Journal of farming systems research-extension

Material Information

Journal of farming systems research-extension
Running title:
Journal for farming systems research-extension
Abbreviated Title:
J. farming syst. res.-ext.
Association of Farming Systems Research-Extension
Place of Publication:
Tucson Ariz. USA
Association of Farming Systems Research-Extension
Publication Date:
Physical Description:
v. : ill. ; 23 cm.


Subjects / Keywords:
Agricultural systems -- Periodicals -- Developing countries ( lcsh )
Agricultural extension work -- Research -- Periodicals ( lcsh )
Sustainable agriculture -- Periodicals -- Developing countries ( lcsh )
serial ( sobekcm )
periodical ( marcgt )


Dates or Sequential Designation:
Vol. 1, no. 1-
General Note:
Title varies slightly.
General Note:
Title from cover.
General Note:
Latest issue consulted: Vol. 1, no. 2, published in 1990.
Electronic resources created as part of a prototype UF Institutional Repository and Faculty Papers project by the University of Florida.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact Digital Services ( with any additional information they can provide.
Resource Identifier:
22044949 ( OCLC )
sn 90001812 ( LCCN )
1051-6786 ( ISSN )

Full Text
Volume 1, Number 2

for Farming Systems
Research- Extension

29 Reerce an xeso.ok eain
Stcphan0 *c~jr n avdK ijni

for Farming Systems

Volume 1, Number 2, 1990

Published by
the Association for Farming Systems Research-Extension

Journal for Farming Systems Research-Extension

Editorial Board
Timothy R. Frankenberger, Editor Timothy J. Finan
Office of Arid Lands Studies Bureau of Applied Research in
The University of Arizona, Tucson Anthropology
The University of Arizona, Tucson
Peter E. Hildebrand Donald E. Voth
Food and Resource Economics Agricultural Experiment Station
Department University of Arkansas, Fayetteville
University of Florida, Gainesville
Harold J. McArthur C. David McNeal, Jr.
Office of International Programs Extension Service, USDA
University of Hawaii, Honolulu
The Journal is sponsored by:
The Farm Foundation
The Ford Foundation, New York
The Ford Foundation, New Delhi
United States Agency for International Development
United States Department of Agriculture
University of Arkansas/International Agricultural Programs Office
University of Florida
The University of Arizona

The Journalfor Farming Systems Research-Extension is published by the Association for
Farming Systems Research-Extension (AFSRE), an international society organized to
promote the development and dissemination of methods and results ofparticipatory on-
farm systems research and extension. The objectives ofsuch research are the development
and adoption through participation by farm household members of improved and
appropriate technologies and management strategies to meet the socioeconomic and
nutritional needs of farm families; to foster the efficient and sustainable use of natural
resources; and to contribute toward meeting global requirements for food, feed, and
The purpose of the Journal is to present multidisciplinary reports of on-farm research-
extension work completed in the field, and discussions on methodology and other issues
of interest to farming systems practitioners, administrators, and trainers. The Journal
serves as a proceedings for the annual international Farming Systems Symposium from
which selected and refereed papers are included. It also welcomes contributed articles
from members of the AFSRE who were unable to attend the symposium. Contributed
articles will be judged by the same review process as invited articles.
Technical Editors: James H. Maish, Emily E. Whitehead, and Daniel Goldstein, Office of Arid
Lands Studies, The University of Arizona
Design and Production: Paul M. Mirocha and Nancy Schmidt, Arid Lands Design, Office of Arid
Lands Studies, The University of Arizona

ISSN: 1051-6786

Journal for Farming Systems Research-Extension
Volume 1, Number 2, 1990


1 The Rice-Wheat Pattern in the Nepal Terai: Issues in the Identification and
Definition of Sustainability Problems
L. Harrington, P. Hobbs, T. Pokhrel, B. Sharma, S. Fujisaka, and C. Lightfoot

29 Relations between Agricultural Researchers and Extension Workers: The
Survey Evidence
Stephan Seegers and David Kaimowitz

47 Agricultural Innovation and Technology Testing by Gambian Farmers: Hope
for Institutionalizing On-Farm Research in Small-Country Research Systems?
Bradford Mills and Elon Gilbert

67 Development and Testing of Integrative Methods to Assess Relationships
between Garden Production and Nutrient Consumption by Low-Income
IngolfGruen, Michel Beck, John S. Caldwell, and Marilyn S. Prehm

81 Farming Systems and Adoption of New Agricultural Technologies: An Eco-
nomic Evaluation of New Sorghum Cultivars in Southern Honduras
Miguel A. L6pez-Pereira, Timothy G. Baker, John H. Sanders, and Dan H.

105 Zoning Survey: Improving the Efficiency of Farming Systems Research-
Extension Diagnostic and Field Activities
Thomas E. Gillard-Byers and Malcolm J. Blackie

123 Agricultural Development in India: An Assessment of the FSRE Approach
Aruna Bagchee

137 Modeling an Agrarian System on a Local Scale as a Tool for Farmer Participa-
tion in Rural Development: Examples from South America
S. Lardon and C. Albaladejo

The Rice-Wheat Pattern in the Nepal Terai:
Issues in the Identification and Definition
of Sustainability Problems'

L. Harrington, P. Hobbs, T. Pokhrel, B. Sharma, S. Fujisaka,
and C. Lightfoot2


The concept of "sustainability" is taking an increasingly central place in the
activities-and reporting-of Farming Systems Research-Extension (FSRE)
practitioners. In the recent literature, examples ofsustainability problems have
been reported with respect to farmers' practices, the agricultural resource
base, policies, and institutional innovations.
Many of these examples are fairly dramatic. For example, Fujisaka and
Garrity (1988) describe the use ofhedgerow technology to reduce soil erosion
and improve cropping-pattern productivity (over the long term) in one site in
northern Mindanao. In this example, damage from erosion was readily
observable, and farmers were vocal in their concern about the problem.
Similarly, Roche (1988) reports the following description of sustainability
problems in Java. "Many of Java's steep upland areas have been classified as
land which has become so degraded that it is, or soon will be, unable to sustain
even subsistence agriculture."

'Paperpresented at the Ninth Annual Farming Systems Research-Extension Symposium,
University ofArkansas, Fayetteville, October 9-11, 1989. It is based on an earlier paper,
The Rice-Wheat Cropping Pattern in the Nepal Terai: Farmers' Practices and Problems
and Needs for Future Research, Fujisaka and Harrington, eds. The opinions expressed
are not necessarily those of the International Maize and Wheat Improvement Center
(CIMMYT), the National Agricultural Research Services Center (NARSC), the Na-
tional Wheat Development Program (NWDP), the International Rice Research Insti-
tute (IRRI), or the International Center for Living Aquatic Resources (ICLARM).
2Harrington is an agricultural economist, CIMMYT Economics Program, Bangkok,
Thailand; Hobbs, an agronomist, CIMMYT Wheat Program, Kathmandu, Nepal;
Pokhrel, a plant breeder, NWDP, Bhairahawa, Nepal; Sharma, an agricultural econo-
mist, NARSC Socioeconomics Research Division, Khumultar, Nepal; Fujisaka, an
anthropologist, IRRI, Los Baios, Philippines; and Lightfoot, a farming systems
agronomist, ICLARM, Manila, Philippines.


When problems related to sustainability are fairly obvious, with dramatic
and readily observable effects, questions of"problem definition" are likely to
seem trivial. For example, when researchers are faced with increasingly severe
and widespread resource degradation, they are unlikely to agonize over
whether a sustainability problem really exists.
Not all problems ofsustainability need be so dramatic, however. This paper
is particularly concerned with sustainability problems that may not be im-
mediately obvious, but that nonetheless can have important effects over the
long term. Less obvious problems ofsustainability can occur when changes in
other factors mask or obscure a long-term decline in productivity. For exam-
ple, when farmers use increasing levels of inputs but yields remain stagnant,
the underlying long-term reduction in productivity may not be immediately
apparent. Nonetheless, this long-term deterioration in productivity can
seriously affect farmers' incomes.
The term "sustainability" has been defined in numerous ways. Without
attempting to comment on the various alternatives, we will restrict our
discussion to "sustainability problems" defined in terms of the longer-term
productivity of farmers' resources, including the resource base.
In this paper, we argue that researchers have tended to overlook hidden
problems of sustainability, and that these problems require relatively high
investments in diagnostic activities. The objectives of this paper are to
(1) discuss problem definition in the context ofsustainability issues; (2) de-
scribe ongoing research in the Nepal Terai which includes an attempt to define
sustainability problems; and (3) discuss further research that may be required
to properly define sustainability problems for the Nepal study area.


Problem Definition
There are numerous approaches to FSRE, each with a somewhat different
vocabulary and suggested sequence of research steps. Many of the differences
between these approaches can be explained in terms of the problem areas, the
agroclimatic environment, and the mixofdisciplines present during a formative
period in which the particular FSRE approach was first developed. Despite
differences, there is broad common ground between approaches, especially
when it comes to the importance of diagnosis, or problem definition (Har-
rington, et al., 1989).
In principle, the definition of sustainability problems (defined here as

Journal for Farming Systems Research-Extension


longer-term productivity problems) differs little from the definition of near-
term productivity problems. In both cases, researchers need to ascertain:

1. Whether the problem really exists. Many diagnostic activities produce
hypotheses about possible problems that prove to be unfounded when further
evidence is accumulated.
2. The likely benefits that could be earned by solving the problem. This
requires some measurement of the productivity loss, broadly defined, asso-
ciated with the problem, and the incidence and frequency of the problem (e.g.,
percent of farmers affected, percent of the farm affected, frequency of loss).
3. The causes of the problem which, in turn, tend to suggest alternative
possible solutions, each with an associated level of research expenditure (Trip
and Woolley, 1989).

Special Considerations in Defining Long-Term Productivity Problems
In practice, the definition of sustainability problems introduces several
complications into the problem-definition process.
Confounding factors: In attempting to ascertain whether a sustainability
problem exists, researchers may have to sort through several confounding
factors. These are factors that tend to mask or obscure a long-term negative
trend in the productivity of farmers' resources. Confounding factors can
include the following:
1. Increased application levels of purchased inputs, for example, fertilizer.
(Gradually increasing levels of fertilizer can maintain yields at a roughly
stagnant level, obscuring the fact that yields would decline at constant input
2. Adoption of high-yielding varieties. (A change to a more productive
variety can mask yield reductions that would otherwise be observed and, in
addition, can actually hasten the longer-term decline in productivity if plant
nutrients are extracted more rapidly, but not replaced.)
3. Changes at the margin in land quality. (Average regional yields of a crop
or a cropping pattern will appear to increase if better-quality lands are
gradually substituted for lower-quality lands. This will be true even if the
productivity of any particular field-at constant levels of purchased inputs-
is gradually declining.)
4. Expansion in irrigated area. (This is a specific example of changes in land
quality, noted above.)

Time paths and discounting: In attempting to ascertain the likely benefits
Vol. 1, No. 2, 1990


of solving a particular long-term productivity problem, several more compli-
cations arise:

1. Assessments of productivity loss are no longer single-cycle "snapshot"
estimates. Rather, the past over time of reductions in productivity (at constant
input levels) needs to be estimated and compared to alternative paths.
2. Estimates of the incidence (number of farmers affected, area per farm,
etc.) and frequency of the problem also need to be expressed in terms ofa path
over time.
3. Appropriate discounting measures need to be included, to estimate the
present value of the productivity loss. Again, this present value can be
compared to the present value of losses associated with alternative strategies.

Identifying causes and proposing solutions: Even when researchers are
convinced that productivity, at constant input levels, is declining over time,
considerable further work may be needed to pinpoint the reasons behind this
decline. Hypotheses may include (but need not be restricted to) increasing
scarcity of macronutrients or micronutrients; buildup of pests or diseases;
buildup of problem weeds; deterioration in soil physical or chemical structure
(including gradual salinization); reduced soil moisture-holding capacity; soil
loss through erosion, etc.
It is not always easy to pinpoint which of these hypotheses is most relevant
to field conditions in a given study area. The close technical identification of
specific reasons for a long-term decline in productivity is likely to require a
combination of systems, commodity and disciplinary expertise. "Diagnostic"
activities for defining sustainability problems may be even more challenging,
complex, time-consuming, and expensive than the ones commonly used to
define near-term problems.

Before continuing with the discussion of sustainability issues (i.e., the
identification and definition of hidden sustainability problems), we will
describe in some detail a joint MARSC-CIMMYT-IRRI project on the rice-

SThis and following sections draw heavily on the NARSC-CIMMYT-IRRI paper The Rice-
Wheat Cropping Pattern in the Nepal Terai: Farmers' Practices and Problems and Needs
for Future Research.

Journal for Farming Systems Research-Extension


wheat pattern in the Nepal Terai. Farm-survey activities, farmers' practices,
system interactions, and near-term problems (along with their causes and
some possible solutions) will be discussed. We realize that this interrupts to a
certain extent the discussion on sustainability, but feel that many readers will
welcome adequate background material on the study area. Readers preferring
to maintain a relatively narrow focus on "sustainability" are invited to skim
over the next few sections, up to the section titled The Rice-Wheat Pattern in
the Nepal Terai: Problems and Causes-The Longer Term.


The rice-wheat cropping pattern is extremely important in South Asia. Wheat
is grown after rice on approximately 8.7 million ha in the region, accounting for
about 25 percent of the region's wheat production. Wheat yields are low (less
than 2 t/ha) even where irrigation is available (Hobbs, Mann, and Butler,
The International Maize and Wheat Improvement Center (CIMMYT) and
the International Rice Research Institute (IRRI) are developing, in partner-
ship with interested National Agricultural Research Systems (NARS), a
collaborative research program on the rice-wheat pattern in South Asia. The
collaborative research program has four main objectives:

1. Conduct adaptive and applied research to define and solve major
problems associated with the rice-wheat pattern in selected, defined study
area. Problems may include near-term productivity issues and longer-term
sustainability issues.
2. Conduct a comparative analysis (over countries) of problems affecting
the rice-wheat pattern in South Asia and identify possible solutions for these
problems, which are effective under a wide range of local circumstances.
3. Improve the understanding ofCIMMYT, IRRI, and participating NARS
on how to address problems of sustainability.
4. Strengthen IARC-NARS linkages.

As an initial step in this collaborative research program, scientists from
CIMMYT and IRRI joined with researchers from the National Agricultural
Research Services Center, Nepal, to study the rice-wheat pattern in Nepal's
Terai. To begin, a diagnostic, exploratory survey was conducted in February
1989. The survey focused on farming systems in Rupandehi District, where

Vol. 1, No. 2, 1990


the rice-wheat pattern is central to farmers' livelihoods.

Survey Objectives and Procedures
The diagnostic survey had three major objectives:

1. Understand local farming systems: the rice-wheat pattern, interactions
between rice and wheat, and interactions between the rice-wheat subsystem
and other subsystems.
2. Define near-term and longer-term problems, and understand their
3. Identify further research needs: to improve the definition of poorly
defined problems, to improve researchers' understanding of causes, and to
identify possible solutions to major problems.

Survey participants were senior researchers from the fields of agronomy,
anthropology, economics, extension, pathology, and plant breeding. Morn-
ing and early afternoon field interviews (conducted independently by each of
three subgroups) were followed by structured discussions, attended by all
participants. Semistructured guidelines rather than formal questionnaires
were used to guide discussions with respondents. Using a sequential approach,
these guidelines were redefined daily, after discussion of new information
obtained and data gaps still existing. Respondents were selected from all parts
of Rupandehi District and included small and large farmers, extension
workers, merchants, and government officials.

Rupandehi District
This district is part ofNepal's Terai and is located 100-200 m above sea level
(Figure 1). The Terai, a part of the Gangetic Plain, represents about 14 per-
cent of Nepal's total land area and 42 percent of the country's cultivated land.
Rupandehi District has a subtropical climate highly influenced by the
southwest monsoon. On average, total annual rainfall reaches around 1,600 mm
and increases from south to north. More than 85 percent of the rain comes in
the period from mid-June to the end ofSeptember. November and December
are the driest months, and light precipitation may be expected in January and
February. Mean temperatures are lowest (15C) in January and highest
(300C) in May. There are occasional strong, hot, dry westerly windstorms in
April and May (APROSC, 1986).
The population of the district was 380,000 in 1981, with a growth rate of
2.23 percent per year. There are about 83,000 ha of cultivable land, of which

Journal for Farming Systems Research-Extension


Figure 1. Wheat Planting Date: Problems and Causes

Vol. 1, No. 2,1990


some 28 percent receives some irrigation.

Land and Soil Types, and Land Use
Farmers were found to use land-type classes that corresponded closely to
technical classifications (APROSC, 1986). Farmers' land-type classes, like the
technical classifications, are based on interrelated variation in soil, topogra-
phy, and hydrology.
Lower terraces (locally khala) are characterized by heavier soils and poor
drainage, and are commonly used for the production oflong-duration (usually
photoperiod-sensitive), traditional rice cultivars. These lands are normally left
fallow after rice. Middle terraces (Osahaniva) are characterized by lighter soils
and fewer drainage problems. Common cropping patterns on middle terraces
include medium-duration rice varieties followed by wheat, wheat mixed with
mustard, or other winter crops. Upper terraces (danda) are well-drained and
drought-prone, and are planted to shorter-duration rice followed by wheat,
various wheat-based crop mixtures, or winter vegetables.

Rice-Crop Management for Middle and Upper Terraces
On upper and middle terraces, rice is grown as a first crop, before the second
crop of wheat. This section describes farmers' management practices for this
first rice crop.
Seedbeds and direct seeding: Farmers prepare and sow seedbeds at the
end of June. Transplanting in irrigated areas is usually finished within 30 to
45 days after seeding. Farmers in rainfed areas do not transplant until after the
rains have started. As a rule, farmers avoid transplanting seedlings more than
60 days old. (Late transplanting, given the varieties commonly used by
farmers, was reported to cause reductions in rice yields.) When rains are
delayed, farmers may use direct seeding. This is more common in the upper
Pests and diseases: Farmers identified ricebug, armyworm, and stemborer
as pests of increasing severity, and blight and blast as important diseases. Rats
are a serious problem in the field and in storage (particularly in stacked rice,
before threshing). Rats are also responsible for damage to bunds and irrigation
infrastructure. Farmers reported using few control measures for insects,
diseases, or rats.
Soil and root samples collected by the team's nematologist contained from
100 to 1,000 Hirschmanniella (cf. oryzae), the rice-root nematode, individ-
uals per liter of soil in all samples. This parasitic nematode appears to be present

Journal for Farming Systems Research-Extension


throughout the district. Farmers take no control measures.
Soil-fertility management: Farmyard manure (FYM) is used in rice
seedbeds and is occasionally applied in small quantities to rice fields prior to
land preparation. Some farmers reported declining rice productivity, possibly
due to low and declining FYM use. Declining yields were also said to be
characteristic of fields with a longer history of intensified cropping.
Farmers using inorganic fertilizers reported applying 50-100 kg/ha of
compound fertilizer (usually 20-20-0) at planting and 30-50 kg/ha urea as a
topdress. This is equivalent to around 25-45 kg/ha of nitrogen and 10-
20 kg/ha of phosphate (below recommended levels). Direct-seeded rice was
said to usually receive only a urea topdress. Several farmers identified zinc
deficiency as a problem, with Saryu-49 said to be particularly sensitive. Good
sources of zinc fertilizer are not available.
Harvest and postharvest: Rice is hand harvested and is usually stacked
(i.e., threshing is postponed) in order to free farmers' labor for wheat land
preparation. Despite this practice, land preparation for wheat can still be
delayed by a late rice harvest (or a need to further field dry rice that is too wet
to stack). Farmers report having few rice-seed storage problems, as seed
storage occurs during the cool, dry season.

Wheat-Crop Management for Middle and Upper Terraces
Tillage: Farmers usually plow four times and plank twice, with the first
plowing requiring relatively more time. More plowings are usually used for
heavier soils. In rain-fed areas where moisture is limited, farmers may reduce
tillage operations or wait for the rains. Turnaround time from rice harvest to
wheat planting requires around 15 to 35 days (mostly for plowing and
planking), given the usual conditions of unfavorable soil moisture and soil
physical condition. Farmers' practices, as described above, may result in over-
tillage with detrimental effects on soil structure.
Timing and method of planting, seed management, plant stand: Most
farmers broadcast seed into plowed soil, then cover the seed by plowing once
again, and planking. Few farmers use line sowing and no farmers were found
using seed drills. Seed rates vary from 120 to 180 kg/ha, with an average of
around 150 kg/ha (compared to the recommended seed rate of 120 kg/ha).
Farmers' stored seed often suffers from the effects of pests and excess
moisture. Farmers are reluctant to buy seed from the government because of
the high cost (around Rp. 8 [US$.46] per kg) and relatively poor seed quality.
If farmers' own stored seed is badly damaged, replacement seed (usually of

Vol. 1, No. 2, 1990


Indian origin, and composed of a mixture of varieties) is purchased from the
Visual observation of numerous fields suggested that only 15 to 20 percent
of fields were planted during the optimum period of mid-to-late November.
Around two-thirds of the fields were planted somewhat late (during the first
two weeks of December) and the remaining 20 percent after mid-December.
Plant stands were observed to be poor (fewer than 200 plants per m2) in about
20 to 25 percent of the fields; fair in 60 to 65 percent of the fields; and good
(more than 300 plants per m2) in the remaining 10 percent of the field.
Variety: Major wheat varieties used by farmers are RR21 and UP262.
Several newly released varieties are gaining popularity (e.g., Siddartha,
Pests, diseases, and weeds: Insect pests are a major problem for stored
wheat seed, but not for the crop in the field. Rats are a field and storage
problem, and farmers have few rat control measures. Helminthosporium blight
often causes significant yield losses. In addition, experimental evidence
suggests that soil fungi and/or nematodes may be causing yield losses.
Although the rice root nematode Hirschmanniella (cf. oryzae) was present in
all soil samples taken, its effect on wheat yields is not known.
Some weeds (especially broadleaf leguminous types) were observed in
farmers' wheat fields, but the effect of these weeds on wheat yields is unknown.
Farmers cut and carry weeds for fodder, as needed. Weeds remaining in the
field at harvest are cut and mixed with the straw for fodder. Some weeds (e.g.,
Phalaris minor and Circium arvense) seem to be increasing and may become
problems as land use is intensified.
Water management: Wheat is grown as an irrigated, partially irrigated, and
rain-fed crop, with the proportion of irrigated wheat increasing over time.
Partially irrigated wheat usually receives only one irrigation. Fully irrigated
wheat is irrigated two to three times, with the first irrigation within a month
of emergence, and the second at flowering.
In the middle terraces, especially in fields with heavier soils, water may stand
in the field (after an irrigation) for more than a day. Waterlogged patches also
occur in poorly leveled fields. The farmers' practice of transferring water from
one field to the next (a practice borrowed from rice cultivation) can also
contribute to waterlogging.
Soil-fertility management: Some (but not all) farmers reported using
inorganic fertilizers on wheat, usually a compound fertilizer (50-100 kg/ha
of 20-20-0) at planting, and urea (30-75 kg/ha) as a topdress at first

Journal for Farming Systems Research-Extension


irrigation. Total application of nutrients, then, is on the order of 25 to 50 kg/
ha of nitrogen and 10 to 20 kg/ha ofphosphate (below recommended levels).
Higher doses are used in irrigated areas. Few farmers apply potash.
Farmers reported using most of their FYM as fuel. FYM remaining after
fuel needs are met is normally reserved for rice nurseries and vegetable fields.
Usually, wheat fields are not regularly fertilized with FYM.
Harvest and postharvest: Wheat is hand harvested. Many farmers reported
that the onset of hot, dry winds from the west tended to curtail the crop
season, often affecting grain filling. Other farmers reported advancing the
harvest (i.e., harvesting before the crop is well-dried in the field) to avoid pre-
monsoon storm damage. These two weather-related problems seem incon-
sistent and are not yet well understood. Threshing is largely by bullock
trampling, although the use of small power-threshers is increasing. Storage
losses due to monsoon weather and insect pests were not measured, but are
probably significant.
Mixed cropping: It was observed that farmers commonly mix mustard
(Brassicaspp.) with wheat. Crop mixtures are more widespread in rain-fed and
partially irrigated areas. Some farmers reported that they would shift to pure
wheat cropping if irrigation were assured. Farmers normally do not plant
mustard alone, apparently because of problems with aphids.

System Interactions: Interactions between Rice and Wheat
Rice-harvest date and wheat-planting date: A major source of interac-
tions between rice and wheat lies in the competition between these two crops
for the farmers' land and labor resources during the rice-harvest/wheat-
sowing period.
Experimental evidence suggests that mid-to-late November is optimum for
wheat planting, with later dates resulting in reduced yields. Farmers report
aiming to prepare and sow wheat fields as soon as possible after the rice harvest
in October or November. Delays in rice harvesting can delay wheat planting.
At first, survey participants hypothesized that delays in rice harvesting might
be due to late rice transplanting, in turn due to late nursery establishment and/
or late arrival of the rains. Farmers, however, reported sowing rice seedbeds
in June (with irrigation if necessary). They also reported avoiding the
transplanting of older seedlings (over 60 days old). When the rains are late,
farmers direct seed a considerable proportion of upper and middle terrace rice
areas. Changes in farmers' rice planting practices, then, are unlikely to
contribute to more timely rice harvest and wheat planting.

Vol. 1, No. 2, 1990


Earlier maturing rice varieties can also lead to an earlier rice harvest, thus
facilitating early wheat sowing. However, farmers on middle and upper
terraces are already using shorter maturity varieties (compared to farmers on
lower terraces). High yield is associated with longer maturity, so these farmers
are already sacrificing a certain amount of rice yield.
Farmers are busiest (labor is most scarce) in November and December
when rice is harvested and wheat fields are prepared and sown. Farmers save
some time during this period by delaying rice threshing until after the wheat
is sown: they stack the rice for later threshing. Rice storage losses to rats are
a consequence (farmers estimated an 8 to 15 percent loss).
The farmers' practices described above (timely rice transplanting; direct
seeding office; use of shorter-duration varieties; delayed rice threshing) all aim
to reduce the competition for land and labor between rice and wheat.
Effect of paddy soils on wheat: The subsurface pan formed by the
puddling of soils for rice cultivation, combined with farmers' wheat land-
preparation methods (intensive, but shallow tillage), seem to reduce wheat
The subsurface pan apparently contributes to two separate problems
related to moisture: it restricts water percolation (and therefore contributes to
waterlogging after irrigation); and it reduces soil moisture-holding capacity
(and therefore contributes to late-season moisture stress). The subsurface pan
also restricts root growth to a narrow soil layer, contributing to the depletion
ofplant nutrients in that layer. The practice ofpuddling destroys soil structure,
which is difficult to reestablish. Effects on wheat of soil chemical and physical
changes due to alternating flooded and dry conditions are not yet well
The problems noted above are specific to soils found after the production
of puddled rice. Dry, direct-seeded rice does not require puddling and seems
to cause fewer problems for subsequent upland crops such as wheat.
Food security for resource-poor farmers: Wheat production appears to
play an important food security role for low-income farm households. Rice
from middle and upper terraces becomes available in October, and traditional
rice from lower terraces in December. Wheat becomes available in March,
when rice begins to get scarce. Although rice is the main staple, wheat is widely
consumed during the months immediately after its harvest.

Other System Interactions
There are a number of complex interactions between rice and wheat, on the

Journal for Farming Systems Research-Extension


one hand, and fuel, fodder, and farmyard manure on the other. Large
ruminants-bullocks, cattle, and carabao-rely on rice and wheat straw as
major sources of fodder. These livestock provide FYM for fertilizer and fuel.
Herd size appears to be limited by fodder availability. Fuel demand is
increasing and, given depletion of fuel wood resources, more FYM is being
used for fuel and less for fertilizer.
Fodder: Rice straw is the major fodder source, and taller, long-duration
rice cultivars (grown on the lower khala terraces) provide the greatest
proportion of straw. Secondary fodder sources include wheat straw, grazing,
and cut grasses and weeds. Seasonally, lower terraces left fallow provide
pasturage for grazing after the rice harvest in November but are depleted by
about March. Wheat straw (usually mixed with rice straw) is available between
March and July. Some farmers report having enough rice straw to last all year,
though many do not. Supplies of rice straw become available in October and
begin to run out by February or March. Farmers agree that fodder is
particularly scarce from July through September. Many farmers feel that
fodder is increasingly scarce all year and, as a consequence, herd sizes are
Fuel: Given the lack of accessible forest areas in this district, firewood has
lost much of its importance as a source of fuel. Dried dung cakes now provide
most local fuel needs. These dung cakes may account for up to 75 percent of
the FYM produced by a farm household's animal herd.
FYM as fertilizer: Increasingly, FYM is used primarily as fuel, but some
is still available for use as fertilizer. Farmers report the following priorities for
FYM as fertilizer: rice seedbeds; cash crops on lighter danda soils (especially
on fields close to the farm house); fields where declining productivity has been
noted; and other rice or wheat fields.


A major objective of the diagnostic survey was to develop hypotheses for
problems affecting the rice-wheat pattern. A "problem" in this context is
defined to include the following: (1) factors that directly reduce yields;
(2) inefficient use of inputs, regardless of the effect on yields; (3) inefficient
cropping patterns or enterprise selection; (4) factors affecting the sustainabil-
ity of rice and wheat productivity.
The first three classes of "problems" are near-term in nature, and can be

Vol. 1, No. 2, 1990


assessed within the time frame of a crop cycle (a few months) or a cropping
pattern (one year). These near-term problems and their corresponding causes
are briefly discussed in this section and are listed in Table 1. (For a more
thorough discussion of these near-term problems, including causes and
suggested possible solutions, see Fujisaka and Harrington, eds., 1989).
The last class of problem (factors affecting sustainability) is long-term in
nature and will be discussed separately.

Problem 1: Late Planting Reduces Wheat Yields
Late wheat planting appears to be a problem in all land-soil types in which
wheat is grown, and especially in the middle terraces. As noted earlier, visual
observation of numerous fields suggested that only 15 to 20 percent of fields
were planted during the optimum period of mid-to-late November.

Problem 2: Early Season Waterlogging Reduces Wheat Yields
This problem, which is most important in irrigated middle terraces with
heavier soils, appears to have two interrelated causes: the subsurface pan left
in the soil by puddled rice culture (and related problems with soil structure),

Table 1. Preliminary List of Problems: Rice-Wheat Pattern, Rupandehi District

Near-Term Problems:

1. Late planting
2. Early season waterlogging
3. Inadequate plant stand
4. Late season moisture stress
5. Nutrient deficiencies (especially N and P)
6. Farmers' wheat varieties are less productive than alternatives

7. Pests (stemborer, planthoppers) and diseases (blast)
8. Mid-season moisture stress
9. Nutrient deficiencies (especially N, P, and Zn)
10. Weed competition (direct-seeded rice)

Longer-Term Problems:

1. Nutrient deficiencies will increasingly limit the yields of both wheat and rice
2. Pests and diseases will increasingly limit the yields of both rice and wheat

Journal for Farming Systems Research-Extension


and farmers' irrigation practices (and related problems with irrigation and
drainage infrastructure and water control) (Figure 2).
In most irrigation systems, farmers are compelled (by poor water-control
structures) to water their wheat crop as if it were rice-by moving water from
field to field. When soils are heavy and when a plow pan is left over from the
previous rice crop, the normal result is water standing in the field, often for
more than 24 hours. This can directly reduce wheat yields, as well as affecting
plant stands.

Figure 2. Waterlogging of Wheat: Problems and Causes

Vol. 1, No. 2, 1990


Problem 3: Inadequate Plant Stands Reduce Wheat Yields
As noted earlier, researchers inspected numerous wheat fields during the
diagnostic survey and found that plant stands were usually poor to fair.
Around a fourth of the fields were observed to have poor stands (less than 200
plants/m2), with over half having only fair stands (200 to 300 plants/m2). The
stand problem was especially acute on rain-fed middle terraces with heavier
soils. There appear to be a number of causes for poor stands, including poor
tilth (combined with broadcast seeding), poor seed quality, and waterlogging
(Figure 3).

Figure 3. Wheat Plant Stand: Problems and Causes

Journal for Farming Systems Research-Extension


Problem 4: Late-Season Moisture Stress Reduces Wheat Yields
This problem occurs on all land-soil types in which wheat is produced. The
average productivity loss and the probability of occurrence, however, are not
well known at this time.
Four major causes of late-season moisture stress were tentatively identi-
fied: (1) late planting, (2) hot, dry winds in February or March that curtail
grain filling, (3) avoidance of late-season irrigation (this tends to exacerbate
lodging problems associated with the strong March winds), and (4) reduced
soil moisture-holding capacity due to the subsurface pan.

Problem 5: Nutrient Deficiencies Restrict Wheat Yields
Nutrient deficiencies (especially of nitrogen and phosphorous) are suspect-
ed to reduce the yields of both rice and wheat. To avoid repetition, the
discussion on the near-term problem of nutrient deficiency is combined with
the section on the longer-term problem of gradually declining soil fertility.

Problem 6: Farmers' Wheat Varieties Are Less Productive than
Alternative Varieties
Most farmers currently grow one of two varieties: RR21 or UP262. Several
newly released varieties (Siddartha, Vinayak) are only slowly beginning to be
used by farmers. There seem to be two interrelated causes for the slow
adoption of newly released varieties: (1) it is difficult for farmers to obtain seed
of new varieties, and (2) many farmers report not being well acquainted with
the new varieties (understandably, given the difficulty they have in getting

Problems 7 through 10: Problems Associated with Rice (pests and
diseases, mid-season moisture stress, nutrient deficiencies, weed
competition in direct-seeded rice)
Less diagnosis was conducted (and fewer problems identified) for the rice
crop within the rice-wheat pattern. This was because the diagnostic survey
being reported was conducted during the wheat season. (Another survey was
conducted during the rice crop season in September 1989, but the results of
this survey are not yet available).
In addition, two problems associated with rice (pests and diseases, nutrient
deficiencies) have both a near-term and a longer-term dimension. To avoid
repetition, the discussions of near- and longer-term issues are combined in the
section on long-term problems.

Vol. 1, No. 2, 1990


Near-Term Problems: A Summary
It should be clear from the preceding sections that there are a number of
serious and complex near-term productivity problems affecting wheat in the
study area that, moreover, interact with each other. The diagnostic survey
made considerable progress in defining these problems. Still, the problem-
definition process is not yet finished.
Productivity loss: Survey results, combined with other sources of data
(experimental results, the body of agronomic knowledge available from other
studyareas) enabled researchers to make some judgments about the productivity
loss associated with each problem (Table 2). These estimates are still exceed-
ingly imprecise, however. Further work (involving both surveys and on-farm
experiments) will be needed to improve estimates of productivity loss (as well
as to assess solutions).
Incidence and frequency: Survey participants gained some feeling for the
geographical incidence of many of the problems (Table 2). For example,
waterlogging was found to be concentrated on irrigated middle terraces, with
heavier soils. Nonetheless, researchers still have only a very imprecise un-
derstanding of the geographical distribution of each problem, and the
proportion of farmers in the study area (and the area per farm) affected.
Closely focused, formal surveys using random sampling will be needed to
quantify these variables.

Table 2. Near-Term Problems: A Summary (Wheat Only)

Problem Yield loss Location Percentage
of farmers

Late planting Moderate Most land types, especially 80 (?)
middle terraces
Waterlogging Serious Irrigated middle terraces, 35 (?)
with heavier soils
Inadequate Serious Middle terraces, with 50 (?)
plant stand heavier soils
Late-season Serious Upper terraces, with 50 (?)
moisture lighter soils
Nutrient Moderate Most land types, especially 80 (?)
deficiencies upper terraces
Variety Moderate Most land types 80 (?)

Journal for Farming Systems Research-Extension



The diagnostic survey aimed to help identify and define problems relating to
the longer-term sustainability of rice and wheat productivity. Compared to
the near-term issues described in the previous section, less progress was made
in defining these longer-term issues. Considerably more diagnostic work is
needed to estimate productivity trends, the relative importance of the differ-
ent long-term problems, and the frequency, incidence, and causes of each one.

Is There a Sustainability Problem?
Some doubt remains as to whether a sustainability problem even exists with
regard to the rice-wheat pattern in the Nepal Terai. What evidence is there to
suggest that rice-wheat farmers face a prospect of slowly declining produc-
Farmer opinion: Farmer opinion on the subject of productivity trends is
somewhat divided. These differences of opinion, however, tend to fall out as
1. Type A farmers: These farmers claim that yields of both rice and wheat
are increasing over time, because of expanded irrigation area, increased use of
high-yielding rice and wheat varieties, and increased use of chemical fertilizer
(on irrigated land). However, Type A farmers tend to have only recently
benefited from the installation of irrigation and tend to have had only brief
experience with the rice-wheat pattern (often less than five years).
2. Type B farmers: These are farmers growing the rice-wheat pattern in
rain-fed upland (ex-forest) areas, usually on light-textured danda soils. They
tend to use few if any inputs, including FYM. Farmers here engage in classical
"soil mining" and recognize that productivity is declining.
3. Type C farmers: These are farmers with longer experience (more than
five years) with the rice-wheat pattern in irrigated areas. These farmers claim
that, with the introduction of the intensified (rice-wheat) pattern, yields of
both rice and wheat initially increased, but then began to decline. Some
farmers have observed a gradual decline in productivity despite the application
of (what they consider) reasonable levels of inputs.
Further evidence from the diagnostic survey on questions of sustainability
is provided in the section titled Long-term Problem 1.
Time series data: Time series data on average regional yields for rice and
wheat are considered to be unreliable. In any event, these data are more likely

Vol. 1, No. 2, 1990


to reflect the confounding effect of land-quality changes (e.g., the substitu-
tion of irrigated for nonirrigated land), or increased levels of fertilizer use,
than long-term declines in productivity at constant input levels. Time series
data on the adoption by farmers of the rice-wheat pattern are available, but are
not reported here. Briefly, these data indicate that rice-wheat is a relatively new
pattern, introduced over the last 10 to 20 years, and has tended to replace rice-
fallow and rice-oilseed.
Experimental data: There are some experimental data that suggest that
the productivity of the rice-wheat pattern is likely to decline over time, given
farmers' current management practices. Here are a few examples.
A long-term fertilizer trial was conducted for seven years under the auspices
of the National Wheat Development Program, on their experiment station in
the Rupandehi District study area. This trial was composed of nine treatments
(involving different combinations ofN, P, K, FYM, and stubble management)
(Table 3). Each treatment was superimposed on the same plots for three crops
per year. (Researchers used a rice-rice-wheat pattern for this experiment, more
intensive than the farmers' practice, which may tend to exaggerate the
observed decline in productivity. Similarly, the trial was conducted on the
khala land type usually reserved for lowland rice, thus raising questions about
the representativeness of the trial.)
The data from this trial indicate that wheat yields have been fairly stable for
all treatments over the seven-year period, but that rice yields-and therefore,

Table 3. Long-Term Fertilizer Trial-Treatments
(Bhairahawa, Rupandehi District, Nepal)

Each Rice Crop (kg/ha) Wheat Crop (kg/ha)
Treatment N P K N P K

1 0 0 0 0 0 0
2 100 0 0 100 0 0
3 100 30 0 100 30 0
4 100 0 30 100 0 30
5 100 30 30 100 30 30
6 100 0 0 100 40 30
7 50 0 0 50 0 0
+ incorporate stubble + incorporate stubble
8 50 20 0 50 20 0
+ incorporate stubble + incorporate stubble
9 10 t/ha FYM 10 t/ha FYM

Journal for Farming Systems Research-Extension


total annual grain yields-have declined drastically for most treatments
(Regmi, 1986). The proper presentation of this data set would require a
separate paper. However, for present purposes, a simple comparison between
yields for the early years (the average of years 1 and 2), versus the later years
(the average of years 6 and 7) is sufficient to illustrate the point. This
comparison is made for the first rice crop and the wheat crop (Table 4). While
these results are presented for illustration only, they support the hypothesis
that long-term declines in productivity are not unlikely.
There are other sources of evidence. Hobbs (1987) reports that data from
the All-India Long-term Soil Fertility Trial series show depletion of P and Zn
in rice-wheat areas. In other areas, deficiencies ofK and B, and problems with
pH were observed.
Comment: In summary, there is no single data set that convincingly and
unambiguously confirms that farmers are facing a long-term decline in
productivity in the rice-wheat pattern. Rather, there are indications from
various sources that the pattern, as currently managed, may not be sustainable.
The indications seem strong enough, however, and the sources sufficiently
consistent, to be cause for alarm.
The discussion of long-term issues has so far focused on soil fertility.
Additional evidence on nutrient depletion, and pest and disease buildup will
be presented in the next sections.

Table 4. Long-Term Fertilizer Trial-Partial Results Grain Yield:
First Rice Crop, Wheat Crop (t/ha)

Treatment Average Yields Average Yields
Years 1 and 2 Years 6 and 7
Rice Wheat Rice Wheat

1 2.5 0.8 0.6 1.1
2 3.8 1.0 1.1 1.2
3 3.8 1.8 2.3 2.1
4 3.9 1.2 1.0 1.3
5 4.0 2.1 2.8 2.5
6 3.7 1.9 1.9 2.4
7 3.2 1.0 0.7 1.0
8 3.3 1.3 2.2 2.3
9 2.9 1.9 2.3 2.2

Source: Based on Regmi (1986), referred to in a personal communication by D.

Vol. 1, No. 2, 1990


Long-Term Problem 1: Nutrient Deficiencies Will Increasingly Limit the
Yields of Both Rice and Wheat
Hypotheses: Field observations made during the diagnostic survey sug-
gested that nutrient deficiencies were already restricting wheat yields, especial-
ly on lighter soils in the upper terraces (Type B farmers). This is not surprising,
given these farmers' soil-fertility management practices. Researchers hy-
pothesized (independently of the long-term trial data presented in the last
section) that the preceding rice crop is probably subject to similar nutrient
stresses. Further diagnostic work is needed, however, to clarify the relative
importance of different nutrients and identify interactions among nutrients.
(Note that nutrient deficiency problems may differ between farmer Types A,
B, and C.)
This problem of nutrient deficiencies seems likely to get worse over time.
Those farmers with longer experience with intensified cropping patterns
(Type C farmers) indicated that yields of both rice and wheat are lower now
than when intensified cropping began. Hypothesized causes of nutrient
deficiencies are listed below. Note that many of these causes are likely to have
cumulative effects over time.

1. The rice-wheat cropping pattern tends to exhaust soil nutrients, com-
pared to the earlier cropping pattern of rice-fallow; as more fields are shifted
to this pattern, nutrient deficiencies are likely to become more common.
2. The subsurface pan (left by puddled rice culture) restricts the rooting
zone of wheat as well as rice, thus upper soil layers are being mined of plant
3. Farmers apply only low levels of inorganic fertilizer to both crops.
4. As FYM supplies decline, and as FYM is increasingly used for fuel,
farmers are reducing to negligible levels the application of FYM to rice and
wheat; many rice and wheat fields receive no FYM at all.
5. Crop residues and weeds are fed to livestock rather than incorporated
into the soil.

Improving problem definition regarding nutrient deficiencies: Future
diagnostic research on the long-term problems posed by nutrient deficiencies
is likely to focus on three interrelated questions: (1) Which nutrients are
limiting? (2) How does productivity decline over time as a consequence of
nutrient deficiencies? (3) What is the distribution of nutrient deficiencies in
the study area? (This question of geographical incidence is likely to be strongly
affected by the farmer types noted earlier.)

Journal for Farming Systems Research-Extension


Nitrogen and
phosphate deficiencies
reduce wheat yields
(near-term problem)

Deficiencies of
minor nutrients are
likely to become
more sever over time

Figure 4. Soil Fertility: Problems and Causes

Vol. 1, No. 2, 1990


Current thinking is that, in the Rupandehi study area, P is usually the
limiting factor for wheat, and both P and Zn for rice-especially in the upper
terraces. These thoughts are suggested by background knowledge of rice and
wheat agronomy, in combination with data from one on-station experiment
in the study area, and a series of experiments in similar areas of India. Clearly,
further research is needed in Rupandehi District itself to clearly identify the
order in which nutrients become deficient, the effects of these deficiencies
over time, and their distribution within the district.
Several sources of data can be used to address these questions. The use of
long-term trials(on-station and/or on-farm) is one obvious approach. Another
approach would involve the use of nurseries of indicator species capable of
flagging well-defined micronutrient deficiencies under farmers' conditions
(D. Saunders, personal communication). Yet another approach would involve
monitoring a panel offarmers over time, to track rice and wheat yields and
relate these to farmers' soil-fertility management practices, including FYM
supplies and management. In addition, single-visit surveys using random
sampling may be needed (in conjunction with soils information) to test some
of the hypotheses on the causes of the soil fertility problem.
It should be clear however, that this three-dimensional uncertainty (which
nutrient? what time path? what geographical distribution?) tremendously
complicates problem definition.
Alternative solutions for the nutrient-deficiency problem: Although
there appears to be no simple solution to nutrient deficiency problems,
researchers might consider the following themes (suggested by the hypoth-
esized causes of the problem):
1. Realistic and profitable doses (and forms of application) of inorganic
fertilizer, possibly including micronutrients (e.g., zinc for rice).
2. Improved FYM management and techniques to combine FYM and
inorganic fertilizer, to increase fertilizer efficiency.
3. Development of alternative sources of fodder, to allow an increase in
animal herd size, increased production ofFYM, and the incorporation of more
FYM and crop residues back into the soil (e.g., fitting multipurpose [grain-
fodder-green manure] legumes into the system).
4. Development of alternative fuel sources, to enable farmers to use FYM
as fertilizer instead of fuel (e.g., agroforestry research to test alternative tree
species as sources of fuel and fodder, etc.).

A suitable research agenda will likely make use of various sources of

Journal for Farming Systems Research-Extension


information, including conventional researcher-managed, on-farm research
trials; farmer-participatory research (especially with respect to agroforestry
and green legume research); researcher and/or farmer-managed, long-term
trials, and monitoring of a farmer panel (required to measure the time path of
benefits associated with different interventions); exhaustive analysis of past
and present data sets (given the expense of setting up new sets of trials).

Long-Term Problem 2: Pests and Diseases Will Increasingly Reduce Rice
and Wheat Yields
There is evidence that a number of pests and diseases reduce wheat yields
in the study area. These include Helminthosporium blight, nematodes, soil
fungus, rats, etc. With respect to rice, farmers report problems with blast, rice
bug, stemborer, and rats. The incidence, frequency, and yield loss associated
with each of these is not yet well understood. Similarly, there is little evidence
on the time path of productivity changes (are these problems getting worse
over time?).
Evidence on pests and diseases includes the following:

* Solarization trials conducted at the NWDP experiment station at Bhaira-
hawa indicate that unidentified biotic factors have a strong negative effect on
wheat yields.

* The rice root nematode (Hirschmaniella sp.) was found in all soil samples
obtained from farmers' fields in the study area during the diagnostic survey.
However, it is as yet unknown whether these cause any yield loss for wheat.

* Evidence from similar rice-wheat areas of Bangladesh suggest that biotic
factors are reducing wheat emergence and plant stand (D. Saunders, personal
communication, reported by P. Hobbs).
Though there is little evidence at this time to support it, there is widespread
concern that these problems may become more severe as time passes. This is
simply because the rice-wheat pattern (which is relatively new to the study
area, and still expanding in size) seems more likely to allow a buildup of pests
and diseases than the earlier rice-fallow pattern.
For biotic factors, problem definition remains at an early stage. As noted,
research is needed to determine which pests and diseases (ifany) are increasing
in severity or frequency. Long-term trials and monitoring may be needed to
trace out time paths, as well as specialized surveys to measure the incidence of
problems in the study area. In addition, on-station research and laboratory

Vol. 1, No. 2, 1990


testing, conducted by trained disciplinary specialists, are likely to be needed.

Discussion and Conclusions
The rice-wheat pattern as cultivated in Rupandehi District of the Nepal
Terai suffers from a number of problems, many of them near-term in nature.
For example, problems associated with wheat stand establishment, water-
logging, planting date, etc., appear to have major effects on wheat produc-
The rice-wheat pattern also appears to suffer from problems of sustain-
ability-problems, moreover, that are not immediately obvious. Available
secondary data do not clearly indicate declining productivity of rice and wheat
over time. Similarly, farmer opinion is mixed, with some farmers reporting that
yields are declining, while other farmers report that yields are stable or
Other sources of information, however, do suggest that rice and wheat
productivity are declining. These sources include:

* Stratification of farmers into farmer types: the only farmers reporting
increasing yields are those with a brief experience with intensified cropping.
Farmers with more experience tend to report declining yields.

* Results from a long-term, on-station fertilizer trial, along with reports of
similar results from similar rice-wheat areas of India, that indicate rapid
declines in rice yields over a seven-year period under a wide array of soil-
fertility management strategies.
* A marked trend among farmers toward the reduction of FYM applications
to rice and wheat, in response to declining herd size (hence, declining FYM
availability) and increased use of FYM for fuel.
* On-station research results suggesting that unknown biotic factors (nem-
atodes? root rots?) have a strong effect on yields, together with the hypothesis
that these biotic factors may tend to become more important in the intensified
rice-wheat pattern, as compared to the earlier rice-fallow pattern.
Much work on problem definition remains, however. The time path of
productivity loss needs to be traced out; the changing incidence of each
problem needs to be identified; and hypothesized problem-cause relationships
need to be tested. All of these are challenging, complex, expensive, and time-
consuming tasks. Moreover, these tasks will require the services of commodity

Journal for Farming Systems Research-Extension


and disciplinary scientists, as well as dedicated "systems researchers."
In the end, researchers will probably have to assemble the best estimates
of yield response and productivity loss from the different available data
sources, then construct a synthetic time path of productivity change for each
of the different soil-fertility management strategies (including the farmers'
practice), and calculate the net present values associated with each of these
strategies. Discounting need not be deferred until the end of the long-term
trials; as data accumulates, it can be fed into successive approximations of the
complete time path associated with each management strategy.
The sustainability problems discussed in this paper may not be as dramatic
as those discussed by some other researchers, e.g., dealing with extensive
deforestation orhighlyvisible soil erosion. However, the very "nonobviousness"
of these problems may tend to delay their definition and obscure their
importance, with ultimately disastrous effects on productivity and farmers'


APROSC. 1986. Semi-detailed soil survey: Report for the Bhairahawa Lumbini Ground-
water Project. Kathmandu: Agricultural Projects Services Centre.
Byerlee, D.B., P. Heisey, and P.R. Hobbs. Diagnosing research priorities for small
farmers: Experiences from on-farm research in Pakistan. Quarterly Journal of Inter-
national Agriculture 28(3/4):254-265.
Fujisaka, S., and D. Garrity. 1988. Developing sustainable food crop farming systems for
the sloping acid uplands: A farmer-participatory approach. Presented at the 4th SUAN
Research Symposium, Khon Kaen, Thailand, July 4-8, 1988.
Fujisaka, S., and L. Harrington, (eds.). 1989. The rice-wheat cropping pattern in the
Nepal Terai: Farmers'practices and problems and needs for future research. Bangkok:
Harrington, L., et al. 1989. Approaches to on-farm client-oriented research: Similarities,
differences and future directions. Presented at the International Workshop on
Developments in Procedures for FSR/OFR, Bogor, Indonesia, March 13-17, 1989.
Hobbs, P.R., C.E. Mann, and L. Butler. 1987. A perspective on research needs for the
rice-wheat rotation. In A. Klatt, technical ed., Wheatproduction constraints in tropical
environments. El Batan: UNDP/CIMMYT.
Regmi, K 1986. Agronomic investigations-plant nutrition aspects of wheat. Presented
at the Wheat Working Group Meeting, NWDP, Bhairahawa, Nepal, September 1-3,
Roche, F. 1988. Java's critical uplands: Is sustainable development possible? Food Re-
search Institute Studies 21(1):1-43.
Tripp, R, and J. Wooley. 1989. The planningstage ofon-farm research: Identifyingfactors
for experimentation. Mexico D.F., and Call, Colombia: CIMMYT and CIAT.

Vol. 1, No. 2, 1990

Relations between Agricultural Researchers
and Extension Workers:
The Survey Evidence

Stephan Seegers and David Kaimowitz'


The objective of this paper is to describe the communication between
agricultural researchers and extension workers and their attitudes toward each
To be effective, agricultural research must be relevant to producers' needs,
and its results, including the necessary inputs and infrastructure, must be made
available to producers. This usually requires specific efforts to extend the new
technology, even though that may not necessarily be accomplished by a
traditional general-public extension service. Much technology is transferred
to farmers by private sector companies, nongovernmental organizations, and
other types of public institutions.
Public agricultural-research institutions often have poor relations with
extension agencies. In 16 out of 20 research projects evaluated by the U.S.
Agency for International Development and in all 12 projects evaluated by the
Food and Agricultural Organization (1984), communication between research
and extension was weak. The World Bank (1985) says that "bridging the gap
between research and extension is the most serious institutional problem in
developing an effective research and extension system."
Previous authors have noted that extension workers see researchers as
working in "ivory towers" and producing technologies that are not applicable
to the farmers with whom they work (FAO, 1984; Samy, 1986). Researchers

SResearcher, Department of Extension Science, Wageningen Agricultural University,
The Netherlands, and coordinator of an international comparative study on the links
between agricultural research and technology transfer in developing countries, Interna-
tional Service for National Agricultural Research (ISNAR), The Hague. The authors
would like to thank Ruben Echeverria, Howard Elliott, and Deborah Merrill-Sands for
their useful comments and Anna Wuyts for her assistance in working with the survey


look down on extension workers and question extension agents' capacity to
perform their jobs (Quisumbing, 1984). Both researchers and extension
agents avoid the tasks that bridge the two activities such as adaptive field trials
and producing written material for extension agents (McDermott, 1987).
Communication between the two groups is limited. These problems are
caused by differences in background, training, experience, responsibilities,
status, institutional setting, and physical location, which promote competi-
tion between the two groups and hinder their ability to understand each other
(Bennell, 1989).
Most writing on the topic has been prescriptive or based on anecdotal
evidence or individual cases. This paper is the first attempt to bring together
the international survey evidence on the subject.
The paper only covers aspects that can be effectively studied using surveys.
It forms part of a larger comparative study of research/technology transfer
linkages currently underway at the International Service for National Agri-
cultural Research (ISNAR). The study is also using qualitative methods such
as case studies and is looking at the broader institutional and structural aspects
of the problem (see Kaimowitz, 1990). Still, surveys can provide unique
lessons for future agricultural research and extension policies in developing
The first section presents our methodology. We then discuss the evidence
on (1) extension input into research, (2) different channels researchers and
extension agents use to communicate with each other, (3) the subjects they
communicate about, (4) the two groups' attitudes toward each other, and (5)
how various personal attributes influence the research-extension relationship.
We then summarize the key conclusions.


The summary tables from 21 surveys of individual agricultural researchers
and/or extension workers with information about relations between the two
groups were collected through an extensive literature review over a three-year
period. These surveys came from 18 countries, including countries in Asia and
Oceania, Latin America and the Caribbean, Africa, the Middle East and the
United States (see Table 1).
Three of the surveys focus exclusively on research and extension for a single
commodity (wool in Australia, coffee in Colombia, and rice in the Dominican
Republic). The rest cover multiple commodities. Although the text consistently

Journal for Farming Systems Research-Extension


Table 1. Summary of Samples in Surveys Useda

Country Author Year Research Extension Regions and
workers workers commodities



Rio et al.


Dominican Rep. Doorman

Dominican Rep. Malkun
Egypt Samy
India Rao







Pap New Guinea Kern
Sierra Leone Lakoh
Taiwan Lionberger

1960 35 47 Buenos Aires
1976 24 25 New South
Wales/sheep &
1987 145 national
1984 145 843 national
1982 175 5 departments/
1985 14 34 Bonao, Mao,
1980 n.d. national
1988 98 64 5 regions
1972 n.d. 429 Punjab, Tamil
Nadu, Adra
1986 52 105 West Java
1970 30 56 national/field
crops, cattle,
1975 21 54 national
1974 27 48 Ife
1988 18 45 Zaria, Ibaden,
1988 50 76 Punjab/wheat
and sugar
1985 105 four provinces
1986 48 northern area
1970 122 484 western part


Lupanga 1986

Thailand Dhandhanin 1984
Trin & Tobago Alleyne 1975
U.S. Jain 1970

coastal and south-
ern highlands

Vol. 1, No. 2, 1990

a More complete information about the samples can be obtained from the authors. We
are grateful to Anna Wyuts for compiling this information.


refers to the surveys by country names, many of the surveys only cover specific
regions within these countries.
The samples varied in size from 48 in Sierra Leone and the Dominican
Republic (rice) to 988 in Colombia, with a median sample size of 108. Two-
thirds of the surveys were conducted after 1982, 10 in the 1970s, and two
(Argentina and Taiwan) in the 1960s. The specific conditions in the countries
where surveys were conducted some time ago have undoubtedly changed, but
there is no reason to believe the general pattern of relations presented in this
paper has varied significantly. Half the surveys come from unpublished Ph.D.
dissertations, the rest from consultants' reports, journal articles, and books.
Statistically speaking, the surveys are not fully comparable. The surveys each
had different samples, questions, and objectives. Thus no attempt was made to
rigorously test statistical hypotheses. Instead we sought to present the general
pattern of research-extension relations. The lack of strict comparability also
made it difficult to present much of the material in summary tables.
We first divided the material into research and extension responses and
organized the survey table by topic. Then we compared the information on that
topic between countries and integrated the research and extension responses.
For any one specific topic, only a subset of the surveys had relevant information.
We were particularly interested in the differences between research-extension
relations in countries or systems often mentioned as having effective extension
systems (Australia, coffee in Colombia, Israel, Taiwan, and Argentina in the
early 1960s) and relations in countries with less effective systems.2 The first
group comes mostly from more developed countries or, as in the case of coffee,
commodity-specific systems supporting politicallysensitive products. Researchers
and extension workers have more similar profiles in this first group.


Extension's Input into Research
Current conventional wisdom says extension agents can and should help
define research problems, provide technical information to researchers, and
give feedback on how research-generated technologies perform in the field.
Since extension input is relatively common, particularly in more advanced

2 Rice extension in the Dominican Republic has also been said to have been effective. This
is reflected in our 1980 Dominican Republic data. However, by 1986 the system was
in a state of decline. Moreover, the 1986 data was drawn from regions where rice
extension has traditionally been weak.

Journal for Farming Systems Research-Extension


extension systems, the survey data support these ideas. It also shows that
although extension input is important, extension workers are not the main
source of research ideas, nor are a majority of them directly involved in
providing input, in any of the countries studied.
In countries such as Argentina and Colombia (coffee), a significant
minority of extension workers provided input into research. In contrast,
extension agents in Pakistan, Sierra Leone, and the Dominican Republic (rice)
had practically no input. Nowhere did a majority of extension workers report
input to research.
Researchers from 10 countries reported input or feedback from extension.
In Egypt 52.3 percent of the researchers surveyed said extension was an
important source ofnew ideas. In Indonesia 61.5 percent thought extension
should help determine research priorities. Feedback from extension workers
and farmers was the source of 23 percent of research projects in the institutes
sampled in India.
Yet in all seven countries with data, researchers said most ideas for research
problems came from the research community itself. In Argentina, Colombia,
Indonesia, Pakistan, and Tanzania, researchers considered farmers a more
important source of input than extension. Moreover, in some countries the
same researchers who said extension input was important admitted devoting
little effort to obtaining it (Taiwan) or found the information extension
provided not to be useful (Tanzania).
Extension workers believe they are competent to help determine research
priorities and want to do more in this regard. This point comes through
strongly in Argentina, the Dominican Republic, Papua New Guinea, and
Sierra Leone. While some extension workers, particularly in Colombia
(coffee), said they had not suggested topics for research because they did not
feel the need to, in three countries extension workers complained that they did
not know how research problems were selected and lacked channels for giving
their ideas.

Personal Contacts between Researchers and Extension Workers
Countries with stronger extension systems and with commodity-specific
extension specialists have substantially more direct personal contacts between
researchers and extension workers (see Tables 2, 3, and 4). In Argentina,
Australia, Israel, Taiwan, and the Dominican Republic (rice in 1980), there
were frequent direct contacts. In Colombia (non-Colombian Agricultural
Institute), Egypt, India, Indonesia, Jamaica, Pakistan, Papua New Guinea,

Vol. 1, No. 2, 1990


Table 2a: Average Number of Times Each Researcher Participated in Selected
Communications Activities during the Year

Personal Meetings Training Trials and
Country contact demonstrations

Australia 27.4 7.5 9.1
Israel 6.2 1.2 2.0
Indonesia 1.8 0.9 0.2 0.5
Egypt 4.4 0.6 1.5
Pakistan 0.7a 0.1 -
Others 0.7- -

a Not including agricultural extension directors.
b Tanzania.

Table 2b: Average Number of Times Each Extension Worker Participated in Selected
Communications Activities during the Year

Personal Meetings Training Trials and
Country contact demonstrations

Argentina 1.5 0.6
Australia 4.2 0.4 -
Israel 9.5 5.2 3.3
Indonesia 1.5 1.8 -
AOa 0.3 0.4 0.2
FEWb 0.2 0.4 0.1
Dominican Republic 2.4 -
Papua New Guinea 3.7 -

a Agricultural officer.
b Field extension worker.

Sierra Leone, and Trinidad and Tobago, such contacts were much less
common. In these cases extension workers depend heavily on relations with
their superiors within the extension services. The Dominican Republic (rice in
1986), ICA in Colombia, Nigeria, Tanzania, and Thailand are intermediate
On average, wool researchers in Australia had 27.4 direct contacts with
extension workers during the year. It was not uncommon for an extension
agent to telephone a research colleague. Almost three-quarters of extension

Journal for Farming Systems Research-Extension .


Table 3a: Percentage of Researchers Involved in Different Communications Activities

Personal Meetings Training Trials and
Country contact demonstrations

Argentina 60 40 -
Experiment Station 39 5 18 14
Research Institute 81 34 43 66
ICAa 27 77 87 64
Indonesiab 12-21 0-25 0-29 0-29
Egypt 44 26 25/18c
Pakistan 12-16 10 10-46 10
Nigeria 15 7 31/40d -
Tanzania 62 -
Thailand 3 -

a Institute Colombiano Agropecuario.
b Range of percentages involved in activities with subject-matter specialists, agricultural
officers, and field extension workers.
c The first number refers to trials, the second to demonstrations.
d The first number refers to training events, the second to seminars.

agents in Argentina used personal contacts to find out about research results.
In Israel researchers reported having direct contact with an average of 12
extension workers during the previous two years; extension agents reported
contacts with an average of4 researchers. Over 90 percent of extension agents
and adaptive researchers in Taiwan reported personal contacts with each
other. Such contacts were the most common means of communication
between research and extension. Contact with applied researchers at Taiwanese
research institutes was less frequent but still important.
In contrast, in Indonesia and Pakistan less than one quarter of researchers
had personal contacts with extension workers and on average these contacts
occurred less than once per year. Even lower percentages of researchers visited
farmers' fields with extension workers or helped them identify or solve
farmers' problems. In Sierra Leone there had been no personal contacts
between the researchers and extension workers surveyed in the previous two
years. In Egypt, India, Jamaica, and Trinidad and Tobago, less than one
quarter of extension workers had significant direct contact. Extension
workers from the Colombian Agricultural Institute (ICA) reported substan-

Vol. 1, No. 2, 1990


Table 3b: Percentage of Extension Workers Involved in Different


Personal Meetings Training Trials and
Country contact demonstrations

Argentina 66 42 86 25
Improvement Station 60 28 95 96
Research Institute 43 74 95 96
ICAa workers with input 0-72 57-68 64 -
from ICA researchers
ICA workers with input 0-30 30-36 27 -
from non-ICA researchers
Indonesia 59 53 -
Egypt 34 70 39/16c
Agricultural officers 11 76 2
Field extension workers 10 79 37
Nigeria 4 58 65
Tanzania 36 34 16 -
Other d -

a Institute Colombiano Agropecuario.
b Range of percentages reflects different types of meetings and contacts.
c The first number refers to trials, the second to demonstrations.
d Dominican Republic-20 percent; India-41 percent general extension officers, 31
percent specialist extension officers.

tial contact with researchers, but only 27 percent of researchers said they had
regular contact with extension agents.

Publications are an important channel for researchers to communicate their
results to extension workers. In Argentina, Egypt, Indonesia, Pakistan, and
Taiwan, between 33 percent and 55 percent of researchers reported writing
articles for extension agents.
Public agricultural researchers in more developed countries dedicate
greater efforts to writing materials for extension. The average number of
extension publications written annually by each researcher varied from .42 in
Pakistan and .63 in Indonesia to .93 for adaptive researchers in Taiwan and 2.3
in Australia.
Research materials take a long time to be published and the field-level

Journal for Farming Systems Research-Extension


Table 4a: Performance of Researchers and Extension Workers from Countries with
Effective Extension Systems in Various Communications Channels

Personal Meetings Training Trials and
Country contact demonstrations

Argentina high medium low low
Australia high medium high high
Israel high high high
Taiwan medium low medium medium
applied res.
Taiwan high medium medium high
adaptive res.

Table 4b: Performance of Researchers and Extension Workers from Countries with
Ineffective Extension Systems in Various Communication Channels

Personal Meetings Training Trials and
Country contact demonstrations

ICA medium medium medium low
non-ICA low low low low
Egypt low high low medium
Indonesia low low low low
Nigeria low low low low
Pakistan low low low low
Tanzania low low low low
Others a low lowc

a Low: Sierra Leone, India, Jamaica, Trinidad and Tobago; medium: Dominican Repub-
lic (Rice,1986) and Papua New Guinea.
c Trinidad and Tobago.

extension agents have trouble getting access to them. In Colombia 93 percent
of researchers said bureaucratic delays in publishing kept them from dissem-
inating their results. Most extension workers in Sierra Leone had trouble
obtaining relevant research findings when needed because of long publication
delays. Eighty-one percent of Egyptian researchers sent their publication only
to extension headquarters, where field agents rarely had access to them.
Extension workers in Colombia (coffee), Egypt, Papua New Guinea, and

Vol. 1, No. 2, 1990


Pakistan complained that publications were difficult to obtain or arrived late.
Extension workers prefer more popular materials such as bulletins, bro-
chures, leaflets, and manuals over scientific research journals. This tendency
is greater when extension workers are less educated.
Extension workers from ICA in Colombia were more interested in receiv-
ing brochures and handouts than journals. Similarly, coffee extension
workers, particularly those with only vocational training, enjoyed technical
bulletins more than scientific journals and found them more interesting.
Only 10 percent of Papua New Guinea extension agents received their
ministry's research journal and even these did not find it useful. Half the
agents received a more popularized ministry publication and 90 percent a
simple publication for farmers, both of which they enjoyed and found useful.
The extension workers clearly preferred simple publications, available in the
local language.3
Taiwanese extension workers regularly used extension materials, adaptive
research publications, and farm magazines. They made less use of research
institute publications. They considered extension publications more handy
and practical than research materials, although less up-to-date and scientific.
Similarly, in Pakistan, Tanzania, and Nigeria, extensionists preferred simple,
more practical publications. Only in Australia, where extension workers are
highly educated, did they use journals more than research reports and other
department of agriculture publications.

Training Events and Research-Extension Meetings
Formal training events and research-extension meetings are common in the
more advanced systems (see Tables 2, 3, and 4). Training events were ranked very
highly by Australian extension workers as channels for gathering information. Joint
research-extension meetings were common in Argentina, Israel, and Taiwan.
These activities are also important in certain countries with weaker extension
systems. In Egypt, for example, formal joint meetings were the principal channel
for informing extension about available technology. Forty-four percent of
researchers and 70 percent of extension agents participated in at least one joint
meeting during the previous year, and most found the meetings useful.
A majority of extension workers in the Colombian Agricultural Institute
and subject-matter specialists in Nigeria and Indonesia had attended courses

SPreference for publications in local languages was also important for extension workers
in Pakistan.

Journal for Farming Systems Research-Extension


or seminars conducted by researchers. In Indonesia, Colombia, and Thailand
more than half of the researchers were involved in extension training. Over 70
percent of the researchers and extension workers surveyed in Tanzania partic-
ipated in joint meetings, seminars, conferences, or workshops. More than three-
quarters of Pakistani extension workers had received training from researchers.
Demonstrations and field days are other common training events. These
exist in most countries, although their importance varies. ICA researchers in
Colombia, adaptive researchers in Taiwan, and subject-matter specialists in
Nigeria reported high participation in these activities (two-thirds or more
participated). Low participation was found among researchers in Egypt,
Indonesia, Pakistan, and Taiwan (applied researchers) and extension workers
in Egypt, Jamaica, and Trinidad and Tobago.4
There is little indication that meetings or training events are frequent of
take up much of either researchers' or extension workers' time in most
countries. A large percentage of those surveyed in Indonesia, Pakistan,
Tanzania, Egypt, and Thailand had only participated once or twice in these
events or said they did not participate frequently.

Research-Extension Field Trials
Joint field trials play a major role in research-extension relations in the more
advanced systems (see Tables 2, 3, and 4). One-third of researchers' contacts
with extension and farmers in Australia focused on cooperative trials. Similarly
in Israel a third of research-extension contacts occurred during joint trials.
Depending on what commodity was involved, joint trials were the first or
second most important setting for extension workers to communicate with
researchers.5 In Argentina 25 percent of researchers used joint on-farm trials
during the previous year, most of which were initiated by research administra-
tors, not individual researchers or extension workers. In Pakistan 37 percent
of agricultural officers and 21 percent of field assistants were involved,
although only 10 percent of researchers participated, and researchers ex-

4Pakistani extension workers reported high participation in demonstrations. 70 percent
of agricultural officers and 55 percent of field assistants said they used them often. Yet
it is not clear if they were referring to joint demonstrations with research or simply
extension demonstrations for farmers. The extension workers made some use of field
days but not often, and many of those who participated questioned their usefulness and
the completeness of the information presented.
SField trials were the most important setting for meetings on citrus and the second most
important for field crops and cattle.

Vol. 1, No. 2, 1990


pressed strong disapproval of extension having adaptive research responsibil-
ities. The average number ofjoint trials per research or extension-person year
in Egypt and Pakistan was far below that in Israel. The lack of information on
this topic in the other surveys may imply joint trials rarely occur in many of the
remaining countries.

Research Information Received or Required by Extension
The most important crop information extension receives from research
relates to field crop varieties and plant protection. Seed varieties and, to a
lesser degree, new pesticides and fertilizers, were the dominant type of
technology received by Egyptian extension workers. Varieties and crop
protection were the most important themes in the publications sent to coffee
extension workers in Colombia. In Sierra Leone extension workers depended
on researchers as their primary source on only one topic-varieties technolo-
gy. For all other technologies they relied principally on their own knowledge.
Crop protection was very high among extension workers' priorities in Egypt,
Israel, Pakistan, and Colombia (including coffee extension). Fertilization and
soils problems were often mentioned but fell far behind crop protection
among extension agents' principal concerns.
Researchers transfer mostly technical information. Extension workers
receive little social science information from researchers, and they give these
issues low priority.
Information flows more easily when both researchers and extension
workers specialize in the same commodity. This comes out clearly in the data
from the Dominican Republic, Colombia, and Israel.

Researchers' Attitudes toward Extension
Data from Argentina, the Dominican Republic (rice), Tanzania, Pakistan,
Indonesia, and Nigeria support the hypothesis that researchers in developing
countries have a poor view of extension. Researchers in these countries felt
extension was ineffective and blamed the problem on insufficient education
and training, poor incentives, and frequent staff turnover (Dominican Repub-
lic, Nigeria, Tanzania). They were also unclear about extension's mandate
(Argentina, Tanzania).
Most ofArgentina's researchers thought the extension agents were incapa-
ble or only partially capable of fulfilling their functions. This feeling was
shared by rice researchers in the Dominican Republic, particularly with respect
to the general-public extension service (as opposed to rice development

Journal for Farming Systems Research-Extension


department). Three-quarters believed experienced farmers had more knowl-
edge about rice than recent graduates working in extension.
Tanzanian researchers said that extension workers didn't appreciate the
complexity of research (65 percent), were not well trained (54 percent), and
did not know much about farming (49 percent). A majority of researchers in
Indonesia, Pakistan, and Tanzania considered extension ineffectiveness a
major cause of non-adoption.6
In none of the six countries just mentioned did researchers see the limited
applicability of their own results as a major cause of low adoption. Those who
did not blame extension mostly said poor adoption was due to farmers'
traditionalism or poor agricultural policies.

Extension Workers' Attitude toward Research
Extension workers do not question the researchers' technical competence,
but many complain that not enough research is being conducted, the research
carried out does not meet their needs, and not enough is being done to
communicate results to extension (Argentina, Papua New Guinea, Pakistan).
Large majorities of the extension agents in the Dominican Republic, Sierra
Leone, and Tanzania had strong doubts about whether the research being
conducted was relevant to farmers' needs. A minority from Argentina
expressed similar concerns.
The agents gave various explanations for the lack of relevance. Researchers
make technical recommendations without considering their profitability
(Argentina). Funding sources with research agendas not relevant to extension
have excessive influence (Sierra Leone). Researchers don't interact enough
with extension agents (Sierra Leone) and know little about farmers' problems
The view from Jamaica was mixed. Researchers were thought to perform well
on (1) choosing appropriate problems for research, (2) making practical
recommendations for farmers, and (3) being committed to solving problems for
small- and medium-scale farmers. They received lower marks on providing
resource materials to extension and following up on research recommendations.

'The Tanzanian data is contradictory. Researchers also agreed with the statements: (1)
"For helping small farmers, the extension worker is more important than the research-
er" (74 percent), (2) "Extension workers have a lot to extend to farmers" (85 percent),
(3) "It is not a waste of time for researchers to consult extension workers" (94 percent),
and (4) "Abolishment of extension in Tanzania would not go unnoticed" (80 percent).

Vol. 1, No. 2, 1990


The Effect of Personal Attributes
Various surveys examined the correlation between the researchers' and
extension workers' personal attributes and their communications patterns and
attitudes. The results, however, are inconclusive.
Age seems to have a positive effect on research-extension relations. Age was
positively correlated with the number of physical and material objects re-
searchers transferred to extension in Egypt. Older researchers in Australia and
Tanzania were more receptive to extension communication. Older extension
workers in Tanzania were more inclined to feel that joint field days, confer-
ences, seminars, and workshops were useful. No correlation was found
between age and other variables studied in Australia, Nigeria, Tanzania, and
the United States.7
Professionals' length of service may partially counterbalance the age effect.
The longer that researchers were in one location, the less receptive they were
to communication with extension (Australia) or participating in joint meet-
ings (Tanzania). Their attitudes toward extension also became more negative
(Tanzania). The number and frequency of contacts declined with length of
service for both research and extension staff in Nigeria and for extension
workers in Israel. Length of service had no effect on the reception of new
technologies by Egyptian extension workers or the propensity of Tanzanian
researchers and extension workers to contact and respond to each other.
For several other variables the results were mixed. More formal education
increased the popularity and intensity ofcommunications between researchers
and extension workers in Israel, but had no impact on researchers' attitudes
toward extension workers in Australia. Coming from a farm background had
a positive effect on the frequency, intensity, or popularity of research-
extension communication in Israel or the number of technologies Egyptian
extension workers receive. There is also contradictory evidence on the effect
of organizational rank and status.


Researchers and extension workers communicate with each other through
meetings, training events, publications, joint participation in trials and dem-

7The evidence from Tanzania on how researchers' age affected their attitudes about
extension is contradictory. Older researchers were more inclined not to blame
extension workers for poor adoption. They were also more likely to believe extension
workers were poorly trained and had little to extend.

Journal for Farming Systems Research-Extension


onstrations, and direct personal contact. Those who work in more developed
systems and commodity-specific systems communicate with each other more.
In particular, they communicate more informally and place greater importance
on joint research-extension trials.
New varieties and crop protection are the major focuses of research-
extension interaction with respect to crops. Crop protection is a key
connection because extension workers and producers concentrate their de-
mands for research on problems they perceive as urgent. They rarely
emphasize long-term or less obvious problems.
The more effective extension services have input into determining research
problems. Researchers in most countries have some doubts about such input,
but are willing to give the idea qualified support. Extension workers actively
want input and feel competent to provide it.
Still, the potential for extension input should not be exaggerated. Evidence
from the more developed systems suggests that extension will probably never
replace the research community as the primary source of research ideas, and
only a minority of extension agents are likely to be involved.
One major reason researchers and extension workers communicate less in
developing countries is the negative attitude they have about each other.
Researchers doubt whether extension agents are competent and motivated to
work, and extension agents question whether the research being done is
Extension workers want researchers to put more effort into communicating
their findings. They also want simpler, more timely, and applicable materials,
written in their local language, and greater efforts to give field-level workers
access to such publications. Research journals are not an effective means to
communicate with extension.
To improve relations between the two groups in developing countries,
researchers will have to perceive extension agents as competent. In many
countries this can only happen if extension staff receive more training and
greater incentives. For its part, research will have to become more applicable,
through a greater emphasis on farmers' constraints, more on-farm research,
and greater input from farmers and extension.
Clear channels and procedures are needed if extension input is to increase.
To produce research materials appropriate for extension will require more
resources for researcher-communications departments and incentives for
researchers to dedicate more time to their extension audience.
Informal, direct, person-to-person communication is probably essential for

Vol. 1, No. 2, 1990


an effective flow of information. This is not surprising given evidence from
communications research elsewhere. It does, however, represent a major
challenge to most developing countries, where extension services are orga-
nized along hierarchical lines, extension workers have limited education, and
there are greater differences between researchers and extension workers.


Akinbode, I.A. 1974. An analysis of interorganizational relationships of agricultural
research, teaching and extension in Western Nigeria. Ph.D. dissertation, University
of Wisconsin.
Alleyne, E. 1975. Training and management policy for agricultural research scientists
in the governmental system: A study in two Caribbean states-Trinidad and Tobago
and Jamaica. Ph.D. dissertation, Cornell University.
Anyanwu, A.C. 1982. Communication behavior of professional disseminators of
scientific agricultural knowledge: A study of the Iowa Cooperative Extension Service.
Ph.D. dissertation, Iowa State University.
Be.nell,P. 1989. Intergroup relationship in institutionalagricultural technologysystems.
Linkages Theme Paper No. 2. The Hague: International Service for National
Agricultural Research.
Bernal, F.C. 1987. Analisis de la estructura de generaci6n y transferencia de tecnologia:
La subgerencia de investigaci6n y transferencia del ICA. Bogoti. Mimeograph.
Dhandhanin, M. 1984. An analysis of organizational structure, role perceptions, and
inter-role relationships affecting the functioningand effectiveness of the agricultural
extension education system in Thailand. Ph.D. dissertation, Comell University.
Doorman, F. 1986. iQue Pasa? Evaluaci6n del sistema de generaci6n y transferencia de
tecnologia en la producci6n arrocera en la Republica Dominicana. Institute Superior
de Agriculture, Santiago, Dominican Republic. Mimeograph.
Elkana, Y. 1970. Research-extension links in agriculture: A study of communications
behavior of professionals. In S. Molho and M. Gitlin, eds., Agricultural extension: A
sociological appraisal. Jerusalem: Keter Publishing House.
Food and Agricultural Organization. 1984. Agricultural extension: A reference manual,
2d edition. Rome: FAO.
Hargreaves, L.E. 1976. Communication between agricultural scientists and extension
workers: A study in the sheep and wool branch of the New South Wales Department of
Agriculture Agricultural Extension. Bulletin No. 1. Department of Agricultural
Economics, University of Sydney.
Hussein, S. 1986. An analysis of the agricultural knowledge system in Indonesia. Ph.D.
dissertation, Cornell University.
Idowu, I.A. 1988. Institutionalization of knowledge flows: An analysis of the links
between agricultural research and extension in Nigeria. Ph.D. dissertation, Technical
University of Berlin.
Institute Colombiano Agropecuario. 1984. Plan national de transferencia de tecnologia.

Journal for Farming Systems Research-Extension


Jain, N.C. 1970. Communication patterns and effectiveness ofprofessionals performing
linking roles in a research dissemination organization. Ph.D. dissertation, Michigan
State University.
Kaimowitz, D. 1990. Making the link: Bringing together agricultural research and
technology transfer in developing countries. Boulder, Colorado: Westview Press.
Lakoh, A.K. 1986. Organizational behaviour and program effectiveness: Analysis of the
Northern Agricultural Development Project extension delivery. Ph.D. dissertation,
Cornell University.
Lionberger, H.F., and H.C. Chang. 1970. Farm information for modernizing agricul-
ture: The Taiwan system. Washington, D.C.: Praeger Publishers.
Malik, W.H. 1988. An analysis of the agricultural knowledge system in Pakistan. Ph.D.
dissertation, Cornell University.
Malkun, J.L., S. Grullon, and J.I. Cepeda. 1980. Analisisdelsistema de extension rural
en la Republica Dominicana, Proyecto de reforzamiento de los sistemas de extension
rural en America Latina. RLA/79/037 PNUD/FAO. Santo Domingo: Secretaria
de Estado de Agricultura.
McDermott, J.K. 1987. Making extension effective: The role of extension/research
linkages. In W. Rivera and S. Schram, eds., Agricultural extension worldwide. New
York: Croom Helm.
Oliviera Moreno, J.E., 1982. Diagn6stico y bases para la evaluaci6n de la distribuci6n
de publicaciones de Cenicaf6 como elements para la difusion de innovaciones. M.A.
thesis, Pontificia Universidad Javerana, Bogoti.
Quimsumbing, E. 1984. New direction in research-extension linkages. In D. Elz, ed.,
Planning and management of agricultural research, A World Bank and ISNAR
Symposium. Washington, D.C.: World Bank.
Rao, R.C. 1972. Communication linkages in transfer of agricultural technology.
Economics and Political Weekly 17(53): A157-A178.
Rio, F., A.M. Arce, and J. Di Franco. 1960. Relaciones entire investigaci6n y extension.
Institute Interamericano de Ciencias Agricolas, Turrialba, Costa Rica.
Ruttan, V.W. 1982. Agricultural research policy. Minneapolis, Minnesota: University
of Minnesota Press.
Samy, M.M. 1986. Linking agricultural research and extension in developing countries.
Paper presented at the Annual Meeting of the Rural Sociological Society, Salt Lake
City, Utah, August 26-30.
Samy, M.M. 1988. Technology transfer in Egypt: The roles of agricultural research and
extension. Ph.D. dissertation, University of Illinois.
World Bank. 1985. Agricultural research and extension: An evaluation of the World
Bank's Experience. Washington, D.C.

Vol. I, No. 2, 1990

Agricultural Innovation and Technology
Testing by Gambian Farmers:
Hope for Institutionalizing On-Farm Research in
Small-Country Research Systems?1

Bradford Mills and Elon Gilbert2


Farmer participation in the identification and adaptation of improved agricul-
tural technologies is generally recognized as a major factor in agricultural
change in sub-Saharan Africa (Johnson, 1972; Vermeer, 1979; Richards,
1985, 1986). In recent years a variety of approaches, including farmer-back-
to-farmer and farmer first-and-last, have been discussed in the literature.
Several approaches have been field tested in an attempt to incorporate farmer
innovation and indigenous knowledge in the design and testing of technolo-
gies (Chambers, Pacey, and Thrupp, 1989). Increasing attention is being
given to cost-efficient, on-farm research (OFR) models that can effectively
serve resource-poor farmers (Chambers and Jiggins, 1986). The Gambia is an
example of a West African country where farmer innovations, in response to
opportunities and adversities, have produced profound changes in the farming
systems of the country over the past 20 years. Major changes include
widespread adoption and adaptation of externally introduced technologies
such as animal traction and fertilizer use, as well as a number of internally
introduced changes in plant varieties and cultural practices.
Yet despite the acknowledged importance of informal farmer experimen-
tation in agricultural change and increasing literature that favors greater
farmer participation in research, there are surprisingly few successful examples
of collaboration between farmers and researchers (Farrington and Martin,
1988). A generation of farming systems research-extension (FSRE) projects

Paper prepared for presentation at the Ninth Annual Farming Systems Research-
Extension Symposium, University of Arkansas, Fayetteville, October 9-11, 1989.
Mills is a graduate student in agricultural economics at the University of California,
Berkeley; Gilbert is an Associate Research Scientist with the Center for Research on
Economic Development (CRED) at the University of Michigan. Both authors were


throughout the developing world have illustrated the difficulties of institu-
tionalizing FSRE in general and farmer participation in particular (Merrill-
Sands, et al., 1989). To sustain substantive farmer involvement requires
resources and motivation that are not consistent with the endowments of most
national research systems outside the episodic help in the form of special
projects. This is especially true of small, low-resource countries such as The
Gambia where efforts to institutionalize on-farm research within the research
services have met with limited success.
Gilbert and Sompo-Ceesay (1988) argue that research systems in small
and/or low-resource countries should focus their efforts on technology
adaptation and testing while relying primarily on International Agricultural
Research Centers (IARCs) and larger research services in adjacent countries
to supply a range of technologies. Further, the formal research activities for
these small systems must be focused on relatively few priorities. Collaboration
between researchers and farmers in the identification and testing oftechnologies
becomes doubly important in small research systems since informal farmer
experimentation can include commodities and subjects that it would not be
possible to deal with otherwise.
This paper examines farmer experimentation, innovation, and adaptation
at the farm level and proposes approaches to improve the linkages between
formal and informal research activities in The Gambia. The second section
reports on a 1988 survey of farmer innovations, adaptations, and adoptions
carried out in two extension districts in the eastern part of the country. The
results provide additional evidence of farmer participation in informal adaptive
research. Further, the farmer-to-farmer spread of innovations, especially from
neighboring Senegal, rivals the importance of formal research and extension
efforts in promoting technologies that increase agricultural productivity. The\
third section reviews the factors influencing farmer innovations and agricul-
tural change in The Gambia and makes suggestions on how research and
development agencies might encourage these processes. The fourth section
proposes the expanded use of existing farmer groups for the technology
testing on farm. Testing of innovations by these farmer groups may be a
resource-efficient approach to expanding the topic and commodity coverage

agricultural economists with the Gambian Agricultural Research and Diversification
(GARD) Project, in the Department of Agricultural Research (DAR), The Gambia, at
the time the research associated with this paper was conducted. The authors wish to
express their appreciation to Joan Robertson, Josh Posner, and Jim Sumberg for their
helpful comments.

Journal for Farming Systems Research-Extension


of the research services. Testing may provide a basis for sustainable farmer-
based, agricultural-technology development in small research systems such as
The Gambia's.


Gambian farming systems have changed significantly in the past two to three
decades in response to opportunities and changes in the availability of
resources such as land, labor, and rainfall. The findings presented in this
section are based on background data and a 1988 survey of innovations and
adoptions by Gambian farmers for two extension districts in eastern Gambia:
Kuntaur District Extension Circle (DEC) in MacCarthy Island Division-
North (MID-N) and Giroba Kunda DEC on the south bank of Upper River
Division (URD). Innovations, adaptations, and adoptions recorded in the
survey are discussed in relation to the changes in resource availabilities.
Both Kuntaur and Giroba Kunda DECs adjoin divisional-administrative
and commercial centers (the towns of Kuntaur and Basse). This, in varying
degrees, contributes to a scarcity of land in both areas. Further, population
pressures are generally high in both areas, although growth rates are higher
(2.7 percent as compared to 2.0 percent between 1973 and 1983) in Fulladu
East (the district encompassing the Giroba Kunda DEC) than in Niani district
where Kuntaur is located. The difference is partly a function of migration,
with MID-N being an area of particularly high net out-migration to urban
centers and areas on the south bank where land availability and rainfall are
generally more favorable (Colvin, 1981). Labor availability during the peak
planting- and weeding-periods is cited as a major constraint to the expansion
of agricultural production in both areas (Boughton, et al., 1987).
In some areas, suitable land is becoming increasingly scarce, although in
MID-N it is still relatively abundant near the River Gambia (Boughton, et al.,
1988). In areas of land scarcity, notably in the uplands away from the river,
farmers continue to reduce fallow periods and to increase fertilizer use to
sustain yields. Fertilizer consumption in The Gambia is currently among the
highest in tropical Africa.3

McIntyre (1985) estimated fertilizer use in The Gambia at 16.43 kg/ha during the
period 1978-82, the third-highest rate in sub-Saharan Africa. In 1988 this had risen to
20 kg/ha.

Vol. 1, No. 2, 1990


Farmers have responded to labor constraints by acquiring and using draft
implements, which have diffused rapidly throughout the area in the last twenty
years. Currently, approximately 80 percent of the dabadas (households) in
URD and MID-N own at least one draft animal and one or more pieces of
equipment (Sumberg and Gilbert, 1988). An even higher percentage has
access to animal traction for a range of operations including land preparation,
seeding, weeding, and lifting of groundnuts. Although efforts of the
extension services to promote ox-plowing in The Gambia undoubtedly served
to sensitize farmers to the merits of animal traction, the phenomenal spread
ofequine traction has occurred largely through farmer initiative. Donkeys and
horses outnumber draft oxen in URD and MID-N by more than two to one
(Sumberg and Gilbert, 1988).
In addition to adopting animal traction, several farmers have adapted
externally introduced implements and tillage systems to the local environ-
ment. In Giroba Kunda the two most commonly reported modifications were
reduction of the blade area on single moldboard plows for compatibility with
donkey draft, and removal of the back two shoes on weeders along with
construction of one large front shoe for more rapid weeding. A few area
blacksmiths also construct implements of prices significantly below those of
implements sold by government-supported cooperatives.
A number of farmers in both areas have adopted a care, or square, pattern
of spacing for coarse grains, which makes it possible to fully mechanize
weeding operations through cross-row cultivation. This innovation is reported
to have first appeared in the area about three to five years ago and has diffused
widely in Giroba Kunda DEC. The practice is present, but less common, in
Kuntaur, where farmers not adopting the practice reported that the labor
saved in weeding barely compensated for the additional labor and draft
requirements for planting. This difference might reflect a greater importance
attached to planting labor time in Kuntaur, which has lower rainfall and a
shorter growing season than Giroba Kunda.
Another recently adopted method of weed control in each of the villages
surveyed in Giroba Kunda DEC was the use of herbicides. Herbicides have
not been promoted in The Gambia since they were felt to be beyond the means
of most farmers. However, a few relatively well-to-do farmers have acquired
the sprayers and chemicals from Senegal.
In the Kuntaur DEC, the most common implement package is a weeder/
seeder combination, and the most common draft source is the donkey. In
terms of tillage, a number of farmers direct-seed a substantial portion of their

Journal for Farming Systems Research-Extension


fields and then mechanically weed with a weeder (sine hoe/occidental hoe).
Farmers have made a number of modifications to the weeder to increase its
effectiveness under this system, including increasing the area of the weeder's
front shoe to allow wider between-row weeding, and increasing the height of
the hoes on certain types of weeders to prevent weeds from jamming in the
carriage. A number of modifications have also been made to enable the weeder
to create ridges earthingng up") after seeding. In addition, farmers in the area
have access to an active implement market in central Senegal.
Perhaps the most significant change in The Gambia has been the decline in
rainfall. In Basse, adjoining Giroba Kunda, the length of the rainy season has
decreased by an average of 14 days in the past 25 years, and the total
precipitation during the "humid period" has decreased from an average of
1,039 mm (1950-65) to 765 mm (1974-87) (Wright, 1988). The newrainfall
pattern has a bimodal distribution, with an increased probability of mid-
season drought.
The rainfall in Kuntaur is significantly lower despite the fact that it is less
than 150 km to the west and slightly to the north of Giroba Kunda. Although
comparable data for 1950-65 is not available, between 1974-87, rainfall
during the "humid period" averaged only 455 mm with a growing season of
85 days, and farmers indicate this is substantially below rainfall levels twenty
years ago (Wright, 1988). Mid-season droughts are even more pronounced
and can have profound effects on crop development.
With changes in the rainfall pattern, farmers now plant all crops as soon as
possible after the rains begin. The closer planting dates, and thus weedings,
accentuated labor bottlenecks. This is especially true in Kuntaur where
farmers have had to forego land preparation for direct seeding. In addition,
many farmers have adopted tillage techniques such as earthingg up" to
conserve soil moisture.
With the decreased growing season, several crop varieties are no longer well
adapted to the environment and farmers have been forced to actively exper-
iment with new, early maturing varieties. Farmers in Giroba Kunda are
experimenting with three varieties of groundnuts, two of which have come
from Senegal and have been diffused from village to village among farmers.
The third variety was given to farmers by an extension agent from Action Aid,
a nongovernmental organization (NGO) operating in the area. Two of the
varieties were described as attractive because they were early maturing while
the third was described as higher yielding than the standard 120-day variety
used by farmers.

Vol. 1, No. 2, 1990


Giroba Kunda farmers also mentioned experimenting with cereal varieties
including sorghum, findo (Digitaria exilis), maize, and rice. The sorghum,
obtained by farmers from southern Senegal, was described as early maturing
and high yielding, but susceptible to pest attack. The findo variety, brought
from Mali by a farmer, was described as early maturing. Farmers obtained
maize varieties from Senegal, the Department of Agriculture, and an area
NGO. All the varieties mentioned, however, were no longer available due to
poor germination or pest attack during the first several years of testing. Rice
varieties, like maize, had been obtained from several different sources and
tested by farmers in specific ecologies. A number of rice varieties have been
adopted from these tests. Finally, Catholic Relief Services, an NGO, recently
has successfully promoted sesame, and many farmers are incorporating it into
their farming system.
In Kuntaur, farmers consider early maturity by far the most important
criterion in variety testing. Two early maturing varieties of groundnuts have
been tested. Both were obtained from farmers in Senegal and one has been
widely adopted by farmers in the area. For cereals, one farmer reported testing
an improved variety of early millet obtained in Senegal, which failed because
of late planting. Two varieties of maize were tested in Kuntaur: a medium-
duration variety from the Senegalese Department of Agriculture and a long-
duration variety from the Gambian research and extension services. Farmers
reported growing the medium-duration variety because it gave high yields,
along with a short-duration variety to assure some yields in low-rainfall years.
Because of inadequate rainfall, few farmers were willing to grow the long-
duration variety despite its recognized potential for higher yields. For rice,
women in two villages reported testing new varieties. One was obtained from
an area NGO (Action Aid) and adopted by all the women in the village. The
other variety (early maturing and yielding well in either flooded or non-
flooded conditions) was obtained from the area district agricultural coordi-
nator, and was adopted by a number of women in that village. Finally, sesame
is being promoted by Catholic Relief Services in Kuntaur, but adoption of the
crop is lower than in the Giroba Kunda area.
Beyond experimenting with different varieties, farmers in both areas
continue to adjust cropping patterns in light of changing circumstances. The
decline in rainfall has led to a shift from sorghum to shorter-duration millets
in MID where Kuntaur is located. This may also represent a response to the
declining soil fertility associated with the reduction in fallowing because early
millet performs relatively well under low-fertility conditions compared to

Journal for Farming Systems Research-Extension


other cereals. By contrast, in URD where Giroba Kunda is located, early millet
is almost absent, partly because higher rainfall favors the longer-cycle cereals,
and pests, particularly birds, pose serious problems for unprotected early
maturing crops.
In conclusion, the study results illustrate the ongoing process of informal
experimentation by farmers in response to environmental challenges and
opportunities. The examples highlight several points:

1. Agricultural technologies exist in a very dynamic environment,
2. Farmers are responding to changes in their environment by rapidly
experimenting with and adopting new agricultural technologies,
3. Farmers obtain new agricultural technologies from a number ofsources,
including other farmers in The Gambia and Senegal, NGOs, and government
agricultural services in both The Gambia and Senegal.

Finally, and perhaps of most importance for the research services, the
informal testing of technologies provides valuable insights into farmers'
concerns and the specific criteria they use in assessing possible improvement


Several factors influence the pace and nature of technological change among
Gambian farmers, including sociocultural factors, availability of inputs, gov-
ernment policies, and linkages with sources of innovations. This section
discusses the importance of each of these factors and suggests ways in which
Gambian research services and development agencies can foster experimenta-
tion by Gambian farmers.

Sociocultural Factors
A broad range of sociocultural factors affect the pace of technological
change among different groups ofGambian farmers. Prominent among these
are gender, ethnicity, and income levels.
Gender: In The Gambia, gender determines the division of agricultural
labor and crops grown. In addition, it affects access to new technologies.
However, gender divisions are not static and new technologies sometimes
serve to accelerate the breakdown of such divisions. For example, von Braun
and Webb (1989) found the introduction of a centralized rice-irrigation

Vol. 1, No. 2, 1990


scheme in MID-South had the unforeseen consequence of turning rice
production from a private women's crop into a communal crop. As a
consequence, women expanded cultivation of private, upland cash crops.
Several development agencies are workingwith groups ofwomen, especially
in the areas ofanimal traction, horticulture, and rice production. These efforts
have been largely geared toward promotion of existing technologies, but
recently extended to include evaluations by women through on-farm, farmer-
managed trials. Women's groups, traditionally used in rice production, are
proving to be effective forums for organizing on-farm testing programs.
Farmer participation can substantially increased by holding regular meetings
with such groups to evaluate current tests and discuss future tests that might
meet women's demands for new technologies.
Ethnicity: Discussions of agricultural systems in The Gambia often refer
to different levels of agricultural technologies among ethnic groups. How-
ever, ethnicity per se is often less important than the range of factors associated
with a particular ethnic group, including language, linkages with sources of
innovations, and resource availability, particularly land. Long-term residents
of The Gambia (primarily Mandinkas and Jolas) are likely to have access to
more and better-quality land than the more recent migrants, which include a
preponderance ofWolofs and Fulas. On the other hand, family and linguistic
ties with Senegal may increase access to external innovations and partially
explain the relatively high levels of mechanization found in many Wolof
communities on the North Bank.
In contrast to other African countries, ethnic barriers in The Gambia do not
substantially inhibit the diffusion of technologies. This improves the potential
for farmer-to-farmer linkages and will be discussed further under "Linkages
with Sources of Innovations."
Income levels: Perhaps the single most important sociocultural factor
influencing the pace of technological change is income. Many innovations are
simply out of reach of poorer farm families. As with ethnicity, differences in
income levels reflect complex factors that collectively explain why some
farmers are more successful than others. These factors include status in the
community, resource availability (land, animal traction, labor), management
skills, and involvement in non-farm activities.
Technical change is often biased with regard to income distribution.
Technologies developed by the research services (domestic and external) tend
to favor the resource endowments of the more well-to-do members of the
community. Promotion efforts also gravitate toward the higher-income

Journal for Farming Systems Research-Extension


groups, which are most capable of adopting technical change.
Credit programs and subsidies can improve the access of low-resource
farmers to improved inputs, but experiences in The Gambia suggest that such
interventions do not lead to increases in earnings. By contrast, farmer
experimentation and farmer networks can be effective in identifying and
adapting innovations suitable for low-resource farmers. There are a number
of examples of innovations spreading without the support of credit programs
(for example, equine traction).

Availability of Inputs
The supply of agricultural inputs within The Gambia is constrained in part
by the lack of a fully developed private market (Langan, 1987). Many new
technologies require a reliable input supply to be feasible on a sustained basis.
Within The Gambia, however, the most commonly used agricultural inputs
are currently marketed through the Gambian Cooperative Union where lack
of timeliness of supply and responsiveness to market demands are sometimes
substantial factors in inhibiting use (von Braun & Puetz, 1987).
The current study found that Senegalese markets were a major source of
new technologies. Development agencies can facilitate farmers' procurement
of inputs by providing technical as well as market information on specific
inputs available in Senegal and by facilitating the establishment of private
input markets within the Gambia. Increased farmer-to-farmer contact can
also enhance knowledge of input availability and contribute to a more
effectively operating input market.

Government policies governing trade, price levels, exchange rates, and
subsidies can affect the attractiveness of different commodities and innovations.
For example, government policies on trade can restrict the availability of
imported inputs and effectively put certain technologies out of reach of most
farmers. Subsidies can increase the attractiveness of certain inputs, but deter
the development of free markets. In the past, availability has been more
important than price levels in influencing the use of agricultural inputs
(Langan, 1987). The policies associated with the Economic Recovery
Program (ERP), however, have let private market prices prevail. It is hoped
that improved availability, through privatization of input markets, will more
than offset the removal of subsidies and lead to a net increase in the use of
agricultural inputs.

Vol. 1, No. 2, 1990


In terms of external markets, Gambian farmers currently benefit from
relatively unrestricted access to Senegalese farmers and their input markets.
Government policies should support, not restrict, cross-border trade in
agricultural technologies.

Linkages with Sources of Innovation
The Gambia is well situated to benefit from technologies developed in
neighboring countries, particularly Senegal. In addition, The Gambia has
rainfall patterns similar to substantial tracts of other countries, which are
served by a network of regional, national, and international agricultural
research institutions. These institutions are valuable sources of information
on a range of potential improvements. Although research linkages with
development agencies and extension are important, enhancing the effectiveness
of these linkages at the national level should also be a priority for a small
research systems such as The Gambia's.
As the results indicate, farmers' linkages to other area farmers provide
sources of innovations. Most farmers have relatives in Senegal and other
countries of the region, which aids in the importation of innovations and
adaptations to The Gambia. Starkey (1986) and Sumberg and Gilbert (1988)
suggest such linkages with Senegal played an important role in the diffusion
of animal-traction technologies. The results of this study have documented
a number of other technical innovations and adaptations which have followed
this route from Senegal.
Development agencies might assist farmer-to-farmer linkages in a number
of ways, including exchanges between farmer groups in different parts of the
country. Further, several of the NGOs operating in The Gambia also have
activities in Senegal and other neighboring countries with whom exchanges
might be arranged, especially where farmers share a common language. Such
exchanges might include representatives of one village group spending an
extended period of time in another village to demonstrate a specific technique
that had been developed/adopted in the first village.
This section has discussed some of the critical factors in the process of
technical change. It has also made a number of suggestions as to how this
process can be enhanced. The final section incorporates these suggestions into
an approach for linking farmer groups, development agencies, and research
services to promote technical change in a manner that addresses the strengths
and constraints of each group.

Journal for Farming Systems Research-Extension



Evolution of On-Farm Research (OFR) in The Gambia
During the past decade, there have been a number of attempts to involve
farmers and extensionists in OFR in The Gambia. These include the
Demonstration Trials Program supported by the Food and Agriculture
Organization's Fertilizer Use Project, and two U.S. Agency for International
Development-supported projects: the Mixed Farming Project which focused
on maize, forage, and range management, and the Gambian Agricultural
Research and Diversification (GARD) Project which sought to improve on-
station and on-farm research within the agricultural research services. The
GARD Project supported the establishment of four Farming Systems
Research Pilot Areas in 1986 with two regional steering committees com-
posed of representatives of research and extension.
A review of these efforts through 1987 suggests that they made only limited
progress in developing a sustainable approach to OFR (Gilbert, Posner, and
Sumberg, 1989). The Pilot Areas program in 1986 made demands on the
researchers' time that were not compatible with their other research re-
sponsibilities, which included a range of on-station trials. Collaboration
between the research and extension services of the Department of Agriculture
within the context of the Demonstration Trials program was de-emphasized
beginning in the 1987 season when the research service opted to focus its off-
station activities in four locations or cluster sites.
The cluster-sites program marked an increase in the quality and quantity of
researcher involvement in OFR because the research programs assumed direct
responsibility for the execution as well as the planning of specific activities.
Most OFR at the clusters consisted of researcher-managed trials and surveys
focused on specific commodities and issues. There were some very productive
interactions with farmers and development agencies, notably in the cases of
the rice and groundnut research activities in the West and the fertility
management studies in the East (Mills and Senghore, 1989). This work
collectively contributed to a better understanding of farmer constraints and
objectives as well as laying the foundations for promotional programs now in
progress and further research on specific technologies that address farmers'
In spite of some accomplishments, there were a number of difficulties with

Vol. 1, No. 2, 1990


the cluster approach. The geographic and subject-matter coverage of OFRat
the clusters was limited by the resource and manpower constraints of the
research system as a whole. Problems included (1) overcommitment of
available manpower; (2) logical constraints, particularly transport; and (3)
lack of an incentive structure for off-station research. A disproportionate
amount of research was conducted by technical assistance staff supported by
GARD and the Overseas Development Administration (ODA). Interactions
with extension were limited and linkages were further weakened with the
creation of a separate Department of Agricultural Research (DAR) in mid-
1988. In the face of continuing difficulties in fulfilling their respective primary
tasks, efforts at collaboration between research and extension in the clusters
often aggravated tensions associated with resource constraints and differences
in objectives. By the end of 1988, the sustainability of the cluster concept, as
originally envisaged, was being increasingly questioned (Bojang, 1989).

The Farmer Innovation and Technology Testing (FITT) Program
In 1988-89, there was a major effort to improve collaboration between
research and development agencies. The collaboration stemmed from the
following factors. First, as mentioned above, some technologies were ready
for promotion or at least on-farm, pre-promotional testing. Second, develop-
ment agencies, including NGOs and extension, were interested in trying
innovations and, in a few instances, had initiated their own testing programs.
Third, the newly-formed National Agricultural Research Board (NARB) and
donor agencies supporting the research services called for an increased flow of
innovations to farmers and greater collaboration with development agencies
(NARB, 1989).
NGOs in particular were receptive to expanded collaboration with research
in OFR. NGO operational philosophies generally favored farmer participa-
tory approaches to many activities. In addition, most NGOs operate with
farmer groups which, in varying degrees, are the key decision-making and
implementing bodies at the village level.
In an attempt to involve farmers, development agencies, and the research
devices in a program of technology testing, the Cropping Systems/Resource
Management (CSRM) program of DAR, in collaboration with the Research
Extension Liaison Unit (RELU), launched the Farmer Innovation and
Technology Testing (FITT) program in April and May of 1989. A key feature
was the involvement of existing farmer groups, which had been formed by
extension and NGOs for promotion and development purposes, in the testing

Journal for Farming Systems Research-Extension


of innovations. The concept was stimulated in part by the experiences of the
Agricultural Technology Improvement Project (ATIP) in Botswana, which
has been using farmer groups for technology testing for several years (Nor-
man, et al., 1988), and in part by the 1988-89 CSRM study of farmer
innovations, adaptations, and adoptions of agricultural technologies. FITT is
designed to enhance the effectiveness of farmer efforts by making available,
through the research services, a wider range of improved practices to a network
of farmer groups for on-farm testing.
FITT addresses past conflicts between the objectives of research and
development agencies in three ways. First, farmer groups and development
agencies decide what technologies they wish to test. The research services
offer suggestions in the form of a list of technologies for which information
and inputs are available. Second, farmer modifications and innovations in
relation to the technologies will be encouraged. Finally, the role of the
research services will be to provide information, associated inputs, and
assistance in the design, monitoring, and evaluation of results. Research per
se is not the primary objective of the proposed activity; rather, it is to assist
farmers and development agencies in evaluating potential technologies.
Development agencies were contacted by representatives of the research
programs and the Research/Extension Liaison Unit to determine their
interest in participating in the FITT program. Discussions with individual
development agencies focused on the following issues:

1. Appropriate objectives for both research and development agencies in
an on-farm testing program,
2. Existence and nature of farmer groups that might participate in the
innovation testing,
3. Innovations that seem most appropriate for these groups and agencies,
4. Availability of development agency staff who can assume responsibility
for working with the farmer groups in technology testing,
5. Training and orientation programs for development agency staff and
farmer groups,
6. Resource requirements and possible sources of support for the farmer
testing program.

Following the initial set of discussions, a workshop was held and detailed
plans were developed with participating agencies. Based on the recommen-
dations of the research services, individual development agencies met with the
different technical scientists of the research services and developed lists of

Vol. 1, No. 2, 1990


technologies to present to their farmer groups for possible on-farm testing.
From those lists, the development agencies and farmer groups met and made
the final choice of technologies. Table 1 gives the technologies chosen by each
development agency.
The designs of all tests were kept very simple so they could be easily
implemented and managed by farmers. Intercropping tests were implement-
ed on an operational portion of farmers' fields, while variety trials were
superimposed on plots within fields. Tests of fruit trees provided for their
gardens by the research services.
Finally, monthly meetings were scheduled between farmer groups, de-
velopment agencies, and research, to discuss:

1. Progress in implementing on-farm tests,
2. Evaluation of performance of technologies in tests,
3. The types of technologies to be tested in coming years.

In addition, visits between farmer groups in different parts of the country,
and possibly in Senegal, were planned to exchange ideas on new innovations.
During the cropping season, it is important that the research services take
an active role in the monthly meetings and assist in the implementation and
monitoring of the tests. Program leaders will be responsible for research
participation in tests emanating from their programs. The development
agencies and their farmer groups, however, will take primary responsibilities
for the field activities. At the end of the season, the research services will assist
development agencies in evaluation and interpretation of results, and an
evaluation workshop will take place to review these results and plan for the
coming season.

Issues and Prospects for Institutionalizing FIIT
The FITT program is in its first year of operation, and it is clearly too soon
to pass judgment on its utility and sustainability. FITT was designed to reflect
the objectives and resource constraints of both the research services and the
development agencies and, thus, hopefully incorporate past lessons. Following
are some of the most important considerations: (1) for a small research system,
active collaboration with development agencies in OFR is possibly the best
hope of finding a sustainable model; (2) linkages with farmer efforts in
experimentation can substantially increase the scope and effectiveness of OFR
activities. The experience with the clusters in 1987 and 1988 demonstrated
that research can operate alone, but only with levels of resources and

Journal for Farming Systems Research-Extension


Table 1: Technologies to be Tested

Development Number and location Technologies
agency of farmer groups tested

Action Aid




Department of

Freedom From

Good Seed


Three groups in MID

One group in Western

One group each in:
URD, MID, and
Western Division

Five groups in URD

Demonstrations blocks
in MID

One group in L.R.

One group in L.R.

One group in N.B.

Rice hand-drawn weeders

Onion varieties and storage
Fertilizer on bitter tomatoes
Live fencing with lime trees
Staggered vegetable planting

Seed dressing on maize
Sesame varieties
Cassava varieties
Fruit tree varieties

Maize varieties
Sorghum varieties

Cowpea varieties
Sorghum varieties

Findo seed
Papaya varieties
Live fencing with lime trees

Findo varieties
Papaya varieties
Cowpea varieties
Cassava varieties

Grafted mangoes
Citrus rootstock

Vol. 1, No. 2, 1990


manpower that are not sustainable in the Gambian context. Collaboration
with farmer groups and development agencies in OFR can enhance capacity,
but this requires a mutual understanding of the objectives of all parties.
While there is general support for OFR and collaboration, FITT has
generated considerable debate among researchers and extensionists. Some of
the issues are specific to The Gambia and FITT, but many are familiar themes
in discussions of OFR. The issues include the following:
1. Is FITT new?: Extension personnel in particular feel with some
justification that FITT is not new. The Demonstration Trials program
contained many of the same features, yet the research service reduced its
involvement in that program in 1987. The extension service remains
interested in collaboration with research in OFR, specifically as a possible
component in the Block Demonstrations Program, which was launched in
1988 with support from the FAO Fertilizer Use Project. At the same time
extensionists are understandably wary of researchers who come bearing OFR
in new acronyms.
FITT is a new dimension to existing farmer-group activities for a number
of NGOs. Further, the emphasis on farmer participation, which is given
specific recognition in the title, distinguishes current activities from most
other OFR efforts where researchers have tended to play the leading role.
Making substantive farmer participation a reality will require a conscientious
effort by all concerned during the initial years. Some possibilities in this regard
are discussed in the section "Fostering Farmer Participation."
2. Direction ofFITT: An explicit objective ofFITT is to give farmers more
control of OFR. As farmer groups become familiar with FITT, it is hoped they
will increasingly take the initiative in seeking assistance from development
agencies and research programs to address their major constraints. Simulta-
neously, researchers and extensionists must accept farmers as valuable col-
laborators in technology development.
Researchers must expect and be able to accommodate deviations from the
standard protocols. This is not easy and there is a danger that researchers
interest in the program will wane where they perceive the research content to
be minimal. On the other hand, it is clearly not desirable for researchers to
assume a dominant role in the management of the on-farm trials. Achieving
a balance that will satisfy the objectives/needs of all concerned-development
agencies, researchers, and farmers-will not be easy.
3. Allocation of resources within the research services: There is continuing
concern among some researchers that FITT will excessively divert resources,

Journal for Farming Systems Research-Extension


specifically researcher time, away from on-station research and other forms of
OFR such as researcher-managed trials. A number of the technologies that
development agencies and farmer groups are requesting have not been the
focus of DAR research to date, and a special effort will have to be made simply
to locate information on these subjects. In short, FITT makes the hard choices
on the allocation of research resources even more difficult. The NARB
research-policy statement has emphasized linkages with development agen-
cies and farmers, and the FITT program is well suited to this task (NARB,
1989). At the same time, FITT requires a functioning research service to
supply it with a continuing flow of innovations for testing. In the process of
screening these innovations, individual research programs will continue to
exercise their discretion in determining the approaches utilized for a specific
commodity or issue.
4. Higher risk of failure: As FITT tries to accelerate the transfer to
technology, the risks associated with the on-farm tests increase. Researchers
are understandably concerned since their reputations rest on their ability to
make valid assessments of new technologies. Ideally, there would be enough
capacity to adequately screen a broad range of innovations before making
them available to extension and farmers. However, a small research system
such as The Gambia's is unlikely to have such a capacity in the foreseeable
future and must look for ways to shorten this process while preserving the
integrity of the research system. Many of the innovations that go out will not
be accepted by farmers, and it is important that all parties understand the
nature of the risks from the outset.
5. Fostering farmer participation: The early stages of FITT make it clear
that a determined effort will have to be made to substantially increase farmer
participation in the identification and testing of technologies. Although
virtually all development agencies favor participatory approaches in theory,
most of the initial innovations to be tested were selected by the development
agencies from proposals put forward by the research programs.
Among the specific mechanisms proposed or in use are (1) periodic
meetings of farmer groups with researchers and development agency staff; (2)
exchange of visits between farmer groups, including groups in neighboring
countries; (3) providing farmers with choices of technologies and being
responsive to their requests for information, even where such information may
not be readily available; (4) requiring researchers and development agency
staff to include farmers' assessments of technologies in their research/
development activities; and (5) giving special recognition to individuals and

Vol. 1, No. 2, 1990


organizations with outstanding performance in this area. The meetings with
the farmer groups are especially critical (Biggs, 1989; Chambers and Jiggins,
As FITT evolves, farmer groups should take larger roles in determining
which technologies are to be tested. The increased contact between the
groups, research, and the development agencies should enable farmers to
more clearly articulate their major constraints and what types of technologies
might mitigate these constraints. As noted earlier, success will depend in part
on the willingness of researchers in particular to allow farmer groups to
determine the direction of their testing.
6. Sustainability: The generally positive response to FITT by the develop-
ment agencies, particularly the NGOs, is encouraging. The NGOs collectively
have very limited research capacity at present. If FITT is successful, however,
they are likely to progressively expand and internalize OFR activities. In the
process, they will draw primarily upon the research and extension services for
additional manpower. The further erosion of the capacities of these services
will likely mitigate against sustaining OFR in any form in government
The Department of Agricultural Research might examine ways in which
researchers can assist development agencies and receive benefits in the form
ofconsultancy fees and expenses while still remaining part of the government
service. There are precedents for such arrangements both in The Gambia and
elsewhere. The dangers are obvious-the core responsibilities of researchers
can be neglected in the face of strong incentives to work for outside agencies.
The feasibility of this approach depends on the extent to which the research
services can regulate themselves in limiting the amount of time individual
researchers devote to outside work and making it conditional upon satisfac-
tory completion of their work programs. Simultaneously, the research services
must continue to improve working conditions and the rewards associated with
the performance of those core responsibilities. Progress has been made in this
direction during the past few years with improvements in compensation and
the agricultural-research management system, but there is still a long way to
go before the research service can arrest the continuing attrition of its most
qualified staff.

Journal for Farming Systems Research-Extension



Dramatic changes have taken place in the farming systems of The Gambia in
the past few decades in which farmer identification, testing, and dissemination
of technologies have played major roles. A review of the factors affecting
technological change suggests several ways that farmer participation in
technology development can be enhanced. The research services, in collab-
oration with development agencies, have initiated a program of testing
innovations with existing farmer groups, which will hopefully prove to be a
workable model for OFR. Difficult decisions are needed on the allocation of
resources within the research system to simultaneously allow researchers time
for participation in this program while increasing the flow of innovations and
improving the credibility of the service. Improvements in the conditions of
service and incentives are required to attract and retain research staff in the face
of increasing demand for research manpower by NGOs. Thus, sustaining
OFR in any form in a small research system such as The Gambia's is
inextricably interwoven with the viability of the entire system.


Biggs,S.D. 1989. Resource-poorfarmerparticipationinresearch:Asynthesis ofexperiences
from nine national agricultural research systems. OFCOR Comparative Study Paper
No. 3. The Hague: ISNAR.
Bojang, M. 1989. Research clusters sites: Performance and options for the future. The
Gambia: Department of Agricultural Research.
Boughton, D., et al. 1988. A rapid village appraisal offarmersoilfertility management
strategies in the Kuntaur and Giroba Kunda cluster areas. Gambian Agricultural
Working Paper No. 4. Yundum, The Gambia.
Chambers, R., and J. Jiggins. 1986. Agricultural research for resource poor farmers: A
parsimonious paradigm. Discussion paper 220. Brighton, U.K.: Institute of Devel-
opment Studies.
Chambers, R, A. Pacey, and L.A. Thrupp. 1989. Farmer first. London: Intermediate
Technology Publications.
Colvin, L.G. 1981. The Gambia. In L.G. Colvin, et al., eds. The uprooted ofthe Western
Sahel: Migrants quest for cash in the Senegambia. New York: Praeger.
Farrington, J., and A. Martin. 1988. Farmer participation in agricultural research: A
review of concepts and practices. London: Overseas Development Institute.
Gilbert, E., and M.S. Sompo-Ceesay. 1988. Dealing with the size constraint: Agricul-
tural research management in small developing countries. Paper presented at
ISNAR/Rutgers Workshop, July 1988.
Gilbert, E., J. Posner, and J. Sumberg. 1989. Farming systems research within a small
research system: A search for appropriate models. Agricultural Systems. In press.

Vol. 1, No. 2, 1990


Johnson, A.W. 1972. Individuality and experimentation in traditional agriculture.
Human Ecology 1: 149-159.
Langan, G.E. 1987. An assessment of agricultural input marketing in The Gambia.
Consultancy report. Banjul: USAID.
Mann, D.R. 1975. The Gambia: Land and vegetation degradation survey: The need for
land reclamation by comprehensive ecological methods. Cape St. Mary, The Gambia:
Department of Agriculture.
McIntire, J. 1985. Constraints to fertilizer use in sub-Saharan Africa. In A. Uzzo
Mokwunye, and P. Vlek, eds., Management of nitrogen and phosphorus fertilizers in
sub-Saharan Africa. Boston: Martinus NijhoffPublishers.
Merrill-Sands, D., P. Ewell, S. Biggs, and J. McAllister. 1989. Issues in institutionalizing
on-farm client-oriented research: A review of experiences from nine national agricul-
tural research systems. QuarterlyJournal ofInternational Agriculture 28(3/4): 279-
Mills, B., and T. Senghore. 1989. The cost effectiveness of fertilizer on manured and
non-manured fields. Gambia agricultural research paper. The Gambia: Department
of Agricultural Research. In press.
National Agricultural Research Board (NARB). 1989. Agricultural research policy for
The Gambia. Government of The Gambia: NARB.
Norman, D.W., D. Baker, G. Heinrich, and F. Worman. 1988. Technology develop-
ment and farmer groups: Experiences from Botswana. Experimental Agriculture
24(3): 321-331.
Richards, P. 1985. Indigenous agricultural revolution: Ecology and food production in
West Africa. London: Hutchinson.
Richards, P. 1986. Coping with hunger: Hazard and experiment in and African rice
farming system. London: Alien & Unwin.
Starkey, P. 1986. Strengthening animal traction research and development in The
Gambia through networking. Consultancy report 12. Banjul: Gambian Agricultural
Research and Diversification Project.
Sumberg, J., and E. Gilbert. 1988. Draft animals and crop production in The Gambia.
Department of Livestock Services, Abulo, The Gambia.
Vermeer, D.E. 1979. The tradition of experimentation in Swidden Agriculture among
the Tiv of Nigeria. In J.M. Frazier and B.J. Epstein, eds., Appliedgeography con-
ferences, Vol. 2. New York: SUNY-Binghamton.
von Braun, J., and D. Puetz. 1987. An African fertilizer crisis: Origin and economic
effects. Food Policy 12(4): 337-348.
von Braun, J., and P. Webb. 1989. The impact of new crop technology on the
agricultural division of labor in a West African setting. Economic Development and
Cultural Change 37(3): 513-534.
Wright, J. 1988. Evaluation of daily rainfall at the DARresearch sites. Paperpresented
to the Agricultural Research Advisory Board, April 1987, Cape St. Mary, The Gambia.

Journal for Farming Systems Research-Extension

Development and Testing of
Integrative Methods to Assess
Relationships between Garden Production
and Nutrient Consumption
by Low-Income Families'

Ingolf Gruen, Michel Beck, John S. Caldwell, and Marilyn S. Prehm2

An implicit goal of Farming Systems Research-Extension (FSRE) projects is
improving the well-being ofsmall-farm households, with anticipated benefits
in food consumption and nutrition for the household members (Shaner,
Philipp, and Schmehl, 1982). Examples of projects that explicitly include
improved nutrition in their goals, however, are rather limited (Frankenberger,
Perguin, and H'Malla, 1986; Prehm, 1987), although improved nutrition
may be considered a worthwhile goal (Frankenberger, 1985; Whelan, 1982;
Streeten, 1979). There may be several explanations for the paucity of
integrated production-nutrition studies employing FSRE methods: (1) FSRE
makes extensive use of informal data-collection techniques, derived in part
from anthropological methods, in setting priorities for on-farm experiments;
comparable informal techniques for assessing human nutrition are less well-
developed; (2) only few techniques that link fqod production and consumption
are time- and resource-efficient (Whelan, 1982).
The purpose of this project was to develop and test data collection and

1 The authors would like to acknowledge the helpful expertise of David J. Parrish and
Thomas Kalb III. The help of the interview pairs and Catherine Sherwood-Nelson and
James Nelson, in particular, is very much appreciated.
2 Department of Human Nutrition, Virginia Polytechnic Institute and State University;
Department of Soil Science, North Carolina State University; Department of
Horticulture and Department of Human Nutrition and Foods, Virginia Polytechnic
Institute, respectively.


analysis techniques for integrating food production and consumption in a fall
garden production project with low-income households. A method for
evaluating the nutritional value of garden crops was also developed and tested.


Data collection methods were tested by developing appropriate nutrition
measures to be used along with production measures in FSRE type of
diagnostic procedures (Caldwell and Walecka, 1987).

Identification of Problems and Intervention Selection
Eleven low-income families with young children were contacted through
a community action group assisting low-income families. Ten of these low-
income families were interviewed using the sondeo method of Hildebrand
(1982) with modifications (Patton, 1980; Caldwell, Rojas, and Neilan,
1984; Gaye, Jack, and Caldwell, 1988). The research team prepared a
guidesheet for each pair of interviewers on topic areas including garden
production; food consumption; dietary problems; and household resources,
constraints, and goals. Each sondeo interview pair, consisting of one pro-
duction science and one nutrition science student, was accompanied by a
local community action worker with whom the families were familiar.
After conducting the interviews, the research team compiled the data into
a summary chart listing the resources, goals, problems, and constraints of all
families. Results indicated that the major family goals were to increase real
income by decreasing food expenditures and to add variety to the diet. The
results of the sondeosled the team to propose a fall garden intervention to seven
of the ten families. All families had previous summer garden experience,
although none had planted fall gardens. Four families agreed to participate
in the actual planting and growing of a fall garden.
For this farm-household-led trial design (Caldwell and Lightfoot, 1986),
18 cool-season vegetable crops were laid out, following standard designs for
horticultural experiments, in an incomplete, random block-design across five
locations (four household-managed gardens, one researcher-managed garden)
with two to eight replications (Steele and Torrie, 1980). Each family selected
6 to 12 vegetables out of the 18 vegetables provided by the research team. All
18 vegetables were planted in the researcher-managed garden. Plot size per
crop varied from 0.6m X 1.2m to 0.6m X 3.0m, depending on the land
available. The research team supplied the seeds or transplants, fertilizer, pest-

Journal for Farming Systems Research-Extension


control material, and technical support in the implementation of the trials.
Research team members helped families plan garden layouts and plant their fall
gardens between the last week of August and the first week of September

Data Collection
Both consumption and production data were gathered. Consumption data
were obtained by modified food-recall interviews during two periods: (1) a
no-garden period after summer gardens had stopped producing and before
the fall gardens started production, and (2) a fall garden production period.
The mother in each household listed the amounts of each type of food
consumed for the dinner meal for herself and one child. The pamphlet The
Four Food Groups, Food for Fitness (Hertzler, 1986) and plastic food models
were used to help respondents estimate portion sizes. The dinner meal was
assessed, because the products of the garden were expected to be consumed
during this meal, since adults often ate lunch away from home, and children
had the benefit of school lunch programs. The number of interviews per
family ranged from four to eight due to differences in levels of cooperation.
Production data were obtained by measuring yields of the 18 vegetables
grown under family management by estimating the weight of the crops by
volume over a 60-day harvest period (October 19 until December 17). Yields
of the researcher-managed garden were determined by weighing the harvests.
The yields for each vegetable were then converted to the amount of nutrients
(vitamin A, vitamin C, calcium, and iron) supplied by the boiled, edible
portions of the vegetable using a nutrient-composition table (Splitstoesser,

Data Analysis
Consumption data. The food-consumption information collected during
recall interviews was converted to the percentages of the Recommended
Dietary Allowances (RDA) (Food and Nutrition Board, 1980; Barton, 1987)
for vitamin A, vitamin C, calcium, and iron. A previous study in the same area
had identified these nutrients as potentially limiting (Hertzler, Caldwell, and
Teo, 1987). The percent RDA of these nutrients was averaged for each meal
and each family. The dietary intake of the four nutrients (expressed as percent
RDA for each family) was then compared for the two different periods using
the General Linear Models program of the Statistical Analysis System (SAS)
(SAS Institute, 1985).

Vol. 1, No. 2, 1990


Production data. Garden total nutrient yield (GTNY) was calculated as a
sum of the nutrient yields of the individual vegetables for each of the four
nutrients. Since the families varied in number of members, a family RDA was
determined for each nutrient by first converting RDA requirements of
individual members of different ages and genders into male-adult-RDA
equivalents. For example: the RDA for iron for a female child, age 11 to 14
equals 1.8 male adult equivalents, because the RDA for iron for an adult male
is 10 mg, but the RDA for iron for a female, age 11 to 14, is 18 mg. Then the
equivalents for each individual member of the family were added together.
The GTNY's were divided by this family measure and the number of harvest
days (60), and multiplied by 100, to obtain family adjusted garden nutrient
yields (FAGNY). FAGNY values express the percent RDA of selected
nutrients each family's garden supplied each day during the 60-day harvest
Garden total nutrient yield [mg nutrient] x 100
60 [days] x Adult Male RDA [mg nutrient/day] x Family Measure

These values are given in Table 1. Correlations between FAGNY and
dietary intake were calculated. Garden size and garden total nutrient yield
(GTNY) were also correlated with the dietary intake data.
The garden production data were also expressed as Crop Total Nutrient
Yield (CTNY) for each crop, location, and the four nutrients. The CTNYs of
each crop in each garden were divided by the plot sizes and by the RDAs of
the four nutrients for a male adult to obtain Crop Nutrient Yield Equivalents

Table 1. Garden Yields for Four Selected Nutrients Adjusted to the Family
Composition (FAGNY) over 60-Day Fall Harvest Period

Family Vitamin A Vitamin C Calcium Iron
(% RDA) (% RDA) (% RDA) (% RDA)

A 30.2 12.8 1.7 0.9
B 66.2 9.8 5.3 2.7

C 92.6 44.3 7.1 2.8

D 8.7 5.8 0.6 0.4

Journal for Farming Systems Research-Extension


Crop Total Nutrient Yield [mg nutrient]
Plot Size [m2] x Adult Male RDA [mg nutrient/day]

Each CNYE value represented the number of days that 1 m2 of a given
experimental unit (plot with a given type of vegetable under a given household
level ofmanagement) could supply 100 percent of the RDA for an adult male.
The CNYEs for the four nutrients were then summed to obtain an Overall
Nutritional Completeness (ONC) value for each crop and location. Since
nutrient deficiencies were not a serious problem in the study area, the CNYEs
for vitamin A, vitamin C, calcium, and iron were summed without different
weighting to calculate the mean (Beck, 1988). ONC then is the average
number of days that a vegetable supplied 100 percent RDA. Nutrient
concentration for individual crops was calculated by multiplying the crop
ONC by 4, dividing it by the crop yield and multiplying by 1,000 to put the
nutrient concentration on the basis of days (male equivalent RDAs) per
kilogram per crop. CTNYs, CNYEs, and ONCs were analyzed using general
linear models and tested for differences using Duncan's multiple range test
(SAS, Institute 1985). The vegetables were ranked according to their ONC


Overall Nutritional Completeness (ONC)
Twelve of the 18 vegetables reached maturity and provided acceptable
yields. Differences among crops in ONC and edible portion yields were highly
significant (P .01) in the researcher-managed garden. Significant differences
among crops for Crop Nutrient Yield Equivalents (CNYEs) for vitamin A,
vitamin C, and calcium (p .01) were found as well as significant differences
among CNYE for iron (p < .05). The results of the family-managed gardens
are not shown here but were similar to those of the researcher-managed
garden. Mean separation by Duncan's multiple range test for the crops grown
in the researcher-managed garden are presented in Table 2. This table, in
which the vegetables were ranked by their ONC values, shows the mean
edible-portion yield per m2, the CNYE, and ONC values of the twelve
vegetables. Chinese cabbage was significantly higher in edible-portion yields
than all other vegetables grown. Pac choi and turnips had intermediate edible-
portion yields and the edible-portion yields of all other vegetables were

Vol. 1, No. 2, 1990


significantly lower. The test for differences between crops for ONC showed
results different from that of the edible-portion yield analyses. For example,
mustard has a significantly lower edible-portion yield than pac choi but was
not lower in its ONC value.

Table 2. Ranking of 18 Vegetables for Selected Nutrients according to Crop
Nutrient Yield (CNYE) and Overall Nutritional Completeness (ONC)

Number of days supplying 100% male MERDA
Vegetable (g/m2) ONC Vitamin A Vitamin C Calcium Iron

Chinese cabbage 1,731 a 24.71 a 16.09 a 4.41 ab 3.20 a 1.01 ab

Mustard 638 c 18.52 ab 11.11 ab 5.17 a 1.10 cd 1.14 a

Pac choi 1,126 b 16.08 abc 10.47 ab 2.87 abcd 2.08 b 0.66 abc

Turnip 1,179 b 15.72 abcd 7.56 bc 6.14 a 1.28 c 0.75 abc

Kale 385 c 13.58 bcd 8.53 be 3.96 abc 0.64 de 0.45 be

Chard 475 c 10.40 bcde 8.15 be 1.34 bed 0.45 e 0.46 bc

Broccoli 337 c 8.37 bcde 2.53 cd 5.21 a 0.37 e 0.26 c

Beet greens 257 c 5.42 cde 3.95 cd 0.65 bcd 0.32 e 0.50 bc

Collards 199 c 5.25 de 3.22 cd 1.52 bcd 0.38 e 0.13 c

Spinach 156 c 5.04 de 3.79 cd 0.75 cd 0.18 e 0.35 c

Radish 423 c 2.34 e Td 1.76 bcd 0.16 e 0.42 bc

Leaf lettuce 163 c 1.77 e 9.92 d 0.48 d 0.14 e 0.23 c

Note: Means separated by Duncan's Multiple Range Test, values accompanied by the
same letters are not significantly different at the 5% level.
No yield was obtained from: cauliflower, kohlrabi, brussels sprouts, cabbage, carrots, and
T = Trace.

Journal for Farming Systems Research-Extension


The comparison of the dietary intake between the garden and no garden
periods shows higher intakes of vitamin A (p .04) and vitamin C (p .07)
in the fall garden period. Calcium and iron intakes were not significantly
different in the two garden periods (p .13, p .15, respectively). The least-
squares means and standard errors of the means are shown in Table 3.
Dietary intake increased linearly with garden size during the fall garden
period, whereas there was no linear relationship in the no-garden period
(Figure 1). There was a significant linear relationship of garden size with
vitamin C (p .02) and calcium (p .01) intake, but not for vitamin A (p
.19) or iron (p 0.81) during the fall garden period (Figure 1). There was
no significant linear relationship between Garden Total Nutrient Yield
(GTNY) and nutrient intake.
When the garden nutrient yield adjusted for family size and composition
(FAGNY) was correlated to the nutrient intake of the family, vitamin C intake
showed a significant (p .02) positive linear relationship with FAGNY (Figure
2). Vitamin A, iron, and calcium showed similar but nonsignificant trends of
linear increase (p .17, p .19, p .23, respectively) (Figure 2) with in-
creasing FAGNY.

Table 3. Least Squares Mean of the Percentage (+ Standard Error)
of Recommended Dietary Allowance Consumed for Selected Nutrients
during Two Time Periods

% RDA consumed
No-garden Fall-garden Probability of
Nutrient period period differences

Calcium 27.6 9.9 45.2 10.0 0.13

Iron 20.7 6.8 39.9 6.9 0.15

Vitamin A 42.9 28.0 124.5 28.4 0.04

Vitamin C 48.5 63.8 170.7 64.7 0.07

Vol. 1, No. 2, 1990



The results of this particular study provide a preliminary test of selected data

collection and analysis methods. The methods are discussed in relation to this

particular case, and areas for further methodological work are suggested.

ONC can be used as a measure of nutrient yields to indicate the relative

nutritional efficiency of garden crops. The ranking of the vegetables by ONC

values is a combined indicator of yield per square meter and the nutrient

concentration of each crop for selected nutrients. Therefore, a high-yielding

crop with relatively poor nutritional value such as Chinese cabbage can

provide higher nutrient yield than a nutritionally better balanced crop such as
broccoli (see Table 2). Hence, under the specific growing conditions of the

study area, growing Chinese cabbage would be preferable to growing broc-

coli. However, the amount of a vegetable that has to be eaten to provide


/ 2 0.19
y 24.7 + (.2x

P < 0.53
S r =0.23
y 67.1 1.5x

* -..-.---- U
U ~

5 10 15 20 25 30


P2< 0.29
r =.50
y = 26.3 + 1.5x t


0 5



70 p2<0.82
r 0.04
60 y 29.9 + 0.6x
40 -
20 P2< 0.25
10 r 0.56
y = 10.1 + 0.7x

0 5 10 15 20 25 30

70 [% RDA]

60 0.98
60 r 0.98
y = 20.6 + 1.5x

40 .- p2< 0.76
S r2 = 0.06
30 y = 30.5 0.2x
0 *

10 15 20 25 30

Figure 1. Relationship of Garden Size (m2) and the Average Family Intake
of Vitamin A, Vitamin C, and Iron (%RDA)

Journal for Farming Systems Research-Extension

I7. Z m),


sufficient nutrients could limit the applicability of the ONC concept. Max-

imum family consumption could be used as an upper limit in calculating ONC

to prevent a crop with low nutrient concentration being ranked high in ONC

because of yield being greater than expected consumption. Particularly in

regions where the intake of specific nutrients is marginal it is important to

consider nutrient concentration of a crop and not simply gross yield. In this

study, none of the families reported problems using the Chinese cabbage for

their meals. When particular nutrients tend to be deficient in the diet, it is also

possible to give limiting nutrients a higher weight when calculating ONC.

The ONC concept may be useful for designing nutritionally efficient

gardens, particularly in marginal growing areas where some knowledge about

the performance of crops has been gathered previously. In addition, the

nutritional potential of gardens prior to harvest can be estimated by using the

ONC. Future research might test the method further by using ONC as the


p 2 0.17
r =0.6 o
y = 58.2 + 1.3x

80 "
70 1,2,
60 y


0 40 60 80 100

360 /p. 02
320 r = 0.96
280 V -30.3 + It.1x

0 10 20 30 40 50


' 0.

S 17.7x


0 .5 1 1.5 2 2.5 3

[x RDA]

p <0.23
r2 0.60
y = 33.8 + 3.1x

0 1 2 3 4 5 6 7 8

Figure 2. Relationship of Family-Adjusted Garden Nutrient Yield (FAGNY) (% RDA)
and the Family Intake of Vitamin A, Vitamin C, Calcium, and Iron (%RDA)

Vol. 1, No. 2, 1990

0 2



variable to be maximized under given garden-size constraints and minimum
and maximum consumption of crops based on dietary preferences. ONC
could be used to answer the question of what crop varieties should be planted
and in what combinations to maximize nutritional benefits.
Similar concepts have been developed and reported in the literature. The
Essential Factor of Nutritive Value of Rinno (1965), which was renamed
Average Nutritive Value by Grubben (1978), describes the overall nutritive
value of the crop by including protein, fiber, calcium, iron, carotene, and
vitamin C in an empirical formula. Even though this empirical formula
includes average requirements for these nutrients, it is inflexible when nutrient
requirements vary, and it does not take into account differences in yield. The
empirical formula also assumes that the nutrient contributions of vegetables
constitute a fixed proportion of total nutrient in the diet. For example, it is
assumed that 50 percent of the iron requirements are provided by vegetables.
However, the proportion of specific nutrients supplied by vegetables may vary
greatly by region.
Implicit nutritional benefits to participants may be one of the anticipated
outcomes of FSRE projects (Shaner, Philipp, and Schmehl, 1982). Since the
assessment of nutritional benefits, such as measurement of nutritional status,
is laborious, an improvement in nutrient intake may be considered an
indication that the intervention was beneficial for the participants' nutritional
well-being. However, it can only be hypothesized that an improved dietary
intake is a result of the intervention. In this project, the results of the dietary
intake comparison indicate a significantly higher intake for two nutrients
(vitamin A, vitamin C) during the fall garden period. These results are
comparable to the results found by Immink, Sanjur, and Colon, (1981) who
reported higher nutrient intakes for vitamin A and vitamin C but not for iron
or calcium ofhomemakers growing a garden. Solon et al. (1979) also reported
higher vitamin A intakes with garden production but did not find higher
vitamin A plasma levels. Such results are at least partial support for the positive
nutritional benefits of gardens to participants. However, in this particular
study, it is difficult with the lack ofa control group and the small sample size
to establish a causal relationship.
While all measures of garden production correlated positively with dietary
intake, the correlation between family-adjusted production (FAGNY) and
dietary intake was the strongest. While garden size is a crude indicator for
production, garden size was more strongly correlated with nutrient intake
than with total nutrient yields of the gardens. Using the family-adjusted

Journal for Farming Systems Research-Extension


production measure (FAGNY), the analyses showed a statistically.significant
relationship between production and dietary intake for vitamin C. The
significant increase of vitamin C consumption during fall garden harvest is
likely to be the result of the increased availability of fresh vegetables from the
For calcium and iron, neither a significant increase in dietary intake nor a
strong relationship between production and dietary intake was found. These
findings are not surprising since the majority of the crops grown were relatively
poor sources of both nutrients (see Table 1).
No statistically significant relationship was found between vitamin A
consumption and garden production, even though vitamin A consumption
was significantly higher during the fall garden harvest period. There are
several possible explanations for the poor correlation:
1. It is possible that the vitamin A production of the garden did not
influence the dietary intake of vitamin A by the families. Considering that the
garden supplied substantial amounts of vitamin A and contributed to the
families' diets (see Table 1), this simple explanation seems the least plausible.
2. One family had a very high vitamin A intake but low vitamin A garden
yields (see Figure 2 and Table 1). With a small sample size, a single divergent
data point could easily mask likely correlations. The divergence can likely be
due to additional sources of vitamin A not derived from the garden. Since no
economic analyses were undertaken, it is not possible to make reference to
differences in purchasing power among families, which might have influenced
the results.
3. The differences among the crops in yields for vitamin C (range = 3.93)
were much smaller than for vitamin A (range = 16.09) (Table 2). This suggests
the vitamin C yield of a garden was less dependent on the crops planted than
the vitamin A yield was. A small garden with high-yielding plants such as
Chinese cabbage and mustard might not yield much less vitamin A than a
much larger garden in which more but lower-yielding plants such as broccoli
or spinach are planted.
This study illustrates the use ofregression analyses to describe the relationship
between production and consumption. However, the relationships of garden
production and dietary consumption are not fully consistent with the results
of the dietary intake comparison. Vitamin A and vitamin C consumption is
significantly higher in the harvest period. However, the relationship between
production and consumption is significant only for vitamin C but not for
vitamin A. In this study, dietary intake increased significantly for only two of

Vol. 1, No. 2, 1990


the four nutrients studied, whereas other projects with garden interventions
show significant increases in the intake of all the nutrients considered in this
study (Gershon, 1985). It is important to note that the families involved in
the present study had adequate dietary intakes during the no-garden period.
Therefore, the gardens may have been budget gardens, only supplementing
purchased food supplies or replacing them with garden staples rather than
making major contributions to the total household food supply (Nifiez,
1985). Other studies have shown that growing a home vegetable garden can
decrease food expenditures rather than improve dietary intake (Utzinger and
Connolly, 1978).
The principal technique used for analyses in this study, linear regression,
can be useful in evaluating the impact of a garden on consumption: the
stronger the relationship (higher r2) the greater the expected impact. A weak
relationship implies either limited impact or that the garden produce replaces
purchased food rather than being used for additional intake. Using this
technique, one can describe the relationship of a garden intervention and
dietary intake. Taking family composition into account is important, because
the use of the crops and the distribution of the nutrients within the household
depend on the family structure, and nutrient requirements vary by gender
and age.
No reference values exist for correlations of garden yields and dietary
intakes. The observations for this study can be used as a baseline for future
studies with these or similar households in the area. With reference values
from well-controlled studies, which can establish relationships under more
controlled conditions, it would be possible to more clearly evaluate the impact
of garden interventions. Instead of assessing the dietary intake during both
periods, the pre-intervention and the intervention period, only the intake
during the intervention period would need to be assessed. The strength of the
relationship between production and consumption would then be used to
evaluate nutritional benefits of the garden. This would decrease project costs
and require less time and cooperation of the families in collecting dietary data.
However, it is not possible to distinguish between seasonal and garden
effects on dietary intake due to lack of a control group. Additional research
is needed to establish the appropriateness of regression analyses in assessing
the relationship between production and consumption and to further test the
usefulness of the technique for partially substituting dietaryintake assessments.
Future studies should use larger sample sizes and control for variables, such
as seasonality. An economic analysis might also be useful to better describe

Journal for Farming Systems Research-Extension


the household uses of the garden. This is of particular importance in cases
where no nutrient deficiencies are expected. The technique may be more
useful in developing countries where a garden is often the primary source of
food and nutrients (Nifiez, 1985), and not a strategy for economizing
(Utzinger and Connolly, 1978).


Barton, J. 1987. Recall; a food-intake nutritional analysis program for the IBM PC.
Virginia Cooperative Extension Service, Virginia Polytechnic Institute and State
University, Blacksburg, Virginia.
Beck, M.A. 1988. Integration of nutritional considerations in a fall garden project
through farming systems research and extension methodology in the New River
Valley, Virginia. BS, Honors Thesis, Virginia Polytechnic Institute & State Univer-
sity, Blacksburg, Virginia.
Caldwell, J.S., and C. Lightfoot. 1986. A network for methods of farmer-led systems
experimentation. FSSP Newsletter4:15-18.
Caldwell, J.S., M.H. Rojas, and A. Neilan. 1984. Interviewing farm families in
Southwest Virginia: A training (videotape) module in rapid rural appraisal. Virginia
Polytechnic Institute and State University, Blacksburg, Virginia.
Caldwell, J.S. and L. Walecka. 1987. Design techniques for on-farm experimentation.
FSRE training units: Volume II. Farming Systems Support Project, University of
Florida, Gainesville, Florida.
Food and Nutrition Board. 1980. Recommended dietary allowances, 9th ed. Washington,
D.C.: National Research Council, National Academy of Science.
Frankenberger, T.R. 1985. Adding a food consumption perspective to farming systems
research. Nutrition Economics Group, Office of International Cooperation and
Development, U.S.D.A., Washington, D.C. (Monograph).
Frankenberger, T.R.,B.Perguin, andF.M.H'Malla. 1986. Farmingsystemsresearchalong
the Senegal River Valley. Food consumption survey in Guidimaka, Gorgol, Brakna and
Trarza regions. Mauritania Agricultural Research Project II, College of Agriculture,
The University of Arizona, Tucson.
Gaye, G.O., I. Jack, and J.S. Caldwell. 1988. Use offarming systems research-extension
(FSRE) methods to identify horticultural research priorities in the Gambia, West
Africa. HortScience 23:21-25.
Gershon, J. 1985. Alleviating vitamin A problems with home gardens. Journal ofPlant
Foods 6:117-124.
Grubben, G. J. H. 1978. Tropical vegetables and their genetic resources. International
Board for Plant Genetic Resources, FAO-UN, Rome, Italy.
Hertzler, A.A. 1986. The four food groups, food for fitness. Virginia Cooperative Ex-
tension Service, Extension Division, Virginia Polytechnic Institute and State Univer-
sity, Blacksburg, Virginia, Publication 348-906.
Hertzler, A.A., J.S. Caldwell, and M.L.M Teo. 1987. Factors related to dietary status
of limited resource farm families: A case study. Journal ofRural Health 3(2):47-60.
Hildebrand, P. E. 1982. Appendix 5-Q. Summary of the sondeo methodology used by

Vol. 1, No. 2, 1990


ICTA. Pages 289-293 in W.W. Shaner, P.F. Philipp, and W.R. Schmehl, eds.,
Farming systems research and development: Guidelines for developing countries.
Boulder, Colorado: Westview Press.
Immink, M.D.C., D. Sanjur, and M. Colon. 1981. Home gardens and the energy and
nutrient intakes of women and preschoolers in rural Puerto Rico. Ecology ofFood and
Nutrition 11:191-199.
Nifiez, V. 1985. Introduction: Household garden and small-scale food production.
Food and Nutrition Bulletin 7(3):1-5.
Patton, M.Q. 1980. Qualitative evaluation methods. Beverly Hills, California: Sage
Prehm, M.S. 1987. Data analysis manual for food consumption/nutrition aspects of
rapid community assessment for planning procedure-Bicol Region Farming Systems
Research and Development Project. Philippines Department of Agriculture and U.S.
Agency for International Development Pili, Carmines Sur., The Philippines. Mim-
Rinno, G. 1965. Die Beurteilung des ernihrungsphysiologischen Wertes von Gemfise.
Arch. Gartenbau 13(5):415-429.
SAS Institute Inc. 1985. StatisticalAnalysisSystem. Cary, North Carolina: SAS Institute
Shaner, W.W., P.F. Philipp, and W.R. Schmehl. 1982. Farming systems research and
development. Boulder, Colorado: Westview Press.
Solon F., T.L. Fernandez, M.C. Latham, and M. Popkin. 1979. An evaluation of
strategies to control vitamin A deficiency in the Philippines. American Journal of
Clinical Nutrition 32:1445-1453.
Splittstoesser, W.E. 1984. Vegetable growing handbook, 2d ed. New York: The AVI
Publishing Co., Inc.
Steele, R.G., and J.H. Torrie. 1980. Principles and procedures ofstatistics, 2d ed. New
York: McGraw-Hill Book Company.
Streeten, P. 1979. From growth to basic needs. Finance and Development 16(3):28-31.
Utzinger, J.D., and H.E. Connolly. 1978. Economic value ofa home vegetable garden.
Plant Science 13(2):148-149.

Journal for Farming Systems Research-Extension

Farming Systems and Adoption of New
Agricultural Technologies:
An Economic Evaluation of New Sorghum
Cultivars in Southern Honduras'

Miguel A. IApez-Pereira, Timothy G. Baker, John H. Sanders,
and Dan H. Meckenstock

The Honduran government, with the cooperation of several international aid
organizations, has tried to improve the income and nutritional condition of
small farmers in the southern region of the country for many years. Attempts
are being made to bring several new agricultural technologies to small farmers
in the region. The most important of these technologies are new drought-
resistant, high-yielding crop cultivars; moderate levels of chemical and organic
fertilizer; and the use of improved hillside soil conservation techniques. The
Ministry of Natural Resources of Honduras (MNR), in collaboration with the
International Sorghum and Millet CRSP (INTSORMIL), has released three
new sorghum cultivars since 1983. These cultivars have shown good potential
for adoption since they have produced high grain and forage yields, resistance
to diseases, and good tortilla quality in experimental station trials and on-farm
demonstration plots during the last three years. (Meckenstock, G6mez, and
Palma, 1987; G6mez, et al., 1989).
The U.S. Agency for International Development in Honduras (AID/H)
has been working closely with the MNR to introduce new agricultural

1 Paper presented at the Ninth Annual Farming Systems Research-Extension Symposium,
University of Arkansas, Fayetteville, October 9-11, 1989. This research was conducted
as part of the dissertation requirements for the Ph.D. degree in agricultural economics,
Purdue University, and was partially funded by International Sorghum and Millet
(INTSORMIL) CRSP, USAID/HONDURAS, and the Ministry of Natural Resources
of Honduras.
2 Research Assistant, Associate Professor, and Professor, respectively, Department of
Agricultural Economics, Purdue University, and Principal Scientist INTSORMIL/
Honduras, Escuela Agricola Panamericana, Tegucigalpa, Honduras.


techniques to improve basic cereal yields and, at the same time, prevent further
erosion on the hillside soils of the south (MNR, 1986). To that effect, the
Natural Resource Management Project (NRMP) was created in 1982, funded
jointly by AID/H and the Honduran government through MNR. The main
objective of the project is to develop and release technologies aimed at
achieving a more efficient use of the land by small farmers. It promotes the use
of techniques to control the process of soil erosion and destruction of forests
and watersheds caused by traditional, usually inefficient methods of
Other Honduran government agencies, through policies and regulations,
are also trying to improve the income level of small farmers in Honduras.
Minimum commodity prices set by the Honduran Institute of Agricultural
Marketing (IHMA); limited credit for cereal production by the National Bank
of Agricultural Development (BANADESA); input price subsidies; and
technical assistance through extension agencies are some of the programs that
have been implemented.
Although the efforts to make these technologies and policies available to
small farmers during the last decade have required substantial amounts of
capital and human resources, their actual or potential impact on the income
of the farmers has not been measured. Therefore, there exists a need to
evaluate the impact of these technologies, as well as the relevant government
policies, on the income of poor, small farmers of the south. This study will
evaluate the potential impact on the income of small farmers in southern
Honduras, of the new technologies being introduced by MNR in collabora-
tion with AID/H and INTSORMIL, and the possible impact of several
government policies on the potential for adoption of these technologies.
An economic study of this kind is important in several ways. It can provide
valuable information to agricultural researchers, administrators, extension
agents, and policymakers. The availabilityofthis information can also contribute
to the improvement of small farmer income through the development of
research and extension programs and government policies that take into
account the conditions faced by the farmers and the factors that affect their
decision-making process. It could also suggest aspects of these programs that
should be emphasized or de-emphasized, and alternative ways to make them
more efficient in improving the livelihood of small farmers. The remainder of
the paper is organized as follows: first, the objectives of the study are stated.
Then, the southern region of Honduras and the general characteristics of the
environment that small farmers face there are described. The next section

Journal for Farming Systems Research-Extension


discusses the new agricultural technologies being promoted in the southern
region. The paper then presents the procedure followed to develop the
economic model and the data used. The results of the model under both
scenarios (with and without the new sorghums) are discussed next. Emphasis
is placed on the income and total cereal-production impact of the two
scenarios and the main constraints that influence the level of the new
technologies in the optimal crop combinations for the farmer. The last section
presents the conclusions derived from the analysis done so far. Policy
implications from these preliminary results are then drawn.

The general objective of this research is to determine the potential impact of
the new technologies introduced by MNR/INTSORMIL, and NRMP/AID,
and the effects ofseveral government agricultural policies, on the productivity
and income of small farmers in southern Honduras. The basic economic and
consumption situation of the small farmer under his/her traditional system of
cultivation will be compared to the scenario where the farmer has the new
technologies as alternatives in the set of crop combinations that can be chosen.
The specific objectives of the study are:

1. To gain a better understanding of the way small farmers in southern
Honduras make decisions regarding their choice of crops and other agricul-
tural activities, as well as, their risk attitude and its effect on crop choice; that
is, to understand their basic traditional farming systems.
2. To determine the potential effect of the new technologies being
introduced by MNR/INTSORMIL and NRMP/AID, on the level of grain
production and income of small farmers in southern Honduras.
3. To determine the effect of government agricultural policies, such as
credit and input/output subsidies, on the potential for adoption of the new
technologies and on small farmers' incomes.

For the first objective, the research approach will be to use a farm-level,
mathematical-programming model to characterize the decision-making process
of a representative small farmer in the region. The approach will determine the
optimal sequence and mix of activities that maximize the farmer's expected
utility of ending cash balances, or some other suitable objective, given
financial, market, labor, land, and minimum cereal-consumption constraints.
Then, to meet the second objective, the model will be expanded to include

Vol. 1, No. 2, 1990


the new sorghum cultivars developed by MNR/INTSORMIL as part of the
alternatives the farmer has available. The results of this second model should
indicate if the optimal portfolio of activities for the farmers includes the new
sorghum technologies. The second model will then be expanded further to
include the same activities but with the use of soil conservation structures on
the hillside lands. The potential income effect of these new technologies will
be determined by comparing the expected value of grain production for the
different models under each scenario. Sensitivity analysis will be used to
estimate how changes in some of the relevant agricultural policies, especially
input/output prices, affect the attractiveness of the new technologies, as well
as their impact on the income of the small farmers. This procedure will lead
to the third objective.
Adoption of the new technologies is hypothesized to increase farm
income. Factors affecting the adoption of the new technologies are expected
to include their level of profitability and risk as perceived by the farmers; the
availability of complementary inputs such as fertilizer, labor, and land; capital
and credit availability; and complementary government policies. Government
policies that facilitate farmers' access to inputs at low prices, and to markets
to sell their grain at better prices, are expected to provide incentives for
adoption. The proportion of new technologies in the optimal crop mix is
expected to be larger when these policies are in effect. The farmers' objectives
are expected to be characterized by the maximization of some utility function
of cash income and the avoidance of risk levels above some minimum, after a
minimum cereal-consumption level for the family has been met.


The area classified as southern Honduras by MNR comprises the departments
of Choluteca and Valle. The main city in the region is Choluteca with a
population of80,000 and located 75 miles south ofthe capital city, Tegucigalpa.
The region covers an area of 2,800 m2 and has a population of 520,000. It
presents a variety of climatic conditions, with altitudes from 0 to 1,600 m
above sea level and average annual rainfall from 1,000 to 2,900 mm. Most of
the rainfall, at least 90 percent in all weather stations considered, occurs
during the six-month rainy season that extends from May through October
(MNR, 1985).
As in the rest of the country, the main activity in the southern region is
agriculture. There are two distinct agricultural sectors. One is the large

Journal for Farming Systems Research-Extension


commercial farms that occupy the rich land of the valleys where sugarcane,
rice, cotton, and watermelons are produced. This agriculture is highly
intensive and mechanized. The other sector is composed of the small
subsistence farmers on the hillsides whose main activity is the production of
maize and sorghum, mostly for home consumption. More than 56 percent of
the land in southern Honduras is composed of mountains with moderate to
steep slopes, usually greater than 15 percent. These soils are usually thin and
susceptible to erosion from runoff and poor farming practices (Chemonics
International, 1985). Small farmers occupy the more accessible of these
slopes and the rest are covered with pine forests. In order to identify
the most important demographic and farming characteristics of the farmers in
the south, an extensive survey was carried out in late 1988 on a sample of
farmers living in the region. The most important results of the survey are
discussed below.

Demographic Characteristics of Small Farmers
The farmers in the sample were classified into large and small based on a
farm size of 5 ha. Small farmers are somewhat younger and less educated than
larger ones, although both groups present very low average-education levels.
The difference between the two groups is more significant in the family size
and farm size categories. Family size, and especially the number of male adults,
is greater in the large farmer group. All members of the household who are 12
years of age or older are considered adults. If the member is 12 years old or
older and still lives at home, the most common situation is that he/she has
either finished elementary school or is no longer attending school and helps
full-time with household chores (females), or with the farm activities (males).
Therefore, larger farmers have more family labor available for farm activities
than smaller ones. This result is consistent with studies that have shown a
direct relationship of farm and family size (The World Bank, 1984).
According to the size classification used here, large farms are on average
over six times larger than small ones (17.0 versus 2.6 ha). Most of the small
farm area is used to grow cereal crops (65 percent). In contrast, large farms use
only 30 percent of their area for cereal crops, reflecting the fact that these are
mostly cattle operations with most of the area used for forage crops and cattle
installations. The main animals owned by farmers in the south are chickens,
pigs, horses, and cattle. Chickens are raised mainly for home consumption of
eggs and meat. Pigs and cattle are the "emergency fund" of the farmer. They
serve as a source of wealth and cash for emergencies. Horses and donkeys are

Vol. 1, No. 2, 1990


used mainly for transportation. Oxen are not common among small farmers
on the hillsides.

Land Tenure, Land Values, and Farm Size
The 1974 agricultural census and results from the 1988 economic survey
indicate a majority of farms which are 5 ha or less in size. This small-farm
group, however, has decreased and larger farms have gained in numbers,
especially the very large ones. Small farms include most of the maize and
sorghum producers who grow cereals for home consumption. Medium-sized
farms also represent a significant number of the farms, and include most of the
permanent cropland (coffee and cashews) and some cattle ranches. The large-
farm category includes the cotton and sugarcane operations and the large
cattle ranches. Most of the farm operations in southern Honduras are owned
by the farmer. Only a few of the farmers interviewed declared that they rent
or co-own their land.
Flat land is worth twice as much as hillside land to the farmers in the sample
($681 versus $322 per ha). With regard to rent values, the average was $30.00
per ha, per year for hillside land. The average total cropland available for grain
crops to the small farmer is approximately 2.6 ha; 0.9 ha comes from rented
land and 1.7 ha is owned land. The rest of the owned land, 0.9 ha, is devoted
to some forage crops, a vegetable garden, the house compound, and some area
around the house for the farm animals.

Crop Patterns by Small Farmers
The crop season in southern Honduras is divided into two periods called
the primer (first) and the postrera (second) seasons. The crop schedule
follows the rainy season, which starts in early May and finishes in late October.
During the cropping years 1986-87 through 1988-89 farmers grew mainly a
maize/sorghum combination and maize monocrop during the first season
(FS). Maize/bean doublecrop and beans and traditional sorghum monocrops
are also common in the FS. Maize and beans are planted in early May and the
sorghum planting is done in late May. The beans are harvested in late July and
maize in earlyAugust during marked short dryperiod (canicula) in the middle
of the rainy season. Sorghum then develops and is harvested in December or
January. The maize harvest marks the end of the FS. The second crop season
(SS) starts after the canicula in late August when the maize and bean
monocrops and the maize/bean doublecrop are planted. Some sole sorghum
is also planted in September and harvested with the FS sorghum in December

Journal for Farming Systems Research-Extension


and January. Postrera beans are harvested in late November and maize in mid-
Therefore, the basic model will include the following traditional technologies
for each season: in the FS the combinations will be sole maize, sole beans,
maize/sorghum, and maize/bean doublecrops. For the SS the available
alternatives will be sole maize, sole beans, sole sorghum, and maize/bean
doublecrop. Other crops grown by the small farmers do not show significant
numbers to be included in the model. The sorghum monocrop grown in the
FS represents mostly the improved sorghums used by a few small farmers. This
alternative will be included in the second stage of the model when the new
sorghum cultivars (Surefo and Catracho) will be added to the available crop
alternatives in the FS and SS.
The discussion above also leads to a natural division of the planning period
for the model. The planning period will be assumed to start with the planting
decision for the FS at the onset of the rains on May 1. The first stage will end
on August 15 when the canicula has ended and the maize has been harvested.
The second stage will start immediately thereafter (August 16) with the
planting for the SS crops. The second stage ends on January 15 when the last
sorghum is harvested. The period from January 16 through April 30 is the
third stage and will involve mostly grain marketing, off-farm labor decisions,
and land preparation in April for the next cycle.

Distribution of Family Labor
Not surprisingly, the head of household (HH) provides most of the farm
labor. The head also supplies more off-farm labor than the rest of the
members. Other male adults also provide significant amounts of farm and off-
farm labor. Female members work very little on the farm and most of their
labor is HH related. There are 2.3 male adults per farm in the sample.
Therefore, it will be assumed that the HH provides six days of farm work per
week and the other 1.3 male members provide a total of three days. This makes
a total of nine man-days of farm work per week per farm during the FS and SS.
With regard to off-farm labor, the head of household works an average of 3.4
months and the other 1.3 members work 1.6 months of the year, for a total
of five months of off-farm labor supplied by the HH.
According to the farmers' responses in the survey and data from MNR, land
preparation and weeding activities account for most of the total time required
to produce the crops shown. More than 50 percent of total labor requirements
are for land preparation and weeding in all crops. Multiple crops require

Vol. 1, No. 2, 1990


significantly more time than sole crops. It is interesting to note the little use
of fertilizers by small farmers in southern Honduras. The percentage of
farmers who use this input is very low. Another interesting result is the use of
herbicides by a significant percentage of farmers, especially for the maize and
sorghum combinations. The results will be used to derive technical coefficients
for the model. Also the coefficients will be used to derive time requirements
for the new sorghum technologies.

Cash Income and Expenses
Off-farm labor provides more than 50 percent of the total household cash
income. Farm animal and cereal sales are the other two important sources of
cash income, with shares of 23 and 9.5 percent, respectively. The total cash
income is just over $1,061 per year which, if we consider the average family
size of 7.4, results in a per capital cash income of $143 per year. The value of
unsold production can be substantial, especially when we consider that most
of the farm harvest is set aside for the consumption of the family and animals.
With regard to cash expenses, the most important items are family-related
expenses, including food (other than cereal grains), clothing, medical, and
school expenses. Interestingly, farm animals and cereal purchases are also
important sources of cash expenses. This may be explained by the fact that
farmers sell a small portion of the cereal production after the harvest to obtain
some cash for other expenses. Household cereal stocks then get depleted
before the next harvest is out and the farmers are forced to purchase grains,
probably from cash obtained with off-farm labor. This situation usually results
in the farmer's selling grains at low prices when supplies are high, and buying
them back in the period before harvest when supplies are low and prices high.
An important implication for model formulation is the very small percentage
of cash expenses from hired labor and purchased inputs. These expenses
account for a combined 3.4 percent of the annual total cash expenses by the
small farmer. Moreover, an important portion ofpurchased inputs is accounted
for by seed, which in most cases is traditional seed. The other significant
expense is for urea and 12-24-12 fertilizer, herbicide, and insecticide for the
maize and sorghum crops. Hired labor is almost totally accounted for by some
land preparation and weeding in the first season.

Credit and Other Financial Constraints
Small hillside farmers start the FS with average cash holdings of $64.00,
which is used mostly for family related expenses. The only agricultural input

Journal for Farming Systems Research-Extension


needed to start the crop season, besides family labor, is seed. Most of the
farmers declared that they save their own seed from the previous harvest. If this
has not been possible, they purchase the seed as part of other grain purchases.
Most farmers declared that they do not have any debt, but almost half of
them stated that they could obtain an average of $463 in credit, mostly from
private lenders, who charged an interest of almost 5 percent per month. Only
four farmers thought they could obtain credit from the national agricultural
bank, although none of them had attempted to obtain that credit. Data also
show the low percentage of farmers who save any cash, and the insignificant
amount of those who did save. Farm assets, including land, were valued on
average at $2,982.

Sequential Decision Making by Small Farmers
Total average rainfall in southern Honduras is high and appears adequate
for the crops grown in the region. However, this high average rainfall is also
characterized by extreme seasonal variation, with short periods of heavy
rainfall and spells of low precipitation. This variation is more important if we
consider that most of their cropland consists of hillsides with little water-
retention capacity and, thus, highly susceptible to water erosion. In addition,
solar radiation is very intense year-round, and moisture evaporates very
quickly, making even short dry periods risky for crop survival. These charac-
teristics of the weather and topography of the south make it a risky environ-
ment for traditional cropping on the hillsides. However, the farmers have
adapted to this environment by rearranging their crop-management decisions
as the rains occur during the season.
Decisions on how much of each crop and which crops to plant during the
SS, for example, depend in part on how the rains have been in the FS. If the
rains have been especially bad (drought or excess rain), the maize crop is
usually entirely lost, and the farmers expect the rest of the year to be as bad.
Thus they may decide to plant only sorghum in the SS since it is more resistant
to bad weather. If the weather is favorable in the FS, the maize harvest is likely
to be good and large enough to provide for family consumption. The farmer
may then decide to plant relatively more beans than maize in the SS to obtain
some cash income selling the beans in the market.
Decisions to store grain, to feed the farm animals with part of the sorghum
harvest, and/or to sell grain in the market, also depend on how good the rain
in the two growing periods has been. The farmer may be able to sell some of
his sorghum harvest and obtain some cash income if the weather has been

Vol. 1, No. 2, 1990


favorable; or he may have to purchase some maize to provide enough for
minimum family consumption when the harvests have not been good. When
questioned about their grain-marketing and home-consumption decisions,
farmers stated that the FS harvest is used to replenish stocks for home and
animal consumption, and a good part of the SS harvest is sold in the market
to obtain some cash income. It all depends on the rains, however, since "...if
the FS harvest is too low or zero, we need to work or sell the animals to
purchase grain and start the SS in September. If the FS harvest is good, we can
even sell some maize or beans in September or October to buy other food
items and clothing...."
The sequential character of the farmer's decisions within a given year, as
well as the stochastic nature of these decisions, can be readily seen. The model
to be used in representing the farmer's decision-making process and adaptations
to the resulting weather conditions as they occur, has to capture all these
characteristics. Specifically, the model should (a) capture the sequential
nature of the decisions in a given planning period; (b) reflect the fact that
several of the events affecting the farmer's decisions are stochastic and, as such,
have more than one possible result; and (c) allow for adaptive changes of the
farmer's earlier decisions as previously uncertain events unfold and become
part of past information used to revise current decisions.


The poor soil fertility of the laderas in the south, combined with insufficient
or zero use of chemical and organic fertilizer, forces the farmer to depend on
slash-and-burn cultivation. This system causes significant soil erosion prob-
lems as portions of the forest land have to be cleared and incorporated into the
system each year. The erosion problems on hillside soils can be greatly reduced
with the use of simple soil-management techniques such as the construction
of ditches and tree barriers. The use of new cereal-crop cultivars in appropriate
rotations such as maize or sorghum followed by beans or another legume
would also improve yields and soil characteristics. It is these two types of
technologies, new sorghum cultivars and water retention on hillside soils, that
are the subject of this study.

New Sorghum Cultivars
MNR/INTSORMIL have released several improved sorghum cultivars

Journal for Farming Systems Research-Extension


with high yield potential and good drought-resistance characteristics. These
are the varieties Tortillero and Sureno and the hybrid Catracho. Tortillero was
released in 1983, Catracho in 1984, and Surenio in 1985. Sureiio is a dual-
purpose variety that can be harvested for grain or forage and has shown good
tortilla quality. Tortilla is the main staple food of the rural poor in Honduras
and is made of maize and/or sorghum. Catracho is a hybrid with high yield
potential. The yield superiority of the improved cultivars over the traditional
varieties is significant. The new sorghums, however, also require higher cash
outlays than the traditional crops, mainly in the form of improved seed and the
chemical fertilizer required to obtain high yields. Table 1 provides an example
of the differences in cash outlays required by the traditional and new
sorghums. In most cases the farmer considers this relatively high cash outlay
as too risky and does not adopt the technology or adopts only part of the
package (e.g., only the seed), which is not very efficient.

Table 1. Partial Budgets for Traditional and New
Sorghum Technologies in Southern Honduras

Description Ma/Be Maicilloa Sureino Catracho
Qty. Total Qty. Total Qty. Total Qty. Total
($/ha) ($/ha) ($/ha) ($/ha)

Cash costs:
Seed (kg/ha)b 10/25 25.30 10.00 2.80 12.00 12.00 12.00 15.36
Fertilizer (kg/ha);
Urea 0.00 0.00 0.00 0.00 45.00 14.85 45.00 14.85
12-24-12 formula 0.00 0.00 0.00 0.00 45.00 17.55 45.00 17.55
Insecticide (kg/ha) 3.00 4.95 2.00 3.30 5.00 6.60 5.00 6.60
Herbicide (gal/ha) 0.00 0.00 0.25 8.25 0.30 9.90 0.30 9.90
Total cash expenses 30.25 14.35 60.90 64.26
Improved sorghum costs as % ofma/be costs 201.32 212.43
Improved sorghum costs as % of maicillo costs 424.39 .447.80

Unit prices used are:
Seed ($/kg): Maize, 0.33; beans, 0.88; maicillo, 0.28; Surefio, 1.00; and Catracho,
Fertilizer ($/kg): Urea, 0.33; formula, 0.39. Insecticide: 1.32 $/kg; Herbicide:
33.00 $/gal.
aMaicillo is the traditional sorghum grown by small farmers in Honduras.
bFor traditional seed, this may not be a cash expense since it is selected and saved
from previous harvests.
Sources: Prices from a 1989 survey of agricultural input/output stores in Choluteca.
Doses from G6mez, et al., and MNR, Chol.

Vol. 1, No. 2, 1990


From an economic standpoint, high-yielding sorghum cultivars can be a
positive addition to the mix of crops grown in Honduras for both national
consumption and for export. These new cultivars are often resistant to
drought and produce higher yields than traditional cultivars. According to
recent statistics (FAO, 1987), Honduras has become a net importer of
sorghum and other cereals because the grains have become important for
human and animal consumption. Total production has not increased to meet
the higher demand for sorghum. An increase in sorghum production,
probably through an increase in yields or acreage, would reduce agricultural
imports with a direct positive effect on the balance of payments. If the farmers
find it profitable to adopt them, higher yields and, thus, increased production,
can be achieved with new cultivars and other more efficient agricultural

Improved Hillside Soil-Management Techniques
A renewed stress on the rational use and conservation of natural resources
in developing countries has been given high priority in recent years. Each year
the international aid agencies place more restrictions on the type of projects
that can be undertaken with development funds. Most of the projects require
an efficient use of the natural resources in the target region. The conservation
of the rain forests, animal species, and water sources is usually required to be
part of these projects. The NRMP of MNR and AID/H have worked jointly
for the last eight years on several projects to protect and control the use of the
watersheds in the southern and east-central regions of Honduras. The work
of NRMP has focused on the small farmers on the laderas, where the damage
to the forest and water sources is potentially greater. Programs for the
construction of rock and live (permanent tree) barriers, ditches, and terraces,
introduction of new crops, use of organic and low doses of chemical fertilizer,
and improved cultivars of traditional crops for hillside lands, have been in farm
trials for several years.
These technologies have the added feature of being almost "free" to the
farmer as the programs are usually accompanied by subsidies to compensate
for the labor and inputs (usually seed) required to build the structures. Once
in place, they require only periodic maintenance and upkeep. The effect of
these structures is usually improved physical soil characteristics and, in some
cases, they provide organic fertilizer and forage for the farm animals. The
addition of these technologies in the model is the next major stage in the
present research. Preliminary economic analysis indicates that a combination

Journal for Farming Systems Research-Extension


of tree and rock barriers and new sorghum technologies with low doses of
chemical fertilizer and compost, has the potential for significant increases in
cereal production by small farmers (L6pez-Pereira, et al., 1989).


Results from the survey discussed above provided the profile of a representa-
tive small farmer of the south. These characteristics, along with historical data
on crop yields and expected yields by the farmers, were used to construct a
whole-farm economic model. In this model, a negative exponential-utility
function of the cash value of ending grain stocks after the second season is
maximized subject to the financial, labor, land, credit, and consumption
constraints described in the survey results. This form of the utility function is
commonly used for decision analysis under risk (Kaylen, Preckel, and Loeh-
man, 1987; Preckel, Featherstone, and Baker, 1987; Gbur and Collins, 1989;
Collins, 1985). The model required a substantial amount of detailed data for
each crop combination and for each season. Data on labor, land, cash
holdings, credit, and production costs, were obtained from the survey results.
Yields for each state of nature were obtained from the average answers given
by the small farmers in the survey, on-farm trials for the new sorghum
technologies, and official government statistics (G6mez, et al., 1989; MNR,
The mathematical model used for the analysis was a discrete stochastic
program (DSP) with two stages (two crop seasons), five states of nature for the
first stage, and three states of nature for the second stage (Cocks, 1968; Rae,
1971a). This procedure is believed superior to other risk-programming
models such as the mean-variance (Markowitz, 1959; Collins and Barry,
1986) and MOTAD (Hazell, 1971; Kaiser and Boehlje, 1980). The main
advantages are that it allows for sequential decisions to be adaptive. For
example, second-season crop decisions are based on first-season crop yields in
the model developed here. DSP also allows for some of the resources or
technical coefficients (e.g., crop yields) to be stochastic. A clear example of this
is the different amounts of labor required for replanting or weeding activities
in different weather scenarios.
Since the small farmers in southern Honduras face a risky environment and
adapt their decisions sequentially as uncertain events unfold, DSP was
considered adequate to model their environment. Although DSP has been
used in several studies in the United States (Rae, 1971b; Tice, 1979; Kaiser

Vol. 1, No. 2, 1990


and Apland, 1989; Featherstone, 1986), the authors know of only one
application to evaluate new technologies in developing agriculture (Adesina,
1988). The main drawback of DSP is the substantial amount of data required
even for relatively simple models. In general, there is a tradeoff between model
detail and size. The size of the model grows exponentially as more stages and
states are included. In order to keep it relatively small, emphasis was placed on
detail for the first season of the DSP developed here. This strategy to keep the
model from becoming too large has been suggested before (Anderson, Dillon,
and Hardaker, 1977).
The first season was assumed to have five possible results of crop yields: very
low; low; medium; high; and very high; with probabilities of 0.15, 0.30, 0.25,
0.20, and 0.10, respectively. For the second season, only three possible yield
outcomes were considered: low, medium, and high, with probabilities de-
pending on the states of nature of the first season. For example, the
probabilities for the second season, given the "very low" state of nature of the
first season, are 0.60, 0.25, and 0.15, respectively. In comparison, for a "very
high" FS state ofnature, probabilities for the SS states of nature are 0.10, 0.20,
and 0.70, respectively. This allows for positive relationships between the first
and second seasons. If the first season has been "very bad," which means
drought or too much rain, a "bad" second season has a higher probability of
occurring than a "good" second season. Therefore there are 15 possible final
outcomes (states of nature) in the model, three in the second season for each
of the five in the first season.
With all the data at hand, the risk-neutral case model was developed. In this
model the farmer is assumed to be indifferent to the level of risk he faces and
his sole objective is to maximize the expected value of ending cash holdings.
This model provides the linear programming results, and is useful to estimate
the impact of including risk in the analysis. In the second model, the farmer
is assumed to maximize a negative exponential utility function of ending cash

Max. E[U(ecash)] = E l P..*(1 -cxp(-ccashi)) (1)

where pji is the probability that the i" state if nature is realized in the SS given
that the j* state of nature was realized in the FS; ecashj is the cash resulting
from the realization of these states ofnature; and is the coefficient ofabsolute
risk aversion. This form of the utility function implies that the farmer is averse
to risk and will prefer returns with low variance. The coefficient Xin equation
1 is the level of risk aversion assumed.

Journal for Farming Systems Research-Extension


The objective function was maximized in each case subject to resource and
consumption constraints; the more important of these constraints are pre-
sented (in non-algebraic form) in equations 2-10.

Initial cash holdings = $70.00 (2)
Total cropland used < 1.7 ha + land rented (3)
Total labor used in crops + off-farm labor < total family labor
available + hired labor (4)
Off-farm labor supplied in crop season < 50 man-days (5)
Credit < $450.00 (6)
Maize consumption + sorghum consumption > minimum
family requirements (7)
Bean consumption > minimum family requirements (8)
Sorghum consumption < 20 percent of total cereal consumption (9)
Grain consumed + grain sold < initial grain stocks + grain
produced + grain purchased (10)

The models were developed and run on a GAMS/MINOS software
program (Brooke, Kendrick, and Meeraus, 1988; Murtagh and Saunders,
1983) at Purdue University. The model output consists of optimal crop
combinations for each of the two seasons of the planning period. It also
includes marginal values of the different resources included in the model. The
effect of changes in the resource and variable coefficients included in the
constraints are also presented in the model solution. Shadow values for such
activities as hired labor for FS planting, amount of credit in SS for each yield
result of FS crops, area of land rented in or out, expected cash value of
production at end of SS, and variance of this expected cash value, are all part
of the output of the models. The model results are presented and discussed in
the next section.


Table 2 presents the optimal crop combinations for the two risk-neutral
models. As mentioned above, in these models the objective is to maximize the
expected cash value of ending grain stocks and other cash-generating activi-
ties. The traditional model is named model I, and the model with the new
sorghum technologies is named model II. The crops selected in model I for
the FS are maize, maize/beans, and maize/sorghum. In comparison, with
model II the crops grown in the FS are maize/beans and Catracho. In the

Vol. 1, No. 2, 1990


Table 2. Optimal Crop Combinations for Traditional and New Sorghum Models,
Risk-Neutral Case-Small Farmer in Southern Hondurasa

Combination Model Ib (ha)

Sole crops:
Maize 0.67
Traditional 0.74 0.74 0.74 0.74 0.74
Double crops:
Maize/beans 0.81 0.74 0.74 0.74 0.74 0.74
Maize/maicillo 0.14
Total area 1.62 1.48 1.48 1.48 1.48 1.48

Combination Model IIb (ha)

Sole crops:
Catracho 0.71 0.70 0.70 0.71 0.71 0.71
Double crops:
Maize/beans 0.71 0.70 0.70 0.71 0.71 0.71
Total area 1.42 1.40 1.40 1.42 1.42 1.42

Note: FS = first season, SS = second season.
aIn these models the objective of the farmer is to maximize the expected value of cash
bModel I includes only traditional crops; Model II includes the traditional crops and
the new sorghum technologies of Sureno and Catracho.
cFS = first season of the year; SS = second season. Crops grown in SS depend on the
yields of FS crops: VL = very low FS yields; L = low; M = medium; H = high; VH =
very high.
Source: Model results.

Journal for Farming Systems Research-Extension