• TABLE OF CONTENTS
HIDE
 Front Cover
 Title Page
 Abstract
 Index
 Introduction
 The assessment of farmer circumstances:...
 Definition of recommendation domains...
 The first cycle of experiments
 The second cycle of experiment...
 Cycle III: Narrowing the research...
 The research strategy for cycle...
 The national response
 Reference






Group Title: On-farm research methodologies at work : : progress report from Les Cayes, Haiti
Title: On-farm research methodologies at work
CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00080074/00001
 Material Information
Title: On-farm research methodologies at work progress report from Les Cayes, Haiti
Physical Description: 44 p. : ; 28 cm.
Language: English
Creator: Yates, Michael
Martínez, Juan Carlos
International Maize and Wheat Improvement Center
Publisher: International Maize and Wheat Improvement Center
Place of Publication: Mexico City Mexico
Publication Date: 1984
 Subjects
Subject: Corn -- Haiti   ( lcsh )
Farmers -- Haiti   ( lcsh )
Agricultural innovations -- Haiti   ( lcsh )
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Spatial Coverage: Haiti
 Notes
Bibliography: Includes bibliographical references (p. 44).
Statement of Responsibility: Michael Yates, Juan Carlos Martínez.
General Note: On cover: Economics program.
 Record Information
Bibliographic ID: UF00080074
Volume ID: VID00001
Source Institution: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 15324848

Table of Contents
    Front Cover
        Front Cover
    Title Page
        Title Page
    Abstract
        Abstract
    Index
        Index
    Introduction
        Page 1
        Page 2
    The assessment of farmer circumstances: First step in the research process
        Page 3
        Page 4
    Definition of recommendation domains and identification of research opportunities
        Page 5
        Page 6
        Page 7
    The first cycle of experiments
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
    The second cycle of experiments
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
    Cycle III: Narrowing the research focus
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
    The research strategy for cycle IV
        Page 38
        Page 39
        Page 40
    The national response
        Page 41
        Page 42
        Page 43
    Reference
        Page 44
Full Text























On-Farm Research Methodologies at Work.
Progress Report from Les Cayes, Haiti

Michael Yates *

Juan Carlos Martinez *


* Economics Program, CIMMYT.









Abstract

On Farm Research Methodologies at Work. Progress Report
from Les Cayes, Haiti

Michael Yates *

Juan Carlos Martinez *

In 1980 the Department of Agriculture (DARNDR) of the
Government of Haiti decided to explore the potential
contribution of on-farm research methodologies (as a liason
and needed complement to traditional station research and
extension activities), to increase the technical and
institutional capacities of the country to provide the
Haitian farmer with technological alternatives appropriate
to his agroeconomic circumstances. In this framework an
area-specific on-farm research program was defined to be
carried out by national staff with CIMMYT's cooperation.
The program started in February 1981 in the Les Cayes area
(southwest Haiti) with maize as a target crop.

The goals of the program were twofold. On one hand it
was to generate appropriate maize technologies for
increasing productivity and income of representative area
farmers in the near term. On the other hand it was intended
to be a source of concrete methodological experiences which
would contribute to guiding the Haitian National Research
Program in establishing OFR operations at national level.
The paper describes the different stages in the
implementation of this program, which led to a final
evaluation of the experience in 1983 by the Government of
Haiti. 'As a result of this evaluation a decision was taken
to extend the Program to other regions and other crops
within Haiti. This should start in early 1985.

Beyond its technical parts, the paper also illustrates
the bottom up CIMMYT strategy, centered on on-farm research,
addressed to strengthening the capacity of national programs
to generate and transfer appropriate technologies for target
groups of farmers.


* Economics Program, CIMMYT.












INDEX

I INTRODUCTION: THE NATIONAL CONTEXT AND THE ROLE
OF CIMMYT COLLABORATION IN ON-FARM RESEARCH. . 1

II THE ASSESSMENT OF FARMER CIRCUMSTANCES: FIRST
STEP IN THE RESEARCH PROCESS. . . 3

III DEFINITION OF RECOMMENDATION DOMAINS AND
IDENTIFICATION OF RESEARCH OPPORTUNITIES. . 5

IV THE FIRST CYCLE OF EXPERIMENTS . . 8

1. Experimental Strategy and Trial Management. 8

2. Analysis of Results . . . 12

3. Integration of Experimental and Survey Results
and Implications for Further Research . 19

V THE SECOND CYCLE OF EXPERIMENTS . . 20

1. Experimental Strategy and Trial Management 20

2. Analysis of Results . . . 21

3. Integration of Results and Implications for
Further Research . . . 28

VI CYCLE III: NARROWING THE RESEARCH FOCUS . 30

VII THE RESEARCH STRATEGY FOR CYCLE IV . . 38

VIII THE NATIONAL RESPONSE . . 41

REFERENCES . . . . 44









On-Farm Research Methodologies at Work. Progress Report
from Les Cayes, Haiti. *


Michael Yates** and Juan Carlos Martinez **


I INTRODUCTION: THE NATIONAL CONTEXT AND THE ROLE OF
CIMMYT COLLABORATION IN ON-FARM RESEARCH


Small-scale agriculture characterizes the Republic of
Haiti. Approximately 75% of the nations' 6 million
inhabitants reside in the rural areas, and over 900,000 has.
are cultivated annually. Population density in the
countryside is quite high, approximately 470 persons/km2
cultivated, and the GNP per capital is among the worlds'
lowest (estimated at U.S. $280.).


Land scarcity, population pressure and the overall
socioeconomic situation of the country suggest that
technological change may play some role as a potential force
contributing to improving the productivity and income of
Haitian farmers.


Within this framework, the Haitian Department of


* This paper draws heavily on the CIMMYT-DARNDR Progress
Report on the Area Specific On-Farm Research Program in
Les Cayes, Haiti (Mexico, April .1984: 168 pages). The
staff involved in the Program included, in addition to
the authors, Eliassaint Magloire, Agronomist DARNDR
(Presently graduate student in the University of
Florida, Gainesville) and Hugo Cordoba, Regional
Agronomist, CIMMYT Maize Program (presently working as
a Maize Breeder at CIMMYT headquarters). The Program
has been funded by the Canadian International
Development Agency (CIDA), the Swiss Development
Corporation, and the Rockefeller Foundation, during the
different stages of its implementation. The views
expressed here are the authors' and not necessarily
those of the MARNDR, CIMMYT, CIDA, Swiss Development
Corporation or the Rockefeller Foundation.

** Economics Program, CIMMYT.









Agriculture wished to explore the potential contribution of
on-farm research methodologies (as a liason and needed
complement to traditional station research and extension
activities) to increase the technological and institutional
capacities of the country to provide the Haitian farmer with
technological alternatives appropriate to his agroeconomic
circumstances. In this context an area-specific on-farm
research program was defined for the Les Cayes region of
southwestern Haiti, to be carried out by the national staff
with-CIMMYT's collaboration.


The goals of the program were twofold. On the one hand
it was intended to generate improved and appropriate maize
technologies to increase productivity and incomes of
representative farmers in the area in a reasonably short
period of time. On the other hand it was to serve as a
source of concrete methodological experiences which could
help guide the Haitian national research program in
establishing OFR operations at the national level. In this
context the project in Les Cayes would be closely monitored
and evaluated by the national program throughout its
development in order to assess its potential role as a model
for the implementation of OFR procedures in other parts of
the country. This would obviously also give it a
potentially important role in an eventual
institutionalization of OFR in Haiti.


Maize was selected as the target crop of the OFR
program, as an important leverage point in the prevailing
farming systems. It is the most important grain in the
country (annual production close to 300 000 tons and 30% of
all cultivated lands) followed by sorghum and rice.









II. THE ASSESSMENT OF FARMER CIRCUMSTANCES: FIRST STEP IN
THE RESEARCH PROCESS


The first step in CIMMYT's strategy / for developing
appropriate technologies is to obtain a thorough knowledge
of farmer circumstances, or the biological and
socio-economic environment in which the farmer makes his
production decisions. This enables the research team to
identify some priority production constraints offering
promising research opportunities, and to define some
tentative "recommendation domains" or roughly homogeneous
groups of farmers for whom a single recommendation may be
made. Information on farmer circumstances is usually
obtained both from secondary sources and from surveys of
farmers in the target area.


A rapid informal survey was carried out in 'the Cayes
Plain by a joint DARNDR/CIMMYT team in February 1981, before
planting the first cycle of on-farm trials. This
information was used to design the initial experiments in
the farmers' fields. A more thorough survey of randomly
selected farmers was done some months later, and it was then
possible to collect quantified data on certain key variables
and to verify some of our earlier hypotheses coming out of
the exploratory survey. This helped "round out" the picture
of farmer circumstances and gave greater precision both to
the definition of our tentative recommendation domains and
to the design of successive cycles of experiments in the
farmers' fields.


1/ The basic criteria guiding such strategy may be seen in
J.C. Martinez, "La Mise au Point D'Une Technologie
Adaptee aux Contraintes et Atouts de L'Agriculteur:
L'Approche du CIMMYT." Document de Travail 01/84.
CIMMYT, Mexico, 1984.
2/ For more details on the concept and its role see:
CIMMYT "Planification de Technologies Appropiees pour
les Agriculteurs: Concepts et Procedes." CIMMYT,
Mexico, 1983.









A brief summary of some relevant farmer circumstances
in the Cayes Plain follows below.


The Les Cayes region receives an average annual
rainfall of approximately 1700-2000 mm, and the principal
rainy season is from February to May. The elevation of the
Plain is sea level to 100 meters above, and the region is
surrounded by hills and mountains. The Plain covers
approximately 20,000 ha., and the population was estimated
at 213,675 in 1979-19S0 (calculated from SNEM data).


Patterns of land tenure are complex in rural Haiti, and
sharecropping is common. Holdings are typically small, and
in 1971 approximately 70% of all farmers in the Cayes Plain
owned less than one hectare of land (FAO 1973;78). Most
lands are rainfed with no irrigation.


Despite the small size of most holdings hiring workers
is common, and workers earn approximately U.S.$0.60-0.80/day
(usually morning only).


The capital market in the Cayes Plain is quite
restricted, though local money lenders are found with
interest rates generally 10% per month. As far as purchased
j.nputs are concerned, most maize farmers do not apply
chemical fertilizers to their fields though many purchase
maize seed at rural markets in the spring. Herbicides are
unknown though some insecticides are used, especially Sevin
and DDT. As with fertilizers, these are sometimes available
for purchase at the DARNDR offices in Les Cayes.


The great majority of small farmers prepare their
fields for planting with animal traction. Most hire others
who own animals and plows to do this task, and costs are
approximately U.S.$60.00/ha. Most farmers sell a portion of
their harvest to itinerant buyers (Madame Sarahs) right from









their households, after first setting aside a supply for
short-term household consumption needs, and women generally
handle the marketing transactions.


Maize is clearly the most important crop grown in the
Cayes Plain, covering 37% of the area cultivated, and the
average area planted in maize per maize farmer was .56 ha.
Prices fluctuate seasonally, and in 1983 post-harvest prices
averaged U.S.$0.24/kg.


Maize is planted both independently and in association
with a wide range of crops, though monocropping is the most
common practice. Patterns of crop rotation vary, though a
typical unirrigated plot is prepared for planting in
February or March. Most farmers in the Plain plant two
maize crops per year, though this is rare on the drier
slopes. The second maize crop is often relayed with sorghum
and pigeon peas which are harvested in December and January.
The field is then plowed once again when the rains begin for
the next cycle of maize. Few farmers in this densely
populated region let much land go out of production, though
some do allow a short-term fallow. This suggested that
chemical fertilizers could have very important impacts on
production in the Cayes Plain.


III DEFINITION OF RECOMMENDATION DOMAINS AND IDENTIFICATION
OF RESEARCH OPPORTUNITIES


The-variability in farmer circumstances and observed
farming systems made it difficult to make a strong
definition of recommendation domains at the initial stage of
the Program; this was particularly true given the limited
planning time before the first set of trials was planted
(one month). However, some basic lines of differentiation
between farmers were identified and used for. this purpose
(later on verified/quantified through the formal survey).









Hilly vs. flat topography was one of these, and this was
highly associated with the political subdivisions within the
region (more mountainous communes vs. those where flat
topography prevails). Irrigated vs. rainfed conditions was
another obvious differentiation to be considered, as well as
the alternative ways in which maize production fitted within
the farming system, given that the crop was planted both in
monoculture and in association w-ith a wide variety of other
crops. The formal survey confirmed that monoculture was the
prevailing system (58%), while all the Idifferent types of
maize associations represented 42% of the reported cases.


With these elements, a decision was taken to focus the
research initially in one recommendation domain, which could
be briefly described as-farmers producing monoculture maize
in flat non irrigated land. Representative farmers and
prevailing practices were accordingly identified for this
recommendation domain with the informal survey and later
verified/quantified with the formal survey. Similarly,
research opportunities were also identified.


The analysis of the information obtained at the initial
planning stage of the program lead a research focus on
variety, fertilizers, plant population and weed control.


The most important local maize planted by farmers was
by far Chicken Corn. / This population not only had low
yields under farmer management but it also appears to be of
low genetic potential, as suggested by some international
trials conducted at the Levy experimental station. This
led to the hypothesis that an appropriate improved variety
could have a significant impact on maize land productivity.


S The formal survey confirmed that 93% of the farmers
were planting the local Chicken Corn.








It was evident that the farmer in the Cayes Plain
strongly prefers a deep yellow hard maize with little flour,
so alternative materials were selected to meet these
specifications. Survey work showed that milling properties,
taste preferences and ease of shelling also had to be taken
into account as they too were expected to play important
roles in the potential adoption of new varieties in the
area.

4/
.Most farmers were not applying chemical fertilizer
to their maize. However survey work revealed that fields
were generally planted successively cycle after cycle and
few farmers allowed more than an occasional short term
fallow, suggesting a steady depletion of key soil nutrients.
Accordingly, the availability of nitrogen and phosphorus in
the soil were hypothesized to be important factors limiting
productivity. Two non-agronomic issues were later
identified which had important interactions with this
research opportunity, a limited availability of these inputs
in the area and land tenancy arrangements, as owners were not
5/
sharing the cost of fertilizers with the sharecroppers. -


Farmer plant population was relatively low, with wide
spacing between rows (90-110cm) and about 44,000 plants per
hectare. This, together with the inclusion of fertilizer as
a research opportunity, led to a more detailed consideration
of plant density. A search for fertilizer-density
interaction then became a complementary research
opportunity.


Finally, perceived problems in weed control completed
the initial identification of research opportunities. The


S73% of them.

SThe formal survey indicated that almost half of the
maize fields were planted by sharecroppers.









prevailing farmer practice was two manual weedings, a very
labor intensive practice. Preliminary evidence suggested
peak-demand labor shortages and problems with timeliness of
weeding. On the other hand a general situation of easy
availability and (accordingly) low price of labor appeared
to prevail, and the interdisciplinary team in charge of the
program did not reach a complete agreement about the
different dimensions of the problem. It was therefore
decided to allocate part of the research resources to
clarify it.


Weed control was included with this qualification as a
research opportunity, and farmer's practices were contrasted
6/
with the application of an herbicide. Later on, the
formal survey showed that labor bottlenecks were actually
localized and not characteristic of the Plains in general.
Most farmers did seem to be doing their weeding on time.


IV THE FIRST CYCLE OF EXPERIMENTS


1. Experimental strategy and trial management


Two groups of experiments, variety and exploratory
trials, were planted in the first cycle. These incorporated
the experimental variables associated with the research
opportunities and the hypotheses formulated at the planning
stage.


The variety trials attempted to screen alternatives to
the local low-yielding Chicken Corn, a maize of the
Chandelle race with intermediate plant and ear height,
widely distributed throughout Haiti. It was recognized that
this population did offer certain agronomic and culinary
advantages including a short vegetative cycle, a deep yellow


6--Before harvesting the experiments.









crystalline grain type (preferred in the area) and a
relatively low flour content.


Accordingly, ten improved varieties were prescreened to
meet both existing agronomic circumstances and these kinds
of farmer preferences. These were tested in the farmers'
fields alongside Chicken Corn and another less popular local
variety, Les Anglais.


The variety experiments, of a completely randomized
block design with 4 repetitions, were planted in 5
localities within Les Cayes (including Levy Farm). Table
IV-1 presents all the varieties tested.


The plots had 4 rows of 5 meters length with 75 cm
between rows. The distance between hills was that most
commonly used by the local farmers (50 cm), and three seeds
were planted per hill to be thinned to two after
germination. The trials were fertilized with 50 kg P20 at
planting and 80 kg N split into two applications (half at
planting, half one month later). Manual weeding and land
preparation was done by the farmers themselves using the
most common farmer practices (e.g. two manual weedings with
hoe, plow and bulls). Locally available insecticides were
applied when necessary to control insect damage.


The exploratory trials incorporated the remaining
experimental variables associated with the research
opportunities identified, namely: 1. Fertilizers (nitrogen
and phosphorus), 2. plant population and 3. weed control. A
randomized incomplete block design was used, with a 2
factorial arrangement. Block effects were confounded with
the third order interaction, and two .repetitions per site
were planted.


One of the levels of the experimental variables was set












Table IV-1
Average Yields and Agronomic Characteristics of 12 Maize
Evaluated in Five Locations, Les Cayes 1981


Varieties


LOCATIONS


Variety
LM 7928
LM 7827
TOC 7926/1
S. Lagoas 7728
PR 7427
PR 7928
Les Anglais
Levy 7835
PR 7726
Across 7726
Chicken Corn
Toc. 7831/1


22 16


21 15


L.S.D. NS 1.004 U.S. 0.548 1.036


Levy
3.3
3.3
3.1-
3.5
4.0
3.2
3.0
3.4
2.6
2.4
3.5
2.4


Laborde
5.0
4.7
3.-3
3.6
4.1
3.5
4.1
3.2
3.5
3.3
2.6
2.6


Beraud
4.3
4.0
4.7
3.8
3.2
3.6
3.4
4.2
3.8
3.8
2.9
3.0


Gerard
3.1
2.8
3.0
2.7
2.0
2.8
2.6
2.3
2.1
2.8
2.3
2.0


Burin
5.2
5.7
4.3
4.2
4.6
4.5
4.5
3.7
4.7
3.8
4.0
3.4


CV%


Combined
Average
4.2
4.1
3.7
3.6
3.6
3.5
3.5
3.4
3.4
3.2
3.1
2.7


Duncan
a
ab
bc
cd
cd
cd
cd
cd
cd
cd
de
e


Plant
Height
(cm)
189
193
179
198
196
186
221
155
174
182
204
156


Ear
Height
(cm)
93
93
-88
100
100
94
123
70
81
81
108
70


Days to
Female
Flower
59
59
55
60
58
57
64
54
55
57
57
51


Lodging
4.2
4.8
4.4
3.4
5.6
6.7
7.1
6.6
5.8
5.8
5.8
4.0


No.
of
Ears
31
30
30
30
28
31
26
29
30
29
33
29


%
Rotten
Ears
11
13
11
18
11
13
14
11
20
15
4
21


L.S.D. NS 1.004 N.S.


0.548 1.036









at the prevailing farmer practice, while the other was set
at a level that was believed to be feasible for the local
farmers and that would still allow the detection of main
effects and interactions, should they exist. The
exploratory trials attempted: 1) to verify the hypotheses
about the importance of the research opportunities
identified and 2) to "explore" the feasibility of new
technological components considered as potential
alternatives to prevailing farmer practices.


The levels of experimental variables were set as
follows (the zero subscript corresponds to the farmer
practice and the subscript one to the alternative):


1. Nitrogen


NO: no nitrogen


N1: 80 kg/ha. source, urea; split half at planting
and half one month later, applied to holes near the
maize plant.


2. Phosphorus


P : no phosphorus
PI: 50 kg/ha. Source, triple superphosphate, all at
planting time applied in holes near the maize plant.


3. Density


Both DO and D1 with farmer
DO: 44,000 plants/ha practice for row spacing, but
D1: 55,000 plants/ha varying in D1 the spacing
between hills within rows.
4. Weed control









H0: Two manual weedings, about three and six weeks
after planting.


H1: 1.5 kg atrazine/ha in preemergence.


All non experimental variables (including variety) were
set at representative farmer's practices.


2. Analysis of Results


Table IV-2 presents main effects combined and by
location from the exploratory trials. A first look at the
Table indicates a highly consistent positive effect across
locations for nitrogen, which on the average gives a yield
increase of .96 ton/ha. This consistency observed for N. is
not paralleled by any of the other technological components.
In particular, the main effect for weed control was pretty
close to zero (with negative and positive values depending-
on location).


Table IV-3 presents the analysis of variance for each
location. (In accordance with previous observations)
nitrogen had a significant effect on yields in all
locations. For phosphorus and density only 25% of the
locations showed significant responses' (in the case of
density one of the significant responses corresponded to a
negative main effect Burin). The case of weed control was
more clear but in the sense of no response; only one of the
locations presented statistical significance with a positive
main effect. Finally, no systematic significant
interactions were found between the variables.


Given the consistency of nitrogen response the economic
dimension was introduced in order to assess this components'
feasibility for the farmer, given the pricing and structural
context they confront in Les Cayes.







Table IV-2

Exploratory Trials. Main Effects by Location
and Across Locations


LOCATIONS
Combined

Levels Pesmerle Boval Beraud Laumond Goin Chantigny Ducis Burin Average


No

N1

difference

P0

P1
difference

DO

D1

difference

H0

.H1


3.29

2.73

0.55

3.17

2.34

0.33

2.89

3.12

0.23

2.82

3.19


2.30

0.61

1.69

1.39

1.52

0.13

1.44

1.47

0.03

1.58

1.34


2.50

1.96

0.54

2.00

2.45

0.45

1.91

2.55

0.64

2.25

2.31


TM/ha 14%

1.99

1.10

0.90

1.49

1.60

0.11

1.45

1.65

0.20

1.54

1.55


H20

2.54

1.39

1.15

1.96

1.97

0.01

1.78

2.15

0.37

1.95

1.98


2.12

1.73

0.39

1.95

1.90

-0.05

1.86

1.98

0.12

1.92

1.93


1.38

0.62

0.76

0.79

1.21

0.42

1.02

0.98

-0.05

1.01

0.99


3.57

1.87

1.71

2.63

2.81

0.19

2.93

2.51

-0.42

2.85

2.58


2.46

1.50

0.96

1.92

1.98

0.20

1.91

2.05

0.14

1.99

1.98


difference 0.37 -0.24 -0.04 0.01 0.04 0.01 -0.01 -0.27 -0.02


-0.04 0.01 0.04 0.01


-0.01 -0.27 -0.02


difference 0.37 -0.24










Table IV-3

Exploratory Trials. Analysis of Variance by Location


Experiirental


DEperirental
Factors

Nitrogen

Phosphorus

Density

Weed Control


Interaccions


.V.


**



1/


Pesmerle

**


14.5%

Significant at 5%
Significant at 10%
Not significant.

Main effect is negative


Boval

**


Beraud

**


** 1/

NP*
PH**
DH*

13.9%


18.8%


Lautond

**


23.7%


Goin

**


19.8%


Chantigny

**


NP**



15.6%


C


Ducis

**


N ~PH**

22.2%


Burin

**


14.5%


14.5%









Table IV-4 presents the basic economic analysis for the
combined results of eight locations. As the table
illustrates two elements are strongly conditioning the
economic feasibility of this input. These elements are
fertilizer availability and land tenure arrangements.


On the latter issue, approximately one half of the
maize farmers interviewed were sharecropping maize. Typical
arrangements compel the sharecropper to give 50% of the
harvest to the owner of the land though fertilizer costs are
generally not shared by the landlords. Sharecroppers
therefore get only half of the benefits even though they
must pay (under prevailing arrangements at the time) 100
percent of fertilizer costs should they decide to use
fertilizer. Under these circumstances, nitrogen returns for
owners and sharecroppers are dramatically different.


The other element affecting the economic feasibility of
nitrogen use was the type of fertilizer available in the
area. The Haitian Department of Agriculture in Les Cayes
was selling fertilizers at a constant price of 10 U$S/100
lbs. for any type (whether a complete formula, ammonium
sulfate or urea). However, even though urea was clearly the
cheapest source of nitrogen, it represented only a small
percentage (4%) of the total fertilizer available for Les
Cayes. Rates of return were accordingly computed for two
nitrogen-pricing scenarios, one using urea as the source of
nitrogen, and the other the most widely available formula
for the area, 18-8-20.


Results shown in Table IV-4 indicate that for the case
of landowners the marginal rate of return of applying 80 kg
of N is on the average 279 or 115%, depending on the source
used (urea or 18-8-20 respectively). In both cases they are
comfortably above the opportunity cost of capital (about 60%
for the crop period). However, while this is true with the








Table IV-4


Economic Analysis Response to Nitrogen *
(NO = Okg N/ha; N1 = 80 kg N/ha; Variety = Chicken Corn
Combined Average
N0 N1

Yield (TM/ha) 1.5 2.5

Gross Benefits (PCM=$182/TM)
--Landowner (GBL) 273.0 455.0
--Sharecropper (GBS= GBL) 136.5 227.5

Variable Costs with Urea -- 48.0
Fertilizer (FPN/Urea) = $.50/kg 40.0
Labor 8 days at 0.50/day -- 8.0
per application (2 app.)

Variable Costs with 18-8-20 -- 84.8
Fertilizer (FPN/18-8-20) = -- 76.8
0.96/kg
Labor 16 days 0.50/day 8.0

Landowner
NB with Urea 273.0 407.0
MRR with urea (%) -- 279%
NB with 18-8-20 273.0 370.2
MRR with 18-8-20 (%) -- 115%

Sharecropper
NB with Urea 136.5 179.5
MRR with Urea -- 90%
NB with 18-8-20 136.5 142.7
MRR with 18-8-20 -- 7%






* Evaluation methodology follows the CIMMYT Manual
"Comment Etablir des Conseils aux Agriculteurs a Partir
des Donnees Experimentales", Richard K. Perrin, Donald
L. Winkelmann, et. al. Bulletin d'information 27.
CIMMYT, Mexico, 1.979.









average values, a marked difference appears if we consider
variability by location. Using urea as a source of N, the
25% on the lower tail of the yield distribution (worst
responses) still give a MRR of about 80%. The picture is
different if we consider 18-8-20 however. In this case the
25% worst cases would be associated with rates of return
which are close to zero or negative. Finally, for
sharecroppers average MRR's are 90% or close to zero (7%)
depending on the source of nitrogen.(urea or 18-8-20).


In summary, the economic analysis confirms that, with
the qualifications considered, nitrogen fertilizer
represents a research opportunity with agronomic and
socioeconomic dimensions that should continue to figure
prominently in the future stages of the program.


A similar analysis was conducted for the variety
trials. A summary of results is presented in Table V-1 The
highest yields across locations were obtained with the
varieties La Maquina 7827 and La Maquina 7928. Both gave
almost identical yields, at 4.10 and 4.19 tons/ha.
respectively, an increase of about 1 ton over the local
Chicken Corn. Considering results by location, both
varieties were above Chicken Corn in all locations with the
exception of Levy, which favoured the local variety by 200
kg.


The combined analysis of variance (Table IV-5) shows
significant differences for "variety" as well as for
"locations" and "location x variety". This means that, all
the varieties considered, they respond differently according
to locations. However this is not true for the differential
effect of both "La Maquinas" since they are above Chicken
Corn in 80% of the locations. On the other hand, where this
does not hold, that is at Levy Farm, there is.only a minor
difference of 200 kg/ha.










Combined Analysis
of Maize


Source of Variation
Repetition (Locs)
Variety
Location
Locs x Variety
Error
C.V.


Table IV-5
of Variance for Yield of
Evaluated in 5 Locations
Les Caves, Haiti.

D.F. M.S. F.C.
14 1463824 [3.46
11 2989129 7.07
4 2293642 54.26
44 725176 1.72
154 422704


12 Varieties
in


Significance
**
**
**
**


18.78


** = Significant at 0.01%


In terms of other agronomic characteristics, little
difference was observed in days to flower between the
improved varieties and Chicken Corn, though the local Les
Anglais was the latest to mature. All the maizes performed
well with respect to lodging, and the highest number of
lodged plants.was recorded for Les Anglais. Chicken Corn
had the smallest percentage of rotten ears (4.33%) among the
varieties tested.


Given these results, the vegetative cycle, quality and
taste of the ground maize, and grain type of both these
improved varieties were also evaluated. These impressed the
collaborating farmers very much (especially La Maquina
7827).


Overall results suggested that both La Maquina 7827 and
7928 appear to offer potential for the Les Cayes region, and
deserve to be considered for further research in the area.









3. Integration of Experimental and Survey Results and
Implications for Further Research


The congruence of experimental results and the
information coming from the surveys (particularly from the
formal survey) permitted us to verify/adjust the research
hypotheses formulated on the basis of the initial informal
survey. This in turn led to a "fine tuning" of the
experimental strategy for the second cycle, using for this
7/
purpose all the information available prior to planting. /
In the first place, the hypotheses about variety and
nitrogen fertilization have been confirmed. Both
technological components should figure prominently in
further research. The following adjustments in the
experimental strategy were proposed for these components in
the second cycle:


1. Incorporate variety as experimental variable in the
exploratory trials, contrasting Chicken Corn with the
most promising alternative (La Maquina 7827) and search
for potential interactions (particularly variety X
nitrogen).


2. Implement variety X nitrogen level trials to have more
precision on the potential recommended levels, as well
as on interactions.


With regard to phosphorus and plant density, both had
inconclusive effects on yields. While they apparently did
not offer as promising a research opportunity as nitrogen
and variety, both components were important in some
locations and further evaluation was considered necessary


S For more details see Magloire, E and Yates, M.
"Recherche Chez Les Paysans", CRDA/FAMV Seminaire Sur
Les Systemes de Production Agricole en Haiti, Limbe,
Haiti, 23 pp. 1984








before a final determination was made on their role in the
research program.


Method of weed control was the other experimental
variable, and as outlined above this factor (i.e. atrazine)
had the least significant effect on yields of all the
components tested. In addition it did not offer important
cost advantages over present methods of manual control,
since, as the formal survey demonstrated, access to manual
labor was not generally an important production constraint.
This means chemical weed control also did not offer
important advantages in terms of timeliness of weeding, and
the experimental results combined with the formal survey
analysis suggested that research resources should not be
applied to the further evaluation of this variable in Cycle
II. Instead, it was replaced by variety in the exploratory
trials.


V THE SECOND CYCLE OF EXPERIMENTS


1. Experimental Strategy and Trial Management


With the basic lines of adjustment of the experimental
strategy described above, three sets of closely related
experiments were planned for the second cycle:


1. Exploratory trials, replacing weed control by variety.


2. Variety trials.


3. Level trials of nitrogen X variety.


However, delays in rainfall and later logistic problems
led to some final adjustments in the experimental program
while planting was taking place. These responded to a need
to reduce the work load in order to be able to finish









planting at the same time as the farmers. The objectives of
the variety and nitrogen X variety experiments were
therefore combined in the variety trials already under way,
and those left to be planted were planted at two levels of
N, zero and 80 kg.


Only two sets of trials were finally implemented
therefore, exploratory and variety, though the latter with
the important modifications described above. In addition to
these five varieties from Cycle I were eliminated based on a
screening with the information available and three new ones
entered the trials. The total number of varieties tested
was reduced to ten (see Table V-l). The exploratory trials
followed the same design and management as described for the
first cycle, the only qualification being that variety was
introduced as an experimental variable in place of weed
control. Here the farmer practice was represented by
Chicken Corn, while La Maquina 7827 was included as the most
promising alternative. Four exploratory trials with these
characteristics were planted in representative sites within
the defined recommendation domain, and this completed the
experimental program of Cycle II.


2. Analysis of Results


Table V.2 summarizes the results of the exploratory
trials. Note that the experiment in Muzaine was affected by
severe drought which reduced yields and gave a high
coefficient of variation for the experiment of 35%. Despite
this nitrogen comes out again with a consistent positive
yield effect in all locations, as can be seen from main
effects in the table. The magnitudes of the effects however
are different by location, and this is confirmed by a
significant interaction for location X N. In spite of this
the response is statistically significant in all locations,
and consequently in the combined analysis as well.









Table V-1
Variety Trials. Average Yields by Location and Corbined


with and without Nitrogen


LEVY

80


Varieties
Toc.

C.C.

P.R. 7827

L.M. 7827

P.R. 7931

Ac 7728

L.M. 7928

PR 7926

Sete L. 7728


3.59

3.27

3.84

4.10

3.66

3.77

3.79

3.59

3.79


TUFFET

80


4.27

2.90

4.71

4.56

3.32

4.23

4.83

3.77

3.89


-- -


LABATHE


TOUITLLARD
Nitrogen
80 0
Tons/ha at 14%

2.72 2.19

2.27 1.90

2.11 2.29

2.25 2.12

2.03 1.44

2.45 2.12

2.54 1.66

2.42 1.97

2.25 2.29

1.75 1.15


CCM BINED

0 80


LAVAL

80
H20

2.75

2.11

2.72

2.55

2.63

2.66

2.92

2.87

2.53


80


4.09

3.40

4.28

3.94

3.29

4.25

4.13

3.57

3.83


10 Les Anglais 4.49 41.2


0


2.18

2.18

2.23

2.24

1.64

2.05

2.14

2.57

2.40


1.98 1.92


3.49 2.87 3.17 1.98


--


0


2.19

2.07

2.17

2.01

1.90

2.05

2.39

2.98

1.84


3.49.

2.75

3.53

3.48

2.87

3.45

3.64

3.24

3.26


2.18

2.57

2.24

2.58

1.58

1.99

2.39

2.77

3.07





Table V-2
Exploratory Trials. Main Effects and ANOVA
by Location and Combined


LOCA


TI ON S


Muzaine Trouillard Tuffet Lebathe Combined
-Tons/ha 14% H20 -

N1 1.43 2.46 3.87 5.03 3.20

NO 0.89 1.53 2.44 4.54 2.35

Dif. 0.54 0.93 1.43 0.49 0.85
ANOVA ) *)(**)() (**)**)
P1 1.12 1.95 3.23 5.02 2.83

p0 1.21 2.04 3.08' 4.54 2.72

Dif. -0.09 -0.09 0.15 0.48 0.11

ANOVA (NS) (NS) (NS) (NS) (NS)
D1 1.13 2.02 3.07 4.77 2.75

D0 1.19 1.97 3.24 4.80 2.80

Dif. -0.06 0.05 -0.17 -0.03 -0.05

ANOVA (NS) (NS) (NS) (NS) (NS)
V1 1.14 2.09 3.37 5.16 2.94

V 1.18 1.90 2.94 4.40 2.61

Dif. -0.04 0.19 0.43 0.76 0.33

ANOVA (NS) (NS) (NS) (**) (**)
Interactions (NS) (NS) (NS) (NS) N X V (*)
Loc (**)
Loc X N (*)
CV % 35% 14% 23% 12% 19%


Significant at 5%
Significant at 1%
Not Significant








The effect of variety is not as strong as nitrogen but
is still positive in all locations with the exception of
Muzaine, where it is close to zero. Only in Labathe were
the differences statistically significant. However the
overall results as captured by the combined analysis were
significant as well, and this is likely due (in addition to
the contribution of Labathe) to the higher statistical
precision associated with increased degrees of freedom.
Results for both nitrogen and variety seem to confirm those
obtained in the first cycle. In addition the interaction V
X N was statistically significant.


Both phosphorus fertilization and plant density, on the
other hand, had only minimal impacts on yield once again,
and their effects across locations are close to zero
(actually negative for plant density). Phosphorus was
statistically significant in only one location, density in
none, and neither were significant in the combined analysis.
This too provides confirmation of the Cycle I results.


An economic analysis of the application of nitrogen
fertilization was also done, taking into account important
changes in both the field price of nitrogen and maize. The
cheapest source of available nitrogen in Cycle II was open
market urea, though this meant a price increase of fully 85%
for nitrogen from one year to the next. The field price of
maize, on the other hand, dropped 28% to U.S.$0.13/kg., and
this in effect meant costs increased while the economic
benefits decreased. Despite these critical constraints the
rates of return for landowners across locations to 80 kg
N/ha were still acceptable for the improved variety La
Maquina 7827 (in excess of the opportunity cost of capital),
though they were not acceptable with the local Chicken Corn
(See Table V-3). In addition site-to-site variability
became more important, and this suggested that priority
should be given to locating a more economically appropriate









Table V-3
Combined Economic Analysis of Response to Nitrogen
(N0=0; N1=80 kg/ha; V0 = "Chicken Corn", V1 La
Maquina 7827)


Yield Ton/ha
Gross Benefits: FPM=$128/ton
Landowner GBL
Sharecropper GBS = (GBL)

Variable Costs
Fertilizer FP/Urea
Laborers 8 days at
$U.S. .80/day
2 applications


Landowners
NB
MRR


Sharecroppers
NB
MRR

* Dominated


2.3


2.4


2.9


3.5


294.4 307.2 371.2 448.0
147.2 153.6 185.6 224.0


-- 81.1


81.1


294.4 307.2 290.1* 366.9
-- (-) 73.6%


147.2 153.6 104.5* 142.9*
(-) (-)


25








level of N offering less risk. This confirmed the need for
level trials (which even though planned for this second
cycle were not implemented due to the difficulties described
above).


With respect to variety trials Table V-1 shows that the
La Maquinas outperformed the local Chicken Corn in every
location, with an average yield differential of
approximately 0.8 ton/ha where fertilizer was applied. No
major differences appeared, however when varieties were
contrasted without fertilizers. Table V-4 presents the
anova by location, in Levy and Tuffet with N=80 kg/ha.
constant for all treatments and with N=0 and N=80 kg in the
other localities. Both sets were then aggregated in a
combined anova (see Table V-5) where N=80 kg has a higher
number of observations than N=0. The results again indicate
nitrogen response in those locations were it was a variable.
The contribution of varieties to yields was significant only
in Tuffet as well as in the combined analysis. LSD .was
for the combined analysis confirming the superior yields of
both La Maquinas with respect to Chicken Corn.


Variety X Nitrogen interaction (here with all the
varieties considered) was not significant. However, La
Maquina 7928 gave the highest differential yield response to
nitrogen fertilization (1.5 TM/ha.) with La Maquina 7827
close behind (1.3 TM/ha), while the lowest was found with
the local Chicken Corn (0.57 TM/ha.). This suggests that the
search for interaction between both variables may have to
continue in the next cycle. In summary, these results
parallel those of the exploratory trials and point again to
a need to evaluate levels of nitrogen by variety with more
precision in the next cycle.










Table V-4

Variety Trials. ANOVA by


Locations


Tuffet Laval
*


Labathe
**


Trouillard
**


Not Evaluated
**


8.5


21.0


11.3


17.4


Table V-5

Variety Trials. Combined ANOVA


for 5 Locations


Source DF MS F Significance
Location 4 28.8272 9.55 *
Error (a) 4 3.0178
V 9 1.0846 3.17 **
Loc. x V 36 0.4163 1.22
Error (b) 45 0.3419
N 1 14.7034 37.76 **
N x V 9 0.3091 0.79
Loc. x N 3.9512 10.15 **
Loc. x N x V 36 0.2232 0.57
Error (c) 50 0.3894
Total 199
C.V. 14.1


* Significant 5%
** Significant 10%


Source
Rep
N
V
NxV
C.V.


18.5


_








3. Integragion of Results and Implications for Further
Research


The convergence of research results from both surveys
and from experiments in the farmers' fields across more than
twenty locations and in two different agricultural years
permitted a considerable advance of the research process
towards formulating farmer recommendations.


The evidence for the potential nitrogen fertilization
has for increasing small farmer maize yields and incomes is
convincing. The combined results from Cycles I and II show
that in 15 of 15 locations the application of 80 kg N/ha had
a statistically significant effect on yield. The marginal
rates of return were also generally good for landowners,
especially with the improved La Maquina 7827, though
site-to-site variability was important with the level of 80
kg N/ha. These facts suggested a clear need to investigate
the possibility of identifying a more economically
appropriate level of N.


The combined results from both Cycles also indicate
that La Maquina 7827 may have potential for the farmers of
the Cayes Plain, as the improved variety out-yielded the
local Chicken Corn in 13 of 15 locations (Chicken Corn
yielded more in only 1). La Maquina 7827 also performed as
well as (or better) than the local maize under difficult
production conditions and where no chemical fertilizers were
applied. However the results have not been conclusive in
terms of interaction variety X nitrogen and future research
should continue to address the issue.


Present levels of both phosphorus and plant density, on
the other hand, do not appear to be important maize
production constraints in the Cayes Plain. The combined
results from both Cycles show that both had a significant









effect on yield in only 3 of 12 locations. It appears these
alternatives may now be dropped for the near future research
with a fair degree of confidence to allow a more intense
concentration on the most promising research components,
nitrogen and variety.


The confluence of the above arguments suggest the
following lines of concentration for the third cycle of
experiments:


1. Level trials in nitrogen X most promising varieties
addressed to a) confirming the yield effects of both
components, b) searching for the most appropriate
economic level of N, and c) attempting to clarify the
issue of N X V interaction.


2. Verification trials in N X V in order to advance both
technological components towards farmer
recommendations. Since these trials are farmer managed,
they would permit us to verify the performance of both
V and N in the hands of farmers themselves.


In addition, the potential pricing scenarios for
maize/fertilizers should be analyzed, as well as the
potential sources of appropriate fertilizers for maize
farmers in the area (whether private market or Department of
Agriculture). A consideration of the potential issues of
seed production associated with La Maquina 7827 should also
begin, in anticipation of an eventual recommendation of the
variety. Results up to this point illustrate how this
approach to OFR effectively "zeros-in" on the two most
promising research opportunities, nitrogen fertilization and
variety, and eliminates further work with less significant
technological components like chemical weed control,
phosphorus, and plant density. This continual process of
narrowing the research parameters while intensifying the








focus of research has been illustrated in Table V-6. As the
priority problems are solved and the alternative solutions
(technological alternatives) reach the farmers, new problems
will press for priority and new research opportunities will
capture the attention of the research.


VI CYCLE III: NARROWING THE RESEARCH FOCUS


The analysis of the Cycle I and II experiments and the
information from the farmer surveys indicated a need for
more in-depth information on the two key technological
components, nitrogen fertilization and variety, and on their
possible interactions, prior to the development of farmer
recommendations. The Cycle III experiments were designed
with these objectives in mind.


In the level of N X V trials four levels of N were
evaluated, 0, 40, 80 and 120 kg/ha (split into two
applications; source = urea). N X V interaction remained as
a research theme after Cycle II, and both La Maquinas and
the local Chicken Corn were contrasted here at four levels
of N.


The components were evaluated in seven locations in a
randomized complete block design with a 3 X 4 factorial
arrangement. Trial management for all the non-experimental
variables followed representative farmers' practices for the
recommendation domain.


Table VI-1 presents the yield response to the two main
effects by location and across locations. Table VI-2 shows
that N is statistically significant in 5 of 7 locations
(though 20 of 22 across cycles) and in the combined
analysis, confirming again the potential of this
technological component for increasing maize yields in the
Cayes Plain. Note too that N had a more important effect on







Table V-6
Narrowing of the Research Focus After Two Cycles of Experiments


CYCLE I (1981)

1. Nitrogen (0, 80-

2. Phosphorus (0, 50)

3. Density (44, 58,000)

4. Weed Control (Manual, Chemical)


CYCLE II (1982)



1. Nitrogen (0, 80)

2. Phosphorus (0, 50)

3. Density (44, 58,000)


5. Variety (80 kg N/ha)
4. Variety (0, 80 kg N/ha)
IM 7928 LM 7928
LM 7827 LM 7827
Toc. 7926/1 PR 7827
S. Lag 7728 a Toc. 7926/1
PR 7427 Across 7728
PR 7928 p S. Lag 7726
Les Anglais PR 7926
SLevy 7835 Les Anglais
PR 7726 PR 7931
Across 7726 Chicken Corn
Chicken Corn o c
Toc 7831/1


STo be conducted.


CYCLE III (1983) 1/





1. Nitrogen (0, 40, 80, 120 kg/ha)

2. Variety

IM 7928, LM 7827, Chicken Corn









Table VI-1. Differential Yield Response (TMl/ha) for Variety and Nitrogen
Levels by Location


(0 kg/ha)
(40 kg/ha)
(80 kg/ha)
(120 kg/ha)
(Chicken Corn)
(La Maquina 7827)
(La Maquina 7928)


Tuffet
1087
2225
2987
3623
2456
2404
2569


Labathe
1967
2295
2988
2888
1992
2641
2968


St. Felix
2290
2276
2300
2636
2362
2276
2523


Mersan
1505
1619
1840
2350
1878
1751
1853


Chantigny
2600
3392
3187
3587
2758
3436
3381


Table VI-2. Summary of Analysis of Variance of Yield for Main
Effects and their Interactions.


Tuffet


LDC
Rep (Loc)
v ..
N
VX N
Loc xV
Loc xN
Loc xVxN
C.V.


Labathe


NS
**
**

NS


St. Felix Mersan


Chantigny


NS
**
**

NS


Beraud Codere Corbined
**


16.3 20.2


No
0
N1
N2
N3

Vl
V2
V3


Beraud
2091
2166
1994
1856
1947
2151
1984


Codere
1148
1795
2515
2652
1852
2124
2107


Average
1832
2253
2547
2799
2172
2401
2493


16.6 17.8


11.2


25.6


27.0









yield than did V, as in the earlier experiments. Variety
was statistically significant in three locations and in the
combined analysis, again providing confirmation of this
research hypothesis.


The yield responses to both N and V were somewhat lower
than in cycles I and II however, and the interaction N X V
was not statistically significant. This may be due to the
very difficult production circumstances in 1983, when a lack
of rainfall at critical periods in the maize cycle lowered
yields in almost all areas. This may have also diminished
the expression of La Maquina 7827's higher genetic
potential, though it is important to note that the
statistical significance of both components across locations
under even these most difficult agronomic circumstances is
an important research finding. Similarly both La Maquinas
outyielded the local Chicken Corn across locations at every
level of N (see Table VI-3). La Maquina 7827 also yielded
more than Chicken Corn in 5 of 7 locations where no nitrogen
had been applied.

Table VI-3
Average Yields by Variety and Levels of N
(V1 Chicken Corn; V2=LM 7827; V3=LM 7928; N0=0
kg N/ha; NI=40; N2=80; N3=120)

Combined

NO N1 N2 N3
V1 1701 2100 2258 2621 2170
V2 1822 2382 2633 2752 2397 (+.227)
V3 1903 2277 2742 3024 2487 (+.090)
Combined N 1809 2253 2544 2799
(+.444) (+.291) (+.255)








Given the earlier experimental results and the survey
data, an economic analysis was performed in order to
identify a more appropriate level of nitrogen fertilization.


Maize prices increased 50% 'from 1982 to 1983, so some
good rates of return were obtained despite the lower
response to N. The analysis demonstrated that N1 or 40
kg/ha increased yields.by .4 TM/ha, providing a marginal
rate of return of 116% for landowners using free market urea
(see Figure 1). This is well above the opportunity costs of
investment capital. The next increment of N or 80 kg/ha
increased yields by another .3 TM/ha, though here the rate
of return fell below the costs of capital (estimated at
60%). The level N3 gave an even lower return, and it is
clear that 40 kg N offers the best level of nitrogen
fertilization. It is also apparent that the rates of return
are unacceptable at all levels of N if the'Department of
Agricultures' price-subsidized 15-15-15 (most available at
that time) is used as the source of N. This is consistent
with the results from Cycles I and II, and serves to
underline to both policy makers and farmers that urea is a
key to increasing maize yields in the Cayes Plain.


There was no statistical significance for the
interaction N X V, and both Chicken Corn and La Maquina 7827
gave good rates of return with 40 kg N/ha (105% and 188%
respectively) though neither provided acceptable returns
with the higher levels N2 and N3. This suggested the
possibility of making a fertilizer recommendation
independent of variety.


La Maquina 7827 also yielded more than the local
Chicken Corn where no nitrogen was applied, and this
suggested it may also be possible to make a variety
recommendation independent of the use of nitrogen
fertilization. This could have immediate relevance to the








FIGURE 1

Net Benefits Curve
Average Returns to Nitrogen Across
Varieties at 4 Levels of Fertilization
(Source of N = Urea at $18.00/100 Ibs)


37%


40 50 60 70 80 90 100 110 120


Variable Costs ($/ha)
Costs of Capital Estimated at 60%


550




500









sharecroppers of the Cayes Plain.


Promising research findings must be confirmed in
farmer-managed verification trials prior to formulating
farmer recommendations however, and these trials completed
the research program for Cycle III. The design of the
verification trials was quite simple. Farmers in the target
recommendation domain were stopped while plowing their.
fields and asked if they wished to try a new maize variety.
Those who did were given La Maquina 7827 seed and were
advised to plant half that field with their own maize and
half with the new variety, giving exactly the same
treatments to both halves of the field. No other management
practices were recommended. These fields were visited some
weeks later and where appropriate (i.e. where management had
been the same across varieties and where no fertilizers had
been applied) 60 kg N/ha. was superimposed across both
varieties but in half of each field, giving four simple
subplots contrasting the two key components, nitrogen
fertilization (0 and 60 kg N/ha) and variety (Chicken Corn
vs. La Maquina 7827). The research team applied the
fertilizer to insure a uniform rate of application across
varieties and across locations, though all other management
was in the hands of the farmers themselves. The use of a
lower level of N reflects the same concerns from Cycle II
for identifying a more economically appropriate dose of
fertilization. All other management was in the hands of the
farmers themselves.


Table VI-4 presents the yields for five locations and
the averages across locations for both varieties, with and
*- without nitrogen fertilization. Note the yield response to
N is again important for both La Maquina 7827 and Chicken
Corn, providing increases between .7 and .8 TM/ha. across
locations. Variety also again appears to be affecting
yields, though the verification results confirm the earlier







Table VI-4
Yields in TM/ha at 15% H20


La Maquina
7827


Chicken
Corn


Canse
N60 1.79


NO 1.60



N 1.03
60


.70


Trouillard LaBorde
2.86 2.16


1.06


.87


Guichard Boudet
1.88 1.79


1.44


1.54


Difference


2.94


1.39


1.93


1.20


1.23


1.32


2.04


.96


Difference
Between
Average Varieties
2.10 + .27


1.30


+ .29


.80


1.83


1.11


( )


( )


Difference


Average
for location


2.06 1.54 1.47


.72


1.58 1.59


1.28








conclusion that this factor is of less importance than is
nitrogen fertilization. Nevertheless the improved variety
both with and without fertilizer outyielded the local maize
in three of five locations.


Note however that yields for both varieties are
substantially lower than those in Cycles I and II, and there
is apparently no important interaction between N and V.
This parallels the results from the level of N X V trials.
The higher maize price meant that the marginal rates of
return to investment capital for landowners with this lower
level of N were still quite good for both varieties however,
despite the lower yields (178% for La Maquina 7827, 150% for
Chicken Corn). The verification trials also confirmed that
open market urea is the only economically appropriate source
of N, and that fertilizer use in maize is not economically
viable for the sharecroppers given present cost-sharing
arrangements between landlords and tenants.


VII THE RESEARCH STRATEGY FOR CYCLE IV


The high degree of consistency of results from site to
site and from year to year suggested a two-pronged approach
for the research program in Cycle IV.


1. On the one hand, recommendations for the two most
promising technological components, nitrogen
fertilization and variety, could now be made to farmers
and policy makers with confidence.


2. On the other hand, as some problems are solved other
production constraints press for solutions, and new
priority research opportunities arei identified.


With respect to formulating farmer recommendations, the
effectiveness of both nitrogen fertilization and variety has









been validated in farmer managed verification trials in the
target recommendation domain. In the case of nitrogen
fertilization, this component had a statistically
significant effect on yield in fully 20 of 22 experiments in
the farmers' fields in three different agricultural years.
Site to site variability in rates of return to investment
capital was important however, and the Cycle III experiments
identified 40 kg N/ha as an excellent low risk/high
profitability option for maize farmers, regardless of
variety. The research has also demonstrated that there must
be two important qualifications to this general farmer
recommendation however;


1. the source of N must be urea, rather than the'MARNDRs'
subsidized blends; as these latter fertilizers
consistently offered rates of return inferior to the
opportunity costs of capital. This conclusion could
have some policy implications as well, as it suggests
that the MARNDR should modify its fertilizer
provisionment policies to include greater supplies of
urea for the maize farmers of the Cayes Plain.


2. The recommendation applies only to fields planted by
landowners, rather than tenant farmers, because present
cost sharing arrangements make fertilizer use
economically unattractive for this latter group.


This conclusion also led to an a posteriori
redefinition of our tentative recommendation domains,
splitting the principal domain in two on the basis of land
tenure for the technological component nitrogen. We may
hypothesize however that these cost sharing arrangements may
eventually change, as more and more landowning farmers
recognize the important economic benefits that nitrogen
fertilization with urea can provide.








As far as the component variety is concerned, the
improved La Maquina 7827 out-yielded the local maize in 16
of 21 locations across three different agricultural years,
for an average yield increase of more than one-half ton per
hectare. La Maquina 7827 also outperformed Chicken Corn
where no nitrogen had been applied (9 of 14 locations), and
this suggests that the interaction N X V is indeed not a
critical issue. The variety may be recommended
independently of the use of nitrogen fertilization, and it
may therefore have immediate relevance to those farmers who
do not intend to fertilize their maize (e.g. the
sharecroppers). Recommendations for variety were also made
at the national level, arguing for an increasing commitment
to seed production and distribution of La Maquina 7827.


The second component in'the research strategy for Cycle
IV is the determination of new research opportunities, and
in this context method of land preparation was identified as
a highly promising area of investigation. Present costs for
animal traction plowing are quite high (approximately $U.S.
60.00/ha.), and the formal survey indicated that only 1 in
10 farmers owns his own plow and bulls. This helped confirm
the hypothesis of the-presence of an important bottleneck in
land preparation at the beginning of each rainy season, with
consequent delayed plantings with potentially reduced
yields. Zero or chemical tillage farming was identified as
a potential alternative, offering substantially reduced
costs when compared with animal traction plowing as well as
other possible benefits (e.g. providing more flexibility in
planting dates, improving soil structure, fertility, and
water holding capacity with mulching). National program
experience with chemical tillage in the Cayes Plain could
also have a valuable "spin-off" effect, providing training
that might serve outside the designated research area to
address soil erosion in the mountains, an important problem
in Haiti.









Four zero-till experiments were recently harvested, and
these included the two recommendations for nitrogen
fertilization and variety. In this way we will be able to
contrast the different costs and yields with zero-till vs.
conventional tillage, and evaluate the yield impact of the
recommended 40 kg N/ha. under these two tillage systems.
The preliminary results from these experiments are most
encouraging, and these will allow us to identify our
research themes with even greater precision in the coming
year.


VIII THE NATIONAL RESPONSE


Finally it is important to note that the Haitian
national program has been enthusiastic about the OFR project
in Les Cayes, and in particular about the research
methodology being used there. They are convinced that
appropriate farmer recommendations can be developed in the
Haitian context by national scientists with appropriate
methodological support, and that such recommendations can be
developed in a reasonably short period of time and with very
low costs. Based on the results from Les Cayes, the MARNDR
has already placed increasing emphasis on the production and
distribution of La Maquina 7827 seed, and an expanded
extension effort linked to the two recommendations is now
under discussion. In addition in keeping with the
Ministry's policy of closely monitoring the progress of the
program in Les Cayes and assessing the effectiveness of the
OFR procedures being used there, the National Director of
Agricultural Research carried out a project evaluation at
the end of 1983. Based in part on fairly rigorous
interviews with over 60 collaborating farmers, the principal
recommendations from the evaluation team to the Secretary of
State for Agriculture and Natural Resources were:


1. This type of program should be carried out in other








areas of the country and with other target crops, and
2. Increasing emphasis should be given to collaborative
CIMMYT/MARNDR training of Haitian researchers in these
particular research procedures, to permit an increasing
national commitment to this kind of research.


CIMMYT hopes to intensify its collaboration with the
Haitian national program in order to help them meet these
important goals, and in particular a more comprehensive
in-country training effort may be initiated in the coming
year (based in part on CIMMYT's recent experiences in
Honduras). This could play an important role in the
institutionalization of OFR in Haiti, and as this paper may
suggest this could have important positive implications for
the Haitian farmer.


Finally, from the perspective of CIMMYT, the Les Cayes
experience, together with similar ones coming from other
countries, 8 reaffirms our conviction that the OFR
procedures involved could play a key liason role between
experimental station research and extension activities. As
the Les Cayes experience suggests, the institutionalization
of these procedures within National Research Programs could
significantly contribute to increasing the capacity of the
country to generate appropriate technology for target groups
of farmers. In this way they might improve the efficiency
of the process of technology generation and transfer. In
addition, part of the results call our attention to a
relatively unexplored potential of OFR, namely that of


S See for example, Juan Carlos Martinez and Jose Roman
Arauz "Developing Appropriate Technologies Through
On-Farm Research: The Lessons from Caisan, Panama."
Agricultural Administration 17 (1984)93-114.









putting the information from surveys and experiments into a
perspective which can confront micropolicy constraints which
may prevent farmers from fully capitalizing on the potential
benefits associated with technological change.










REFERENCES


CIMMYT. "Planification de Technologies Appropiees pour les
Agriculteurs: Concepts et Procedes." CIMMYT, Mexico,
1983.


Magloire, E. and Yates, M. "Recherche Chez Les Paysans",
CRDA/FAMV Seminaire Sur Les Systemes de Production
Agricole en Haiti, Li-mb, Haiti, 23pp. 1984


Martinez, Juan Carlos and Jose Roman Arauz. "Developing
Appropriate Technologies Through On-Farm Research: The
Lessons from Caisan, Panama." Agricultural
Administration 17: 93-114, 1984.


Martinez, Juan Carlos. "La Mise au Point D'Une Technologie
Adaptee aux Contraintes et Atouts de L'Agriculteur:
L'Approche du CIMMYT." Document de Travail 01/84.
CIMMYT, Mexico, 1984.


Perrin, Richard, K and Donald L. Winkelmann et. al.
"Comment Etablir des Conseils aux Agriculteurs a Partir
des Donnees Experimentales." Bulletin d'information
27, CIMMYT, Mexico,, 1979.




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs