients or water with maize. Crop bands and hedgerows were in the middle because they harvested water, but did compete for resources with maize. Better yield from rock wall terraces compared to hedgerows has precedent in Haiti, but gradually the biomass contributed by trees changes the yield ranking in favor of hedgerows. An on-station experiment on 17 to 21% slopes (Isaac et al. 1995) reported that rock wall plots produced greater maize yields than hedgerows in the first year after establishment, equaled the hedgerow treatments during the second year, then yields steadily declined in rock walls but remained steady in hedgerows. Isaac et al. ( 1 996b) reported that during two years and four maize harvests there was a slight increase in yield in all Leucaena hedgerows, and a steady decrease in all other treatments, including rock walls. Increased yield in farmer-managed sloping hedgerow plots has also been shown. Lea (1995e) reported 40% greater sorghum yields in hedgerow plots compared to untreated control plots in experiments done in two regions of Haiti, including farms close to those reported on here. In other regions having different soils and rainfall distribution, hedgerow plots did not show improved yield over traditional cultivation (Lea 1 995e). However, on-farm experimental results can be complicated by management factors outside the intended experimental design. An on-farm study in the same region as the plots in this experiment that reported 70%) higher sorghum yields from rock wall terraces compared to control plots without soil conservation (J.D. Lea, personal communication, Sept. 1995), was followed up by this author. The follow up showed that the sorghum harvest had been measured correctly by technicians, but the farmers who owned the plots had managed the protected plots differently than the control plots, which

PAGE 122

107 were not physically adjacent or always physically the same as the treated plots. It was found that the rock walls served to stop sorghum seed from washing out of the plots during heavy rainfall. These seeds sprouted above the structures and the farmers later transplanted the seedlings back into the alleys, resulting in a higher plant density compared to the control plots. In addition, farmers practiced better, deeper, tillage on rock wall plots in some cases, tended to weed them earlier than control plots and were more likely to thin hills (planted 10 to 15 seeds/hill) after germination in the rock wall plots, resulting in larger panicles. This points out a major difference between farmer-managed onfarm trials and researcher-managed on-farm trials, at least for treatments involving relatively complex inputs such as soil conservation structures. Farmers adjust their management intensity according to their investment in the gardens and the economic potential of the gardens, whereas in researcher-managed trials this difference is eliminated. Conclusions Conclusions based on these data are limited to differences in the pattern of maize plant development among soil conservation treatments. There was a greater variation in the pattern of maize development between type of soil conservation structure than in relative nearness to the structure or to position of the alley on the slope. Lack of competition between plants serving as the structural components of hedgerows {Leucaena) and crop bands (sugarcane and pineapple) is the probable cause of faster maize development in rock wall terraces and untreated control plots. The numerically smaller yield in the control plots compared to the plots having soil conservation structures, although not statistically different, is probably due to the water harvesting effect of the contour structures. A much larger sample size and row by row harvest over several years would be required to draw clearer conclusions regarding the effect of slope position within alleys in plots with such great variability in soil type, soil fertility, soil depth, and rainfall regime.

PAGE 123

CHAPTER 6 LAND AND HOUSEHOLD CHARACTERISTICS IN RELATION TO ADOPTION AND MANAGEMENT OF HEDGEROWS Introduction and Objectives The two previous chapters described research regarding competition between trees in hedgerows and adjacent crops, with a focus on soil water. If soil water competition were an important factor in reducing crop yield in hedgerows, then farmers might consider it in their adoption and management decisions on farm. Results from the on-station research described in Chapter 4 showed that soil water competition between hedgerow trees and maize can reduce maize yield under certain rainfall conditions. The on-farm research in Chapter 5 was not able to show the same phenomenon at work under the highly variable environmental conditions found in hillside farm plots, but did show differences in maize growth between different types of soil conservation structures. Those differences in maize development may have been due to differences in the way soil conservation structures exerted competitive pressures on the crop. The research described in this chapter focuses on how farmers adjust their adoption and management decisions based on the technical aspects of soil conservation structures, the land and human resources available to the adopting household, physical and tenure characteristics of the plot, and farmers perceptions of potential benefits and problems associated with the soil conservation practices. The links between the work described in this chapter and the on-station and on-farm studies are: (1) hedgerows remain the principal focus of the research, and (2) farmers were asked their opinion regarding the severity of 108

PAGE 124

109 tree/crop water competition in hedgerows to explore whether the degree of that perception was correlated with management practices. However, the adoption and management questions explored here included farmers' responses to other soil conservation structures as well as hedgerows, and this proved to be useful in revealing differences in farmers' decision-making strategies. The hypotheses tested in this chapter are • Farmer household characteristics and farm resources influence the adoption and management of hedgerow intercropping. • Plot characteristics influence the adoption and management of hedgerow intercropping. • Problems and benefits with hedgerow intercropping as perceived by the adopting farmer influence management quality. This study seeks to discover relationships between farmers' decision to install an agro forestry intervention and three groups of variables: (1) characteristics of the technology, represented by the decision to install either hedgerows or other agroforestry practices (crop bands, rock walls, gully plugs, or trees) with different characteristics; (2) farm household characteristics representing wealth, education, and household size; and (3) physical characteristics of the plot and the mode of access through which the farmer worked it. Having installed a particular practice on a plot, some measures of management quality of the soil conservation practices are compared to the same three groups of variables. Finally, hedgerow management quality is compared to farmers' opinions regarding the benefits and problems associated with the practice. Studies on the adoption of soil conservation technologies began in the 1950s (Ervin and Ervin 1 982). Study results predict whether farmers adopt, the rate at which they adopt, or identify the characteristics of adopters compared to non-adopters. This

PAGE 125

110 information is used to design interventions (Bonnard and Scherr 1 994), evaluate policy or adjust extension messages (Earle et al. 1979, Murray 1980). or to report on and evaluate extension projects (Pierre et al. 1995, Scherr 1994, Shultz et al. 1997, Smucker 1988. Sunderlin 1997. Swanson 1993, White 1992a). Adoption studies generally followed the model described by Rogers ( 1995). that diffusion of adoptions is a process having four components: (1) the innovation. (2) the social system in which the innovation is being adopted. (3) channels of communication (the ways that adopters pass on information about an innovation), and (4) the time over which a social system adopts an innovation. Initially adoption models were static, with the underlying assumptions that the adoption decision is binary; there is a fixed, finite ceiling on adoption; the rate and degree of diffusion through a population is fixed; the innovation is not modified once it is introduced and its diffusion is not affected by other innovations; there is one adoption per adopting unit; and the geographical boundaries of the adopting social system stays constant over the diffusion process (Knudson 1991). However, some of these assumptions are not useful in understanding the adoption of a complex technology such as hedgerows. Adoption researchers began to recommend dynamic models that separated determinations of diffusion over time as the technology being investigated was affected by other technologies and as the adopting population changed in various ways (Knudson 1991), and to integrate physical, economic, and other factors (Femandez-Comejo et al. 1994). Feder et al. (1985) defined final adoption of agricultural technology, at the level of the individual farmer, as the degree of use of a new technology in long-run equilibrium when the farmer has ftiU information about the new technology and its potential. They recommended studying both the extent and intensity of use of the new

PAGE 126

Ill technology throughout the adoption process to allow for changes in parameters affecting farmers' decisions. Specifically for hedgerow adoption studies, Dvorak (1991) suggested dividing hedgerow adoption into three phases (establishment, maintenance, and productive) to avoid pronouncements of success or failure based only on the initial establishment. Some agroforestry adoption studies are regional overviews based on cost/benefit analysis (Current and Scherr 1995). but most are based on case studies using combinations of farmer interviews and plot monitoring (David 1995, Dvorak 1991, Fujisaka et al. 1995, Murray 1980, Neef et al. 1996, Smucker 1988. Wiersum 1994) or questionnaire surveys (Kessey and O'Kting'ati 1994, Lea 1994, Pierre et al. 1995, Singhai and Kumar 1997, Swanson 1993, Vaval et al. 1997). Usually more than one method of data collection is required to interpret the meaning of adoption (Dvorak 1991, Sunderlin 1997, White 1992a). Once the data is collected, analysis methods vary from multiple regression (Ervin and Ervin 1982), logistic regression (McNamara et al. 1991, Norris and Batie 1987, Sureshwaran et al. 1996). ethnographic interpretation (Murray 1980. Smucker 1988. Swanson 1993). or cost/benefit analysis (Nelson et al. 1997). Adaptability analysis is another methodology specifically developed to incorporate the wide diversity of biophysical and socioeconomic conditions found on developing country farms. It uses regression techniques, but has a focus on finding technology solutions for multiple environments rather than simply defining and describing adopters and nonadopters (Hildebrand and Russell 1996). Variables Used to Predict Adoption The dependent variables used to represent adoption of soil conservation practices are commonly the binary decision to adopt or not adopt, the number of practices adopted.

PAGE 127

112 and estimated rates of erosion (Ervin and Ervin 1982). The author did not encounter any studies that used the quality of management of soil conservation structures as a dependent variable. The independent variables used to make the predictions/descriptions can be economic characteristics of the technology (Byerlee and Hesse de Polanco 1986. Jarvis 1981, Stark 1996), socioeconomic characteristics of the farm or farm household (Earle et al. 1979, Harper et al. 1990, Kessey and O'Kting'ati 1994, Shullz et al. 1997, Singhai and Kumar 1997, Stark 1996, Sunderlin 1997, Sureshwaran et al. 1996, Vaval et al. 1997, White 1992a), physical or tenure characteristics of the plot (Smucker 1988, Sureshwaran et al. 1996, Vaval et al. 1997. White 1992a). or farmer attitudes regarding actual or potential benefits or problems associated with a technology (David 1995, Earle et al. 1979, Fujisaka et al. 1995, McNamara et al. 1991, Norris and Batie 1987. Wiersum 1994). A combination of these is the norm. Findings from studies using these categories of variables are discussed below. Technologv Characteristics Farming technologies differ in several ways, including complexity, adaptability, cost of installation and management, potential return from the investment, timing of the return, and the inputs (land, farmer knowledge, labor, chemicals, seed) involved. Some technologies are simple, relatively low in initial cost, and have a short return on investment (e.g., a new variety of crop). Others are complex, involve relatively high investments in installation and management, and can take several years before benefits are realized (e.g., hedgerows). The success of some technologies depends on the adoption of others. A study of an improved barley variety and attendant inputs in Mexico found that farmers rarely adopted complete technology packages, but adopted them in sequence

PAGE 128

113 based on predictions of profitability and riskiness. Most adopted one component at a time, those giving the highest return on capital were adopted earliest (Byerlee and Hesse dePolanco 1986). In Kenya a study of 3.000 farmers found they decreased risk associated with new agroforestry practices through incremental adoption and adaptation, and cost and riskreducing modifications to the technology design (Scherr 1995). The authors noted that farmers adopt agroforestry practices only when they bring economic gain. This is a widely noted, and hardly surprising, observation. Soil conservation programs in developing countries have not had much success because they are often too costly to implement and maintain, so only the better-off farmers are be able to adopt the best practices, poor farmers have to modify them to reduce the cost (Napier 1991). In Rwanda, where a high population of farmers intensely crop marginal soils on steep slopes, soil conservation practices proposed to them were not attractive because of the time lag between initial investment and return (Steiner and Scheidegger 1994). Successful alley cropping adoption in Sri Lanka depended on profitability, time to profit, labor requirement and seasonal availability (Nuberg and Evans 1993). Farmer Household Characteristics Household characteristics are labor. land, and knowledge resources that influence adoption of technologies. These characteristics are not consistent in the way they influence adoption, but vary according to the technology. Problems occasionally arise because the correlations between some household variables can be influenced by other variables so that the correlation includes the spurious effect of the others (Feder et al 1985). In Africa, where larger farm size and greater extension contact were important to adoption of alley cropping, both were highly correlated to level of education (Tripathi

PAGE 129

114 and Psychas 1992). Ervin and Ervin (1982) rejected farmer age as a household variable in a study of adoption of soil conservation in Missouri, USA, because of a high correlation with other independent variables. A study of an integrated pest management (IPM) practice in Florida. Michigan, and Texas in the US found a significant and positive relationship between adoption and farm size and availability of family labor; a significant and negative correlation between adoption and the importance of livestock (Femandez-Cornejo et al. 1994). Farm size and income had a significant and positive impact on soil conservation expenditures in Virginia; off-farm employment, debt level, and tenure had significant and negative impacts (Norris and Batie 1987). In a soil conservation project in El Salvador, adoption of soil conservation and agroforestry practices was greater among younger than older farmers, among land owners than tenants, and increased with more extension visits (Shultz et al. 1997). A model based on a survey predicted intention toward soil conservation to be stronger as farm size, family income, and education level increased (Earle et al. 1979). A study of adoption of sweep nets (IPM) in the US found that higher education was correlated with lower adoption, but farm size was not a significant factor (Harper etal. 1990). A study of agroforestry adoption (tree and grass associations) in Tanzania found farm size, household size, number of animals, and distance traveled in fuelwood collection, were significantly correlated with adoption. Number of days used for fodder collection was not significant (Kessey and O'Kting'ati 1994). Farm size, household size, and total number of animals were positively correlated with adoption of silvopastoral associations in the Garhwal Himalaya region of India. Age of head of household and

PAGE 130

115 distance traveled for fuelwood collection were negatively correlated with adoption (Singhai and Kumar 1997). A survey in the Philippines found the main constraint to hedgerow adoption was high labor demand for installation and maintenance. As the number of persons per household decreased and the percentage of female members increased, adoption decreased; women did not usually help in hedgerow establishment (Stark 1996). A model of hedgerow-based soil conservation adoption in the Philippines suggested that government assistance, land size, farmer age. land intensity, and tenure impacted adoption, but income and education did not (Sureshwaran et al 1996). Plot Characteristics A commonly held opinion is that farmers will invest more resources on a plot of land that is held under a secure form of tenure. It has been recommended in some areas that alley cropping should be targeted first to farmers who own inherited or purchased land (Tonye et al. 1994). Other researchers note that tenure insecurity, but not lack of title itself, could be a disincentive to adoption of agro forestry practices (Current et al. 1995). Apparent effects of tenancy, however, could be caused by indirect relationships between tenure and access to credit or other inputs (Feder et al. 1985). In West Africa tenure plays a significant role in adoption of alley cropping, land tenure security is favorable to it (Lawry et al. 1994). In Benin, tenants, the landless, and most women are worse off than landowners in their ability to adopt agro forestry practices because of tenure insecurity among other reasons (Neef and Heidhues 1994). In the US, land tenure was found to be not significant to IPM adoption because it was an investment in human capital, not in the land (Femandez-Comejo et al. 1994) — so the impact of tenure depends on the kind of technology.

PAGE 131

116 Physical characteristics of the plot sometimes impact adoption by themselves, but they usually interact with other characteristics. Size of the farm holding could be a surrogate for other factors, such as access to credit, information or wealth (Feder et al. 1985). In Benin, there was a significant difference between owned and leased fields with regard to distance from the compound (Neef et al. 1996). An adoption study in Indonesia found that non-adopters had significantly less area in sloping land than adopters, but the slope percent of the sloping land was not significant (Fujisaka et al. 1995). Another study in Benin found the key parameters determining adoption of farming technology included the function and history of the field, tenure, and the field's position with regard to fertility flows in the farming system (Koudokpon et al. 1994). Farmer Attitudes Regarding Problems and Benefits of the Technology Perceived problems and benefits are sometimes reported as observations and sometimes used as independent variables in adoption models. However, it can be difficult to reconcile expressed attitudes with farmer actions. Dvorak (1991) reported in her survey farmer responses about alley cropping were generally favorable, but in no case had a farmer, household member or neighbors extended an alley or planted a new alley farm. She noted that two apparently enthusiastic farmers even uprooted the alley trees, and concluded this reflected the difficulty in using a one-time survey to evaluate a complex system (Dvorak 1991). Nevertheless, farmer attitudes can help understand adoption. Farmers" expressed benefits of hedgerows in Indonesia were decreased erosion, flattening alleys for cropping, and ability to use fertilizers without loss; drawbacks were animal damage. Most important reasons for non-adoption were work demands on lowland plots, high labor in the hedgerows, offfarm work opportunities, lack of draft animals, and lack of

PAGE 132

117 capital for labor and inputs (Fujisaka et al. 1995). A model based on survey data predicted intention toward soil conservation was stronger as perception of soil erosion as a problem increased (Earle et al. 1979). Perception of erosion did have a significant and positive impact on soil conservation expenditures in Virginia (Norris and Batie 1987). Farmer attitudes may be contrary to the expectations of researchers. After 12 years of extension education in Georgia with peanut farmers showing increased income and environmental benefits, only 31% of farmers adopted IPM. The study found promoting the environmental impact of IPM did not affect adoption (McNamara et al. 1991). In Rwanda, small farmers rarely mentioned erosion as an important production constraint (Steiner and Scheidegger 1994). In a risky production environment in Kenya, where crop yields fluctuate constantly with the amount of rainfall, soil fertility was not an urgent concern to farmers (David 1995). Study of hedgerow adoption in Indonesia noted that some farmers adopted not because of its productive benefits, but as a means to gain access to land or credit, or to demonstrate allegiances to social networks (Wiersum 1994). Findings from Adoption Studies in Haiti The independent variables used in this study, and others that were not included, have been used by other researchers in Haiti. Smucker (1988) found that farmers planting tree seedlings distributed by a project were older, more protestant, more often married, and better schooled; had larger households and more children; were more involved in wood-related occupations and more likely to own large animals; hired more labor and sold their own labor less; and had more land and more securely-held land, and had more land in fallow than did nonplanters. White (1992a) found that religion, age.

PAGE 133

118 and wealth were not significant to participation in group soil conservation practices in Haiti, but that group membership and cooperative labor tendencies were very important. Both White (1992a) and Vaval et al. (1997) studied soil conservation and hedgerow adoption in the Maissade region of Haiti and found that land tenure of plots was not important to participation in cooperative watershed management or soil conservation adoption. There were, however, differences of plot tenure among farmers of different economic levels within adopters and non-adopters (Vaval et al. 1997). Haitian farmers' perceived problems and benefits associated with hedgerows have been surveyed in several studies. The most frequently cited problem was animals destroying the hedgerows, especially during the period after crop harvest when goats are let free to forage (Pierre et al. 1995, Swanson et al. 1993a, Swanson et al. 1993b, Villanueva 1993). Animal problems and mode of access to the plot can be related. Farmers in the northwest of Haiti reported to Swanson et al. (1993a) that, on inheritedundivided plots, they often tie animals nearby their hedgerows to permit direct grazing because other family members will do so in any case. Plot characteristics and labor required to install hedgerows are also cited as problems. White (1992b) reported that 39% of farmers not adopting hedgerows and other soil conservation practices cited lack of time as the reason, 21% said it was due to not owning land, 17% said they owned inappropriate land. A study following up a terminated soil conservation project in the south of Haiti cited farmers as saying hedgerows were too difficult to manage, and that they could not extend them because it was too expensive and took too much time and labor to repair them (Villanueva 1993). Only one study reported farmers' perceived benefits. Over 80% of 105 farmers interviewed in the south of Haiti said that hedgerows increased crop production by 25 to 50%, and some said they felt more secure in their

PAGE 134

119 access to the plots because of the positive reaction of the landlords to hedgerow construction (Pierre et al. 1995). This literature review has shown that variables representing characteristics of the technology, farm household resources, plot characteristics, and farmer attitudes toward the technology are sometimes useful in predicting adoption. However, the predictive utility of each variable is case specific. A variable might be useful in predicting adoption in one case but not another, or the influence of the variable might be positive to adoption in one case and negative in another. The most consistent variables influencing adoption of hedgerows appear to be economic gain and lag to profit, and threshold cost of installation, especially labor cost. Farmers are more likely to adopt when it makes economic sense to do so. The studies done on adoption of soil conservation in Haiti included mode of access to the land as a variable. In some cases the conclusion appeared to be that land tenure was important, but that it was not in other cases. This apparent contradiction probably means that it was not important to a farmer's participation, in some form, in a project, because at least one of the farmer's plots is likely to be suitable for some activity that counts as participation. However, tenure could be important in deciding what technology to adopt on a particular plot. It is noted that no other studies were discovered that used management quality of installed agroforestry practices as a dependent variable. Materials and Methods A survey of farmers participating in the PADF PLUS project was carried out in spring 1996. The objectives of the survey beyond those of interest to this research were explained in Chapter 4. Sampled farmers were those who had first participated in the project before 1 January 1995, and who, therefore, had been working with the project for

PAGE 135

120 at least 16 months at the time of the survey. In each of the four project regions {Camp Perhn in the southwest, Marigot/Palmiste a Vin in the southeast. Cap Haitien in the north, Mirebalais in the lower central plateau), we selected 35 participating farmers from the geographic areas of responsibility of eleven PADF agronomist technicians by using a random number list to pull farmer dossiers from the regional files. The technicians visited 1,540 farmers. 5.6% of all eligible farmers. During the whole period of data collection, other project staff carried out a continuous check of the technicians' work, rotating among the four PADF/PLUS field regions. Data cleaning was performed by sorting variables in the database, and checking suspect data with original questionnaires. The Household Questionnaire The survey data were recorded on two questionnaires, one completed in the home of the sampled farmer and one in the garden plots. Technicians completed a two-page household questionnaire recording a description of all plots owned and cultivated by the household during 1995, and a description of all the members of the household (Appendix D). Plot information recorded included distance from the residence (home garden, nearby, or distant), topographic position (top of slope, mid-slope, foot of slope, or plain), and slope steepness (flat, sloping. ver>' steep). The size of the plot and the mode of access (purchased, inherited and separated among siblings, inherited and unseparated. share-cropped, rented from a private party, rented from the state) under which the plot was held were also recorded. For each individual member of the household, age, sex, highest level of education, kinds of PLUS project participation, and kinds of economic activities participated in were recorded. Finally, the head of household was asked whether he or she was originally from the locality, and if not. when did he arrive there. The farmer being interviewed defined the composition of the household and identified

PAGE 136

121 who was the head of the household. The information recorded on the household questionnaire was collected during an in-home interview. The Plot Questionnaire During the home interview, the farmer was asked to indicate in which of the plots listed on the household questionnaire he or she had installed PLUS-inspired agroforestrv' practices. The farmer and the technician then visited those plots. A total of 2.295 plots were visited, an average of 1 .5 plots per farm. The technician recorded information on the physical properties of the plots, the yield of crops during the previous twelve months and the average yield before soil conservation structures were built, and management and attitude information for each of the PLUS agro forestry interventions in the plots (Appendix E). Data collected about the physical characteristics of the plot included the fallow status of the plot, slope (taken by the technician with a plumb line clinometer attached to a clipboard), slope aspect, an estimate of the elevation of the plot to within 100 m, the number of erosion gullies greater than 20 cm depth, parent material of the soil (basaltic or calcareous), and the local name for the soil type on the plot. The number of trees per hectare (of any species from any source) whose trunk diameter was greater than 10 cm at 1.3 meters from ground level (DBH = diameter at breast height) was recorded. These trees were counted by the technician conducting the interview, using a plywood caliper with a 10 cm opening to define countable trees. The distance in minutes from the interviewed farmer's house to the plot was estimated. The farmer was then asked four qualitative questions about the soil on the plot; the answers were recorded as a number between one and five. These were regarding soil fertility (1 = infertile, 5 = fertile), soil

PAGE 137

122 depth (1 = shallow, 5 = deep), the "hotness" or ''coldness" of the soil' (1 = hot. 5 = cold), and the degree of erosion on the plot (1 = no erosion, 5 = severe erosion). The farmers' qualitative analyses of soil fertility were followed up with a laborator}' analysis. Soil samples were taken from 1 75 gardens selected at random from all gardens visited. 35 each from gardens having fertility classes 1 through 5. The samples were taken by technicians using a machete from the 0-20 cm layer; a composite of four positions for each sample. They were put into ziplock bags and labeled, then transported to the University of Florida for laboratory analysis. Complete methodology and results for the soil analysis are discussed in Appendix F. Farmers were asked about the yield of the principal crops they harvested during 1995 and an average yield of the same crops before soil conservation structures were installed on the plot. This information was analyzed by SECID for reporting on PLUS project impact to USAID, and so was not used by the author in this study. Questions about the number of family members working on the plot, number of person days of labor from outside the family used on the plot, and the kinds of agro forestry and soil conservation structures installed on the plot completed the general questions. Then followed a separate page of questions for each of seven possible kinds of PLUS project activities installed on the plot: hedgerows, crop bands, rock walls, gully plugs, trees planted on the plot during 1995, top-grafted fruit trees, and vegetable gardens. The appropriate page was used according to the types of structures or techniques used on the plot. The questions asked on these pages varied with the technique. For the soil 'Most writers agree that "hot" soils (CHO) are dry, well-drained soils and "cold" soils (FWET) have more soil water available for plants. Other concepts are integrated into this system as well, including soil parent material, slope, orientation, and vegetative cover (McClain and Stienbarger 1988, Murray 1981). Some crops grow better in hot soils, some in cold, but in general cold soils are preferred (Smucker 1981).

PAGE 138

123 conservation practices they included: the amount present as measured by the technician (number of rows and total length of each row), qualitative assessments (1 to 5 scale) by the farmer of the importance of a series of potential benefits and problems associated with the practice, who of the household participated in the construction and repair of the practice, the number of breaches in the structures over a specified size, whether or not the breaches were repaired, and a row-by-row rating by the technician of well-managed. adequately-managed, or poorly-managed. The management ratings integrated the technicians' opinion about hedgerow plant vigor, pruning height, density, and number of breaches. Agroforestrv Practices Compared on the Plots Farmers participating in the PLUS project are asked to protect at least one plot of land with an agroforestry practice before they gain access to crop seed banks operated by their community-based organization (CBO) with PLUS assistance. The practices are discussed during training sessions with PADF technicians and CBO extension agents.. The farmer decides what practice to install, and where to install it. It is noted that this project requirement is probably complied with, in many cases, only to gain access to crop seeds and tools. Some hedgerows, therefore, have been installed only to satisfy this rule, by farmers who judged the value of the seed and tool credit to be greater than the cost of trying the new technology. Although the central focus of this study was the hedgerow, several other agroforestry practices were included in the analysis to contrast how farmers change their installation and management strategies to accommodate characteristics of the technology. The agroforestry practices vary in their general types. Hedgerows, crop bands, and rock wall terraces are built across the slope in contour rows. The area between rows, the

PAGE 139

124 interspaces or alleys, is planted to field crops. Gully plugs are confined, obviously, to ravines or gullies. Trees may be planted anywhere. The usual planting configurations are perimeter plantings, small woodlots. or in widely spaced rows in farmed plots. Top grafting is done on existing fi-uit trees, planted before the PLUS project began. Hedgerows Hedgerows were defined in Chapter 3. They consist of a single or double row of densely planted trees (e.g.. Leucaena leucocephala, Gliricidia sepium), perennial food crops (e.g., sugar cane, pineapple), or other perennial (e.g., perennial cotton, castor bean) planted on the contour as a physical barrier to soil erosion, as a soil fertility enrichment structure, and a source of diverse products of economic interest to farmers. Annual crops are planted in the alleys between hedgerows. Crop bands (bann manie) These evolved from hedgerows built from sugar cane and pineapple. They are similar to hedgerows in that they are installed as contour rows of perennial vegetation in a farmed plot, leaving room for field crops in the alleys between rows, with the goal of producing useful vegetation and holding soil and water on the plot. The differences are: 1) crop bands have a larger dimension — they are a band of one or two meters width as opposed to a single or double row of plants found in hedgerows, 2) the perennial plants that serve as structural components are food crops, such as sugar cane, pineapple, and plantain, and 3) annual crops are also planted in the band, such as yams and sweet potatoes. While differences between hedgerows and crop bands are not always clear — hedgerows can be made with sugar cane or pineapple, for instance, and often have an annual crop planted on the uphill side of the hedgerow that differs from the field crops planted in the alleys — crop bands are wider, contain a greater number of crop species.

PAGE 140

125 and usually do not contain a woody perennial. The number of crop bands is still small (4.4% of PLUS plots had them in 1995). but farmers are interested and the practice is growing. Rock walls Rock walls are constructed of fieldstones place in rows across the slope contour. Two general kinds are built. The most common has a footing dug into the soil, and the stones are stacked carefully with flat sides and top {mi sek). The other kind requires less skill; the stones are simply piled in rows along the contour without preparing a footing {kodon pye). Gully plugs These are constructed across narrow, actively eroding ravines. They are built from rocks where available {sey woch), but also from stakes cut from trees, interwoven with smaller branches {kleyonaj.fasinaj). If they are built strongly, they will accumulate a considerable amount of eroded soil in a short time. The soil trapped by the gully plug is quite valuable, because it is deep, well drained, and collects water from adjacent slopes during rainfall. The usual practice is to plant high-value crops uphill from the structures, often plantains and taro. as soon as enough soil accumulates. They can be costly to install and maintain, but the income from the crops can be high and a well-protected ravine does less damage to the farmer's property. Trees Two kinds of trees are referred to in the analysis: those planted on the plot as seedlings by a farmer during 1995 as a project activity, or mature trees of any species or source over 1 cm in diameter at breast height.

PAGE 141

126 Top-grafted fruit trees These are mature low-value mango or citrus that were grafted, either by the farmer or by a project extension agent, with high-value buds. Statistical Analysis Data transformation and statistical analyses were done using SPSS version 9. Statistical tests were considered significant at the 95% level of probability. Because the distribution of most of the variables recorded on the questionnaires was not normal and because variables having several factor levels usually had great differences in the numbers of observations among the factor levels, nonparametric analysis was performed on the questionnaire data. The Kruskal Wallis test of mean ranks was used instead of analysis of variance. Because SPSS version 9 is not able to do mean separation after a significant Kruskal Wallis result, Mann-Whitney comparisons were done on the ranked mean pairs, followed by a Bonferroni correction (the pvalue of the MannWhitney tests were multiplied by the number of factor levels, R. Littel. personal communication, September 1999) (Ott 1992, Zar 1984). Cross tabulations were tested with Pearson chisquare; bivariate correlations were tested with Spearman's rho. Limitations of the Analvses The regions where the PLUS project was working were selected because they are smallholder hillside farming areas with above-average agricultural potential. Because only farmers participating in the PLUS project were used as the sample population, conclusions based on the data produced by the survey may not apply to Haiti generally, and they may not apply to farmers not participating in PLUS. It is also noted that the information regarding participation of household members in PLUS agro forestry activities and other economic activities was given by one person, usually a male head of

PAGE 142

127 household, and may therefore be biased. The analysis presented here is limited in that no plots were sampled that 1) were controlled by PLUS participants but had no agroforestry practices installed, 2) were controlled by farmers not participating in the project, or 3) had PLUS-inspired practices installed by secondary adopters (farmers not receiving PLUS extension visits or subsidies). Results and Discussion Results are presented in three sections. The first section describes the household members and all the garden plots under household control during 1 995 based on the inhome interview questionnaires, and then the 2,295 plots visited by PADF technicians during the survey. The plots visited by the technicians were those where interviewed farmers had installed at least one project-sponsored agroforestry practice. The second section addresses the decision to install hedgerows and other related agroforestry practices. Several types of household resources bearing on land, education, and labor are compared between households that installed and households that did not install particular agroforestry practices. Physical characteristics of plots having the practice are then compared to those of the plots where that practice was not installed. The third section discusses the relationships between the quality of management applied to agroforestry practices on the plots, and three categories of independent variables: 1) household resources, 2) plot physical characteristics, and 3) problems and benefits associated with hedgerows as perceived by the farmer. Characteristics of the Household and Household Plots Numbers of family members by gender The 1,540 households interviewed reported a total of 8.584 members, of which 52 percent were male and 48 percent female. The mean number of people per household

PAGE 143

128 was 5.6. Table 6-1 shows the percent of household members by gender in seven age classes. Males and females have about the same distributions among the age classes. Heads of household Eighty-five percent of heads of household were male; fifteen percent were female. The mean age of male heads of household was 46 years, and 50 years for female heads of household. Eightyfour percent of heads of household reported that they were from the area where they now lived, and sixteen percent moved into the locality from outside. There was a small but significant age difference between local (46 years) and immigrant (49 years) heads of household. A higher percent of females (18.6%) than males (15.6%) were immigrants, but the difference was not statistically significant. The number of years of schooling for heads of household was 2.7. but it was not distributed normally. About half of them had no schooling whatsoever; the median number of years was zero. There was a significant difference in schooling between female and male heads of household, the mean number of years for females was 1.2, and for males 3.0. Table 6-1 : Age distribution by gender of members of 1,540 farm households interviewed in the four PADF/PLUS project areas in Haiti, 1996.

PAGE 144

129 Years of school for household members Increased education could facilitate access to extension material, or increase the likelihood that a farmer might be selected as a project extension agent, and thereby increase his probability of adopting an agroforestry technology. The mean total combined number of years of school for all members of the household was 16 (median = 12). The average male household member had gone to school for 3.3 years, significantly more than the average female household member, who had 2.6 years of school. The distribution of years of education is better understood when broken down by age, as shown in Table 6-2. Table 6-2: Number of years of school by gender in seven age classes of 1,540 households participating in the PADF/PLUS project, 1996

PAGE 145

130 received more education than women did. but that has apparently changed for the sampled population and now there is no difference. Participation of household members in economic activities The interviewers asked PLUS farmers if each household member participated in economic activities other than PLUS soil conservation, tree culture, and crop improvement practices. A list of categories of activity was produced from the results, but this list does not include normal agricultural work, childcare. cooking, or home maintenance. The mean number of economic activities was three per household. Table 6-3 shows the percent of female and male household members engaged in various categories of economic activity. Small businesses include furniture construction, tailoring, plowing fields, and other activities requiring specialized training or equipment. Local official positions include persons elected to government posts and schoolteachers. Two categories, fishing and factory work, are not shown because they included less than 0.5 percent of household members. Table 6-3: Percent of female and male members of 1,540 project households who participated in various economic activities during 1995-1996. % of Members Participating Activity Female Male Animal raising 6 25 Marketing agr. products 6 Marketing other products 20 3 Small businesses 2 9 Handicrafts 1 1 Agricultural day labor 0.5 1 Charcoal or lime production 2 Local official position 1 1 Leaf doctor 1 Sawyer 2 Other day labor 1 Sample size 4097 4479

PAGE 146

131 Table 6-3 confirms the commonly made observation that women are more involved in marketing (Murray 1981) and men in animal raising. The percent of women involved in marketing, men selling charcoal, and men selling agricultural day labor appears to be very low. This might be an indication that sampled farmers, being project participants, were wealthier than average. Table 6-4 presents the same information by age class. The two youngest age classes are not included because of very low participation. Only 2 percent of the 6-10 year old boys and 1 percent of the 6-10 year old girls were active in raising animals. Table 6-4: Percent of female and male members of 1 .540 project households who participated in various economic activities during 1995-1996, by age class.

PAGE 147

132 more into the younger and older age classes, especially for males. Table 6-4 confirms that more boys than girls participate in animal raising, and likewise more old men participate than do older women. The counterpart activity for women is marketing. They begin this activity at an earlier age and continue it longer into older age than do men, and more women than men work in marketing. Men are more active in small businesses, charcoal and lime production, and sawing wood than are women. Participation of household members in PLUS agro forestry practices A greater percent of males than females participate in the installation and management of agro forestry practices. The PADF field staff commonly report that mainly females participate in vegetable gardens, but Table 6-5 shows that males are equally involved. As in the case of other economic activities, persons in the 21 to 60 year age class (usually over 70% of those participating) do most of the agro forestry work. However, the highest percent participation within an age class is seen in the 61 to 70 year olds, where 79% of the males and 36% of the females participate in soil conservation structures on slopes (Table 6-6). Table 6-5: Percent of female and male members of 1 ,540 households who participated in PLUS project activities during 1995-1996. Activity Female Male Soil conservation on slopes 15 39 Gully plugs 2 7 Vegetable gardens 4 5 Fruit tree grafting 1 3 Tree planting 7 20 Sample size 4097 4480

PAGE 148

133 Table 6-6: Percent of female and male members of 1 ,540 project households who participated in PLUS project activities during 1995-1996, by age class. Age classes (years) 11-15 16-20 21-60 61-70 >70 Activity FMFMF MFMFM Soil conservation on

PAGE 149

134 of the sizes of each plot type, considered separately, are skewed, so the means are not good predictors of the average value. Each distribution has a number of extreme values (plots much larger than average). The sum of the median values (1 .5 ha) is close to the 1 .3-hectare median value for the total household plot area. The average size of each type of plot (as opposed to the total amount of land area in all plots of each type as shown in Table 6-7) is shown in Table 6-8. Table 6-7: Average total area of three categories of plots held by 1 .540 project households based on the relative distance from the plots to the residence.

PAGE 150

135 absolutely, is strengthened. The distant plots were significantly larger than the other two plot types in Regions 1,3, and 4. In Region 2. the mean area of the near plots was smaller than that of the other plot types. Table 6-9: Mean plot area of three categories of plots held by 1.540 project households during 1995 in four PLUS project regions, based on the relative distance from the plots to the residence.

PAGE 151

136 Table 6-10: Percent of three categories of plots held by 1,540 project households during 1995 under various modes of access.

PAGE 152

137 Table 6-11: Area of 5.660 plots held by 1.540 households during 1995, by mode of access. Mean plot area (ha) N_ Rented, state Caretaker Purchased Rented, private Inherited, unseparated Sharecropped Other Inherited, separated 0.90 a

PAGE 153

138 Tenure and slope class . A cross tabulation of land tenure by slope class shows results similar to those for topographic position, as it should since topographic position and slope are related (Table 6-13). Land rented from the state is much less likely to be flat, and much more likely to be very steep than other tenure types. Again, the opposite is true for caretaker plots, which are less likely to be on very steep land than are other tenure types. Table 6-13: Percent of all plots held by 1,540 households during 1995 in eight modes of access categories by slope steepness .

PAGE 154

139 practices. If we consider only the plots having PLUS practices, each plot had two agro forestry interventions, one of which was a soil conservation structure. Distance from the residence. There were small but statistically significant differences in distance between the residence and the plot, measured in minutes, among tenure classes" (Table 6-14). Table 6-14: Distance from residence of plots having project agro forestry practices held by 1.540 households during 1995 in the five most common mode of access categories Mean distance (minutes) N Sharecropped 16 a 189 Rented, private 14 ab 284 Inherited, unseparated 13 b 481 Purchased 13 b 948 Inherited, separated 12 b 324 From the data in Table 6-10, we might have expected purchased plots to be significantly nearer the residence than plots of other tenure types. This was the case; rented and sharecropped plots were more distant than purchased or inherited plots. However, the range of distances is so small that the result is not important. The seven plots in the sample that were rented from the state averaged 25 minutes away from the residence. Tenure and size of plot. Comparison of tenure categor>' by plot area done only on the plots having PLUS interventions (Table 6-15) agreed with the analysis done on all plots controlled by the household (Table 6-11). Purchased plots were again significantly larger than the other tenure types. Inherited, separated plots were the smallest. Three tenure classes having very few observations (State land, 7; Caretaker, 32; Other, 29) are eliminated from statistical analyses of PLUS plots.

PAGE 155

140 Table 6-15: Area of plots having project agro forestry practices held by 1,540 households during 1 995 in the five most common mode of access categories Mean plot area (ha) N Inherited, separated 0.36a 324 Rented, private 0.41a 284 Inherited, unseparated 0.41a 482 Sharecropped 0.42a 189 Purchased 0.54b 948 Tenure and elevation. We estimated elevation within plus or minus 1 00 meters above sea level for the PLUS subset of plots (Table 6-16). Elevation did not correspond to relative topographic position as described by farmers in Table 6-12 (for example, it is possible to have a high elevation plain), but the results were similar. There were no significant differences in elevation among tenure categories. This might indicate that the technician's estimates were not accurate enough, that farmers" plots tend to be found within 100 meters of each other, or a combination of both. Although elevation of plots rented from the state was significantly higher, there were only seven plots in that category. State land in PLUS areas was more commonly found in higher elevations. Table 6-16: Elevation of plots having project agro forestry practices held by 1,540 households during 1995 in seven mode of access categories N 948 324 481 189 32 284 29 7

PAGE 156

141 Tenure and slope . Slope percents measured by technicians on the subset of plots with PLUS interventions (Table 61 7) shows caretaker plots as having the gentlest slopes. State land, however, is also shown as being on gentler slopes, which is contrary to the finding in the cross tabulation done for all household plots (Table 6-13). Table 6-17: Slope of plots having project agroforestry practices held by 1,540 households during 1995 in seven mode of access categories Slope (%) N Caretaker 22a 3 1 Rented, state 26b 7 Inherited, separated 31b 324 Purchased 32b 947 Rented, private 33b 284 Inherited, unseparated 33b 482 Sharecropped 34b 189 Other 34b 29 The Decision to Install Agroforestrv Practices This section considers the 2,295 plots where the 1 ,540 surveyed households had installed at least one project agroforestry practice. These plots are referred to as "PLUS" plots, meaning the practices were installed under the auspices of the PLUS project. The characteristics of all PLUS plots having a given agroforestry practice were compared to those of all PLUS plots not having that practice. The plots excluded from consideration are the 3,356 plots accessed by the households during 1995 that did not have at least one agroforestry practice installed. Using all household plots in the comparison would have been preferable, as it would have given a better idea of how farmers make decisions based on plot criteria, but technicians visited and collected data only from PLUS plots.

PAGE 157

142 However, it is still possible to see differences in plot characteristics using only the PLUS plot data. Household resources and the decision to install The independent variables of interest in this section pertain to the amount and security of access of the land controlled by the household (number of plots, total plot area, number of purchased or inherited-separated plots, total plot area of purchased or inherited-separated plots); the number of family members available as labor (total number of household members, total number of household members of ages of 21 to 60, number of females of ages 21 to 60. number of males of ages 21 to 60); level of education (total years of education for all household members, years of education of the head of household); and the amount of labor used on the plot of interest during 1995 (person-days of agricultural labor purchased in the plot, number of household members working in the plot). The reasons for including these variables were that the probability of adoption might increase as the amount of total land and land under secure tenure increased and as the number of family members available to work the land increased. Farmers with more land and labor might be more able to risk land for new technologies and with a lower cash outlay for labor if family members were available. Level of education was of interest because more education could allow a farmer easier access to extension material or increase the likelihood that he or she be selected as an officer in the farmer group or as an extension agent. Either of these outcomes could increase the probability of adoption. Households that had installed a particular practice (hedgerow, crop band, rock wall, gully plug, project trees planted during 1995, top-grafted fruit trees) on at least one of the visited plots were compared to those who had not installed that practice. Very few

PAGE 158

143 of the comparisons between adopting and nonadopting households were significant, and those that were did not have very large numerical differences. In spite of the small differences, interesting trends were revealed: Hedgerows . Households installing hedgerows had 0.3 ha less total plot area than those not installing (p = .012), and had 0.2 fewer family members working on the plot during 1995 (p <.001). Both of these findings might appear to be counter intuitive, since less land and labor available to a household should decrease the likelihood of adoption. However, the comparisons were made only within the population of project participants, all of whom installed at least one soil conservation practice. Since hedgerows were the least expensive of the structures to install, it is possible the poorest participants installed hedgerows. Crop bands . Households installing crop bands had 0.7 more plots (p < .001), 0.6 more secure plots (p = .005), 0.27 ha more area in secure plots (p = .011), but heads of household having 0.9 fewer years of school (p = .01 1) than did households not installing crop bands. This indicates that households adopting crop bands were better off compared to nonadopters. Crop bands are more expensive to install that hedgerows because they require large quantities of perennial crop cultures (e.g., sugar can, pineapple) and they take up more space in the plot than hedgerows. Rock walls . Households installing rock walls had 0.3 more household members (p = .001), 2 years less of total schooling (p = .006), and 0.2 more family members working on the plot in 1995 (p < .001 ) than did households not installing rock walls. Rock wall adopters apparently have more labor available than non-adopters, but no difference in amount of land. Rock walls are costly to install, but the cost is exclusively

PAGE 159

144 in labor as no crop germplasm is used. Returns to labor are seen more quickly than for hedgerows, since rock walls function immediately to slow erosion and collect water. Gully plugs . Households installing gully plugs had 0.5 more plots (p < .001), 0.5 more total hectares (p < .001), and purchased 4 more person-days of labor to work in the plot during 1995 (p = .001) than did households not installing gully plugs. This appears to indicate that better-off farmers built gully plugs. Gully plugs built of stone can be costly to install, but they often create new agricultural land (because they collect sediment on the uphill side) where water accumulates from the side slopes. This can happen quickly (within one rainy season), especially in sandy, basaltic soils. Farmers typically plant plantains and taro in these microsites. As for the installation, getting the crop germplasm can be expensive but the profit can be substantial. PLUS project tree seedlings . Households planting PLUS project tree seedlings on at least one plot during 1995 had 0.2 more plots (p = .012), 2.2 more total years of school (p = .005), and 0.2 more family members working on the plot during 1995 (p = .004) than did households not planting project tree seedlings. Top-grafting . Households having top-grafted fruit trees on at least one plot had 0.5 more plots (p = .001), 0.4 more hectares (p = .012), and 3.4 more total years of school (p = .006) than did households not having top-grafted fruit trees. It is interesting that households planting project tree seedlings and those grafting fruit trees had more total years of school than households not engaging in those activities. This was not true for any of the other agroforestry practices. These findings are similar to those found by Smucker (1988), but the reasons for it are not clear. Tree seedlings are either provided free or grown by the farmer, and the first four grafts are also either done by the farmer or the extension agent without charge, so costs of adoption are minimal.

PAGE 160

145 The small number of significant results and the small numerical differences, even when significant, between households having and not having the indicated practices means there were small household resource differences in the sampled population. This was probably because nonproject farmers were not included in the sample. However small the differences, though, it appears that • households installing hedgerows had less land than households installing crop bands and gully plugs or top grafting; • households installing crop bands had more land in secure tenure than other households; • households installing gully plugs invested more in purchased labor on the plot; and • households planting project trees or top-grafting attended more years of school than other households. These associations appear to make sense based on the costs of installation of the agroforestry technologies, as explained above. Plot characteristics and the decision to install The plot characteristics of interest were tenure security, soil fertility, slope, and distance from the plot to the residence. The assumption was that these influence the probability that farmers adopt agroforestry practices in various ways. Land held under secure modes of access would be more likely to be used when the practice is expensive to install and potential returns are high and extend over time. Plots having fertile soil would be used for practices that could provide a quick return to that resource. In contrast, plots having infertile soil might be used for practices that were previously unknown to the farmer and looked risky. Slope might influence adoption in a similar way to soil

PAGE 161

146 fertility — the best (less sloped) plots would be used for practices requiring larger investment, steeper slopes for untried practices. Distance from the residence might be an issue if theft of potential products (e.g., trees, fruit) was likely. Direct measurements or observations were used to determine tenure, slope, and distance. Purchased plots and inherited divided plots are referred to as "secure". Farmers assessed soil fertility on a qualitative scale, with the soils rated as 4 and 5 on that scale referred to as "fertile." Analyses showed there was a correspondence between farmers' ratings and laboratory tests: soils rated as fertile had a lower pH, more potassium, and more organic carbon (Appendix F). Plot characteristics for six agroforestry practices were considered in the 2,295 gardens visited during the survey: hedgerows, found in 60% of the gardens; crop bands, found in 4%, rock walls, found in 30%; gully plugs, found in 17%; tree seedlings, found in 46%; and grafted trees, found in 9%. The importance of tenure security and soil fertility on the decision to install a practice is shown in Tables 6-1 8 to 6-21 . These tables present two-by-two cross tabulations comparing tenure security (Tables 6-18 and 6-19) and soil fertility (Tables 6-21 and 6-21) of plots having the indicated practice to those not having the practice. Three kinds of results were seen. A lower percentage of plots with hedgerows were in secure tenure (52%) compared to plots without hedgerows (61%). However, a higher percentage of plots with crop bands, tree seedlings, and tree grafting were in secure tenure compared to plots without those practices. Tenure status of plots did not differ for rock walls or gully plugs.

PAGE 162

147 Table 6-18: Percent of 2,295 surveyed plots with and without three kinds of agroforestry practice held under secure (purchased or inherited/divided) or not secure (all other) tenure Hedgerows Crop bands Rock walls with without with without with without % of plots having secure tenure 52 61 66 55 54 56 % of plots not having secure tenure 48 39 34 45 46 44 Number of plots 1,365 930 100 2,195 686 1.609 2-by-2 chi-square P-value <.001 ^30 .349 Table 6-19: Percent of 2,295 surveyed plots with and without three kinds of agroforestry practice held under secure (purchased or inherited/divided) or not secure (all other) tenure Gully plugs Tree seedlings Tree grafting . with without with without with without % of plots having secure tenure 57 55 64 48 69 54 % of plots not having secure tenure 43 45 36 52 31 46 Number of plots 385 1,910 1.050 1,245 207 2,088 2-by-2 chi-square P-value ^604 <.001 <.001 All cross tabulations for the effect of soil fertility on adoption showed significant differences from the expected cell values. A lower percent of plots with hedgerows had fertile soil (41%) compared to plots without hedgerows (51%). The opposite was true for all other practices, where a higher percent of the plots with them were rated as having fertile soil compared to plots without them. Table 6-20: Percent of 2,295 surveyed plots with and without three kinds of agroforestry practice having fertile or not fertile soil Hedgerows Crop bands Rock walls with without with without with without % of plots having fertile soil % of plots not having fertile soil Number of plots 1,365 930 100 2,195 686 1,609 2-by-2 chi-square P-value <.001 ^07 .001 41

PAGE 163

50

PAGE 164

149 Table 6-22: Percent of 2,295 surveyed plots with and without three kinds of hedgerows held under secure (purchased or inherited/divided) or not secure (all other) tenure Tree-based Crop-based Grass-based with without with without with without % of plots having secure tenure 49 59 54 56 48 56 % of plots not having secure tenure 51 41 46 44 52 44 Number of plots 701 1.594 507 1.788 44 2,251 2-by-2 chi-square P-value <.001 .543 .300 Table 6-23: Percent of 2.295 surveyed plots with and without three kinds of hedgerows having fertile or not fertile soil Tree-based Crop-based Grass-based with without with without with without %ofplots having fertile soil 34 50 51 43 43 45 % of plots not having fertile soil 66 50 49 57 57 55 Number of plots 701 1,549 507 1.788 44 2,251 2-by-2 chi-square P-value <.001 ^01 .810 The results for tree-based hedgerows were the same as for hedgerows in general — a lower percent of plots where they were installed were in secure tenure and fertile soil compared to plots where they were not installed. However, when perennial crops were added as structural components of hedgerows the results were similar to those for crop bands, with a higher percent of plots having crop-based hedgerows in fertile soil compared to plots without them. Tenure security was not significant for crop-based hedgerows; neither tenure security nor soil fertility were significant for grass-based hedgerows. There were statistically significant differences (a = 0.05) in the slope of plots having agroforestry practices compared with those not without them, but the differences were numerically small. Hedgerows (34% vs. 30%) and crop bands (37% vs. 32%) were

PAGE 165

150 on plots having steeper slopes compared to plots without those practices. Rock walls (31% vs. 33%) and grafted trees (26% vs. 33%) were on plots having less steep slopes compared to plots without those practices. There were no differences in slope for gully plugs or tree seedlings. The differences in distance from the residence to the plots having agroforestry practices and those without them were also statistically different (a = 0.05), but the numerical differences were only from one to three minutes, except for grafted trees. Grafted trees were found on plots averaging seven minutes walk from the residence, while plots without grafted trees were fourteen minutes distant. It is possible to gain additional insights on the relationship between plot characteristics and the decision to invest in agroforestry practices by considering all trees on the plot whose trunk diameter is greater than 1 cm DBH. These trees may or may not have been planted in association with PADF projects, and may be either fruit or hardwood species. There were a significantly greater number of mature trees on purchased plots, and on inherited but separated plots, than on inherited unseparated, sharecropped. or on rented land (Table 6-24). More (but not statistically different) trees were found on inherited separated land than on purchased land, probably because the inherited land had been held longer by the household than the purchased land. Table 6-24: Number of all trees larger than 10 cm diameter on PLUS plots, by plot tenure Mean no. of trees/ha" No. of plots Inherited, separated 103a 324 Purchased 88 a 946 Inherited, unseparated 69 b 48 1 Sharecropped 61 b 189 Rented, private 56b 283 "Kruskall-Wallis p-value <.001

PAGE 166

151 A similar analysis showed that more mature trees were found on plots having fertile soil (Table 6-25). Table 6-25: Number of all trees larger than 10 cm diameter on PLUS plots, by soil fertility Qualitative fertility classes Mean no. of trees/ha'' No. of Plots Very fertile

PAGE 167

152 whether the farmer repaired breaches by observation and by asking the farmer, since this kind of repair is normally done just before crops are planted and might not have been done yet at the time of the visit. As for the decision to install tests, the four management quality indicators were tested against plot slope, distance from the residence in minutes, mode of tenure/access to the plot, and farmers* qualitative estimates of soil fertility. Household resources and management quality The statistical tests used in this section were bivariate correlations (percent well managed, percent poorly managed, number of breaches) and t-tests without the assumption of uniform variances (whether breaches were repaired). Similar to the section on household resources and the decision to install, there were statistically significant correlations and t-tests, but the correlation coefficients and differences between means were small. Hedgerows . The percent of well-managed hedgerows decreased as the number of plots held by the household (p = .004) and the number of plots in secure tenure (p = .054) increased. The percent of well-managed hedgerows increased as the number of males of age 21-60 (p = .003), the number of household members of age 21-60 (p .016), and the number of person-days of labor purchased in the plot during 1995 (p = .050) increased. The number of breaches in the hedgerows decreased as the number of plots in secure tenure (p < .001), number of hectares in secure tenure (p < .001), number of person-days of labor purchased in the plot during 1995 (p = .012), and the number of family members working in the plot (p = .002) increased. Households repairing breaches in hedgerows had 0.3 more family members working in the plot than did households not repairing breaches (p = .002).

PAGE 168

153 Crop bands. The percent of well-managed crop bands increased as the number of years of school attended by the head of household increased (p = .043). The number of breaches in crop bands decreased as the number of secure plots (p = .036), the number of person-days of labor purchased in the plot during 1995 (p = .017). and the number of years of school attended by the head of household increased (p = .027). Households repairing breaches in crop bands purchased 19 person-days of labor in the plot, households not repairing breaches bought 1 person-days (p = .012). Rock walls . The percent of poorly managed rock walls decreased as the number of females of age 21-60 increased (p = .021). The number of breaches in rock walls increased as the number of plots per household increased (p = .034). Households repairing breaches bought four more person-days of labor in the plot (p = .049) and had 0.3 hectares less total plot area (p = .037) than households not repairing breaches. Gullv plugs . The percent of well managed gully plugs increased as the number of secure plots (p =^ .025), the number of hectares in secure tenure (p = .041), the number of household members (p = .004), the number of females age 2 1 -60 (p = .002), and the number of household members age 21-60 (p = .006) decreased. The number of poorlymanaged gully plugs increased as the number of females age 21-60 (p == .012) and the number of household members age 21-60 (p = .039) increased. Plot characteristics and management quality Although plot tenure security appears to have been an important factor in farmers' decisions to install agroforestry practices on a plot, it was not related to the quality of management of the practices. None of the tests of land tenure status against management quality (percent well-managed, percent poorly-managed, number of breaches, repairs breaches) were significant for any agroforestry practice.

PAGE 169

154 Hedgerows. Hedgerows were installed on plots having less-secure tenure compared to other practices (Table 6-18). but, once installed, the best management of the hedgerow was practiced on plots having the most fertile soil (Table 6-26). Note that the very fertile category appears to have poorer management (fewer percent well-managed rows and more breaches) than the above average category. This kind of anomaly in the order of soil fertility categories is found in many of the analyses, and is probably due to the inexact and qualitative nature of this variable. Table 6-26 indicates that the bestmanaged hedgerows were found on the most fertile soil, the worst managed hedgerows were found on the most infertile soils, and that farmers were more likely to repair breaches on more fertile soils. Table 6-26: Soil fertility and hedgerow management quality. Qualitative soil Percent well Percent poorly No. of breaches Repairing (%) fertility classes managed rows managed rows per 100 m Yes No infertile

PAGE 170

155 distance between the house and the plot; there are more breaches in hedgerow plots closer to the house (p < .001). These results seem contradictory. It seems reasonable to suppose steeper slopes would have more breaches because it is harder to establish hedgerow trees on steeper slopes, and more difficult to work at repairing them. The case of more breaches in closer more gently sloping plots might be explained by more foot traffic closer to the house cutting paths through the hedgerows, and the possibility that animals might be picketed in plots close to the house. When the three kinds of hedgerows are tested separately, as they were in the previous section discussing farmers' decisions to install practices, we find a similar but weaker result (as compared to the aggregated hedgerow results in Table 6-26) regarding management quality and soil fertility. For tree-based hedgerows, the soil fertility vs. management quality Kruskal-Wallis tests (percent well managed rows, percent poorly managed rows, number of breaches, and whether or not the breaches were repaired) showed significant and similar differences in management quality as all hedgerows combined — better management on more fertile soil. However, the mean separation tests were either not significant or did not produce a rank order that could be interpreted. Crop-based hedgerows were better managed in plots having fertile soils. The percent welland percent-poorly-managed variables increased and decreased in the same order as soil fertility ratings increased (p = .006) and decreased (p < .001), respectively. Again, however, mean separation was not significant in the case of percent well managed. Slope was not an important factor in management quality in tree-based hedgerows, but significant correlations were found for crop-based hedgerows. The percent of well-

PAGE 171

156 managed rows increased as slope became steeper (p = .009), and the number of breaches decreased as slope became steeper (p < .001). The mean slope of repaired crop-based hedgerows was slightly, but significantly (p = .007), steeper (36%) than the slope of unrepaired ones (32%). In only two cases was distance from plot to residence significant. Number of breaches decreased with increasing distance for both tree-based (p = .003) and crop-based (p = .009) hedgerows. No differences were found in any of the tests of management quality for grass-based hedgerows. Differences in hedgerow management quality among PADF field regions were at least as important as differences among plot characteristics (Table 6-27). Table 6-27: PADF/PLUS field region and hedgerow management quality. PADF/PLUS Percent well Percent poorly No. of breaches Repairing (%) Field Region managed rows managed rows per 100 m Yes No 1 : Les Cayes

PAGE 172

157 Crop bands . The small number of crop band gardens in the survey (about 100) made it necessary to recode the soil fertility ratings into three categories instead of five. Even so, only five crop band plots were in the infertile category. Farmers apparently eliminated infertile plots during the installation, leaving few to be analyzed for differences in management quality. Table 6-28 indicates that very fertile plots had fewer breaches and a larger percent of farmers repairing breaches. The test for percent of rows well managed and soil fertility was significant, but mean separation tests were not. Table 6-28: Soil fertility and crop band management quality. Qualitative soil

PAGE 173

158 included in the table. All other teams had more than 30 crop band plots each, but because of missing data for some of the variables, not all of the plots were included in the analyses. Table 6-29: PADF/PLUS field region and crop band management quality. PADF/PLUS Field Region

PAGE 174

159 Table 6-30: Soil fertility and rock wall management quality. Qualitative soil Percent well Percent poorly No. of breaches Repairing (%) fertility classes managed rows managed rows per 100 m Yes No infertile

PAGE 175

160 There is less difference in rock wall management quality among PADF regions than there is for plant-based structures. However, rock walls in regions 1 and 4 appear to be better managed (Table 6-31). Table 6-3 1 : PADF/PLUS field region and rock wall management quality. PADF/PLUS Percent well Percent poorly No. of breaches Repairing (Vo) Field Region managed rows managed rows per 100 m Yes No 1 : Les Cayes

PAGE 176

161 gardens in which the interviews took place, and were not observations about hedgerows in general. The percentages of ratings falling in each qualitative category for six potential benefits are shown in Table 6-32. Table 6-32: Percent of 1,362 plots where potential benefits of hedgerows (fuel, charcoal, construction wood, fodder, soil fertility, soil conservation, and crop production) were classified according to importance by farmers on a five-point scale. l=not important. 5=very important.

PAGE 177

162 compare stated importance of benefits to how well farmers manage hedgerows, on the theory that hedgerows on a plot where a potential benefit is highly-rated would be better cared for. The results of Kruskal-Wallis tests comparing the mean ranks of the percentage of well-managed hedgerows in gardens where the potential benefits were rated are shown in Table 6-33. Table 6-33: Percent of hedgerows classified as well-managed by technicians in 1.362 plots where potential benefits of hedgerows (fuel, charcoal, construction wood, fodder, soil fertility, soil conservation, and crop production) were classified according to importance by farmers on a five-point scale, l=not important, 5=very important.

PAGE 178

163 only 5% of the total. There was apparently some substance to farmers categorizing soil fertility and soil conservation as of highest importance in a large number of gardens, since in those gardens were found substantially greater proportions of well managed hedgerows compared to gardens rated lower for those benefits. The opposite trend should show up when the same analyses are done for proportion of poorly managed hedgerows and for the number of breaches per 100 m of hedgerow. This is the case, as shown in Tables 6-34 and 6-35. All Kruskal-Wallis tests were significant for percent poorly managed vs. benefit rating, however mean separation for the charcoal benefit was not significant at a=0.05. Table 6-34: Percent of hedgerows classified as poorly-managed by technicians in 1,362 plots where potential benefits of hedgerows (fuel, charcoal, construction wood, fodder, soil fertility, soil conservation, and crop production) were classified according to importance by farmers on a five-point scale, l=not important. 5=very important.

PAGE 179

164 were highly rated but the differences are not great and the trend, according to the results of the mean separations, are not always clear. Table 6-35: Number of breaches per 100 m of hedgerow counted by technicians in 1,362 plots where potential benefits of hedgerows (fuel, charcoal, construction wood, fodder, soil fertilit}'. soil conservation, and crop production) were classified according to importance by farmers on a fivepoint scale, l=not important, 5=very important.

PAGE 180

165 Table 6-36: Percent of gardens where hedgerow breaches were repaired or not repaired for three potential benefits of hedgerows.

PAGE 181

166 vs. whether the breaches were repaired, the problem rating categories were reduced from 5 to 3, with the old categories 1 and 2 (the problem is not important) grouped together and old categories 4 and 5 (the problem is very important) grouped together. Only two cross tabulations produced significant results, as shown in Table 6-38. As reduced space for picketing animals in hedgerows and labor cost of maintaining hedgerows became more important, fewer farmers repaired breaches in the hedgerows. Table 6-38: Percent of gardens where hedgerow breaches were repaired or not repaired for three potential problems of hedgerows.

PAGE 182

167 participated in some phase of the installation or maintenance of soil conservation structures. The survey found no significant differences between tenure of plots in terms of elevation, topographic position, or severity of erosion. There were small differences in slope (caretaker plots were less steep, all others were about 33%) and distance from the residence (sharecropped and rented plots were one or two minutes more distant) by tenure category. Purchased plots averaged 0.53 hectares, significantly larger than divided, undivided, and sharecropped plots. There were statistically significant differences in soil fertility (data not shown). A higher percent of purchased plots (49%) were in the high fertility category compared to other plots (42-44%), with rented plots having the lowest percent in high fertility. Decision to Install Soil Conservation Technologies There were indications that household resources influenced decisions about what kind of soil conservation practice to install. Households installing hedgerows had smaller farms and fewer family members working on plots than did households not installing hedgerows. Households installing crop bands accessed more plots, had more plots in secure tenure status, and had more land area in secure tenure status than households not installing crop bands. Households installing rock walls had more household members and more family members working on the plots than households not installing rock walls. There were strong indications that farmers considered plot characteristics in their decision to install hedgerows and other agro forestry practices. Figure 6-1 shows the relative influence of plot characteristics. The numbers for tenure security and soil

PAGE 183

fertility are the percent of plots having the indicated practice that are held in secure tenure or that have fertile soil, respectively. Tenure security Soil fertility Slope (%) Distance from residence (min) Less 52

PAGE 184

169 • Fruit trees were top-grafted on more secure, more fertile plots of gentler slope, much closer to the residence. • More mature trees of any species were found on more fertile plots. There were a significantly greater number of mature trees on purchased plots, and on inherited but separated plots, than on inherited-unseparated. sharecropped, or on rented land. Management Quality Management quality of agro forestry practices, as defined by the condition of the structures at the time of the survey, was not correlated with level of household resources. The strongest trend noted was that quality of management for hedgerows, crop bands, and rock walls increased as the level of labor employed on the plots increased. Although tenure security was an important factor in the decision to install an agro forestry practice, it did not influence the quality of management of those practices once they were installed. The management quality of hedgerows, crop bands, and rock walls did not differ according to the tenure of the plots. The most important variable appeared to be soil quality as perceived by the farmer. All soil conservation structures were better managed on plots having better soil fertility. Management quality increased as slope decreased. The effect of distance between the plot and the residence was not strong. These results could not be compared to those from other studies, because no other studies used management quality as a dependent variable. Most farmers did not consider fuelwood, charcoal, or construction wood to be important benefits of hedgerows. Soil conservation and soil fertility improvement were considered important by substantial numbers of farmers, as were, to a lesser extent, crop yield improvement and fodder. Management quality of hedgerows was significantly better on plots where any of the potential benefits was highly rated. Very few farmers said any of the listed potential drawbacks of hedgerow was important — 94% said water

PAGE 185

170 competition between hedgerows and adjacent food crops was not a serious problem. There was little correspondence between opinions regarding severity of hedgerow problems and management quality. Differences Among Technologies Differences among technology adoption revealed in the survey are better understood if discussed in the context of investment, risk, and timing of benefits to the farmer. Agroforestry practices have a cost of installation and a cost of management that can be measured in labor and materials. The farmer must compare these costs to the resources commanded by the household, and then to the amount and timing of the anticipated benefits, which must in turn be compared to the needs of the household. All this must be balanced against the risk of failure. Risk might be higher if the farmer or his peers have no previous experience with the practice, or if the practice is sensitive to environmental factors, such as free ranging animals or drought. Risk is diminished when some of the costs are subsidized by projects, or when there is a continuing outreach program. The survey did not provide cost and benefit data, but some general comparisons among practices can be made based on experience of the project staff. In the matrix shown in Table 6-39, the relative costs, benefits, and risks of agroforestry techniques are compared. The ratings shown in Table 6-39 cannot be fixed exactly, because they change with the length of the farmer's experience, with the available resources and needs of the household, with the stage of development of the practice, and with the degree of project subsidy. For example, tree-based hedgerows are at greater risk of animal damage in the early stages of growth, and are less prone to animal damage if there are perennial crops associated with the trees. In that case, neighboring farmers would be more likely to

PAGE 186

171 contain their animals. Tree seedlings are much more vulnerable to loss due to accidental weeding by paid laborers and by animal damage in the early stages of growth. Crop bands containing sugar cane, pineapple, or plantains could be expensive to install in some areas, unless the plant material is subsidized by a project. Table 6-39: Estimated relative costs, benefits, and risk of agroforestr\' practices Cost of Cost of Amount of Timing of Risk of AF practice installation management benefits benefits loss Hedgerows low-med. med.-high med.-high med.-long med.-high Crop bands med.-high med.-high high short med. Rock walls high low low-med. short low Gully plugs med.-high low-med. med.-high short-med. low-med. PLUS trees low low med.-high med.-long low-med. Grafted low-med. low-med. med.-high med.-long low-med. trees Hedgerows It is apparent that hedgerows were installed on less-desirable plots. There are several possible reasons for this. Hedgerows are a relatively new agroforestry practice in Haiti, even though there are similar indigenous practices such as trash barriers and living fences, and farmers tend to install new practices on their worst plots (Murray 1 979). They are not difficult to install, and the PLUS project sometimes provides the seeds. Hedgerows are a relatively easy way to comply with the PLUS project requirement that participants install soil conservation structures on some part of their holdings in order to get access to crop seed banks operated by their group— about 60% of the sampled PLUS plots had hedgerows. Although the cost of installation is reasonable, hedgerows require regular maintenance in the form of repairing breaches, top-pruning the hedgerow trees, and eliminating invasive hedgerow seedlings from the area between the rows. The

PAGE 187

172 timing and amount of returns to hedgerows depend on the management objective of the farmer. If they are being managed for soil fertility, three to five years may be required to make a noticeable contribution. If they are being managed for animal production, useful amounts of fodder can be taken after the first year. Tree-based hedgerows exhibited the same plot characteristics (Tables 6-22. 6-23) as all types of hedgerows were considered together (Tables 6-18, 6-20). They were on plots of less security, lower fertility, and steeper slope. Tree-based hedgerows showed no difference in distance from the residence compared with all other plots. The most striking result displayed in Tables 6-22 and 6-23 is that when even one food crop species entered the hedgerow, plot characteristics changed, and became similar to those of the crop band plots: they were installed in plots having soils of higher fertility and slightly steeper slopes. Unlike crop bands, however, crop-based hedgerow plots did not differ in tenure security from other plots, and there was a slight, but significant tendency for cropbased hedgerows to be farther from the residence. The key plot parameter affecting the farmers' decision in this case seems to be soil fertility, since 51% of the plots having crop-based hedgerows had fertile soils, while only 34% of the plots having tree-based hedgerows were rated as fertile. Crop bands Because of their larger dimension and greater number of food crops, crop bands are more expensive to install than hedgerows, and need more fertile soils and regular rainfall to flourish. Quite often in the PLUS project, the food crop plants are distributed to new participants in a crop multiplication arrangement. The farmer must return the same amount of plant material to his or her group as was given, and it is then spread to other farmers under the same arrangement. This underwrites the cost of installation, and

PAGE 188

173 reduces risk to the farmer. Crop bands were planted on steeper plots of higher fertility compared to plots having other interventions. Land tenure appears to be a factor in the choice of plots, as it is in the case of hedgerows, except that crop bands are installed on more secure plots. High soil fertility is required to sustain crop growth, and to reduce the risk of losing the investment in valuable crop germplasm. The choice of steeper slopes possibly reflects the extension message that crop bands are a soil conservation technique. Rock walls Rock wall terraces are well known to most farmers because they have been promoted for many decades by various projects. They require the most skill and labor to install of all the soil conservation practices promoted by the PLUS project. Once installed, however, they are relatively undemanding in their upkeep. They produce no vegetation to take care of, and they are stable if large animals are kept out of the garden and the slope is not too steep. About 30% of the sampled gardens had rock walls installed. The availability of rocks prevents farmers from constructing rock walls in some areas, especially where the soils are derived from basaltic parent material. Analysis of plots having rock walls show them to be similar in most characteristics compared to others, except they are located on soils of higher fertility. This may reflect a choice of the farmer — since rock walls are an expensive investment they might want to install them on the best soils — or it might be that soils having enough rocks for constructing the terraces are of higher fertility than normal. Although slope of rock wall plots was less steep and distance from the residence was greater compared to other plots, the differences were small. Land tenure was not a factor in choice of plots for rock walls.

PAGE 189

174 Gullv plugs Tenure and slope were the same on plots with and without gully plugs. Soils on gully plug plots were more fertile, but it is not clear whether the farmer was judging the fertility of the soil collected by the structure, or the soil on the plot in general. Gully plug plots were slightly farther from the house than are other plots. It could be that houses are located away from gullies for safety. Number of mature trees on the plot There was clearly an association between the number of trees on a plot and the security of tenure. More securely-held plots had greater numbers of trees. There was also a very slight but statistically significant negative correlation between the numbers of trees on a plot and the distance of the plot from the house. There tended to be more trees on plots closer to the house. There was not a significant correlation between slope of the plot and numbers of trees. There was also a significant difference in the numbers of trees found on plots of different soil fertility, as shown in Table 6-25. The number of trees increased with increasing soil fertility. From observation, most trees nurtured by Haitian farmers are fruit trees or trees whose timber is valuable. Valuable trees are planted near the house to avoid theft. Home garden plots were more likely to be in secure tenure categories (Table 6-10). Greater numbers of trees could be found on soil of higher fertility because they survive and grow better there. In addition, animals are often tied under trees near the home, and this, along with the accumulation of tree litter, enriches soil fertility. Project tree seedlings planted during 1995 Trees planted with project assistance during 1995 exhibited similar plot characteristics as existing large trees (Tables 6-19 and 6-21). They were located on plots

PAGE 190

175 of more secure tenure, slightly greater soil fertility, closer to the house, but with no difference in slope. Overall, more trees were planted and maintained on more fertile, more secure plots close to the residence. Establishment and maintenance cost little, and potential returns are high. These plot characteristics reinforce the idea of trees as a store of value. Top-grafted fruit trees The decision to graft fruit trees appears to have been made according to specific plot characteristics. Alternatively, the original decision of the farmer to plant or nurture the low value fruit trees could be reflected in these plot characteristics. All four plot characteristics are significantly different from those of plots not having grafted trees. Grafted trees are found on plots with greater tenure security, more fertile soils, gentler slopes, and much closer to the house. This clearly indicates that grafted trees are mainly found in or near the home garden, the plot where the house is found. Conclusions Previous studies found differences in household resources between adopters and non-adopters. These differences were weak in this study, but appeared to indicate that hedgerow adopters were less well off (less land and household members), while crop band adopters were better off (more land and more securely held land). This could have been due to regional differences (most of the crop bands were in the north of the country), or because they are more demanding in their requirements for space, soil fertility, and rainfall. Possibly only better off farmers can make crop bands succeed. Farmers apparently based their decisions regarding installation of agro forestry practices on characteristics of the plot and the technology. There were differences in the selection of plots between the different agro forestry practices. Overall, the survey

PAGE 191

176 indicated that farmer decisions to adopt new technologies are correlated with several plot characteristics. Tenure appears to influence adoption in five of six technologies surveyed. Trees and grafted fruit trees were more common on purchased and divided inheritance plots. The value of tree products increases over time, so farmers need to protect their rights to harvest. Crop bands and gully plugs are also more common on purchased or divided inheritance plots. This is likely attributable to the high value of perennial food crops in crop contour bands (pineapple, plantain, sugar cane) and the economically important crops planted in soil collected by gully plugs (plantains, taro). Hedgerows are more commonly found on plots with other modes of access. Hedgerows are relatively easy to install, so this may reflect a strategy of risk minimization when trying a new practice or fulfilling project requirements to install soil conservation measures. Correlation between technology adoption and soil fertility was as important as the relation between adoption and tenure status. This is perhaps to be expected since tenure status and soil fertility were also related (more purchased plots were classified as very fertile); however, farmer assessments of fertility also appear to integrate other productive factors not measured by laboratory analysis of soil nutrient levels. Tenure and soil fertility were both associated with adoption in parallel fashion. Technologies (crop contour bands, gully plugs, trees, top-grafted fruit trees) more common on purchased and divided inheritance plots were also more common on fertile plots, and conversely (hedgerows). Although overall analysis of the data indicated that mode of access to land was an important variable, the data showed no definitive relationship between tenure status and adoption. The evidence does not allow clear separation of the relative influence of tenure and fertility on adoption; therefore, it is not

PAGE 192

177 possible to determine which is more important in a particular decision to adopt new technology. The analysis found no association between tenure status and differences in management; but soil fertility was important to management quality, with the best management found on the best soils. Based on these results, an extension program would not need to target technologies towards or away from any particular tenure type. It might, however, want to work with farmers on cost-effective ways to improve soil fertility while simultaneously ensuring a supply of dry-season fodder. Farmers did not consider soil water competition between hedgerows and adjacent crops to be a problem. Only 16 of the 1,540 farmers considered it to be a serious problem. This could be because there was adequate rainfall in 1995, the success or failure of a crop is dominated by other factors that effect the whole field, such as the timing of the rainfall compared to crop demands, or because very few farmers rated any of the problems on the questionnaire as severe. This study revealed relationships between plot characteristics and farmers' decision to install various soil conservation structures, and how those structures were managed. However, the findings should be considered as indicative, and not definitive, until follow-up interviews clarify the relationships, for several reasons: 1 . Decisions to install various technologies appeared to correlate with both tenure and soil fertility in the same way, the relative contribution cannot be determined from the data in this study. 2. Management quality of soil conservation practices varied significantly between regions of the country at least as much as differences due to soil fertility, but it

PAGE 193

178 is not clear if this was a reflection of differences in regional soil fertility, extension methods, or the length of time since a technology had been introduced into the region. 3. The analysis mixed together farmers just recently installing a practice with those having several years experience. Hedgerows can require several years to show benefits, and they evolve as farmers get more experience with them. Analysis should be separated by age of the practice. 4. The correlations between farmers' expressed problems and benefits regarding hedgerows and the management quality of hedgerows should have been compared as well to their pattern of building or removing hedges after the initial installation. In general, getting good responses to these kind of questions might require a more intensive type of data collection from a much smaller sample of farmers as a complement to the larger survey.

PAGE 194

CHAPTER 7 CONCLUSIONS AND RECOMMENDATIONS The over-arching objective of this series of studies was to produce information that could improve extension and adoption of agro forestry technologies in Haiti, particularly of hedgerow intercropping. The experimental designs employed reflected the range of information needed to understand agroforestry systems, and included onstation and on-farm agronomic studies as well as a large-scale questionnaire-based socioeconomic survey. The objectives of the three studies comprising the dissertation were to determine the importance of soil water competition between hedgerows and adjacent maize on station and on farm, and to examine the factors that influence farmer adoption, adaptation, and management of hedgerow intercropping. These objectives were attained, but with some limitations as indicated in the following discussion based on the hypotheses tested in each study. The on-station trial showed that soil water competition caused by Leucaena hedgerows could reduce maize yield substantially, however the percent reduction in yield depended on how limiting rainfall was during the season. The mitigating effect of installing root barriers between the maize and the hedgerows was temporary, as Leucaena roots re-grew under the barriers and into the alleys. Severing Leucaena roots during the installation of the barriers resulted in a loss of 1.600 kg/ha Leucaena biomass over a seven-month period. Thereafter, the hedgerow trees recovered and there was no 179

PAGE 195

180 subsequent difference in biomass yield between plots having root barriers and those without barriers. The on-farm trial of maize development did not detect any differences at various distances from three types of soil conservation structures and at various positions on the slope. There were differences in maize development between types of soil conservation structures and the untreated control, with the maize growing in plant-based structures (hedgerows and crop bands) developing more slowly than maize in rock wall terraces and plots without soil conservation structures. No statistical differences in maize yield were detected, even though the untreated control plots produced numerically less than the treated plots. The plots having plant-based structures yielded less maize than the rock wall plots, indicating competition between maize and the plants in the hedges, but again the difference was not significant. It is the opinion of the author that differences in maize growth and yield do exist between flat land and sloping land. The failure to show differences was probably due to the small sample size and short duration of the on-farm trials. No direct recommendation for the use of hedgerow root barriers on farms can be made based on these results. Assuming the increased maize grain yield due to root barriers on station could be realized on farm by a similar technique, say trenching on the uphill side only of the hedgerows before each maize crop, it still might not make economic sense to the farmer. On the station trial, increased maize grain yield due to barriers was about 350 kg/ha. If a typical hillside half-hectare plot would realize an extra 175 kg of maize grain, the cost of installation would be about halfway between the range of prices the extra maize would bring at the farm gate. In addition, the maize crop might fail completely due to poor rain up to 30% of the time, so the risk to the farmer might be

PAGE 196

181 too great for the expected benefit, unless another crop of higher value were planted or the trenches bring some other value to the garden plot, such as increased water harvesting. The on-farm survey hypotheses were supported by the results, but within a limited range because the sampled farmers were all project participants and therefore similar. Level of household land and labor resources in some cases influenced the choice of technologies and the management quality applied to agro forestry practices. Associations were detected that appeared to confirm that more available labor and land improved management quality, and that farmers with better resources invested in more expensive agroforestry practices. Farmers based their decisions about where to install different soil conservation structures on characteristics of the plots. Mode of access and soil fertility were important in the decision to install contour soil conservation structures (hedgerows, crop bands, rock walls) but their relative influence could not be separated. Farmers' decisions to install hedgerows on plots having relatively poorer soil fertility and less secure tenure could indicate they have not yet seen that practice as one returning sufficient benefits to risk better plots, whereas the short-term benefits taken from the crops bands apparently justify installing them on better plots. Soil fertility appeared to be the most important determinant of management quality of the structures, with the best management occurring on the most fertile plots for all practices. Tenure status of the plot did not affect how well farmers managed hedgerows or any other agroforestry practice. Hedgerow plots where farmers cited any of several potential benefits (wood products, fodder, soil fertility and conservation, improved crop production) as being very important were better managed. However, very few farmers rated wood products as important hedgerow benefits. Over half of the interviewed hedgerow farmers cited soil fertility and soil conservation as being very important hedgerow benefits. Because plots where that

PAGE 197

182 opinion was expressed were significantly better managed, the farmers were probably revealing their true attitudes toward hedgerows, as opposed to parroting an extension message. The conceptual thread of tree/crop soil water competition was not successfully traced through the three studies described here. This was because the study designs recorded different scales of data from environments of increasing variability. The smallscale phenomenon of soil water competition close to hedgerows seen on the station was not encountered on the four hillside farms comparing different soil conservation techniques in the second study. A larger sample of farmers over several seasons, with yield measured row by row might have picked it up. Farmers claimed that hedgerow/crop soil water competition was a serious problem on less than one percent of hedgerow plots surveyed. From observation, drought over the whole garden causing crop failure probably covers up tree/crop competition for soil water on hillside farms. These three studies represent components of what should be a broader program of multidisciplinary trials and field exercises to understand the dynamics of hedgerow performance and adoption in Haiti. The resuhs are indicative of some aspects of competition and adoption, and the links between them, but they are by no means complete. It is evident that by focusing on a specific biophysical phenomenon on station, such as soil water competition between hedgerow trees and adjacent maize, one can not necessarily usefully apply those results to a system as complex as Haitian hillside farms. There are three overriding, and interrelated, characteristics of hillside agriculture in Haiti that make that so: complexity, variability, and need for cash. Complexity is manifested in the number of plots of varying quality worked, the several modes of access through which the plots are worked, numbers of crops

PAGE 198

183 grown/animals raised, numbers of extra-agricultural enterprises undertaken by family members, and the social networks and dependencies nurtured by the family. Great variability is seen in soil quality and other physical properties within and between plots, timing of rainfall compared to the needs of the crops, the quantity and timing of agricultural labor available to the family from year to year, sale price of crops, and the variability of crop yield. The need for cash for both cyclic and cataclysmic expenses in a cash poor economy guides farmer decisions in a complex and highly variable farming system. Complexity can be a response to soil and rainfall variability and the need for cash, as a risk management strategy, such as holding several plots in different microsites under different modes of access. But complexity in a new technology, such as hedgerows, can make it difficult for a farmer to plan investment strategies that allow him to manage it optimally, given his level of family resources and cash constraints. This is because farmers' price expectations from a plot contain both a mean and a cyclical component, and farmers are unlikely to invest in a practice, like hedgerows, unless they perceive it as affecting the mean component. Time of adoption for a given farmer occurs when his or her beliefs about the expected value of profits from the new technology is at least as great as the return from the old technology. But it takes time for farmers know the variance of profits with the new technology. Because of the initial delay in realizing benefits from hedgerows, due to loss of cropping space and time required to establish the hedges, and because this takes place in a highly heterogeneous and variable cropping system, the farmer has difficulty in discerning the mean component from the seasonal variations. Hedgerows are an evolving, complex, relatively expensive agroforestry practice being experimented with by very poor farmers in a highly variable and risky social and agricultural environment. When new technologies are constantly being

PAGE 199

184 modified, adoption equilibrium may never be attained because the parameters affecting farmers' decisions are changing as well. Methods for studying adoption under these conditions were not completely met by the research undertaken in this study. Additional farmer focus groups to follow up soil water competition on farm and to separate the influence of mode of access on adoption from that of soil fertility would have been useful. Most socioeconomic adoption models have not incorporated the distinctive features of agroforestr>', such as multiple outputs, production variability, the economic role of trees, impact of off-farm employment, and the sociocultural context of farmer decision-making. On farm methods of testing dynamic agroforestry technologies have not been adequately tested. However, in spite of the shortcomings of the methodology used in these studies, useful information was developed that might guide future investigations. Some recommendations based on these studies are: • It is more important to focus on stabilizing the soil and water flow on the garden plot as a whole than to reduce competition at the tree/crop interface. Tree/crop soil water competition was not observed in on-farm trials, nor did farmers consider it to be a problem. It is unlikely that farmers would make large investments to prevent soil water competition at the tree/crop interface. Farmers should be occasionally surveyed to see if tree/crop soil water competition becomes important, and under what conditions. • Analysis of survey data should compare farmers according to how long they have been using particular agroforestry practices and whether or not they are changing the structure and management of the practice. Nonadopters should be included in surveys to better understand how household resources limit adoption. • Survey data should be supplemented with other data collection and interpretation methods, such as farmer focus groups or a small number of ethnographic studies of household economics. • Farmers' assessment of soil fertility appears to be important in the decision to adopt a practice and in how well it is managed. Since improving soil fertility makes agronomic sense, a substantial proportion of farmers with

PAGE 200

185 hedgerows say that soil fertihty is a very important benefit of hedgerows, and since farmers expressing that opinion manage their hedgerows better than those who do not, extension programs should put more focus on developing appropriate ways of increasing soil fertility.

PAGE 201

APPENDIX A POT STUDY OF SOIL WATER DEPLETION VS. MAIZE LEAF WATER POTENTIAL I established a pot study in January 1994 to determine how soil water depletion at the ODH site correlated with plant water stress. I planted 12 pots of maize on 2 January 1994 with soil taken from the top 25 cm at the trial site, one plant per pot. Field capacity of the soil measured at the trial site was about 41%. I stopped watering the maize in the pots on 4 March, 60 DAS. Pre-dawn leaf water stress was measured with a pressure chamber. Figure A1 shows the result of four sets of measurements taken from 3 to 12 days after watering stopped, 63 to 72 DAS. After 72 DAS, the maize leaves were too dry to give pressure chamber readings. Most of the plants were wilted by that time, and some had fired leaves. After the final measurement at 74 DAS, I rewatered six of the pots. Two pots did not recover from the wilting, the other four did. Figure A-1 shows the mean values of the 12 pots as they dry from field capacity to just under 25% soil water over the four observation dates. Figure A-2 shows the regression relationship between leaf water stress and percent soil water. When soil water percent in the root zone decreases below about 25%, leaf water pressure passes below 10 bars. Maize plants begin to suffer drought stress at this point. 186

PAGE 202

187 -SM(%) -PMS 64 66 68 70 Days after sowing 72 74 Figure A-1 : Changes in maize leaf water pressure as soil water percent decreases, 63 to 72 DAS (n= 12) (0 k. re (A w 0) k. M L. 0) re re 0) 35 30 25 20 15 10 I 5 y = 2E+07X-' '°'

PAGE 203

APPENDIX B ROOT DISTRIBUTION OF A LEUCAENA LEUCOCEPHALA HEDGEROW This study was done in November 1990 on a nursery /demonstration site owned by a community church group in the town of Mirebalais, located in Haiti's lower central plateau. There were no repetitions (and therefore no statistical analyses), just one transect 3 m long through one Leaucaena hedgerow planted on a deep, sandy soil of gentle slope (about 5%) on a river terrace. I mapped eight sections 1 .2 m wide by 1 m deep parallel to the hedgerow, beginning 200 cm uphill of the row, cutting through the row of trees, and ending 100 cm downhill. A plexiglass plate inscribed with a 10 by 10 cm grid was pinned to the cut surface. The positions of the roots were drawn on plastic sheets clipped over the Plexiglas plate, a separate symbol was used for each of four diameter classes: very fine (10 mm). From 70 to 96% of the very fine roots were found in the top 30 cm of the soil, the lowest percents being found 150 cm and 200 cm uphill from the hedgerow (Figure B-1). A lesser percent of the fine roots were found in the top 30 cm (20 to 83%), with the least (20%) at the 200 cm uphill position. Very few medium and large roots were found, but of those that were, 50% or more were deeper than 30 cm. 188

PAGE 204

189 100 80 60 O ^ 40 20 upslope Hedgerow I L I I I ]^ ^ very fine Dfine medium H large 200 cm 150 cm 100 cm 50 cm 10 cm 10 cm 50 cm 100 cm Distance from hedgerow Figure B-1. Percent of very fine (10 mm) Leucaena root intersections in the 0-30 cm soil layer, Mirebalais.

PAGE 205

APPENDIX C P-VALUES OF ANOVAS PERFORMED ON STATION

PAGE 206

c c o o 'o (/I n o 3 "o N 'ra S (<. o c I c o c o c o a
PAGE 207

192 3 c o o u X)

PAGE 208

193 T5 3 C O o X)

PAGE 209

194 c D. E o o k. '5 c C3 1) _3 03 > U

PAGE 210

195
PAGE 211

196 1 o V I u i

PAGE 212

APPENDIX D HOUSEHOLD QUESTIONNAIRE

PAGE 213

D O O E D O O (U E E o S o Ei Ql E o c a> E a. o a. 03 <^ Q. Q. V « 2 m o ^ >>^ Is] S o := O c

PAGE 214

199 c n c 0> *^ B n 3 O E 3 O lA C o >l « E o <^ c < CO c .— 2 > s c nj •a ra m a. « a 0) 1 ns o

PAGE 215

APPENDIX E GARDEN PLOT QUESTIONNAIRE

PAGE 216

o 0) in (A 9 S UJ re Si -I o 5: I Lk re < §> ii ?SI

PAGE 217

202 let z < tUJ o I/) 3 -J 0. ^ a < 0.
PAGE 218

9C 203 z < K
PAGE 219

204 ^ s a Bill m W m iTi ir> iTt u^ \r> D D ^lioBS^ am'^tzaK'^B^^ ro ro rn ro ro (NJ f\J CvJ CM OJ iiii li^ lTi ID m lii E V I 1 m m (^ (^ o 6 2

PAGE 220

205 a. in at a> V) z < 3 Q < c & a u o i Z _: 9 > ^

PAGE 221

206 a. i 2 Iff z V) >£. LU V) 3 O < Q. > i o 2 z = S; C n > <

PAGE 222

207 hi < Q to D IT) 0> CO D c/> 3 < Q. D D D _j D lo IT) in in in ld 11. D < D D E

PAGE 223

208 o. 2 u s o —I D D in z < < a. DDD D _j D E £

PAGE 224

a. S 209 S < 3 < 0. D D D D X z hi O

PAGE 225

210 a. z < I(A 3 O < a < z bJ a < 3 o l/> -J < a: O « Z _: 9 X < Q a < < E o z

PAGE 226

APPENDIX F SOIL NUTRIENTS. ORGANIC CARBON STATUS. AND pH OF 175 HILLSIDE GARDENS As a part of the interview that took place during the garden visits (n=2,295), technicians asked the farmers to evaluate several soil parameters of the plot. A qualitative scale of 1 to 5 was used. The parameters were soil fertility (l=poor, 5=ver>' fertile), soil depth (l=shallow, 5=deep), an evaluation of the hot/cold scale' (l=hot, 5=cold), and signs of soil erosion between the rows of soil conservation structures on the plot (l=none, 5=many). In addition, the farmer was asked to give the local Creole name for the soil type, and the technicians noted the type of parent material (calcareous or basaltic). The relationship of Haitian indigenous soil classification systems to the western systems is not well understood (McLain and Stienbarger 1988). Since these qualitative opinions about soil were used as independent variables in analyses of farmers' decision to install and manage various agro forestry practices, it is necessary to understand if they relate to measurable fertility parameters. Method Cross tabulations of farmers" soil names and ratings were done using information from all 2,295 gardens visited. Soil samples were taken for laboratory analysis from 175 'Most writers agree that ""hot" soils {cho) are dry, well-drained soils and "cold" soils {fu'et) have more soil water available for plants. Other concepts are integrated into this system as well, including soil parent material, slope, orientation, and vegetative cover (McLain and Stienbarger 1988, Murray 1981). Some crops grow better in hot soils, some in cold, but in general cold soils are preferred (Smucker 1981). 211

PAGE 227

212 gardens selected at random from all gardens visited, 35 each from gardens having fertility classes 1 through 5. Two samples were mislabeled and could not be used. The samples were taken by technicians using a machete from the 0-20 cm layer: a composite of four positions for each sample. They were put into ziplock bags and labeled, then transported to the University of Florida for laborator}-' analysis. Inductively coupled argon plasma spectroscopy (ICAP) was used with the Mehlich 3 procedure for Ca. Mg, K, and P. ICAP was also used to determine water extractable pH. Percent organic matter was determined using the Walkley-Black procedure. Statistical analyses were done using SPSS; nonparametric tests (chi-square, MannWhitney) because distributions were not normal and numbers of observations were not equal for all factor levels. The 1 to 5 scales were recoded by combining ratings 4 and 5 into one category (the "best" end of the scale) and ratings 1, 2, and 3 into another category (the "worst" end of the scale). The comparisons were therefore fertile vs. not fertile, deep vs. not deep, not eroded vs. eroded, and cold vs. not cold soil. Results Farmers used 21 different names for their soils in the 2,279 gardens for which they gave that information. Of these, four are based on color, six refer to parent material or texture, six refer to the fertility, water-holding ability, or other capacity of the soil to produce crops, and five refer to miscellaneous or unknown properties. Table F-1 summarizes these, and gives the percent of each named soil given the top two fertility ratings (categories 4 and 5).

PAGE 228

213 Table F-1 : Haitian farmer soil classifications Group

PAGE 229

214 ratings shows that 63% of the 431 gardens given the top two cold ratings are also given the top two fertility ratings. However, 41% of the not cold gardens (n= 1,864) are also rated as fertile, so the distinction is not clear-cut.

PAGE 230

215 would probably respond to phosphorus and nitrogen, and possibly micronutrients (R. Tucker, pers. comm.).Table F-2: pH. nutrient, and percent organic carbon for 1 73 gardens pH P (mg/kg) K Mg Ca % OC (t^^/kg) (mg/kg) (mg/kg) Mean 7.7 5.6 143.6 435.5 14,783 1.7 Median 8^0 3^6 117.0 261.0 13.500 1.5 There were statistically significant differences between soils rated as fertile and those rated as not fertile (Table F-3). Fertile soils had a lower pH, more potassium, and more organic carbon. However, since the difference in pH is very small and potassium is abundant in both classes of soils, the difference in organic carbon is the most interesting result. These differences are indicative, but doubtless do not reveal everything farmers integrate into the concept of fertility. Table F-3 : Nutrients, pH, and % organic carbon for soils rated as fertile and not fertile

PAGE 231

216 mg/kg less magnesium than not eroded soils (p = .030), and cold soils had 30.5 mg/kg more potassium than did not cold soils (p = .081). These differences do not appear to be important. Limestone substrate underlies 80% of the land area in Haiti; the rest is basaltic or alluvial (Ehrlich et al. 1985). The soils found in basaltic areas are visually different than those found in limestone areas, they are of a coarser texture and they erode faster. Only limestone (68%) and basaltic (32%) soils were recorded in this survey. Do farmers' fertility ratings differ between these two categories of soils, and are the differences in fertility shown in Table F-3 reflecting the differences in parent material? A cross tabulation revealed that 72% of fertile soils (n=1032) and 65%) of not fertile soils (n=1262) were of limestone origin. This appears to indicate that limestone soils are more likely to be considered fertile. Table F-4 shows the differences as found in laboratory analyses. Table F-4: Nutrients, pH, and % organic carbon for basaltic and limestone soils

PAGE 232

217 and limestone soils is somewhat greater than the difference between soils rated as fertile and not fertile (Table F-3). Conclusions Soil fertility ratings as expressed by farmers in the survey appear to be in agreement with opinions expressed in the literature (Murray 1981. McLain and Stienbarger 1988. Smucker 1981), for example the hot/cold fat/thin scales. They also generally conform to laboratory results, but the soil analysis does not completely explain them. The ratings are relative, subjective, and relate to other nearby plots, not to plots over the whole country. Although farmers rate limestone soils as fertile more often, the basaltic soils in the north of the country often produce better crops because rainfall is higher and more regular. Farmers integrate rainfall, regularity of rainfall, slope, water holding capacity and other undiscovered factors into their fertility ratings. However, their ratings as expressed in the survey appear to be well founded enough to be used in analysis of adoption and management choices.

PAGE 233

REFERENCES Agus F, Garrity DP, Cassel DK and Mercado A ( 1 999) Grain crop response to contour hedgerow systems on sloping oxisols. Agro forestry' Systems 42: 107-120. Akonde TP, Leihner DE and Steinmuller N (1996) Alley cropping on an Ultisol in subhumid Benin. Part 1 : long-term effect on maize, cassava and tree productivity. Agro forestry Systems 34: 1-12. Anderson LS and Sinclair FL (1993) Ecological interactions in agroforestry systems. Agroforestry Abstracts 6: 57-91. Ashley MD (1986) Illustrations of gardens and land use representing agroforestry systems in Haiti. UMO/AFORP. Port au Prince, Haiti: USAID. Balasubramanian V (1983) Alley cropping: can it be an alternative to chemical fertilizers in Ghana? Third national maize workshop. Kumasi. Ghana: Kwadeso Agricultural College. Balzano A (1997) A Haitian community: ten years after. (Unpublished) Banda AZ, Maghembe JA, Ngugi DN and Chome VA (1994) Effect in intercropping maize and closely spaced Leucaena hedgerows on soil coservation and maize yield on a steep slope at Ntcheu, Malawi. Agroforestry Systems 27: 17-22. Bannister ME and Nair PKR (1990) Alley cropping as a sustainable technology for the hillsides of Haiti: experience of an agroforestry outreach project. American Journal of Alternative Agriculture 5: 51-57. Bellerive PA (1991) A financial analysis of selected hedgerow operations in Haiti's southern and northwestern regions, pp. 27-3 1 . SECID/Aubum Agroforestry Report. Belsky AG (1994) Influences of trees on savanna productivity: tests of shade, nutrients, and tree-grass competition. Ecology 75: 922-932. Benge MD (1980) Agro-forestation systems using Leucaena as erosion control barriers. 1980. Washington, DC, Office of Agriculture, Agro-forestation, Development Support Bureau, USAID. 218

PAGE 234

219 Benge MD (1985) Memorandum to USAID/Haiti regarding Leucaena seed dated 25 September 1985. Bloch P. Lambert V and Singer N (1987) Land tenure issues in rural Haiti. Review of the evidence and research and policy agendas. Draft for comments-May 1987. DAN-530 1-1-00-5025-00. 1987. Port au Prince, Haiti: USAID Bohringer A and Leihner DE (1997) A comparison of alley cropping and block planting systems in sub-humid Benin. Agroforestr>' Systems 35: 1 17-130. Bonnard P and Scherr S (1994) Withing gender differences in tree management: is gender distinction a reliable concept? Agroforestry Systems 25: 71-93. Brochet M (1993) Les strategies de lutte contre I'erosion et I'amenagement des bassinsversants en Haiti. Tiers Monde 34: 423-436. Byerlee D and Hesse de Polanco E (1986) Farmers' stepwise adoption of technological packages: evidence from the Mexican altiplano. American Journal of Agricultural Economics 68: 519-527. Cannell MGR, Van Noordwijk M and Ong CK (1996) The central agroforestry hypothesis: the trees must acquire resources that the crop would not otherwise acquire. Agroforestr>' Systems 34: 27-31. Carter J (1995) Alley farming: have resource-poor farmers benefitted? ODI Natural Resource Perspectives. Chamshama SAO, Mugasha AG, Kbvstad A, Haveraaen O and Maliondo SMS (1998) Growth and yield of maize alley cropped with Leucaena leucocephala and Faidherbia albida in Morogoro. Tanzania. Agroforestry Systems 40: 215-225. Chery Y (1990) Les haies vives et leur place dans la production agrocole: Experience du Projet Quest Jacmel. pp. 18-22. Port au Prince. Haiti, Groupe de Recherches, d'Etudes et de Formation en Environment, Secretariat Technique a lAmenagement des Bassens Versants (STABV/MARNDR, USAID). Actes du Colloque sur les Haies Vives pour une Agriculture Continue et Soutenue, 12 December 1989. Chirwa PW, Nair PKR and Kamara CS (1994a) Soil moisture changes and maize productivity under alley cropping with Leucaena and Flemingia hedgerows at Chalimbana near Lusaka. Zambia. Forest Ecology and Management 64: 23 1-243. Chirwa PW, Nair PKR and Nkedi-Kizza P (1994b) Pattern of soil moisture depletion in alley cropping under semiarid conditions in Zambia. Agroforestry Systems 26: 89-99.

PAGE 235

220 Comia RA. Paningbatan EP and Hakansson I (1994) Erosion and crop yield response to soil conditions under alley cropping systems in the Philippines. Soil & Tillage Research 31: 249-261. Corlett JE, Black CR. Ong CK and Monteith JL (1992a) Aboveand below-ground interactions in a leucaenaymillet alley cropping system. II. Light interception and dry matter prodution. Agricultural and Forest Meteorology 60: 73-91 . Corlett JE. Ong CK, Black CR and Monteith JL (1992b) Aboveand below-ground interactions in a leucaena/millet alley cropping system. I. Experimental design, instrumentation and diurnal trends. Agricultural and Forest Meteorology 60: 53-72 Cunard AC and Lorcy MR (1990) Soil erosion estimations from the cropping systems trial at the Brutus site, 1990. Associates in Rural Development. Port au Prince, Haiti: USAID. Current D. Lutz E and Scherr SJ (1995) The costs and benefits of agro forestry to farmers. World Bank Research Observer 10: 151-180. Current D and Scherr SJ (1995) Farmer costs and benefits from agroforestry and farm forestry projects in Central America and the Caribbean: implications for policy. Agroforestry Systems 30: 87-103. Davids (1995) What do farmers think? Farmer evaluations of hedgerow intercropping under semi-arid conditions. Agroforestry Systems 32: 15-28. de Geus JG (1973) Fertilizer Guide for the Tropics and Subtropics. Centre d'Etude de lAzote, Zurich. de Young M (1958) Man and Land in the Haitian Economy. Gainesville, FL. University of Florida. University of Florida Latin American Monographs No. 3. Dijkman MJ (1950) Leucaena~a promising soil-erosion-control plant. Economic Botany 4: 337-349. Dvorak KA (1991) Methods of on-farm, diagnostic research on adoption potential of alley cropping. Agroforestry Systems 15: 167-181. Earle TR. Rose CW and Brownlea AA (1979) Socio-economic predictors of intention towards soil conservation and their implication in environmental management. Journal of Environmental Management 9: 225-236. Ehrlich M, Conway F, Adrien N, LeBeau F, Lewis L, Lauwerysen H, Lowenthal I, Mayda Y. Paryski P, Smucker G, Talbot J and Wilcox E (1985) Haiti Country Environmental Profile: a field study. Port au Prince, Haiti: USAID.

PAGE 236

221 Ehui SK (1988) Evaluating the impact of soil erosion and productivity loss on the relative profitability of alternative land use systems in the tropics. Evidence from Western Nigeria. Lima. Peru, CIP. S econd biennial meeting of the Rockefeller Foundation social sciences research program. Emmanuel PC (1990) Une strategic en matiere de defense et restauration des sols~le cas de la zone Perches/Acul Samedi (Nord-est d'Haiti). Port au Prince. Haiti. Ervin CA and Ervin DE (1982) Factors affecting the use of soil conservation practices: hypotheses, evidence, and policy implications. Land Economics 58: 277-292. Ewel J (1977) A report on soil erosion and prospects for land restoration in Haiti. Port au Prince. Haiti: USAID. Feder G, Just RE and Zilberman D (1985) Adoption of agricultural innovations in developing countries: a survey. Economic Development and Cultural Change 33: 255-298. Feder G and O'Mara GT (1982) On information and innovation diffusion: a Bayesian approach. American Journal of Agricultural Economics 64: 145-147. Fernandes EC, Davey CB and Nelson LA (1993) Alley cropping on an acid soil in the Upper Amazon: mulch, fertilizer, and hedgerow root pruning effects. In: Ragland, J and Lai, RJ (eds) Technologies for sustainable agriculture in the tropics: Proceedings of two international symposia held in San Antonio TX and Denver. CO, 1990 and 1991 respectively, pp. 77-96. American Society of Agronomy. Femandez-Comejo J. Beach ED and Huang W-Y (1994) The adoption of IPM techniques by vegetable growers in Florida. Michigan and Texas. Journal of Agronomy and Applied Economics 26: 158-172. Friedman JRP (1955) Developmental planning in Haiti: a critique of the U.N. report. Economic Development and Cultural Change 4: 39-54 Fujisaka S (1991) Thirteen reasons why farmers do not adopt innovations intended to improve the sustainability of upland agriculture. In: Dumanski. J, Pushparajah. E, Latham, M, and Myers. R (eds) IBSRAM proceedings No. 12 (2). Evaluation for sustainable land management in the developing world. Volume 2: technical papers. Fujisaka S (1993) A case of farmer adaptation and adoption of contour hedgerows for soil conservation. Experimental Agriculture 29: 97-105. Fujisaka S and Cenas PA (1993) Contour hedgerow technology in the Philippines: not yet sustainable. Indigenous Knowledge & Development Monitor 1:14

PAGE 237

222 Fujisaka S, Mercado AR, Jr and Garrity DP (1995) Farmer adaptation and adoption of contour hedgerows for soil coservation. In: Kang, BT. Osiname, AO, and Larbi. A (eds) Ibadan, Nigeria, Alley Farming Network for Tropical Africa. Alley farming research and development. Proceedings of the International Conference on Alley Farming, 14-18 September 1992, Ibadan, Nigeria, pp. 547-555. Gaddanakeri SA, Gumaste SK, Hunshal CS and Angadi SS (1994) Effects of Leucaena live bund on growth and yield of companion crop of rabi sorghum. Journal of Maharashtra Agricultural Universities 19: 244-247. Garrity, DP (1994) The fate of organic matter and nutrients in agro forestry systems. In: Lefroy, RDB. Blair. GJ. and Craswell. ET (eds) Soil organic matter management for sustainable agriculture: a workshop held in Ubon, Thailand. 24-26 August 1994, ACIAR, pp. 69-77. Garrity DP (1996) Tree-soil-crop interactions on slopes. In: Ong CK and Huxley PA (eds) Tree-Crop Interactions, a Physiological Approach, p 299-3 1 8. Wallingford, UK: CABI. Garrity, DP and Mercado, AR, Jr (1994) Nitrogen fixation capacity in the component species of contour hedgerows: how important? Agroforestry Systems 27: 241258. Garrity DP and Van Noordwijk M (1995) Research imperatives in conservation farming and environmental management of sloping lands: an ICRAF perspective. Bogor, Indonesia: ICRAF George SJ, Kumar BM, Wahid. PA and Kamalam NV (1996) Root competition for phosphorus between the tree and herbaceous components of silvopastoral systems in Kerala. India. Plant and Soil 179: 189-196. Ghosh SP, Mohan Kumar B, Kabeerathumma S and Nair GM ( 1 989) Productivity, soil fertility, and soil erosion under cassava based agroforestry systems. Agroforestry Systems 8: 67-82. Govindarajan M, Rao MR, Mathuva MN and Nair PKR (1996) Soil-water and root dynamics under hedgerow intercropping in semiarid Kenya. Agronomy Journal 88:513-520. Haggar JP and Beer JW (1993) Effect of maize growth of [on] the interaction between increased nitrogen availability and competition with trees in alley cropping. Agroforestry Systems 21: 239-249. Harper JK, Rister ME, Mjelde JW, Drees BM and Way MO (1990) Factors influencing the adoption of insect management technology. American Journal of Agricultural Economics 72: 997-1005.

PAGE 238

223 Hauser S and Gichuru MP (1994) Root distribution of Dactyladenia (Acioa) barter i and Senna (Cassia) siamea in alley cropping on Ultisol. II. Impact on water regime and consequences for experimental design. Agroforestr>' Systems 26: 9-21. Hauser S and Kang BT (1993) Nutrient dynamics, maize yield and soil organic matter in alley cropping with Leucaena lencocephala. In: Mulongoy. K and Merckx. R (eds) Soil organic matter dynamics and the sustainability of tropical agriculture: proceedings of an International Symposium, Leuven, Belgium. Hernandez CC (1961) A new method of farming hillsides. Coffee and Cacao Journal 14:5. Herrero MP and Johnson RR (1981) Drought stress and its effects on maize reproductive systems. Crop Science 21:1 051 1 0. Hildebrand PE and Russell JT (1996) Adaptability Analysis: A Method for the Design, Analysis and Interpretation of On-Farm Research-Extension. Ames: Iowa State University Press. Holdridge LR (1947) Determination of world plant formations from simple climatic data. Science 105: 367-368. Isaac L and Shannon DA. (1994) Evaluation of tree species adaptation for alley cropping in four environments in Haiti. B. First year of pruning. SECID/Aubum PLUS Report No. 15. Port au Prince, Haiti. Isaac L, Shannon DA, Brockman FE and Bernard CR. (1995) The effects of Leucaena hedgerow management on maize and hedgerow biomass yields over two years of cropping. SECID/Aubum PLUS Report No. 27. Port au Prince, Haiti. Isaac L. Shannon DA. Brockman FE and Bernard CR (1996) The effects of alley cropping and other soil conservation practices on maize {Zea mays) yields over two years of cropping. SECID/Aubum PLUS Report No. 30. Port au Prince, Haiti. Jama B, Nair PKR and Rao MR (1995) Productivity of hedgerow shrubs and maize under alley cropping and block planting systems in semiarid Kenya. Agroforestry Systems 31: 257-274. Jarvis LS (1981) Predicting the diffusion of improved pastures in Uruguay. American Journal of Agricultural Economics 63: 495-502. Jeanes K W, Gutteridge RC and Shelton HM ( 1 996) Competition effects between leucaena and maize grown simultaneously in an alley cropping system in subtropical Australia. Experimental Agriculture 32: 49-56.

PAGE 239

224 Jonsson K. Fidjeland L, Maghembe JA and Hogberg P (1988) The vertical distribution of fine roots of five tree species and maize in Morogoro, Tanzania. Agro forestry Systems 6: 63-69. Kamasho JA ( 1 994) Agro forestry in the southern highlands of Tanzania. Norwegian Journal of Agricultural Sciences 21, Supplement: 87-92. Kang BT, Reynolds L and Attah-Krah AN (1990) Alley farming. Advances in Agronomy 43: 315-359. Kang BT and Wilson GF (1987) The development of alley cropping as a promising agroforestr}' technology. In: Steppler HA and Nair PKR (eds) Agroforestr>': a decade of development, pp 227-243. Nairobi, Kenya: ICRAF. Kang BT, Wilson GF and Lawson TL. (1984) Alley cropping: a stable alternative to shifting cultivation. Ibadan, Nigeria: IITA. Kang BT, Wilson GF and Sipkens L (1981) Alley cropping maize (Zea mays L.) and leucaena (Leucaena leucocephala Lam.) in southern Nigeria. Plant and Soil 63: 165-179. Kass DL, Araya JS, Sanchez JO. Pinto LS and Ferreira P (1995) Ten years of experience with alley farming in Central America. In: Kang BT. Osiname AO and Larbi A (eds) Alley Farming Network for Tropical Africa. Alley farming research and development. Proceedings of the International Conference on Alley Farming. 14-18 September 1992, pp. 393-402. Ibadan, Nigeria. Kessey JF and O'Kting'ati A (1994) An analysis of some socioeconomic factors affecting farmers involvement in agroforestry extension projects in Tanzania. Annals of Forestry 2: 26-32. Khaleque K and Gold MA (1993) Pineapple agroforestry: an indigenous system among the Garo community of Bangladesh. Society &. Natural Resources 6: 71-78. Kiepe P (1995) Effect of Cassia siamea hedgerow barriers on soil physical properties. Geoderma66: 113-120. Knudson MK (1991) Incorporating technological change in diffusion models. American Journal of Agricultural Economics 73: 724-733. Korwar GB and Radder GD (1994) Influence of root pruning and cutting interval of Leucaena hedgerows on performance of alley croped rabi sorghum. Agroforestry Systems 25: 95-109. Korwar GB and Radder GD (1997) Alley cropping of sorghum with Leucaena during the post-rainy season on vertisols in semi-arid India. Agroforestry Systems 37: 265-277.

PAGE 240

225 Koudokpon V. Brouwers J. Versteeg MN and Budelman A (1994) Priority setting in research for sustainable land use: the case of the Adja Plateau. Benin. Agroforestry Systems 26: 101-122. Lai R (1989) Agroforestry systems and soil surface management of a tropical alfisol: I: Soil moisture and crop yields. Agroforestry Systems 8: 7-29. Lawry S. Stienbarger D and Jabbar MA (1994) Land tenure and the potential for the adoption of alley farming in West Africa. Outlook on Agriculture 23: 183-187 LeaJD (1993) Initial financial evaluation of hedgerows. SECID/Aubum PLUS Report No. 17. Port au Prince. Haiti. Lea JD (1994a) Project PLUS baseline information (revised). SECID/Auburn PLUS Report No. 18. Port au Prince, Haiti,. Lea JD (1994b) Rates of adoption of PLUS project interventions, Northwest Haiti. SECID/Aubum PLUS Report No. 22. Port au Prince, Haiti. Lea JD (1995) PLUS project hedgerow case study report. PADF zones: Gayita. Banat, and Mirebalais; care zone: Bombardopolis. Harvests of December 1994 February 1995. SECID/Aubum PLUS Report. Port au Prince, Haiti. Lea JD (1996) Analysis of CARE 1995 Grande Anse Baseline Data (draft). SECID/Aubum PLUS Report. Port au Prince, Haiti. LeybumJG (1941) The Haitian People. New Haven, Connecticut: Yale University Press. Livesley SJ. Gregory PJ and Buresh RJ ( 1 997) Approaches to modeling root growth and the uptake of water and nutrients. Agroforestry Forum 8: 24-27. LocherU (1996) USAID Food Security Team Report (draft). Port au Prince, Haiti: USAID. Lundahl M (1997) Income and land distribution in Haiti: some remarks on available statistics. Journal of InterAmerican Studies and World Affairs 38: 109-126. MARNDR (1990) Gid plantasyon mayi [Maize planting guide]. Port au Prince, Haiti: Ministry of Agriculture (MARNDR). Pwoje Entansifikasyon Kilit Vivriye/P.I.C.V. McLain RJ and Stienbarger DM (1988) Land tenure and land use in southem Haiti: case studies of the Les Anglais and Grande Ravine du Sud Watersheds. LTC Research Paper No. 95. Madison: University of Wisconsin-Madison.

PAGE 241

226 McNamara KT. Wetzstein ME and Douce GK (1991) Factors affecting peanut producer adoption of integrated pest management. Review of Agricultural Economics 13: 129-139. Mercer DE and Miller RP (1998) Socioeconomic research in agro forestry: progress, prospects, priorities. Agro forestry Systems 38: 177-193. Metzner JK (1976) Lamtoronisasi: an experiment in soil conservation. Bulletin of Indonesian Studies (Canberra. Australia) 2: 103-109. MintzSW (1962) Living fences in the Fond-des-Negres region. Haiti. Economic Botany 16: 101-105. Mugendi DN, Nair PKR, Mugwe JN. O'Neill MK and Woomer PL (1999) Alley cropping of maize with calliandra and leucaena in the subhumid highlands of Kenya. Part 1 : soil-fertility changes and maize yield. Agro forestry Systems 46: 39-50. Mureithi JG, Tayler RS and Thorpe W (1994) The effects of alley cropping with Leucaena leucocephala and of different management practices on the productivity of maize and soil chemical properties in lowland coastal Kenya. Agroforestry Systems 27: 31-51. Murray GF (1979) Terraces, trees, and the Haitian peasant: an assessment of twenty-five years of erosion control. Port au Prince, Haiti: USAID. Murray GF (1980) Haitian peasant contour ridges: The evolution of indigenous erosion control technology. HIID Development Discussion Paper No. 86. Cambridge, Massachusetts: Harvard University. Murray GF (1981) Peasant tree planting in Haiti: a social soundness analysis. Port au Prince, Haiti: USAID. Murray GF (1991) The tree gardens of Haiti: agroforestry among Caribbean peasantsfrom extraction to domestication. In: Challinor D and Frondorf MH (eds) Johns Hopkins University. Social forestry: communal and private management strategies compared. Case studies presented at a conference, February 14, 1991, Johns Hopkins University. Murray GF (1997) A Haitian peasant tree chronicle: adaptive evolution and institutional intrusion. In: Krishna A. Uphoff N and Esman MJ (eds) Reasons for hope: instructive experiences in rural development, pp 241-254. West Hartford, Connecticut: Kumarian Press.

PAGE 242

227 Mutsotso D and Gicheru L (1995) Development/extension in alley farming. In: Kang B T, Osiname. AO and Larbi A (eds) Ibadan. Nigeria. Alley Farming Network for Tropical Africa. Alley farming research and development. Proceedings of the International Conference on Alley Farming, 14-18 September 1992. Ibadan. Nigeria, pp. 507-510. Nadagouda VB, Manjappa K, Desai BK. Palled Y and Kalaghatagi SB (1994) Management o^ Eucalyptus tree rows for sustained agro forestry system. Journal of Maharashtra Agricultural Universities 19: 308-309 Nair PKR (1993) An Introduction to Agroforestry. The Netherlands: Kluwer Academic Publishers. Nair PKR (1998) Directions in tropical agroforestry research: past, present, and future. Agroforestry Systems 38: 223-245. Napier TL (1991) Factors affecting acceptance and continued use of soil conservation practices in developing societies: a diffusion perspective. Agriculture, Ecosystems & Environment 36: 127-140. Neef A. Haigis J and Heidhues F (1996) L'impact du pluralisme institutionnel et juridique sur introduction de mesures anti-erosive-Le cas du Niger [Impact of institutional and legal pluralism on the introduction of erosion control measures— the case of Niger]. Buerkert B, Allison BE and von Oppen, M (eds.) Wind erosion in West Africa: the problem and its control. Margraf Verlag. Proceedings, University Hohenheim, Germany, 5-7 December 1994., pp. 283296. Neef A and Heidhues F (1994) The role of land tenure in agroforestry: lessons from Benin. Agroforestry Systems 27: 145-161. Nelson RA. Cramb RA. Menz KM and Mamicpic MA (1998) Cost-benefit analysis of alternative forms of hedgerow intercropping in the Philippine uplands. Agroforestry Systems 39: 241-262. Nelson RA, Grist PG, Menz KM, Cramb RA. Paningbatan EP and Mamicpic MA (1997) A cost-benefit analysis of hedgerow intercropping in the Philippine uplands using the SCUAF model. Agroforestry Systems 35: 203-220. Norris PE and Batie SS (1987) Virginia farmers' soil conser\'ation decisions: an application of Tobit analysis. Southern Journal of Agricultural Economics 19: 79-90. Nuberg IK and Evans DG (1993) Alley cropping and analog forests for soil conservation in the dry uplands of Sri Lanka. Agroforestry Systems 24: 247-269.

PAGE 243

228 Nye PH and Greenland DJ. (1960) The Soil under Shifting Cultivation. Technical Communication No. 51. Harpenden, UK: Commonwealth Bureau of Soils. Ojyewotu LOZ, Ogigirigi MA and Stigter CJ (1994) A study of competitive effects between a Eucalyptus camaldulensis shelterbelt and an adjacent millet {Pennisetum typhoides) crop. Agriculture, Ecosystems & Environment 5 1 : 281-286. Ong CK. Corlett JE, Singh RP and Black CR (1991) Above and below ground interactions in agroforestry systems, Forest Ecology and Management 45: 45-57 Ong CK and Leakey RRB (1999) Why tree-crop interactions in agroforestry appear at odds with tree-grass interactions in tropical savarmahs. Agroforestry Systems 45: 109-129. Ongprasert S and Turkelboom F (1995) 20 years of alley cropping research and extension in the slopes of northern Thailand. In: Turkelboom F. van LookRothschild K, and van Keer K (eds) Highland farming: soil and the future? Proceedings of a conference, Chaing Mai, Thailand, 21-22 December 1995. OttRL (1992) An Introduction to Statistical Methods and Data Analysis. Belmont. California: Duxbury Press. PADF (1986) Pwoje Pyebwa 1986 Annual Report. Agroforestry Outreach Project. Report for U.S. Agency for International Development (USAID). Port au Prince, Haiti: PADF (Pan American Development Foundation). PADF (1990) Pwoje Pyebwa 1990 Annual Report. Agroforestry II Project. Report for U.S. Agency for International Development (USAID). Port au Prince, Haiti: PADF (Pan American Development Foundation). PADF (1999) PLUS Project 1999 Annual Report. Report for U.S. Agency for International Development (USAID). Port au Prince, Haiti: PADF (Pan American Development Foundation). Paningbatan EP, Ceisiolka CA, Coughlan KJ and Rose CW (1995) Alley cropping for managing soil erosion of hilly lands in the Philippines. Soil Technology 8: 193-204. Pierre F, Lea JD and St-Dic R (1995) Gestion et impacts des haies vives PADF/Camp Perrin. SECID/Aubum PLUS Report No. 24. Port au Prince, Haiti. Rane PV, Sagare BN and Rewatkar SS (1995) Available soil moisture storage and nutrient uptake by cotton as influenced by vegetative barriers. Annals of Plant Physiology 9: 139-141.

PAGE 244

229 Rao MR. Nair PKR and Ong CK (1998) Biophysical interactions in tropical agro forestry systems. Agro forestry Systems 38: 3-50. Rasmussen SD and Shapiro CA (1989) Crop responses to root plowing along multi-row broadleaf windbreak borders. Forty First Annual Meeting. Great Plains Agricultural Council Forestry Committee, June 12-15, 1989, pp. 123-125. Great Plains Agricultural Council. Regis G (1990) Resultats de deux ans de recherches sur les pertes de sols effectuees au Centre de Formation en Amenagement de Limbe. Groupe de Recherches, d'Etudes et de Formation en Environment. Secretariat Technique a I'Amenagement des Bassens Versants (STABV/MARNDR, USAID). Actes du Colloque sur les Haies Vives pour une Agriculture Continue et Soutenue. Proceedings of a conference held at Cap Haitien, 12-15 December, 1989, Port au Prince, Haiti, pp. 63-76. Rogers EM (1995) Diffusion of Innovations. New York: The Free Press. Romanoffs, Voth D and Douglas M (1995) Further assessment and refinement of the PLUS M&E system. SECID/Aubum PLUS Report No. 25. Port au Prince, Haiti. Roose E and Ndayizigiye F (1993) Agroforestry, water and soil fertility management to fight erosion in tropical mountains of Rwanda. Soil Technology 11: 109-1 19. Salazari A. Szott LT and Palm CA (1993) Crop-tree interactions in alley cropping systems on alluvial soils of the Upper Amazon Basin. Agroforestry Systems 22: 67-82. Sanchez P A (1995) Science in agroforestry. Agroforestry Systems 30: 5-55. Scherr SJ (1994) Agroforestry in the East African highlands: a sustainable strategy for land-use intensification. Research & Exploration 10: 145-157. Scherr SJ (1995) Economic factors in farmer adoption of agroforestry: patterns observed in western Kenya. World Development (Oxford) 23: 787-804. Scherr SJ and Current D (1997) What makes agroforestry profitable to farmers? Evidence from Central America and the Caribbean. Agroforestry Today 9: 10-15. Schroth G, Balle P and Peltier R (1995a). Alley cropping groundnut with Gliricidia sepium in Cote d'lvoire: effects on yields, microclimate and crop diseases. Agroforestry Systems 29: 147-163. Schroth G and Lehmann J (1995) Contrasting effects of roots and mulch from three agroforestry tree species on yields of alley cropped maize. Agriculture, Ecosystems & Environment 54: 89-101.

PAGE 245

230 Schroth G, Poidy N. Morshauser T and Zech W (1995b) Effects of different methods of soil tillage and biomass application on crop yields and soil properties in agroforestry with high tree competition. Agriculture, Ecosystems &. Environment 52: 129-140. Schroth G and Zech W (1993) Root length dynamics in agroforestry with Gliricidia sepium as compared to sole cropping in the semi-deciduous rainforest zone of West Africa. Plant and Soil 170: 297-306. Shannon DA and Isaac L (1993) Evaluation of tree species adaptation for alley cropping in four environments in Haiti. A. Establishment phase. SECID/Aubum. PLUS Report No. 6. Port au Prince, Haiti. Shaw RH (1988) Climate Requirement. In: Sprague GF and Dudley JW (eds) Com and Com Improvement. Madison, Wisconsin: ASA/CSSA/SSSA Publishers. Shively GE (1998) Impact of contour hedgerows on upland maize yields in the Philippines. Agroforestry Systems 39: 59-71. Shultz S, Faustino J and Melgar D (1997) Agroforestry and soil conservation: adoption and profitability in El Salvador. Agroforestry Today 9: 16-17. SiederP (1994) Haiti-traum und wirklichkeit. Ein tragisches beispiel aus der forstlichen enwicklungshilfe [Haiti-dream and reality. A tragic example of forestry development aid]. Forst und Holz 49: 40-46. Sinclair FL (1999) A general classification of agroforestry practice. Agroforestry Systems 46: 161-180. Singh RP, Van den Beldt RJ and Hocking D (1986) Korwar GR. Alley farming in the semi-arid regions of India. In: Kang BT and Reynolds L (eds) Alley farming in the humid and sub-humid tropics. Proceedings of an international workshop held in Ibadan, Nigeria, IDRC. Singh V. Rajbahadur. Hooda MS and Balkrishan (1996) Pattern of root distribution in 30-month old five tree and two shrub species of an arid region of northwestern India. Indian Forester 122: 323-330. Singhai RM and Kumar V (1997) An analysis of some socio-economic implications influencing people's participation in silvi-pastoral systems of a micro watershed in Garhwal Himalaya. Indian Forester 123: 136-142. Sitompul SM. Syekhfani MS and Van der Heide J (1992) Yield of maize and soybean in a hedgerow intercropping system. Agrivita 15: 69-75.

PAGE 246

231 Smucker GR (1988) Decisions and motivations in peasant tree farming: Mome Franck and the PADF cycle of village studies. Port au Prince. Haiti: PADF/Haiti Agro forestry Outreach Project. Smucker GR and Dathis N (1998) Organisations Paysannes en Haiti. Tendences et Implications. Port au Prince, Haiti: Editions La Ruche, MES Impressions et Publicite. Smucker GR and Thomson J (1999) Social capital and development in Haiti. Report to ADF-Asosye. Port au Prince, Haiti: ARD-PACTE, ASSET. Smucker GR and Timyan J (1995) Impact of tree planting in Haiti: 1982-1995. SECID/Aubum PLUS Report No. 23. Port au Prince, Haiti. Smucker GR, White TA and Bannister ME (2000) Land tenure and the adoption of agricultural technology in Haiti. International Food Policy Research Institute (IFPRI). Solera CR (1993) Determinants of competition between hedgerow and alley species in a contour hedgerow intercropping system. Doctoral dissertation. University of the Philippines at Los Banos. Ssekabembe CK, Henderlong PR and Larson M (1994) Soil moisture relations at the tree/crop interface in black locust alleys. Agroforestry Systems 25: 135-140. Stark M (1996) Adoption and adaption of contour hedgerow farming in the Philippine uplands: results of an early case study. Tropenlandwirt 97: 3-16. Steiner KG and Scheidegger U (1994) Participatory technology development for soil fertility management in tropical highlands, Montpellier, France. In: Recherchessysteme en agriculture et developpement rural: Symposium international. 21-25 Novembre 1994, pp. 310-315. Sumberg JE and Atta-Krah AN (1988) The potential of alley farming in humid West Africa—a re-evaluation. Agroforestry Systems 6:163-168. Sunderlin WD ( 1 997) An ex-post methodology for measuring poor people's participation in social forestry: an example from Java, Indonesia. Agroforestry Systems 37: 297-310. Sureshwaran S, Londhe SR and Frazier P (1996) A logit model for evaluating farmer participation in soil conservation programs: sloping agricultural land technology on upland farms in the Philippines. Journal of Sustainable Agriculture 7: 57-69. Swanson RA (1993) Farmer needs assessment exploratory surveys. Field information acquisition guide and methodology. SECID/Aubum PLUS Report No. 8. Port au Prince, Haiti.

PAGE 247

232 Swanson RA. Gustave W, Jean Y and Conde G (1993a) Farmer needs assessment exploratory surveys. PADF Cap Haitien Region 3. SECID/Auburn PLUS Report No. 9. Port au Prince, Haiti. Swanson RA, Gustave W, Jean Y. Saint-Die R (1993b) Farmer needs assessment exploratory surveys. CARE Northwest Regions 2, 3, & 4. SECID/Auburn PLUS Report No. 10. Port au Prince. Haiti. Swanson RA. Gustave W, Jean Y. Saint-Die R (1993c) Farmer needs assessment exploratory surveys. PADF Jacmel Region 2. SECID/Auburn PLUS Report No. 1 1 . Port au Prince, Haiti. Swanson RA. Gustave W, Jean Y, Saint-Die R (1993 d) Farmer needs assessment exploratory surveys. PADF Les Cayes Region 1. SECID/Auburn PLUS Report No. 13. Port au Prince, Haiti. Swanson RA, Gustave W, Jean Y, and Saint-Die R (1993e) Farmer needs assessment exploratory surveys. PADF Mirebalais Region 4. SECID/Auburn PLUS Report No. 12. Port au Prince, Haiti. Tacio HD (1993) Sloping agricultural land technology (SALT): a sustainable agroforestry scheme for the uplands. Agroforestry Systems 22: 145-152. Tassin J, Perret S and Cattet R (1995) Alteration in the porosity of an andisol in Reunion under a hedge of Calliandra calothyrsus. Bois et Forets des Tropiques 245: 91-99. Tilander Y, Ouedraogo G and Yougma F (1995) Impact of tree coppicing on tree-crop competition in parkland and alley farming systems in semiarid Burkina Faso. Agroforestry Systems 30: 363-378. Tonye J, Duguma B and Tiki-Manga T (1994) Stepwise approach to alley cropping technology development and transfer in the forest zone of Cameroon. Agroforestry Systems 28: 269-278. Tonye J and Titi-Nwel P (1995) Agronomic and economic evaluation of methods of establishing alley cropping under a maize/groundnut intercrop system. Agriculture, Ecosystems & Environment 56: 29-36. Toussaint RJ (1989) Les haies vives arbustives: etablissement, gestion et production— contraintes et limitations techniques dans I'aire du Projet LORD I a Maissade. Port au Prince, Haiti. Groupe de Recherches, d'Etudes et de Formation en Environment, Secretariat Technique a I'Amenagement des Bassens Versants (STABV/MARNDR. USAID). Actes du Colloque sur les Haies Vives pour une Agriculture Continue et Soutenue. Proceedings of a conference held at Cap Haitien. 12-15 December, 1989, pp. 101-113.

PAGE 248

233 Treadwell BD and Cunard AC (1992) Perennial herbaceous leguminous plants as permanent contour land improvements for Haitian hillside farms: results & observations from a 2-year field study and current recommendations, pp. 14-24. Port au Prince, Haiti, Associates in Rural Development. Pwoje Sove Te Livestock Working Document Series. Tripathi BR and Psychas P (1992) The AFNETA alley farming training manual. Volume 1 . Core course in alley farming. Addis Ababa: ILCA. United Nations Development Programme (1996) PNUD Haiti-Econet [newsletter] January 1996, vol. 2, no. 1. USAID/Haiti (1996) US AID Haiti Monitoring Report. Vol. 4, No. 6. Port au Prince, Haiti: USAID/Haiti. Vandermeer J (1998) Maximizing crop yield in alley crops. Agroforestry Systems 40: 199-206. Vaval F, Soto Pinto L, Nelson K and Serrano Galvez E (1997) La percepcion de los productores sobre el impacto de la practica agroforestal "barrera viva" en Maissade, Plateau Central, Haiti. VillanuevaMR (1993) A review of PDAI and ADSIl project technologies. SECID/Aubum PLUS Report No. 2. Port au Prince, Haiti. Weaver J (1920) Root development in the grassland formation. A correlation of the root systems of native vegetation and crop plants. Washington, DC: Carnegie Institution of Washington. Weil TE, Black JK, Blutstein HI. Johnston KT, McMorris DS and Munson FP (1973) Area Handbook for Haiti. Washington, DC: U.S. Government Printing Office. Welke SE, Tanyi E and Kang BT (1995) Effect of Leucaena leucocephala (Lam.) de Wit pruning frequency on alley cropped maize/cassava. In: Kang BT, Osiname AO and Larbi A (eds) Alley Farming Network for Tropical Africa. Alley farming research and development. Proceedings of the International Conference on Alley Farming, 14-18 September 1992, Ibadan, Nigeria, pp. 21 1-225. White TA (1992a) Peasant cooperation for watershed management in Maissade, Haiti: factors associated with participation. EPAT/MUCIA Working Paper No. 4. White TA (1992b) Peasant initiative for soil conservation: case studies of recent technical and social innovations from Maissade, Haiti. EPAT/MUCIA Working Paper No. 3.

PAGE 249

234 White TA and Quinn RM (1992) An economic analysis of the Maissade. Haiti. Integrated Watershed Management Project. [Working Paper No. 2, July 1992], 24. 1992. EPAT/MUCIA. Wiersum KF (1994) Farmer adoption of contour hedgerow intercropping, a case study from east Indonesia. Agro forestry Systems 27: 163-182. Wirjodarmodjo H and Wiroatodjo P (1983) Leucaena leucocephala: the Indonesian experience. Bangkok: FAO Regional Office for Asia and the Pacific. Yadav JP, Sharma KK and Khanna P (1993) The effect of Acacia nilotica on mustard crop. Agroforestry Systems 21: 91-98. Young A and Muraya P ( 1 990) Soil Changes Under Agroforestry. Computer program with User's Handbook. Version 2. Nairobi, Kenya: ICRAF. ZarJH (1984) Biostatistical Analysis. Englewood Cliffs. New Jersey: Prentice-Hall, Inc. Zoschke A, Ebner L and Quadranti M (1990) Conservation farming in Malaysia-first experiences with alley cropping. Malaysian Plant Protection Society 6: 272-283. Zuvekas C (1979) Land tenure in Haiti and its policy implications: a summary of the literature. Social and Economic Studies 28: 1-30.

PAGE 250

BIOGRAPHICAL SKETCH After serving four years in the U.S. Navy, Michael Bannister earned a B.S. and Master of Forestry (M.F.) in forest management from Oregon State University in 1976 and 1 98 1 , respectively. Between the two degrees he served as a Peace Corps volunteer in Guatemala, establishing two community pine/oak nurseries and promoting soil conservation. In 1981 he accepted a position working in agro forestry extension in Haiti with the Pan American Development Foundation (PADF). PADF has implemented three USAID-funded agroforestry projects in Haiti from 1981 through 2000. Mr. Bannister worked on all three projects as regional team leader, research and documentation coordinator, and assistant director for agroforestry; the latter position he still holds. Since 1981 he has also done short consultancies for PADF in the Dominican Republic, Honduras, El Salvador, and Panama. Mr. Bannister began his Ph.D. at the University of Florida in 1987, taking three years off from his professional work for required classes. He then returned to Haiti as a full-time employee of PADF while simultaneously doing his field research, analysis, and much of the writing. This process, aggravated by political and civil turmoil in Haiti, has taken 14 years. 235

PAGE 251

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy p. K. Ramachandran Nair, Chair Professor of Forest Resources and Conservation I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy HiJ^ Kenneth L. Buhr Assistant Professor of Agronomy I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy 'Ai^-v MarylL. Diryea \ j Profes^r-<5f Forest Resources and Conservation I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully_aileqjuate, in scope and quality, as a dissertation for the degree of Doctor of PhilosopF Hildebrand Professor of Food and Resource Economics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy Gerald Murray Murray Associate Professor of Anth/opology

PAGE 253

This dissertation was submitted to the Graduate FacuUy of the School of Forest Resources and Conservation in the College of Agricultural and Life Sciences and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. May 2001 yCiyOy^ /^ //wc^d^^^^^lDirector/rorest Resources and Conservation Dean, Graduate School

PAGE 254

82BF1 07/30/0i ^^