An ecological analysis of soil and water conservation in hillslope farming systems


Material Information

An ecological analysis of soil and water conservation in hillslope farming systems Plan Sierra, Dominican Republic
Plan Sierra, Dominican Republic
Physical Description:
xiii, 420 leaves : ill., maps ; 28 cm.
Rocheleau, Dianne E
Publication Date:


Subjects / Keywords:
Farms, Small -- Economic aspects -- Dominican Republic   ( lcsh )
Soil conservation -- Research -- Dominican Republic   ( lcsh )
Water conservation -- Research -- Dominican Republic   ( lcsh )
Geography thesis Ph. D
Dissertations, Academic -- Geography -- UF
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1984.
Bibliography: leaves 394-419.
Statement of Responsibility:
by Dianne E. Rocheleau.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 030502015
oclc - 11698486
System ID:

Full Text








This work could not have been completed without the kindness and

able assistance of family, friends and professional associates too

numerous to mention. Special thanks are due to Dr. Gustavo Antonini

for his confidence in me, his personal concern and his active support

of my work; to Dr. Katherine Ewel for her guidance and encouragement

throughout my interdisciplinary program at the University of Florida;

to Dr. Helen Safa for her moral support, financial and intellectual

contributions during the last year of dissertation work; to Dr. Robert

Marcus and Dr. James Henry for their careful review and suggestions;

to Dr. H. T. Odum for sparking my imagination; and to Dr. Manuel

Paulet for direction and support in Santo Domingo.

Several offices of the University of Florida contributed

financial and logistic support, including the Graduate School, the

Center for Latin American Studies, the Department of Geography, and

the International Programs Office of the Institute of Food and

Agricultural Sciences. The latter covered research expenses in the

field with fund provided by a Title XII grant from USAID. My research

stipend was provided by a traineeship from the OAS for a period of 18

months. The State Secretariat of Agriculture, principally through

Plan Sierra, provided extensive logistic and financial support for the

field research, as well as employment for a period of six months.

My research benefitted substantially from the professional

dedication and warm friendship of many colleagues at Plan Sierra,

especially Angel Liriano S., Victor Montero, and Geuris Martinez.

Invaluable data and assistance were also provided by the National

Hydrology Institute of the Dominican Republic (INDRHI), the National

Cartographic Institute, the Dominican Electric Company (CDE), and the

Departments of Meteorology and Land and Water, of the State

Secretariat of Agriculture (SEA).

Final data processing, drafting, and typing tasks were completed

with the able assistance of Nelly Mogallon, Kim Feigenbaum, Beth

Higgs, and Pat French.

The warmest appreciation is reserved for those closest to home,

especially Mickie, Nelly, Gustavo, Marie, Mom and Dad. More than any

other, I owe the successful completion of this work to my husband,

Luis, who helped me tap my own energies and gave selflessly of his

time and effort as computer consultant, data processing technician,

editor, critic, and nurturer.




ACKNOWLEDGMENTS ................................................ ii

LIST OF TABLES ................................................. vi

LIST OF FIGURES.................................................. ix

ABSTRACT....................................... ................ xii

CHAPTER I INTRODUCTION..................................... 1

The Problem.. .................................... 1
The Sierra Region..................o.................. 7
Purpose and Scope of Work............................. 11

PROBLEM ................... ............ ............ 15

Overview ................................. ....... 15
Erosion and Sedimentation Research in Geomorph-
ology, Biogeochemistry and Ecology................. 15
Models of Erosion and Sedimentation.................. 17
A Review of Relevant Findings in Experimental
Watersheds and Erosion Plots.................... 24
Qualitative and Informal Analyses of Land Use
and Erosion in the Caribbean and Similar Environ-
ments............ ......................... .... .... .... .. 44
The Role of Farming Systems Research in Soil and
Water Conservation.............................. 45
Farming Systems, Agroecosystems and Agroforestry
Research ............. ................. ...... ...... 46
Central American Research............................ 51

CHAPTER III METHODOLOGY....................................... 58

The General Approach............................. 58
Materials and Methods.......................... ... 60

CHAPTER IV RESULTS AND DISCUSSION.............................. 101

Regional Profile................................. 101
Study of Large Watersheds..........o............. 130
Study of Small Watersheds......................... 164
Erosion Plot and Household Studies.................. 207












Application of Erosion and Runoff Coefficients
to the Small Watersheds ........................
Sediment Delivery Ratios .........................

CONCLUSIONS ......................................



INDRHI .................................... .......







DATA FROM EROSION PLOTS..........................

LITERATURE CITED ...............................................

BIOGRAPHICAL SKETCH ............................................

















1 Relative location and land cover of erosion plots........ 65

2 Equations for system model................................ 123

3 Land use systems in the region............................ 126

4 Rainfall and river discharge in the Amina and Mao
Watersheds ................................................ 144

5 R values for regression analyses of subwatershed
rainfall vs. river discharge and sediment load........... 146

6 Summary of regression analyses of river discharge and
sediment concentration.................................... 155

7 Sedimentation transport in Mao and Amina Rivers.......... 158

8 Sedimentation in the Mao River basin estimated from May
1980 measurements......................................... 160

9 Water and sediment yields estimated from 1980 and 1981
data...................................................... 162

10 Characteristics of the small watersheds................... 167

11 Physical characteristics of erosion plot sites........... 179

12 Sediment transport in five watersheds..................... 193

13 Discharge rates measured for low flow conditions in small
watersheds ................................................ 194

14 Flood events yielding peak sediment discharge during the
study period .............................................. 195

15 Analysis of variance of stream discharge and sediment
transport for all streams................................. 200
16 Maximum recorded concentrations (g L ) per flood event
for all streams............................................ 201
17 Average sediment concentration (gm L ) per flood event,
for all streams.. .......................................... 202


18 Analysis of variance of stream discharge and sediment
transport comparing streams draining coffee stands and
streams draining food crops and pastures .................. 204

19 Peak discharge ha per flood event for all streams,
results of the a posteriori test of the means............. 205
20 Peak sediment discharge rate ha per flood for all
streams, results of the a posteriori test of the means.... 206

21 Land use, land tenure and production, by household........ 208

22 Total annual storm runoff and soil loss rates, by plot.... 219

23 Relationship of total annual rainfall and storm runoff
in erosion plots........................................... 221

24 Analysis of variance and runoff and sediment losses for
all plots at site Los Montones............................. 229

25 Runoff and sediment losses for plots at the Los Montones
site: Results of the aposteriori tests of the means...... 230

26 Analysis of variance of runoff and sediment losses for
plots grouped by land use at the Los Montones site........ 232

27 Runoff and sediment losses for plots at the Los Montones
site, results of the Duncan Multiple Range Test, by
land use.................................................. 233

28 Analysis of variance of runoff and sediment losses for
all plots at site Pananao.................................. 236

29 Runoff and sediment losses for plots of the Pananao site,
results of the Duncan Multiple Range Test, by land use.... 237

30 Analysis of variance of runoff and sediment losses for
plots grouped by land use at site Pananao................. 238

31 Runoff and sediment losses for plots of the Pananao site,
results of the Duncan Multiple Range Test, by land use.... 239

32 Analysis of variance of runoff and sediment losses for
plots in pasture grouped by site........................... 240

33 Analysis of variance of runoff and sediment losses of
forested plots grouped by site............................. 241



34 Analysis of variance of runoff and sediment losses for
plots planted in crops grouped by site..................... 242

35 Runoff and sediment losses for plots in Los Montones and
Pananao sites with crops, results of the Duncan Multiple
Range Test........................................ ......... 243

36 Soil loss coefficients by land use and conservation
practice ......................................... ........ 247

37 Comparison of measured and predicted erosion losses....... 252

38 Comparison of soil loss coefficients derived from
USLE and from empirical data.............................. 255

39 Storm runoff coefficients, total runoff estimates and
runoff/rainfall ratios, by watershed...................... 271

40 Calculation of soil loss coefficients and soil loss
estimates by watershed.................................... 272




1 Systems model of land use and erosion in the Caribbean.. 5

2 Impact area: Plan Sierra. 9

3 Model of applied research process........................ 25

4 Flow chart of research activities........................ 26

5 Input-output diagram for interview notations and
monitoring............................................... 53

6 Organization of research activities...................... 64

7 Diagram of Thiessen polygons superimposed on a map of
the Mao and Amina watershed subdivisions................. 77

8 Sampling sites for large watersheds...................... 78

9 Uppsala-type manual sampler for instantaneous measure-
ment of sediment concentrations in streams.............. .81

10 A. Illustration of velocity-area method. Person "a"
releases float at time t1 and person "b" records time
it takes float to move 2 m. B. Cross-section of
stream showing placement of float to measure velocity
and area of three sections of the stream................ 86

11 Equipment installed in streams to measure sediment
concentrations at different levels of flooding.......... 88

12 Illustration of erosion plot with runoff and sediment
collectors............................................... 93

13 Illustration of alternative erosion plot design with
three subsections .......................................... 95

14 Diagram of sediment and runoff collector indicating the
points at which samples were taken....................... 97

15 Plan Sierra geologic subregions.......................... 103

16 Plan Sierra region with study sites...................... 104


17 Plan Sierra life zones................................... 106

18 Plan Sierra land use systems............................. 109

19 System model of the Sierra............................... 120

20 Land use model........................................... 125

21 Monthly discharge of the Mao River....................... 132

22 Monthly discharge of the Amina River..................... 133

23 Monthly rainfall at the San Jose de Las Matas
climatological station, #1............................... 136

24 Monthly rainfall at the Moncion climatological station,
#2 ....................................................... 137

25 Monthly rainfall at the San Jose climatological station,
#1, for two periods...................................... 138

26 Monthly rainfall at Moncion climatological station, #2,
for two periods of record................................ 139

27 Plan Sierra region: Mean annual rainfall for 1967-1979. 142

28 Plan Sierra region: Annual rainfall for 1980........... 143

29 Time series of sediment concentrations during selected
flood events ....... ..................................... .. 153

30 Small watersheds in coffee region........................ 165

31 Land use in the Prieto watershed......................... 168

32 Land use in the larger Prieto watershed.................. 169

33 Land use in the Upper Bajamillo watershed............... 170

34 System model of small watershed in coffee region........ 172

35 Land use model for watershed model of coffee producing
region................................................... 173

36 Land use in Hondo watershed.............................. 184

37 Land use in Pananao watershed............................ 185

38 System model of small watershed in pasture-field crop
association .............................................. 187


39 Land use model of small watershed model of pastures-
field crops association.................................. 188

40 Pananao watershed........................................ 190

41 Hondo watershed.......................................... 191

42 Illustration of equivalent sediment discharge according
to distinct regimes...................................... 198

43 Infiltration rates in erosion plots at Carrizal and
Pananao ................................................... 213

44 Infiltration rates in erosion plots at Los Montones..... 214

45 Monthly rainfall, storm runoff, and sediment loss at
Carrizal plots ........................................... 222

46 Monthly rainfall, storm runoff, and sediment loss at
Los Montones plots........................................ 223

47 Monthly rainfall, storm runoff, and sediment loss at
Pananao plots............................................ 225

48 Model of coffee farms: Large and small holdings........ 257

49 Model of dairy and cattle farm, Pananao.................. 260

50 Bitter manioc production on a small holder plot, Pananao 261

51 Model of a mixed production on a well integrated farm... 264

52 Evaluated small watershed submodel of land use, erosion,
and sedimentation: Prieto stream........................ 266

53 Evaluated small watershed submodel of land use, erosion,
and sedimentation: Upper Bajamillo stream.............. 267

54 Evaluated small watershed submodel of land use, erosion,
and sedimentation: Greater Bajamillo stream............ 268

55 Evaluated small watershed submodel of land use, erosion,
and sedimentation: Hondo stream......................... 269

56 Evaluated small watershed submodel of land use, erosion,
and sedimentation: Pananao stream....................... 270

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




August 1984

Chairman: G. A. Antonini
Major Department: Geography

The purpose of the study was to develop and test an

interdisciplinary methodology for applied research in soil and water

conservation in hillslope farming systems. The specific objectives

were to collect baseline data on erosion and sedimentation in the

Sierra region of the Dominican Republic and to evaluate soil

conservation technologies and cropping systems recently introduced

into the area.

Erosion, runoff and sedimentation were measured at three scales

of analysis: on-farm experimental plots (22 x 2 m ), small watersheds

2 2
(1-30 km ), and large watersheds (300 km ). Erosion losses in the 16
-I -1
plots ranged from 0.05 to 0.10 tons ha yr under pine forest, 1 to
-i -i -I -i
3 tons ha yr in pasture, 0.5 to 3 tons ha yr in coffee
-1 -l
plantations, and 6 to 70 tons ha yr for plots in mixed food crops.

Annual storm runoff varied from 1% of precipitation under forest to

12% in an eroded continuously cropped plot. Infiltration and soil

profile analyses and erosion measurements at the plots showed a


pronounced influence of intensity and longevity of cultivation at the

site. Erosion rates also changed dramatically with the phases of the

cropping cycle on the coffee and food crop plots.

The recently introduced slope modification (hillside ditches) did

not significantly reduce erosion rates at the test plots. The minimum

tillage field trial showed more substantial reductions.

Field data were combined with photogrammetric analyses of land

use to compare erosion and sedimentation rates at the watershed level

for production systems based on coffee and pasture with annual crops,

respectively. The contribution to river sedimentation and flooding

did not differ significantly between the two systems at the watershed

level. In both cases it is the association of annual crops with the

dominant commercial land use that determines the erosion rate. The

planting of coffee for soil conservation was ineffective in most cases

because of high labor and related food crop demands.

Suggested alternatives to coffee include reduction of competition

for land and labor between commercial and subsistence production by

substituting tree crops that meet household demands at the local and

national level. Recommendations for land in pasture and annual crops

are the reduction of tillage in annual crop plots, the mixture of more

food crops into annual cash cropping systems, and the combination of

grazing and tree crops in large holdings.



The Problem

Almost 10 years after the first dramatic success of the Green

Revolution, the technological breakthroughs remain largely

inaccessible to small farmers of the underdeveloped world. Increased

yields have occurred primarily in large scale commercial or state

enterprises (Harris, 1973; Greenland, 1975; Stevens, 1977). Millions

of small farmers who produce commercial and subsistence food crops and

cash crops for export have maintained or increased production only at

great cost to themselves and to the natural environment (Crosson and

Frederick, 1977; Eckholm, 1976).

These farmers have expanded into ever more marginal areas (often

arid, semiarid, or hillslope environments) or have intensified

cultivation of existing plots, often already located in marginally

productive lands (Brush, 1981). The intensification has been for the

most part without benefit of new technology or capital inputs. It is

achieved through increasingly higher inputs of labor, and often

through practices that damage the long-term fertility and stability of

the soil (Geertz, 1972; Lagemann, 1977) and disrupt hydrologic and

geologic cycles in watersheds (Greenland, 1974; Kellman, 1969;

Pereira, 1973; Rapp, 1977).

Given this situation there is an immediate need for research to

adapt technologies to needs of small farmers within the limits of the

available factors of production and environmental constraints (Crosson

et al., 1978; Hildebrand, 1981; Lagemann, 1977; Makhijani and Poole,

1975; Novoa and Posner, 1981). The "external costs" of watershed

degradation must be considered and tested within the context of such

research (Erickson, 1974; Novoa and Posner, 1981). Beyond the need

for an inventory of the current magnitude of the problem, there is a

real need to describe and test the complex functional relationships

between land use, production, erosion and sedimentation under field

conditions, in order to explore viable alternatives to the causes of

the problem.

It is widely recognized that the combined processes of erosion

and sedimentation pose a serious threat to sustained production in

both upper and lower portions of tropical watersheds (Farvar and

Milton, 1973). The decreasing depth and fertility of the soil limit

the productivity of small upland farms (Greenland and Lal, 1977;

Morgan, 1979), while quantity, quality, and regularity of surface

water supplies limit both urban and agricultural development in the

lowlands downstream (McPherson, 1974; Nelson, 1973; Odum and Odum,

1976). This is especially critical in tropical and subtropical humid

montane environments subject to intense population pressure by farmers

using traditional and semitraditional methods (Antonini et al., 1975;

Floyd, 1969; Santos, 1981; Sheng and Michaelson, 1973; Wilson, 1976).

While subsistence farmers, landless laborers and their

traditional technology are often blamed for deforestation and

environmental degradation under such circumstances (Rodriguez, 1980),

they are constrained by limited access to land and lack of

alternatives (Plumwood and Routley, 1982). The settlement of rugged

hillslopes and the near total deforestation of upland watersheds

represent a choice by default rather than a free choice between

rational alternatives for sustained production. Moreover, there is

intense pressure for continuous cropping and/or establishment of

pastures on cleared land. Large local landowners, as well as urban

and foreign markets, play a major role in this process (Amin, 1977;

Cultural Survival Inc., 1982, Hildyard, 1982; Nations and Komer, 1982;

Plumwood and Routley, 1982).

Small farmers and shifting cultivators in such areas often

practice forms of management that can be sustained well at lower

population densities or within more hospitable environments to which

they have no access (Bailey, 1982; Grainger, 1980; Nations and Komer,

1982). Although these small farmers in marginal areas are referred to

by many as "subsistence farmers," they usually produce some surplus

food crops. In some cases this sector is a major source of staple

food production for domestic markets (Brush, 1981; Novoa and Posner,

1981). The same farm families often function as a seasonal labor

force in coffee, lumber, and other cash crop harvests (Beckford, 1972;

Frucht, 1967). These subsistence farmer/farmworker populations in

marginal lands highly susceptible to erosion form an integral part of

the regional economy and ecosystem. As such, the problem must be

treated as a complex phenomenon that not only affects the larger

downstream and lowland production systems, but is partially

conditioned by them.

The question remains as to how production can be maintained or

increased (at a sustained rate) with minimum damage to both portions

of the watershed. The need for an answer to this question is

particularly urgent in the Caribbean because of high population

pressure on hillslope lands (Antonini et al., 1975; Santos, 1981)

coupled with high demand for food crops and relegation of large lowland

tracts to cash crop cultivation (Beckford, 1972; Rankine, 1976; Wilson,


The resource base in the Caribbean is subject to intensifying

multiple demands by commercial and subsistence sectors for food, cash

crop, wood, fuel and mineral production, as well as protection of the

watershed for downstream development. From a national perspective, the

upland watershed's most important export crop may well be water, needed

for irrigation and hydroelectric projects for downstream development

(Swedforest, 1980). The various types and rates of production demanded

are often competitive in nature, if not mutually exclusive (Crosson and

Frederick, 1977; McPherson, 1974). In many cases the upland regions'

internal situations also clearly indicate the need for change (Brush,

1981; Chaney and Lewis, 1980; Ferreiras, 1979; Hildebrand, 1981; Reiche

and Lee, 1978; Santos, 1981; SEA, 1978).

A Conceptual Model of the Problem

Based upon the information presented above, a model is postulated

that describes the interaction of significant elements in Caribbean

land use systems with regard to the problems of soil erosion and rural

poverty (Fig. 1). The model shows the interaction between population,

land use and the condition of soil, water, and vegetation both within

and outside the study region.

The model diagram follows the format developed by Odum for energy

modelling of ecosystems (Odum, 1971). The tank-shaped symbols (Fig. 1)


represent storage of land, soil, biomass, water, and economic assets

within the system. The bullet-shaped symbol indicates a

transformation process (in this case photosynthesis), while the

circles define forcing functions from outside the agroecosystem. The

arrow-shaped devices are workgates, which regulate the interaction

between the various mass and energy flows. Changes in the storage

are determined by the input and output flows, indicated by solid

lines. The imports and exports from the system are indicated


In this case the export of soil (soil erosion/sedimentation) is

of special interest, along with changes over time of crops, other

vegetation storage, and-human population. The hypothesis implicitly

stated in the model is that upland land use controls both the export

of soil from the system, and the production of food and income for the

population, both in the uplands and the lowlands.

The problem remains to reconcile internal regional development

goals with national priorities to define desirable changes. At best,

the conflicting needs will be met by careful optimization of land and

water use within entire river basins (Pereira, 1973; White, 1977).

The determination of what is optimal implies a client group within a

defined spatial and temporal context. Decisions must be made to

reconcile differences in "optimal" solutions at local and regional

scales, as well as to adjust short-term and long-term costs and

benefits. Policy makers also need information on the distribution and

nature of costs and benefits among various sectors of the population

in order to formulate "optimal" strategies. Resulting rural

development programs must include short- and long-term incentives at

the local level for the widespread adoption of resource management

practices that will benefit the population of the region as a whole

over the long term (Gladwin, 1981; Hildebrand, 1981; Santos, 1981).

Traditional cost-benefit analysis will not suffice since it

excludes environmental as well as social aspects of the system, assumes

a static system, and has been developed for application over relatively

short time periods (Amin, 1977). The problem requires a more holistic

theoretical and methodological approach that will evaluate

environmental and human concerns on their own terms and within the

total system rather than by econometric criteria.

The Sierra Region

The Sierra is a rugged montane area in central Dominican Republic

that has been subjected to the traditional practice of shifting

cultivation, as well as to extensive exploitation of primary resources

such as timber and mineral deposits. It is a relatively underdeveloped

region within an underdeveloped country where the area under production

in the country (27,000 km ) already has surpassed the area of land

classified as suitable for agriculture (22,000 km ) (OAS, 1967;

Swedforest, 1980; USAID, 1974). Production increases necessary to meet

the national demands for food and income have come from increased

yields in areas already under production, or from expansion into more

marginal areas such as the Sierra. The latter strategy has dominated

among the poor and landless members of the peasantry (Beckford, 1972;

Antonini et al., 1975), and has been a last resort for the former

employees of mining and sawmill camps and furniture shops, most of

which had closed by 1979 (Ferreiras, 1979). The major alternative,

emigration to the capital city of Santo Domingo, and to New York City,

provided an outlet for a large segment of the population during the

1960s and early 1970s.

The impact area of Plan Sierra (Fig. 2), an integrated rural

development project within the Sierra, offered a unique opportunity to

examine the apparent conflicts between agricultural development and

natural resource conservation (Santos, 1981). Plan Sierra is a joint

venture between the State Secretariat of Agriculture and the private

sector to initiate and coordinate development efforts in the region in

several sectors: agriculture, livestock, credit associations, health

services, transportation, handicrafts, university programs, and

natural resource management (Quezada, 1977; Antonini and York, 1979;

Plan Sierra, 1979). The Plan's objectives are: 1) to improve the

quality of life of the inhabitants of the Sierra; 2) to manage soil,

water and forest resources; and 3) to promote participation by local

people in the development process (Antonini and York, 1979; Santos,


Several hydroelectric and irrigation projects are planned for the

study area (Jorge, 1981). The impoundments will serve the Cibao

Valley downstream. Sedimentation from this area is already a problem

in the Tavera Reservoir, completed in 1977 (Cepeda, 1980). The

magnitude and distribution of erosion in the uplands, however, has

received little attention until recently (Vasquez, 1980).

Plan Sierra has promoted specific farming practices and changes

in land use that are intended to reduce sediment export. A parallel


I V / ^ -^^^"^^ So-- 0E

iX 2S^\ ^
o eo


0 4-)
1 0
o M

objective of the program is the development of viable sustained yield

agricultural systems. Existing land use in the region consists of a

combination of forest, coffee, bush, small plots of annual crops, and

pastures. The relative rates of erosion and sediment yields from the

prevailing land use systems in this area have been estimated, but not

measured (Antonini et al., 1975). Neither have the effects of erosion

within the upland system been investigated. In the absence of basic

data Plan Sierra had to experiment with tentative conservation

practices and crop changes to curtail erosion and to maintain or

improve production. The widespread adoption of hillside ditches and

various forms of terracing was strongly encouraged by credit

incentives to large and small landowners alike. The conversion of

pasture, bush, and crop land to coffee also was promoted through

credit incentives and supported by extension and education programs.

The research project described in the following chapters sought

to define the magnitude of the sedimentation and erosion problems in

the region, to measure the variation of erosion rates under existing

land use systems, and to test the effectiveness and feasibility of

specific land use changes and conservation practices promoted by Plan

Sierra. The study was integrated into the Plan Sierra soil, water and

conservation program. Logistic support as well as the active

collaboration of paratechnical, technical and professional staff made

it possible to conduct the study at a regional scale and with an

intensity of effort that would otherwise have proven difficult if not


Purpose and Scope of the Work

The purpose of the study was to develop and test an

interdisciplinary methodology that addresses the needs cited above

within the context of the Caribbean. Such a methodology, and the

theoretical framework within which it is developed, should meet the

following criteria:

1. Treat the problem in a holistic way, building upon existing

research in several disciplines, and incorporating both physical and

cultural aspects of erosion and land use problems;

2. Be flexible enough to include analyses within a broad range

of temporal and spatial scales of observation and to allow for

differentiation by client groups;

3. Facilitate the participation of clients in research and

extension programs;

4. Be applicable in areas with limited data bases;

5. Yield practical short-term results at the topographic scale

while working toward more basic solutions over the long-term;

6. Be amenable to integration into rural development programs,

from preliminary research to subsequent extension efforts.


The specific objectives of the study included short-term

practical achievements at the local and regional level. The latter

were readily accomplished within the larger task of methodology

development and testing and contributed to the development of the

theoretical aspects of this study. The objectives, stated in

chronological order of completion, were as follows:

1. The study will provide useful information and services for

farmers, rural communities and regional policymakers in the study

area, the Sierra region of the Dominican Republic. This includes the

collection of basic data about the magnitude, distribution, impact and

causes of the problem. (In this case, the basic data will include

knowledge and perceptions of the residents about the study area and

the problems of erosion and sedimentation.) Further, the study will

develop and adapt practical field methods and data analysis techniques

that are suitable for future use by local personnel trained within the

project. The project should become a vehicle for farm extension and

community education in management of natural resources.

2. The study will develop and test a theoretical model of the

relationship between land use, runoff, erosion and sedimentation in

the Sierra. The hypotheses implied in the model will be tested and

the results will contribute to the evaluation and modification of the

model. The refined model will be used as a point of departure for

future simulation and for planning of field trials in the study


3. A general version of the resultant model will be proposed for

application to the problem in the Caribbean. The methodology for

testing the model and applying it to research and extension efforts in

the region will be presented.

The Theoretical Context of the Study

Systems analysis provides a theoretical framework and analytical

tools appropriate to the proposed study and adequate for the complex

interdisciplinary nature of the problem. It allows the researcher to

focus on the interactions and mutual causality so difficult to include

in other analyses and so critical to understanding the links between

land use, poverty, and erosion. The relationship of the upland areas

to the larger watershed and the national context is also readily

included within this perspective by using the concept of nested

hierarchies in systems of various scales (Antonini et al., 1975; Odum

and Odum, 1976). This feature of systems analysis also allows the

inclusion of boundary conditions on subsystems that may be determined

by the larger system or some elements thereof. As such, the

"structural determinants" of land use and production relationships

(Beckford, 1972) can be considered within what has often been viewed

by structural determinists as a noncompatible methodology. The

theoretical assumptions underlying this approach are in fact

consistent with the world-system paradigm in anthropology (Nash,


Moreover, within the broad framework of systems analysis it is

possible to combine elements of such fruitful and diverse approaches

as the ecosystems-energy analyses developed by Odum and Odum (1976)

and the farming systems research and extension methodology developed

by Hildebrand (1981). Other related lines of research which can be

incorporated include the basic watershed-ecosystems analysis (Hart,

1980), agroecosystems testing and development conducted by Ewel

(1981), farming systems research conducted by Ruthenberg (1976), Flinn

(1980), and Lagemann (1977), and regional characterization of tropical

agroecosystems (Posner et al., 1981).

Systems analysis offers the potential for experimenting with a

variety of "futures possible" through simulation of dynamic nonlinear

models (Ewel et al., 1975; Forrester, 1971; Lagemann, 1977; Meadows et

al., 1972; Odum, 1971). Such analyses allow us to "optimize" over

various time scales and according to different criteria. Ideally,

this provides planners, policymakers, and their clients with

information about the probable outcome of different resource

management alternatives; the net result can be viewed from the

perspective of the farm household to the national level.

The accuracy and quality of the conclusions drawn from the

qualitative or quantitative analysis of such models depend upon the

accurate conceptual design of the model and the quality of the

information used to evaluate its components (Amadeo and Golledge,

1975; Harvey, 1969; Kuhn, 1962). The model, as a selective

simplification of reality, must be constructed by one who is famil-iar

enough with both the system-at-large, and the problem of interest, to.

choose which elements to include, and to describe their relationships


A researcher trained in ecosystem analysis and modelling can

perform this combined function by modelling the entire system and by

adding questions of ecological concern and ecological monitoring

methods to interdisciplinary adaptive research in farming systems and

resource management. A general reconnaissance survey of the region,

and interaction with researchers from complementary disciplinary

backgrounds, as well as with project clients, can provide the kind of

information and broad perspective needed to properly define the




Various aspects of the relationship between erosion,

sedimentation and land use have been treated by geographers as well as

by researchers in several other fields. Although some geographers

(Antonini et al., 1975; Haggett, 1961; Kellman, 1969; More, 1967;

Morgan, 1979; Pereira, 1973) and other scientists (Geertz, 1972;

Greenland, 1974, 1975; King, 1978; Lagemann, 1977) have spanned both

ends of the research spectrum, the literature is generally divided

along these lines and will be reviewed as such.

Erosion and Sedimentation Research in Geomorphology,
Biogeochemistry and Ecology

The theoretical aspects of erosion and sedimentation mechanics at

the micro-scale are fairly well established and understood and already

have been summarized for general reference use by applied scientists

(Brady, 1974; Chow, 1964; Gregory and Walling, 1973; Morgan, 1979;

Toy, 1977; USDA, 1979). What are less clear, and still require a wide

range of basic and applied, theoretical and empirical research, are

the more complex causes and effects of erosion and sedimentation at a

larger scale, under field conditions.

The study of erosion and sedimentation and the mechanisms that

determine their occurrence at the meso- and regional scales under

field conditions is a proper subfield of physical geography since

geomorphology is a traditional and well-developed avenue of inquiry

within the discipline (Gregory and Walling, 1973; James, 1972; Keller,

1968; Morgan, 1979; Stoddart, 1965; Strahler, 1964). Systems theory

has been applied widely in studies of watersheds and other

geomorphological units by British geographers (Chisholm, 1967;

Chorley, 1962; Kirkby, 1978; Oilier, 1968) as well as by numerous

other geomorphologists (Leopold et al., 1964; Leopold and Langbein,

1962; Toy, 1977) and hydrologists (Chow, 1964; Vemuri and Vemuri,

1970). This general approach offers the advantage of integrating form

and process, by accounting for their interactive relationship

(Chorley, 1962, 1969).

The open systems approach allows the inclusion of the

quantitative hydrology/geomorphology tradition that dates from Horton

(1935, 1938) and continues in the work of Strahler (1964) and other

physical geographers. The use of systems theory in geomorphology also

facilitates the study of the human use of the earth, a longstanding

focus of geographic research (Harvey, 1969; James, 1972; Marsh, 1964;

Thomas, 1974) that could lend itself well to quantification within a

systems framework (Stoddart, 1965).

The application of systems theory to the study of watersheds has

also been tested and developed within ecology in recent studies of

biogeochemistry (Bormann and Likens, 1979; Likens et al., 1977) and

resource management (Cooper, 1971; Hall and Day, 1977; Hopkinson and

Day, 1980; Patton, 1971; Thomas, 1974; Van Dyne, 1969). In such

studies the watershed defines the boundaries of the ecosystem and the

interacting elements whose mutual influences are observed within the

system include physical, chemical, biotic and cultural entities (Odum,

1971, 1982; Odum and Odum, 1976).

The general conceptual framework of systems analysis is well

suited to the study of mutual influence between land type, land use,

plant and animal productivity, and erosion in the upland watersheds of

the Caribbean. Incorporating the broad perspective of the man/land

tradition in geography, the research in open systems geomorphology and

ecosystem analysis also offers valuable methodological examples for

application to interdisciplinary research on the topic.

Models of Erosion and Sedimentation

The range of erosion and sedimentation models in use includes

stochastic and deterministic models, statistical as well as

parametricc" models, and combinations of all of the above. These

models have been developed and applied within scales of analysis

ranging from individual plots to large watersheds.

One school of research focuses on modelling processes and

interactions between the various parts of watersheds as complex

systems (Chorley, 1962; Likens et al., 1977), and the other major

thrust of erosion and sedimentation studies has been to develop

empirical predictive equations (Wischmeier, 1975; Wischmeier and

Smith, 1978). The latter relate land use and management to erosion

loss and sediment yield and have been developed for management

purposes, primarily for use in soil and water conservation programs.

Empirical Models

The best known and most widely used of the empirical models of

erosion loss is the Universal Soil Loss Equation (USLE), an empirical

formula for estimating potential soil loss by sheet and rill erosion

for individual plots (Wischmeier, 1975; Wischmeier and Smith, 1978).

The USLE is used primarily to predict erosion losses based on a

combination of inherent site characteristics and variables subject to

human intervention and management. Tons of soil lost is the dependent

variable. It is determined by a combination of six independent

variables as follows:

A = RKLSCP, where
-i -i
A = soil loss per unit area (tons acre or tons (m) ha ) over

a given time

R = erosive potential of rainfall (based on total energy and

intensity of rainfall)

K = an index of soil erodibility (a measure of soil suscepti-

bility to erosion based on physical properties)

L = length-of-slope factor

S = inclination-of-slope factor

C = vegetative cover factor, integrating type of cover and

management of crops or other vegetation

P = conservation practice, including structural alteration of

site and/or contour planting.

The apparent disadvantages of this model reside mainly in the

costly and time consuming local calibration required for its proper

application. This is particularly important in underdeveloped

countries where research funds and personnel are limited and the

complexity of the rural agricultural landscape in the uplands requires

extensive calibration of the model. Rapid changes in farming

practice, crop types and level of technology in many areas would

require almost a continuous update of the calibration experiments.

Such a program of research not only represents a large investment in

and of itself, it implies a diversion of resources from alternative

avenues of theoretical and applied research directed toward cumulative

growth of knowledge about the processes in question.

Another weakness of the USLE is the failure of the conceptual

framework to account for the difference between land cover and land

use systems. There are no economic or social aspects to the model,-.

yet it is proposed as a practical tool for farm and regional

conservation planning purposes. The premises under which the equation

is applied often can be misleading, resulting in serious errors in

planning and management decisions. Aside from the logical pitfalls

inherent in the use of the model, there is the technical drawback of

its inability to predict the feedback effects of current erosion rates

on future land use and productivity which in turn affect future

erosion rates.

The USLE can be a useful tool for prediction of erosion rates

under specific known conditions, given prior calibration for the full

range of conditions in a region. Used alone, however, it does not

constitute an adequate basis for management decisions at the farm

level, much less for regional planning purposes.

The USLE has also been adapted for prediction of sediment yields

at the watershed scale (McElroy et al., 1976; Onstad and Foster, 1975;

USDA Forest Service, 1980; Williams, 1975) by estimating the sediment

delivery ratio based on drainage area (Holtan and Lopez, 1971; Roehl,

1962). The models derived from the USLE have been used widely in

economic and land use planning studies for evaluation of specific

cropping systems and/or conservation practices (Kling and Olson,

1975). The weaknesses of the approach include the inherent

limitations of the USLE as well as the questionable realism of

sediment routing techniques (Skopp and Daniel, 1978).

One empirical model that does not derive from the USLE is used by

the United States Bureau of Reclamation (Flaxman, 1975; Skopp and

Daniel, 1978; Strand, 1975) as well as by international research

organizations (Rapp, 1977). The technique combines flow duration

curves with sediment concentration, the latter derived from either a

sediment rating curve or a power function. Both flow duration and

sediment rating curves describe empirical relationships that must be

determined on site.

The model is useful for prediction of sediment yield over the

long term, given a continuation of current land use conditions, but is

less amenable to integrated watershed management based on land use and

land treatment programs. The flow duration/rating curve model,

however, could be calibrated to particular watersheds or groups of

watersheds for specific land uses and treatments, as has been the case

in paired watershed studies conducted in experimental catchments.

A number of sophisticated digital computer models of runoff

and/or erosion and sedimentation predict water and sediment discharge

by relating hydrologic and physical characteristics of the source

areas (Chapman and Dunin, 1975; Mein, 1977; Fleming, 1968; Negev,

1967; Ebumive and Todd, 1976; Donigan and Crawford, 1976).

Theoretical Models of Physical Processes

Research conducted by Elwell (1979a) in Zimbabwe resulted in a

simple model similar to the USLE in some aspects, but based on

rational rather than empirical parameters for estimating sheet erosion

from arable land. Rainfall energy measured in 10-day increments

defines erosivity while the protective value of crops and cropping

practices is assessed according to the percentage of seasonal rainfall

energy "i" intercepted by the vegetative canopy and ground cover.

Potential interception is determined in a fashion similar to the leaf

area index measured by ecologists (Odum, 1971).

Application of the model by Elwell (1979a) in Zimbabwe field

trials confirmed the importance of mulches for erosion control and

demonstrated the potential for reducing soil loss by increasing crop

yields. Moreover, important seasonal relationships were identified

between various protective crop covers and erosivity of rainfall in

given areas. This allowed the identification of the crops or crop

combinations that best protected the soil during periods of intense


The model is particularly applicable to the seasonally dry

tropics where single storm events cause a large proportion of the

total annual erosion loss. The practical application of the model is

aided by a description of the techniques for calculation and field

measurement (Elwell, 1979a).

Many of the theoretical models describing erosion and sediment

transport processes are mechanistic approximations of reality

(Bennett, 1974). Most of these constructs take systems theory as a

point of departure, although either combined or dispersed systems may

be postulated. The combined system models assume a uniform watershed,

with average physical features and composite land use, and do not

consider distance between source and outlet. In dispersed systems the

runoff and sediments are generated in spatially distinct source areas

and are then routed spatially and/or chronologically through

successive increments of the watershed (Holtan and Lopez, 1971;

Fleming and Leytham, 1976; Frere, 1978; Renard and Lane, 1975; Skopp

and Daniel, 1978).

The models within the theoretical research tradition generally

offer better conceptual realism than the empirical models (Holtan and

Lopez, 1971) but the theoretical models tend to be much stronger in

the explanation of physical and chemical processes and remain largely

underdeveloped in the treatment of biological and cultural phenomena.

These models are just now beginning to be tested in systems that

include human populations, and management options are being added to

the range of variables to be tested. Meanwhile, the vast majority of

field data collected on erosion, sedimentation and land use has come

from empirical studies designed to calibrate predictive equations that

are, essentially, site-specific black box models of complex systems

and processes (Boughton, 1967; Hayward, 1967; Holtan and Lopez, 1971;

Skopp and Daniels, 1978).

Ecosystem Models of Watershed Processes and Land Use

Two types of ecosystem models that are less frequently applied to

erosion and sedimentation studies warrant special consideration for

their broad applicability and holistic approach. The first is a model

developed to analyze the biogeochemistry of a forest ecosystem at the

Hubbard Brook Experimental Forest in New Hampshire (Likens et al.,

1977). The second is an energy model used to simulate the complex

interaction of population, land use and sedimentation in the Central

Mountains (Cordillera Central) of the Dominican Republic (Antonini et

al., 1975).

The ecosystem analysis studies conducted at Hubbard Brook focus

on quantification of nutrient budgets through monitoring of

meteorological inputs and geologic outputs of nutrients in small

watersheds. The model used for the study places a strong emphasis on

precipitation, runoff, and solute and sediment export from the areas

of interest (Likens et al., 1977). The variables are measured first

under primary undisturbed forest cover, and later under disturbed

conditions. The final product of this analysis is an annual

hydrologic and nutrient budget that includes sediment yields.

Another form of ecosystem analysis is based on the evaluation of

energy pathways within communities of plants and animals, and on the

relationship between those pathways and the physical environment

(Antonini et al., 1975; Odum, 1971). Recent studies using the energy

approach have extended this concept to include the complex transfers

of matter, energy and information in ecosystems that include human

populations. Diverse interactions, ranging from mineral cycling,

photosynthesis, runoff and sedimentation to economic transactions, are

interrelated by the common denominator of energy flow (Antonini et

al., 1975; Odum, 1971; Rappaport, 1971).

Models using the energy flow language developed by Odum have been

applied to analyses of watershed ecosystems in several environments.

The most pertinent case is a model of the interaction between land

use, erosion and sedimentation in the rugged uplands of the Dominican

Republic (Antonini et al., 1975). As illustrated in the diagrams

(Figs. 3 and 4), the models are dynamic non-linear systems models that

can be simulated on an analog or a digital computer by simultaneous

solution of differential equations. The equations describe the change

in the landscape variables over time. The model indicates mutual

causality between variations in land use and population but does not

account for similar relationships between erosion and land use, since

erosion is not considered as a separate process. The model simulation

demonstrates the implications of current and alternative trends in

land use for future conditions of population, land use and reservoir

sedimentation. The two-part model developed in this study provides

the conceptual point of departure for construction of a single energy

flow model of hillslope land use systems, upland erosion and

downstream sedimentation in the Caribbean.

A Review of Relevant Findings in Experimental Watersheds
and Erosion Plots

During the past 10 years the international scientific community

and policymakers at various levels have focused greater attention on

the problems of erosion and sedimentation, particularly in the tropics


neati jve I

EX P N' L Pr' ':ES




(formal representat.Iion
oF thile izIo()


(definition, classifica-
Lion, mcasuremnent)


f -

success f I


posi tive


Fig. 3. Model of applied research process (after Harvey,


(statistical tests,


H ilt-1

Test best alternatives
at other sites

Fig. 4. Flow chart of research activities.

(Brown, 1980; Henkes, 1982). However, a large proportion of the

published literature on the topic refers to studies conducted in the

United States. While the findings and methods are not all directly

applicable to the Caribbean, they supplement the less extensive data

base and cumulative research experience available at present from

Third World and Caribbean sources.

Summary of Recent Research in the United States

Three landmark studies in experimental watersheds have set the

methodological and technical trends for studying the impact of land

use on sediment yields in small watersheds. The Coshocton, Ohio,

watershed studies (Harrold et al., 1962; Mustoneu and McGuiness, 1968)

focused primarily on sedimentation rates under various cropping and

land treatment conditions in agricultural areas. Research conducted

at Coweeta, North Carolina, and Hubbard Brook, New Hampshire,

addressed both methodological and theoretical aspects of watershed

biogeochemistry (including land treatment) in forested ecosystems

(Bormann and Likens, 1979; Douglass and Swank, 1975; Likens et al.,

1977). These experiments served as models for a series of applied

watershed studies recently initiated in forests, agricultural land,

and rangeland.

The more recent research has been conducted primarily under the

auspices of federal environmental legislation that mandates

documentation of non-point sources of pollution, including sediment

discharge (Haith and Dougherty, 1976; USDA Forest Service, 1980;

Jewell and Smith, 1976; Rao, 1980; Reikerk et al., 1978). This

research focuses more on entire watersheds and less on individual

plots, in contrast with prior work conducted by the Soil Conservation

Service at the farm level (Ackerman, 1966). Most of the watershed

experiments combine monitoring of precipitation with recording rain

gauges, continuous monitoring of stream discharge at weirs, sampling

of sediment discharge at weirs, and/or collection of sediments in weir

ponds in watersheds with areas less than 20 km2 (Hewlett et al., 1969;

Ward, 1971).

A major topic of the earlier studies was the role of the

undisturbed forest in regulating the hydrologic cycle and sediment

export (Douglass and Swank, 1975; Helvey, 1967; USDA Forest Service,

1980). Among the more important findings were: the importance of

litter versus canopy in protecting the soil against the erosion

potential of rainfall (Table A-I); the impact of forest vegetation on

stream discharge (Dils, 1957; Johnson and Swank, 1973) (Table A-2);

and the association between undisturbed forest cover and low sediment

concentration in streams (Table A-3).

Streamflow, sediment concentrations and mass transport from

forested watersheds showed dramatic changes after harvesting, various

site clearing and management operations, or conversion to other uses.

Several studies reported heavy increases in suspended sediment and

nitrate concentrations after clearcutting (Bormann and Likens, 1979;

Douglass and Swank, 1975; Hewlett and Nutter, 1969; Likens et al.,

1977; Monk, 1976).

Water yield increments proportional to percent area in cleared

openings were reported for several gauged watersheds (Likens et al.,

1977; Sopper, 1975). This was attributed to increased storm flow as

well as to reduced water consumption by evapotranspiration. Reported

streamflow increases in Georgia, South Carolina, and Oregon ranged

from 40 to 50% (USDA Forest Service, 1980).

The combination of increased sediment concentration and higher

streamflow resulted in dramatic increases in sediment transport after

clearing (Monk, 1976; Sopper, 1975). By contrast, little or no change

was reported in a patch-cut watershed in the Oregon Coast Range, while

much higher figures are reported elsewhere with use of conventional

treatments (Brown, 1982; Monk, 1976).

Results from east Texas forested watersheds identify sediment as

the major pollutant in streamflow. At the same sites, in a study of

harvest practice and related impact on water quality and mass

transport, the highest sediment transport rates occurred in

association with harvesting along streams. The design, construction

and maintenance of roads often are cited as major determinants of

water quality in forested watersheds (Fredriksen, 1970; Pavoni, 1977;

Texas A&M, 1979; Ursic, 1978).

Water yield and mass transport of minerals and sediments from

nonforested watersheds differ markedly from forested sites. A study

of nutrient yields from various categories of land use in the

watersheds of 24 Connecticut lakes estimates contributions of

phosphorus from agricultural and residential-commercial land at 200

and 1100%, respectively, of forest contributions (Hill and Frink,

1978). National averages of erosion rates for various categories of

land use in the United States (Table A-4) agree well with the results

from the Connecticut study. Transport rates from harvested forests

are extremely high, even in comparison to agricultural uses. The high

rates, however, are offset by the fact that forest harvests are

periodic events that only occur once every 15 to 40 years, even in the

fast-growing pine plantations of the Southeast and the Caribbean. By

contrast, many agricultural uses are sustained continuously on a given

parcel of land.

Recent studies in agricultural watersheds in the United States

concentrate on cropping systems and practices in large scale

commercial farms or ranches. Pollution of surface waters by chemical

fertilizers (Haith and Dougherty, 1976) and pesticides often

overshadows sediment pollution as a subject of public concern (USDA

Soil Conservation Service, 1980). Pathogens entering the waterways

from feedlots and grazing lands (Jewell and Smith, 1976) also attract

more attention, although sediment is the major pollutant, by volume,

discharged into surface waters from agricultural lands. Suspended

sediments in streams and rivers carry pathogens as well as chemical

pollutants. Much of the current research, however, emphasizes the

chemical by-products discharged into waterways in solution from

agricultural non-point sources (Rao, 1980).

Studies of erosion in individual plots offer more information on

the variation in erosion rates with changes in cropping systems, farm

management and conservation practices. Data analyses by Wischmeier

and Smith (1978) for cropped and clean-tilled plots corroborate the

conclusions of studies in forest ecosystems. The canopy cover and

ground cover on the site determine how much rainfall energy reaches

the soil surface. As in the forest, both canopy and mulch (comparable

to litter) intercept raindrops, but mulch does this so close to the

surface that the drops regain no fall velocity. Mulches also obstruct

runoff flow and reduce its velocity and sediment transport capacity

(Wischmeier and Smith, 1978). Surface roughness of the soil also

influences the velocity and transport capacity of runoff. Thus,

tillage practices, crop yields and crop rotations, as well as above-

and below-ground architectural characteristics of particular plants,

influence the degree to which given cropping systems reduce erosion

relative to a standard clean-tilled plot (Mannering and Meyer, 1963;

Meyer and Mannering, 1961).

Annual row crops vary widely but erosion rates generally range

between 5 and 50% of those measured in clean-tilled plots. For

example, a field tilled with chisel and disk plows, rotated from wheat

to meadow to corn, with one crop annually, loses an average of 9% of

the total loss for a control plot (Wischmeier and Smith, 1978).

Pasture, rangeland, bush and woodland reduce erosion to between 1

and 10% of clean-till soil losses, and undisturbed forest further

reduces the loss to between 1 and 0.01% of the bare fallow. Losses

under harvested, mechanically prepared woodland sites, however, vary

between 10 and 90% of the clean-till figures. Average annual erosion

losses for cropland in the United States bear out the trends reported

by Wischmeier and Smith (1978) (Table A-5).

The same results have been extrapolated to the watershed level

with the use of sediment delivery ratios (USDA Soil Conservation

Service, 1980; Wischmeier and Smith, 1978). The proportion of total

eroded soil that arrives at a given outlet ranges from 33% in a 1 km2

2 2
area to 18% for a 25 km area to 10% for a 250 km drainage area

(Roehl, 1962).

Studies in watersheds and erosion plots in the United States

cannot be extrapolated directly to tropical and subtropical montane

watersheds. Beyond the difference in the ecosystems themselves there

is the even greater divergence in level of technology and the greater

complexity of land cover associations within upland land use systems

of the tropics. The general relationships established between land

cover, erosion, and sedimentation must be tested further and compared

with results from the Caribbean and similar regions.

Erosion and Sedimentation Research in the Caribbean and Similar

Although few in number, studies of erosion, sedimentation and

land use have been conducted in the Caribbean in Jamaica, Puerto Rico,

Trinidad-Tobago and the Dominican Republic. Information also can be

found from study areas in New Zealand, Australia, Yugoslavia, Kenya,

Tanzania, Zimbabwe, Malaysia, the Philippines, Costa Rica, Guatemala,

and Colombia. These span a wide spectrum of environmental and

economic conditions, but each has in common some combination of

topographic, climatic, cultural and cropping system characteristics

with the upland forest and farming areas of the Caribbean. The

methods and materials used for watershed monitoring and other data

collecting activities offer proven alternatives to the more

sophisticated, expensive experiments in the United States and other

developed nations (Pereira, 1973).

Watershed and sedimentation studies

Erosion, sedimentation and land use research in tropical

hillslope lands falls into two major categories. The watershed

studies focus primarily on the interaction of climate, topography, and

land use throughout the drainage area in determining river regimes and

erosion and sedimentation rates. By contrast, reservoir sedimentation

studies focus on the identification of sediment source areas as well

as on the immediate protection of reservoir facilities, which often

presupposes an emphasis on the development of physical infrastructure

at various points throughout the watershed.

Representative basins and experimental catchments. Studies

initiated under the auspices of the International Geophysical Year

(IGY) and the International Hydrological Decade account for a large

proportion of the work conducted in the tropics. The representative

basin studies emphasize comparative description of diverse river

schemes and watershed ecosystems on a global basis, while experimental

catchment studies focus more attention on the effects of alternative

land cover and land treatment in a given area.

Among the more important findings to date are the contrasting

characteristics of tropical climate and hydrology when compared to the

more temperate regions. Water balance data, including ratios of

runoff and evapotranspiration to total precipitation, are available in

reports from empirical studies (Golley, 1972; Holdridge, 1967, 1982;

Odum 1970b; Pereira, 1973; Thornthwaite and Mather, 1959). The

seasonal distribution as well as the amount of precipitation varies

substantially from the pattern of temperate areas. Bimodal rainfall

and river discharge distributions are common. The proportion of total

precipitation that leaves the watershed as evapotranspiration is

higher than in temperate areas and the ratio of surface water

discharge to total precipitation generally is lower, except when

deforestation occurs. The overall amount as well as the intensity of

rainfall usually is higher, making the erosive potential greater than

in most areas of temperate or cooler climates. High erosive potential

of climate often coincides with high erodibility of soils in the

seasonally dry tropics (Elwell, 1979a). The international comparative

statistics on river sedimentation bear out the implications of high

erosivity combined with readily eroded soils and high population

densities in such areas (Douglas, 1968; Holeman, 1968; Stoddard,

1965). Measurements of sediment yield in several catchments in

Malaysia demonstrate the relationship of land use to sediment

transport (Table A-6).

In addition to providing baseline information, the representative

basin and experimental catchment research tests various methods,

materials and modelling strategies. Experimental catchment studies in

New Zealand (Campbell, 1962; New Zealand Ministry of Works, 1968a,

1968b, 1968c, 1970) and Australia (Australian Water Resources Council,

1969; Chapman and Dunin, 1975) emphasize methodology and techniques,

from mapping of sediment sources (Mosley, 1980) and evaluation of

suspended sediment data (Campbell, 1962) to calibration of catchment

models (Mein, 1977; Wood and Sutherland, 1970). Multiple catchment

experiments in New Zealand (New Zealand Ministry of Works, 1970)

provide examples for measuring the impact of farming, forestry, and

range management practices on sediment yield in hilly terrain.

Reports from Kenya, Tanzania, and Uganda (Blackie, 1972; Pereira,

1973; Pereira et al., 1962, 1967; Rapp, 1977) present useful data for

comparison with some of the land use systems of the Caribbean,

including coffee and other cash crop plantations, grazing, subsistence

farming and forestry. Paired watershed studies spanning four years or

more demonstrated the impact of both land cover and specific

management practices on runoff and erosion as well as on the harvest

within the watershed. In all of the cases cited, researchers

collected frequent stream discharge and precipitation measurements.

Some cases also include continuous monitoring of the above, as well as

sampling of suspended sediments in streamflow. The results of grazing

and range improvement trials in experimental watersheds in Uganda

include a more than twofold increase in depth of penetration of

rainfall into the soil, and a concurrent reduction in peak streamflows

after restoration of overgrazed grasslands (Pereira et al., 1962). -

Data from experimental sites in Kenya (Blackie, 1972; Pereira, 1973)

document the effects of replacing tall evergreen forest with tea

plantations. The mean water yields over an 11-year period

effectively were equal for the forested control watershed and an

adjacent area planted in tea. The floods resulting from peak storm

events, however, varied substantially. The minimal impact of the tea

plantation reflects in part the stringent conservation measures

employed during its establishment. By contrast, clearing of

indigenous bamboo forest without immediate replacement by tree crops

increased streamflow 16% (Blackie, 1972). In Tanzania, a cleared

forest planted to a maize and vegetable single-crop system yielded a

50% increase in runoff, measured as streamflow (Edwards and Blackie,

1975; Edwards, 1977).

Findings from two other study areas in Tanzania, one a cultivated

montane catchment and the other a series of catchments in semi-arid

savanna, offer comparative data on river regimes as well as on

suspended sediment concentrations under changing land use practices

(Rapp, 1977). The catchments represent a complex mosaic of forest,

farm, pasture and bush, which is comparable to many upland catchments

in the Caribbean. During flood peaks, sediment concentrations ranged

from 2000 to 3500 mg L in the upland areas, and from 15,000 to
75,000 mg L in the semi-arid catchments. Flash floods and high

sediment loads in both the montane and savanna areas were attributed

to land use. A comparison of the relationship between drainage area

and sediment delivery ratio in the United States and Tanzania shows

much less reduction in sediment yield with increased drainage area in

the Tanzanian catchments.

Total streamflow and sediment yield were determined by the use of

flow duration and sediment rating curves along with stream gauge and

sediment concentration data. Sampling was carried out with a home-

made point-integrating hand operated sampler (Nilsson, 1969; Rapp,

1977) and an automatic multi-stage sampler designed for ephemeral

streams. Both the instruments and the methods of analysis used in

this study are feasible for use in the Caribbean.

Reservoir sedimentation. Studies of reservoir sedimentation and

other aspects of regional water utilization and management often

approach the situation as an engineering and economic problem, either

in terms of reservoir design or in subsequent maintenance of the

completed structure and the watershed. Many studies of this type have

been carried out in Latin America (Casco de Aviles, 1979; Crosson and

Frederick, 1977; Farvar and Milton, 1973; Rabinovitch, 1979; United

Nations, 1979) including several studies in the Caribbean (CDE, 1981;

IBRD, 1972; Lahmeyer, 1967; McHenry and Hawks, 1966; Noll, 1953).

Most of these works analyze the feasibility of proposed

impoundments or document sedimentation rates in existing structures.

The major hydroelectric and irrigation projects planned or under

construction in the Caribbean lack empirical data on sediment delivery

by watershed subdivisions. Measures of erosion rates within the

watershed also are seldom included. Soil conservation programs, when

they exist, usually are initiated in response to reservoir

sedimentation problems (Floyd, 1969; Gomez, 1980; Paulet, 1980;

Rocheleau, 1980; Santos, 1981; SEA, 1978; Vasquez, 1980).

Several reservoirs are in danger of filling up in half the time

projected for the useful life of the structure (CDE, 1981; McHenry and

Hawks, 1966; Noll, 1953; INDRHI, 1981), and some reservoirs already

require expensive dredging procedures to continue functioning (CDE,

1981). Serious erosion problems are related to land use in the upper

watersheds as well as in the immediate vicinity of impoundments

(Antonini et al., 1975; Carmona, 1980; Cepeda, 1980; CDE, 1981; de

Leon, 1980).

Erosion rates at the farm and plot level

While experimental watershed research has concentrated primarily

on undisturbed forested areas, the impact of deforestation and

resultant reservoir sedimentation, erosion studies in diverse

hillslope environments throughout the world have documented soil

losses at the scale of individual farms or small plots. These results

are available for varying crop types, rotation and conservation

practices as well as for a wide range of natural environmental


Research on erosion rates in the Caribbean and similar regions

consists primarily of experiments in standard runoff and erosion plots

under various land covers and management practices. Methods tested

and adapted in similar environments offer alternatives to the more

expensive and time consuming instrumentation often applied. An

adaptation of the standard erosion plot with sediment and runoff

collectors tested in hillslope experimental fields in Yugoslavia

provides a simple design for applications in other studies (Djorovic,

1977). Dunne (1977) describes simple erosion plot designs and

summarizes several inexpensive techniques for erosion measurement

without the use of standard plot structures. The Gerlach trough

(Gerlach, 1967) is easy to install and to use for soil loss and runoff

measurements (Morgan, 1979). It has been employed successfully in New

Zealand (Soons and Rainer, 1968), the Philippines (Kellman, 1969),

Israel (Yair, 1972), the United Kingdom (Morgan, 1977), and the

Carpathian Mountains (Gerlach, 1967).

Results from erosion studies in the Caribbean and Latin America.

Results from the Caribbean and neighboring Latin American nations

consistently show that erosion rates vary significantly (up to three

orders of magnitude) with changes in vegetation cover and management

practice (Table A-7) (Ahmad and Breckner, 1973; Barnett et al., 1972;

Bertoni, 1966; Marques et al., 1961; Noll, 1953; Rocha, 1977; Sheng,

1973; Sheng and Michaelson, 1973; Smith and Abruna, 1955; Suarez de

Castro, 1952; Suarez de Castro and Rodriguez, 1955, 1962; Vincente-

Chandler, 1976; Uribe, 1966). Clean-tilled fallow consistently

yielded soil losses in the range of 100 to 200 tons ha-1 yr- 1, while

land in annual crops lost approximately 20% of that amount, and

pasture land lost 10% of the cropland losses and about 2% of the

losses sustained in bare fallow plots. Undisturbed forest loses 500

to 1000 times less than the clean-till control plots (Sheng and

Michaelson, 1973). Results from Colombian coffee plantations indicate

fairly low rates of erosion, varying between the ranges common for

forest and field crops, depending on the age of the stand, the method

of establishment, and management practices (Suarez de Castro and

Rodriguez, 1955, 1962). The relative importance of farming practice

is also illustrated by the five- to ten-fold decreases in erosion

reported for various conservation practices tested in hillslope food

crop plots in Jamaica (Sheng and Michaelson, 1973).

Research on erosion potential in the Caribbean is based upon the

factors in the USLE. Studies conducted in Puerto Rico and the

Dominican Republic have defined the erosive potential of rainfall (R)

(Paulet, 1978), the erodibility of soils (K) (Barnett et al., 1972;

Bonnett and Lugo-Lopez, 1950; Lugo-Lopez, 1969) and the cropping

factor (C) (Santana, 1980) for parts of the region, but the

applicability of the USLE to these areas is questionable (Barnett et

al., 1972). More field data collection is needed on erosion and

runoff and their relationships to land use within a variety of land

use and ecosystem types throughout the Caribbean.

Results of erosion studies in similar environments. Basic data

on erosion rates are scarce in Latin America and the Caribbean, in

comparison with the humid tropics of Asia and Africa (Lal, 1977b). A

brief summary of selected erosion studies in these regions provides a

broader frame of reference for work already completed or in progress

in the Caribbean.

Many of the crops, the small farm technology and some elements of

the natural ecosystems of West Africa bear a strong resemblance to

parts of the Caribbean. Reports of experiments conducted by Lal et

al., (1979) and others at the International Institute for Tropical

Agriculture (IITA) in Ibadan, Nigeria, indicate a clear relationship

between vegetation cover and erosion rates, with results approximating

those from the Caribbean. Losses from clean-tilled fallow range from
-l -l -l -I
11 tons ha yr at 1% slope to 230 tons ha yr at 15% slope. On

the average, soil erosion varies much more with slope than does runoff

(Lal, 1977a).

Experiments in Senegal (Charreau, 1972) on gentler slopes in the

savanna demonstrate a similar range of soil loss as from the medium

slopes (10%) studied by Lal (1977b). Runoff in cropped plots exceeds

that in natural bush by a factor of 20 to 35 depending upon the crop,

while soil loss increased 30 to 50 times for the same crops as

compared to natural vegetation (Table A-8). Similar results are

reported for other sites in Ivory Coast and Upper Volta.

Lal (1977b) tested the effectiveness of various types of mulch as

well as several variations of minimum tillage. While mulch had less

effect on runoff, it stopped soil erosion, even on the 15% slopes

(Lal, 1977b). Experiments in Nigeria demonstrated the effectiveness

of alternative methods of field preparation and planting. While

croplands with ridges oriented downslope yielded 28% runoff and 20

tons ha1 of soil loss, alternate tied ridges across the slope reduced
-1 -I
runoff to 13% and soil losses to 6 tons ha yr (Kowal, 1970).

One West African study reported on the continuous measurement of

erosion in the same plots over several years. Lal (1977b) found that

slope effects may be reversed after a few years. After a rapid

initial loss of the topmost layers on steep inclines the erosion rate

decreases, while gentler slopes maintain a more constant erosion rate.

This indicates the importance of documenting the land use history of

hillslope sites so as to account for the influence of past soil loss

and profile modification.

More detailed surveys of the published West African soil erosion

literature have been complied by Lal (1977a), Okigbo (1977), Fournier

(1967), and Jones and Wild (1975). Projects in progress include

minimum tillage and multiple cropping experiments in plots at IITA

(Lal et al., 1979).

There is also substantial similarity between some of the lowland

dry forest and montane ecosystems of East Africa and the Caribbean.

The farming systems have some crops and practices in common, though

fewer than in the case of West Africa.

Erosion plot studies in Uganda yield similar results to the

experiments already cited in West Africa and Latin America (Table A-9)

(Sperow and Keefer, 1975). The major difference is in the magnitude

of total soil loss under bare fallow and annual crops, which is

attributable to lower annual rainfall. The relationship between

vegetation types and soil loss, however, is the same.

Similar experiments in Tanzania and Zimbabwe showed the same

range of soil loss for annual crops, bare fallow, and pasture. Losses
-I -i
under maize in Zimbabwe varied from 4 to 10 tons ha yr (Hudson,

1957). Mosquito netting placed above the soil reduced losses on
-i -i
clean-till plots to 1.2 tons ha yr demonstrating the importance

of interception by canopy and ground cover (Hudson, 1957; Lal, 1977b).

Early erosion studies in Tanzania (Staples, 1939; Rensburg, 1955)

compared sorghum with grass cover and sorghum/grass strip cropping

(Okigbo, 1977; T-rriple, 1972). Soil losses varied between 9 and 116
-i -i
tons ha yr under sorghum, depending upon site and cultivation
-1 -I
practice. Grasslands yielded 1.2 tons ha yr and sorghum strip-
-1 -i
cropped with grass yielded from 4 to 60 tons ha yr .

Findings in Kenya confirm that infiltration is greatly reduced by

grazing (Stephens, 1971; Thomas, 1974) and that while cultivation may

increase initially, the effect is temporary. Similar results have

been observed in Tanzania with a six-fold increase in peak runoff rate

after conversion from forest to annual crops (Wrigley, 1969).

Hutchinson et al. (1958) recorded a ten-fold runoff increase in clean-

tilled land converted from natural grassland. The end result is

higher runoff and erosion rates under both grazing and cultivation

(Ahn, 1977).

A less typical, broader study of erosion in Tanzania demonstrated

the soil conservation potential of many traditional farming methods,

including mulching and intercropping of field crops with bananas as

well as other crop association and rotation schemes. The contrasting

erosion plot sites in the Uluguru highlands and the semi-arid central

plains formed part of the same catchment study mentioned above (Rapp,

1977). Although some of the findings closely resemble the results for

other African and Latin American sites (Temple, 1972), high losses

were recorded for clean-weeded coffee, exceeding losses in nearby

plots'with maize (Anderson, 1962). The somewhat atypical results in

this case demonstrate that perennials do not necessarily conserve soil

better than annuals (Ahn, 1977). A wide range of sediment yields is

also reported for tea plantations in East Africa, with differences

attributed to management variables (Othieno, 1975).

Erosion in tree crop plantations is recognized as a major

contribution to regional sediment yields in Southeast Asia, where long

experience with rubber and tea plantations has demonstrated the wide

variation due to management of canopy and ground cover as well as

tillage practices (Coulter, 1972; Edwards, 1977). While tea

plantations yielded approximately one-seventh the soil loss from bare

fallow (Hasselo and Sikurajapathy, 1965) in two cases in India (Table

A-10), the erosion in tea was double that reported for forested plots
-i -l
in Malaysia and reached rates of 40 tons ha yr in Sri Lanka prior

to implementation of conservation practices (Lal, 1977a, 1977c).

Measurements from several land use and land treatment types in

upland Mindinao in the Philippines (Table A-ll) illustrated the

relationship of both land cover and land use rotations to runoff and

erosion rates (Kellman, 1969). The plantations had relatively little

impact in comparison with logging and farming uses. The results from

long established rice and corn plots suggest cumulative

destabilization of soil structure under permanent cultivation.

Qualitative and Informal Analyses of Land Use and Erosion
in the Caribbean and Similar Environments

Development and technology transfer projects in erosion prone

areas have yielded useful information on land use and erosion

problems. A few examples of interest include: studies of erosion and

overgrazing in the Bolivian highlands (LeBaron et al., 1979); a

summary of the programs of the Yallahs Valley Land Authority in

Jamaica (Floyd, 1969); the progress reports, project summaries and

consultant reports from Plan Sierra in the Dominican Republic

(Antonini and York, 1979; Chaney and Lewis, 1980; Montero, 1979;

Swedforest, 1980), and development agency communiques on resource

management projects in Cajamarca, Peru (Nicholaides and Hildebrand,

1980b), the southwestern slopes of the Dominican Republic, the

interior of Jamaica, and Haiti (Murray, 1977; Zuvekas, 1978).

Published and mimeographed reports of the government agencies charged

with soil conservation in the Dominican Republic (Gomez, 1980; Lopez,

1980; Paulet, 1980; Russo, 1980; Vasquez, 1980), Puerto Rico, Jamaica,

Trinidad, Tobago, and other areas of the Caribbean (Henriquez, 1962)

also provide useful information.

The qualitative information gathered from such sources can help

link land management variables to the onset, severity, and persistence

of various erosion features and sedimentation problems. The

settlement and development history of a region is often indicative of

the changing rates of deforestation and the intensity of cultivation

over time. This provides a basis for relating current erosion rates

to the succession of land use systems in an area. Future trends in

land use and erosion rates can be estimated more realistically with

the aid of this type of background information.

The Role of Farming Systems Research in Soil and
Water Conservation

It is important to view the effects of land use and erosion

problems within the source areas rather than simply calculating the

net export of sediment to downstream areas (Carmona, 1980). The key

to changing the situation is to be found in the internal workings of

the upland land use systems (Quezada, 1977) and in their relationship

to the larger system (Antonini and York, 1979). Any changes in

management of the uplands must take into account the causes of current

practices and the practical feasibility of proposed technological or

land use changes (Morgan, 1979; Russo, 1980).

For example, the proposed solutions to erosion problems may

involve specific technical innovations such as mulching or terracing

(Sheng and Michaelson, 1973), or a change of land use may be

suggested. Terracing and mulching with increased crop cover

alternately have received priority in various conservation projects

with contrasting and somewhat unpredictable results. Reports from

Zimbabwe demonstrate the effectiveness of increased crop cover and

mulches (Ahn, 1977; Hudson, 1957), while climatic and farming system

constraints in Kenya make terracing a more attractive alternative

(Ahn, 1977; Thomas, 1974). In Tanzania, terracing in inappropriate

situations increased the erosion hazard from landslides (Temple and

Rapp, 1972). Both approaches have been tried in the Caribbean (Sheng

and Michaelson, 1973; Wilson, 1976) though the structural

modifications of slope predominate in projects in Jamaica and the

Dominican Republic (Bonilla, 1980; Vasquez, 1980). The advisability

of this approach is questionable.

The evaluation of proposed technological changes must be measured

against the existing practices as well as against the more drastic

option of land use change on a broad scale. For such analyses a

knowledge of erosion, sedimentation and their variation according to

crop and vegetation type will not suffice. The proper choice of

conservation measures depends upon a full understanding of farming and

related land use systems in a region.

Farming Systems, Agroecosystems and Agroforestry Research


The use of the systems approach in agricultural development

efforts is a relatively recent phenomenon (Dent and Anderson, 1971;

Duckman et al., 1974). Unlike the study of erosion and sedimentation,

the research in this field has been conducted mainly in the tropics.

Ruthenberg (1980) carried out much of the pioneer work in farming

systems, primarily in smallholder farming districts in Kenya, Tanzania

and West Africa. Most of the research in farming systems in Africa

(Ruthenberg, 1980) and Central America has focused on small, limited

resource farmers.

Studies have been conducted by interdisciplinary and often

international teams of agronomists, agricultural economists,

anthropologists and ecologists. The individual farm enterprise and

the farm household have been the preferred units of analysis. Systems

concepts have been consistently employed, although the methodology and

topical emphasis vary with the regional and disciplinary orientation

of the researchers and institutions involved. The methodological and

substantive contributions of this avenue of research provide a solid

point of departure from which to expand the treatment of ecological

aspects of the problem and to extend the scale of analysis beyond the

farm level.

Selected Examples from Africa

Most of the African research in cultivation and grazing systems

has focused on systems that represent transitions from traditional to

more commercialized forms of production. Collecting, or hunting and

gathering, are largely ignored, as is forestry. Cultivation and

grazing systems receive the greatest emphasis. Ruthenberg (1980)

divides cultivation into shifting cultivation, fallow systems

regulated by farming (pasture-crop rotations), permanent upland

cultivation, arable irrigation farming, and perennial crops. Within

grazing systems pastoral nomadism and ranching are considered.

Ruthenberg (1976) and Flinn and Lagemann (1980) analyzed resource

utilization by farmers, their impact on the quality and condition of

the resource base, and the future implications for sustained

production in the area. Carrying capacity was determined by inherent

qualities and current condition of the environment, level and type of

technology, and standard of living (Lagemann, 1977). Within this

context, Lagemann tested and criticized Boserup's theory of

agricultural innovation (Boserup, 1965) with respect to environmental

response to intensification of agriculture. The relationship of

current population densities to environmental carrying capacity under

different production systems was demonstrated by simulation models.

The simulations extrapolate current bush fallow practices (in West

African examples) to predict net environmental degradation,

diminishing net yields and decreasing production per unit labor input,

all due primarily to declining soil fertility.

Patterns of land use, spatial organization of cropping, farm

level resource management and farm level economic analyses are

emphasized within this tradition. Methods for evaluating technical

innovations for low resource farmers have also been presented in

recent studies (Flinn and Lagemann, 1980; Flinn et al., 1980). Many

of the recent studies have been conducted in Nigeria in conjunction

with the International Institute for Tropical Agriculture (IITA).

Farming systems research and extension programs also have been

developed in East Africa (Collinson, 1981).

Studies of shifting cultivation systems in West and Central

Africa indicate a potential for maintaining shifting cultivation

indefinitely at a lowered but sustained rate of production, relative

to undisturbed forest systems. Soil fertility is reduced to

approximately 75% of the value for undisturbed forest soils. The

successful attainment of adequate sustained production hinges on the

rotation cycle, which must vary between 20 to 50 years of forest

fallow per year of cultivation. The implications for carrying

capacity are clear. While the system itself may work, shifting

cultivation cannot support more than about 20 to 50 persons km ,

taking into account the required fallow. Further experimental work at

IITA by Greenland, Lal, and others has explored alternatives to this

system, emphasizing soil management and conservation under bush fallow

and continuous cropping (Lal et al., 1979; Lal, 1977a) and continuous

mixed-cropping systems (Greenland, 1975; Ruthenberg, 1976).

Bioeconomic modelling has been proposed to evaluate alternative soil

conservation practices and cropping systems (Dumsday and Flinn, 1977).

Agroforestry research in Africa has combined experiments with

commercial forestry and subsistence agriculture (King, 1968). The

taungya system features mixed cropping of commercially harvested and

replanted forest tracts, with the tenant farmers caring for the

seedlings and saplings as well as their food crops over a period of

about four years (Dubois, 1979; King, 1978). The field of

agroforestry has further developed to include diagnostic and

experimental work with existing subsistence and commercial production

systems that feature some combination of trees, livestock production

and/or field crops (Brookman, 1976; Douglas and Hart, 1976; Olawoye,

1975; Parry, 1957; Raintree, 1982; Lundgren, 1982). Both cocoa and

oil palm production on small farms have been studied within this

context (Flinn, 1980; Grinnell, 1977; Lagemann, 1977; Letouzay, 1955)

as well as many traditional systems of shifting agriculture that

include management of tree crops (Dubois, 1979; King, 1968). In

general, mixed tree crop/annuals production systems are more diverse

and more stable, both in economic and ecological terms (Lagemann,


Tree crops can provide fuel wood, high protein forage, lumber,

fiber, food, mulch, and cash crops (Douglas and Hart, 1976). While

establishment of such a stand requires more capital, labor and

management than a plot of annual crops, the products are often of

higher value and can be used on the farm to fill a wide range of

subsistence needs (Lagemann, 1977). Soil fertility and structural

stability are enhanced by partial tree crop cover, providing some of

the ecological benefits of forest fallow without sacrificing economic

production (Nair, 1983). In hillslope environments the combination of

trees and row crops is particularly advantageous since tree crops

offer the soil greater protection from erosion (Douglas and Hart,


While the study of agroforestry is still relatively new, the

field is developing rapidly, in part as a response to the need to

increase small farm production in marginal lands while maintaining or

rehabilitating watersheds and forest resources (King and Chandler,

1978). Current research at the International Council for Research in

Agroforestry in Kenya (ICRAF) focuses on the elaboration of a

methodology for diagnosis of new ones. Rapid survey diagnostic

techniques and combined research/extension programs are being

developed and tested in field sites that include farm level and

community level work as well as experimental plot studies (Raintree,

1982). The multidisciplinary staff includes foresters, agricultural

economists, agronomists, anthropologists and ecologists, among others.

The international client areas for the methodology being developed by

ICRAF include the hillslope farms of the Caribbean, as well as the

Andean highlands, the Amazonian lowlands, and the African savannas,

and many other fragile and/or marginal environments under cultivation

(King and Chandler, 1978).

Central American Research

Farming systems research in Central America has been conducted at

the Center for Teaching and Research in Tropical Agriculture (CATIE)

in Costa Rica, and the Institute of Agricultural Science and

Technology (ICTA) in Guatemala. Both have sought to serve small

farmers and to modify, rather than replace, traditional systems of

agriculture (including hillslope farming).

Farming systems research conducted at CATIE has emphasized

economic and agronomic description of existing cropping systems

through community and regional level questionnaires and surveys, using

standard sampling procedures (Navarro, 1979). Experiments have

focused on mixed-crop combinations for optimization of production

given the available natural and economic inputs. The general

perspective as well as some specific techniques of ecosystems and

energy analysis have been applied (Hart and Pinchinat, 1980; Moreno,


Several precedents exist for ecosystem and energy analysis

applied to production systems, including examples from India (Revelle,

1976; Odend'hal, 1972), the United States (Burnett, 1977; Ewel, 1973),

Indonesia (Geertz, 1963, 1972), Guatemala (Carter and Snedaker, 1969),

the Dominican Republic (Werge, 1975) and elsewhere (Lugo and Snedaker,

1971; Odum, 1967). Energy budget analyses more or less independent of

the systems approach also have been conducted by researchers in the

United States (Lockeretz, 1975; Pimentel, 1973), England (Leach,

1976), and Asia (Makhijani and Poole, 1975; Smil, 1979). The energy

studies at CATIE, however, use the systems approach and focus on plant

and animal production subsystems, as opposed to the socioeconomic or

environmental subsystems (Hart, 1980; Holle, 1979). The majority of

the cropping systems research does not address questions of erosion

and watershed protection although some studies have been carried out

on erosion rates under hillslope cropping systems (Bermudez, 1979) and

under agroforestry systems (Apolo, 1979).

The application of energy analysis to individual farms (Hart,

1980) represents a distinct and particularly useful subset of the

general body of research at CATIE. The energy analysis methodology

developed by Odum (1971) and Odum and Odum (1976) has been adapted by

Hart (1980) for use in the rural farming districts of Central America.

This method can be used to identify and describe existing successful

adaptations. It can also be used to highlight the functional

relationships found in the average case, in order to choose the

critical points in the system where specific (and viable) changes will

have a major impact on total production (Fig. 5).

Hart (1980) has elaborated further a field survey methodology

using systems concepts and energy flow diagrams in rapid surveys of

rural farming communities, to characterize existing farming systems in

qualitative and quantitative terms. Still lacking are inclusion of

environmental inputs, outputs and storage, and the extrapolation from

the farm to larger scales of analysis.

'. -
44~4 4 4 4 A

r 'wir^/pooo *aooUri

knia~ar~~ri* 1LH4

-**'T'o /ri^/fTp^~



a" aa s

o *. C.
\i a .i *
* *. 0

k f
I 3 S 1 a.

* a c s

I^ .I s "
0 0 <

0 - 3 0j a
< .4 4


4 CO
3 a


0 0-

CO 0

a o
* U


a~upo.~ ~

twTT/T~~7~r *ri~

daB0 3BT ,poc;

poo.Q ;o *ev-..

5 .
0 -"

5- a
S -


0 6
-, z

0 0 6 -0
-- 4 -"'
i.~~~~~ &a a. < fi o c

o *~ -1 0
0 0

o .a o -0o* * j 3~ ; ^S3
S ^ -S0 S ^













A similar approach has been applied to the environmental

subsystem in a National Science Foundation project in the same area

(Berish, 1982; Brown, 1982; Ewel, 1981). The project focuses on

nutrient cycling in small farm and successional plots, and the

research is directed toward improved management of nutrient cycles in


Agroforestry research at CATIE has combined with forestry and

watershed protection research (Budowski, 1977; Combe, 1979). The

major experiments to date emphasize the use of laurel (Cordia

alliodora) and poro (Erythrina poeppigiana) trees to provide shade,

reduce erosion, and produce firewood or lumber within coffee

plantations in montane environments (Beer, 1979; Bermudez, 1979;

Rosero and Gewald, 1979; Russo, 1983). Combinations of grazing with

plantations of alder (Alnus acuminata), Eucalyptus sp. and cypress

(Cypressus lusitanica) also have been tested in multiple field trials

(Combe, 1979; Gonzales et al., 1979). One test of the taungya system

has been conducted, using Gmelina arborea interplanted with maize and

beans (Rosero, 1979). Most of the trials have proven successful when

judged in terms of mixed ecological and economic criteria by

researchers, but widespread adoption by small farmers has not yet

occurred beyond the original centers of innovation.

The planting of nitrogen-fixing leguminous trees in or around

pastures and within coffee and cocoa plantations is common practice in

many parts of Central America, but the density and distribution of the

trees often is very limited. Moreover, the need for the introduction

and/or development of agroforestry systems often is greatest in the

marginal lands where small farmers are dedicated exclusively to

cultivation of annual crops in slash and burn or bush fallow rotations

(Lagemann, 1978).

The farming systems research conducted at the Institute of

Agricultural Science and Technology (ICTA) in Guatemala, concentrated

more on the socioeconomic subsystem and on the integration of research

and extension (Reiche and Lee, 1978). Extensive contact with

hillslope farmers in Guatemala's rugged uplands also produced some

work related to erosion control (Hildebrand, 1981).

The most significant result of the ICTA research was the

development of a practical multidisciplinary approach that merges

research and extension efforts in farming systems. Farming systems

research and extension (FSR/E) integrates farmers, researchers and

extensionists into an effective group to identify and solve problems

and promote dissemination of the -solutions (Hildebrand, 1981; Gladwin,

1981). This is reflected in the rapid survey technique (sondeo)

developed for use by interdisciplinary research and extension teams

with small, low-technology farmers (Swisher et al., 1982).

The survey itself consists of a series of informal interviews

conducted by an interdisciplinary team that includes a social

scientist (usually an anthropologist), an agronomist or related

technical specialist in agriculture, and an economist (Hildebrand,

1981). The approach borrows heavily from traditional anthropological

field methods with reliance on key informants and on corroboration of

information from various sources. The use of open-ended interviews

with chosen informants contrasts with the random sampling of the

population so often used with questionnaires or structured interviews.

The latter, more formal approach is a more common and preferred form

of data gathering in many disciplines. However, it is also expensive

and time-consuming, and it presupposes an accurate population census

or property survey on which to base the sample. Formal surveys and

questionnaires also inherently limit the categories of information to

be treated. Little room is left for the definition in the field of

problems not already recognized prior to the survey. Opportunities to

explore the unique relationship between various aspects of a problem

in an area are also constrained by the format. The major

considerations in using the sondeo technique are quality versus

quantity of information and cost of the survey versus useful

information obtained.

The sondeo provides a practical and effective means of

reconnaissance and data gathering. It also lays the groundwork for

future extension programs in the area. During the intensive one-to-

two-week survey, the knowledge and needs of the farmer are

incorporated into the design (form and content) of subsequent on-farm

research projects. This methodology relies heavily on the judgement

of the research team, the local populace, and individual farmers to

define farming systems and their problems and to choose representative

or exceptional cases for further study, according to the goals of each

project (Hildebrand, 1981).

The experiments themselves are on-farm trials which may or may

not be replicated at experimental stations. The sondeo and the field

trials are supplemented by farm record-keeping. A family member keeps

an account of the farm's inputs and outputs as well as of activities

and movement of materials within the farm itself. These records also

assist in the evaluation of on-farm trial results and provide basic

data for further trials and/or discussions with farmers. The success

of a new technology is judged at least in part by the farmer's

perception of its performance and by its subsequent adoption by him

and other farmers in the area. This indicates t6 some extent the

"fitness" of a technology for the farming system as a whole, at the

farm level (Swisher et al., 1982).

The sondeo as well as the subsequent on-farm trials and farm

record-keeping of the FSR/E approach are readily adaptable to soil and

water conservation research in hillslope environments in the

Caribbean. The major elements lacking are greater attention to

natural resource management at the farm level and a method for

predicting and evaluating the success of a given technology at the

watershed or regional level.


The General Approach

The methods reported in the literature review include a variety

of techniques that can be incorporated into a consistent and

appropriate methodology. The end product, however, must be more than

a method or a collection of techniques. The methodology proposed and

tested in this study is an adaptation of the scientific method in

general, and systems analysis in particular, to the interdisciplinary

study and treatment of the problems of land use, erosion, and

sedimentation in the underdeveloped nations of the tropics. The

research approach combined elements of farming system and ecosystem

analysis, drawing most heavily on the work of Odum (1971, 1982),

Antonini et al. (1975), Bormann and Likens (1979), Hildebrand (1981),

and Hart (1980), all described in greater detail in Chapter II.

The research tested specific management-related hypotheses under

field conditions. The cases studied required immediate action based

on tentative solutions from experimental results, prior to extensive

repetition and replication. The general research model included

direct outputs from verification to policy and production sectors and

a feedback from verification to further experimentation (Fig. 3).

The feedbacks in the research program imply an ongoing process of

learning. The time constraints in applied research were handled by

using these feedback loops to continually test and modify the

tentative solutions already proposed. This iterative approach has

already been tested in farming systems research and extension programs

(Hildebrand, 1981).

The usual concept of applied research is one of a finite activity

to be carried out and completed by specialists, after which they will

offer a set of definitive conclusions to be implemented. In this

case, the study area was viewed as a system in a flux, constantly

adjusting to changes in internal and boundary conditions. The object

of study in this case also had a subjective component. The role of

people in determining the direction of ecosystem evolution was taken

into account. Residents of the region contributed to the

investigation as both informants and participants in data gathering,

experimental and verification procedures. The researcher participated

in an on-going experiment in which people living in the area sought

short-term relief and long-term solutions to problems at least

partially perceived and defined by them.

The study was designed to accommodate the distinct priorities and

information needs of local clients, scientists, and the regional

policy sector. The experimental design and data analyses tested

multi-faceted hypotheses concerning technology and land use

alternatives for the region. Each experiment included: a primary

hypothesis as to the biophysical or economic performance of a

particular alternative; a secondary hypothesis concerning how the

proposed change would fit into the existing system; and a third

hypothesis that the system, as such, could and should be sustained,

with modifications.

The primary hypotheses were tested and judged jointly by researchers

and farmers, according to objective, quantitative criteria. The

secondary postulate was tested and judged by the farmer according to

subjective criteria, based on overall practical performance. Researchers

provided a posteriori explanation and interpretation of the farmers'

experience. The value judgement as to the fitness of the existing system

was considered the joint prerogative of the local residents and the

policy sector (clients) while the researcher determined the system's

sustainability by ecological analysis of current trends.

Materials and Methods

The study was conducted at four scales of analysis within a nested

hierarchy of spatial units, including: the Plan Sierra impact area (2500

2 2 2
km ); watersheds of 500 to 100 km ; small watersheds of 1 to 20 km ; and

individual landholdings ranging from 0.5 to 500 ha.

Chronologically the research activities were grouped into three

phases of increasingly finer levels of resolution and greater detail of

analysis. The regional reconnaissance and refinement of problem

definition was followed by detailed characterization of the study sites

at the watershed and plot level. The third phase consisted of monitoring

runoff, erosion, sedimentation, and production under varying land use and

soil conservation practices (Fig. 4).

The conceptual model of the Caribbean region (Fig. 1) formed the

basis for the overall research design, while at each successive level of

resolution a conceptual model was proposed, evaluated through field data

collection, and refined or modified based on empirical evidence and

testing of specific hypotheses inherent in the model.

Regional Reconnaissance

The inventory of existing land use systems and the condition of

soil and water resources within the Plan Sierra impact area provided

the data for refinement of the problem definition and for subsequent

application of the research design within successively smaller units

of analysis.

Most of the reconnaissance activities were completed between

January and March 1980. A regional description and summary of land

use systems was complied from library and field research. A review

and synthesis of maps, aerial photographs, statistical summaries and

literature relating to the study area preceded the field surveys. The

area was stratified into multitopic subregions based on cartographic

analysis of physical, biotic and socioeconomic characteristics mapped

at scales of 1:250,000 and 1:50,000. The major criteria for zonation

were life zones (Holdridge, 1967; OAS, 1967), topographic

characteristics, and land use characteristics, the latter reflecting

population density as well as condition and productivity of the land.

Maps were compiled by Plan Sierra cartographic and project staff from

topographic and thematic maps at 1:250,000 and 1:50,000 (Jennings,

1979a; OAS, 1967; Swedforest, 1980), and from aerial photographs at


The field survey was similar to the general procedure outlined by

Hildebrand (1981). Several rapid reconnaissance surveys of erosion

features, land cover and land use systems were conducted within the

regional subdivisions outlined above, as part of Plan Sierra program

development in soil and water conservation. Survey teams varied in

composition, but usually included the author and one to four

specialists and paraprofessionals in engineering, forestry, agronomy,

and soil conservation.

The surveys included formal and informal interviews with

residents, as well as field mapping of land cover, land use and

evidence of erosion and sedimentation in the various areas visited.

The selection of sites for more detailed observation reflected a

strong reliance on the knowledge of agronomists, foresters and

conservationists already familiar with the area, as well as the

opinion of residents as to what areas constituted typical or extreme

examples of particular physical and socioeconomic characteristics.

Field survey records included numerical data, maps, and interview with

residents, as well as the impressions and observations of the survey


A synthesis of the cumulative results of prior field

reconnaissance by interdisciplinary teams of consultants (Chaney and

Lewis, 1980; Georges, 1981; Hart, 1981; Montero et al., 1981; Navarro,

1981; Nicholaides and Hildebrand, 1980b; Safa and Gladwin, 1981;

Santos, 1981; Swedforest, 1980) supplements the information gathered

from the author's survey. Written reports and personal communication

from the consultants and visiting scientists contributed to updates of

the regional profile.

Refinement of the Research Design

Based on the reviews of regional information and the completed

field reconnaissance, the conceptual model of the region was modified

to include relationships previously omitted or incorrectly defined.

The research design then was developed to test the major hypotheses

implied in the model. The discharge and sediment yield of two large

watersheds were measured over a 15-month study period. During the

same interval subwatersheds were described and monitored in greater

detail to relate differences in discharge and sediment yield to

varying physical and land use characteristics. Erosion plots

constructed within the subwatersheds provided comparative data on

runoff, erosion and production under different land uses, each with

varying conservation practices.

The 18-month period of study for phases two and three extended

from 1 Apr. 1980 to 30 June 1981. This period included a full

hydrologic year, from 1 Apr. 1980 to 1 Apr. 1981, and also allowed

repetition of sampling and monitoring during the time of peak

rainfall, from April through June.

The spatial and logistical organization of research activities

are illustrated in diagram and tabular form (Fig. 6, Table 1). The

chronological order of analytical and data collection procedures

parallels the general case described in Fig. 4. The choice of study

sites reflects the insights gained from the review and reconnaissance

survey, as well as the information needs of Plan Sierra.

Study Sites

The study sites selected within the Plan Sierra impact region

included two large watersheds (500 to 100 km 2), five small watersheds

(1 to 20 km2) nested within the two larger units, and 16 plots on nine

landholdings situated within three of the small watersheds.





= S:

I.. -






~ S

'I *f.



Cn U)
0 0

S0 0 m 0
U0 0 ( )-
W 00
0 0) 0) 0
U - H 4-1 -Ii

0) )0 -

0 0
0 0 0
01 ri 0,

0H d 0 0
I>C 0 f

o 0

0 411 ( 03


0 + O (:DO 0 0C 0O OO
r-i f0 c I

(3) 4-)
ro 4
0 -') 00)

0 M
0) 0)
0; 44- )4-)
4-I o- 0) i 0
0 0 0 0 0) :00
..-i >-4

0) > d-^ 0)
o u ~~~~ro a nt
04 4 U 44)

0 0U 3
T3 O : 0)(UO +
C4 0) >- o)

>-I 4.44-40)4J 0d 4-) 4)J 4
0 L-i i- tu 0) U) M U) t 0
V0 0 0 0 0-I 0 4-3.H.-4(T
zC 4-) f04 41

0 0

0 I- *0

iCU 0
U r 0 i-CN 44 4-4 ) 4 d 4 W -
0 0)40 4- CO 0 (n z U) 0
r"- a) 0-4 0
r. 4C U 4M 4C4P L
0) U)

- 4-4
4(t 0
$4 4) m I: U, o I,
0 Q0 O0

4- 0 0 *- -
n> -r-I 'O ^ ^ a. -

0) < 0 4-r
.- 00
4- 0 0 000000
(U C ;0 0~ r--l

4 0 0 (V
aJ) zr-.H 0 0 -
0 00 000
r-9 (1) (1 ro ro ra ra ro ra ra ra
43U) rq -H r. z z z z z z r- 0
(3 W 0 0 0 0 0 0 0 0 z
P4 P4 M~ = = = = = = = +-

The criteria for selection of the large watersheds for further

study were as follows:

1. Availability of historical data on climate and river

2. Current coverage of the area by rain gauges and water level

3. Need for erosion and sedimentation data in the area; and

4. Presence of Plan Sierra personnel and facilities in the

The Plan Sierra region is drained by three major rivers, Mao, Bao, and

Amina. All three watersheds are partially contained within the impact

area and empty into the Yaque del Norte River. Each river is a

current or future source of water for hydroelectric and irrigation

projects planned within the borders of the study area. The study

concentrated on the Mao and Amina Rivers.

Within each of the two large watersheds, subwatersheds of

critical concern were selected based on the land use systems, erosion

features, climate, soil, and topography. Ease of access and presence

of soil conservation personnel also played a part in the decision.

Five small watersheds were chosen to test the following

hypotheses: 1) that land use is the major determinant of erosion and

sedimentation in the region; and 2) that the combination of pasture

and field crops common in the densely populated central region

contributes more heavily to sedimentation than land use practices

based on coffee which is more typical in the highlands. The size of

the watersheds ranged from 1 to 30 km with two replicates in the

first size category (1-2 km ) and three in the second group (10-30

km2). The examples represented the smallest scale of analysis in
km ). The examples represented the smallest scale of analysis in

which the land use matrices characteristic of the given subregion

could be measured. In each case a cluster of settlements was

included, along with a combination of subsistence and commercial land

uses. This unit of analysis also was compatible with community level

socioeconomic analyses conducted in the region (Georges, 1982).

The smallest scale of analysis focused on landholdings associated

with individual households. Sixteen experimental plots were installed

on property belonging to nine separate households, all situated within

the watershed study areas mentioned above. As depicted in the

research design (Fig. 6, Table 1), two distinct physical subregions

and four major land use types were represented in the three clusters

of plots. The plots were chosen to reflect average characteristics of

microclimate, slope, soil and land use within each watershed. The

first three variables were held constant, while land use was varied.

Nested within the comparison of land uses was a comparison of soil

conservation practices in plots planted to annual crops and a

comparison of stages of development in coffee and forest plots.

Characterization of Watersheds and Plots Selected for Further Study

Large watersheds

The portions of the watersheds included within the study were

delineated and measured on 1:50,000 scale topographic maps (U.S. Army

Map Service, 1962). Cartographic analysis of drainage networks,

geologic subregions, life zones, topography and land use indicated

areas whose composite characteristics favor high rates of runoff,

erosion and sedimentation.

The choice of sampling sites for more detailed study of discharge

and sedimentation was based on several criteria, including ease of

access, proximity to homes of observers, established hydrologic

monitoring sites, proposed dam sites, and regularity of the

longitudinal and cross-sectional profiles of the channel in the

vicinity of the potential sampling points. Profiles of the river

cross-sections were surveyed. The relative heights of the bridges and

other large structures along the stream bank also were measured to

provide reference points for reporting maximum flood stages. The

profiles are included in Appendix B.

Small watersheds

The small subwatersheds selected for further study were analyzed

by standard cartographic and photogrammetric methods to determine

total area, average slope, and area and distribution of land cover

types. Land use and socioeconomic characteristics of the watersheds

and their settlements were determined by field observation,

interviews, and a review of statistics available at the community


The average slope was determined by the Wentworth (1930) method,

using topographic maps at the 1:50,000 scale. Each watershed was

mapped separately and overlain with a 1-km grid subdivided into four

cells each, with two diagonal cross sections per grid. The slope from

the center toward each corner was determined by the change in

elevation over the four 0.5-km transects. The average slope (%) was

calculated for each small grid cell (0.25 kg2), then for each larger

cell (1 km 2), and finally for the watershed as a whole.

Total area was determined by planimetry. Area and distribution

of land cover types were determined in a three step process, beginning

with the interpretation of black and white aerial photographs at the

1:20,000 scale, taken in January and March 1980. The watershed

boundaries, the stream, its tributaries (if any) and ma3or features

such as roads and paths were delineated on outline maps enlarged to

the same scale, as well as on the aerial photographs. Land cover

subdivisions were outlined on the photographs, based on differences in

color, texture, and pattern. In many cases the fine texture of land

cover subdivisions required aggregation into units of mixed land cover

recognizable by plot or property boundaries. The land cover units

previously outlined were classified according to land use, based on

prior field observations in the study area during the reconnaissance

survey. After transfer to a map of known scale and projection, the

areas of major land use types were measured by planimetry.

The incidence, type and distribution of erosion features in each

small watershed were noted during the initial field visits and during

the 15-month period of study. The number, severity and distribution

of landslips and landslides, gullies, and signs of rill and sheet

erosion were compared between watersheds included in the study as well

as with several other watersheds visited regularly in the course of

related studies. Residents often were questioned about new features,

or noticeable changes in pre-existing gullies, rills and landslides.

Whenever possible, the immediate causes, such as a particular storm

event, agricultural and construction activity, were identified. This

type of qualitative analysis provided a background for further

refinement of the research design (Mosley, 1980) and for more informed

interpretation of subsequent results.

Detailed analyses of channel erosion, deposition, and range of

flood stages were conducted during field inspection of the stream

courses. Variations in depth and areal extent of sediment deposits

gave some indication of sedimentation rates within the basins. The

general form of channel terraces and floodplains, as well as the

height of residual debris lodged in trees and rocks on the stream

banks, provided indicators of peak annual flood stages. The specific

topographic indicators included erosion features on the stream banks

and breaks in slope along the cross-sectional profile. The cross-

sections were surveyed by the stadia transit method (see Appendix B).

Estimates of peak floods for longer periods were obtained from

interviews with elderly residents of the area. Independent

questioning of several persons established the margin of error in the


The sampling and monitoring locations within each watershed were

chosen according to several criteria:

1. Situation relative to settlements and the land use to be

2. Ease of access;

3. Security of equipment from vandalism

4. Availability of natural footings and fastenings against
flash floods and transported debris; and

5. Regularity of channel longitudinal and cross-sectional
profiles in the vicinity.

Interviews with residents of the communities within the

watersheds were conducted at local meeting places, in the field and in

the homes of agricultural laborers and landholders. The Plan Sierra

soil conservationists living in the area also were interviewed, and

they in turn questioned residents about the settlement history, land

use and farming practice, past and present production levels, and

sources of income. Discussions with two anthropologists conducting

land use and migration research in the region also provided valuable

information and insights into the character of the communities in the

study areas (Pessar, 1981; Georges, 1982). Plan Sierra social

workers, agronomists and foresters familiar with the area of interest

also contributed to the socioeconomic profile of the small watersheds.

Description of individual landholdings

Selection of sites for measurement of runoff and erosion in

experimental plots was based on uniformity and degree of slope, as

well as type of land use, management practice and easy access for

construction and sampling purposes. Wherever possible, replicates of

land use and treatment were established within the same watershed and

also in another watershed to determine the margin of error in

measurement and to compare the relative difference in runoff and

sediment yield under varying conditions of site and land use.

The experimental design further subdivided the categories of

forest, pasture, coffee and annual crops to compare undisturbed and

secondary forest, new and established coffee stands, different types

of annual crops, and use of minimum tillage and hillside ditches for

erosion control in fields planted to annual crops. These subsets of

land use type were tested in paired plots in the same or adjacent land

holdings to guarantee duplication of all other conditions except the

variables of interest. Individual plots were chosen based on field

observation of the above mentioned criteria. The choices were

confirmed after consultation with agronomists, conservation personnel

and residents of the area to determine if the plot in question

constituted a representative example of management relative to the

surrounding watershed.

Formal and informal interviews with owners, residents, neighbors

and local Plan Sierra personnel served to outline the settlement, land

use and production history and the variability of natural conditions

for the individual plots. Detailed information on past and present

crop associations, rotations, yields, labor and material inputs, and

ratios of commercial to subsistence production came from intensive

interviews with the persons directly responsible for management of the

site for a period of 10 years or more. These interviews often spanned

two or three visits by one or more members of the research team. The-

format was open-ended to allow the participants to elaborate on their


Farmers were encouraged to discuss their problems with respect to

subsistence and commercial production and to volunteer insights and

judgements as to potential solutions. The women and children working

at each site also were interviewed, usually on separate occasions, to

obtain accounts of their roles in production and natural resource

management as members of the farm household. Their perceptions of

problems and suggestions for changes also were solicited.

The baseline information to be obtained was outlined in diagram

form (Fig. 5). This helped the interviewers to keep track of the

subject matter covered. It also provided a convenient format for

recording and summarizing responses during or following the

discussion. Information noted on the diagrams and tables served to

evaluate the farm level models prior to initiating the erosion plot

experiments and watershed monitoring activities.

All sites were chosen to reflect variation in land use and

treatment, while slope and soil conditions were held constant and as

close as possible to the average for the watershed. Slope

measurements along the downslope transect were made prior to final

siting of all experimental plots.

Soil profile descriptions, characterization of soil samples by

laboratory analysis, and taxonomic classification constituted part of

the site description at each plot. Rectangular trenches at least 1 m

deep, 1 m long and 0.5 m wide were cut for observation and sampling.

Measurement and description of profile stratification, with detailed

description of color, texture, structure, and uniformity, by horizon,

were carried out according to the procedures outlined in the Soil

Survey Manual (USDA, 1951).

Munsell color charts were used for wet and dry color

determinations in the field (Munsell Color Co., 1951). Laboratory

analyses for N, P, K, and organic matter content followed standard

methods for determination of Kjeldahl N and Bray P by colorimetry, and

for determination of K by atomic absorption (USDA, 1975).

The North District Research Laboratory (CENDA) of the State

Secretariat of Agriculture conducted all laboratory tests for the

project, including physical and chemical characterization of soil

samples. Soil classifications were confirmed and refined by soil

survey specialists from the Secretariat's South District Laboratory.

The relative infiltration rates of soils at the various sites

were determined by measurements with ring infiltrometers (Gregory and

Walling, 1973; Wisler and Brater, 1959). The inner ring was cut to a

25 cm diameter and the outer ring measured 40 cm across. After

placement in the ground with a minimum of soil displacement, the outer

ring was filled to form a barrier of saturated soil around the inner

ring which was filled to a 10-cm depth. Throughout the next 4 hours a

nearly constant head of 10 cm was maintained while measurements of

water added were recorded at increasingly longer intervals. The form

used for the field measurements is included in Appendix H.

The frequency, distribution and severity of erosion features on

and around the plot sites were observed and noted prior to

construction of experimental plots. Wherever possible, the

developmental sequence of such features was determined from accounts

by the residents or neighbors and from repeated observation and

photographic records kept over the 15-month study period.

Detailed Analysis and Measurement of Key Parameters

The full characterization of the study areas at all three scales

of analysis served as a point of departure for the third phase of the

study, the measurement of water and soil exports from the individual

plots and from the nested sets of watersheds. Erosion, runoff, and

sediment transport were related to daily and continuous precipitation

measured at 15 stations in and near the Plan Sierra impact area.

In the large watersheds total sediment transport was measured and

sediment yield was calculated in order to estimate the magnitude of

the erosion problem on the watershed, to predict the future

sedimentation rates of the proposed dams, and to compare the losses

per unit area between the study areas and other sites for which

sediment yields have been reported. The measurement in the small

watersheds showed the integrated effects of land use and physical

characteristics in each study site. Subsequent tests of the data by

multiple analysis of variance indicated the relative influence of land

use and physical factors.

Experimental plots were included to demonstrate the impact of

varying specific crop types, land treatments and conservation

practices on erosion and runoff. The plots also provided the erosion

rate data necessary to calculate sediment delivery ratios for the

watersheds. Erosion plots were included because they allow

observation of the problem within the context of individual

landholdings and related households. The experiments were conducted

under the same conditions that limit and influence the management of

individual landholdings. While the most striking effects may be

expressed at the watershed level, the management decisions are made at

the household level. The degree to which such decisions are

constrained by pressures from the larger system does not change the

fact that these decisions directly determine, in turn, the condition

of the larger watershed. Any proposed changes must be tested within a

holistic framework at the level of the land managers.

Precipitation, discharge and sedimentation in the large watersheds

Records from 15 climatological stations collected by SEA's

Department of Meterology, and INDRHI provided daily precipitation

values as well as continuous data on rainfall amount and intensity at

five of the stations in the region. A multiple correlation analysis

of daily data from all 15 sites was performed to check for duplication

and overlap of information and to determine the variability of

rainfall over the study area (SAS, 1979).

The distribution of rainfall over each watershed and its

subdivisions was determined by the Thiessen polygon method (Wisler and

Brater, 1959). The daily rainfall values from the climatological

stations were extrapolated to the surrounding areas, then aggregated

into units more relevant to the study, such as subwatersheds. The

Thiessen polygons defining the area of influence around each station

were delineated and superimposed on a map of the Mao and Amina

watershed divisions, at a scale of 1:250,000 (Fig. 7). The watersheds

were defined by cartographic analysis of topographic maps at the

1:50,000 scale, and Thiessen polygons were constructed according to

standard methods summarized by Kenah (1980). After determining the

proportion of each subwatershed that fell within the polygon assigned

to each station, a weighted average of daily rainfall was calculated

for each subwatershed.

Stage and discharge measurements. On both rivers, hydrometric

stations maintained by INDHRI were located conveniently near the

downstream borders of the Plan Sierra impact area (Fig. 8), and

relatively close to concrete bridges. The river cross-sections were















4 0

4 -H

* *

J 'c-.

0 > x

S 0,
c o -

E .D *
- 0 3
(5 $ cn


- /
- r
- 1




gauged and calibrated periodically, such that the stage measurements

recorded at 0700 and 1700 hrs daily, as well as maximum flood stage,

could be converted directly to discharge rates based upon a nomograph

constructed by INDRHI hydrologists (see Appendix C). The stage-

discharge graphs relate discharge rates obtained through periodic

field measurements using the area-velocity method (Grover and

Harrington, 1943; Herschy, 1978) to the simultaneous reading of river

stage on a simple gauge.

For purposes of this study the nomographs of each river were

segmented and the equations for each segment were determined by simple

linear regression using the Statpak package of statistical programs

(MUSIC, 1967). The gauge readings recorded for morning, evening, and

flood peaks during the study period were converted to discharge rates

using the appropriate equations derived from the INDRHI nomographs.

Stage measurements made at the bridges by project personnel were

compared to simultaneous measurements at the hydrometric stations. A

simple linear regression equation converted the measurements made at

the bridge at sampling time to a gauge reading for the station. This

procedure replaced independent velocity-area discharge measurements

and allowed the substitution of a simple indirect method for the more

difficult and lengthy procedure used initially.

Comparison of rainfall and discharge during the study period to

the historical period. The total monthly rainfall totals over the

study period were compared to the average monthly rainfall totals over

the full period of record at stations with 11 or more years of data.

The SAS (1979) means program was used to compare the data from each

station for 1980 and 1981 with the prior records. The comparison of

river discharge measurements during the study period and the full

period of record followed the same procedure.

Measurement of suspended sediment concentration. Direct

measurement of sediment concentration required periodic sampling of

river water over the full 15-month study period, spanning three wet

seasons. This allowed sampling under a variety of conditions from low

flow to flash floods. The timing of observations was designed to

sample as wide a range as possible of the variation in discharge and

sediment concentration. In some cases time series samples were drawn

to cover the rise and fall of a particular flood.

The samples were extracted at the bridge site with a modified,

locally built Uppsala-type sampler (Fig. 9), as described by Nilsson

(1969) and Rapp (1977). The finished product resembles the USDA-48

Wading Sampler (USDA, 1979). The diameter of the sample intake nozzle

(0.64 cm) limited the particle size to a maximum of approximately 0.3

cm, which probably did not exclude any suspended sediments. Bedload

was not sampled. The emphasis on suspended sediment load as based on

the high average ratio of suspended sediment to total sediment load

(Gregory and Walling, 1973) and on the even higher ratios reported for

turbulent conditions.

While the sampling sites were not ideal by hydrometric criteria,

the accessibility and safety of bridge crossings outweighed other

factors (Herschy, 1978). Wading for samples from steep banks at river

narrows entails undue risk, particularly in areas subject to flash

flooding. The accessibility of the sampling site also can limit

and water intake


Nolgene plastic
collecting bottle

Fig. 9. Uppsala-type manual sampler for instantaneous
measurement of sediment concentrations in
streams (Rapp, 1977; Nilsson, 1969; USDA, 1979).

the number of observations. The bridge sites allowed simple grab-

sampling during flood conditions.

Initial results from multiple samples stratified throughout the

river cross-sections showed little variation across the section but

pronounced differences with depth. Subsequent samples were taken in

pairs at 10 and 30 cm below the surface (water depth permitting) at

the center of the section. The samples normally were drawn by

extending the sampler downward from the upstream side of the bridge,

directing the intake into the flow, and extracting the sample just

prior to filling the bottle, to avoid sampling error.

The 1-2 L samples were stored at room temperature at field

headquarters, then shipped periodically to the CENDA laboratory.

Sediment content by dry weight was determined and reported as
concentration (g L ). Analyses were conducted according to standard

methods for the determination of suspended solids in water. After the

sample volume was recorded, the samples were shaken, then poured into

Gooch crucibles lined with pre-weighed Whatman No. 5 fiberglass filter

paper. After the sample was strained into a flask, under a slight

vacuum, the filter and collected sediments were oven-dried at 105C

for 24 hrs, then weighed. The final weight minus the previously

determined paper weight gave the new weight of the sediments for a

given sample. The latter then was divided by the sample volume to

obtain the sediment concentration.

Sediment discharge rates and total sediment transport.

Instantaneous rates of sediment discharge in tons sec were
calculated by multiplying total river discharge (m sec ) by the

sediment concentration (tons m 3):

-l 3 -l -3
Sq (tons sec ) = Q (m sec ) x Cs (tons m )

where Q = river discharge, Cs = concentration of sediment, and Sq =

sediment discharge rate.

Daily sediment discharge for non-flood sampling days was

calculated by averaging morning and evening discharge rates from stage

measurements. This was multiplied by a time conversion factor to

obtain total river discharge for the day. The total discharge times

the concentration approximates total sediment transport past the

sampling point for that day.

For samples drawn during or very close to short-lived peaks, the

flood peak duration was estimated from field records and observations

as well as from reports by the hydrometric station operator and other

nearby residents. River discharge is derived from measurements or

estimates of the flood stage, using the stage-discharge equations

described above. The instantaneous rates of sediment discharge were

calculated by multiplying sediment concentrations times the discharge

at the time of sampling. In cases of time series sampling during

flood events the average discharge rate for each time interval was

converted to a discharge value, then multiplied by the sediment

concentration. The sum of the river discharge and sediment discharge

over the sampling period provided empirical measures of sediment

transport for flood events of a given magnitude.

Analysis of relationships between discharge, sediment

concentration, sediment transport and rainfall. The frequency

distributions of all variables were tested by frequency analysis (SAS,

1979). Based on the results of the preliminary analysis the

relationships between discharge, an independent variable, and sediment

concentration and transport (dependent variables) were tested by

simple linear regression of raw and log-transformed data (SAS, 1979).

The relationships between river stage and sediment transport were

used to estimate sedimentation rates of the dams to be constructed

just downstream of the sampling points in both rivers. The daily

discharge for a full hydrologic year during the period of record was

used to generate daily sediment discharge values based on the

relationship established in the previous analyses. The total was

multiplied by a correction factor (ratio of average annual discharge

to the discharge for the 1980-1981 hydrologic year) to predict the

sediment discharge for an average year, as opposed to the study


The relationship between the amount and distribution of rainfall

on the watershed (independent variable) and the discharge and sediment

concentration in the rivers (dependent variables) was tested by simple

and multiple linear regression of raw and log-transformed data (SAS,

1979). The same procedure was repeated by subwatershed. Critical

areas for further study were singled out by the relative strength of

association between rainfall in each subarea and the subsequent river

flood stages and sediment concentration. This information, combined

with field reconnaissance, contributed to assignment of research

priority by subwatersheds.

The relationships between total daily rainfall on the whole

watershed and discharge and sediment concentration also were tested by

simple linear regression. The total volume of rainfall on the

watershed was compared to total volume of discharge, by month, and the

remainder of rainfall minus discharge was attributed to storage and

evaporation. The results were compared to water balances previously

calculated for weather stations in or near the study area.

Precipitation, discharge, sedimentation, and production in small

The proximity of three climatological stations to the respective

study areas allowed direct application of the rainfall data from these

stations. To supplement the existing monitoring network, small

plastic water gauges with 5-cm apertures were mounted at eye level on

wooden supports installed in well-exposed open areas. The amount,

intensity, and duration of rainfall on a daily basis during peak

rainfall periods were recorded.

Discharge and sediment concentration. Discharge measurements

made under relatively low flow conditions followed the procedures

prescribed by the velocity-area method (Herschy, 1978). The surveyed

cross-sections in each stream were segmented and a velocity

measurement was made at the center of each segment.

Surface flow velocity was measured with floats and chronometer as

illustrated in the diagram (Fig. 10). Each velocity measurement

consisted of three to five readings (sec), that were converted to flow
rates (m sec ), then averaged. One complete velocity measurement was

made for the center of each segment, except in cases where the cross-

section was treated as a single segment. The discharge rate (Q) is

equal to the velocity (V) times the cross-sectional area (A), or a

segment thereof (Ax):

Q(m3 sec-1) -1 2
Qmsec ) = V(m sec ) x A(m ).

S..... ... ....

*. . . . . .

. '.. . . .
. . ...





) 0

*rJ -1

4-) 0-

4J E
(10 1

0 U)
r U)


E Ou



(0 ,-
z >

a) 0


ra 0
0 r


CI -4

*H) E

o --

04 )
I) 0)

> '0

-1 "

0 U
> ro



H- 0



.. .. .. ... .
. .........
. ........
.. .. .. .. .