• TABLE OF CONTENTS
HIDE
 Front Cover
 Title Page
 Table of Contents
 Measuring sustainability: Issues...
 Indigenous knowledge and fertilizer...
 Gender issues in African farming:...
 A regional perspective on cassava,...
 Extension agent surveys for defining...
 The role of farmers in the Jordanian...
 Methodology for designing and evaluating...
 Some aspects of land management...
 Farming systems research approach...






Group Title: Journal for farming systems research-extension.
Title: Journal of farming systems research-extension
ALL VOLUMES CITATION PAGE IMAGE ZOOMABLE PAGE TEXT
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00071921/00006
 Material Information
Title: Journal of farming systems research-extension
Alternate Title: Journal for farming systems research-extension
Abbreviated Title: J. farming syst. res.-ext.
Physical Description: v. : ill. ; 23 cm.
Language: English
Creator: Association of Farming Systems Research-Extension
Publisher: Association of Farming Systems Research-Extension
Place of Publication: Tucson Ariz. USA
Publication Date: 1990-
 Subjects
Subject: Agricultural systems -- Periodicals -- Developing countries   ( lcsh )
Agricultural extension work -- Research -- Periodicals   ( lcsh )
Sustainable agriculture -- Periodicals -- Developing countries   ( lcsh )
Genre: periodical   ( marcgt )
 Notes
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.
Funding: Electronic resources created as part of a prototype UF Institutional Repository and Faculty Papers project by the University of Florida.
 Record Information
Bibliographic ID: UF00071921
Volume ID: VID00006
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 22044949
lccn - sn 90001812
issn - 1051-6786

Table of Contents
    Front Cover
        Front Cover
    Title Page
        Title Page 1
        Title Page 2
    Table of Contents
        Table of Contents
    Measuring sustainability: Issues and alternatives, by L. W. Harrington
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
    Indigenous knowledge and fertilizer strategies in Leyte, Philippines: Implications for research and demonstration trials, by Daniele Perrot-Maitre and T. F. Weaver
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
    Gender issues in African farming: A case for developing farm tools for women, by R. N. Kaul and A. Ali
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
    A regional perspective on cassava, famine, and seasonal hunger in humid and subhumid Africa, by Steven Romanoff
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
    Extension agent surveys for defining recommendation domains: A case study from Kenya, by Steven Franzel
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
    The role of farmers in the Jordanian combined Sondeo process, by D. L. Galt and A. F. Al-Kadi
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
    Methodology for designing and evaluating comparative cropping systems, by Luanne Lohr, Oran Hesterman, James Kells, Douglas Landis, and Dale Mutch
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
        Page 111
        Page 112
        Page 113
        Page 114
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
        Page 120
        Page 121
        Page 122
        Page 123
        Page 124
        Page 125
        Page 126
        Page 127
        Page 128
        Page 129
        Page 130
    Some aspects of land management by small farmers of southwestern Nigeria, by M. A. Adewole Osunade
        Page 131
        Page 132
        Page 133
        Page 134
        Page 135
        Page 136
        Page 137
        Page 138
        Page 139
        Page 140
        Page 141
        Page 142
        Page 143
        Page 144
    Farming systems research approach in risky agroecosystems of North Bihar, India, by S. Saran, P. Mishra, A. Kumar, and A. Salman
        Page 145
        Page 146
        Page 147
        Page 148
        Page 149
        Page 150
        Page 151
        Page 152
        Page 153
        Page 154
        Page 155
        Page 156
        Page 157
Full Text
Volume 3, Number 1
1992



journal
for Farming Systems
Research-Extension



L.W. Harht 0o 0


R.N Kau an. Ali *
47 Csaa aie n esnlHne nArc



10 *oprtv C ropn System







Journal
for Farming Systems
Research- Extension


Volume 3, Number 1, 1992


Published by
the Association for Farming Systems Research-Extension








Journal for Farming Systems Research-Extension


Editor
Timothy R. Frankenberger
Office of Arid Lands Studies
The University of Arizona, Tucson

Associate Editors
Nancy Schmidt and Daniel Goldstein
Office of Arid Lands Studies
The University of Arizona, Tucson

Production and Layout
Nancy Schmidt, Eliza Cain, and Diedre Muns
Arid Lands Design, Office ofArid Lands Studies
The University of Arizona, Tucson

Sponsors
Michigan State University Foundation
United States Agency for International Development
The University of Arizona




The Journal for 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 of participatory on-
farm systems research and extension. The objectives of such 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
fiber.
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.


ISSN: 1051-6786









Journal for Farming Systems Research-Extension
Volume 3, Number 1, 1992






CONTENTS


1 Measuring Sustainability: Issues and Alternatives
L.W. Harrington

21 Indigenous Knowledge and Fertilizer Strategies in Leyte, Philippines:
Implications for Research and Demonstration Trials
Danile Perrot-Mattre and T.F. Weaver

35 Gender Issues in African Farming: I. A Case for Developing
Farm Tools for Women
R.N. Kaul and A. Ali

47 A Regional Perspective on Cassava, Famine, and Seasonal Hunger
in Humid and Subhumid Africa
Steven Romanoff

71 Extension Agent Surveys for Defining Recommendation Domains:
A Case Study From Kenya
Steven Franzel

87 The Role of Farmers in the Jordanian Combined Sondeo Process
D.L. Gait and A.F. Al-Kadi

105 Methodology for Desiging and Evaluating Comparative Cropping Systems
Luanne Lohr, Oran Hesterman, James Kells, Douglas Landis, and Dale Mutch

131 Some Aspects of Land Management by Small Farmers of Southwestern Nigeria
M.A. Adewole Osunade

145 Farming Systems Research Approach in Risky Agroecosystems
of North Bihar, India
S. Saran, P. Mishra, A. Kumar, and A. Salman









Measuring Sustainability: Issues and
Alternatives'

L. W. Harrington 2



ABSTRACT
There is widespread agreement among agricultural scientists on the
importance ofsustainability as an objective for research. Many scientists
acknowledge that the ability to quantify sustainability is critical to making
the concept operational. There is less agreement, however, on appropri-
ate methods for its measurement. This paper discusses alternative ap-
proaches to the measurement ofsustainability, given the differing ways in
which the concept is used. Three major interpretations ofsustainability
are discussed: agroecology, equity, and sustainable growth. A number of
nonquantitative and quantitative approaches to the measure of sustain-
ability are then presented. It is concluded that work on measuring
sustainability is still at an early stage and that much remains to be done by
FSRE practitioners in partnership with disciplinary specialists.


BACKGROUND AND OBJECTIVES

There is widespread agreement among agricultural scientists on the impor-
tance ofsustainability as an objective for research. Many scientists acknowl-
edge that the ability to quantify sustainability is critical to making the concept
operational (e.g., Hildebrand and Ashraf, 1989). There is less agreement,
however, on appropriate methods for its measurement. This is hardly
surprising given the many ways in which "sustainability" has been character-
ized and defined.
The objective of this paper is to discuss alternative approaches to the
measurement of sustainability, given differing uses of that concept. First,
broad issues associated with measuring sustainability are discussed. Then
three major interpretationsofsustainability (agroecology, equity, and sustain-
able growth) are presented, along with a classification of sustainability
problems (internal vs. external, reversible vs. irreversible, agricultural produc-

1 Paper presented at the 11th Annual Association for Farming Systems Research-Extension
Symposium, Michigan State University, East Lansing, October 5-10, 1991.
2 Regional Economist for Asia, CIMMYT Economics Program, Bangkok, Thailand.






HARRINGTON


tivity vs. public health). The implications of each of these for the measurement
of sustainability are explored. A number of nonquantitative and quantitative
approaches to the measurement of sustainability then are listed and discussed.


ISSUES IN THE MEASUREMENT OF SUSTAINABILITY

The notion of sustainability has been defined and characterized in vastly
different ways-from the resilience of individual agroecosystems to food
security in the face of global climate change. Approaches to measurement will
be heavily conditioned by how "sustainability" is interpreted. Still, some
general issues cut across all approaches to measuring sustainability.

General Issues
Predicting the future. Assessments of sustainability necessarily imply
drawing conclusions-or at least stating probabilities-about future events.
When an agroecologist warns of agroecosystem breakdown as a system
becomes less diversified, a forecast is being made. Similarly, when farmers'
soil-eroding practices are portrayed as unsustainable, predictions are implicitly
being made about future levels of soil depth and fertility and crop productivity.
More obviously, when LISA ("low-input sustainable agriculture") propo-
nents advocate a switch to wholly renewable resources, tacit forecasts are being
made about the future availability and prices of agricultural inputs.
Like all forecasts, these necessarily contain a degree of uncertainty. Fore-
casts about the future effects of rapid soil erosion on crop yields may be
considerably more reliable than predictions about regional changes in temper-
ature and rainfall due to global warming. Thus the degree to which
sustainability can be measured and quantified depends very much on the
ability of analysts to predict reliably relevant future events.
Time frame. Difficulties associated with measuring sustainability are
exacerbated by the wide range of time frames that apply to different sustain-
ability issues. Some problems are best studied in the medium-term, within a
time frame of 5 to 20 years (e.g., problems of soil nutrient depletion; build-
up of weeds, pests, and diseases; rapid soil erosion, etc.). Other problems are
best studied within a longer time frame of 20 to 100 years. These include
slower forms of land degradation (e.g., gradual processes of erosion, saliniza-
tion, or desertification) and some expected changes in the external environ-
ment (e.g., initial effects of global warming). Still other problems are best
"studied" (if that word still makes sense in this context) in terms of very long


Journal for Farming Systems Research-Extension






MEASURING SUSTAINABILITY


time frames of 100 to 1,000 years and beyond. These would include questions
about the "ultimate" sustainability of agriculture.
State versus control variables-Problems and causes. In some approaches to
sustainability measurement, only "state variables" descriptorss of environ-
mental or resource quality) are quantified. In other approaches, both state
variables and "control variables" (variables that directly influence the level of
a state variable) are quantified. For example, the control variable "tillage
practice" influences the state variable "soil depth remaining after erosion."
There is typically a problem-cause relationship between control and state
variables.
When only state variables are measured, considerable doubt can remain
regarding causes of any observed changes. For example, it may be observed
that per capital food production (a state variable) is declining over time. The
causes (Increasing population? Lower input levels? A switch to nonfood cash
crops? Declining yields?), however, cannot be ascertained unless appropriate
control variables are also measured. Satisfactory measurement of sustainabil-
ity, then, is likely to require the simultaneous measurement of a number of
state and control variables, linking problems with their causes.
Continuous versus discrete measurement. There seems to be little discussion
in the literature about continuous vs. discrete measurement of sustainability.
Ifsustainability is conceived of as discrete, then an agroecosystem (in theory)
can be described as being either sustainable or not. Measuring sustainability
comes down to ascertaining which of these two states prevails. Ifsustainability
is conceived of as continuous, however, it is possible to visualize different
degrees of sustainability, opening the way to quantitative comparisons be-
tween alternative systems. Most proponents of increased quantification seem
to assume implicitly that continuous measurement is possible.
Level of measurement and possibilities for substitution. It is frequently
assumed that sustainability is best measured at the plot level. Sustaining a
particular cropping pattern in a specific location is often taken to mean
maintaining the productivity of that pattern (and the quality of the resources
devoted to the production of that pattern) indefinitely. Yet other cropping
patterns may come along that are more attractive to farmers. At a more general
level, it is not always necessary to insist on the sustainability of all system
components-it is possible to employ some resources in excess of sustainable
levels while maintaining the overall productivity of the resource base by using
substitution possibilities among resources (Graham-Tomasi, 1990). A major
issue, then, is deciding exactly what it is that we are trying to sustain.


Vol. 3, No. 1, 1992






HARRINGTON


Sustainability of What? Three Interpretations
"Sustainability" and sustainable agricultural development have been con-
ceptualized and defined in numerous ways. Indeed, there is an overabundance
of definitions. Most of these, however, appear to fall into one or more of three
distinct categories: agroecology, equity, and sustainable growth.
Agroecology. Some definitions focus on sustainability in terms of system
resilience, or the ability of an agricultural system to "maintain its productivity
when subject to stress or perturbation" (Conway, 1986). Sustainability in the
agroecological sense is enhanced through system diversity. A diversity of
enterprises over time and space fosters the recycling of nutrients, an increased
efficiency in the use of moisture, nutrients, and sunlight, and a reduction in
the incidence of weeds, pests, and diseases (Altieri, 1987). Modern monocul-
ture, characterized by low levels of diversity, is viewed as having a fragile
ecological equilibrium, with control coming from external inputs rather than
internal feedback mechanisms (Ingram and Swift, 1989).
In this view, then, the sustainability of agriculture can be improved by
increasing system diversity and by fostering nutrient and energy cycling (and
thereby reducing the use of external inputs) through the development of
suitable new farming systems (Francis, 1986; Altieri, 1987). Consequently,
monitoring changes in system diversity and in the internal cycling of nutrients
and energy is perceived as fundamental when measuring the sustainability of
an agricultural system.
Equity. Other definitions focus on sustainability in terms of equity,
especially intergenerational equity and the rights of nonhuman species (Batie,
1989). The emphasis is on stewardship, or the proper care and protection of
resources (Barker and Chapman, 1988). This conceptualization is founded
on the belief that future generations have the right to an environment and a
stock of renewable and nonrenewable resources in no worse condition than
that enjoyed by the current generation.
In theory, the efficient intertemporal use of resources can be assessed by
means of benefit/cost analysis (Schmid, 1989). However, intertemporal
efficiency can lead to extinction of renewable resources and exhaustion of
nonrenewable resources (Clark, 1976). Discounting involves making judge-
ments concerning the value of current versus future consumption. However,
when these judgements are made by the current generation of humankind on
the behalf of future generations, serious ethical questions emerge (Batie,
1989; IFPRI, 1989). Moreover, agricultural and economic development
sometimes are seen as inherently unsustainable simply because geometric


Journal for Farming Systems Research-Extension






MEASURING SUSTAINABILITY


growth rates (e.g., in demand for food) are ultimately incompatible with
absolute scarcities (e.g., of resources or of the environment to absorb
pollution; Heilbroner, 1980; Batie, 1989).
In this conceptualization, the sustainability of agriculture can best be
enhanced by slowing down economic development, stabilizing human pop-
ulation growth, and fending offthe exploitation of natural resources (especial-
ly common property resources; Barbier and McCracken, 1988; Durning,
1990). Proponents of the equity view, then, argue that measurements of
sustainability must somehow capture the quantity and quality of natural
resources expected to be available for future generations.3
Sustainablegrowth. A third major view ofsustainability focuses on the need
for continued growth in agricultural productivity, while maintaining the
quality of resources devoted to agriculture. It implies the following: using
renewable resources at rates less than the rate at which they can be continu-
ously generated; emitting wastes at rates less than the rate at which they can
be absorbed by the environment; and optimizing the efficiency with which
renewable resources are used (Barbier and McCracken, 1988).
This view of sustainability takes into account foreseen increases in the
demand for food arising from continuing population and income growth. It
is this view that has inspired the definition ofsustainable agriculture proposed
by the Technical Advisory Committee of the Consultative Group on Intera-
tional Agricultural Research (CGIAR): sustainable agriculture "should in-
volve the successful management of resources for agriculture to satisfy
changing human needs while maintaining or enhancing the quality of the
environment and conserving natural resources" (CIMMYT, 1989).
The sustainable growth perspective can be operationalized (and measured)
at several different levels, among them the regional level (where sources of
agricultural productivity growth are compared with expected growth in
demand for agricultural products; e.g., Byerlee and Siddiq, 1989; Rosegrant
and Pingali, 1991) and the plot level (where changes in yields and total factor
productivity are explained in terms of changes in levels of inputs, technical
change, and changes in resource quality; e.g., Lynam and Herdt, 1988).4
3 Given the possibility of irreversible deterioration in resource quality, the nonlinear way in which
resource quality can vary (e.g., global climate change), and the very long time periods involved,
sustainability measurement in the equity interpretation is clearly challenging-if not impossible.
4 Clearly, the levels are related: The ability of food supply to keep up with growth in demand
increasingly hinges on resolving plot-level constraints to increased yields. Similarly, it is not
always necessary to insist on the sustainability of all system components-it is possible to employ
some resources in excess of sustainable levels while maintaining the overall productivity of the
resource base by using substitution possibilities among resources (Graham-Tomasi, 1990).


Vol. 3, No. 1, 1992






HARRINGTON


Plot-level issues can be further subdivided in accord with the importance of
externalities or common property resources as causal factors.

Classes of Sustainability Issues
An immense number of issues are raised regularly in relation to the
sustainability of agriculture. An incomplete listing of these issues might
include soil erosion, global warming, salinization of irrigated areas, deforest-
ation, deterioration of soil structure, reduction in biodiversity, exhaustion of
soil nutrients, desertification, pest and disease buildup, environmental pollu-
tion from agricultural chemicals, and reduced future availability of agricultural
inputs (including fossil fuels). Many of these issues, especially those having to
do with land degradation or the maintenance of soil quality, have been studied
in some depth by disciplinary and subject matter specialists. Other issues, such
as global warming, are relatively new.
For simplicity, researchers should aim to group sustainability issues into
categories. Different ways to measure sustainability may be needed for each
of these categories.
External versus internal. External issues of sustainability are those associ-
ated with changes in farmers' external circumstances. Global warming and
future climate change, future availability and prices of fertilizers and other
purchased inputs, and changes in global biodiversity are examples. These
issues are beyond the farmers' control, or exogenous. Internal questions of
sustainability are associated directly with farming system operations and
farmers' decision making, e.g., soil erosion, nutrient mining, buildup of pests
and disease, salinization, environmental pollution from agricultural chemi-
cals, etc. These questions may or may not be partly associated with issues of
common property resources and externalities.5
Reversible versus irreversible. Sustainability problems may be distinguished
as reversible and irreversible. The permanent effects of irreversible problems
cause special concern. When future demands for a resource are uncertain and
the effects ofan irreversible change are not well known, the present generation
may perceive a value ("option demand") in maintaining an option for future
uses of a resource (Johnston, 1988).6
5 Not all issues can be unquestionably classified as either internal of external. Farm operations, for
example, undoubtedly contribute (although in a relatively subordinate way) to global warming
(Pretty and Conway, 1989). Moreover, most internal issues are conditioned to a certain extent
by farmers' external circumstances. Nonetheless, the distinction helps by highlighting the
relative importance of farm-level decisions in addressing sustainability problems.
6 Irreversibility has been most studied in relation to species extinction (e.g., Krutilla and Fisher,
1975; Bishop, 1978). The concept of a "safe minimum standard" was developed to identify

Journal for Farming Systems Research-Extension






MEASURING SUSTAINABILITY


Some of the problems commonly associated with the sustainability of
agriculture are not obviously irreversible, for example soil nutrient depletion,
loss of soil structure, or buildup of pests and diseases. On the other hand,
severe soil erosion or massive deforestation can often be considered reversible
only under the most optimistic-and unlikely-assumptions about future
land use over exceedingly prolonged periods of time.
Public health versus agricultural productivity. A few of the issues often
included under the rubric of sustainable agriculture have little to do with the
sustainability of agricultural productivity as such. Rather, they deal with the
effects of agricultural practices (e.g., pesticide application) on public health.
There is no doubt that these questions are important and that adjustments in
agricultural technology can be effective in addressing them. However, they
are different from other problems in that they do not deal with threats to
future agricultural productivity or food security.

Interpretations and Categories: Implications in Measuring Sustainability
The measurement ofa complex notion like sustainability is, to say the least,
challenging. It is unlikely that a single approach to measuring sustainability-
equally useful regardless of the concept interpretation or the category of
problem under consideration-will ever be found. To this extent, the idea of
"measuring sustainability" is not very meaningful.
More specifically, the measurement of the sustainability of agricultural
productivity when threatened by externalproblems (climate change, reduced
availability of external inputs) is likely to depend greatly on the work of
disciplinary specialists. Within resource economics, for example, there are
those who specialize in assessing future availability and prices of natural
resources (e.g., Chapman, 1983; U.S. Department of Interior, 1989). Agri-
cultural field scientists would do well to monitor (without feeling compelled
to duplicate) the work of these specialists. Plausible farmer adaptations to
expected increases in the prices of external inputs (e.g., adjustments in input
use, shifts in enterprise selection, adoption of low-input technologies) can
then be assessed.7 The work of outside specialists on global climate change
minimum resource quality levels (e.g., sufficient population and habitat of a species to assure
survival) and to argue in favor ofa particular rule: "Extinction [and, presumably other irreversible
forms of resource degradation] should be avoided unless social costs are unacceptably large"
(Bishop, 1978). These concepts irreversibilityy, option demand, and safe minimum standards)
can be applied to other issues of resource degradation.
7 Developing adaptations to possible shifts in farmers' external circumstances is a worthwhile
activity. However, it should be distinguished from the broader theme of"measuring sustainabil-
ity," which requires considerable information on the direction and extent to which those external
circumstances will, in fact, change.
Vol. 3, No. 1, 1992







HARRINGTON


and implications for agriculture will take on a similar importance. Some
studies of this issue have already been conducted (e.g., Arthur, 1988; Jodha,
1989) but much remains to be done.
Measuring sustainability, in the agroecological interpretation of system
resilience, will depend on the development of reliable quantitative indicators
of resilience and diversity that can be readily measured. To date, there has
been little progress in formulating such indicators (Tisdell, 1988). In contrast,
there has been considerable work, following the "sustainable growth" inter-
pretation, on approaches to measuring the sustainability of agricultural
productivity when threatened by internalproblems. A number of approaches
have been proposed, typically relying in one way or another on trends in yields
or total factor productivity (state variables), with or without complementary
evidence on resource degradation. Discussion of some of these approaches
(Table 1) constitutes the rest of this paper.

Table 1. Approaches to Measuring Sustainability: Some Comparisons.*


Approach
Directional
measurements

Forecasts on
future availabil-
ity and prices of
external inputs
Indicators of
global climate
change and effects
on agriculture
Indicators of
system resilience

Indicators of
groundwater
pollution
Production or
yield trends

Trends in per
capital production

Yield trends in
relation to inputs
applied


Interpretation
All


Agroecology;
equity


Problem category

Potentially
all


External,
irreversible


Equity, External,
agroecology irreversible


Agroecology Internal,
plot-level,
productivity
Equity Internal,
public health

Sustainable External or
growth internal;
productivity
Sustainable Internal,
growth regional,
productivity
Sustainable Internal,
growth plot-level,
productivity


Comments

Assumes proportional
relation between
problems and causes
Available from disciplin-
ary specialists;
uncertainty in forecasts

Available from disciplin-
ary specialists;
uncertainty in forecasts

Suitable indicators not
clearly identified

May not deal with
resource degradation,
food security
No information on
causes; easily confounded

No information on
causes; easily confounded

Approximation to total
factor productivity; more
information on sources


Journal for Farming Systems Research-Extension






MEASURING SUSTAINABILITY


METHODS FOR MEASURING SUSTAINABILITY

Nonquantitative Approaches
Rejecting quantification. Before proceeding, it should be recognized that
some scientists reject the very notion that sustainability can or should be
quantified. For example, MacRae et al. (1989) argue that quantification tends
to distort the research process, inducing researchers to choose quantifiable
(but less relevant) variables at the expense of other nonquantifiable (but
conceptually more important) ones. They are especially skeptical of numerical
modeling of biological systems, arguing that the internal consistency of these
models does not compensate for their lack of realism.
This rejection ofquantification is linked to a similar rejection of"reduction-
ism." It is usually not possible, MacRae et al. (1988) maintain, to analyze
complex systems by examining a few variables and then applying the results
over a broad area, nor is it usually possible to find direct, single-cause-and-
effect relationships between factors. Given that sustainable processes are

Table 1. Approaches to Measuring Sustainability: Some Comparisons (cont.)."
Approach Interpretation Problem category Comments
Total factor produc- Sustainable Internal, Little information on
tivity trends growth plot-level, causes, less confounding
productivity
Trends in total Sustainable Internal, Decomposition may be
factor productivity growth plot-level, difficult or impossible
or yields, decomposed productivity in practice
into trends in input
use, technical change,
or resource degradation
Evidence from Sustainable Internal, Good information on
long-term trials growth plot-level, causes, does not take
on declining yields productivity account of possibilities
of substitution
Other physical Equity, Internal No clear link to issues
evidence on re- sustainable of food security
source degradation growth

(Measurement not Equity; Potentially all Claims sustainability
possible) agroecology cannot be measured
SListing of approaches to measuring sustainability is incomplete. For full description of interpre-
tations of sustainability and categories of sustainability problems, see text. Note that few if any
of these approaches take account of possible farmers' adaptations to sustainability problems, e.g.,
shifting to other enterprises or cropping systems, or substitution among inputs.


Vol. 3, No. 1, 1992






HARRINGTON


location specific, they argue that these processes are inherently difficult or
impossible to quantify.
There is undoubtedly a certain amount of truth to these arguments. Yet,
as a practical matter, it is not humanly possible to deal with any problem (not
just sustainability problems) in all of its real world complexity. In terms of
science, "we have to simplify to survive" (McCall and Kaplan, 1985). In
addition, the experience of farming systems research suggests that often it is
possible to quantify and model complex biological systems without unaccept-
able loss of realism.
In contrast, lack of quantification can lead to circular reasoning, with the
relative sustainability of systems being assessed in terms of the degree to which
they use practices that have been defined a priori as "sustainable." This
increases the probability of self-deception and virtually eliminates the ability
to systematically compare alternative systems, examine sustainability-produc-
tivity trade-offs, or gauge progress being made towards specific goals.
Directional measurements. Most proponents of sustainable agriculture
probably would not agree that measuring sustainability is utterly impossible,
or that trying to measure it is a bad idea. Many, however, would be content
with "directional" measurements. A directional measurement as used here is
one that explicitly measures only the direction of change in the sustainability
of a system, not the magnitude of that change.
Directional measurements may be most attractive when it is felt that a
proportional relationship exists between control and state variables. The
implicit assumption is that the sustainability of an agroecosystem is changed
in rough proportion to changes in those practices felt to most strongly
influence trends in the system's future productivity (and/or its ability to deal
with stresses and perturbations). For example, in this approach an agroeco-
system suffering from gradually declining levels of plant nutrients in the soil
is thought to become more sustainable in rough proportion to the amount of
these nutrients that are generated or recycled within the system or applied
from external sources. Insofar as the levels of these nutrients are increased, the
system is assumed to become proportionately more sustainable.
Note that in this approach, current levels of sustainability need not be
measured. In fact, cardinal units of measurement are unnecessary. This
approach implicitly agrees that "sustainability" is a continuous, not a discrete,
variable but finds that measuring levels of sustainability in cardinal terms is
unnecessary. Insofar as the assumption of proportionality between control


Journal for Farming Systems Research-Extension






MEASURING SUSTAINABILITY


and state variables is incorrect, of course, this approach can lead to thoroughly
misleading results.

Quantitative Approaches
Purpose. Suitable methods of quantification are necessary in order for
researchers to answer questions such as the following:
- Is System A sustainable or not?
- Is System A becoming more or less sustainable over time?
- Is System A more or less sustainable than System B?
- By what percent is System A more sustainable than System B?
- Is the relative sustainability of System A with respect to System B increasing
or decreasing over time?
- What are the trade-offs between longer-term sustainability and current
levels of productivity of System A?
- Is the current productivity of System A more or less sensitive than that of
System B to technical changes aimed at enhancing sustainability?
Analysis of regional trends.s Trend analysis uses time series information
from the recent past to forecast likely events for the near future. Most of the
examples of trend analysis described below focus on a particular state variable:
output, yields, total factor productivity, per capital production, per capital
RAVC, etc. The aim of much of this analysis is to measure the extent to which
a system has already become less sustainable, and the corresponding debate
focuses on choice of state variable. Which ones most precisely and most
accurately capture a decline in system sustainability? Which ones tend to
confound trends in system sustainability with other factors? Which ones are
relatively easy and economical to use in practice?
Aggregate trends in output and yields. There is an understandable tempta-
tion to measure levels of system sustainability in terms of trends in production
or yields. These trends typically can be estimated through published data at
the aggregate (e.g., district or provincial) level. When maize yields show a de-
clining trend, for example, maize researchers understandably become appre-
hensive about possible degradation of resources devoted to maize production.
However, problems of sustainability can be present-and worsening-
even when published data indicate a rising trend in output and yield at the
regional level. Similarly, sustainability problems may be entirely absent when

8 Apart from the conceptual issues associated with trend analysis that are discussed below, there
are other issues of statistical estimation (e.g., how many years of data to use, choice of functional
form, etc.) that are not discussed in this paper.


Vol. 3, No. 1, 1992





HARRINGTON


aggregate data show declining trends in these variables. Changes in aggregate
production or yields may be due to changes in input levels, input quality, or
in the enterprises assigned by farmers to different land types, not to changes
in resource quality (Harrington et al., 1989). For example, yields of a
particular crop may appear to be declining over time because more attractive
crops have replaced it in the more favorable production environments, or
because farmers are using lower levels of purchased inputs. Researchers
should be careful not to confound trends in the productivity of a particular
crop or enterprise with trends in system productivity, nor productivity trends
with trends in resource quality.
Total factor productivity. Lynam and Herdt (1988) suggest that sustain-
ability be measured in terms of trends in total factor productivity (TFP), where
(1) TFP = O / I
with O being the total value of all outputs and I being the total value of all
inputs. A sustainable system would feature a non-negative trend in TFP.
Monteith (n.d.) notes, however, that the "total value of all inputs" can be
a somewhat arbitrary quantity with diverse components whose relative value
may be hard to assess. A declining trend in TFP (as defined above) might be
due to resource degradation or to declining product prices and higher input
prices caused by gradual shifts in government policy. Note that this approach
takes no explicit account of changes in the quality of the agricultural resource
base and does not feature hypotheses on technical factors that may be causing
an observed decline in TFP. As a practical matter, measuring the total value
of all outputs and all inputs used in a farming system is likely to be expensive,
considering that input/output measurement may be needed at several times
during the year (to minimize recall error), for a reasonably large number of
farmers (to minimize sampling error), over an indefinite number of years.
Finally, Monteith notes that this approach focuses on sustainability at the
plot level, while avoiding assessment at the regional level. Increasing levels of
TFP may mean little if population increase outruns productive capacity.
Similarly, past gains in TFP may mean little if made at the expense of system
resilience or in ways that ultimately degrade farmers' resources. "Turning
points" in trends have always been the bane of those who would predict the
future from the record of the past.

9 Induced, for example, by changes in input or product price policies. The confounding effect
described here stems directly from the ability of farmers to switch resources among enterprises
and to substitute one input for another.


Journal for Farming Systems Research-Extension






MEASURING SUSTAINABILITY


Trends in per capital production. Monteith (n.d.) proposes a somewhat
different variation on trend analysis. He argues that to be sustainable, a system
should maintain per capital benefit levels from year to year (and in principle
from generation to generation) and should not itself deteriorate as a conse-
quence of being used. He summarizes this in the following rule: "A system
is sustainable over a defined period if outputs do not decrease when inputs are
not increased."
This conceptualization is intuitively attractive and is fairly transparent as
long as farmers use the same cropping patterns and associated livestock
enterprises year after year on the same fields without increasing input levels.
It seems aimed at subsistence economies with few opportunities for enterprise
diversification or use of external inputs-that is, it ignores opportunities for
substitutions within farming systems.
With input levels held constant, per capital production is a function ofyields,
harvested area, and population density, where yield changes are driven by
"sustainability" factors (resource quality), not by varying input levels or land
use shifts; thus
(2) C = Y (A/P)
where C = per capital production, Y = yield per unit area, A = harvested area,
and P = population density. By differentiating with respect to time, percent-
age changes become additive in the following manner:
(3) dC/dt)/C = (dY/dt)/Y + (dA/dt)/A (dP/dt)/P
where (dC/dt)/C is the percentage change in per capital production with a
small increment in time. In other words, the percent increase in per capital
production is the sum of the percent increase in harvested area and the percent
increase in yields, less the percent increase in population density.
For example, if yields are growing at 3.1% per year, with harvested area
declining at 0.2% per year, and population increasing at 2% per year, then per
capital production is increasing by 3.1 + (-0.2) 2.0 = 0.9% per year. Note that
this approach assumes that parameter values do not vary over time.
Because this approach defines sustainability as the maintenance of per
capital net benefits from year to year (net benefits vary in direct proportion to
gross benefits because inputs are held constant), declining trends in per capital
production are used to identify sustainability issues. However, this approach,
like the previous two approaches, provides little information on the technical
dimensions of any decline there may be in the quality of the agricultural


Vol. 3, No. 1, 1992






HARRINGTON


resource base-the "causes of unsustainability."
This approach can conceivably be applied to whole systems. However,
given the emphasis in the analytical model on harvested area and yield, its use
in enterprise-specific analysis seems virtually inevitable. Indeed, Monteith
himself uses enterprise-specific examples in his paper. However, as noted in
the discussion of the first approach ("aggregate trends in output and yields"),
this increases the danger of confounding "declining trends due to resource
degradation" and "declining trends due to other factors," e.g., shifts of the
enterprise by farmers from one land type to another (less productive) land
type.
Finally, the feasibility of this approach hinges on an ability to hold input
levels constant. This may be difficult in practice. In controlled trials, input
levels can be kept from increasing. At the farm level, however, input levels are
usually found to vary. An approach that cannot assess the sustainability of
systems where both product and inputs are increasing does not seem terribly
helpful.
TFP revisited. The definition of total factor productivity described above
is not the only one-it's not even a standard one. The concept of TFP can be
strengthened by explicitly linking it to a production function.
An acceptable method of quantifying sustainability should be capable of
distinguishing between: (1) yield changes due to changes in levels of pur-
chased inputs (movements along a production function), (2) increases in total
factor productivity due to technological change (e.g., upward shifts in the
production function due to farmer adoption of an improved variety, or earlier
planting), and (3) reductions in total factor productivity due to resource
degradation (e.g., downward shifts in the production function due to nutrient
depletion). This latter case is hypothesized to take the form of stagnant yields
despite continuously increasing input levels, or actual yield reductions given
constant input levels. Note that in this sense, resource degradation can be
regarded as "technological change" with a negative sign.
Total factor productivity has been widely used for the purposes of empirical
measurement of "the effect of technological change." TFP has been opera-
tionalized in several different forms. Samuelson and Nordhaus (1985), for
example, use the following:
(4) TFP = Q SL(L)- SK(K)
where TFP = total factor productivity (% change per year), Q = output growth
rate (% per year), L = labor input growth rate (% per year), K = capital input


Journal for Farming Systems Research-Extension






MEASURING SUSTAINABILITY


growth rate (% per year), SL = (constant) labor factor share, and SK= (constant)
capital factor share.
TFP as defined here, then, is a residual after accounting for the effects of
increased input levels on increased output. As noted above, it includes (and
thereby confounds) positive effects of technological change as well as negative
effects of resource degradation. If it were possible to explicitly identify shifts
in the production function attributable to technological change, then, after
subtracting these from TFP, the new residual would estimate the specific effect
of resource degradation on productivity.
A farmer monitoring program recently begun in Nepal by the National
Agricultural Research Center, International Maize and Wheat Improvement
Center, and International Rice Research Institute takes this approach to
measuring sustainability. A farmer panel is monitored by local research and
extension workers twice per year. Input/output data are obtained, along with
information on field-level productivity problems and assessments of resource
quality. Yields and TFP are then explained (through a set of recursive
regressions) in terms of changes in input levels, technical change, and changes
in resource quality (with weather information included to reduce unexplained
variability). This project is still at an early stage of development. However,
this approach, like the "trends in per capital production approach," focuses on
the plot level, not the regional level. It says nothing about the race between
increased demand on the one hand and sources of productivity growth on the
other.
Yield trends in relation to inputs applied. Not everyone is captivated by the
idea of estimating TFP. Direct estimation of the contribution of different
factors to yield increase might be an unpretentious approximation.
Cardwell (1982), for example, estimated the relative contributions of a
number of factors, both positive and negative, affecting Minnesota corn yields
from the 1930s to the present. Each factor was assessed separately. First, the
contribution of a particular factor in kg/ha or kg/ha/year was estimated
synthetically. Area and numbers of years corresponding to that factor were
used to estimate its current year contribution to yield change, which was then
expressed as a percentage of the current yield.
Cardwell found that the switch from open-pollinated varieties to hybrids,
improved weed control through herbicide use, increased plant densities, and
earlier planting (allowed by a combination offall plowing, better soil drainage,
and herbicide use) accounted for most of the increase in yields. An increase
in nitrogen fertilizer use also accounted for part of the yield increase, but much


Vol. 3, No. 1, 1992






HARRINGTON


of this was merely a substitute for lower levels of manure and reduced levels
ofN from mineralized organic matter. Soil erosion was found to have reduced
yield potential by 8 percent over the 50-year time horizon studied.
Byerlee and Siddiq (1989) followed a more elaborate variation of this
approach in their assessment of sources of growth in wheat production and
yields in the Pakistan Punjab. They identified three major sources of growth:
Increased irrigated area relative to rain-fed area, adoption of HYV's (high-
yielding varieties) in both irrigated and rain-fed areas, and increased HYV
yields in irrigated areas. They also identified factors tending to depress yields:
earlier planting, declining groundwater quality and increased field saliniza-
tion, an increase in problem weeds, and lower fertilizer efficiency. Secondary
factors also were explicitly included and measured. For example, the increase
in irrigated wheat area relative to rain-fed wheat area was found to be partly
due to an increased share of wheat in total cropped area-a shift in cropping
patterns.
This approach is powerful in that it explicitly measures trends in both state
and control variables, controls for land type changes and cropping pattern
changes, controls for changes in input use levels, and identifies both positive
and negative factors affecting yields. This allows researchers more scope for
reliable forecasting of future events.
At this level ofdisaggregation, for example, it may be clear that some of the
past sources ofyield increase (e.g., adoption ofHYVs) have been fully used and
can no longer support further growth. In contrast, it may be found that some
of the negative factors (e.g., salinity) are increasing in importance, threatening
future productive capacity. By integrating all of these factors into a single
model, a powerful tool is forged for assessing yield and production growth for
the near future. These can then be compared with expected demand changes
to develop regional-level assessments of sustainability.
A disadvantage of this approach, however, is that it is extremely data
intensive. It requires a combination of time series data from secondary
sources, and microlevel data from farm surveys and from on-farm and on-
station experiments. In many cases this data will not be available and the
approach will be unusable without a substantial investment in data generation.
Moreover, the approach is even more difficult to apply to complex farming
systems, at the level of the system. The example given, focusing on wheat in
the context of a relatively simple system, was already somewhat elaborate.
Finally (and this is a comment that applies equally to all of the approaches
to analysis of trends), it interprets sustainability in terms of efficiency, not


Journal for Farming Systems Research-Extension






MEASURING SUSTAINABILITY


"resilience," and shows little sensitivity to the possible virtues of diversity as
a generalized solution to the negative trends that were identified. That is,
principles ofagroecology seem to have little place either in the analysis or in
the conclusions.
Linear programming. Where trend analysis forecasts the near future based
on information from the recent past, linear programming can be used to
simulate possible future events given parametric changes in farmers' access to
land and other assets. When farmers can choose between several activities that
have differential effects on sustainability (resource quality), the conditions
under which they might shift from a more to a less sustainable activity become
interesting.
One recent study examined this very question (Hildebrand and Ashraf,
1989). Several alternative cropping activities were assessed, with some of
them assumed to have more beneficial carryover effects than others. Farm size
was parametrically reduced to reflect likely changes arising from population
pressure. An estimate was made of the minimum farm size needed to meet
family food requirements, while maintaining soil fertility through fallow,
alley-cropping, or fertilizer application strategies. Not surprisingly, it was
found that minimum allowable farm sizes were larger when soil fertility
maintenance depended on traditional fallowing and alley-cropping activities.
Activities featuring the use of chemical fertilizer allowed farms sizes to decline
much further without reducing soil fertility below critical levels. Results
specifically highlighted a trade-off between fertilizer application and bush
fallow area. It was not possible, however, to compare the sustainability of one
strategy versus another, due to a lack of time series data.


SUMMARY

In order to operationalize the concept of sustainability, it is necessary to
develop ways to measure it. This concept, however, has taken on a vast array
of definitions and interpretations. In this paper, three distinct interpretations
of sustainability have been discussed: agroecology, equity, and sustainable
growth. Moreover, threats to the sustainability of agriculture take on many
forms. Problems may be internal or external; they may be reversible or
irreversible; or they may affect public health as opposed to the quality of
resources devoted to agriculture.
Approaches to measuring sustainability are sensitive to the interpretation
given to sustainability and the category of sustainability problem under


Vol. 3, No. 1, 1992






HARRINGTON


consideration. Approaches may be nonquantitative (e.g., directional) or
quantitative. The measurement of most external problems (e.g., reduced
availability of external inputs, global climate change) is likely require consid-
erable input from disciplinary specialists. Measurement ofsustainability from
the agroecology perspective is hindered by lack of indicators of system
resilience and diversity.
Considerable attention recently has been given to measuring sustainability
at the plot-level, focusing on resource quality and productivity (as opposed to
public health), in the context of the sustainable growth interpretation. This
paper has discussed several methods that have been used, including the use of
production or area trends; trends in total factor productivity (with or without
decomposition into the effects of changes in input use, technical change, and
resource quality); trends in per capital food production; yield trends in relation
to inputs applied; and others. Some advantages and limitations of these
different approaches have been discussed.
Work on measuring sustainability is still at an early stage. Much remains to
be done by FSRE practitioners in partnership with disciplinary specialists.


REFERENCES

Altieri, M. 1987. Agroecology: The scientific basis of alternative agriculture. Boulder,
Colo.: Westview Press.
Arthur, L. 1988. The greenhouse effect and the Canadian prairies: Simulation of future
economic impacts. In Natural resource and environmental policy analysis: Cases in
applied economics. Boulder, Colo.: Westview Press.
Barbier, E., and J. McCracken. 1988. Glossary ofselected terms in sustainable economic
development. IIED Gatekeeper Series SA7. International Institute for Environment
and Development.
Barker, R., and D. Chapman. 1988. The economics ofsustainable agriculturalsystems in
developing countries. Agricultural Economics Working Paper 88-13. Cornell Univer-
sity, Ithaca, N.Y.
Batie, S.S. 1989. Sustainable development: Challenges to the profession of agricultural
economics. Presidential Address, AAEA Summer Meeting, Baton Rouge, La., July 3-
August 2.
Bishop, R. 1978. Endangered species and uncertainty: The economics of a safe minimum
standard. American Journal ofAgricultural Economics 60(1):10-18.
Byerlee, D., and A. Siddiq. 1989. Sources of increased wheat production and yields in
Pakistan's irrigated Punjab, 1965-2000. CIMMYT Economics Working Paper 90/
04. International Maize and Wheat Improvement Center, Mexico City.
Cardwell, V.B. 1982. Fifty years of Minnesota corn production: Sources of yield increase.
Agronomy Journal 74(November-December).
Chapman, D. 1983. Energy resources and energy corporations. Ithaca, N.Y.: Cornell
University Press.


Journal for Farming Systems Research-Extension






MEASURING SUSTAINABILITY


International Maize and Wheat Improvement Center (CIMMYT). 1989. Toward the
21st century: Strategic issues and the operational strategies of CIMMYT. Mexico
City.
Clark, C. 1976. Mathematical bioeconomics. N.Y.: John Wiley.
Conway, G. 1986. Agroecosystem analysisfor research and development. Bangkok, Thai-
land: Winrock International.
During, A. 1990. How much is enough? World Watch 3(6).
Francis, C. 1986. Resource efficient systems for Third World farmers. Seminar
presentation, Kansas State University, Manhattan, April.
Graham-Tomasi, T. 1991. Sustainability: Concepts and implications for agricultural
research policy. In Pardey, Roseboom, and Anderson, eds., Agricultural resource
policy: International quantitative perspectives. N.Y.: Cambridge University Press.
Harrington, L., P. Hobbs, T. Pokhrel, B. Sharma, S. Fujisaka, and C. Lightfoot. 1990.
The rice wheat pattern in the Nepal Terai: Issues in the identification and definition
of sustainability problems. Journal for Farming Systems Research-Extension 1(2):1-
27.
Heilbroner, R. 1980. An inquiry into the human prospect. N.Y.: Norton.
Hildebrand, P., and M. Ashraf. 1989. Agricultural sustainability as an operational
criterion. Paper presented at the Ninth Annual Farming Systems Research-Extension
Symposium, Fayetteville, Ark., October 9-12.
IFPRI. 1989. Environmental policy for agricultural sustainability: An IFPRI research
thrust for the 1990s. Unpublished draft.
Ingram, J., and M. Swift. 1989. Sustainability of cereal-legume intercrops in relation to
management of soil organic matter and nutrient cycling. Paper presented at the
Intercropping Conference, Lilongwe, Malawi, January.
Jodha, N. 1989. Potential strategies for adapting to greenhouse warming: Perspectives
from the developing world. In N. Rosenberg et al., eds., Greenhouse warming:
Abatement and adaptation. Washington, D.C.: Resources for the Future.
Johnston, G. 1988. The role of economics in natural resource and environmental policy
analysis. In Natural resource and environmental policy analysis: Cases in applied
economics. Boulder, Colo.: Westview Press.
Krutilla, J., and A. Fisher. 1975. The economics of natural environments: Studies in the
valuation of commodity and amenity resources. Baltimore, Md.: Johns Hopkins
University Press.
Lynam, J., and R. Herdt. 1988. Sense and sensibility: Sustainability as an objective in
international agricultural research. Paper presented at CIP-Rockefeller Conference
on Farmers and Food Systems, Lima, Peru.
MacCrae, R., S. Hill, J. Henning, and G. Mehuys. 1989. Agricultural science and
sustainable agriculture: A review of the existing scientific barriers to sustainable food
production and potential solutions." BiologicalAgriculture and Horticulture 6:173-
219.
McCall, M., and R. Kaplan. 1985. Whatever it takes: Decision-makers at work. Engle-
wood Cliffs, N.J.: Prentice-Hall. 25 pp.
Monteith, J., ed. n.d. Can sustainability be quantified? Conference Paper No. 538.
International Crop Research Institute for the Semi-Arid Tropics. Submitted.
Pretty, J., and G. Conway. 1989. Agriculture asaglobalpolluter. IIED Gatekeeper Series
No. SA11. International Institute for Environment and Development.


Vol. 3, No. 1, 1992







HARRINGTON


Rosegrant,M.,and P. Pingali. 1991. SustainingriceproductivitygrowthinAsia: Apolicy
perspective. Social Science Division Paper No. 91-01. International Rice Research
Institute, Manilla, the Philippines.
Samuelson, P., and W. Nordhaus. 1985. Economics. N.Y.: McGraw Hill.
Schmid, A. 1978. Property, power and public choice. N.Y.: Praeger.
Schmid, A. 1989. Benefit-cost analysis: A political economy approach. Boulder, Colo.:
Westview Press.
Tisdell, C. 1988. Sustainable development: Differing perspectives of ecologists and
economists, and relevance to LDC's. World Development 16(3):373-384.
U.S. Department of the Interior. 1989. Estimates of undiscovered conventional oil and
gas resources in the United States. U.S. Government Printing Office, Washington,
D.C.


Journal for Farming Systems Research-Extension








Indigenous Knowledge and Fertilizer
Strategies in Leyte, Philippines:
Implications for Research and
Demonstration Trials'

Daniele Perrot-Maitre and T.F. Weaver2



ABSTRACT
Indigenous resource taxonomies obtained by ethnographic interviews of
lowland rice farmers in Leyte, Philippines are used to develop an unders-
tanding of farmers' fertilizer allocation decisions. The analysis shows that
farmers have complex taxonomies of soils and paddy types and that a
knowledge of these taxonomies is necessary for understanding farmers'
rationale for fertilizer management. How farmers perceive fertilizer
response, the rules of thumb they follow in allocating fertilizer, and their
fertilizer-related concerns are addressed. The paper concludes with a
series of suggestions for incorporating farmers' concerns and knowledge
into the design, management, and evaluation of fertilizer trials.


INTRODUCTION

Lightfoot (1985) has argued that indigenous research is "a viable resource for
developing on-farm experimental methods that...promote farmer
participation...in adapting technologies to specific farming conditions and
providing feedback on basic research needs." He stresses, however, that one
of the main drawbacks of the approach is that indigenous knowledge is hard
to elicit and that it will require researchers to develop new skills and new
procedures for documentation and interpretation of this information.
This paper gives an example of how data from ethnographic interviews are
useful for understanding farmers' knowledge of soils and fertilizer, which in
turn is used to understand their rationale for making resource allocation
1 Paper presented at the 11 th Annual Farming Systems Research-Extension Symposium, Michigan
State University, East Lansing, October 5-10, 1991.
2 Research Associate and Professor, respectively, Department of Resource Economics, The
University of Rhode Island, Kingston. This paper is based on the first author's doctoral
dissertation, "The use of indigenous knowledge in understanding resource allocation at the farm
level: The case of lowland farms in Leyte Island, Philippines. "Rhode Island AES Contribution
No. 92-14.






PERROT-MAITRE AND WEAVER


decisions. In the process, insights are gained that may be used in designing
fertilizer trials and research that are relevant to farmers' concerns.


ETHNOGRAPHIC INTERVIEWS

Ethnographic interviews are a series of conversations for gathering informa-
tion that describes the respondents' view of the world around them. The
approach is based on the premise that naming and classifying things are
fundamental principles of human thinking, enabling individuals to organize
concepts and reduce them to a manageable set (Tyler, 1969). The categories
of these classification systems or taxonomies are assumed to be operationally
significant to an individual in day-to-day decison-making. Hence discovering
them and the rationale behind their construct enables us to understand
decisions regarding the allocation and the manipulation of the items that
constitute the taxonomy. For example, if we know how a farmer classifies his
or her land and the rationale behind the classification, we may gain consider-
able insight into his or her management decisions regarding this resource.
During ethnographic interviews, the task is to ask the "right" questions:
those that will elicit relevant information from the informant about the areas
of interest to the researcher. The sequence of questions is not predetermined
as with conventional surveys, but follows directly from the informant's
answers and knowledge. Discussion and examples of the types of questions
used in this study are presented in Appendix A.


LOCATION OF THE STUDY SITE AND SAMPLE SELECTION

The research was conducted in Hibunawan, a barrio comprised of 131
households in the municipality of Baybay on the west coast of Leyte,
Philippines (Figure 1). This part of Leyte lies in the humid typhoon belt with
a climate characterized by abundant rainfall with no pronounced dry season.
Ninety percent of the 131 households in the barrio are involved in
subsistence and semisubsistence agriculture. The main crops are rice, coconut
palms, and corn, which occupy, respectively, 47, 41, and 12 percent of the
total 254 ha of cultivated land in the barrio. Ethnographic interviews were
conducted with 57 farm households selected by stratified random sampling.
Rice, which is double cropped by 97 percent of these households, is the
economically dominant crop. Half the paddy area is rain-fed whereas the rest
receives supplementary irrigation from the local pump-irrigation system.


Journal for Farming Systems Research-Extension






INDIGENOUS KNOWLEDGE AND FERTILIZER STRATEGIES


Tacloban


Republic Leyte
rC ofthe |
.c Philippines N

I I
300 mi
Figure 1. Location of the Study Site in the Philippines
The rice farms in the barrio are small, averaging 0.9 ha (standard deviation
of 0.98). The area devoted to paddy production ranges from 0.23 to 7.5 ha
with 70 percent of the households in the sample cultivating 1 ha or less.


CLASSIFICATION OF PADDY LAND

For their rice lands, farmers have a taxonomy of paddy types that describes
hydrological conditions, which reflect and are related to soil type, position in
the toposequence, water source, and special features such as proximity to a
canal, a creek, ora hillside. In this classification (Table 1) banika paddies refer
to drought-prone upper paddies with low water-holding capacity. If rainfall
is below normal, these fields may be unsuitable for rice cultivation unless
irrigation is available. Binog refers to lower paddies with bad drainage. Many
of these fields are prone to periodic flooding and/or waterlogging. The term
lanoddescribes an intermediate situation. The risk of crop failure, considering
both drought and flood, is the lowest in these fields. Generally fields lower
in the toposequence are considered the best for growing rice, especially when
the rainfall is not heavy and the risks of flooding are low. This is not only
because of the better water situation but also because the fields benefit from
nutrient runoff from higher elevations.


Vol. 3, No. 1, 1992






PERROT-MAITRE AND WEAVER


Table 1. Characteristics of Paddy Types and Management Practices by Type of Paddy
According to Farmers in Hibunawan.

Paddy type Banikaa Lanodb Binogc


LITTLE RAINFALL
Topography Highest/high
Drainage Dries quickly
Yield Poor
Tillers density Few tillers
Planting
sequence Planted last
Land preparation
technique Carabao only

Seedbed type Dapog
Transplanting
method In rows w/marker
Weeding With knife only
Fertilizer Once or none


ABUNDANT RAINFALL
Yield Good
Tillers density Good number
Planting
sequence Planted second
Plowing
intensity Once or twice
Land preparation
technique Hand-tractor/
carabao
Seedbed type Dapog
Transplanting
method In rows w/mar
Weeding Mechanical/
manual
Fertilizer Once or twice


rker


High/low
Does not dry easily
Good if irrigated
Good number

Planted second

Hand-tractor/
carabao
Dapog

In rows w/marker
Mechanical/manual
Once or none


Good
Good number

Planted first

Once or twice

Hand-tractor/
carabao
Dapog

In rows w/marker
Mechanical/
manual
Once or twice


Low/lowest
Does not dry
Good if no kohol snail
Good number

Planted first

Hand-tractor if too
deep
Wet bed if too deep

Random
Manual only
None


Poor if floods
Few tiller if floods

Planted last

None

None

Wet bed

Random
None or manual

None


SGenerally pilit soils.
b Generally pilit, bukagay, or yutae soils.
c Generally yutae soils.
Source: 1987-88 Field Survey, N=57.


SOIL CLASSIFICATION AND FERTILITY DIFFERENCES

The farmers of Hibunawan distinguish differences in soil fertility both
between and within fields. They associate differences with three identified soil
types: (1) pilit (clay), (2) yutae (clay with silt), and (3) bukagay (clay with
sand). As shown in Table 2, the concept of soil type encompasses texture,
color, fertility, water-holding capacity, workability, and husbandry practices,
including plowing requirements, weeding, and fertilization strategies. Ac-

Journalfor Farming Systems Research-Extension


t






INDIGENOUS KNOWLEDGE AND FERTILIZER STRATEGIES 25

cording to the farmers, yutae soils are the best for growing rice and require very
little or no fertilizer. Pilit soils generally are thought to require two
applications of fertilizer whereas bukagay soils are intermediate to the other
two in terms of their fertilizer needs. Operationally these differences in
fertilizer need are associated as much with paddy types as with soil types. A
Chi-square test of the association between soil and paddy types identified by
farmers showed a strong relationship (Chi-square=29.7, 4 d.f., and 171
observations).3
Based on laboratory analysis, there was no significant difference across
fields between the different types identified by the farmer in the level of
organic matter (and therefore nitrogen (N)), phosphorus (P), potassium (K),
and pH. This finding suggests that the terms for the three soil types (pilit,
yutae, and bukagay) may be used by farmers in a relative sense based on their
personal experience in their fields. Unlike paddy types that are identified by
location variables that are directly observable and easily shared between
individuals, soil types are dependent mostly on texture, which is evaluated by
feel rather than by sight. Without the existence of a shared standard such as
that taught to soil scientists for their textural classifications, the notion that
farmers' soil types are relative concepts seems a reasonable hypothesis. The
hypothesis is supported by the fact that no correlation was found between the
farmers' and soil scientists' soil types4 based on textural classifications (Chi-
square=2.630, 6 df, and 33 observations) and that attempts to produce a soil
map based on the farmers' soil types resulted in a random mosaic. At the same
time the soil map for the area produced by the soil scientists has the ordering
typical of these sorts of efforts. Consistent with this line of reasoning, as
already noted, there was no difference in fertility across fields identified as
more fertile by farmers. However differences in relative fertility that farmers
identified within the same field were supported by laboratory analysis.5





3This association between soil and paddy types is illustrated by the fact that 60 percent of the binog
fields have yutae soils, 67 percent of the banika fields have pilit soils, and 56 percent of the lanod
fields have pilit soils, 30 percent have bukagay soils, and 14 percent have yutae soils.
4 These categories are clay, clay loam, silty clay loam, and sandy clay loam.
5 Approximately 83 percent of the sections of rice fields described as more fertile by the farmer
showed higher percentages of organic matter (six observations) than sections said to be less
fertile. All had higher levels of P (although in only two-thirds of the sample was the difference
significant enough to affect the P status of the soil) and two-thirds had significantly higher levels
of K.


Vol. 3, No. 1, 1992






PERROT-MAITRE AND WEAVER


Table 2. Rice Soils and Associated Cultivation Practices According to Farmers in
Hibunawan.
Soil types
Bukagay Pilit Yutae
Texture Clay with sand Clay Clay with silt
Color Brown/yellow/black Red/brown/black Black/brown
Fertility Intermediate Exhausted soil; least Most fertile
fertile
WHCa Dries very easily Dries easily Does not dry easily
Workability Friable Hard if no water Soft and muddy
Plowing Can plow dry or wet Cannot plow without Can plow dry or wet
water
Weeding Can pull weeds even Need water to weed Intermediate
if dry
Fertilizer Intermediate Heavy None or very little
application (once or twice) (twice) (none or once)
a Water holding capacity
Source: 1987-88 Field Survey, N=57.

FARMERS' FERTILIZER STRATEGIES

Based on ethnographic interviewing, the fertilizer practices of the farmers are
more related to fertilizer interaction with other inputs such as water and with
the perceived risk of crop loss associated with different paddy types than with
soil type itself.
Farmers do not allocate fertilizer according to the marginal productivity
rule. Rather, they decide upon the quantity of fertilizer to apply by following
what is known in the barrio as the ekonomia strategy. Using this strategy,
farmers try to keep the crop "healthy" at a minimum cost. If growth is
favorable, little or no fertilizer will be applied during the entire growing
season. However, if growth is unfavorable and there is enough water in the
field, then cultivators will tend to apply fertilizer in an attempt to have the crop
"catch up." Farmers will not apply high doses of fertilizer on crops that are
constrained from catching up, such as weed-infested fields.
The farmers' taxonomy of paddy land serves as the basis for understanding
fertilizer-allocation decisions. As a rule, lower-lying fields (binog) are not
fertilized because farmers say they are more fertile (partly because of the
abundant water and nutrient runofffrom upper paddies) and because the risk
that the fertilizer will be washed away (hunob) is high. More fertilizer may be
applied on upper fields (pilit and bukagay) to compensate for unfavorable
water conditions over the entire growing season (provided there is enough
water at the time of application and that the field is not infested byweeds). For


Journal for Farming Systems Research-Extension






INDIGENOUS KNOWLEDGE AND FERTILIZER STRATEGIES 27

example, one farmer applies the equivalent of 79 kg of N per hectare on an
upper, drought-prone, less fertile field, and no fertilizer in the lowest,
waterlogged field. The farmer's two other paddies, which present an
intermediate situation, receive the equivalent of 33 to 73 kg of N per hectare.
In summary, farmers' fertilizer allocation strategies are based on their paddy
land taxonomy, which incorporates characteristics in addition to soil types.


TYPES AND CHARACTERISTICS OF COMMERCIAL
FERTILIZERS
The most widely used commercial fertilizer is urea (grade 46-0-0) and, to a
lesser extent, ammonium sulfate (21-0-0). Only one of the farmers applied
complete fertilizer (14-14-14) in the dry season and, according to this farmer,
this was an experiment.
All respondents indicated that too much fertilizer can be detrimental to
plant growth and that diminishing returns occur if too much fertilizer is
applied. Too much fertilizer, it is thought, will cause the plants to lodge
(tomba), grow with fewer panicles, or bear empty panicles (tahopon). Forty-
one percent of the farmers who usually apply fertilizer claimed they knew the
maximum level of fertilizer to apply on their fields. These levels vary widely,
averaging 69 kg of N per hectare (standard deviation of 45.5, 23 observa-
tions). This amount is close to official recommendations from the Philippines
Council on Agriculture and Resources Research (PCARRD, 1979), which
calls for 60-90 kg of N per ha depending on the season (and 30 kg of P and
0-30 kg of K depending on the soil type). But survey data indicate that
respondents who applied N during the two cropping seasons studied applied
on average respectively 40 and 34 kg of N per hectare for the two seasons
(standard deviation of62.8 and 68.6), or 66.3 percent and 56.8 percent of the
recommended 60 kg per hectare. These average levels are higher than usual
because in the survey year farmers who cultivated irrigated land benefited from
a government fertilizer subsidy6 (47.4 percent of the farmers in the first
cropping season and 32.3 percent during the second cropping season bene-
fited from this subsidy). Even with the subsidy, approximately 88 percent of
the farmers applied less than the recommended 60 kg of N per hectare.
6 The goal of the subsidy program, referred to as "Buy One Take One," was to compensate farmers
for the low harvest following the 1987 drought and to boost production levels by encouraging
irrigated rice farmers to apply greater amounts of fertilizer. The program targeted irrigated rice
farmers only and stipulated that for every bag of a type of fertilizer purchased (urea or ammonium
sulfate), farmers would get one bag ofanother type of fertilizer free. In practice, depending on
stock availability, farmers may have received two bags of the same type.
Vol. 3, No. 1, 1992






PERROT-MAITRE AND WEAVER


None of the farmers surveyed applied K and P to their fields. However
results from a laboratory analysis of 33 soil samples indicated that 82 and 79
percent of the soils in the rice fields were deficient in P and K, respectively. On
the other hand, 88 percent of the samples had adequate levels of organic
matter, and therefore did not need N application. Because farmers do not
recognize K and P deficiency as distinct from N deficiency and because
complete fertilizer is more expensive, farmers are reluctant to apply Kand P.
There is almost total agreement among farmers concerning the appropriate
use of the different commercial fertilizers. Urea is said to be good for
"fruiting" (forming full panicles), whereas ammonium sulfate favors tillering.
Huijsman (1986) reported similar findings among rice farmers in Iloilo
province. Urea is considered best for rice because, although it is not
immediately effective (it takes four to six days to see the effects compared to
three days with ammonium sulfate), it has a longer lasting effect (about 90
days) than ammonium sulfate (about 21 days). Farmers in Hibunawan say
that the long-lasting effect of urea means fertilizer need only be applied once,
while ammonium sulfate, whose effect is believed to last only three weeks,
must be applied at least twice during the growing season. Agronomists at
PCARRD also reported that ammonium sulfate has a more immediate effect
on crops than urea (PCARRD, 1979); however, none of their recommenda-
tions address that issue. Recommendations are worded in terms of kg of NPK
per ha and give the optimal time and quantities for split application, without
addressing how quantities, method, and timing of applications may vary
depending on the type of fertilizer used.
According to all farmers surveyed, if only one application of urea is used,
the optimal time to apply it is about a month after transplanting following the
last weeding, because by then weed competition for nutrients has been
eliminated. This is the reproductive stage (mabdos) and the rice plant needs
to be "helped" to form panicles. This is partly consistent with recommenda-
tions from the International Rice Research Institute (IRRI, 1987), which
suggest, among other practices, an application of N five to seven days before
or at panicle initiation.
A widespread belief among Hibunawan farmers is that longer maturing
(120 days) rice varieties need less fertilizer because they have more time to
grow and recover from drought or pest infestation than do quicker maturing
varieties. As a result, the optimal timing of fertilizer application differs
according to the maturity of the variety. Some farmers claim that the best
strategy is to use a split application with the second application earlier for the


Journal for Farming Systems Research-Extension






INDIGENOUS KNOWLEDGE AND FERTILIZER STRATEGIES 29

short-season varieties, but there is no consensus on the timing. Some advocate
that the second application should be 45 days after transplanting for short-
season varieties and 60 days after transplanting for longer maturing ones.
Others favor 21 days and 30 days in these two situations.
PCARRD recommendations are somewhat confusing regarding any rela-
tionship between rice varieties and fertilizer application. There are two sets of
recommendations. One set (PCARRD, 1979) indicates the optimal level of
NPK per hectare according to the percentage of organic matter, K, and P found
in the soil, and the "group" to which rice varieties are assigned. The number
and timing of applications or the effect of maturity on application strategies are
not addressed. Also rice varieties are not classified by maturity, and the
characteristics that distinguish them are not specified, so that it is impossible for
the uninformed to extrapolate into which group new varieties will fall.
The second set of recommendations for rice (PCARRD, 1979) addresses
the issue of timing and number of applications according to the method of
crop establishment (transplanted versus direct-seeded), the water regime
(irrigated versus rain-fed), the season (for irrigated fields only), and the soil
types (clay versus loamy soils for the wet season only). The effect of rice
varieties on the method of application is not addressed in these recommenda-
tions. All of these recommendations follow from basal application before the
last harrowing without mentioning how recommendations might change
when farmers do not apply fertilizer that early in the season. Survey data
indicate that, in practice, no farmer in Hibunawan applies fertilizer before the
last harrowing. Farmers explain that the risk of crop failure is too great that
early in the season to invest in fertilizer, even in irrigated fields.
In summary, the recommendations, from both IRRI and PCARRD, do not
address the issues raised by the farmers, such as the location of the fields, weed
density, and risk factors associated with the quantity, timing, and number of
fertilizer applications. Likewise, no recommendations address the possible
effect of the type of fertilizer (urea or ammonium sulfate) on fertilizer strategies.
The fact that the two sets of recommendations available during the research
period did not match was a source of confusion for farmers and added to the
fact that these types of recommendations are largely irrelevant to their concerns.
In general, the farmers of Hibunawan do not readily discuss abstract
situations related to farming, i.e., situations they have not encountered in their
fields. The levels ofN application they are willing to discuss are the ones with
which they have experience. Using ethnographic interviews, farmers were
asked to recall, for specific fields, what fertilizer levels they had experience with


Vol. 3, No. 1, 1992






PERROT-MAiTRE AND WEAVER


and how different fertilizer levels affected yields. The descriptive experience
question was worded with no specific quantity of fertilizer because, given the
diversity of fertilizer levels and field areas experienced by farmers, it was
impossible to word a general question suitable to all situations.
Twenty-one percent of the farmers reported that over the years they have
experimented with more than one level of fertilizer application. From these
farmers, an equal proportion (25 percent) reported that under higher levels
of fertilizer they observed greater yields for all their fields, or higher yields in
some fields and lower yields in other. A third of these farmers said their yields
were lower in all their fields, and 17 percent reported no change in yield level.
However, farmers with higher yields typically attributed them to the type
rather than the quantity of fertilizer, the season, or a switch in rice variety.
Farmers with lower yields explained it by the presence of tungro, and/or by
application of too much fertilizer, so that in the end no farmers attributed the
observed yield changes to the quantity of fertilizer itself.
Given that farmers who experimented with different levels of fertilizers did
not explain the change in yield by the effect of fertilizer alone, an attempt was
made to determine indirectly the perception of fertilizer response by asking
farmers what were the best and worse yields they experienced in each field, and
what they thought were the reasons for these yields. These results are given
by order of importance in TalSle 3.
Water, rice variety, pests, and tungro infestation rank the highest in farmers'
list of variables affecting yields. Good water and/or good rice variety were
cited in 90 percent of the cases as the reason for high yield; inadequate water,
tungro, and pests were mentioned in 77 percent of the cases as an explanation
for poor yields.7 Fertilizer-related reasons were mentioned in only six percent
of the cases as being responsible for high and low yields.
Again, fertilizer was never cited alone but always in combination with other
inputs. Water, rice variety, and fertilizer levels interact and farmers recognize
this interaction. According to the farmers, if there is little water during the
reproduction stage, the crop is the most affected both because of the lack of
water and because it is impossible to apply fertilizer. Therefore low application
alone does not explain low yields. This reinforces the conclusion that farmers
do not separate out the effect of fertilizer because they tend to apply this input
in combination with water.
Other impressions and conclusions gleaned from ethnographic interviews
were reinforced at a seminar given in May 1988 by an agronomist from the
7 This percentage does not include cases where water, pest, and tungro interact with other
variables.
Journal for Farming Systems Research-Extension






INDIGENOUS KNOWLEDGE AND FERTILIZER STRATEGIES 31

Table 3. Farmers' Perception of Factors Affecting Rice Yields, Hibunawan.
Factors Percent mentioning
HIGHEST YIELDa
Good water 55
Good variety and good water 25
Good variety 10
Other 10

LOWEST YIELDc
Drought or not enough water 53
Tungro 18
Pest 6
Other 26
a Highest yields reported for 98 parcels.
b Includes 'right quantity of fertilizer and good seeds,' 'Good water and dean field,' 'Right type
of fertilizer and good rice variety,' 'good water, good seeds, and dean field,' 'Good water, good
seeds, and right quantity of fertilizer,' and 'Seedbed type, spacing, and good water.'
c Lowest yields reported for 119 parcels.
d Includes 'Drought and pest,' 'Drought and not enough fertilizer,' 'Typhoon,' 'Drought and
tungro,' 'Drought and bad rice variety,' 'Waterlogged field,' 'Wind before harvest,' 'Weed
infestation,' 'Several transplantings,' 'Tungro and other pests,' 'Field not well prepared,' and
'Drought, no fertilizer, and no time to weed.'
Source: 1987-88 Field Survey, N=57.

Visayan State College ofAgriculture about the use of fertilizer. The questions
farmers raised revealed that they were not as concerned about optimal
quantities of fertilizer as they were about the best way to apply a given quantity
of fertilizer. The questions were of the type "How do I apply one sack of urea
fertilizer if all I can afford is one sack?," "Is it okay to mix urea and ammonium
sulfate together?," "Is it better to apply in two applications instead of one?,"
and "If there is no rain, how long should I wait for the second application of
fertilizer?" Farmers also showed great concern about applying too much
fertilizer, which they believe results in empty panicles, exhausted soil fertility,
and increased soil acidity, and gets them caught in a vicious circle of needing
to apply more and more fertilizer to maintain fertility.


IMPLICATIONS FOR FERTILIZER TRIALS

Overall, the ethnographic interviews with the farmers of Hibunawan provide
information that strongly suggests that farmers' concerns are quite different from
the issues traditionally addressed in fertilizer trials. Because recommendations are
developed from the results of these trials, they bypass many of the factors the
farmers consider to be relevant and are therefore of limited practical use. The
concerns raised by farmers in this study suggest new directions for designing and


Vol. 3, No. 1, 1992






PERROT-MAITRE AND WEAVER


evaluating fertilizer trials as well as for research. In addition, the ethnographic
approach has provided an understanding of the cultivators' view of the produc-
tion process that can provide the context for the discussion of current practices
or the introduction of new ones through workshops and published materials.
The following are examples of the types of fertilizer trials and research that
are suggested by the practices and concerns of the farmers of Hibunawan.
1. Farmers' knowledge ofmicrodifferences in their fields should be taken
into account when locating fertilizer and demonstration trials. In Hibu-
nawan, trial design might include the paddy types recognized by the farmers
(banika, lanod, or binog).
2. The rules of thumb farmers use to manage their fields should be tested,
such as applying more fertilizer on banika paddies although the marginal
productivity of fertilizer is probably lowest on these fields and applying less
fertilizer on the (supposedly more fertile8) low-lying binog paddies because of
the high risk of flooding. At the present time there is insufficient information
either to confirm or refute the farmers' rules of thumb relating to their percep-
tion of the marginal value product from the application of N to their rice crop.
3. Field trials or demonstrations might investigate farmers' perceptions
that urea and ammonium sulfate have different effects on the crop. For
example, the source of N is said to have an effect on tillering and panicle
initiation with urea being better for panicle development and ammonium
sulfate for promoting tillering. Fields trials also could demonstrate whether
urea and ammonium sulfate require different timing of application.
4. Farmers' confusion over the implications of mixing types of N fertilizer
might be clarified through field trials and/or published materials.
5. An analysis ofHibunawan farmers' apparently universal opinion that the
application of fertilizer at planting time is too risky to recommend the practice
might be instructive to a wide audience.
6. The effect of maturity of rice varieties on fertilizer strategies needs to be
demonstrated and explained to the farmers.
7. A central issue to Hibunawan farmers facing budget constraints is how
to make the best use of quantities of fertilizer that are less than the recom-
mended optimal levels. For example, farmers can be instructed as to how they
can best use one bag of urea and one bag of ammonium sulfate and under
which circumstances they can be mixed together.
8 This view was not supported by the soil sample analysis for organic matter, K, and P, which showed
no difference on average for the three soil types. It is possible that differences in microelements
not tested for, such as zinc, might explain the higher fertility attributed to the lower fields. Another
possibility is that the risk of crop failure from flooding is great enough that indeed it does not pay
to fertilize the crop.
Journalfor Farming Systems Research-Extension






INDIGENOUS KNOWLEDGE AND FERTILIZER STRATEGIES 33

CONCLUSION

This paper shows that the technique of ethnographic interviewing is useful in
eliciting farmers' resource taxonomies that provide insights into the farmers'
rationale for input allocation. The insights gained are relevant to farmers and
can be directly incorporated into FSRE activities. The relevance to the design
and evaluation of fertilizer trials and research has been addressed in this paper.
The discussion focused on how information gathered during ethnographic
interviewing was useful in developing an understanding of the farmer's
rationale for managing and applying fertilizer to their rice crops. The
implications for fertilizer trials and research are important. Incorporating
farmers' knowledge and practices not only into farmer-designed but also into
farmer-managed trials would be a fruitful way to develop technologies that are
truly relevant to farmers' circumstances. Because they are developed and
tested according to the farmer's perspective, these technologies present a
familiarity to the farmer that makes them more readily adoptable.
Perhaps of greater significance is the fact that by knowing the farmers'
perceptions of their resources, a basis for communication between extension
workers, scientists, and farmers has been established. This wll surely help
facilitate the development of new technologies and management practices.


REFERENCES

Huijsman, A. 1986. Choice and uncertainty in a semi-subsistence economy. Wageningen, The
Netherlands, 336 pp.
International Rice Research Institute (IRRI). 1987. Rice production course. Volume 1,
Training and Technology Transfer Department, IRRI, Los Bafios, Philippines.
Lightfoot, Clive. 1985. Farmer participation in on-farm trials. Paper presented at the Fifth
Annual Farming Systems Research-Extension Symposium, Manhattan, Kans., October
13-16.
Perrot-Maltre, D. 1992. The use of indigenous knowledge in understanding resource
allocation at the farm level: The case of lowland farms in Leyte, Philipines. Unpublished
Ph.D dissertation, University of Rhode Island, Kingston.
Philippines Council for Agriculture and Resources Research and Development (PCARRD).
1979. The Philippines recommends for soil fertility management, 1979. Los Bafios,
Laguna, 108 pp.
Spradley, J.P. 1969. The ethnographic interview. N.Y.: Holt, Rinehart and Winston, 247
pp.
Tyler, S.A., ed. 1969. Cognitive anthropology: Readings. N.Y.: Holt, Rhinehart and
Winston, 521 pp.


Vol. 3, No. 1, 1992






PERROT-MAITRE AND WEAVER


APPENDIX A: EXAMPLES OF ETHNOGRAPHIC QUESTIONS.

Ethnographic interviewing uses three main types of questions (Spradley,
1979): (1) descriptive questions that aim to identify classification systems, (2)
structural questions that aim to verify taxonomic relationships, and (3)
contrast questions that elicit the differences between folk terms. The primary
varieties of questions that were employed in this study were four types of
descriptive questions (grand tour, mini tour, experience, and native language
questions), three types of structural questions (verification, cover term, and
included term questions), and three types of contrast questions (verification-
contrast, dyadic, and rating questions). Examples of the different types of
questions are given below.

Descriptive Questions
Grand tour question. "I would like to know what your farm is like. Could
you please describe it for me?"
Mini tour question. "Could you explain to me what you do when you plow
your field?"
Experience question. "Could you explain to me what happened when you
applied fertilizer to your rice field?"
Native language question. "Could you explain to me what the word binog
refers to?"

Structural Questions
Verification questions. "Are there different kinds of soils in your farm?"
(domain verification question). "Are there other ways to plow a field?"
(included term verification question). "You mentioned earlier that sandy soils
were hot. Then could you say that clay soils are cold?" (semantic relationship
verification question).
Cover terms questions. "Are there different kinds of fertilizers?"
Included terms questions. "Are there other kinds of soils in your farm?"

Contrast Questions
Contrast verification question. "Would you say that all binog paddies are
fertile and all banika paddies unfertile?"
Dyadic contrast question. "What are the differences between pilit and yutae
soils?"
Rating contrast question. "Which type of paddy is best for growing rice?"


Journal for Farming Systems Research-Extension








Gender Issues in African Farming:
I. A Case for Developing Farm Tools for Women'

R.N. Kaul and A. Ali 2



ABSTRACT
Women in Africa contribute up to an estimated 70 percent of the labor
involved in food production and nearly 100 percent in rural food
processing. Their range of daily activities (food preparation, water and
fuel transportation, farm weeding, crop threshing, and dairy product
preparation) is far wider than men's and demands high energy input.
Despite awareness of this, most technologies developed for small-scale
farmers are geared to men with no concern for their appropriateness for
women, who possess different physiques and energy capabilities in
comparison to men.
This paper examines farm tools and processes thatarepredominantly used
by women in Africa and explains, with available research data, the
inadequacies of such tools and processes when linked to women's
physiques and energy levels. Specific examples of tool design-parameter
selection for operations such as weeding and rural milk processing are
highlighted to show the applications of the research. They also illustrate
the need to design and develop farm tools specifically for female operators
in order to improve overall ease of use, safety, and effective integration of
women in farming system innovations.


INTRODUCTION

Farming in developing countries is a complex interaction ofmany components
that have evolved into various systems in different ecological regions. One
thrust in farming systems research-extension (FSRE) is to understand and
unravel these complicated systems so that, wherever possible, innovations may
be introduced without disturbing the equilibrium of the systems.



1 Paper presented at the Ninth Annual Farming Ssytems Research-Extension Symposium,
University of Arkansas, Fayetteville, October 9-11, 1989.
2 Institute for Agricultural Research, Ahmadu Bello University, Zaria, Nigeria.






KAUL AND ALI


Gender issues in farming have emerged as a strong focus in FSRE. It is
reported that two-thirds of the work in the world is done by women. In return
for their work they receive a mere 10 percent of all income and own only 1
percent of all means of production (Moller, 1987). Many studies have
established that women perform nearly every possible task in rural African
communities. Some of these activities, such as transportation and the
drudgery associated with it, are not fully appreciated. An analysis of transport
activities of able-bodied adult males and females, for example, revealed that an
average female spends 1,000 hours in Ghana and 1,600 hours in Tanzania per
year physically transporting items. This constitutes 70 percent of transport in
terms of time and 80 percent in terms of effort (Harrison and Howe, 1989).
Studies by Adeyokunnu (1981) showed that 40 percent of all women
studied in Nigeria could be classified as farmers, 28 percent as processors, and
52 percent as traders. In another study on Yoruba women in Nigeria, a time
budget analysis showed that all village women spent some of their time
farming (Spiro, 1981). Abalu (1984) identified the "cropping chain" as
typical of farming systems in sub-Saharan Africa, where women constitute a
vital link in the chain for optimal resource management.
The issue is not strictly one of gender differentiation but involves a much
broader scope: Who does what, why, when, and where (Poats, 1988)?
Women's work domain is far wider than men's. Agricultural modernization
programs have tended to increase women's work load but seldom have
attempts been made to introduce labor-saving devices for women's use in their
traditional tasks. In fact, when a farm labor-saving gadget is developed, it is
usually introduced for use by male workers. Sometimes, after a new gadget
is introduced, a task formerly done by women is taken over by men, resulting
in the loss of livelihood for women, who otherwise constitute a vital earning
member of the family.
Upsetting gender roles in traditional tasks will have negative effects in the
long run. Women's participation in rural development needs to be treated as
a goal in its own right. This has become a topic of great concern in FSRE.


THE GENDER ISSUE

As has been briefly mentioned, the so-called gender issue is not only gender
related-it involves looking at a typical system in which men, women, and
children are integrated into a working unit with an unwritten understanding
of the division of labor. The system, which is skewed towards greater total


Journalfor Farming Systems Research-Extension






DEVELOPING FARM TOOLS FOR WOMEN


work hours by women, is operational although not fully efficient. When we
introduce modernizations without taking gender issues into account, the
system becomes more disadvantageous for women. For example, introduc-
tion of better seeds and fertilizer has increased the work load of women
because of closer plant spacing and extra hours needed to weed and process
the additional food produced. A similar pattern is observed in the animal
,production sector and involves farming and domestic activities that may
necessitate a 16-hour work day (Hoskins and Weber, 1985). In the African
context, there is a sexual distribution of farm labor that appears to be task and
crop specific, as reflected in the work of Igben (1980) and Dixon-Mueller
(1985). (See Tables 1 and 2, respectively, which are based on their work and
which show the division of labor by activity and crop between the two
genders.) The involvement of the whole family in the various tasks and the
variations in labor participation for the different tasks and crops are significant.
A different pattern may apply to other ecological areas but the division of labor
prevails. It is therefore important that any technological improvement
acknowledge this division of labor so that there is no imbalance created in the
existing system by having only males benefit from new technology.
Most attempts to introduce better farming tools further complicate this
issue. The tasks that were traditionally done by women are passed on to men,
who with the new equipment find the task easy and remunerative. This leaves
the more difficult tasks, such as grinding and internal transportation, for
women to perform. Thus a new dimension is introduced into this system,
displacing women from their traditional tasks and leading to subsequent loss

Table 1. Proportion of Labor Input Contributed by Family Members in Plateau
State.

Person-hours contributed bya
Activity Men Women Children Relations Total
Land clearing 39.38 26.94 19.17 14.51 100.00
Land tilling 43.18 23.96 20.33 12.53 100.00
Planting 30.41 22.22 31.58 15.79 100.00
Weeding 39.00 27.76 18.00 15.25 100.00
Fertilizer
application 36.67 23.33 19.33 20.67 100.00
Harvesting 36.13 24.51 22.26 17.70 100.00
Transporting 35.77 30.66 18.97 14.60 100.00
Total average 38.21 25.72 20.75 15.32 100.00
a In percentages.
Source: Igben, 1980.


Vol. 3, No. 1, 1992






KAUL AND ALI


Table 2. Female and Male Labor Contributions to Staple Crops Among the Tiv,
Central Nigeria.'

Crop Field Planting Weeding Harvesting Processing Storage
preparation
Yams F: 50 F: 80 F: 100 F:100 F: 100 F:100
M: 50 M: 20
Millet F: 20 F: 100 F: 50 F:100 F: 50
M: 100 M: 80 M: 50 M: 50
Sorghum F: 20 F: 100 F: 50 F: 100 F: 100
M: 100 M: 80 M: 50
Cassava F: 25 F: 75 F: 100 F: 75 F: 100 F: 100
M: 75 M: 25 M: 25
Maize F: 25 F: 90 F: 100 F: 90 F: 100 F: 100
M: 75 M: 10 M: 10
Rice F: 10 F: 100 F: 100 F: 50 F: 50 F: 100
M: 90 M: 50 M: 50
Sesame F: 50 F: 100 F: 50 F: 100 F: 100
M: 100 M: 50 M: 50
Water- F: 25 F: 25 F: 100 F: 25 F: 100 F: 100
melon M: 75 M: 75 M: 75
Cowpeas F: 25 F: 25 F: 100 F: 100 F:100 M:100
M: 75 M: 75


a All measurements are percentages.
Source: Dixon-Mueller, 1985.


F = female; M = male.


of income. In the African context, this involves extra hardship for the family
because women contribute greatly to family income and, in most cases,
control their income for specific needs not shared with men.
The issue, therefore, is to find appropriate technology for women within
the existing farming system so that they can perform their traditional tasks as
well as upgrade male tasks. This approach would elevate the existing system
to a higher level of efficiency without creating a gender imbalance in terms of
tasks. The other strongly related issue in such an approach concerns social and
governmental policies relative to rural development. This paper, however, will
concentrate on appropriate technology aspects.


GENDER-SPECIFIC TOOLS

In view of the fact that nearly 8 0 percent ofThird World farmers are small-scale
farmers, there is a global focus on developing appropriate and relevant
technology for this group. However, even when developing "relevant"
technology there is a tendency to ignore the well-established gender-based
task division of small-scale farmers, as highlighted above. Therefore, ironical-


Journal for Farming Systems Research-Extension






DEVELOPING FARM TOOLS FOR WOMEN


ly, the so-called "relevant" technology usually is not acceptable to female
users. Because of their socially underprivileged status, the women simply
abandon the task to men or continue performing the task with the male-
oriented equipment that is difficult for them to use. A few illustrations support
this argument.
1. Governmental and other agencies have focused attention on rural water
supply to "ease the hardship of women." However, in practice many pumps
are not technically satisfactory for constant use and require increased energy
beyond most women's capacities. Also, some unfamiliar movements, such as
foot pumping, generally are not acceptable (Hoskins and Weber, 1985).
Similarly, a pedal-operated grinding machine that was disseminated in central
Africa was rejected because women had to sit astride it. This exposed part of
their legs, a posture that is considered indecent for women in this area.
2. Although the introduction of the water pump may have eased lifting
activities by women, little attention has been paid to the main chore of
transporting the water after it is pumped. Also, the long line at the pumping
point further delays women's work schedules.
3. While trying to encourage women to use a commercial maize sheller driven
by a foot-operated pedal, it was noted that women's output was about half that
of the male operators because women found the drive hard to use and also felt
uncomfortable feeding the hopper, which was too close to chest level
(Vermeulen and Kaul, 1979).
4. A hydraulic palm press in Nigeria was adopted initially by 72 percent of the
villages but after a year the figure dropped to 24 percent because the machine
was too taxing for the women to operate and because fiber obtained in the
traditional extraction method, which was used as a source of heat, was not
produced by the improved press. Also the new method upset the schedule for
other tasks.
5. Many "improved stoves" were rejected by women because they did not
cater to the prevailing pot sizes. Other varied cooking devices required an
unacceptable standing posture or were difficult to light and maintain.
6. Any innovation introduced must ensure that it does not radically create an
imbalance in the tasks allocated between men and women. For example, as
a result of the introduction of a hard press for palm nuts, the sphere of activity
shifted to men, whose work hours increased from 60 (before introduction of
the device) to 1,050 while women's and children's work hours decreased from
1,450 to 992 (Afonja, 1984).


Vol. 3, No. 1, 1992






KAUL AND ALI


7. Women have been relegated to using technologies with lower levels of
efficiency, for example, manual transplanting, collecting firewood, and even
walking to the market while men ride bicycles or drive carts.

Isolated Examples of Recognition of the Specific Needs of Women
Some projects have recognized the specific needs of women in developing
new technologies. Concerned about depriving women of weeding as a source
of income, the International Crop Research Institute for the Semi-Arid
Tropics (ICRISAT) did not promote chemical weed-control measures in
India. Instead, a rather labor-intensive double-cropping technology for
vertisols was developed (Mueller and von Oppen, 1985). In developing small-
scale threshing machines, the separation of chaff from grains was not incorpo-
rated into several prototypes because separation and winnowing are not tiring
tasks. The cost of the machine also was reduced because incorporating a
separation device would have involved sieves and fans and their regulated
motions (Kaul, 1987).

Ergonomic Considerations in Developing Small Equipment
Any improved farming tool should involve a design criteria based on
engineering and biological material (such as some crop or soil), the existing
machine design, and the gender of the operator. The three factors (material,
machine,person) are interrelated and must be matched for optimal effective-
ness. Often the role of operator is not fully addressed, i.e., the operator usually
is assumed to be a man.
Ergonomics considers two main aspects:
1. The energy demand of the tool should be within the limits of both male
and female operators' capacities. Energy assessment can be done by using a
physiological index, such as oxygen consumption, heart rate, ventilation rate,
etc. (Karpovich, 1971).
2. Anthropometric aspects use the dimensions of segments of the human body
(arms, legs, etc.) and the tool configuration and dimensions to match the
operator's dimensions for ease of operation. Broadly, it involves developing
machines to fit the operator's convenience.
Such considerations are well established in the aircraft industry (cockpit and
passenger seating), automobile design, and even workshop tools. Lately, this
has been extended to agricultural equipment in the form of tractor cabins,
tractor seats, location of controls, etc. Unfortunately, in the realm ofmanually
operated and small equipment (motorized or animal drawn), this has seldom


Journal for Farming Systems Research-Extension






DEVELOPING FARM TOOLS FOR WOMEN


been recognized, even though most farmers in the developing countries use
these tools. Blacksmiths, who typically fabricate the tools, seldom pay
attention to the specific needs of women due to their own bias towards men
as users. Women either avoid the improved tool or use it under duress, to the
detriment of their health and with reduced performance.
Some efforts at the Institute for Agricultural Research (IAR), Samaru,
Nigeria recently have been directed towards gender issues, and researchers are
looking more specifically at the special tool needs of women.

Male-Female Body Segments and Their Use in Tool Design
In anthropometric terminology, most body segments can be expressed as
a segment-link in terms of a person's overall height. Link parameters are used
to build a segment-link ratio diagram for a known sample population.
Basically, this diagram expresses the length of various linkages in the body
(such as arm length) in terms of the total height.
Wagami (1983) studied the parameters of 450 male and 450 female
workers in nine localities in the savannah region of northern Nigeria. Table
3 gives the difference in male and female workers based on an average of 450
persons in the nine locations. Taking a 95t percentile value of the user
population, the dimensions suiting a male or female segment ratio can be
established. This also can be related to appropriate energy levels and work-
performance criteria-that is, a particular work posture will demand a specific
energy level.
Just as sitting and standing posture demand different energy levels, the
work posture induced by a tool (due to its shape, weight, etc.) can demand
different human energy levels. For example, Nwuba and Kaul (1985, 1987)
reported that long-handled manual tillage tools were less energy demanding
than short-handled tools. However, by bending over, it was possible for

Table 3. Difference in Typical Segments of Male and Female Workers.
Segment/parameter Male Female
Weight (kg) 63.18 53.69
Height (cm) 166.11 154.19
Arm reach (cm) 76.48 71.38
Hand length (cm) 19.08 17.72
Hand width (cm) 9.99 9.24
Forearm (cm) 28.87 26.93
Elbow to ground (cm) 100.58 91.69
Pelvic height (cm) 100.70 92.87


Vol. 3, No. 1, 1992






KAUL AND ALI


farmers to be more effective in shaking weeds lose from the soil after they were
uprooted by a hoe.
The optimal ratios of typical tool-handle dimensions (length, diameter,
etc.) can be used to ensure convenience in operation. Proper tool-operator
matching can indirectly reduce stresses on the back and other joints, which
otherwise can result from bad work postures imposed by poorly designed farm
tools. For instance, in a particular sample population it was observed that
choosing a tool handle length of 0.60H gave convenience of operation. If the
95th percentile height (H) of the user population is 154.19 cm, the correct
handle length is 92.5 cm (0.60H or 0.6 0 x 154.19). This would involve
increasing the existing tool length by over 50 percent. Similarly, tool diameter
can be related to the circumference of the circle that a hand (palm) traces when
making an approximate circle while gripping the handle. This ensures a firm
grip on the tool handle. Taking the length ofhand as 0.11H, the tool diameter
can be calculated.
Using the above criteria, optimal values of existing and proposed dimen-
sions have been worked for some tools (Table 4).
Table 5 gives comparative energy basis requirements for performing certain
farm tasks for both male and female workers under three environmental
regimes (Abdulkarim and Kaul, 1986). Using existing tools, the energy
requirements for female operators are very high and, in fact, exceed the
permissible limits of 4.5 kJ/min for most operations.
Table 4. Typical Existing and Proposed Values for Selected Segments and Tools for
Female Workers.
Linkage Segment- Existing Proposed
link ratio (cm) (cm)
Handle length
for weeding hoe 0.6 Ha 59.98 92.51
Handle length
for sickle 2 x 0.06H 12.65 18.50
Length of
pounding pestle 0.6 H b92.51
Diameter of weeding
hoe handle 0.11H 2.28 5.40
Diameter of sickle
handle 0.11H 2.59 5.40
Grip length of
grinding stone 0.33 x 0.11H b 5.60
a H= Height.
b 'Not included in survey.
Source: Abdulkarim, 1986; Bassi and Kaul, 1986.


Journal for Farming Systems Research-Extension






DEVELOPING FARM TOOLS FOR WOMEN


Table 5. Energy Requirement per Unit Output for Three Different Tasks With Male
and Female Operators Under Three Work Environment Regimes.
Men Women
Environment Output Energy/ Output Energy/ Higher
output output value for
(kJ) (kJ) women (%)
Task: Poundinga
20/48 42.6 1.03 30.3 1.30 20.8
25/85 35.2 1.01 22.8 1.65 38.8
36/63 26.5 1.50 17.8 2.07 27.5
Task: Weedingb
20/48 12.5 3.76 9.8 4.94 23.9
25/85 11.3 4.02 9.2 5.57 27.8
36/63 10.1 4.76 8.6 6.25 28.2
Task: Ridginga
20/48 2.0 32.85 0.8 68.66 52.2
25/85 1.8 35.14 0.9 63.82 44.9
36/63 1.6 40.99 0.9 62.02 33.8
a Outputin m3
b Output in m2.

Efforts are underway to redesign the parameters of these tools for women
and then to undertake field evaluations. A prototype of an improved weeding
tool has been fabricated. Three female subjects were used for comparative
evaluation between the new tool and the old conventional type. A summary
of preliminary results is given in Table 6.
The results show that even though the traditional type covered more area,
its energy expenditure was more than the improved type for the same work
duration. Subjective responses confirmed that the improved tool was less
fatiguing but that the output was lower because the tool was still unfamiliar
to the users. Additional tools are being developed to test further the concepts.
Work scheduling and concepts of rest intervals are fairly well established in
the industrial sector. Unfortunately in the agricultural sector these consider-
ations are seldom viewed seriously. Preliminary farm studies (Abdulkarim and
Kaul, 1986) have revealed that ridging drudgery was reduced by 16 and 8
percent, respectively, for male and female workers by introducing rest pauses
and simultaneously gave an output advantage of 48 percent for male and 39
percent for female workers over the traditional work schedule.
In Nigeria most milk is produced by the Fulani, a nomadic tribe with about
9 million head of cattle. Milk yields are very low-about 0.74 1 per cow per
day. The domestic milk production-to-total consumption ratio (self-suffi-
ciency) has been declining, from, for example, 64.2 percent in 1970 to about


Vol. 3, No. 1, 1992






KAUL AND ALI


Table 6. Comparative Evaluation of Modified Weeding Hoe and Conventional Hoe.
Conventional hoe Modified hoe
Subject Heart rate Area Subject Heart rate Area
(beats/min) (m2) (beats/min) (m2)
1 140 1.8 1 144 1.2
2 168 2.6 2 152 1.6
3 128 3.2 3 120 1.6
Average 145 2.5 Average 138 1.47
1 132 1.8 1 124 1.6
2 160 2.8 2 156 1.8
3 136 2.6 3 132 2.4
Average 142.6 2.4 Average 137.3 1.93
1 144 2.0 1 140 1.6
2 156 2.6 2 120 1.8
3 126 2.2 3 120 2.0
Average 142 2.27 Average 126.6 1.8



35.6 percent in 1983 (Nwoko, 1986) due to neglect by Fulanis of milk
production. Milk is processed mostly by Fulani women, who use very crude
tools and techniques. Little fresh milk is sold due to low yields; instead it is
stored and processed into cream and butter using traditional tools. For
example, a bottle-like gourd (local calabash) is used for shaking cream
collected from milk left overnight in another open vessel. The milk is shaken
for 20-35 minutes depending on weather conditions. Shaking takes longer
during the cold weather, and usually the Fulani women take 2-3 minutes for
rest after 5 minutes of continuous shaking. The churned cream is then stirred
with a primitive milk stirrer for collection of butter globules, which are pressed
by hand into butter.
The whole process has a very low output and high energy demand. In a
study by Eleseku (1989) on energy evaluation of a selected Fulani subject at
IAR, the churning exercise required an energy level of 2.54 kJ/l of cream
churned. An improved device using a simple barrel, which could be rotated
axially to induce limited centrifugal action, was developed, reducing the
energy level to only 0.89 kJ/l of cream, with an advantage of 64 percent over
the traditional tool. Limited data on Fulani segment-link measurements was
used to pick parameters for the improved device. An improved stirrer also was
developed that gave a 47 percent time advantage over the traditional method
and increased butter recovery by nearly 60 percent.


Journal for Farming Systems Research-Extension






DEVELOPING FARM TOOLS FOR WOMEN


CONCLUSIONS

The gender issue in small-scale farming as practiced in most developing
countries involves a complex sex-based division of farm tasks that has evolved
over a long period of time. Any innovation introduced into this system of
farming must take into account the importance of gender so that improve-
ments are planned simultaneously for both sexes.
Introduction of new farm tools is an innovation that affects the gender issue
greatly. Past experience has shown that such introductions were skewed
towards male workers, thereby upsetting the gender balance for farm tasks.
Most tools introduced require high energy levels for women and are not
operationally convenient based on body type.
Because women are the principal force for farm operations such as weeding,
planting, harvesting, threshing, transportation, food storage, and food pro-
cessing there is a vital need to ensure that tools are developed specifically to
their needs. Women's roles in animal husbandry tasks, especially milk
handling and processing, is no less important because the women (like the
pastoral Fulani) who process most of the milk suffer from the lack of
appropriate tools. Finally, farming operations are only some of the activities
of typical rural women, who also perform domestic tasks and are responsible
for child bearing. For effective upgrading and improvement of rural systems,
the role of appropriate tools for women cannot be overemphasized.


REFERENCES
Abalu, G. 1984. Women, the cropping chain, and farming systems research in sub-
Saharan Africa: Some experiences from Nigeria. Farming Systems Support Program
Seminars, University of Florida, Gainesville, October 16.
Abdulkarim, A.M. 1986. Comparative ergonomic studies of male and female operators
on selected tasks. Unpublished M.S. thesis, Department ofAgricultural Engineering,
Ahmadu Bello University, Zaria, Nigeria.
Abdulkarim, A.M., and R.N. Kaul. 1986. The effect of controlled work schedules on the
performance of a hoe farmer. Paper presented at the Annual Conference of the
Nigerian Society of Agricultural Engineers, Ife, September 20.
Adeyokunnu, O.T. 1981. Women and agriculture in Nigeria. Research Studies for
African Training, Rockefeller Foundation, N.Y.
Afonja, S. 1984. The historical evolution of the sexual division of labour in Nigeria.
Pages 129-135 in Proceedings of the Meeting on Theoretical Framework and Method-
ological Approaches to Studies on the Role of Women in Africa. International Labour
Organization, Geneva.


Vol. 3, No. 1, 1992






KAUL AND ALI


Bassi, S.Y., and R.N. Kaul. 1986. Selected anthropometric parameters of northern
Nigerian farmers. Paper presented at the Annual Conference of the Biotechnology
Society of Nigeria, University of Ilorin, May 16.
Dixon-Mueller, Ruth. 1985. Women's work in Third World agriculture: Concepts and
indicators. Women, Work and Development Paper 9. International Labor Organi-
zation, Geneva.
Eleseku, E.M. 1989. A study of milk processing tools used by Fulanis. Project report,
Department of Agricultural Engineering, Ahmadu Bello University, Zaria, Nigeria.
Harrison, P., and J. Howe. 1989. Measuring the transport demands of the rural poor:
Experiences from Africa. German Appropriate Technology Exchange (GATE) No. 1,
May.
Hoskins, M., and F.R. Weber. 1985. Why appropriate technology projects for women
fail. ECOFORUM 10(2):April.
Igben, M.S. 1980. The economics ofpeasant maize production in Plateau State. Occasion-
al Paper. National Institute for Socio-Economic Research (NISER), Ibadan, Nigeria.
Karpovich, P.V., and W.E. Sinning. 1971. Physiology of muscular activity. Philadelphia:
W.B. Saunders Company.
Kaul, RN. 1987. Agricultural equipment prototypes. Report IAR/AM/87/10. Institute
for Agricultural Research, Samaru, Ahmadu Bello University, Zaria, Nigeria.
Moller, S. 1987. Women make up over half the work force in the informal sector.
German Appropriate Technology Exchange (GATE) March, pp. 3-5.
Mueller, R.A.E., and M. von Oppen. 1985. Developing agricultural technologies for
diverse target groups in the semi-arid tropics. Pages 41-57 in Women and agricultural
technology: Relevance for research II. The Rockefeller Foundation, N.Y., and Interna-
tional Service for National Agricultural Research Council, The Hague, Netherlands.
Nwoko,S.G. 1986. Dairyimports in Nigeria: Development and policies. AlpanAfrican
Livestock Analysis Network.
Nwuba, E.I., and RN. Kaul. 1985. The effect of working posture on the Nigerian hoe
farmer. Journal of Agricultural Engineering Research 33:179-185.
Nwuba, E.I., and R.N. Kaul. 1987. Energy requirement of hand tools for wood cutting.
Journal of Agricultural Engineering Research 36.
Poats, S.V. 1988. Gender analysis in farming systems research and extension projects.
Keynote address at the Eighth Annual Farming Systems Research-Extension Sympo-
sium, University of Arkansas, Fayetteville, October 7-9.
Spiro,H.M. 1981. Thefifth world: Women's rural activities and time budgets in Nigeria.
Occasional Paper No. 19. Department of Geography, Queen Mary College, Univer-
sity of London, U.K.
Vermeulen, G.D., and RN. Kaul. 1980. Maize sellers test report DAE 79/7. Institute
for Agricultural Research, Ahmadu Bello University, Zaria, Nigeria.
Wagami, S.Y. 1983. Selected ergonomic studies of commonly operated farm tools.
Unpublished M.S. thesis, Department of Agricultural Engineering, Ahmadu Bello
University, Zaria, Nigeria.


Journal for Farming Systems Research-Extension








A Regional Perspective on Cassava, Famine,
and Seasonal Hunger in Humid and
Subhumid Africa

Steven Romanoff1



ABSTRACT
Data from agricultural surveys by national scientists in eight countries of
humid and subhumid Africa, conducted as part of the multidisciplinary
COSCA project of IITA, yield the following information for each
country: occurrence of famine, year of latest famine (food shortage
causing death or movement), major cause of famine, month of planting
of major crops, months of seasonal food shortage, relationship between
planting and shortage seasons, relationship between latitude and hunger
schedule, months ofconsuming cassava, and relationship between cassa-
va schedule and shortage season. Conclusions include (1) famine is not
limited to semiarid regions, where it occurs in different years than in
subhumid areas; (2) seasonal food shortages are general and timed in
relation to the agricultural cycle and natural factors; and (3) in tropical
areas with more than 700 mm of annual rainfall, cassava is a major buffer
to seasonal shortages. Besides cassava consumption, wage labor within
the village may be a behavioral response to seasonal shortage. No
evidence is found thatwage labor outside the village, importation offood,
or hunting is more widespread during seasonal shortages. The organiza-
tion of survey data from eight countries at the continental level by a
geographic information system complements local studies and macro-
economic methods of studying agriculture and social phenomena.
Agricultural and social scientists are asked to code their data for latitude
and longitude.


INTRODUCTION

This article is intended to enhance the ability of agricultural scientists and
planners to perceive regional patterns of food crises and local responses, to
verify the role of cassava in food security, and to better focus agricultural

1 Consultant, Agricultural Sciences, The Rockefeller Foundation. A draft of this report was
presented at the 3rd meeting of the COSCA steering committee in London, August 13-15, 1990.






ROMANOFF


research and nutritional interventions. The emphasis on cassava reflects that
crop's general importance in the humid and subhumid regions of tropical
Africa below the Sahara. In that area, cassava is the leading or second staple,
providing more than half the calories in the diet of millions of people (FAO,
1988). Farmers value it for its high yields and as a reliable insurance crop
(Jones, 1959; Longhurst and Lipton, 1989). The crop has many special
agronomic traits that contribute to its rapidly growing importance in the
region (Romanoff and Lynam, in press), especially in areas where market
access is restricted, food-getting strategies are fewer, kinship groups are
important, and rainfall is plentiful (Romanoff et al., in press).
This article draws on data from a continuing agricultural survey of cassava-
producing areas across the tropical belt of Africa. The survey began in 1988
and continues in 1992 with the third visit to sample villages. This study uses
data from the first visits to sample villages in eight countries. It combines
regional scope with village-level farmer interviews, and the method is designed
to add perspective with moderate detail to the study of agricultural systems.
By covering hunger in the humid and subhumid zones, this article
complements the more usual geographic focus of discussions of food security
in the Sahel, the Sudan zone, and the drier parts of East and Southern Africa,
where the problem is most severe (e.g. Nicholson, 1986; Harrison, 1988;
Shipton, 1990). Other authors discuss whole countries using aggregated
statistics that do not distinguish among zones (Jamal, 1988). The millions
of people living in the wetter areas of the continent also face food scarcities,
and their coping patterns are sometimes different from those of the drier areas.
As a result, existing social science research is less relevant than it should be to
the efforts of agricultural scientists and planners to improve the food security
of people in humid areas.
This article also will document seasonal food shortages in ways that
complement other reports. Seasonal deficits have received considerable
attention (Glantz, 1989; Messer, 1989; Sahn, 1989; see also the description
of the annual cycle of activities in most ethnographies). "While the evidence
is scattered and somewhat inconclusive," says one researcher, "it does support
the conclusion that [seasonal] food deficits are a priority concern for African
smallholders" (Moris, 1989, p. 234). The contribution of this article is to
move forward from scattered evidence to systematic consideration, showing
that farmers in many villages consistently identify seasons of hunger that
change in regular ways from one region to another.


Journal for Farming Systems Research-Extension






CASSAVA, FAMINE, AND SEASONAL HUNGER IN AFRICA 49

The role of cassava in mitigating seasonal shortages has been noted by
several authors (Jones, 1959; Guyer, 1989; Moris, 1989). This article will
attempt to prove that the function of cassava cultivation is nearly ubiquitous
in the humid and subhumid belt. Other behaviors that have been proposed
as adaptations to seasonal hunger will be shown to be less clearly patterned in
the current sample.
Thus, with the advantage of new survey data from an unusually wide area,
this paper documents the incidence of famine outside the Sahel, relates
seasonal hunger in the humid and subhumid zones to natural and agricultural
cycles, and shows the degree to which cassava consumption and other adaptive
behaviors are related to seasonal food shortages.


METHODS

This report is part of the Collaborative Study of Cassava in Africa (COSCA),
undertaken by national agricultural research institutions and the International
Institute for Tropical Agriculture (IITA) in eight countries. COSCA is
designed to discover and/or document broad differences in farming systems
at the national or regional levels, in order to help agricultural scientists to plan
research more effectively. COSCA is also designed to identify and weigh
factors related to agricultural development. Focused primarily on cassava, the
study covers farming systems based on other crops, food processing, market-
ing, consumption, and related social, cultural, or economic conditions
(Nweke, 1988).
The data for this article are drawn from COSCA research in eight countries:
Cote d'Ivoire, Ghana, Malawi, Nigeria, Tanzania, Uganda, Zambia, and
Zaire. In each country, a stratified random sample ofplaces was used to choose
between 30 and 70 villages; a few cases were also included from the Congo.
The sampling frame, part of a geographic information system (GIS) devised
by the Centro Internacional de Agricultura Tropical (CIAT),2 divided the map
of Africa into 10 km2 grid cells. Grid cells were selected without weights
within strata based on population density, road access, and climate type. The
village nearest the center of the cell enters the sample (Carter and Jones,
1989). Most countries have excluded areas where cassava is not a staple, parks,
areas that national agricultural scientists consider too dangerous to visit, and
some considered to be uninhabited.
2 Both IITA and CIAT are international agricultural research centers of the CGIAR system. The
COSCA study is led by Dr. Felix Nweke of IITA.


Vol. 3, No. 1, 1992






ROMANOFF


Survey Villages
The selected villages represent the broad mid-African belt where cassava is
a primary staple: 75 percent of the continent's total cassava production occurs
in the study countries (FAO, 1988). Most of the villages have less than six dry
months (a "dry" month experiences less than 100 mm mean rainfall during the
month); mean growing season temperature over 22 degrees Celsius; daily
temperature range less than 10 degrees; and isothermic seasonality (Carter
and Jones, 1989). In the current data, all villages have more than 700 mm of
rain, although COSCA hopes to obtain data from Angola or Mozambique,
where cassava production is important in drier areas. The sample does not
include many villages in the drier zone of West Africa, where sorghum, millet,
or cattle predominates. Culturally, the sample emphasizes rural settlements
more than would be the case if it had been drawn with probability of selection
proportionate to population.
COSCA survey teams, staffed by personnel from national research institu-
tions in each country, make three visits to each village in the sample: first, for
general group interviews; later, for observations of farmers' fields; finally, for
a household survey. Each interview team includes at least two professionals,
typically agricultural economists or breeders. In some countries, entomolo-
gists have participated to observe conditions in farmers' fields. Data from the
first phase of the survey-group interviews, which have been completed in all
eight countries-are the basis for this article.
In each village, the group interview actually consists of three concurrent
sessions: one requires a small group of patient farmers to answer detailed
questions about village agricultural practices; another asks more general
agricultural questions; and the third asks about cassava processing and use, as
well as other questions about the village. The sessions typically last between
two and three hours. Earlier versions of the questionnaire had only two
sections and exhausted the patience of most villagers.3
Questions on food scarcity and hunger refer to general patterns in the
village, and interviewers tried to attain consensus responses. The questions
included the following:

3 About 60 percent of the interview subgroups had between 10 and 24 adults; about 25 percent
had 25 or more adults; the rest had fewer than 10. In the median case, the adults of each subgroup
represented 4 percent of the estimated adult population of the village. Most of the subgroups
included both men and women, and in some cases the groups split along gender lines on
particular questions before reaching consensus. Although the written questionnaires are in
English and French, most interviews were conducted in local languages or other non-European
languages, such as Swahili, commonly used for trade and communication between different
ethnic groups.
Journal for Farming Systems Research-Extension






CASSAVA, FAMINE, AND SEASONAL HUNGER IN AFRICA 51

1. In which months of the year do people in this village suffer most from food
shortage?
2. Has this village ever suffered from famine in the past? [Famine was defined
as a shortage of food causing death or causing people to leave the village.]
3. In which years did famine occur in this village and what were the causes?
[Both "major cause" and "other cause" were elicited].4
Other questions concerned the months when the villagers ate each staple,
when they undertook specific agricultural tasks, when they were busiest, and
when they migrated for work.

Climate
Information on climatic patterns is drawn from a digitized map of 4,000
meteorological stations. The computer map, another part of the geographic
information system (GIS) created by CIAT's Agroecological Studies Unit
with a presentation program written by IITA, affords researchers immediate
access to monthly averages ofprecipitation and other climatic traits for any site
in Africa.


ANALYSES

The incidence of famine and seasonal hunger are demonstrated by simple
tables or calendars of occurrence, using unweighted data. Because sampling
proportions varied from one country to another, the unweighted data are
generally presented by country, except in a few general tables where little bias
results from pooling the data.

The Database
The village-level COSCA interviews are the basis for an evolving database.
The study began in Uganda, Tanzania, Zaire, Nigeria, Ghana, and Cote
d'Ivoire. It was expanded when the East and Southern Africa Root Crops
Research Network (ESARRN) adopted the COSCA methodology for diag-

4 Both the use of group interviews and presentation of the data by country affect some results.
Because they concern each village as a whole and are addressed to groups, the questions are
unlikely to elicit information on such causes of hunger as illness or loss of employment, or on
issues such as food allocation in individual households. There may be significant individual
differences, for example, in agricultural and economic strategies among the households of a
village (e.g., Alverson, 1990). Data on household variation and on events related to household
production and shortages are being gathered in the second and third phases of COSCA. Scale
of analysis also affects the results presented here: Seasonal patterns of hunger or behavior evident
in countries considered severally may not coincide with patterns found in the region as a whole.


Vol. 3, No. 1, 1992






ROMANOFF


nostic work. Malawi and Zambia have completed village-level surveys, and
other ESARRN countries are in the field. When IITA conducted a diagnostic
study in Sierra Leone and Liberia, it used many of the COSCA questions as
well. Congo decided to use the COSCA questionnaire for diagnostic studies
related to a development project, and some countries, such as Zaire and
Tanzania, have undertaken to expand the original sample.
As new countries join the COSCA study, they determine sample size and
whether or not to include areas without cassava production, while retaining
the use of stratification by climate, population, and access. The Malawi
sample, for example, covers the whole country, and Sierra Leone and Liberia
have large samples relative to the size of each country.
Hence, this article presents descriptive results disaggregated by country
and, for the moment, postpones statistical analysis of relationships among
variables. Such studies (e.g., of the factors associated with susceptibility to
famine or seasonal hunger) would be of great importance, but they await
COSCA decisions on weighting cases to avoid bias and on other methodolog-
ical issues. Future tables using weighted data may appear different from those
presented in this article, which are based on unweighted data.

Graphs and Maps
Frequency graphs (by month and country) are used to illustrate calendar-
based data. In a typical case, the graphs illustrate the variation in the number
ofvillages reporting hunger in a given month. The graphs clearly demonstrate
that the months when hunger is reported with the greatest frequency vary
from one country to another.
Graphs also serve to reveal and verify functional relationships among
behaviors, conditions, or natural events. Graphs demonstrate, for example,
the relationships among date of planting, food scarcity, and cassava consump-
tion. In this case, the graphs show that, in a given country or region, these
variables are most common or concentrated in the same months.
Maps are used sparingly in this article, pending the availability of better
maps as the COSCA data base continues to expand over the next several years.
The seasonal map of hunger, included to illustrate the striking annual
progression north and south of reported food shortages, should therefore be
interpreted as showing only general patterns.


Journalfor Farming Systems Research-Extension






CASSAVA, FAMINE, AND SEASONAL HUNGER IN AFRICA 53

BACKGROUND: SUBSISTENCE BASE OF THE SAMPLE
VILLAGES

Residents of the sample villages ranked their crops by land area (Table 1).
Because most villages in the sample were drawn from regions where cassava
is a staple, it is not surprising to find cassava ranked first in more villages than
any other crop. A large number of villages, however, report cassava as a
secondary crop. Maize is reported as the dominant crop in Malawi, and in a
substantial number of villages in many countries.
In terms of its importance as a staple food, rather than in terms of land area,
cassava is even more frequently cited in first place in the COSCA sample. The
difference may be due to cassava's very high yield compared to maize, even
when measured on a dry weight basis. The sole exception is in western
Nigerian villages (Table 2).
Irrigation is virtually unreported in the sample villages. Although owner-
ship of goats or sheep is common, cattle herding is practiced in less than one-
third of the villages of West Africa and Zaire. In East Africa, in contrast, cattle
are herded in more than halfof the villages ofTanzania, Uganda, Zambia, and
Malawi, where goats and sheep are correspondingly less important. The
virtual absence of irrigation and the minor importance of cattle, together with
the near absence of millet and sorghum as primary crops, confirm that the
subsistence basis of the sample villages is sharply distinct from those of the
Sahel or Sudan zones, in conformity with regional patterns observed by others
(Davies, 1973).
The survey shows that residents of almost all the villages import some staple
food, while farmers in almost all of the villages sell food. The most common
food import is rice; cassava products are frequently traded, more often locally
than in urban markets. Of the few villages that do not import food, most are
located in zones of high rainfall and low population density. The degree to
which the villages participate in the cash economy varies considerably.


RESULTS

The survey data show the incidence of famine in the villages, the patterns of
seasonal hunger, and the relationship between cassava consumption and
hunger.


Vol. 3, No. 1, 1992






ROMANOFF


Table 1. Leading Crop According to Land Area Planted, by Country.
Villages Leading Crop Total

Cassava Maize Yams Sorghum Rice Other
Cote # 2 6 4 28a 40
d'Ivoire % 5.0 15.0 10.0 70.0
Ghana # 8 3 8 1 10 30
% 26.7 10.0 26.7 3.3 33.3
Nigeria # 12 3 2 4 21
(west) % 57.1 14.3 9.5 19.0
Nigeria # 7 5 16 10 3 3 44
(east) % 15.9 11.4 36.4 22.7 6.8 6.8
Tanzania # 10 10 1 2 7 4 34
% 29.4 29.4 2.9 5.9 20.6 11.8
Uganda # 12 4 16 32
% 37.5 12.5 50.0
Zaire # 35 8 8 51
(west) % 68.6 15.7 15.7
Zaire # 5 1 4 10
(east) % 50.0 10.0 40.0
Zambia # 18 14 2 4 38
% 47.4 36.8 5.3 10.5
Malawi # 8 58 1 67
% 11.9 86.6 1.5
Congo # 6 6
% 100.0
a Principally cocoa
Source: COSCA database as of October 20, 1991; excludes Sierra Leone and Liberia;
includes all of Malawi with non-cassava areas; unweighted data.

Famine
Although they are generally far from the Sahel and enjoy greater rainfall,
most of the villages of the sample report experiences of famine: hunger
resulting in mortality and/or migration from the village (Table 3). Of the four
WestAfrican countries or areas, only western Nigeria reports that most villages
experienced no famine. In Uganda, plagued by civil war, and in Malawi, with
very low per capital income, almost all villages report famine. The famines
reported are not long-ago or mythological events: two-thirds of the villages
reporting famine (half of all villages) referred specifically to events of the last
15 years (Table 4).
Drought, distantly followed by insect damage, war, and plant diseases, was
most frequently cited by villagers as the proximate cause of famine (Table 5).
As contributory causes, insect pests were most frequently cited, followed by
fire of natural or human origin (the questionnaire did not make the distinc-
tion), drought, and war. Overall, natural causes (including fire on the


Journal for Farming Systems Research-Extension






CASSAVA, FAMINE, AND SEASONAL HUNGER IN AFRICA 55

Table 2. Relative Importance of Cassava in the Diet, by Country.
Villages Rank'

1 2 3 4 5 6 7to 11
Cote # 4 20 7 5 2 1
d'Ivoire % 10.3 51.3 17.9 12.8 5.1 2.6
Ghana # 16 8 5 1
% 53.3 26.7 16.7 3.3
Nigeria # 8 10 1 1 1
(west) % 38.1 47.6 4.8 4.8 4.8
Nigeria # 15 8 2 5 5 5 4
(east) % 34.1 18.2 4.5 11.4 11.4 11.4 9.1
Tanzania # 24 4 2 1 2 1
% 70.6 11.8 5.9 2.9 5.9 2.9
Uganda # 16 6 6 1 2
% 51.6 19.4 19.4 3.2 6.5
Zaire # 47 2 1 1
(west) % 92.2 3.9 2.0 2.0
Zaire # 8 1
(east) % 88.9 11.1
Zambia # 18 5 4 2 1 1 4
% 51.4 14.3 11.4 5.7 2.9 2.9 11.4
Malawi # 9 15 6 10 10 8 2
% 15.0 25.0 10.0 16.7 16.7 13.3 3.4
Congo # 5
% 83.3
a 1 = most important; 11 = least important.
Source: COSCA database as of October 20, 1991; excludes Sierra Leone and Liberia;
includes all of Malawi with non-cassava areas; unweighted data.

assumption that it was of natural origin or abetted by dry weather) were
blamed in two-thirds of the reports, agricultural causes were cited in a quarter,
and human causes (including the "other" category) in the remainder.
At least one famine was reported for almost each of the 60 years between
1930 and 1990 (Table 6). At least some famines, evidently, remain vivid in
the memories of villagers (Shipton, 1990). Reports of only a single famine in
many villages may reflect respondents' fatigue and/or the failure ofinterview-
ers to probe sufficiently. The actual incidence of famine therefore may be
greater than reported.
Famine was particularly widespread in some years: 1945, even 45 years after
the event, figures in 13 ofa total of 376 famine reports. In Malawi, 1949 was
a hungry year in 54 villages, but not in other countries. More recently, famines
in 1977 were remembered in 20 villages, most of them in Cote d'Ivoire; in
1980 in 27 villages, mostly in Uganda and Malawi; and in 1984 in 26 villages
in many countries.


Vol. 3, No. 1, 1992







ROMANOFF


Table 3. Villages Reporting Famine, by Country.
Village Famine

Not reported Reported Row total

Cote d'Ivoire # 0 40 40
% 0 100.0
Ghana # 2 28 30
% 6.7 93.3
Nigeria (west) # 13 7 20
% 65.0 35.0
Nigeria (east) # 6 38 44
% 13.6 86.4
Tanzania # 15 17 32
% 46.9 53.1
Uganda # 1 31 32
% 3.1 96.9
Zaire (west) # 18 30 48
% 37.5 62.5
Zaire (east) # 8 2 10
% 80.0 20.0
Zambia # 18 16 34
% 52.9 47.1
Malawi # 6 61 67
% 9.0 91.0
Congo # 4 2 6
% 66.7 33.3
Source: COSCA database as of October 20, 1991; excludes Sierra Leone and Liberia;
includes all of Malawi with non-cassava areas; unweighted data.

The worst year was 1983, cited in reports from 52 villages. Famine was
most widely reported in that year in West Africa, but other regions were also
affected. Fire was mentioned prominently as a contributing cause in 1983. In
contrast, the years between 1968 and 1973, which were notable famine years
in the Sahel, were reported as famine years by modest numbers of villages in
the African cassava belt.

Seasonal Food Shortages
The hungry months, when exceptionally large numbers of villages report
substantial or extraordinary food shortages, vary from country to country.
The COSCA village data show the clear relationship between reported hunger
and the agricultural cycle. In each country, the months when most villages are
planting their main crop fall within a month either way of the onset of the
hunger season. Graphically, this relationship appears quite strong (Figure 1),
although variation in the data warrants further analysis at the village level.


Journal for Farming Systems Research-Extension






CASSAVA, FAMINE, AND SEASONAL HUNGER IN AFRICA 57

Table 4. Reported Year of Latest Famine, by Country.,

Villages Date of Latest Famine Total
Pre-1955 1955-64 1965-74 1975-84 1985-on
Cote d'Ivoire # 2 1 29 5 37
% 5.4 2.7 78.4 13.5
Ghana # 27 1 28
% 96.4 3.6
Nigeria # 3 1 3 7
(west) % 42.9 14.3 42.9
Nigeria # 6 2 1 12 12 33
(east) % 18.2 6.1 3.0 36.4 36.4
Tanzania # 3 1 3 5 4 16
% 18.8 6.3 18.8 31.3 25.0
Uganda # 2 2 23 4 31
% 6.5 6.5 74.2 12.9
Zaire # 1 4 12 9 26
(west) % 3.8 15.4 46.2 34.6
Zaire # 1 1 2
(east) % 50.0 50.0
Zambia # 6 3 2 1 4 16
% 37.5 18.8 12.5 6.3 25.0
Malawi # 39 1 14 7 61
% 63.9 1.6 23.0 11.5
Congo # 1 1
% 100.0
a Only villages reporting a famine and giving a year are included. Years refer to most recent famine
only, not to all famines reported.
Source: COSCA database as of October 20, 1991; excludes villages not reporting
famine or not providing date; excludes Sierra Leone and Liberia; includes all of
Malawi with non-cassava areas; unweighted data.
Table 5. Major Cause of Last Famine.
Cause Reported as Major Cause Reported as Contributory Total
(# villages) (# villages)
Drought 252 13' 265
Flood 15 15
Hail 2 1 3
Fire 7 19 26
Insect pests 49 23 72
Plant disease 17 5 22
Animal damage 7 8 15
Infertile soil 2 1 2
Land shortage 1 2 3
War 19 12 31
Government policy 5 4 9
Human disease 1 1 2
Marketing 1 1
Source: COSCA database as of October 20, 1991; excludes Sierra Leone and Liberia;
includes all of Malawi with non-cassava areas; unweighted data. The causes refer to
primary and secondary causes of all famines reported in the village.


Vol. 3, No. 1, 1992







ROMANOFF


Table 6. Incidence of Famine, as Reported in Group Interviews.


Year

1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962


Villages reporting famine Year Villages reporting famine


+
+++++++++++++
+++
+++++
++++
++++++++++++++++++//+++
+


1963 2
1964 7
1965 0
1966 2
1967 0
1968 3
1969 3
1970 4
1971 3
1972 2
1973 4
1974 9
1975 5
1976 5
1977 20
1978 6
1979 9
1980 27
1981 10
1982 2
1983 52
1984 26
1985 10
1986 9
1987 13
1988 12
1989 6
1990 3


++
+++++++

++

+++
+++
++++
+++
++
++++
+++++++++
+++++
+++++
++++++++++++++++++++
++++++
+++++++++
++++++++++++++++++++++++
++++++++++
++
+++++++++++++++++++//+++
++++++++++++++++++++++++
++++++++++
+++++++++
+++++++++++++
++++++++++++
++++++
+++


Source: COSCA database as of October 20, 1991; excludes Sierra Leone and Liberia;
includes all of Malawi with non-cassava areas; unweighted data. Villages may report
up to three events.

The planting season is related to the onset of the rainy season, which the
IITA Agroclimatology Unit defines as the first month with more than 60 mm
of rain. The CIAT meteorological data, which cover most of Africa, confirm
the relationship of the rainy season to the hunger season, as graphed in Figure
1. Most villages begin to report hunger between 0 and 3 months after the first
rainy month. At the country level, the average lag between rain and hunger
in the COSCA data is 1.5 months.
While this general relationship between onset of rains and increased
incidence of hunger is clear at the country level, important details remain to
be added. The sample villages in the study are located largely in the humid
and subhumid zones. The pattern found here-hunger at the beginning of
the rainy season-may be somewhat distinct from that discerned by Moris


Journal for Farming Systems Research-Extension






CASSAVA, FAMINE, AND SEASONAL HUNGER IN AFRICA 59

(1989), who saw the "energy-deficit period overlapping the end of the long
dry season and the beginning of a new cycle." Alternatively, it may be useful
to redefine the onset of the rainy season as a month with 100 mm ofrain, since
in practice villagers tend to plant somewhat later than the first month with 60
mm of rain.
In any case, the onset of rains proves to vary somewhat with latitude, and
to a lesser degree with longitude, in that at a.given latitude the rains often
begin a bit later toward the east. An unsystematic inspection of meteorolog-
ical data reveals some general patterns with variations due to local factors, such
as the proximity to seacoasts and their currents. Purely for illustration of the
magnitude ofpatterns, one might say that at 70 N ofthe Equator, the rains tend
to begin in June to the west or July to the east. At 50 N, depending on
longitude, the rains begin in April or May. At 30 N, the starting dates are
advanced to March or April.
At the Equator, the rains begin in September toward the west, while a
bimodal pattern prevails to the east. At 3 S ofthe Equator, rains begin in the
fourth quarter, between September and November. At 5 S, they advance
nearer to year-end. At 70 S, the rain dates fall between October and December.
These generalizations, inescapably, oversimplify reality, but they indicate the
broad seasonal patterns that differ markedly with latitude.
The patterns suggest, for any date, where villages with adaptations like the
ones found by COSCA are likely to be in their annual planting-hunger cycle.
Patterns would be different for villages with different productive patterns, like
those depending more on cattle or on cereals, and there may well be variation
among the COSCA villages that depend on degree of market involvement on
the cycle of special crops. Still, if the most marked differences in onset of rains,
planting, and hunger are determined by latitude, broad similarities should
exist in the hunger calendar within groups of northern countries: Ghana, Cote
d'Ivoire, Nigeria, and Uganda. A group of southern countries-Tanzania,
Zambia, Malawi, Zaire-should also display a pattern, but distinct from that
of the north. In fact, as Figure 2 demonstrates, this is precisely the case.
Maps of the COSCA data clearly track the annual progression of hunger
reports, starting in the south in February, moving north, and reaching the
westernmost sites in June and July. By September, the reports center farther
south, becoming increasingly concentrated in the south through February,
when the cycle begins anew (Figure 3). This map series resembles those of
rainfall and green vegetation.


Vol. 3, No. 1, 1992






ROMANOFF


Jan. June Dec.

Ghana


June

Zambia


Jan. June

West Nigeria


Jan. June Dec.

Tanzania


Hunger

Plantg


Jan. June Dec.

East Zaire


Figure 1. Planting Season and Hunger in Eight African Countries.
Source: COSCA.


Journal for Farming Systems Research-Extension






CASSAVA, FAMINE, AND SEASONAL HUNGER IN AFRICA


Jan. June Dec.
West Zaire


Jan. June
East Nigeria


Jan. June Dec.
Cote d'Ivoire


18 -
16
14
12-
10
8
6-

2
0
Jan. June Dec


Hunger

Pbnting


Uganda



Figure 1. Planting Season and Hunger in Eight African Countries (cont.).
Source: COSCA.


Vol. 3, No. 1, 1992


50
40
r 30
20
10


June
Malawi


Dec.


id






ROMANOFF


60- 30

40- 20-
S30 15

0 5-

Jan. June Dec. Jan. June Dec.
South Range North Range

-- Tanzania -- Zambia -a- Ghana --- Uganda
Malawi -- Zaire -0- W. Nigeria Cote d'lvoire
--- E. Nigeria

Figure 2. Hunger Months for Northern and Southern Ranges.
Source: COSCA.

Adaptations to Hunger: The Role of Cassava and Other Crops
If the hunger season coincides with planting, the season of plenty begins
with the harvests of such staples as plantain, banana, cocoyam, rice, potatoes,
and sometimes beans. Between the planting season and the harvest of the local
mix of multiple staples in each region, the only local production depends on
a limited set of crops: first, the leafy vegetables, sometimes including cassava
leaves; then cassava itself; then, in some cases, maize.
Except for cassava leaves, the leafy vegetables are secondary crops; presum-
ably they do not yield enough food to eliminate hunger entirely. Their dietary
value is enhanced, however, by their availability just when there is little else to
eat. Cassava leaves, which begin to regrow on established plants at the onset
of the rains, are important in East Africa, along with leafy vegetables, during
the hunger period. In Zaire, cassava leaves are eaten regularly throughout the
year.
Where cassava itself is not a year-round staple, cassava consumption is most
widespread during the hunger season. Cote d'Ivoire is typical of the countries
where most cassava is eaten as a hunger-breaker; East Nigeria is similar in this
regard (Figure 4). In such countries, a minority of villages report cassava as
the major crop. At the opposite extreme are countries or areas like western
Zaire, western Nigeria, and Zambia, where cassava is consumed year-round
(Figure 4) and where the majority ofvillages report cassava as their major crop
by land area.


Journal for Farming Systems Research-Extension






CASSAVA, FAMINE, AND SEASONAL HUNGER IN AFRICA 63

Some countries exhibit some correlation between cassava use and hunger,
but also include many villages that eat a lot of cassava in all seasons. These
countries consume cassava in a pattern that is intermediate between that of
countries where cassava is only a hunger breaker and that of countries where
consumption is constant. Most of the East African countries-Uganda,
Malawi, Tanzania, and East Zaire-are intermediate. The data for these
countries are adjusted in Figure 5 to show the number of villages reporting
high cassava consumption, minus a constant equal to the minimal number of
villages reporting consumption in any month. In these countries, a substantial
minority of villages, ranging between 29 and 50 percent (except in Malawi),
report cassava to be the number-one crop. Cassava consumption in Ghana
does not clearly conform to any of the patterns.
Despite their differences, all of the patterns described share a reliance on
cassava during the hunger period. Other starchy staples, like banana, plantain,
and yam, do not play this crucial role, and some, like sweet potatoes, are not
widespread. In some areas, maize and/or one or another of the fast maturing
legumes may also become available at about the same time as cassava. In West
Africa, for example, leafy vegetables precede cassava during the hungry season
and ground-nuts follow cassava by one month; in East Africa, beans peak a
month before cassava, and ground-nuts and millet arrive in the same month.
Cassava is especially suited to provide protection against hunger because it
can be harvested over a very long period, from about 7 to 36 months after
planting. Because the period from the peak of the planting season to the peak
of the hunger season varies from 12 to 18 months, cassava's indifference to
time is important. Breeders often evaluate cassava for yield at 12 months, and
precocity is valued; the patterns found here suggest that the ability to be
available in the ground remains desirable.
Eating cassava is not the only alternative to going hungry mentioned in the
literature. Villagers can import food into the village, work for wages within
or outside the village, or harvest wild resources. In Cote d'Ivoire and Malawi,
there is a substantial calendrical correlation between hunger and food imports
into the villages; that is, most villages face food scarcity and import increased
amounts of food in the same months. In none of the other COSCA countries,
however, do COSCA data reveal a similar correlation. Other factors, such as
the custom of importing rice into West African villages for holidays in
December, account for spikes in food-import activities in some villages. The
evidence does not suggest that imported food rivals cassava as a seasonal
response to hunger.


Vol. 3, No. 1, 1992






ROMANOFF


January


*t
S /










S /


* *


February






March





April





May






June


Figure 3. Annual Northward and Southward Progression of Hunger.
Source: COSCA.


Journal for Farming Systems Research-Extension


dwV
(S






CASSAVA, FAMINE, AND SEASONAL HUNGER IN AFRICA


July


N



August


It



S September






,. .** October






^ November







SDecember





Figure 3. Annual Northward and Southward Progression of Hunger (cont.).
Source: COSCA.


Vol. 3, No. 1, 1992



















Jan. June Dec.
Cote d'Ivoire













Jan. June Dec.
West Zaire


Jan. June Dec.
Jan. June Dec.


30


20


10



Jan. June Dec.

East Nigeria


Jan. June Dec.

West Nigeria









->4

Cusovn


Zambia
Figure 4. Cassava and Hunger Months in Four African Countries.
Source: COSCA.


Demand for labor at the village level is strongly seasonal, peaking during
or just before the months of food scarcity. COSCA data show that the busiest
time is during land preparation and planting. Because we know from other
COSCA data that some of the labor is hired, particularly for land preparation,
there may be a peak of transactions, in cash or kind, during the months when
hunger is most widespread.


Journal for Farming Systems Research-Extension


ROMANOFF






CASSAVA, FAMINE, AND SEASONAL HUNGER IN AFRICA


Jan. June Dec.

Uganda


Jan. June Dec.
Tanzania


40
30

20
10-


Jan. June Dec.
Malawi


East Zaire


Hunger



Figure 5. Adjusted Cassava and Hunger in Four African Countries.
Source: COSCA.


Working outside the village is another possible strategy during the months
of food shortage. The data show only a weak correlation between the two
phenomena in Cote d'Ivoire and possibly Ghana, whereas in the villages of
eastern Nigeria and Uganda, outside work appears actually to increase during
the months of plenty. Other countries display no clear pattern. This lack of
clear correlation suggests that the relationships between working outside the
village and food scarcity, ifany, may be locally specific. Similarly, COSCA data
show no correlation between food scarcity and consumption of bush meat.


CONCLUSIONS

The majority of villages of the sample, drawn from the cassava belt of Africa,
report famine in the last 15 years. Village residents attribute the famines to


Vol. 3, No. 1, 1992





ROMANOFF


drought, agricultural pests, or war. Because the sample is drawn mostly from
areas with less than six months of dry season and rainfall above 700 mm
annually, these results demonstrate that famine is not limited to the Sahel,
Sudan, and dry parts of Southern Africa. Coping strategies such as slaughter-
ing or selling cattle or relying on irrigation, both well known in the Sahel and
Sudan, are unimportant in the sample villages simply because cattle are rare
and irrigation virtually absent.
This has implications for researchers and planners. Among them are that
breeders should maintain the traits of cassava that give it its special role in
famine prevention. Indicators of famine should be regionally specific: sales of
cattle would be an inappropriate indicator for the area studied here, and the
price of cereals may or may not be appropriate. Menus of responses to drought
or famine should also be as regionally specific as possible. Relief efforts may
be needed in wetter areas, adapted to prevailing human ecological patterns.
Studies in support of famine prevention should reflect the climatological and
social diversity of the region, including both drier and wetter zones.
The data confirm that local patterns of seasonal hunger, linked to cropping
patterns and thus to the onset of the rainy season, occur in a yearly cycle that
moves north and south with the seasons. The pattern has been demonstrated
to a degree that warrants further analysis to predict more precisely the usual
onset of the hunger season in relation to natural, agricultural, and social
factors. If that is done successfully, the patterns may be projected using
meteorological and remote sensing data bases.
Eaten year-round or mostly during the hunger season, cassava plays a
special role in the yearly calendar ofagriculture and nutrition in the vast region
studied by COSCA. There may be additional functions of cassava production
in evening out seasonal labor demand, but that is beyond the scope of this
paper (see Stone et al., 1990). Maize also plays an important role at the end
of the hunger season as well as generally, and the two vie for first place among
African food crops. Other behaviors that may be related to seasonal hunger
were not so clearly shown to be related to seasonal shortages. Importing food
and seeking work outside the village may be alternative responses to food
scarcity, but the analysis of COSCA data did not demonstrate that function.
Many authors have indicated dissatisfaction with using aggregate statistics
to study problems of hunger and farming systems (e.g., Berry, 1984; Hill,
1985). Such studies do not perceive regional variations and may conjecture
local dynamics from national data. Hill (1985) has suggested a taxonomic
solution, and identification of domains of recommendations is standard


Journal for Farming Systems Research-Extension






CASSAVA, FAMINE, AND SEASONAL HUNGER IN AFRICA 69

practice in farming systems research. The volume of Huss-Ashmore and Katz
(1989) suggests complementing "macroperspectives" with "microperspec-
tives." This paper has used a wide regional perspective that incorporates and
summarizes village-level data, an approach that is becoming more common
(Downing, 1991).
We may be confident that eventually methods will be adopted to incorpo-
rate agricultural and social science survey data into geographic contexts. Such
a process would be assisted by scientific surveys and trials that are coded by
latitude and longitude.
In viewing human strategies for coping with famines or seasonal shortages,
scientists should presume that each type of response is distributed differently
in space and time. Beginning with rich ethnographic or farming systems
descriptions and catalogues of responses to food scarcities, agricultural and
social science researchers can develop continental data bases to map the
occurrence of different strategies, in preparation for identifying their environ-
mental and cultural correlates. The close-up focus of ethnography or farming
systems research and the regional data base are complementary perspectives.


REFERENCES
Alverson, H. 1990. The class context of agriculture in Botswana: Some policy
implications. Culture and Agriculture 40:11-16.
Berry, S. 1984. The food crisis and agrarian change in Africa: A review essay. African
Studies Review 27(2):59-112.
Carter, S.E., and P.G. Jones. 1989. COSCA site selection procedure. Collaborative Study
of Cassava in Africa (COSCA) Working Paper No. 2. International Institute of
Tropical Agriculture, Ibadan, Nigeria.
Davies,H.R.. 1973. TropicalAfrica:Anatlasforruraldevelopment. Cardiff: University
of Wales Press.
Downing, T.E. 1991. Assessing socioeconomic vulnerability to famine: Frameworks,
concepts and applications. Brown University, The Alan Shawn Feinstein World
Hunger Program, Providence, R.I.
Food and Agriculture Organization (FAO). 1988. FAO production yearbook. Food and
Agriculture Organization of the United Nations, Rome.
Glantz, M. 1989. Drought, famine, and the seasons in sub-Saharan Africa. In Rebecca
Huss-Ashmore and S.H. Katz, eds., African food systems in crisis. N.Y.: Gordon and
Breach Science Publishers.
Guyer, J. 1989. From seasonal income to daily diet in a partially commercialized rural
economy (southern Cameroon). Pages 137-150 in D.E. Sahn, ed., Seasonal variation
in third world agriculture: The consequences for food security. Baltimore, Md.: Johns
Hopkins University Press.
Harrison, G.A., ed. 1988. Famine. Oxford: Oxford University Press.


Vol. 3, No. 1, 1992






ROMANOFF


Hill, P. 1985. The practical need for a socio-economic classification of tropical agrarian
systems. Pages 117-130 in Socialanthropologyanddevelopmentpolicy. N.Y.: Tavistock
Publications.
Huss-Ashmore, Rebecca, and S.H. Katz, eds. 1989. African food systems in crisis. Part
One: Microperspectives. N.Y.: Gordon and Breach Science Publishers.
Jamal, V. 1988. Special issue: The African food crisis, food security and structural
adjustment. International Labour Review 127:6.
Jones, W.O. 1959. Manioc in Africa. Palo Alto: Stanford University Press.
Longhurst, R., and M. Lipton. 1989. The role of agricultural research and secondary
food crops in reducing seasonal food insecurity. Pages 285-298 in D.E. Sahn, ed.,
Seasonal variation in third world agriculture: The consequences for food security.
Baltimore, Md.: Johns Hopkins University Press.
Messer, E. 1989. Seasonality in food systems: An anthropological perspective on
household food security. Pages 151-175 in D.E. Sahn, ed., Seasonal variation in third
world agriculture: The consequencesforfood security. Baltimore, Md.: Johns Hopkins
University Press.
Moris, J.R. 1989. Indigenous versus introduced solutions to food stress inAfrica. Pages
209-234 in D.E. Sahn, ed., Seasonal variation in third world agriculture: The
consequences for food security. Baltimore, Md.: Johns Hopkins University Press.
Nicholson, S.E. 1986. Climate, drought, and famine in Africa. Pages 107-128 in A.
Hansen and D.E. McMillan, eds., Food in Sub-Saharan Africa. Boulder, Colo.: Lynne
Rienner Publishers, Inc.
Nweke, F. 1988. COSCA project description. Collaborative Study of Cassava in Africa
(COSCA) Working Paper No. 1. International Institute of Tropical Agriculture,
Ibadan, Nigeria.
Romanoff, S., and J. Lynam. n.d. Cassava and African food security: Some ethnographic
examples. Ecology of Food and Nutrition, in press.
Romanoff, S.,S. Carter, and J. Lynam. n.d. Cassava production andprocessing in a cross-
cultural sample of African societies. Behavior Science Research, in press.
Sahn, D.E., ed. 1989. Seasonal variation in third world agriculture: The consequencesfor
food security. Baltimore, Md.: Johns Hopkins University Press.
Shipton, P. 1990. African famines and food security: Anthropological perspectives.
Annual Review of Anthropology 19:353-394.
Stone, G.D., R.McC. Netting, and M.P. Stone. 1990. Seasonality, labor scheduling, and
agricultural intensification in the Nigerian savanna. American Anthropologist92( 1):7-
23.


Journal for Farming Systems Research-Extension









Extension Agent Surveys for Defining
Recommendation Domains:
A Case Study From Kenya

Steven Franzell



ABSTRACT
There is little consensus on when and how to define recommendation
domains in FSR. Manyrecommend using secondarysources and informal
farmer surveys; this paper evaluates a brief survey of extension agents in
central Kenya for tentatively defining recommendation domains over a
larger area than would normally be covered in an informal survey. The
extension agent survey was evaluated by comparing the data obtained
with those ofa formal survey of farmers in the same area during the same
year.
Compared to using an informal survey to define recommendation
domains, an extension agent survey has three important advantages. First,
it can provide information on recommendation domains for areas much
larger than can be covered effectively in informal surveys and at much
lower cost. Second, by delimiting boundaries between the defined
domains, little time is wasted during the informal survey interviewing
farmers not in the targeted domains. Third, byinduding extension agents
in the information-gathering process, researchers can gain their confi-
dence and active participation in on-farm research. However, as useful as
an extension agent survey is for tentatively defining recommendation
domains, it cannot serve as a substitute for an informal survey of farmers
for understanding farming systems.











1 Agricultural Economist, International Centre for Research in Agroforestry, Nairobi,
Kenya.






FRANZEL


INTRODUCTION

The concept of "recommendation domain" has emerged as an important tool
in farming systems research. A recommendation domain is defined as "a group
of roughly homogeneous farmers with similar circumstances for whom we can
make more or less the same recommendation" (Byerlee and Collinson, 1980).
Defining recommendation domains in an area assists scientists to focus their
efforts on the particular problems of different farmer groups, which may vary
according to agroecological or socioeconomic variables. This approach con-
trasts with a more common one in which researchers seek to increase crop
productivity without an awareness of the different categories of farmers they
serve and the varying problems they face.
Defining recommendation domains in a particular area serves four purpos-
es, as outlined below:
1. Helps researchers to decide where, and for which farmers, to focus their
efforts (CIMMYT, 1979; Franzel, 1981; Kean and Singogo, 1988).
2. Assists researchers to assure that their activities are targeted towards the
needs and circumstances of specific categories of farmers.
3. Makes the research process more efficient. By defining boundaries
between domains, researchers can ensure that only farmers within the domains
under study are included in the surveys and experiments, saving valuable time
and resources.
4. Assists policy makers in development planning. For example, greater
emphasis at the national level on foreign exchange earnings will require
policies and resources to be shifted to those domains that have the greatest
potential for producing export crops (Collinson, 1982).
There are varying opinions in the literature as to when and how researchers
should define recommendation domains. Many recommend making hypoth-
eses concerning domain boundaries at the beginning of the research process,
based on reviews ofsecondary data or during the informal survey2 (Harrington
and Tripp, 1984; Hildebrand and Poey, 1985). These hypotheses are then
tested and refined during the surveys and experiments that make up the on-

2 Informal surveys, also called exploratory or rapid reconnaissance surveys, are surveys
in which researchers interview farmers using informal, unstructured techniques in
order to encourage dialogue and probing of issues. This is often followed by a formal
survey during which data are collected by means of a questionnaire, which is usually
administered by enumerators to a randomly selected sample of farmers. The formal
survey is used to verify the findings of informal surveys and to quantify important
parameters (Byerlee and Collinson, 1980).


Journal for Farming Systems Research-Extension






DEFINING RECOMMENDATION DOMAINS


farm research process. Collinson (1982, 1987) proposes a distinct survey
exercise preceding the informal survey in which researchers administer a brief
questionnaire to extension agents or other key informants. The purpose of this
exercise is to elicit information for tentatively defining recommendation
domains over a larger area than can be effectively covered in an informal
survey.
The objective of this paper is to evaluate a brief survey of extension agents
in Kenya to define recommendation domains, similar to that proposed by
Collinson. The method has several potential advantages over the two options
discussed above for making a preliminary definition of recommendation
domains, i.e., using secondary information or an informal survey. First, in
many developing countries, secondary information is lacking, is of question-
able reliability, or is aggregated across widely different agroecological zones.
Second, using the informal survey to begin defining recommendation do-
mains is not cost-effective, because only small areas can be covered and much
researcher time might be wasted interviewing farmers who are not in the
recommendation domains that are eventually chosen for focus. The extension
agent survey can help to ensure that the recommendation domains are defined
fairly accurately before the survey begins, thus ensuring that informal surveys
are conducted more cost-effectively. Third, the area selected for the informal
survey may not be representative of larger areas; the extension agent survey can
provide information useful for selecting an area for the informal survey that is
as representative as possible of larger areas.
However, the extension agent survey also has potential drawbacks. The
technique is open to a wide range of error because prior to the informal survey,
researchers might know little about the area and might have trouble formu-
lating appropriate questions. Furthermore, the extension agents might give
inaccurate responses. If the errors are significant, the exercise will be wasteful
and misleading.
In this paper, the extension agent survey is evaluated by comparing the
information obtained with that of a formal survey of farmers carried out in the
same area during the same cropping season. First, the design and results of the
extension agent survey are presented and sources of error are assessed. Next,
data from the extension agent survey are compared to data from the formal
survey. The comparison is used for evaluating the effectiveness of the
extension agent survey for (1) providing preliminary information about
farmers in each domain and (2) demarcating recommendation domain
boundaries.


Vol. 3, No. 1, 1992






FRANZEL


DESIGN OF THE EXTENSION AGENT SURVEY
In 1981, the mid-altitude area (800 to 1,400 m) of Muranga, Embu, and
Kirinyaga Districts in Kenya was selected as the study area for an extension
agent survey to define recommendation domains by the Ministry ofAgricul-
ture in collaboration with the International Maize and Wheat Improvement
Center (CIMMYT). The study area covered roughly 4,000 m2 and was
populated by about 400,000 people. The objective of the survey was to define
recommendation domains and to select a target area to conduct surveys and
on-farm trials.
The exercise took three weeks and involved four steps. First, secondary data
on the area were assembled. These proved to be of limited use because most
of the data had been averaged across entire districts, which ranged from several
hundred to several thousand meters in altitude. Second, researchers, govern-
ment officials, and local leaders familiar with the area were interviewed
informally about agroclimatic differences in the area and socioeconomic
differences among farmers. Third, changes in agroecological zones and
cropping patterns were noted while touring the study area. Fourth, a brief
two-page questionnaire was drawn up based on the interviews and observa-
tions and was administered to extension agents in each sublocation, the
smallest administrative unit in Kenya. The respondents were asked about the
principal characteristics of the farming systems in their area and characteristics
that differentiated farmers between and within areas. Individual questions
concerned such aspects as physical environment, cropping patterns and
practices, livestock, and sources of income. Separate questionnaires were
completed for low-income and high-income farmers.
The interviews were conducted by the researcher; each interview lasted
about forty minutes. Twenty-two questionnaires were completed, most of
these when extension agents attended monthly meetings at their district
office.


RESULTS OF THE EXTENSION AGENT SURVEY
The results of the extension agent survey were used to select the medium
altitude areas ofKirinyaga District (called Middle Kirinyaga in this paper) for
further survey work and on-farm experiments. Middle Kirinyaga was selected
because the area (1) had high potential for maize production, (2) had a fairly
low adoption rate for improved technologies, and (3) was accessible through-
out the year.
Journal for Farming Systems Research-Extension






DEFINING RECOMMENDATION DOMAINS


Middle Kirinyaga has an area of about 170 km2 and a population of
approximately 35,000. Principal agroecological and socioeconomic charac-
teristics include 1,100 mm of rainfall distributed between two seasons; red
loam soils and a flat to mildly sloping terrain; farm size averages of 2 to 4 ha;
principal food and cash crops of maize and beans, which are almost always
intercropped; and farmers who either own or rent oxen for plowing. The
boundaries of Middle Kirinyaga represent fairly distinct changes in the agro-
ecological and socioeconomic environment. For example, to the north of
Middle Kirinyaga, altitude increases, rainfall is higher, and the terrain is
steeper. The southern boundary of Middle Kirinyaga is a fairly distinct line
representing a change in soil type from red loam to heavy, black clay. To the
east and west of Middle Kirinyaga are dry, hilly areas with less fertile soils,
lower population densities, and more uncultivated land.
The task of assessing socioeconomic differences among farmers in Middle
Kirinyaga was more difficult than defining agroecological differences. Two
factors were considered in the extension agent survey:
1. Characteristics of the farming systems. Researchers sought to determine
whether there were important differences in the way farmers managed their
farms, the resources they used, and their constraints. Important differences in
management, it was felt, would probably indicate differences in research needs
and development opportunities.
2. Potential. Researchers were concerned with the farmers' relative poten-
tial for adopting new technology, in terms of resource availability. Two
farmers may be operating their farms in a similar manner, but have varying
potential for adopting new technology because of differing resource availabil-
ity.
In Middle Kirinyaga, it appeared that level of cash income was an important
distinction among farmers determining both the characteristics of the farming
system and the potential for change. For example, cash income influenced
whether the farmer undertook certain high-cost, risk-increasing enterprises,
such as owning exotic breed cattle, or used purchased inputs such as hybrid
maize seed. These differences in the way farmers operate their farms affect the
type of experiments to be planned for farmers of different income levels. For
example, the number and range of experimental variables would generally be
greater for high-income farmers, since low-income farmers lacked cash for
purchasing improved inputs.
Another issue concerned the number of income categories to establish and
how to differentiate among them. Before the extension agent survey, research-


Vol. 3, No. 1, 1992






FRANZEL


ers classified farmers into two recommendation domains-high income and
low income. In subsequent informal and formal surveys, a set of proxies for
income was drawn up to differentiate high-income farmers from low-income
farmers. The proxies for high-income farmers included ownership of exotic
breeds of cattle, house type, past land purchases, and type of off-farm income.
A subjective weighting of these variables was used to allocate farmers between
the two groups. In only a few cases was there any uncertainty as to which group
a farmer belonged.


EFFECTIVENESS OF THE EXTENSION AGENT SURVEY

This section is divided into three parts. First, sources of error in the extension
agent survey and measures taken to reduce their effect are examined. Next, the
quality of data obtained in the survey is evaluated by comparing data obtained
with data collected in the formal survey of farmers. Finally, the delineation of
recommendation domains, based on the extension agent survey, is compared
with the categories arrived at following the formal survey.

Sources of Error
Eight sources of error that were encountered in the extension agent survey
are listed below. It is important to examine these sources so as to understand
why some estimates were incorrect and to avoid or correct for these in future
exercises. Five of the sources stem from the extension agents and three from
the researcher.
Respondent-based sources of error.
1. Bias towards male, "progressive" farmers. Extension agents tend to
frame their answers about farmer circumstances with the high-income, male,
"progressive" farmers in mind. To reduce this bias in the extension agent
survey, the respondent was repeatedly reminded of the difference between
male, "progressive" farmers and the range of other farmers in the survey area.
2. Bias resulting from attempts to impress investigators. Extension agents
tend to overestimate the number of farmers who have adopted recommended
innovations, perhaps because they are trying to impress the researcher or
because they are afraid their work is being evaluated. To counter this bias,
respondents were reminded that the researcher's goal was not to evaluate their
work, but to understand farmer circumstances. Respondents were also re-
minded that low adoption rates may be caused by many factors other than poor
extension.


Journal for Farming Systems Research-Extension






DEFINING RECOMMENDATION DOMAINS


3. Inaccurate information caused by inability of respondent to deal with
percentages. Some extension agents have only primary school education and
some are not comfortable dealing with percentages. Some progress was made
by using a continuum of all-most-half-some-few instead of percentages.
4. Inaccurate information caused by respondent's lack of knowledge. The
extension agents may think they know the answer to a question but are
mistaken. In other cases, they are afraid to say they do not know the answer
and give a false one.
5. Bias associated with answers that refer only to area that respondent
knows best. Some extension agents may give an answer that applies to the area
that they know best, usually the area around their home, but not to other parts
of the area they serve. The effect ofthis bias was reduced by using a map during
the interview and assuring that the responses were relevant to all areas within
the respondent's jurisdiction.
Researcher-based sources of error.
1. Bias resulting from attempts to compensate for perceived respondent
biases. In some cases, the researcher discounts or alters an extension agent's
response because it is inconsistent with the responses of other respondents or
because it contradicts the researcher's own observations. However, these
compensatory biases may be incorrect or may overcompensate for respondent
bias.
2. Improper phrasing of question or inappropriate narrowing of possible
responses. In the early stages of an investigation, researchers lack sufficient
knowledge to appropriately phrase questions; thus, they may inadvertently
preclude realistic possibilities. For example, in this study, the researcher asked
respondents to name the methods that farmers used to prepare their land,
without considering the possibility that some farmers practice no-till. It is not
clear that extension agents were aware that some farmers practice no-till, but
the manner in which the question was phrased did not encourage them to offer
this information.
3. Bias stemming from incorrect assessment of composition of a particular
recommendation domain. The researcher believed that most high-income
farmers were immigrants to the area and thus oriented the questions about
high-income farmers towards immigrants. Later it became clear that most
high-income farmers were natives of the area. Immigrants differ from other
high-income farmers in some important respects; responses concerning such
characteristics were biased towards immigrants.


Vol. 3, No. 1, 1992





FRANZEL


Evaluation of Data Describing Recommendation Domains
Table 1 presents the accuracy ratings and sources of inaccuracy for
estimates3 of 29 variables from the extension agent survey. The table shows
that over one-third of the estimates made were highly accurate and three-
quarters were ofhigh or moderate accuracy for both income groups. Appendix
1 shows the estimates of the 29 variables made following the extension agent
survey and those made from data collected in the formal survey.
The system of accuracy ratings was based on the degree of correspondence
of an estimate in the extension agent survey to sample estimates from the
formal survey. Formal surveys have inherent biases but are generally effective
for collecting data on phenomena that change slowly (e.g., farm size) or one-
time or infrequent occurrences (e.g., crops planted; Shaner et al., 1982).
Nearly all of the variables examined in the extension agent survey fall into one
of these categories (Appendix 1). Furthermore, the formal survey data are
believed to be accurate on the grounds that the survey (1) was administered
to a random sample of farmers, (2) used female enumerators to interview the
approximately one-third of household heads who were female, (3) used a
questionnaire, thus a set of standardized questions for all respondents, and (4)
was preceded by a comprehensive informal survey, which was important for
developing an understanding of the local terms of reference critical for
questionnaire development
Correspondence was measured in three different ways, depending on the
type of variable considered. If the variable estimated in the extension agent
survey was the percentage of farmers with a particular characteristic, such as
the percentage of farmers growing coffee, then the following system was used:
A margin less than 10 percentage points away from the sample percentage in
the formal survey was considered to be highly accurate, a margin of between
10 and 20 percentage points was considered to be moderately accurate, and
a margin of greater than 20 percentage points was considered to be of low
accuracy.
The particular boundaries between the ratings are arbitrary but broadly
reflect the degree of closeness and the effects that an error would have on the
understanding of the farming system and the defining of recommendation
domains. For example, the formal survey indicated that 85 percent of low-

3 In the extension agent survey and the formal survey, "estimates" of quantitative
variables (e.g., maize yields) or percentage variables (e.g., percentage of farmers
growing coffee) refer to the means of responses. For nominal variables (e.g., main
method of land preparation), the estimate is the one most frequently mentioned by the
agents.
Journal for Farming Systems Research-Extension






DEFINING RECOMMENDATION DOMAINS


Table 1. Accuracy of Estimates From Extension Agent Survey.

High-income farmers Low-income farmers
Number of Percent Number of Percent
Estimates Estimates
Level of accuracya
High 10 37 11 39
Moderate 10 37 10 36
Low 7 26 7 25
Total no. of estimates 27 100 28 100
Sources of inaccuracy
Respondent-based
Lack of knowledge 8 47 11 65
Progressive farmer bias 2 12 8 47
Researcher-based
Phrasing 1 6 2 12
Compensating for perceived
progressive farmer bias 1 6 2 12
Composition of
recommendation domains 8 47 1 6
Total of inaccurate estimates 17 100 17 100
a Accuracy means degree of correspondence with results from the formal survey.
Accuracy ratings and sources of inaccuracy are defined in the text.
b Figures do not sum to 100, since more than one type of bias may be encountered for
a particular estimate.
Source: Data are derived from the appendix.

income farmers intercrop their maize and beans. If the estimate from the
extension agent survey had been off by 10 or less percentage points it would
have made little or no difference in the understanding of the farming system
or demarcation of recommendation domains. The conclusion would still be
that the overwhelming majority of all low-income farmers intercrop. If the
estimate from the extension agent survey had been off by 30 percentage
points, i.e., if it had been estimated that 55 percent of farmers intercrop, the
researcher may have decided to demarcate recommendation domains accord-
ing to whether or not farmers intercrop. This would have been a fundamental
error, because the number of farmers who do not intercrop is so small that this
variable should not receive priority in defining recommendation domains.
For estimates of quantities, such as maize yields per hectare, an estimate
within 10 percent of the sample estimate was considered to be accurate, 10 to
20 percent of moderate accuracy, and off by more than 20 percent of low
accuracy. For nominal data, such as the two principal months in which farmers
experience labor shortages, a "correct" estimate was considered to be highly
accurate. A partially correct estimate, say estimating March and April to be the


Vol. 3, No. 1, 1992






FRANZEL


principal months of cash shortage when the formal survey indicates that April
and May are the principal months, was considered to be moderately accurate.
Estimates that were not even partially correct were considered to be of low
accuracy.
Surprisingly, the levels of accuracy of the data from the extension agent
survey are nearly identical for each income group (Table 1). Extension agents
tend to have more contacts with "progressive" and high-income farmers
(Leonard, 1977); thus it had been hypothesized that the accuracy ratings
would be higher for high-income farmers than for low-income farmers.
Indeed, extension agents were responsible for twice as many errors concerning
low-income farmers as for high-income farmers. This result is as expected:
extension agents showed a greater lack of knowledge and more progressive
farmer bias in responses concerning low-income farmers than in responses for
high-income farmers. Researcher-based errors were more numerous for high-
income farmers than for low-income farmers because of the researcher's
incorrect assessment of the composition of the high-income recommendation
domain. Thus, whereas respondent-based biases were the principal factor
lowering the quality of information about low-income farmers, researcher-
based biases were the principal source of inaccuracy concerning high-income
farmers. The net results were similar levels of accuracy in information
concerning both groups.
Two other, less important, researcher-based biases also distorted informa-
tion about farmers. First, inappropriate phrasing and formulating questions in
a way to exclude certain appropriate responses were more of a problem in
asking about low-income farmers than about high-income farmers. This
indicates, not surprisingly, that the researcher knew less about low-income
farmers' circumstances and practices than those of high-income farmers.
Second, modifying estimates to compensate for perceived respondent bias
sometimes decreased the accuracy of the estimate. Five estimates concerning
high-income farmers and seven concerning low-income farmers were revised
for such reasons. In six cases the modification improved the accuracy rating,
in four cases the modification lowered the accuracy rating, and in two cases
there was no change. Two examples typify the advantages and disadvantages
of this measure. First, extension agents reported that about half of the low-
income farmers purchased hybrid seed each year. However, a seed retailer
assured the researcher that low-income farmers rarely, if ever, bought hybrid
seed. In the report on recommendation domains the estimate was lowered,
and formal survey results confirmed that the modification was correct. A


Journal for Farming Systems Research-Extension






DEFINING RECOMMENDATION DOMAINS


second example shows how one can easily make mistakes using the compen-
satory bias tool. Extension agents estimated that one-third of low-income
farmers grew coffee. This figure was revised downward because ofa perceived
progressive-farmer bias on the part of the extension agents. In fact, the
extension agents' estimates were accurate, as confirmed in the formal survey.
A further issue concerns the reasons underlying respondent inaccuracies.
Estimates concerning crops grown, livestock, labor use, and income sources
were generally fair to good; poor estimates were made on manure use, maize
varieties, methods of land preparation, maize yields, and percentage of high-
income farmers intercropping. In several instances, it is possible to explain why
some estimates were poor. For example, estimates on manure use were poor
because of respondents' lack of knowledge; manure use is not part of the
extension agents' recommendations to farmers and thus they were unaware of
farmers' manure practices. Progressive farmer bias and lack of knowledge were
important sources of bias for the other four variables that were incorrectly
estimated, and improper phrasing of questions contributed to inaccuracy in
two of the cases.

Accuracy in Defining Recommendation Domains
Concerning agroecological differences, there were no important modifica-
tions of the preliminary set of boundaries established during the extension
agent survey. The only modification following the exercise involved excluding
a part ofRukanga sublocation, which included approximately 4 percent of the
population and 10 percent of the area of Middle Kirinyaga. During the
informal survey, it was apparent that rainfall was lower and farm size and
livestock numbers were greater in Rukanga than in Middle Kirinyaga. Thus
the area was excluded.
The attempt to define socioeconomic differences proved to be more
complicated and somewhat inaccurate. Two recommendation domains were
defined in the extension agent survey on the basis of income level and these
domains were maintained throughout each ensuing stage of analysis. But the
composition of the domains was modified following the informal survey. The
correspondence between full-time farming and low-income was dropped,
since some full-time farmers were found to have lucrative operations such as
dairy or tobacco and were thus high-income farmers. Similarly, the correspon-
dence between part-time farming and high-income was omitted because many
low-income farmers were found to be farming only part-time; their off-farm
activities were not profitable enough to make them high-income farmers.


Vol. 3, No. 1, 1992






FRANZEL


In summary, the exercise defining recommendation domains was reason-
ably accurate in defining and demarcating recommendation domains and in
providing preliminary information about the farmers in each recommendation
domain. This result is somewhat astonishing, because so little time and
resources were allocated to this exercise. The primary source of inaccuracies
were researcher-based biases concerning high-income farmers and respon-
dent-based biases concerning low-income farmers.


IMPROVING THE ACCURACY OF THE EXTENSION AGENT
SURVEY

By examining the sources of inaccuracy in the exercise it is likely that their
effects can be minimized in future investigations defining recommendation
domains. In addition, three important lessons arise from the exercise.
First, interviewing local leaders, e.g., sub-chiefs, in addition to, or in place
of, agricultural extension agents, may be useful in obtaining more valid
information about low-income farmers. This is also the finding of Njobvu
(1985) who used a similar survey in Zambia to define recommendation
domains. In Middle Kirinyaga, respondent-based biases were generally caused
by extension agents' lack of orientation towards low-income farmers. In
several cases, sub-chiefs were able to provide better information about their
sublocation because (1) they were natives of the area and (2) they had less of
a stake in over-reporting the use of recommended inputs than did extension
agents. However, their education level was generally lower than those of
extension agents. Thus, many had difficulties with such concepts as estimating
the proportion of farmers having a particular characteristic. Moreover the fact
that all were male and high-income may have contributed to biases against
low-income and female-headed households.
Second, more effort should have been made to identify the possible range
of responses to questions in the questionnaire, particularly concerning low-
income farmers. Because of the biases towards progressive, high-income
farmers, there were more phrasing errors concerning low-income farmers than
high-income farmers.
Finally, the exercise highlights the importance of correctly identifying
socioeconomic differences among recommendation domains before design-
ing questionnaires. In the case reported in this paper, a more thorough
investigation of the composition of the high-income recommendation do-
main through discussions with key informants should have been carried out


Journal for Farming Systems Research-Extension






DEFINING RECOMMENDATION DOMAINS


before administering the extension agent survey.


CONCLUSIONS
The extension agent survey described in this paper was found to be reasonably
effective for:
1. Tentatively classifying farmers into recommendation domains. Few
revisions were made in the domain boundaries as a result of the informal and
formal surveys that followed the extension agent survey.
2. Developing some preliminary information about farmers in the respec-
tive recommendation domains. About three-quarters of the estimates of
variables in the extension agent survey were moderately or highly accurate.
The extension agent survey also had important advantages over other
methods (e.g., secondary information or the informal survey) fora preliminary
demarcation of recommendation domains. Defining recommendation do-
mains using secondary data was impossible, because what data was available
was aggregated over a large area composed of several agroecological zones.
Compared to using the informal survey to define recommendation domains,
the extension agent survey had three important advantages. First, by defining
recommendation domains in the study area before the informal survey began,
researchers were able to decide which recommendation domains they wanted
to target during the informal survey that followed. Second, by delimiting
boundaries between the defined domains, little time was wasted during the
survey interviewing farmers who were not in the targeted domains. Third, by
including extension agents in the information gathering process, the research-
ers were able to gain their confidence and active participation in on-farm
research activities. Furthermore, the cost of the extension agent survey was
very low, lending further support for its use before beginning an informal
survey. Only three weeks of one researcher's time was required to cover an area
of 4,000 km2 and 400,000 people; this includes data collection, analysis, and
completing a report on the findings.
Finally, it should be noted that an extension agent survey cannot be used
to substitutefor an informal survey of farmers for understanding farming
systems. The exercise described in this paper involved fairly simple questions
about characteristics of farmers and areas that extension agents were capable
of an-swering. To identify problems and propose solutions, researchers must
care-fully observe local conditions and conduct detailed interviews with
farmers.


Vol. 3, No. 1, 1992







FRANZEL


ACKNOWLEDGMENTS

The author wishes to thank the International Maize and Wheat Improve-
ment Center (CIMMYT) and the Scientific Research Division, Ministry of
Agriculture, Kenya, for support of the research reported in this paper. The
advice and assistance of M. Collinson, E. Crawford, S. Haggblade, S.
Mwaniki, N. Njeru, J. Murage, and S. Goldmark are also gratefully acknowl-
edged. The views expressed in this paper do not necessarily reflect those of the
institutions or individuals mentioned above.


REFERENCES

Byerlee, D., and M. Collinson. 1980. Planning technologies appropriate tofarmers: Concepts
and procedures. International Maize and Wheat Improvement Center, El Batan, Mexico.
71 pp.
International Maize and Wheat Improvement Center (CIMMYT), Eastern Africa Economics
Programme. 1979. Deriving recommendation domains for Central Province, Zambia.
Demonstrations of an Interdisciplinary Approach to Planning Adaptive Agricultural
Research Programmes No. 5. Nairobi, Kenya. 38 pp.
Collinson, M. 1982. Farming systems research in Eastern Africa: The experience of CIMMTT
and some national agricultural research services, 1976-81. Michigan State University
International Development Paper No. 3. Michigan State University, East Lansing. 67 pp.
Collinson, M. 1987. Farming systems research: Procedures for technology development.
Experimental Agriculture 23:365-386.
Franzel, S. 1981. Identifying farmer target groups in an area: Methodology and procedures.
Farming Systems Newsletter 4:13-25. International Maize and Wheat Improvement
Center, Eastern Africa Economics Programme, Nairobi, Kenya.
Franzel, S. 1983. Planning an adaptive production research program for small farmers: A case
study of farming systems research in Kirinyaga District, Kenya. Unpublished Ph.D.
dissertation, Michigan State University, East Lansing. 279 pp.
Harrington, L., and R. Tripp. 1984. Recommendation domains: A framework for on-farm
research. Economics Program Working Paper 2/84. International Maize and Wheat
Improvement Center, El Batan, Mexico. 27 pp.
Hildebrand, P., and F. Poey. 1985. On-farm agronomic trials infarming systems research and
extension. Boulder, Colo.: Lynne Rienner Publishers, Inc. 162 pp.
Kean, S., and L. Singogo. 1988. Zambia organization and management of the adaptive
research planning team, Ministry ofAgriculture and Water Development. On-farm Client
Oriented Research Case Study No. 1. International Service for National Agricultural
Research, The Hague, Netherlands. 302 pp.
Leonard, D.K. 1977. Reaching the peasantfarmer: Organization theoryandpractice in Kenya.
Chicago, Ill.: University of Chicago Press. 223 pp.
Njobvu, C.A. 1985. Factors influencing recommendation domain boundaries of the farming
systems and levels of agricultural development in Lusaka Province, Zambia. Paper
presented at the Farming Systems Research-Extension Symposium, Kansas State Univer-
sity, Manhattan, October 13-16.
Shaner, W., P. Philipp, and W. Schmehl. 1982. Farming systems research and development:
Guidelines for developing countries. Boulder, Colo.: Westview Press. 414 pp.


Journal for Farming Systems Research-Extension







DEFINING RECOMMENDATION DOMAINS


APPENDIX 1. COMPARISON OF ESTIMATES OF VARIABLES
FROM EXTENSION AGENT SURVEY AND FROM FORMAL
SURVEY, MIDDLE KRINYAGA, 1981.


High-income farmers
Extension Formal survey
agent survey


Low-income farmers
Extension Formal survey
agent survey


Land and enterprises
Farm size (ha)
Area cultivated long
rains (ha)
Farmers who have
purchased land (%)
Main food crops
Main cash crops
Farmers growing coffee
(%)
Farmers growing sun-
flower (%)
Area under maize (%)
Area under beans (%)
Farmers owning Zebu
cattle (%)
Farmers owning exotic
cattle (%)
Farmers owning cattle
(%)


40 23
maize-beans maize-beans
maize-beans beans-maize


Maize-bean husbandry
Farmers intercropping
maize-beans (%) 20
Method of land preparation (%)
hire ox 60
own ox 20
borrow ox 0
rent tractor 20
no-till 0


Farmers growing long rains maize varieties (%)
hybrid 67 36
2nd gen. hybrid 10 28
katumani 33 49
local 0 59

Farmers growing short rains maize varieties (%)
hybrid 10 0
2nd gen. 0 38
katumani 90 51
local 0 51


Vol. 3, No. 1, 1992


3.0
2.5


0
maize-beans
maize-beans

15

5
90
90

71

15

80



88

50
45
0
5
0


0
5
10
55


0
30
30
50


2.4
1.4


2
maize-beans
beans-maize

25

0
91
78

56

2

62



85

0
27
33
0
21


0
21
69
77


0
29
60
71







FRANZEL


APPENDIX 1. COMPARISON OF ESTIMATES OF VARIABLES
FROM EXTENSION AGENT SURVEY AND FROM FORMAL
SURVEY, MIDDLE KRINYAGA, 1981. CONT.

High-income farmers Low-income farmers
Extension Formal survey Extension Formal survey
agent survey agent survey
Farmers using fertilizer
(%) 20 18 5 2
Farmers using manure
(%) 30 82 10 49
Average maize yield
(kg/ha) 2,500 1,200 1,900 1,100

Income, labor, and other
Farmers getting income
from relatives (%) 0 0 10 11
Farmers having salaried jobs
or businesses (%) 75 69 20 21
Busiest months March-May April-May March-May April-May
Activity at busiest time P, W W P, W W
Farmers hiring labor (%) 90 74 10 17
Farmers having zinc roofs
(%) 100 97 50 72
Main distinguishing factor
between domains part-time high income full-time low income
farming/off- farming/no
farm income off-farm income
Farmers in this domain
(%) 20 40 80 60
a P = planting; W = weeding.
Source: Survey data, as reported in Franzel (1983).


Journal for Farming Systems Research-Extension








The Role of Farmers in the Jordanian
Combined Sondeo Process'

D.L. Gait and A.F. Al-Kadi 2



ABSTRACT
A rapid rural appraisal (RRA) technique called the "combined sondeo"
evolved in Jordan to help the staffof the National Center for Agricultural
Research and Technology Transfer (NCARTT) of the Ministry of
Agriculture (MOA) to prioritize their annual research and extension
activities by directly incorporating farmers' priorities and problems on a
subnational basis. The purpose of this paper is to explain how this RRA
technique allows explicit incorporation of farmers' ideas, opinions, and
suggestions into the annual national work plan process for research and
technology transfer and to describe briefly how this process is evolving
and becoming a sustainable part of the National Agricultural Research
System (NARS).


INTRODUCTION

The FSRE approach officially began in Jordan in January 1987, with a
"training of trainers" workshop (Gaudreau et al., 1989), sponsored by the
U.S. Agency for International Development (USAID)-funded Jordan Na-
tional Agricultural Development Project (JNADP) and implemented by the
Consortium for International Development (CID) through Washington
State University (WSU) as the lead university. This training was followed by
a second JNADP-sponsored training of NCARTT researchers and technolo-
gy-transfer (i.e., extension) staff in August 1989 (Gaudreau et al., 1989).
During both training, participants carried out socioeconomic field research.
The second research training included use of the classical sondeo technique
of questioning farmers about their prevalent systems and problems (Hilde-
brand, 1981). The field work also included adaptations of relevant experienc-
es from the Nepalese Samuhik Bhraman process (Mathema and Gait, 1988,
1 Paper presented at the Tenth Annual Association for Farming Systems Research-Extension
Symposium, Michigan State University, East Lansing, October 14-17, 1990.
2 Head, Monitoring and Evaluation Unit, MCARTI, Ministry of Agriculture, Jordan, and
Sociometric and FSRE Technical Assistance Advisor to NCARTT through JNADP, Jordan,
respectively.






GALT AND AL-KADI


1989; Chand and Gibbon, 1990). The Jordanian combined sondeo is another
example of how RRA techniques continue to be incorporated into full-scale
FSRE approaches (Chambers, 1981; Chambers and Ghildyal, 1985; Mathe-
ma and Galt, 1988, 1989; Chand and Gibbon, 1990). Most importantly,
however, the evolving combined sondeo in Jordan is a particularly cost-
effective way to collect and analyze rural data quickly (Kumar, 1987).
During the second training, participants interacted with farmers in Jordan's
six regional centers. These six Regional Agricultural Service Centers, or
RASCs, are located in Ramtha (in the north), Khaldieh (in the northeast), Deir
Alla (in the Jordan Valley north of the Dead Sea), Mushaqar (just south of the
Amman area), Rabba (between the southern most RASC and Mushaqar), and
Shoubak (covering approximately one-half of the area of Jordan and located
in the southern half of the Kingdom). Recently, a seventh region, Al-Balca,
which includes the national research headquarters in the Al-Baqa'a Valley, has
been added as a RASC. These seven RASCs represent the beginning of
administrative decentralization of research and technology transfer in Jordan.
These interactions with area farmers lasted one week. During this week,
individual farmer interviews occupied the first two or three days. Next,
participants spent one or two days listing (1) priority farming systems and (2)
farmer-identified problems in these systems. At the same time, they suggested
tentative solutions to those problems judged to be most important.
The final day of the field week was set aside to host a joint farmer-
researcher-extension agent meeting to discuss prioritized problems and
tentative solutions to them. Finally, the research and extension participants
were responsible for producing modified work plans for 1989/90 for each
RASC, based on (1) 1988/89 work plans and (2) the assimilated input
obtained from farmers. At some RASCs, a few trial design details were also
included in the draft research work plans (Gaudreau et al., 1989).
During the spring of 1990, staff at each RASC (led by the RASC director,
the research supervisor, and the extension supervisor and assisted by the
socioeconomist) hosted follow-up meetings with groups of invited farmers.
Also attending these meetings were representatives from appropriate regional
Directorates of Agriculture and interested commodity section heads or
representatives. More details concerning these joint farmer-researcher-
extension agent meetings are presented in Al-Kadi and Gait, 1990b.
The results of these joint meetings allowed RASC staff to plan research and
extension activities for 1990/91, taking farmer priorities directly into ac-
count. Because it culminates with the annual NCARTT planning meeting, the


Journal for Farming Systems Research-Extension






FARMERS' ROLE IN COMBINED SONDEO PROCESS


process is institutionalized. The institutionalization of explicit farmer opinion
in the annual NCARTT planning process makes the Jordanian FSRE approach
sustainable into the future.
A key to the process is that neither research nor extension are able to ignore
the priorities placed on activities in crop or livestock-specific areas, because
these farmer-prioritized lists are used in follow-up meetings. This does not
allow enough time for refiltering the priorities through normal research and
extension disciplinary thought processes, so no unintentional reprioritization
takes place when the combined sondeo process is followed. The rest of this
paper describes the methodology of this process in more detail and presents
some of its dynamics.


METHODS

Initial Farmer Meetings
NCARTT commodity and disciplinary research is organized around nine
sections or units: (1) Stations and RASC Section, (2) Cereals Section (with
responsibility for research on field crops), (3) Communications Section, (4)
Fruit Trees Section, (5) Plant Protection Section, (6) Range and Livestock
Section, (7) Soils and Irrigation Section, (8) Vegetable Section, and (9)
Monitoring and Evaluation Unit. As the two major components of agricul-
tural development, research and extension (or "technology transfer" in
Jordan) are separated into their own divisions. Organizationally, these nine
sections and units are in the research division.
During initial field site visits using the Jordanian combined sondeo process,
researchers selected by Section Heads from each NCARTT commodity
section join regional researchers and technology-transfer staff in farmer
interviews. Interviews are carried out by two-person teams. The initial
combined sondeo field activities take one week to complete.
The methodology used consists of three major components: (1) interview-
ing individual farmers during the first few days of the week; (2) listing
predominant systems and problems within systems, discussing these lists, and
advancing tentative solutions to them (during the latter part of the week); and
(3) convening a meeting at the end of the week that includes farmers,
researchers, and extension agents. Output from the process includes (1) lists
of predominant farming systems for the area, (2) lists of problems prioritized
by the frequency with which they are mentioned by farmers, and (3) a draft
RASC research and technology-transfer work plan.


Vol. 3, No. 1, 1992






GALT AND AL-KADI


Individual farmer interviews. During two or three days of individual
farmer interviews, farmers are interviewed where they are found--either in
their homes or their fields. Lists offarmer's systems, problems within systems,
and farmer's suggestions to overcome some of these problems are noted by the
interview team during and immediately after each interview.
The civil service working hours in Jordan are not conducive to completing
field work. The Jordanian work week consists of six, six-hour work days
(Saturday through Thursday), from 8:00 A.M. to 2:00 P.M. To extend this
time to a meaningful length (so that either farmer interviews or group
discussions can continue beyond 2:00 P.M.), it is necessary to provide a meal
for the participants. There is no national mechanism in place to provide meals.
Also, because governmental transportation leaves from each regional center
at 1:55 P.M. each day, additional transportation must be ananged in advance
to allow professionals to stay beyond 2:00 P.M. Thus, in the Jordanian
context, use of outside funds has been necessary to finance part of the extra
transportation costs, to purchase fuel for this transportation, and to pay for
meals for professional staff. Therefore, the combined sondeo is not currently
sustainable in Jordan unless outside donor funds are made available on a
continuing basis to support it, or the appropriate national laws are changed.
Listing systems, problems, and solutions. After interview teams return to the
RASC and finish lunch, their notes on predominant systems and farmer's
problems within these systems are transferred to master lists on flip charts in
an informal group setting. These lists, in turn, redirect research and technol-
ogy-transfer plans for the following year.
These lists then are used by the professional staff to prioritize farmer-
identified problems on a regional basis, using the simple method of farmer-
reported frequencies: The more often a problem is mentioned by farmers, the
higher its priority is assumed to be. Problems then are divided into three
subgroups: those to be addressed by (1) researchers, (2) extension, or (3)
other policymakers.
The remaining time is spent in group brainstorming sessions to arrive at
tentative solutions to the first two sets of problems. This consists of (1)
proposing trials to solve research-related problems and (2) proposing exten-
sion activities to address problems for which research solutions now exist, but
for which extension information has not yet been provided to farmers.
Final joint meeting and draft regional work plans. The mechanism for
convening the farmer group meeting is as follows. During the first two to
three days of individual farmer interviews, each farmer interviewed is invited


Journal for Farming Systems Research-Extension






FARMERS' ROLE IN COMBINED SONDEO PROCESS


to attend the closing group meeting. In our experience, an individual farmer's
propensity to attend this final group meeting was a function of four factors:
his or her (1) schedule for that particular day, (2) distance from the RASC, (3)
access to public or private transportation, and (4) personal interest in the
combined sondeo process. In 1989, across the six RASCs, a total of 238
individual farmers were interviewed during the three days devoted to this
combined sondeo activity. Of these, 94 farmers (or 39%) participated in the
final group meetings (Gaudreau et al., 1989).
During this final meeting, the summarized priorities and problems are
reviewed for the farmers by means of RASC research and technology-transfer
staffpresentations. RASC staff then outline next year's proposed research and
technology-transfer draft work plans. Following these informal presentations,
farmers offer their opinions on any part of the revised program. RASC staff
take note of the suggestions that can be incorporated into improving their
proposed activities. Occasionally, for problems that fall into the policy area
outside of the mandates of both research and extension, the group is able to
agree on a set of problems, tentative solutions, and a plan of action to solve
them. Results of these discussions are noted by the socioeconomic staff, and
this information is provided to other MOA decision-makers at a later date
(Galt and Al-Kadi, 1989).
A practical example of farmer-generated policy issues occurred in Deir Alla.
The farmer-researcher-extension agent group reached agreement on the
following two points: (1) styrofoam containers (for transporting fruits and
vegetables to market) are too expensive, pollute the environment to a great
and very persistent degree, and should be replaced by an alternative (wooden
or pasteboard) container and (2) farmer representatives must be included if
and when fruit and vegetable grading standards are proposed and developed
for the Kingdom. The group then proceeded to discuss how these problems
could be addressed by other policymakers in the government. The results of
this meeting were summarized, written up, and forwarded to the MOA (Gait
and Al-Kadi, 1989).
This type of informal group meeting format allows for direct farmer input
into the formation of research and technology-transfer activities, assisting in
the fine-tuning of solutions that may already be formulated for the problems
they identified earlier in the week. The output of the process is sets of annual
work plans that have been modified to take high-priority, farmer-identified
problems and constraints into account. Headquarters-based research and
extension personnel have less chance to reprioritize these lists than personnel


Vol. 3, No. 1, 1992






GALT AND AL-KADI


in the normal sondeo process, during which trial design and extension activity
generation are physically separated from the farmer-interview process. This
point, though subtle, is a key to incorporating real-not reinterpreted-
farmer priorities into research and extension work plans. Finally, as shown
above, the informal group setting provides a way for the whole group to
forward concrete suggestions to other governmental decision-makers for
solving policy issues that adversely affect agriculture in the region.

The Recurrent Nature of the Combined Sondeo
The initial combined sondeo represents only the first step of the process.
It is followed by annual, recurring combined sondeos. The recurrent
combined sondeo is the first part of a sustainable FSRE approach in Jordan.
It is designed to accomplish two things: (1) to update the list of prioritized
problems annually through a series of area meetings with farmers and (2) to
allow each RASC to directly incorporate these reprioritized problems into a
revised draft research and extension work plan for the following year.
Follow-up meetings withfarmers. The second part of the combined sondeo
methodology involves updating work plans annually to capture new farmer
concerns by fine-tuning on-going research trials and extension activities.
These follow-up combined sondeos occur each spring at each RASC. A
meeting is arranged by RASC staff to host farmers from each region. Those
NCARTT commodity sections with sufficient time and interest assign staff
members to participate in these meetings. Meetings last either one or two
days, depending on the complexities of the systems and/or the frequency of
problems in the particular area. More details of the dynamics of these
meetings have been provided elsewhere (Al-Kadi and Gait, 1990b).
Follow-up survey meetings start with a briefintroduction to the region and
to the research and technology-transfer activities underway there. Next, a list
of those problems identified by farmers during the previous combined sondeo
are presented on flip charts (e.g., Gaudreau et al., 1989). An open discussion
of these problem follows. During this discussion, existing problems are
clarified, amplified, refined (divided into two or more subproblems), or
eliminated if they no longer exist or if they are no longer considered important.
Finally, new problems are discussed, clarified, and added to the master list.
Drafting revised annual RASC work plans. Following this discussion, two
approaches have been used to move toward potential solutions to problems.
These approaches are (1) prioritization of all problems and (2) prioritization
of problems after separation into major commodity groups.


Journal for Farming Systems Research-Extension






FARMERS' ROLE IN COMBINED SONDEO PROCESS


Approach 1: Overall prioritization ofproblems. In 1990, all regions except
Ramtha prioritized problems directly from the revised master problem list.
Problems were not separated into major commodity groups. For example, the
problem of "insufficient feed for livestock" was considered along with the
problems of "excessive drop of citrus fruits," "not enough laborers," and
"poor agricultural roads." While this is a less time-consuming method, it is
not the preferred method as it involves combining unrelated topics.
Approach 2: Prioritization ofproblems by commodity type. The preferred
method is to divide farmer-identified problems into lists by major commod-
ities research sections; for example, "insufficient feed for livestock" would be
considered along with all of the other problems being dealt with under the
category of "Range and Livestock." Likewise, "not enough laborers" can be
considered either a socioeconomic research problem (to be addressed by the
Monitoring and Evaluation Unit), or it can lead to more research into those
crop production processes that are hypothesized to be mechanized most easily
in the Jordanian setting. Finally, problems such as "poor agricultural roads"
are placed on the public policy list to be forwarded to upper-level govern-
mental decision-makers.
The only region to separate problems by section commodity types in 1990
was Ramtha. Here, four commodity types (cereals, fruits, vegetables, and
range and livestock) were selected, and problems were first grouped into
separate master lists by these four groups. During the balloting process
(detailed in the following paragraphs), votes were collected and analyzed
separately from farmers, researchers, and technology-transfer staff for each
group.
Although this second approach takes more time to complete (about four
times as much time is required for data entry, for example), it leads to more
logical lists of common problems. Indeed, when approach 2 was used in
Ramtha in 1990, it was easier for the group to focus on tentative solutions to
problems that should be addressed by either research or extension. This was
because it was easier to concentrate on these more homogeneous groupings
than it was to jump from topic to topic (e.g., from livestock to cereals to
vegetables and back to livestock topics).
Once the master problem list was revised to everyone's satisfaction, simple
ballot forms were distributed to farmers. These ballots were designed to allow
each participating farmer to vote for the three most important problems that
affect his or her system. After the ballots were distributed, the meeting
facilitator gave a brief explanation of how to complete the ballots.


Vol. 3, No. 1, 1992






GALT AND AL-KADI


At some RASCs, not enough farmers attended the meeting. At these
RASCs (Khaldieh, Ramtha, and Deir Alla), both the researchers and the
extension agents joined in the voting to prioritize problems. In such cases,
each ballot cast was coded to identify the background of the voter: "Fl" =
"farmer one;" "R3" = "researcher three;" "E4" = "extension agent four;"
"A2" = "administrator two." No attempt was made to compare problem
rankings between these subsets of voters to see if their perceptions ofproblem
importance are similar or different, because the sample size was quite small.
It is significant that Jordan returned to a parliamentary democracy in
November 1989, following 22 years ofemergency authoritarian rule. For this
reason, the voting process, in addition to being extremely quick, was very
popular with both farmers and NCARTT staff. Such a process should also be
popular in other parts of the world now working in a more democratic mode.
By conducting an open meeting followed by individual problem prioriti-
zation through a secret ballot process, the combined sondeo allows a full-
blown, open group discussion with minimized dominance of strong
personalities, vocal farmers, and farmers of high local political standing. The
over-dominant impact of these three subgroups of farmers has long been the
major draw-back to holding farmer group meetings to obtain "averages" in
regard to important rural issues.
When the ballots had been filled out, data from them were entered directly
into a software program using a Zenith portable laptop computer. The
software program used, Decision Pad (Apian Software, 1990),3 is one of
several programs developed to facilitate the process of making difficult group-
based decisions. It allows the user to identify up to 250 alternatives for action,
each ofwhich can be scored against up to 250 criteria. Criteria can be grouped
together and/or assigned different weights. To make group decision-making
easier, up to 60 different people can "vote" on the importance of each
alternative. If the need arises, different weights can be assigned to different
voters.
Depending on the number of voters and how data were input, data entry
took between 30 minutes and one hour to complete. This time estimate does
not include preparation time for data entry. Data entry template files were
created and set up for each RASC in advance of each meeting. To record votes,
very little time is required, because only the values 1,2, or 3 are possible across
a single criterion (or vote, in this case)w and because farmers were limited to

3 Names of products are included for the benefit of the reader and do not imply endorsement or
preferential treatment by the AFSRE.


Journalfor Farming Systems Research-Extension






FARMERS' ROLE IN COMBINED SONDEO PROCESS


voting for their top three problems. Data entry time naturally varies also by
the number of ballots cast per region and by the "keyboarding experience" of
the person performing data entry.
In addition, there is the methodological difficulty of working with ranks of
numbers where the top-ranked problem is 1, the next most important
problem is 2, etc. Using this natural way of ranking means that any number
greater than "3" must be entered into every cell for which a problem received
no vote (i.e., was not ranked among the top three problems). This is because
mathematically a "0," or a blank cell, is ranked higher than 1, whereas these
values really mean that the problem is not in the voter's top three in
importance. Thus, it was necessary to enter a "4" in each cell for problems that
did NOT receive either a 1, 2, or 3 vote by each farmer.
Data entry was slightly more rapid if"4"s were pre-entered into all cells of
the anticipated number of farmer voters during the early hours of the joint
meetings. Where this procedure was followed, the staffmember entering data
only had to change three of the 4s to 1, 2, and 3, corresponding to each
farmer's vote for first, second, and third most important problems on a ballot-
by-ballot basis.
Decision Pad scores all votes and reranks problems automatically. After
data entry, a single key-stroke produces a ranked listing of problems on the
screen. A hard copy print-out of these reranked problems then is produced
from the program's "report" menu. This report is used by the meeting
facilitator to assign problem ranks to the problems on the master list, using a
contrasting colored marker pen. The report is also given to RASC staffto assist
them in documenting changes in their draft work plan.
When the master list of problems on the visual flip-chart pages has been
updated by indicating the new priorities based on frequency ofvote, a second
open discussion follows. This time, tentative solutions to some of the most
important problems are proposed by participants and discussed by the whole
group.
If time permits, the research and technology-transfer staff reconvene late in
the afternoon to work toward technical solutions to the most important
research and extension problems. In situations where the complexity of
systems was especially high (e.g., both Deir Alla and Ramtha) and/or there
were a large number of problems (e.g., in Deir Alla)-approximately 30 or
more-an additional day was required to complete this step of the follow-up
combined sondeo process.


Vol. 3, No. 1, 1992






GALT AND AL-KADI


RESULTS

Initial Combined Sondeo
In 1989, as in 1987, Jordanian farmer opinion had an impact on some
planned research and technology-transfer activities of NCARTT. In 1989,
those farmers who attended the final group meeting of the initial combined
sondeo acted as an ad hoc advisory panel to the staffs of each of the six regional
centers.
The lists of problems assisted RASC staff attending the FSRE training
workshop to revise RASC work plans for 1989/90. The results of this
combined sondeo were then summarized, published, and distributed to a
wide range of MOA decision-makers on a Kingdom-wide basis (Galt and Al-
Kadi, 1989; Gaudreau et al., 1989). In addition, problems were separated
into those that have an impact on different commodity sections. These were
summarized, published, and distributed to NCARTT commodity sections
through a series of Research Information Bulletins (Al-Kadi and Galt, 1990a).

Follow-up Farmer Meetings
Overview of the process. Short and efficient follow-up combined sondeos
were held in each region in the spring of 1990. Master problem lists from the
initial 1989 combined sondeo were presented on flip charts to a group of
invited farmers. This farmer group assisted NCARTT, RASC, and headquar-
ters research and technology-transfer staff to reprioritize these problems
according to 1990 realities.
It should be noted that the initial combined sondeos were carried out as the
Jordanian dinar (JD) was undergoing a major devaluation to bring an
overvalued currency more into line with other world currencies and to
promote exports. Consequently, farmers interviewed during the initial
sondeo in August 1989 expressed a great deal of uncertainty about the future
of the JD and their own future farming plans. The follow-up combined
sondeos carried out in the spring of 1990 took place after the JD had stabilized
against the U.S. dollar at about US$0.68 = JD 1.000 (down from US$0.34
= JD 1.000 during the fall of 1988). In 1990, farmers expressed less concern
about economic unknowns.
However, this experience reveals some of the difficulties of attempting to
extrapolate from socioeconomic research that occur when there is a high
degree of social uncertainty, unless such uncertainty can be considered
"normal." Based on our experiences in Jordan, we recommend that follow-


Journalfor Farming Systems Research-Extension




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

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