Farming systems research for agroforestry extension

Material Information

Farming systems research for agroforestry extension
Hildebrand, Peter E.
Food and Resource Economics Dept., University of Florida,
Publication Date:
Copyright Date:

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:
191870883 ( OCLC )

Full Text
6yV 031


Peter E. Hildebrand 2
Barbara C. Bellows 3
Paul Campbellt
Bashir Jama Adan*


Farming systems research and extension (FSRE), as used by the
global Association for Farming Systems Research-Extension,
applies to a family of methodologies used to generate, evaluate
and disseminate agricultural technologies in association with
farmer participation. FSRE shares many attributes with D and D
as practiced in agroforestry. The history of FSRE is traced from
1965 to the present showing the formalization of the methodology
and its critical use in sustainable agricultural technology
development. In on-farm research, a primary basis for FSRE,
research and extension merge in practice. The definition of
recommendation domains is based on analysis and interpretation of
multi-environmental research results as evaluated by varied
evaluation criteria. The results of three research projects
demonstrates the nature of farmer criteria for evaluation.
Modified Stability Analysis (MSA) is used to demonstrate the
relationship of on-farm research to specific extension messages.
Design of on-farm research to make it amenable to analysis by MSA
is discussed.

Key Words: farming systems, recommendation domains, modified
stability analysis, sustainable agriculture.

t Florida Agricultural Experiment Station, Journal Series No.

$ Prepared for presentation at the International Conference on
"Agroforestry for Sustainable Development: Past
Achievements and Future Possibilities." Universidad
Aut6noma Chapingo, Chapingo, M6xico August 24-28, 1992.

2 Professor, Food and Resource Economics Department, University
of Florida, Gainesville, Florida 32611-0240 USA.

3 Recently completed a PhD in Soil, and Water Science at the
University of Florida. Now with the SANREM CRSP at the
University of Georgia.
t M.S. Candidate in Forestry (Agroforestry) at the University
of Florida.
$ Recently completed a PhD in Forestry (Agroforestry) at the
University of Florida.

Farming Systems Research (FSR), Farming Systems Research and
Extension (FSRE), or simply "Farming Systems" are widely used,
often misunderstood, and frequently misused terms. As used by
the Association for Farming Systems Research-Extension, they
apply to a family of methodologies used to generate, evaluate and
diffuse agricultural technologies in association with farmer
participation. The use of the term "family of methodologies"
implies that there is no one methodology. Rather, there are
closely related methodologies that are modified according to
local needs and resources.

It is probably safe to say that most practitioners of farming
systems methodology use either the shortened form of farming
systems or attach the E on the end of FSR. The use of FSR is
mostly with the donor community or the CGIAR (Consultative Group
on International Agricultural Research, that overseas the program
and work of the 18 international agricultural research centers -
IARCs worldwide). In the case of the CGIAR, this is probably
because their charge is with research and not extension, there
not being any "international farmers" for them to work with.

In this paper, the terms FSRE and "Farming Systems" will be used
interchangeably. The paper covers three main topics: A brief
history of FSRE, the relationship between FSRE and "Extension"
and a discussion of the socioeconomic and biophysical FSRE
methodologies relevant to agroforestry extension.


1965 1980 (Origin)

In the late 1960s and early 1970s, a number of individuals,
national research and extension organizations, donors, and
development agencies were becoming concerned that Green
Revolution technology was bypassing many of the most needy
farmers in the developing world. These were the farmers who were
in isolated areas where purchased inputs were difficult to
obtain, who had few resources with which to purchase the inputs
necessary for the Green Revolution technology to be productive,
or did not possess the quality of natural resources (soils and
climate) that let the high potential yields manifest themselves.

Interestingly, the beginnings of what we now call FSRE began
simultaneously, but in isolated cases, in Africa, Asia and Latin
America. In Asia, much of the influence came from the multiple
cropping work of Richard Bradfield (1966) and later Richard
Harwood (1975) at the International Rice Research Institute in
the Philippines. This work was largely biological in nature, but

4 This section taken from Hildebrand (1992).

initiated the interest in production systems more akin to what
many Third World farmers practiced. This interest also spawned
the work at CATIE in Costa Rica from which Robert Hart and others

In eastern Africa, with Michael Collinson (1972) in Kenya, and in
western Africa, with David Norman (1975, 1976) in Nigeria,
farming systems interest stemmed from the perspective of farm
management economists. This was also true in my own case, in El
Salvador, where the agricultural economics program was influenced
by Bradfield's multiple cropping activities at IRRI (Hildebrand
and French, 1974). In Guatemala, the effort was to incorporate
socioeconomics into an agricultural research organization
(Hildebrand, 1976). In South America, (CAqueza, Colombia),
Hubert Zandstra a soil scientist, worked with agricultural
economists Ken Swanberg and Carlos Zulberti (Zandstra et al.,
1979). In Mexico, Don Winkelmann and other agricultural
economists were instrumental in generating the CIMMYT farming
systems activities in Latin America (Byerlee et al., 1982)

1975 1990 (Formalizing)

One of the earliest organized efforts to capture these and other
similar activities was the USAID/Washington contract to Colorado
State University and the University of Hawaii (via the Consortium
for International Development, CID). Part of the work in this
project involved a survey of farming systems activities
worldwide. The result was a set of "guidelines" for
practitioners of farming systems research and extension to follow
(Shaner et al., 1982).

In 1982, upon termination of the "Guidelines" project,
USAID/Washington funded the Farming Systems Support Project,
FSSP, whose objectives were to support USAID-funded farming
systems projects with training, networking and technical
assistance activities on a world-wide basis (FSSP, 1987). This
project, headquartered at the University of Florida, had the
collaboration of 21 US universities and four consulting firms.
Among other activities, the FSSP helped sponsor the annual
International Farming Systems Symposium which originated, and
held its first six meetings at Kansas State University, then
passed to the University of Arkansas for three years. In 1992 it
held its third of three meetings at Michigan State University.

As a direct result of the networking created and supported by the
FSSP and the Farming Systems Symposia, the Association for
Farming Systems Research-Extension was created in 1989. This
international society publishes the Journal for Farming Systems
Research-Extension and the AFSRE Newsletter, the successor to the
FSSP Newsletter which became, in the interim, the FSRE
Newsletter. The association also supports the annual Farming

Systems Symposium which is scheduled to hold its thirteenth
meeting in Europe in 1994.

During this period, many USAID missions initiated projects which
had some aspects of farming systems incorporated in them.
Others, partly because of the interest in USAID/Washington,
created projects called "farming systems" projects, but were that
in name only. Farming systems became a "buzzword" that helped
projects to obtain funding. Honest efforts, however, were
established in many countries in Africa, Asia and Latin America.
Canadian, German and French organizations also became involved in
farming systems activities.

As projects became more numerous and the FSSP gained momentum,
there was a great deal of effort toward creating and modifying
FSRE methodology. Rapid rural appraisals (Chambers, 1981), often
called Sondeos (Hildebrand, 1981) in Latin America and much of
the world, were more and more widely used. Concerns with
formalizing the participation of farmers in analysis, generation
of alternatives, testing and evaluation were manifested with
increasing mention in papers and presentations and used in

Because women are so important in agriculture, both as farmers in
their own right and in support of the rural family, gender issues
and gender analysis (Poats, et al., 1988) easily became part of
accepted FSRE methodology. The concept of research,
recommendation and diffusion domains was established (Wotowiec,
et al., 1988). Statistical methods to cope with the problems and
variabilities of on-farm research became more formalized
(Hildebrand, 1984; Stroup et al., 1991). Means of incorporating
livestock in on-farm trials were studied (Nordblom et al. 1985).
And led by CIMMYT and CIAT, means of selecting and prioritizing
alternative solutions to farmers' problems became more meaningful
and usable (Tripp and Woolley, 1989).

It is strange that in the face of a near global diffusion of
farming systems methodology, and its use becoming more and more
widespread even in more conventional research and extension
organizations, its demise was often discussed and/or predicted.
It is quite clear that those who indicated that FSRE was dead or
dying did not understand the nature of the methodology and were
primarily basing their statements on the frequency of buzzwords
heard in the bureaucratic halls of such capitals as Washington.
Perhaps the most often suggested "replacement" was "sustainable

1985 2000 (Sustainable agriculture)

Increasing concerns with the environmental effect of development
rightly has brought the issue of sustainable agriculture into
clear focus. However, the issue that sustainable agriculture

would replace farming systems not only is misguided, but
nonsensical and can only be espoused by the seriously

Sustainable agricultural technology will have to be technology
that fits the environment. Much of the Green Revolution
technology in the Third World, and the "broadly adaptable"
technology in use on large, mechanized farms in the developed
world depend on dominating the environment to fit the technology.
FSRE was developed to provide technology for small scale, limited
resource farmers who are unable to dominate the environment.
Thus, FSRE methodology has been developed to generate technology
that fits the environment (whether the base biophysical
environment or the socioeconomically caused environment) and is
especially suited to develop sustainable agricultural technology
which must be environment- or location-specific in nature. In
fact, two new sustainable agriculture initiatives emanating from
Washington 7 insist on the use of farmer participatory methods
and on-farm research, both of which are clearly farming systems

During this period of time there has been relatively little use
of FSRE methods in agroforestry research and extension, although
much of what is known as D and D (Raintree, 1987) is very
similar. Partly this has been because of the background of FSRE,
coming from annual crops and agronomy while agroforestry is based
heavily multipurpose trees. It is also, no doubt, because
research, particularly on-farm research, is much more difficult
with animals and perennial species than with annual crops. FSRE
practitioners had their hands full with annual cropping systems.

Notwithstanding the "E" on the end of the term, FSRE has been
widely criticized for having a weak extension "linkage" (Merrill-
Sands et al., 1989). When these criticisms refer to linkage with
the institutional extension services, they are usually justified.
Although some FSRE projects are organized within extension
services, most are not. When they are not, there is usually
little institutional connection with extension. On the other
hand, as will be seen in the next section, the extension function
works quite well in most FSRE projects.

7 The Low Input, Sustainable Agriculture (LISA) project from
the US Department of Agriculture, and the Sustainable Agriculture
and Natural Resource Management Collaborative Research Support
Project (SANREM CRSP) from SAID.


On-farm research, the basis of FSRE methodology, uses many of the
farmers' resources, can risk their crops and livelihood, and
usually tries their patience. However, farmers are usually
willing to participate as informants, advisors and purveyors of
resources, if there is a promise of an improved technology
resulting from this participation. On-farm research is a process
to evaluate new technology. Extension is a process disseminate
knowledge of the new technology. Hence, extension becomes a
necessary partner in on-farm research. Furthermore, it is not
really possible to do good (defined from the farmer's
perspective) on-farm research without doing extension. On-farm
research serves the function of demonstration even when it is not
intended as a "demonstration". Additionally, good (from the
farmer's perspective) extension requires that appropriate (from
the farmer's perspective) new technology be available. The best
way to assure good extension, then, is through good on-farm
research. In practice, research and extension functions merge in
FSRE and become nearly indistinguishable.

The concept of Recommendation Domains (Perrin et al., 1976;
Byerlee, Collinson et al., 1980) has been developed in FSRE as a
tool to facilitate the broad dissemination of new technologies.
As is the case with most concepts and methodologies in FSRE, the
definition of this term has been undergoing change over time.
Initially it was equated with homogeneous farming systems
(Hildebrand, 1981) which could be defined with rapid appraisal or
Sondeo procedures. Research was then conducted and technology
extended within these defined recommendation domains. Generally
it was thought that recommendation domains should refer to
farmers rather than fields because it was the farmers who made
decisions about their systems (Harrington and Tripp, 1984).

More recently it has been argued that recommendation domains can
pertain not only to fields, but to parts of fields (micro
environments), and should be defined as the result of both
socioeconomic and biophysical research (Stroup et al., 1991).
They are defined based on the responses of treatments to specific
environmental conditions (the combination of the natural
biophysical environment and the socioeconomically-created
modifications to these natural conditions) and evaluated using
the potential adopting farmers' own evaluation criteria.
Fortunately we have efficient socioeconomic and biophysical
research methods (design, analysis and interpretation) amenable
to making recommendations and extension messages with this degree
of specificity. These will be discussed in the following


On-farm research is, of course, not limited to biological
experiments or trials. Socioeconomic research is a necessary
complement to the biological research and both are critical
inputs into extension programs. One of the most crucial aspects
of on-farm socioeconomic research is to ascertain the farmers'
goals, their scarce and abundant resources, and ultimately, their
evaluation criteria. Three examples will serve to illustrate the
usefulness of on-farm socioeconomic research.

In a bean producing zone of Costa Rica, Bellows (1992), at the
time a soil science graduate student with a farming systems minor
in her PhD degree program, found that introduced bean production
systems based on cleared and burned fields and purchased inputs
acted more as a complement to rather than a substitute for the
traditional system. In this latter system, bean seed is
broadcast into standing fallow or monte and then the vegetation
is felled to cover the seed, and protect against erosion and weed
infestation. The traditional system yields more beans per unit
of labor during the production period (when labor is used for
coffee harvest) and per unit of cash input, since purchased
inputs are seldom used. It is a low-cost and low-labor means of
producing food. The introduced system produces more kg ha'1, and
is considered more of a cash crop. It uses more labor during the
growing season and is based on purchased inputs. However, it is
more amenable to production during the first growing season (when
cleared vegetation can be burned) than the second when the
traditional system is normally planted. The traditional system
depends on abundant land (which is fast disappearing), so kg ha"'
is not yet a relevant evaluation criterion. In these cases, it
is the denominators of the evaluation criteria that are changing
depending on the resources considered scarce and abundant by the
farmers. The numerator, kg of beans produced, remains the same.

In the Machakos area of Kenya, Bashir Jama Adan, then an
agroforestry PhD student with a farming systems minor in his
program of studies, found that farmers could be divided into two
groups depending on their interests in hedgerow products: fodder
or mulch. Interest in fodder was associated with wealth:
farmers with coffee and farmers who used oxen to plow their land.
These farmers purchased less food for their families, used more
fertilizer and manure, hired more labor and weeded more
frequently than the "mulch" group. In this case, the numerator
of the evaluation criterion changes (fodder vs mulch) rather than
the denominator, as in the case of the Costa Rican bean
producers. However, fodder also can be evaluated in terms of the
reduction in purchased feed, while mulch can be evaluated in
terms of increased crop yield.

In an isolated area near Lascahobas in eastern Haiti, Paul
Campbell, then an agroforestry MSc student also with a farming
systems minor, found that farmers had about as many trees on
their farms as they could use. Remaining space was needed for
other crops. The product or products depended on the nature of
the tree (fruit, coffee, firewood, building materials or lumber).
Trees were mostly standing alone, so productivity was measured in
terms of the tree, itself. That is, the denominator in the
evaluation criterion was "tree".

It is obvious that evaluation criteria used by farmers are many
and varied. It should also be obvious that research and
extension workers need to know and understand farmers' evaluation
criteria if they hope to design and evaluate technology
efficiently and to interest farmers in it by way of specifically
directed extension programs. The use of a traditional research
criterion (such as kg ha-1) to the exclusion of other criteria can
mean at best that farmers are not interested in the products
generated and disseminated by the established research and
extension organizations. At worst, using the wrong criterion can
result in discarding technology which may have been useful to the
farmers for whom it was supposedly being generated (Hildebrand,
1990). Knowledge of farmers' evaluation criteria plays an
important role in the analysis of on-farm biophysical research
data and in the dissemination of the resulting technology.


On-farm research has a different purpose than on-station research
and must be based on a design that is acceptable for the
appropriate statistical analyses (Stroup et al., 1991). On-farm
research is organized to help search for technologies most
appropriate for specific environments. Besides mixed model
analysis of variance, Modified Stability Analysis is an effective
means of determining specific recommendation domains.
Modified Stability Analysis or MSA is a procedure for designing,
analyzing and interpreting on-farm research data. The method can
serve as a basis for an entire research-extension program
(Hildebrand and Russell, 1992). This section demonstrates some
of the basic procedures for analyzing and interpreting data taken
from an on-farm research example for which an appropriate design
was used (Singh, 1990).

Four terms important in MSA need to be defined.

Environment The natural biophysical and socioeconomically
created conditions existing for plant or animal
growth in the location of the trial.

Environmental Index, El A convenient measure of the environment
at the location of the trial. For a specific
location, it is the average response of all the
treatments at that location, usually based on
physical yield per hectare.

Evaluation criterion The measure or measures used to compare
the treatments in a trial. Can reflect a
researcher's concern (Mg ha1), for example, or a
farmer's (kg/kg seed, among many others).

Recommendation domain The situations for which specific
treatments or technologies will be recommended.
They are defined by a combination of environmental
considerations and farmer-specific evaluation


A prerequisite to thoroughly analyzing and interpreting on-farm
research data is that the design of the trial is adequate and
amenable to this type of analysis. On-farm research can have
various functions and be managed by researchers, extension
workers and/or farmers (Hildebrand and Poey, 1985). The most
appropriate for incorporating farmer participation is a simple
(few treatments), non-replicated design which has one to three
treatments to be compared with the farmers' own technologies.
For purposes of demonstration, a trial conducted in the Amazon
basin of Brazil (Singh, 1990) with four treatments and on eight
environments, without replication, will be used. Additional
information on design will follow discussion of the steps in

Response of Treatments to Different Environments

The term "environment" is used here rather than "farm" or
"location" because on a single farm, or even in a single field,
more than one environment can exist for the production of the
livestock or crops grown. Similar farmers do manage different
environments differently just as different farmers may manage
similar environments differently. Furthermore, making technology
conform to varying environments, rather than the contrary, is
more in keeping with sustainable agriculture.

Measure of environments: Environmental Index. El

The factors which influence the environment for raising crops or
livestock are many and complex and are very difficult to assess.
A convenient substitute measure of the quality of each
environment where a trial has been conducted is the average
"yield" of all the treatments included when, and only when, the
same treatments are included in all the sampled environments.
The first step is to calculate this index, EI, which provides an
effective measure of the environmental differences in the
research domain represented by the range of Els.

To facilitate further analyses, it is convenient to sort the data
by descending (or ascending) values of this index. The data in
Table 1 are in descending order of the environmental index EI.

Relation of treatment response to environment

The yield data for each treatment should be related to the
environmental index. The second step is to view the observations
by graphing the results of one treatment against El as in Figure
1. The data analyst must decide if the relationship is linear or
curvilinear, and some estimate of the relationship must be made.
It is satisfactory to simply draw a line or a curve through the
data, and with practice this can become fairly precise. Linear
regression can be accomplished easily with many inexpensive
calculators, and linear or curvilinear regression can be
estimated with a computer. The estimated relationship shown in
Figure 2 is from linear regression.

Interaction of treatments with environment

When all treatments have been related to, or regressed on EI, the
third step is to compare the response of the treatments to
environment, as in Figure 3. No interaction exists if all the
lines are parallel. If no treatment by environment interaction
exists, the treatment which is greatest over all environments is
the best for the criterion used, here Mg ha-1. However, if the
lines are not parallel, such as in this case, treatment by
environment interaction exists and different treatments may be

best for different environments. Notice that the values of the
El are shown in the lower part of Figure 3. This will be useful
in helping to designate recommendation domains.

Multiple evaluation criteria

The evaluation criterion used to calculate the EI, above, is Mg
ha-1, the most common criterion used by agronomists in crop trials
and appropriate in most cases as the basis for calculating the
El. However, few farmers use this criterion when making
production decisions. If seed, labor or cash are most scarce,
more appropriate criteria are kg/kg seed, kg/day of labor, or
kg/dollar of cash cost, respectively. MSA easily lends itself to
analysis using multiple criteria. The fourth step is to compare
alternative evaluation criteria. Figure 4 is based on the
farmers' criterion of kg/$ cash cost. Notice that the same El is
used -- this does not change with changes in the evaluation
criterion. The El values used to form the X-axis do not change.
The criteria used on the Y-axis do change. The same procedures
were used to obtain these relationships as were used to obtain
the relationships based on the researchers' criterion, Mg ha1.

Notice that very different conclusions result when the evaluation
criteria change. This is important because it relates to the
recommendations that will be made and to the extension messages
used for further dissemination of the recommended technology.

Defining recommendation Domains for Diffusion

The fifth step is to interpret the results and define
recommendation domains. This involves three sub-steps: 1)
characterize the environments, 2) interpret multiple evaluation
criteria, and 3) assess risk. Recommendation domains, the
situations for which specific treatments or technologies will be
recommended, depend upon both the characteristics of the
environments and the choice of evaluation criteria.

Characterizing the environments

Environments can be characterized using both biophysical and
socioeconomic factors that may, at the same time, be both
quantitative and qualitative in nature. Data obtained for the
environments in the Amazon example include soils characteristics
and a category called "land class", Table 2. The soils
characteristics are self explanatory. Land class refers to the
kind of forest that was cleared (P = primary, S = secondary) and
the number of years it has been cropped (1 = first year, etc.).
The term WL refers to land that had been cleared by bulldozer at
the time of colonization and is, essentially, waste land.

Because the data in Table 2 have been sorted by EI, it is easy to
assess the relationship between El and these characteristics.
Lower Els are associated with lower pHs, lower phosphorus levels,
lower ECECs and higher aluminum saturation. If desired, these
relationships can also be graphed and/or estimated by regression
with El being the dependent variable as was done with pH in
Figure 5.

Perhaps the most useful for farmers and extension agents is the
land class characteristic, because farmers in these conditions
seldom, if ever, have detailed soil information on their fields.
It can be seen that both the nature of the forest that was
cleared for the field and the number of years in use are closely
associated with EI.

Choice of evaluation criteria

Two evaluation criteria have been demonstrated: Mg ha"' in Figure
3 and kg $-1 in Figure 4. In the case of the former, using the
researchers' criterion, TSP would be recommended for the two
highest Els and CM would be recommended for the remaining
environments. The two top Els are PF,. But the fourth highest
is also PFi. The difference is that the top two have a pH higher
than 5 and phosphorus levels above 7.0 ppm. But farmers will not
have this kind of information, so it may be necessary to group
all PFi in one recommendation domain and all other classes in the

For the farmers' criterion, kg $-, none of the amendments are
superior to the farmers' practices for land cleared from either
secondary or primary forest in first year of crop production.
Thus, based on this criterion, none of the amendments would be
recommended for farmers producing maize on land being used the
first year after clearing. Beyond the first year of production,
either CM or TSP would be recommended.

Risk considerations

The probability of low values of the selected criterion for any
of the technologies being assessed in the trial can be estimated
by means of a distribution of confidence intervals based on the
treatment results in the specific recommendation domains. As an
example, consider the choice between CM and TSP in the second or
third year of use, and based on the farmers' criterion, kg $".

The equation:
y t, (s/Vn) (1)
gives the confidence interval for the a level of probability from
a two-tailed "t" table for n-1 degrees of freedom and where s =
sample standard deviation for the observations in the potential
recommendation domain. A probability level a = 0.4 in the "t"

table means that 40% of the values lie outside the interval and
60% lie inside the interval defined by the equation. The lower
value of this equation:
y t. (s/Vn) (2)
provides information on the level of risk associated with the
technology in this recommendation domain, that is, the
probability that yield or other evaluation criterion values would
fall below the value represented by the second equation. Figure
6 shows the risk levels for CM and TSP for the farmers'
criterion, kg $-, and for maize cropping after the first year of
use, using equation (2). In this case, CM is less risky (has a
lower probability of low values) than TSP so would be recommended
for the relevant environment.

Recommendation domains

Based on the above analyses, interpretations and judgements,
several recommendation domains can be specified based on this
single on-farm trial. The treatment to be recommended depends on
the environment and of the evaluation criterion of the farmer.
The former requires information about the kind of field the
farmer is going to plant, when it will be planted, and/or other
factors which may have appeared in the analysis to be important.
The latter depends on the scarcity of resources available to the
specific farmer and what that farmer would like to maximize for
the specific crop or livestock in question. For the example used
here, Table 3 summarizes the options available to an extension
agent for use with different farmers in the research domain. If
Mg ha'1 is a relevant criterion, TSP would be recommended for land
taken from primary forest and in first year of production. CM
would be recommended for all other land. If kg $-1 is relevant,
as it would most likely be to the farmers, what they already do
(FP) is the best on all land in first year of use. If farmers
want or need to produce maize a second year, CM would be
recommended. The conversion of on-farm research to multiple
extension messages for different environments and varying
evaluation criteria is illustrated in Figure 7. It would be
expected that adoption of recommended technology should be high
over a wide range of farms.

Design of On-Farm Trials

It is now time to return to consideration of the design of on-
farm trials appropriate to generate the kind of data used in this
example and amenable to interpretation using Modified Stability

Treatments. Treatments should be few in number to
facilitate participation of farmers in the trial. This enhances
diffusion of acceptable results but also helps research and
extension workers understand the farmers' evaluation criteria,

needed for analysis of the data. All environments should have
the same treatments, or at least a common set of them. Note that
differences in farmer management become factors affecting
environment (not treatments) and have a positive, rather than a
negative impact on trial design.

Replications. No replications are needed for analysis by
MSA. If replications are desired by the research or extension
worker to help assure that a trial will not be lost at a specific
location, two blocks is a sufficient number.

Environments. The number of environments is more important
than the number of replications in each environment. The
following, from Stroup et al., provides a simple guide, based on
number of treatments in the trial:

S the number of environments required for
estimation of treatment by environment response in
research domains and verification in recommendation
domains is not excessive. In order to have at least 20
degrees of freedom in the error term, and allowing for
estimation of both linear and quadratic responses . .
if 8 treatments are included in the trial . 6
environments is an adequate number. For 4 treatments,
10 environments would be required, and in a
verification trial with only two treatments (the
recommended treatment and the farmer check, for
example) 23 environments is adequate. These
suggestions, of course, are approximate. (pp. 13-14)


The "Farming Systems Research Movement" (Tripp, 1991) has been
evolving for about 20 years. On-farm research, the basis of
Farming Systems Research and Extension (FSRE) necessarily merges
research and extension functions. An effective and efficient
FSRE on-farm research method, Modified Stability Analysis (MSA),
utilizes both socioeconomic and biophysical data and procedures
to provide information upon which to define recommendation
domains. Using MSA, recommendation domains, the situations for
which specific treatments or technologies will be recommended,
are defined by a combination of environmental considerations and
farmer-specific evaluation criteria. This same information is
used to create multiple extension messages tailored to individual
farmers' situations.

By using MSA as a basis for the design, analysis and
interpretation of on-farm research, agroforestry practitioners
should be able to capitalize on the long FSRE investment in on-
farm research methods and enhance both research and extension
aspects of their projects.


Adan, B.J. 1993. Soil fertility and soil productivity aspects
of alley cropping Leucaena Leucocephala and Cassia Siamea under
semi-arid conditions at Machakos, Kenya. Unpublished PhD
dissertation. University of Florida, Gainesville.

Bellows, B.C. 1992. Sustainability of bean (Phaseolus vulgaris
L.) farming on steep lands in Costa Rica: An agronomic and
socioeconomic assessment. Unpublished PhD dissertation.
University of Florida. Gainesville.

Bradfield, Richard. 1966. Toward more and better food for the
Filipino people and more income for her farmers. A/D/C Paper.
Agricultural Development Council, Inc. New York.

Byerlee, D., Collinson, M. et al. 1980. Planning technologies
appropriate to farmers: Concepts and procedures. Mexico:

Byerlee, D., L. Harrington and D.L. Winkelmann. 1982. Farming
systems research strategy and technology design. Amer. J. of
Agric. Econ. 64:899-904.

Chambers, R. 1981. Rapid rural appraisal: rationale and
repertoire. Public Administration and Development. 1:95-106.

Collinson, M. 1972. Farm management in peasant agriculture.
Praeger Publishers, Inc. (Republished in 1983 by Westview Press,
Inc. Boulder.)

FSSP. 1987. Training, networking and technical assistance: The
product and process of the Farming Systems Support Project, 1982-
1987. University of Florida. Gainesville.

Harrington, L.W. and R. Tripp. 1984. Recommendation domains: A
framework for on-farm research. Working Paper 02/84. Mexico:

Harwood, R.R. 1975. Farmer-oriented research aimed at crop
intensification. pp. 12-32. In Proceedings of the Cropping
Systems Workshop, IRRI, Los Bafos, Philippines.

Hildebrand, P.E. 1976. Generating technology for traditional
farmers: a multidisciplinary methodology. Paper presented at
the Conference on: Developing economies in agrarian regions: a
search for methodology. The Rockefeller Foundation Conference
Center, Bellagio, Italy. ICTA, Guatemala.

Hildebrand, P.E. 1981. Combining disciplines in rapid
appraisal: The sondeo approach. Agric. Administration 8:423-

Hildebrand, P.E. 1984. Modified stability analysis of farmer
managed, on-farm trials. Agronomy Journal, 76:271-274.

Hildebrand, P.E. 1990. Modified stability analysis and on-farm
research to breed specific adaptability for ecological diversity.
pp. 169-180 In: Kang. M.S. (Ed.). Genotype-by-environment
interaction and plant breeding. Louisiana State University.
Baton Rouge.

Hildebrand, P.E. 1992. The FSRE approach: The meaning and
value of FSRE to promote agricultural production and the welfare
of rural families. Presented at the Southern African Farming
Systems Conference "Farming Systems Research-Extension in the
1990s: Its relevance as a practical approach to solving the
problems faced by farmers in Southern Africa. Johannesburg, RSA.
February 25-27.

Hildebrand, P.E. and E.C. French. 1974. Un sistema salvadoreho
de multicultivos. Departamento de Economia Agricola, Centro
Nacional de Tecnologia Agropecuaria (CENTA), Ministerio de
Agriculture y Ganaderla. San Salvador, El Salvador.

Hildebrand, P.E. and F. Poey. 1985. On-farm agronomic trials in
farming systems research and extension. Lynne Rienner Publ.
Inc., Boulder, Colorado.

Hildebrand, P.E. and J.T. Russell. 1992 (Draft). Modified
stability analysis: A method for the design, analysis and
interpretation of on-farm research-extension. Anticipated
publication in 1993 by Lynne Rienner Publishers, Inc. Boulder,

Merrill-Sands, D., D. Kaimowitz, K. Sayce and S. Chater. 1989.
The technology triangle: Linking farmers, technology transfer
agents, and agricultural researchers. ISNAR. The Hague.

Nordblom, T.L., A.K.H. Ahmed and G.R. Potts. 1985. Research
methodology for livestock on-farm trials. IDRC. Ottawa.

Norman, D. W. 1975. Rationalizing mixed cropping under
indigenous conditions: The example of northern Nigeria. Samaru
Research Bulletin 232. Institute for Agricultural Research,
Samaru, Ahmadu Bello University. Nigeria.

Norman, D.W. 1976. The social scientist in farming systems
research. Invited paper for the Workshop on Farming Systems
Research in Mali, Institut d'Economie Rurale, Bamako.

Perrin, R.K., Winkelmann, D.L., Moscardi, E.R. and J.R. Anderson.
1976. From agronomic data to farmer recommendations: An
economics training manual. Mexico: CIMMYT.

Poats, S.V., M. Schmink and A. Spring (eds.). 1988. Gender
issues in farming systems research and extension. Westview
Press. Boulder and London.

Raintree, J.B. 1987. The state-of-the-art of agroforestry
diagnosis and design. Agroforestry Systems 5:219-250.

Shaner, W.W., P.F. Philipp and W.R. Schmehl. 1982. Farming
systems research and development: Guidelines for developing
countries. Westview Press. Boulder.

Singh, B.K. 1990. Sustaining crop phosphorus nutrition of
highly leached oxisols of the Amazon Basin of Brazil through use
of organic amendments. Unpublished PhD Dissertation, University
of Florida, Gainesville.

Stroup, W.W., P.E. Hildebrand and C.A. Francis. 1991. Farmer
participation for more effective research in sustainable
agriculture. Food and Resource Economics Dept. Staff Paper SP91-
32. University of Florida, Gainesville.

Tripp, R. 1991. Planned change in farming systems: Progress in
on-farm research. John Wiley & Sons. Chichester.

Tripp, R. and J. Woolley. 1989. The planning stage of on-farm
research: identifying factors for experimentation. Mexico, D.F.
and Cali, Colombia: CIMMYT and CIAT.

Wotowiec, P., S.V. Poats and P.E. Hildebrand. 1988. Research,
recommendation and diffusion domains: a farming systems approach
to targeting. In: Poats, S.V., M. Schmink and A. Spring. 1988.
Gender issues in farming systems research and extension.
Westview Press. Boulder and London.

Zandstra, H., K. Swanberg, C. Zulberti and B. Nestel. 1979.
CAqueza: Living rural development. IDRC. Ottawa.

Table 1. Response of maize to three soil amendments and the
farmers' practices from on-farm research results in Amazonas,
Brazil (Singh, 1990). FP is farmers' practices, PCW is processed
city waste (from Manaus), CM is chicken manure, and TSP is triple
super phosphate. For more details see Singh, 1990.

Farm No.
















Table 2. Environmental characteristics of the sites for the on-
farm maize trials in Amazonas, Brazil (Singh, 1990).

El Land Type







Al sat 205




Table 3. Summary of the recommendation domains and specific
technology recommended, based on the data for maize in Amazonas,
Brazil (Singh, 1990).

Land Type


kg $-' FP FP CM CM None




................................................................................................. .....................................
1 .... . . . ...............................................................................................

-1 T I I I T
0 0.5 1 1.5 2 2.5 3 3.5

Figure 1. Observed response in Mg ha' of the TSP treatment to
environment (EI) for maize in Amazonas, Brazil (Singh, 1990).


1 1.5 2 2.5


Figure 2. Estimated linear response in Mg ha' of the TSP
treatment to environment (EI) for maize in Amazonas, Brazil
(Singh, 1990).



Figure 3. Estimated responses in Mg ha-1 of the four treatments
to environment (EI) for maize in Amazonas, Brazil (Singh, 1990).






1 1.5 2 2.5

Figure 4. Estimated
environment (EI) for

responses in kg $' of the four treatments to
maize in Amazonas, Brazil (Singh, 1990).






3.5- ------
SR 2 ............. ................................. .............................. . .................................................................................................9 4

3.8 4 4.2 4.4 4.6 4.8 5 5.2

Figure 5. Relationship of soil pH to El of the sites
farm maize trials in Amazonas, Brazil (Singh, 1990).

for the on-

WL, SF2, PF2

10 15

Figure 6. Risk
CM and TSP from

levels for the criterion kg $" and the treatments
the maize trials in Amazonas, Brazil (Singh,

of Farms









Not recommended











Multiple Environments






Level of Adoption High
I INone

Figure 7. Extension messages for multiple environments and
several evaluation criteria with MSA.

B02/AFCONF (Text)
Singh5.WQ1 (Fig.1-4)
Envph.WQ1 (Fig.5
Conint.WQ1 (Fig. 6)