• TABLE OF CONTENTS
HIDE
 Front Cover
 Title Page
 Table of Contents
 The concept of the average farmer...
 An evaluation of soybean planting...
 Systems of production and production...
 Evolving crop-livestock farming...
 Growth, forage content, and biomass...
 Survival and sustainability in...
 Circumstances of rapid spread of...
 Gender differences in livestock...
 Institutional linkages that enhance...
 Farm-level evaluation of adoption...






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/UF00071921/00010
 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: VID00010
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
        Page i
        Page ii
    Table of Contents
        Page iii
    The concept of the average farmer and putting the farmer first: The implications of variability for a farming systems approach to research and extension
        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
    An evaluation of soybean planting methods for small-scale farmers in central Province, Zambia: A farming systems approach
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
    Systems of production and production of knowledge: Reflections on the basis of Ivorian and Mexican experiences
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
    Evolving crop-livestock farming systems in the humid zone of west Africa: Potential and research needs
        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
    Growth, forage content, and biomass yield response to on-farm fodder tree species trials in Nepal
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
    Survival and sustainability in the midwestern hills of Nepal
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
    Circumstances of rapid spread of cultivation of improved cassava varieties in Nigeria
        Page 93
        Page 94
        Page 95
        Page 96
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        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
    Gender differences in livestock production management in the Chitwan district of Nepal
        Page 121
        Page 122
        Page 123
        Page 124
        Page 125
        Page 126
        Page 127
        Page 128
        Page 129
        Page 130
        Page 131
        Page 132
        Page 133
        Page 134
        Page 135
        Page 136
    Institutional linkages that enhance the value of on-farm research for smallholder farmers: The Zimbabwe experience
        Page 137
        Page 138
        Page 139
        Page 140
        Page 141
        Page 142
        Page 143
        Page 144
        Page 145
        Page 146
    Farm-level evaluation of adoption and retention of maize variety in central Province, Cameroon
        Page 147
        Page 148
        Page 149
        Page 150
        Page 151
        Page 152
        Page 153
        Page 154
        Page 155
        Page 156
        Page 157
        Page 158
Full Text

Volume 4, Number 3

t 1994



o urnal

for Farming Systems

Research- Extension

1 Conep of th Avrg Fame an utn heFre is
Ja e Beebe. I
17 A Evauatin of Soba PlnigM tos o ml-cl
Famrsi CetalPoicZm a
M. B 0unh F.I Olen P.T Gison a0 I S. *t
31~~~~~' SsesoPrdcioanPrdcinfKoweg



M.A jleab S.*
61 Got, Foag Cotet an Bioms ildRsos


S so K VadaadDvdGbo
93 CrcumtanesoRSid Spread ofImroedCs saa
Varieties ~ inNgei
Fei I. Nive .K an an B.O S 0g 0 v


Eno M. *i.m
147 FamLee Evlaino dpionadReeto
of Maiz Varet in Cenra Prvicaero
*ewylc Mosi and Auusi I *







Journal
for Farming Systems
Research-Extension


Volume 4, Number 3, 1994


Published by
the Association for Farming Systems Research-Extension









Journal for Farming Systems Research-Extension


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

Associate Editors
Jennifer J. Manthei, Claude Bart, and M. Katherine McCaston
Office of Arid Lands Studies
The University of Arizona, Tucson

Production and Layout
Jennifer J. Manthei and Sonia Telesco
Arid Lands Design, Office of Arid Lands Studies
The University of Arizona, Tucson

Sponsors
Ford Foundation
The University of Arizona




The Journalfor Farming SystemsResearch-Extensionis 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 require-
ments 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 4, Number 3, 1994




CONTENTS




1 The Concept of the Average Farmer and Putting the Farmer First:
The Implications of Variability for a Farming Systems Approach to Research
and Extension
James Beebe
17 An Evaluation of Soybean Planting Methods for Small-Scale Farmers in
Central Province, Zambia: A Farming Systems Approach
M. Bezuneh, F.J. Olsen, P.T. Gibson, and K.S. Chanda
31 Systems of Production and Production of Knowledge: Reflections on the
Basis of Ivorian and Mexican Experiences
Jean-Philippe Colin
47 Evolving Crop-Livestock Farming Systems in the Humid Zone of West
Africa: Potential and Research Needs
M.A. Jabbar
61 Growth, Forage Content, and Biomass Yield Response to On-Farm Fodder
Tree Species Trials in Nepal
Madhav B. Karki and Michael A. Gold
75 Survival and Sustainability in the Midwestern Hills of Nepal
Ashok K. Vaidya and David Gibbon
93 Circumstances of Rapid Spread of Cultivation of Improved Cassava
Varieties in Nigeria
Felix I. Nweke, S.K. Hahn, and B.O. Ugwu
121 Gender Differences in Livestock Production Management in the Chitwan
District of Nepal
Pradeep Tulachan and Asha Batsa
137 Institutional Linkages that Enhance the Value of On-Farm Research for
Smallholder Farmers: The Zimbabwe Experience
Enos M. Shumba
147 Farm-Level Evaluation of Adoption and Retention of Maize Variety in
Central Province, Cameroon
Menwuyellet Moussie and Augustin Foaguegue








The Concept of the Average Farmer and
Putting the Farmer First:

The Implications of Variability for a Farming

Systems Approach to Research and Extension 1

James Beebe 2


ABSTRACT
As the Green Revolution has brought more farmers into cropping species with
greater variability of yield, the analysis of the conditions under which people
farm has become more critical to understanding and improving livelihoods.
Data from Malawi and the Philippines demonstrate that when even a few
variables are used together to assess the characteristics of "average" farmers,
the differences quickly become more significant than the similarities; they
often range into the hundreds of percentage points. The use of "recommen-
dation domains" to focus research and extension cost effectively on problems
faced by many has been counterproductive for extension to the degree that it
has encouraged practitioners to overlook important variations, assume ho-
mogenous groups, and prescribe farming innovations for them. By contrast, in
a "farmer-centered" approach, extension agents offer information and train-
ing to support farmers' decisions about agricultural innovations. Only such an
approach can deal successfully with variability.

INTRODUCTION
Extension programs in many places are premised on the identification of the
"average" farmer. Extension services then attempt to get all the farmers in a
"recommendation domain" a group of farmers believed to be homogeneous
to adopt practices that research has indicated should be better than the existing
practices of the farmers. This paper examines the validity of assumptions about
"average" farmers and the relationship of these assumptions to the relatively
1 This paper was prepared during a Reverse-Joint Career Corps Assignment at Oregon State
University. The cooperation of the Office of International Research and Development and the
Department of Anthropology at Oregon State University are appreciated. The paper was
presented at the Tenth Annual Symposium Farming Systems Research-Extension, October
14-17, 1990, Michigan State University, East Lansing, Michigan, USA.
2 Director, Office of Economic Development, United States Agency for International Develop-
ment (USAID)/South Africa. Comments are welcome and can be addressed by INTERNET
electronic mail to jbeebe@usaid.gov or mailed to USAID/Pretoria (ID), Department of State,
Washington, DC 20521-9300, USA. The opinions and views expressed are those of the author
and not necessarily those of USAID.






BEEBE


passive role many extension systems assign to farmers. This paper will (1)
review evidence on variability among farmers, with special attention to
examples from Malawi and the Philippines, (2) consider the implications of
this evidence for the concepts of "average" farmers and recommendation
domains, and (3) suggest the need for a significant change in the relative roles
of farmers and extension workers.
Research variability, whether biological or social, implicitly assumes vari-
ability and uses it to test hypotheses. Possibly because variability is so central
to research, it has received only limited explicit attention; the literature on
farmers and international agriculture includes very few explicit references to it.
Jock Anderson and Peter Hazell's book, Variability in Grain Yields:
Implications for Agricultural Research and Policy in Developing Countries
(1989), summarizes much of the literature about variability of grain yields. It
provides some reference to variability among farmers, while illustrating how
rarely variability has been an explicit issue. Papers in this book examine how
grain yields have become' more variable even as production has increased.3
Although total cereal production for the world grew at an average yearly
rate of 2.7 percent between 1960 and 1982, absolute variability around this
trend also increased substantially and significantly. Variability has increased
faster than average production.
Increases in variability in grain yields have been associated with the "Green
Revolution." Mehra (1981) and Barker, Gabler, and Winkelmann (1981)
argue that a large part of the increased instability in grain yields is due to
widespread adoption of improved seed and fertilizer-intensive technologies
since the Green Revolution of the mid-1960s. They argue that introduction
of the new varieties results in a reduction in the number of traditional varieties
planted, and that this increases variability in yield over time. Some of the new
varieties appear to be particularly sensitive to weather and disease, and the
narrowing of the genetic base increases their susceptibility to pest, disease, and
weather conditions. If a variety is planted widely, and is susceptible to a given
pest or disease, the results can be disastrous. An often-cited example is the
devastating impact on corn in the United States in 1970 southern corn leaf
blight (Helmintho-sporium Maydis) (Hargrove et al., 1979). Another example
is the decline in rice production by about 30 percent from the previous two-
year average yield in the Philippines in 1971-72, due primarily to tungro virus
damage (Barker et al., 1981:70). The new varieties of rice are particularly
sensitive to variability in rainfall. Their relatively poor performance compared
to the local varieties under rain-fed conditions appears to be related to the
shorter growing period of the new varieties.
3 Economics divides yield variability over time into trend, cyclical and year-to-year components.
Year-to-year variability is measured in terms of the deviation of yield from trend. The standard
deviation of the distance of observations from the trend is calculated and this term is divided
by the mean to determine the coefficient of variation. The standard deviation is a measure of
absolute variance and the coefficient of variation is a measure of relative variance (Barker et al.,
1981:55).


Journal for Farming Systems Research-Extension






CONCEPT OF THE AVERAGE FARMER


Although increases in yield variance are partly due to the widespread
adoption of improved seed and fertilizer-intensive technologies since the mid-
1960s, the increases in yield variance can also be attributed to other factors
such as increased price variability (Hazell, 1989:31), increase in the areas
cropped in more marginal land (Nguyen, 1989:78), changes in policies
(Tarrant, 1989:60) and decreases of crop diversity within regions (Hazell
1989:32). Analysis ofworld cereal production provides clear evidence that (1)
yields vary significantly between years, (2) variability in yields has increased
significantly, and (3) the improved seed and fertilizer-intensive technologies
have been a major factor in this increase in variability.
Variability over space (cross-sectional data), sometimes referred to as
"adaptability," should be differentiated from variability over time (time/series
data), sometimes referred to as "stability" (Evenson et al., 1978:4). Evenson
et al. note that much of the existing literature on variability does not make a
clear distinction between the two concepts, and that they are presumed to be
highly correlated (1978:5). Limited evidence, however, suggests that variabil-
ity over time and space may not be highly correlated at all. Analysis of the
International Rice Research Institute's Yield Nursery trials for 22 rice varieties
at 22 locations for three years (with replications for each site) indicates no
evidence that adaptabilityand stability are correlated (Evenson et al., 1978:14).
Experience in the United States with yield data on wheat and response to levels
of nitrogen fertilizer indicates that, depending upon the method used for
estimation, variance was two to three times greater for cross-sectional data at
nine locations than in time-series data for five trials (Watson and Anderson,
1977:80-82).
Soil type and rainfall affect variability over space. There is evidence that
rainfall can vary significantly for places only a few kilometers apart and that
recorded rainfall patterns from a station may not provide a good estimate for
a region (Sharon, 1978; Patrinos et al., 1979; Nguyen et al., 1981; Gurovich
and Ramos, 1985). There can also be significant variability in soil character-
istics, especially soil hydraulic properties that result in differences in the
amount of water infiltrated into the soil. Since crop response to rainfall is a
function of the actual water depth stored in the soil profile after a rain event,
differences in hydraulic properties can increase the variability over space
resulting from variation in gross rainfall (Gurovich and Ramos, 1985:14).
Varietal responses to water stress further complicate an already complex
situation.
The implications ofvariability for individual farmers are significant. Village-
level data from southern India suggest fluctuations of net household income
and household food grain consumption of about 35 percent over five cropping
years (Walker and Jodha, 1986:34).


Vol. 4, No. 3, 1994






BEEBE


VARIABILITY AT THE MACRO LEVEL: RAINFED
AGRICULTURE IN MALAWI

A survey of farming in Malawi suggests the complexity of African rainfed
agriculture.4 The Malawi data indicate how few farms or farmers can be
considered "average," especially when two or more variables are used to define
the average. One of the most obvious differences among farmers is in the total
area farmed by the household. The "average" area farmed by the 1,026
households in this study is 1.7 hectares (ha), with a standard deviation (s.d.)
of 1.3. However, only 27 percent of households farm 1.5 -2 ha, and seven
percent farm less than 0.4 ha. From the perspective of the farmers with only
0.8 ha, the farmers with 1.6 ha have very large farms.
Households differ significantly in the amount of farm labor available from
within the family, the extent to which individuals within the household work
part time on other farms or in other jobs, and the extent to which labor is hired.
Approximately ten percent of the households have only one member over the
age of 14, about 50 percent have two, and about 40 percent have three.
Although not all household members over the age of 14 can be assumed to be
available for farm labor, the number of household members represents the
maximum amount of labor available within the household. Children aged 5-
14 provide significant amounts of the labor available to the farm. However,
very young children can be labor liabilities. Only two out of the sample of
1,026 households in this study have no children present under the age of 14.
The number of children ranged from one to thirteen, with about 25 percent
of the households having four or more children present. Labor availability
from within the family changes rapidly over time as children become old
enough to work, get sick, start school and end their formal studies. Although
most pregnant women continue to farm until the child is delivered, and often
resume farm activities relatively quickly, their level of activity can be reduced
significantly during pregnancy.
Off-farm work can both decrease family farm labor and increase resources
for farming, thereby increasing variability. The size of the farm and off-farm
work are linked. Although the "average" respondents were full-time farmers,
almost one-fourth of all respondents reported that they also engaged in off-
farm work. Part-time farmers who are also shop owners have farms that average
more than twice the size of full-time farmers. Part-time farmers who are
engaged in traditional farm labor on other farms have significantly smaller
farms than full-time farmers. Households with farms larger than average are
more likely to use traditional labor or hire labor commercially.
4 Malawi data were collected by Carol J. Culler, Helen Patterson, and Isabel Chikagwa Matenje.
Results are reported in Culler et al., 1990. Their survey focused on the role of women in
agricultural production, and the inclusion of gender as a variable makes this data set especially
relevant for understanding variability in Africa. The study was conducted in conjunction with
the Malawi Agricultural Research and Extension Project and was funded in part by the United
States Agency for International Development/Malawi. See also Swanson et al., 1986.


Journal for Farming Systems Research-Extension






CONCEPT OF THE AVERAGE FARMER


The demography of the Malawi farmers' households varies significantly.
Whereas the "average" farmer, or about half of all farmers, are between 30 and
50, about one-fourth are less than 30 and about one-fourth are more than 50.
A fifty-year-old farmer is likely to have more than double the farming
experience of a thirty-year-old. Older farmers tend to have larger farms.
Although the "average" farmer has had little or no schooling, one-third of the
farmers have had four years of schooling or more. Although the vast majority
of farmers (almost 90 percent) indicated they need credit, only about one-
fourth were able to get it. Whereas the "average" farmer is male, slightly more
than one-fourth of the farm households in this sample are headed by women.
The average farm for female-headed households is less than two-thirds the
average size of all farms, and about half the size of farms of male-headed
households. Whereas one-third of all farmers have had four years of schooling
or more, only one-fifth of females who headed households have had this
amount of schooling.
Given the degree of variability in background characteristics outlined
above, it is not surprising that there is tremendous variation in farm practices.
Farmers in this study report cultivating 25 different crops, with almost three-
fifths of farmers planting two or three different crops, less than one-fifth
planting only one crop, and slightly more than one fifth planting four or more
crops. Whereas the "average" farmer plants local maize, ten percent do not.
Whereas the "average" household reports having livestock, chickens are the
only livestock for more than half of the households with livestock. Only one
in four households has one or more goats, and only about ten percent of
households have pigs.
Household income from crop-related activities ranges from zero to more
than Malawi Kwacha (MK) 1000 (at the time of the study 1 MK = US $0.40).
Almost half of all households had incomes between MK 100 and 299, but
approximately ten percent had crop-related income of less than MK40 and five
percent had more than MK 1000. The variability in income, with differences
of hundreds of percent between individuals, is consistent with the variability
in background characteristics, quality of land, size of farm, on-farm activities,
labor, and off-farm employment.
The problem with the concept of the "average" farmer in Malawi is that
relatively few farmers are able to meet all criteria when more than one criterion
is used to define the average.
If sex and age are used, the "average" farmer in this study would be a male
between the ages of 30 and 49. Of this sample, however, only 17 percent meet
both of these criteria.
Programs are often designed to assist full-time farmers with average size
holdings. Although 60 percent of farmers have between 0.8 and 1.6 ha, only
41 percent of farmers with this amount of land are also full-time farmers. If a
program also requires access to credit, then the program would only be
relevant to less than ten percent of the farmers.


Vol. 4, No. 3, 1994






BEEBE


If a program assumes an average size holding and availability of an
"average" amount of additional labor from within the household, about half
the farmers would meet both the size and labor criteria would be about 50
percent. If the additional criterion of planting more than one crop is added,
only 33 percent can be considered "average".
The examples presented above illustrate how quickly the number of
"average" farmers or farms can be reduced when as few as three variables are
used at the same time. The use of additional variables further reduces the
number of cases that meet all of the criteria, until very few cases remain. These
examples also illustrate the problem with trying to assign a given farmer or farm
to a category, especially when the criteria being used are subject to change. An
individual's access to credit, for example, might change in response to actions
taken by the individual, and access to labor in the family will certainly change
with the health status of family members and the age and school status of
children.

VARIABILITY AT THE MICRO LEVEL: AN IRRIGATED RICE
PRODUCING VILLAGE IN THE PHILIPPINES

An intensive study of an irrigated, rice-producing village in the Philippines
suggests tremendous variability among neighbors. The village consists of 100
households, with 84 farmers farming 204 ha of contiguous irrigated rice land
divided into 159 parcels. All land is irrigated, and all farmers use the improved
seed and fertilizer intensive technology associated with the Green Revolution
in at least one of their parcels. From a distance, it appears that variability is
minimal within this micro-environment.
Closer analysis, however, reveals a number of important variations. One of
the most obvious differences is the area farmed by each farmer. Although the
average farm is 2.4 ha, total area ranges from 0.5 ha to 7.6 ha, with a standard
deviation of 1.4. The eight farmers with the smallest area have an average of
0.8 ha, whereas the eight farmers with the largest area have an average of 5.6
ha, seven times larger.
Almost as significant as differences in land area are differences in the quality
of the land. Farmers' indigenous soil classification system divides the soil into
four categories; They pointed out very subtle differences in the color of fields
on an aerial photograph taken during the dry season. Their system reflects, to
some extent, relative field height and proximity to what was once a stream.
Fields that are only a few centimeters higher can be very difficult to irrigate,

5 Philippine village-level research was conducted in 1976 in Barrio Gugo, Calumpit Bulacan in
Central Luzon and is reported in Beebe, 1978. The research was supported by a fellowship
granted by the Social Sciences Research Council and the American Council of Learned
Societies. Conclusions in this paper are based on a reanalysis of the data, and some minor
differences in results reflect the exclusion of data from one respondent and from two parcels
of land because of problems with missing data.


Journal for Farming Systems Research-Extension





CONCEPT OF THE AVERAGE FARMER


whereas fields that are relatively lower can have problems with waterlogging
and drainage. The indigenous soil classification system also reflects differences
in chemical properties. There are statistically significant differences for pH,
organic matter, and phosphorus between the different categories (at greater
than the 0.5 level, based on one-wayanalysis of variance for the 36 parcels with
available soil analysis data). Yield per ha for the two better categories is almost
15 percent higher than for the other two categories.
All farmers have irrigation, which often reduces yield variability (Boggess et
al., 1983). The quality of service varies significantly, however. Fields near the
ends of canals receive less water, and are more likely to receive it at times other
than when needed. Thirty percent of the farmers indicated that the preparation
of their main fields was delayed because of problems with irrigation, and 20
percent indicated delays were two weeks or longer. In addition to delays in
preparation because of problems with irrigation, one-fifth of the farmers
report water shortages of seven or more days during the growing season.
Production of transplanted rice is labor-intensive, and household labor is
never entirely sufficient. All households hire labor for transplanting and
harvesting. Only household labor, however, is used for some tasks-application
of inputs such as fertilizer and insecticides, and in almost all households for
weeding-and household labor can be a limiting factor for the utilization of
aspects of the new technology that are labor-intensive. There are major
differences in labor available in the household to work the land. Even though
there is an average of 6.9 individuals per household, there is only an average
of two household members available to help with farm work in addition to the
head of the household. Approximately one-fourth of households have no one
available for farm labor, whereas another one-fourth have only one individual
available. Women are almost half of the additional labor available in house-
holds with only one additional laborer. Especially for these households, being
pregnant or having very young children reduces the labor available. Sixty
percent of households have children present who are too young to go to
school.
Significant demographic variability has implications for the strategies
adopted by individual households. The "average" farmer (indeed, 83 of 84
cases) is male, but women play critical roles in decision-making and providing
labor. Women are "bankers" for the families, usually physically hold any cash
that is available, budget household and farm expenses, and hire labor. The
"average" farmer (55 percent) is between 30 and 49, with 15 percent younger
than 29, and 30 percent older than 50. About 40 of the farmers have ten years
or less of farming experience, and about 25 percent have 25 years or more
experience. Farmers have an average of five years of schooling, but 35 percent
have four years or less and only 20 percent have completed the sixth grade. In
the Philippines, it is often assumed that four years of schooling is necessary to
achieve functional literacy. Literacy in Tagalog of the farmers was evaluated
using an instrument that identifies the grade level at which an individual is able


Vol. 4, No. 3, 1994






BEEBE


to read. Only four of the 84 farmers are completely illiterate. Sixty percent of
the farmers are able to read and understand grade four or higher material.
Thirty percent of farmers who completed grade four are not able to read at the
grade four level and 25 percent of individuals who completed less than grade
four are able to read at this lever or higher. The comparison of actual reading
abilitywith years of schooling illustrates the problems with using a variable like
years of schooling as an indicator of functional literacy.
Access to credit is often critical to farmers who adopt new technology and
is a factor in the level of inputs used. The "average" farmer (77 of 84 cases)
indicated the need to borrow money during the growing season under
investigation. Only 58 percent were able to borrow money through a govern-
ment-sponsored credit program. The remaining farmers had to borrow from
landlords, money-lenders and other informal sources.
The tremendous variations in farmers' accomplishments reflect the differ-
ences in their resources and backgrounds. The average yield for the 159 parcels
was 4.3 metric tons (mt) of paddy per ha (with s.d. of 1.4), but yield ranged
from 6.8 mt per ha for the ten parcels with the highest yield to 1.1 mt per ha
for the ten parcels with the lowest yield. Since all farmers are using the new
seeds in at least one of their parcels, a willingness to experiment with the new
technology does not appear to be the issue here.
Mostly, farmers explain the differences in crop yield in terms of two
elements: (1) fate, defined by the farmers as things they do not control, and
(2) the farmer's actions in response to his or her fate. The individual is seen as
having control over some decisions but also facing important constraints,
some of which are viewed as a matter of "luck" and some of which are viewed
as the result of the actions of others.
The most important factors over which the farmer is thought to have little
or no control are the size and quality of his or her land holding. One of the 84
farmers in the village owns all the land he farms, and several others are part
owners of land. All the others are tenant farmers, working the land of absentee
landlords. Farmers have traditionally had almost no choice of land for farm-
ing.6
Soil types are critical for farmers' decisions about varieties and inputs,
including fertilizer. Varieties with longer growing periods tend to be planted
in areas with greater possibility of water problems, and less fertilizer is used in
these areas. The strategy appears to be to invest more in the areas with less risk,
and to limit exposure in the higher-risk areas. Given the widespread use of
high-yielding, fertilizer-sensitive varieties, it is not surprising that there is a
relatively strong correlation between use of nitrogen fertilizer and yield
(r=0.48, significant at the 0.05 level). Use of nitrogen fertilizer on all the fields
6 Tenant farmers will almost always accept any land, regardless of its size or quality, to which
they are given access. In recent years, a system has developed by which tenant farmers buy
"rights" to farm land from other tenant farmers, and then, make separate arrangements with
the landlord to farm the land as tenants. Final decisions rest with the landlord, and very few
farmers have been able to actually change the amount of land they farm.


Journal for Farming Systems Research-Extension






CONCEPT OF THE AVERAGE FARMER


averaged 60 kg nitrogen per ha (s.d.=30), with no fertilizer used in 8 parcels
(and less than 20 kg per ha of nitrogen used in another 10 parcels) and more
than 100 kg nitrogen per ha used in ten parcels. Nitrogen fertilizer application
was more than 25 percent greater in the two better soil categories than in the
other categories.
Total production ofpaddy ranged from 2.3 mt to 41.3 mt. (s.d.=6.29). The
eight farmers with the smallest production have an average production of 3 mt,
whereas the eight farmers with the largest production have an average
production of 24 mt. The average value of the rice crop produced was
Philippine Pesos 11,610 (s.d.=6,945) and ranged from Pesos 2,500 to Pesos
45,500 (at the time of the study, 1 Peso=US $0.13). Eleven farmers report that
they did not sell any of what was produced, and more than one-fourth of
farmers estimate that they will be forced to purchase rice before the next
harvest.
Relatively few fields or farms can be considered "average" in this village
when more than one criterion at a time is used to define the average. Although
50 percent of farmers have a total area between two and four hectares, and 60
percent of the farmers' largest fields are from the two better categories of the
indigenous soil classification system, only 26 percent of the fields meet both
criteria. If fields with problems with irrigation service are excluded, only 18
percent remain.
Maximum benefit from new technology in this village required a combina-
tion of fields from the two better categories of the indigenous soil classification
system, labor availability and the ability to borrow through government
programs. Although 60 percent of the farmers' largest fields are from the two
better categories of soil and 75 percent have one or more household members
available to provide labor, only 48 percent of the fields meet both of these
criteria. When the ability to borrow through government programs is added
as a criterion, only 18 percent-one in five fields-meet all three of these criteria.
Specific innovations associated with new technology are sometimes espe-
cially sensitive to demographic characteristics. For example, younger, better-
educated farmers are far more likely to use herbicides than older, less-educated
farmers. More than half of the farmers are under the age of 50 and more than
half are able to read at the fourth grade level or higher, but only 29 percent of
farmers fall into both categories. Only farmers with access to production credit
used herbicides, and when access to credit through government programs is
added as a criterion, fewer than one in five of the farmers meet all three criteria.
In addition, more fertilizer is likely to be used in fields from the two better
categories of soil and where irrigation is not a problem. Even when both of
these criteria have been met, as they were with 40 percent of farmers' largest
fields, a significant number of the fields receive only minimal fertilizer-24
percent of these fields received 50 kg of nitrogen per ha or less.
As in Malawi, the data from a village in the Philippines suggest that
differences, both of physical characteristics of fields and of demographic and


Vol. 4, No. 3, 1994






BEEBE


socioeconomic characteristics of farmers, are more important than similarities
for understanding what farmers do. Differences in many of the factors
associated with agricultural productivity in these two situations not measured
in a few percentage points change from the mean, but are often in terms of
hundreds of percentage points. Comparisons of the ten percent at either end
of the spectrum suggest that the observed differences are not the results of a
few abnormal cases or "outriders." For both the Malawi and the Philippine
data (1) some of the criteria used to define the "average" are subject to change
(for example labor availability, and irrigation services), and farmers or fields in
one category at a given time may belong to another category at a different time;
and (2) as more than one criterion is used to define an "average" group, the
number of farmers or fields that remain in the group can be reduced to very
small fractions of the entire population.

RECOMMENDATION DOMAINS

The extent of the variability described above has implications for the entire
agricultural development process, but would seem to have the greatest
implications for extension. An important element of a farming systems ap-
proach to research and extension has been the identification and use of
"recommendation domains." These were a response to variability in rural areas
and to the need to divide farmers into homogeneous groups for both research
and extension purposes (Weise, 1985). However, as variability increases, the
utility of recommendation domains declines.
An early definition and discussion of recommendation domains can be
found in a 1976 Centro Internacional de Mejoramiento de Maiz y Trigo
(CIMMYT) training manual:
It is impossible to conduct experiments on each farm to make recommenda-
tions tailored to each farm. Instead, you must define a target group of
farmers, conduct experiments under conditions representative of their farms,
and make recommendations which are applicable to the entire group. We
shall call such a group of farmers a recommendation domain. Generally, a
recommendation domain will consist of farmers within an agroclimatic zone
whose farms are similar and who use similar practices (Perrin et al., 1976:1).
The argument was that if farming systems research was to be cost-effective,
research activities had to address problems of, and provide solutions for,
relatively large numbers of people. It became necessary therefore to ". ..
classify farmers with similar circumstances into recommendation domains-
groups of farmers for whom we can make more or less the same recommenda-
tions" (Byerlee et al., 1980:71).
Some of the people at CIMMYT who originated the term were careful to
define a recommendation domain as a group offarmers, and not a geographical
area or land type. "Domains are composed of farmers because farmers, not land
types, take decisions on new elements of technology" (Harrington and Tripp,


Journal for Farming Systems Research-Extension






CONCEPT OF THE AVERAGE FARMER


1984:5). CIMMYT scientists recognized that by defining recommendation
domains in terms of groups of farmers, it is possible to consider socioeconomic
criteria. They also realized that some domains could not be mapped, that
neighboring farmers could belong to different domains and that a given farmer
could belong to more than one domain (Harrington and Tripp, 1984:5).
Many of the people who have adopted the term, however, have tended to
use it to describe physical areas. Thus most domains are defined in terms of soil
type, topography, and climate. Social, economic, and cultural factors such as
farm size, market access, and ethnic identity have been considered secondary
factors and generally have not been included (Cornick and Alberti, 1985:3).
Even the scientists at CIMMYT have tended to use the term to describe
physical areas, and to focus on the homogeneity of areas. "It may be that the
area is homogenous enough to constitute a single recommendation domain.
If not, there are usually one or at most a few key circumstances that can be used
to define domains" (Harrington and Tripp, 1984:13). For many, recommen-
dation domains are virtually synonymous with site selection and the identifi-
cation of target groups (Cornick and Alberti, 1985:3).
The concept of a homogenous group may be useful for research even if the
concept of a recommendation domain is counterproductive for extension. At
the beginning of a research process, the selection of target and research areas,
problem identification and the development of a research base are appropriate
and necessary (see Hildebrand, 1982:290). It is necessary to recognize groups,
but important not to forget the variability that exists. Labeling these areas as
"recommendation domains" is not necessary and may make the research
process less effective than itwould be otherwise. For research purposes, groups
of farms or farmers who share important characteristics could be called
"research domains" instead of "recommendation domains."
Cornick and Alberti conclude that recommendation domains are of little
utility in the first year or two of a farming systems project because early
establishment of recommendation domains forces assumptions of relatively
homogeneous farmer groups operating under similar conditions (1985:25).
They note that many practitioners of farming systems research give nominal
recognition to the variability found in small farm systems, but rarely deal with
it in a systematic way. They also note that early definition of recommendation
domains often exclude gender as a variable, and conclude that this is because
the time frame for establishing domains is too brief to permit an understanding
ofintra-household dynamics and the importance women may have in a system.
As a tool for extension, the concept of recommendation domains has been
counterproductive to the extent to which it has allowed practitioners to make
invalid assumptions about homogeneity within domains and to largely ignore
variability. This has been so especially when recommendation domains have
been based on agroclimatic characteristics and soils, and have ignored socio-
economic factors. The result has been attempts to disseminate technologies
that are relevant to only very small percentages of farmers. Keenly aware of


Vol. 4, No. 3, 1994






BEEBE


variability in their local situation, most farmers have, when given the oppor-
tunity, modified innovations. Experience from many locations worldwide
indicates that only rarely will farmers adopt a "technology package" in its
entirety and without modification. Many extension programs, however, con-
tinue to be evaluated on the extent to which farmers adopt the entire package.
This introduces tension between a system based on farmers following instruc-
tions from extension agents and farmers who are forced by variability in their
local situation to modify technology as they use it.
The focus of extension programs based on recommendation domains is to
get farmers to adopt technology, and farmers are reduced to a passive role of
accepting what is being offered by the system. The only truly active role in this
approach is that of the outsider who defines the domains and assigns farmers
to the already defined domains. The evidence on variability presented in this
paper suggests that this is an almost impossible task. A combination of criteria
that is subject to change and infinite possible combinations of criteria is bound
to result in many farmers being assigned to the wrong domain.

PUTTING THE FARMER FIRST

An alternative approach to extension begins with the assumption that farmers
know more about the unique conditions they face, ranging from physical land
types to socioeconomic conditions, than researchers and extension agents ever
can know. Although farmers often may not be able to articulate these issues,
they know the constraints and are willing to experiment with possible solutions
(Bunch, 1989:55-61). When presented with technology in the form of
options, farmers can be expected to assess options against local conditions.
Farmers need sufficient information to decide initially on whether to try an
innovation and may, in a few cases, need help in developing specific skills to
implement it. There is convincing evidence that successful farmers have
traditionally been willing to experiment (Lightfoot, 1987:80). There would
not be tens of thousands of traditional varieties of rice unless farmers were
willing to experiment with new varieties. A major thrust of a new approach to
extension should be to help farmers develop skills for experimentation (Ashby
et al., 1989; Bunch, 1989). Once a farmer in an area has experimented with an
innovation, other farmers are far more likely to learn from him or her than from
government extension workers. The problem is how to identify a cost-effective
way of presenting farmers with options, helping them develop their skills for
undertaking experiments, and getting a few farmers to actually experiment
with an innovation. In many places, mass media, especially radio, may be a
cost-effective way to make individuals aware of innovations and create interest
(Hornick, 1982; Perraton et al., 1983; Meyer, 1985; Ray, 1985). For some
innovations, radio may be sufficient for adoption (Rogers, 1983). There is a
need to consider a wide range of cost-effective activities that can increase


Journal for Farming Systems Research-Extension






CONCEPT OF THE AVERAGE FARMER


awareness and interest. A beginning point has to be increased attention to
traditional communication and knowledge systems.
Before some farmers will experiment with an innovation, they have to see
that it works with other farmers. Sometimes farmers have to develop new skills.
Short-term farmer training, maximizing learning by doing, can provide
farmers with direct exposure to an innovation and can provide specific skills
needed to implement a technology. An important aspect ofany training should
be strengthening skills in assessing local conditions and evaluating proposed
innovations relative to the local situation. Training could also strengthen
traditional systems for experimentation and for on-going evaluation of inno-
vations.
The role of the traditional government extension agent needs to change
radically. Extension agents should be more involved in research, especially
identification of problems and the implementation ofon-farm trials (Shaner et
al., 1982:158). They should be trained in the methodology of farming systems
and should become regular members of field teams. The best-qualified
extension agents could play key roles in information dissemination and farmer
training. Extension agents should be responsible not for trying to get farmers
to adopt specific innovations, but rather for providing farmers with informa-
tion on options with which the farmers could then choose to experiment.
Placing the farmer at the center of the process offers two significant
opportunities for dealing with variability. First, farmers are in a far better
situation than anyone else to assess their local situation and to determine
whether a given innovation is relevant. Second, farmers retain responsibility
for further experimentation with an innovation under local conditions, and for
adapting the innovation to their local conditions.
A farming systems approach to research assumes that farmers should play a
lead role in identifying problems and, through farmer-managed trials, play a
critical role in finding and validating solutions (Fernandez and Salvatierra,
1989). A farming systems approach to extension, based on providing options,
would place the farmers at the center of the extension system.

CONCLUSION
This paper argues that insufficient attention has been given to the implications
of variability for extension. Although the literature dealing explicitly with
variability is limited, there is evidence of significant variability over time and
space, and evidence that the new technology associated with the Green
Revolution has increased variability, especially in grain yields. Data from
studies of rainfed agriculture in Malawi and irrigated rice agriculture in the
Philippines illustrate the very wide ranges in the physical characteristics of the
land and labor resources available to farmers and in the farmers' demographic
and socioeconomic characteristics. This tremendous variability in production


Vol. 4, No. 3, 1994









and productivity, results in differences in the nutrition and income of house-
holds.
Recommendation domains have become an important element in a farming
systems approach to extension. Because recommendation domains largely
ignore variability within the group, their use may be counterproductive to
extension. Once it is recognized that outsiders cannot categorize farmers and
tell them what to do, an alternative conceptual model is needed. This paper
argues that only a farmer-centered approach, with farmers involved in identi-
fying problems, testing of solutions and disseminating options, can deal with
variability. Such an approach can lead to a radically different extension model
based on a combination of information dissemination and farmer training with
a limited role for a specialized extension service.

REFERENCES

Anderson, Jock and Peter Hazell, eds. 1989. Variability in Grain Yields: Implications
for Agricultural Research and Policy in Developing Countries. Baltimore: The Johns
Hopkins University Press for the International Food Policy Research Institute.
Ashby, Jacqueline A., Carlos A. Quiros, and Yolanda M. Rivers. 1989. Farmer
participation in technology development: Work with crop varieties. Pages 115-122
in Robert Chambers, Arnold Pacey, and Lori Ann Thrupp, eds., Farmer first:
Farmer innovation and agricultural research. London: Intermediate Technology
Development Group.
Barker, Randolph, Eric C. Gabler, and Donald Winkelmann. 1981. Long-term
consequences of technological change in crop yield stability. Pages 53-78. in
Alberto Valdes, ed., Food security for developing countries. Boulder, Colorado:
Westview Press.
Beebe, James. 1978. Resources, information and decision making skills: A study of
Filipino rice farmers in a Central Luzon Barrio. Ph.D. Dissertation, Stanford
University.
Boggess, W.G., Lynne, J.W. Lores, and D.P. Swaney. 1983. Risk-return assessment of
irrigation decisions in humid regions. Southern Journal of Agricultural Economics
15(1):135-143.
Bunch, Roland. 1989. Encouraging farmers' experiments. In Robert Chambers,
Arnold Pacey, and Lori Ann Thrupp, eds., Farmer first: Farmer innovation and
agricultural research. London: Intermediate Technology Development Group.
Byerlee, Derek, et al. 1980. Planning technologies appropriate to farmers-Concepts and
procedures. Mexico: Centro Intemacional de Mejoramiento de Maiz y Trigo.
Cornick, Tully RV., and Amalia M. Alberti. 1985. Recommendation domains recon-
sidered. Paper presented at the Fifth Annual Association for Farming Systems
Research-Extension Symposium, October, Kansas State University, Manhattan,
Kansas.
Culler, Carol J., Helen Patterson, and Isabel Chikagwa Matenje. 1990.' Survey of
women in agriculture in Malawi. Malawi: Women's Programme Section, Ministry
of Agriculture in conjunction with the Malawi Agricultural Research and Extension
Project.


Journal for Farming Systems Research-Extension







CONCEPT OF THE AVERAGE FARMER


Evenson, R.E., J.C. O'Tool, R.W. Herdt, W.R. Coffman, and H.E. Kauffman. 1978.
Risk and uncertainty as factors in crop improvement research. Research Paper Series
no. 15. Philippines: International Rice Research Institute.
Fernandez, Maria E., and Hugo Salvatierra. 1989. Participatory technology validation
in highland communities in Peru. Pages 146-150 in Robert Chambers, Arnold
Pacey, and Lori Ann Thrupp, eds., Farmer first: Farmer innovation and agricultur-
al research. London: Intermediate Technology Development Groups.
Gurovich, Luis A., and Roberto Ramos. 1985. Spatial variability of rainfall and yield
on maize in a semiarid region. Agricultural Water Management 10:13-29.
Hargrove, T.R., W.R. Coffman ,and V.L. Cabanilla. 1979. Genetic interrelationships
of improved rice varieties in Asia. Research Paper Series no. 23. Philippines:
International Rice Research Institute.
Harrington, L.W., and R. Tripp. 1984. Recommendation domains:A framework for on-
farm research. CIMMYT Economics Program Working Paper 02/84. Mexico:
CIMMYT.
Hazell, Peter B.R. 1989. Changing patterns of variability in world cereal production.
In Jock Anderson and Peter Hazell, eds., Variability in grain yields: Implicationsfor
agricultural research and policy in developing countries. 13-34. Baltimore: The
Johns Hopkins University Press for the International Food Policy Research Insti-
tute.
Hildebrand, Peter E. 1982. Summary of the sondeo methodology used by ICTA. In
W.W. Shaner, P. F. Philipp, and W.R. Schmehl, eds., Farming systems research and
development: Guidelines for developing countries. Boulder: Westview Press.
Lightfoot, Clive.1987. Indigenous research and on-farm trials. AgriculturalAdmin-
istration and Extension 24:79-89.
Mehra, Shakuntla.1981. Instability in Indian agriculture in the context of the new
technology. Research Report 25. Washington DC: International Food Policy Re-
search Institute.
Nguyen, Hung.1989. Agricultural planning policy and variability in Syrian cereal
production. Pages 78-90 in Jock Anderson and Peter Hazell, eds., Variability in
grain yields: Implicationsfor agricultural research and policy in developing countries.
Baltimore, Maryland: The Johns Hopkins University Press for the International
Food Policy Research Institute.
Nguyen, V.T., J. Rouselle, and M.B. Mac Pherson.1981. Evaluation of areal versus
point rainfall sparse data. Canadian Journal of Civil Engineering 8:173-178.
Patrinos, A., N. Chen, and R. Miller.1979. Spatial correlation of monthly rainfall:
Applications in climatology and weather modification experiments. Journal of
Applied Meteorology 18:719-732.
Perraton, H. 1983. Basic education and agricultural extension. Washington, DC: The
World Bank.
Perrin, R.K., D.L. Winkelmann, E.R. Moscardi, and J.R. Anderson.1976. From
agronomic data to farmer recommendations: An economics training manual. Mex-
ico: CIMMYT.
Ray, H. 1985. Incorporating communication strategies into agricultural development
programs. Washington, DC: Academy for Educational Development.
Rogers, E.M. 1983. Diffusion of innovations. New York: Free Press.


Vol. 4, No. 3, 1994






BEEBE


Shaner, W.W., P.F. Philipp, and W.R. Schmehl.1982. Farming systems research and
development: Guidelines for developing countries. Boulder: Westview Press.
Sharon, D. 1978. Rainfall fields in Israel and Jordan and the effects of cloud seeding
on them. Journal ofApplied Meteorology 17:40-48.
Swanson, Burton E., Andrew Sofranko, Kathleen Cloud, John B. Claar, Earl D.
Kellogg, and Clarence J. Kaiser. 1986. Agricultural technology system in Malawi.
Urbana: International Program for Agricultural Knowledge Systems (INTER-
PAKS) Office of International Agriculture, University of Illinois at Urbana-
Champaign.
Tarrant, John R. 1989. An analysis of variability in Soviet grain production. Pages 60-
77 in Jock Anderson and Peter Hazell, eds., Variability ingrain yields: Implications
for agricultural research and policy in developing countries. Baltimore: The Johns
Hopkins University Press for the International Food Policy Research Institute.
Walker, Thomas S., and N.S. Jodha. 1986. How small farm households adapt to risk.
Pages 17-34 in Peter Hazell, Carlos Pomareda, and Alberto Valdes, eds., Crop
insurance for agricultural development: Issues and experience. Baltimore: The Johns
Hopkins University Press for the International Food Policy Research Institute.
Watson, W.D. and J.R. Anderson.1977. Spatial v. time-series data for assessing
Response risk. Review of Marketing and Agricultural Economics 45(3):80-84.
Weise, Karen. 1985. Farming systems research: Issuesfor project evaluations. Washing-
ton, DC: United States Agency for International Development, PPC/CDIE.


Journal for Farming Systems Research-Extension








An Evaluation of Soybean Planting Methods
for Small-Scale Farmers in Central Province,
Zambia: A Farming Systems Approach

M. Bezuneh, F.J. Olsen, P.T. Gibson, and K.S. Chanda '


ABSTRACT
Soybean [Glycine max (L.) Merr.] is a relatively new crop in Zambian farming
systems, particularly for traditional (subsistence) and small-scale farmers. In
recent years, improved institutional infrastructure has given farmers an added
incentive to produce soybeans as a cash crop. These improvements have
included the availability of new cultivars, greater opportunity to obtain credit,
easier marketing, and an attractive guaranteed price. However, soybean yields
have not improved, due partly to the lack of an appropriate planting method
to produce optimal population densities. The present method of planting
soybeans in the Kabwe Rural District of Central Province is by plowing and
planting in the same operation, dribbling the seed behind the ox-plow. This
method leads to uneven planting depths, resulting in poor seedling emergence
and stands.
The Adaptive Research Planning Team for Central Province (ARPT-CP)
evaluated various soybean planting techniques on farmers' fields to determine
which one would provide the most uniform soybean stand, highest yield, and
greatest economic returns to labor and capital invested. The agronomic and
economic results from three years of on-farm research suggest that farmers in
Kabwe Rural should replace the common practice of planting soybean by
dribbling seed behind the plow with either hand seeding following a plow-
harrow operation or using a modified ox-drawn maize planter ("Taparia").

INTRODUCTION
Soybean [ Glycine max (L.) Merr.] is a relatively new crop to the traditional and
small-scale farmers in Zambia (Zambia Ministry of Agriculture and Water
Development, 1985). Recently, financial credit for small-scale soybean pro-
duction units has been provided by the Lint Company (LINTCO). Addition-
ally, two unique soybean cultivars have been identified, Hernon 147 and
1 Mesfm Bezuneh is Associate Professor and Chair, Dept. of Economics, Clark Atlanta
University, Atlanta, GA 30314, USA; Farrel J. Olsen and Paul T. Gibson are Professor and
Assistant Professor, respectively, Department of Plant and Soil Science, Southern Illinois
University, Carbondale, IL 62901 USA; and Kefi S. Chandais an Agronomist for the Adaptive
Research Planning Team, Kabwe Regional Research Station, Kabwe, Zambia.





BEZUNEH ET AL.


Magoye, that form nodules in association with indigenous rhizobia (Javaheri,
1985). These cultivars are well suited for traditional and small-scale farmers,
who do not have easy access to rhizobium innoculum. Recently, the price of
soybeans has been very attractive. As a result of these developments, farmers
have shown considerable interest in producing soybeans as a cash crop.
Labor shortage during the peak labor period of late October-January, when
all cash and staple crops are planted and weeded, was identified as the major
constraint in the Kabwe Rural District in Central Province, Zambia (Zambia
Ministry of Agriculture and Water Development, 1983). This constraint
contributes to rough seedbed preparation, resulting in poor tilth with large
clods often leading to poor crop emergence and erratic stands. Dribbling seed
behind an ox-drawn plow is the most common method of planting the major
crops in this system. This practice is favored because it is fast and releases labor
for other farming operations. However, with this planting technique, seeding
depth varies widely and often is too deep, resulting in poor seedling emer-
gence. Fehr et al. (1973) reported an average soybean emergence of73 percent
from planting at a five cm depth and 44 percent from a 10 cm depth. Similar
results were observed earlier by Khan and Ashley (1965) on pigeon peas
[Cajanus cajan (L.) Milliop].
An alternative method of soybean planting was recommended by the
Zambian Ministry of Agriculture and Water Development for small-scale
farmers (called the LIMA recommendation, from "kulima"-"to cultivate" in
Icibemba). This method involves preparing the land by hand-hoe, opening
furrows 50 cm apart by hand, distributing the seed along the furrow, and
covering by hand. The LIMA method is the most common way of planting
soybeans among small-scale farmers, with the variation that initial land
preparation is sometimes by oxen.
Weaver (1984) noted that zero-tillage was less costly for growing soybean
than conventional/tillage or furrow subsoiling in deep sand soils in Florida,
USA. Maize (Zea mays L.) grown under a zero-tillage system in Traditional
Recommendation Domain #3 (Mkushi District), Central Province, allowed
farmers to plant their late-maturing hybrid cultivars with the first rains,
resulting in higher yields than conventionally planted maize hybrids (Zambia
Ministry of Agriculture and Water Development, 1985). Javaheri and Nkum-
bula (1984) compared the performance of the soybean cultivars Hernon 147
and Magoye using two planting methods (drilling and hill planting) and found
no significant yield differences. Broadcasting, hill-planting, and dribbling rice
(Oryza sativa indica) resulted in no significant differences in grain yields
(Waterworth, 1985). However, it took more time to hill plant than to
broadcast or dribble seed.
The objective of this research was to evaluate the plant stands at harvest,
yield, and economic benefits of different soybean planting methods for the
small-scale farmers in Central Province, Zambia (Figure 1). For this purpose,
both agronomic and economic analyses were carried out.


Journal for Farming Systems Research-Extension






SOYBEAN PLANTING METHODS IN ZAMBIA


Figure 1. Location of Central Province, Zambia.


MATERIALS AND METHODS
This is a detailed account of an agronomic and economic study initiated by
Chanda et al. (1990). It was conducted during the rainy season (November-
April) for a three-year period (1985-88) in Kabwe Rural District, Central
Province, Zambia. On-farm research was carried out using farming systems
methodology (Perrin et al., 1976). There are two broad soil groupings in the
area under consideration. In the northern part of the district, the soils are
primarily Mushemi series (Kandilestalf, Clayey, Kaolinitic, Isokyperthermic).
These soils are deep, well drained, and highly leached, with a sandy loam
topsoil texture (approximately 20 to 30 percent clay). The soils are strong
brown in color and have a low cation exchange capacity. In the southern part
of the district, the dominant soils are of Mutwale series (oxic Palesustult,
Clayey, Kaolinitic Isokyperthermic). These soils are also deep, well drained,
and redder than the Mushemi series. They are highly leached,with clay texture
(approximately 45 percent clay in the topsoil). They have a low cation
exchange capacity, and both soils are highly acidic (pH 5.0).
Treatments were planted by hand or ox-drawn equipment in plots 5 m x 20
m in a randomized complete block design with four replications. The row
width was 50 cm. Two sites were used in 1985/86, with four sites in each of
the two subsequent years. The number ofsites was limited due to the large size
of the plots, number of replications per site, and the logistics involved. The
experimental sites were located on fallow or virgin ground, and included both
broad soil groupings. Planting dates ranged from mid-December to early
January. Weeding was done by hand as needed.


Vol. 4, No. 3, 1994





BEZUNEH ET AL.


Data were analyzed for each trial site separately and for combined locations.
The analysis of variance and Duncan's Multiple Range Test were used to
identify and separate significant differences among treatments. If the treat-
ment by location interaction was significant (P = 0.05), it was used as the error
term to test treatments in place of the pooled within site error.
The soybean cultivar Hernon 147 was used at a seeding rate of 500,000 seed
ha-1. This cultivar is unique in that it forms nodules in association with the
indigenous rhizobia found in Zambia soils, and thus is well suited for small-
scale and traditional farmers. Soybean was hand harvested, threshed, and
tested for moisture. Yields were adjusted to 12.5 percent moisture. Stand
counts were obtained at harvest.
Since the farming system in the Kabwe Rural District is characterized by
widespread ownership of cattle for draft purposes and access to planter (as
owner or renter), emphasis for assessing soybean planting techniques was
focused on methods involving animal power and planters. The following five
soybean planting techniques were evaluated during the 1985/86 cropping
season:
Dribbling seed behind the plow. Seeds were placed in every second furrow as
the oxen plowed and covered with soil thrown by the plow as it opened the next
furrow.
"Taparia" planter. This ox-drawn equipment was originally designed for
planting maize. Ox-plowing and harrowing were followed by planting in the
prepared seedbed in a separate operation using the seed box of the Taparia
planter.
Plow-harrow. The seedbed was prepared using oxen; seed was then dribbled
into a furrow made by an ox-drawn implement and covered using a light ox-
drawn harrow.
LIMA method. The seedbed was prepared by hand-hoe; seed was placed in
furrows, opened by hand, and then covered.
"Sebele plow planter." Originally designed for maize, this planter was used
for seeding behind the plow.
Neither the Taparia nor the Sebele maize planters worked satisfactorily for
planting soybeans. The Taparia planter was modified for the 1986/87 season.
This planter has a seed box at the front and a fertilizer hopper in the back. A
small piece of metal (1.5 cm2) was removed from the center of the bottom of
the fertilizer lever that adjusts the fertilizer flow. Soybean seed was placed in
the fertilizer hopper instead of the seed box and flowed freely through the
orifice made in the lever. The seeding operation was calibrated to obtain the
desired seeding rate. The Sebele planter was dropped from the study due to the
lack of availability in Zambia and because it could not be modified easily to
improve its performance. Some small-scale farmers started planting maize
using no-till methods. Thus in 1986/87 zero-tillage was added to the study.
In addition, hill planting was also evaluated.


Journal for Farming Systems Research-Extension






SOYBEAN PLANTING METHODS IN ZAMBIA


Zero-tillage. Hills were opened in non-tilled land using hand hoes, and six
seeds were dropped in hills 25 cm apart. Chemical weed control involved
paraquat (1-1 dimthyl 1-4-4 hipyridylionion) at 1.0 kg ai ha-' and metlachlor
(achlorob-ethyl N) (2-methoxy-l-methlethyl-0-acetotoluidide) at 1.8 kg ai
ha1 mixed together and sprayed by hand using a knapsack sprayer.
Hill planting. The land was plowed and harrowed by oxen. Six seeds were
dropped in hills 25 cm apart.
Although the LIMA method resulted in good plant stands, it was very labor
intensive. Hill planting and zero-tillage were also quite labor intensive.
Additionally, zero-tillage required considerable capital input. Hence, during
the 1987/88 cropping season, only the following three soybean planting
methods were evaluated: (1) dribbling seed behind the plow; (2) Taparia
planter as modified in 1986/87; and (3) plow-harrow where small furrows
were made using an ox-drawn plow.

Economic Analysis
One of the objectives of the soybean on-farm research was to determine
which method ofplanting generated the highest marginal rate of return to cash
and labor invested. The basic concerns of all traditional (subsistence) and
small-scale farmers in adopting any new technology are often expressed in
terms of additional benefits and the stability of the benefits. More specifically,
farmers' questions are: (1) Does the new alternative (method) offer enough
additional benefit to warrant additional investment? and (2) How risky (or
stable) is the new alternative or method in relation to the existing practice? In
this study, marginal and risk analyses were carried out for the 1986/87 and
1987/88 cropping seasons in order to compare and assess the different
soybean planting methods and deliver recommendations for farmers' adop-
tion.

Results and Discussion
The rainfall distribution during the growing seasons (1985 through 1988)
is presented in Table 1. In the 1985/86 growing season, the area experienced
four percent more precipitation than normal, resulting in satisfactory yields in
some treatments (Table 2). However, there was 38 percent less precipitation
than normal in the 1986/87 growing season, and 30 percent less than normal
in 1987/88. In 1986/87, February was extremely dry, resulting in very low
yields (Table 3), whereas in 1987/88, the below-normal rainfall was favorably
distributed, resulting in good soybean yields (Table 4).

Agronomic Results
Although dribbling soybean seed behind the plow offers a considerable
savings in labor compared to other planting methods, plant stands at harvest
in 1985/86 were very poor, probably due to the uneven planting depths


Vol. 4, No. 3, 1994






BEZUNEH ET AL.


Table 1. Rainfall distribution for the 1985/86-1987/88 growing seasons in Ka-
bwe Rural District, Central Province, Zambia.
RAINFALL (MM)
MONTH 1985/86 1986/87 1987/88
October 29.2 (+8.2)* 25.3 (4.3) 1.3 (-19.7)
November 53.2 (-42.8) 52.6 (-43.4) 20.4 (-75.6)
December 316.7 (+51.7) 153.0 (-112.0) 278.8 (+13.8)
January 356.5(+118.5) 286.7 (+48.7) 179.2 (-58.8)
February 126.5 (-72.5) 34.9 (-164.1) 134.3 (-64.7)
March 114.4 (-0.6) 47.9 (-67.1) 63.2 (-51.8)
TOTAL 996.5 (+34.5) 600.4 (-361.6) 667.2 (-294.8)
* Numbers in parentheses indicate the deviation from the long-term average in a particular
month.
(Table 2). The LIMA method of soybean planting resulted in good stands but
was very labor intensive. The Sebele planter and the non-modified ox-drawn
Taparia planter had much lower plant densities than the LIMA and plow-
harrow methods.
Table 2 shows that soybean yields in 1985/86 appeared to be related to
plant population at harvest (r2 = 0.80, P = 0.11). Dribbling soybean seed
behind the plow gave significantly lower grain yields than any other planting
method except the Taparia planter. The LIMA and the plow-harrow method
had the best stands at harvest and also the highest soybean yields.
Dribbling soybean seed behind the plow resulted in the lowest plant
population again in 1986/87 (Table 3). For plant density, there were no
significant differences among the other treatments, but the modified ox-drawn
Taparia planter ranked first, followed by the LIMA method. Soybean seed
yields on all treatments were low due to a lack of adequate precipitation during
the growing season. Dribbling seed behind the plow resulted in the lowest
yield.
Three soybean planting techniques involving the use of animal power were
evaluated during the 1987/88 growing season. The plow-harrow method of
planting soybeans resulted in the best soybean plant stand at harvest as well as
the highest yield (Table 4). Dribbling seed behind the plow ranked lowest for

Table 2. Summary of treatment effects: Soybean planting methods, Kabwe Rural
District, Central Province, Zambia (1985/86).
PLANTING POPULATION
PLANTING METHOD AT HARVEST (1,000 PLANTS HA') YIELD (KG HA')
Dribbling seed behind the plow 133 b* 569 d
Ox-drawn planter (Taparia) 179 b 788 cd
Plow-harrow 337 a 1407 ab
Lima 303 a 1758 a
Sebele planter 189 b 1116 bc
* Numbers followed by a common letter within columns are not significantly different at the
0.05 level of probability according to Duncan's Multiple Range Test.


Journal for Farming Systems Research-Extension






SOYBEAN PLANTING METHODS IN ZAMBIA


Table 3. Summary of treatment effects: Soybean planting methods, Kabwe Rural
District, Central Province, Zambia (1986/87).
PLANTING POPULATION
PLANTING METHOD AT HARVEST (1,000 PLANTS HAK) YIELD (KG HA')
Dribbling seed behind the plow 110 c* 396 b
Modified ox-drawn planter (Taparia) 305 a 861 a
Plow-harrow 2
68 a 793 a
Lima 273 a 859 a
O-Tillage 234 a 828 a
Hill planting 230 a 851 a
Numbers followed by a common letter within columns are not significantly different at the
0.05 level of probability according to Duncan's Multiple Range Test.

yield, although treatment differences were not significant. Dribbling soybean
seed behind the plow did not result in consistently good soybean stands,
especially on fallow or virgin land. The modified ox-drawn Taparia planter
performed well.

Economic Results
The various soybean planting methods differed greatly in the amount of
labor required to plant and weed (Table 5). The most labor-intensive planting
method was the LIMA (108 man-days ha-). Dribbling seed behind the plow
is the favored planting method for small-scale farmers in Zambia because it is
quicker (three person-days ha-'). However, this planting method results in
poor seedbeds requiring greater amounts of labor for weeding.
The economic analysis presented in Table 6 indicates that in 1986/87, the
modified ox-drawn planter (Taparia) generated the highest mean net return
(ZK 733 ha'-), minimum net benefit (ZK 458), and "average lowest 2" (ZK
531) with the least variations in net benefit (Table 6).2 Although this method
did not meet the desired rate of return (40 percent), it was the only undom-
inated treatment.3 The plow-harrow technique was the next-best treatment.
The return to labor from the Taparia method was 82 percent higher than the
plow-harrow method (i.e., ZK 30.50 compared to ZK 5.30 per person-day).
The main difference in the net return to labor between these two methods was
the large labor use of the plow-harrow method for hand planting in 1986/87.
All other methods, including the farmers' own dribbling seed behind the plow,
showed negative minimum net benefit.
Results of the marginal and risk analysis for 1987/88 are presented in Table
7. Again, the modified ox-drawn planter (Taparia) planting method gave the
highest net benefit (14 percent more than dribbling seed behind the plow). It

2 Economic analysis was not carried out for the 1985/86 season because relevant data were not
collected.
3 An undominated treatment is a treatment for which there is no alternative treatment with
respect to both lower variable cost and higher net benefit.


Vol. 4, No. 3, 1994






BEZUNEH ET AL.


Table 4. Summary of treatment effects: Soybean planting methods, Kabwe Rural
District, Central Province, Zambia (1987/88).
PLANTING POPULATION
PLANTING METHOD AT HARVEST (1,000 PLANTS HA'1) YIELD (KG HAl )
Dribbling seed behind the plow 192 c* 2149 a
Modified ox-drawn planter (Taparia) 234 b 2302 a
Plow-harrow 290 a 2504 a
* Numbers followed by a common letter within columns are not significantly different at the
0.05 level of probability according to Duncan's Multiple Range Test.

also generated the highest minimum net benefit and almost equal variability
in net return compared to that of the farmer's own practice and the plow-
harrow planting method (Table 7A). The marginal analysis further revealed
that the modified ox-drawn planter (Taparia) was the most dominant planting
method. Its rate of return for the additional investment is 10.55 percent (i.e.,
ZK 10.55 return for every additional Kwacha invested). This is over and above
the next best alternative ofplanting soybeans, the plow-harrow method. Using
a target rate of return of 40 percent, both the modified ox-drawn planter and
the plow-harrow methods have the potential of being selected over the
farmer's own practice.4 However, both the risk and the marginal analysis
suggest that the modified ox-drawn planter (Taparia) is superior (Table 7B).

CONCLUSION

Although dribbling seed behind the plow is used widely for maize planting in
Zambia, it does not consistently result in good soybean stands, especially on
fallow or virgin land. In trials conducted over three years with different rainfall
patterns in the two major soil types of the district on virgin or fallow ground,
both the plow-harrow method and the modified ox-drawn Taparia planter
out-performed the traditional method of dribbling seed behind the plow. In
all three years, the plow-harrow method was significantly superior in stand
establishment to dribbling seed behind the plow, and gave significantly greater
yield in two of these years. Prior to modification, the Taparia ox-drawn planter
was no better in stand or yield than dribbling seed behind the plow, but was
superior in stand establishment in the two years after modification, with
greater yield than the dribbling method in one year. In 1986/87, when
drought severely reduced yields, these two alternative methods were superior
in stand and yield to the traditional method, giving almost three times the
plant population and roughly double the yields. Both the plow-harrow and the
modified Taparia planter resulted in optimal stands with a minimum number
of total person-days required through the season. All methods using oxen for
planting were faster than hand-planting methods and less costly than zero-till.
4 It is often assumed that the traditional and small-scale farmers generally do not commit any
of their resources unless there is at least a 40 percent rate of return.


Journal for Farming Systems Research-Extension





Table 5. Average labor demand (person-days ha') for six soybean planting methods in Kabwe Rural District,
Central Province, Zambia (1986/87).


OPERATION


'0
Land preparation,
including dribbling


DRIBBLING
SEED BEHIND
THE PLOW


OX-DRAWN
PLANTER
(TAPARIA)


seed behind plow 3 (Ox)+ 3 (Ox)

Harrowing 1 (Ox)

Opening hills or furrows

Planting -1 (Ox)

Covering seed

Harrowing to cover seed

Herbicide application

Weeding 23 (H) 9 (H)

Harvesting 4 (H) 10 (H)

Threshing and cleaning
(one person-day
/90 kg bag)

TOTAL 30 24
(Ox) Indicates an operation done by oxen; (H) Indicates an operation done by hand.


PLOW
HARROW


HILL
O-TILLAGE


3 (Ox)

1 (Ox)

25 (H)

21 (H)


1 (Ox)


7 (H)

9 (H)




67


28 (H)


25 (H)

21 (H)

17 (H)


7 (H)

10 (H)




108


25 (H)

21 (H)

17 (H)


4 (H)


9(H)




76


0

PLANTING Z


3 (Ox)

1 (Ox)

13 (H)

21 (H)

17 (H)


5(H)

10 (H)




70


--








Table 6. Summary of net benefit and risk analysis: Soybean planting methods in Kabwe Rural District, Central Province, Zambia
(1986/87), Number of Sites = 4.*
RISK TABLE
TREATMENT MEAN STANDARD INDEX OF MINIMUM AVERAGE
NET BENEFIT DEVIATION VARIABILITY NET BENEFIT LOWEST 2
(ZK/HA- ) (PERCENT) (ZK/HA-I)

Dribbling seed behind the plow 100.24 259.35 258.73 -144.56 -92.72

Modified ox-drawn planter (Taparia) 733.04 372.78 50.85 457.60 530.72
Plow-harrow 357.56 383.80 107.34 75.68 106.64

Lima 140.32 502.31 357.98 -222.92 -196.28

O-Tillage -338.52 367.62 -108.60 -650.64 -588.72
Hill planting 164.08 294.87 179.71 -89.00 -38.60
* Marginal analysis was not carried out since none of the treatments met the desired target rate of return (40 percent).






- Table 7. Summary of risk and marginal analysis: Soybean planting methods in Kabwe Rural District, Central Province, Zambia (1987/
4 88), Number of Sites = 4).*


TREATMENT
tS


MEAN
NET BENEF


A. RISK TABLE
STANDARD
IT DEVIATION
(ZK/HA-1)


INDEX OF
VARIABILITY
(PERCENT)


MINIMUM AVERAGE
NET BENEFIT LOWEST 2
(ZK/HA-')


Dribbling seed behind plow 3971.33 1420.13 35.76 2495.87 2794.11

Plow-harrow 4204.68 1665.97 39.62 2674.45 2764.47

Modified ox-drawn planter (Taparia) 4603.41 1792.04 38.93 2749.46 3298.28
B. MARGINAL RETURN TABLE (NON-DOMINATED TREATMENTS ONLY) ZK/HA"
RANK TREATMENT NET BENEFIT TOTAL RATE MARGINAL NET MARGINAL COST MARGINAL RATE
VARIABLE COSTS BENEFIT BENEFIT OF RETURN

1 Modified Ox-drawn
planter (Taparia) 4603.41 827.22 398.73 37.78 1055.4*

2 Plow-harrow 4204.68 789.44 233.35 99.05 235.6*
3 Dribbling seed behind plow 3971.33 690.39 -
* These treatments meet or exceed target rate of return (40 percent).






BEZUNEH ET AL.


The total person-day requirements were less for the plow-harrow and the
modified Taparia planter than dribbling seed behind the plow because the
improved seedbed preparation and stand establishment helped suppress weeds
so that fewer days were required for weeding. In practice, many small-scale
farmers weed their fields inadequately due to the labor shortage. If the farmer
did little or no weeding, the plow-harrow and the modified ox-drawn planter
methods would probably incur less yield loss from weed competition than
would the traditional method.
Economic analysis also showed that these two methods had the highest
mean net benefits, lowest index of variability, and highest return to labor. The
plow-harrow method does not require the purchase of any additional equip-
ment, whereas the Taparia planter requires a modest capital investment. Both
agronomic and economic analyses indicate that farmers with oxen should
adopt either the plow-harrow method of planting soybeans or use the modified
Taparia planter. The decision of which of these methods to use will depend
upon the individual farmer's preference.

ACKNOWLEDGMENT

The authors wish to acknowledge the assistance ofMoffat Shamambo, Ernest
Shingalili, and George Swallow of ARPT-CP in carrying out the on-farm
research.
The on-farm research component of this study was supported jointly by
USAID and Government of Zambia under the Zambia Agricultural Research
and Extension Project (Contract Number 611-02).

REFERENCES

Chanda, KS., M. Bezuneh, P.T. Gibson, F.J. Olsen, and RE. Hudgens. 1990. An
agronomic and economic evaluation of soybean planting methods in the Central
Province of Zambia, Experimental Agriculture 26(4): 441-445.
Fehr, W.R, J.S. Burris, and D.F. Gilman. 1973. Soybean emergence under field condi-
tions. Agronomy Journal 65:740-742.
Javaheri, F. 1985. Soybean varieties for small-scale farmers in Zambia. Soybean Genetics
Newsletter, Zambia.
Javaheri, F., and S. Nkumbula. 1984. Hill planting: An alternative approach to soybean
planting by small-scale farmers in Zambia. Farming in Zambia December.
.Khan, T.N., and J.M. Ashley. 1965. Factors affecting plant stand in pigeon pea.
Experimental Agriculture 11:315-322.
Perrin, Richard K, D.L. Winkelman, E.R. Moscardi, and J.R. Anderson. 1976. Agronomic
data to farmer recommendations: An economics training manual. Bulletin 27. Mexico:
Centro Internacional de Mejoramiento de Maize y Trigo (CIMMYT).
Waterworth, J. 1985. Rice planting method study. Adaptive Research Planning Team,
Eastern Province, Zambia. Annual Report.


Journal for Farming Systems Research-Extension






SOYBEAN PLANTING METHODS IN ZAMBIA


Weaver, M. 1984. Economic variables of conventional vs. in-row subsoil and drilled no-
tillage soybeans in Gilchrist County. Proceedings ofthe seventh annual no-tillage systems
conference. Pages 133-136.
Zambia Ministry ofAgriculture and Water Development. 1983. Formal survey. Tradition-
al Recommendation Domain #3. Central Province.
Zambia Ministry of Agriculture and Water Development. 1985. Annual Report. Adaptive
research planning team. Central Province, Zambia.


Vol. 4, No. 3, 1994








Systems of Production and Production of
Knowledge:
Reflections on the Basis ofIvorian and
Mexican Experiences

Jean-Philippe Colin 1


ABSTRACT
This paper discusses the links-so important in farming systems analysis-
between the type of data collection techniques used and the type of knowledge
produced, in terms of causal explanation. The major point argued in the paper
is that if one aims to understand and explain farmers' socioeconomic and
technical practices, rapid information collection with low personal involve-
ment of the researcher may not be the best suited methodology. This point is
illustrated through two research experiences: an in-depth study of a village
economy in Lower Ivory Coast, and a study based on a classic questionnaire
survey in various Mexican villages.

INTRODUCTION
The concept of production system, whatever its meaning, is an intellectual
construction intended to facilitate the perception and the interpretation of the
real world. This perception and interpretation requires the "operationaliza-
tion" of the concept, the implementation of a whole set of information
collection techniques, to provide an empirical content. A fundamental point,
then, is to recognize that strong bonds link the type of data collection
technique used and the kind of knowledge produced.
This article provides an illustration of this point through considering the
methodology of two research experiences: the first in the Ivory Coast between
1983 and 1985 (Colin, 1990), the second in Mexico in 1990-91 (Colin,
1992). These research programs were both based on the economic analysis of
agricultural production system (APS); however, they used very different data
collection techniques. The first was an in-depth study of a village economy in
Lower Ivory Coast; the second was a study based on a classic questionnaire
survey in various Mexican villages.
1 Agricultural economist with the ORSTOM (French Scientific Research Institute for Develop-
ment through Cooperation), Visiting Professor with the Colegio de Postgraduados Sistemas
de Producci6n y Desarrollo Agricola (Texcoco, Mexico). I would like to thank Eric Crawford,
from Michigan State University, for his comments on a first draft of this paper. I remain
responsible for its imperfections.






32 COLIN

I will not present here the APS concept (see Badouin, 1987); rather, I will
liken the APS with what would be an economic understanding of the farming
system, emphasizing three points:
1. Research on "farming systems" or "production systems" (in a generic
sense) is usually likened to "Farming Systems Research" (FSR);2 however, it
is important not to confuse them (Crawford, 1981). The main differences
between FSR and Agricultural Production Systems Research (APSR) are the
following:
FSRinvolves an interdisciplinary approach, whereas APSRis explicitly an
economic research program.
FSR is explicitly problem-solving research; its objective is the generation
and dissemination of relevant technologies through on-farm research3 -even
if, according to Tripp et al. (1990), the principal contribution ofFSRhas more
to do with methodological improvement than technology development. APSR
does not have this aim. I would have called it subject matter research (SMR),
but according to Johnson's definition (1986), SMRmust be multidisciplinary,
and this is not the case. We sometimes make the distinction, in France, between
cognitive research (recherche cognitive), aimed at understanding a given
subject without any explicit orientation toward action to change the situation
under study, and problem-solving research (recherche-action); APSR as pre-
sented here would be labeled as cognitive research.
The scope of FSR is limited-in practice more than discourse, showing
what Baker (1991) describes, from an economic viewpoint, as an over-
investment in technology production, with little interest in improved institu-
tional performance, or in feedback relating to the effects of policies and
development programs on producers. As defined, the scope ofAPSRis broader
as far as socioeconomic issues, and much more limited regarding technological
issues.
FSR stresses quick data collection, such as rapid rural appraisal, whereas
APSR, even if better implemented with more data collection, does not set up
any prescription regarding fieldwork techniques.
These differences do not preclude similarities such as a same-system
orientation (which does not mean a real "system science" perspective), on-
farm research, the recognition of farmers' goals and the relationships between
human and technical factors, and the recognition of local specificity and
heterogeneity.

2 On FSR methodology, I rely on Dillon (1976), Gilbert et al. (1980), Norman (1980),
Crawford (1981), CIMMYT Economics Staff (1984), Maxwell (1986), Collinson (1987),
Byerlee and Tripp (1988), Tripp ct al. (1990), Worman et al. (1990), Baker (1991), and
Crawford and Baker (1992). See Portires (1950) for a precursory analysis of the logic of
African traditional farming systems and sustainability, in terms of systemic relationships.
3 The systematic parallel drawn between on-farm research and FSR may be questioned. See
Sebillotte (1974, 1987) for an excellent epistemological analysis of on-farm research as an
essential component of fundamental agronomy.


Journal for Farming Systems Research-Extension






PRODUCTION OF KNOWLEDGE


2. The APS includes explicitly, as a principle point of analysis, the social
organization of production-that is, the way in which production units func-
tion: internal decision structure (who decides what?), conditions of access to
productive resources (land tenure system, labor relationships, etc.), and
relationships between the farm and its economic environment (parastatals,
cooperatives, markets, etc.).
The economic approach followed here is close to the American "Old
Institutionalist" stream (Colin and Losch, 1992). Research is not restricted to
the analysis of resource allocation but rather considers the social conditions of
access to resources. The economic calculation in terms of production costs and
factor productivity is set within the institutional context that gives it its
meaning. The economic calculation is not in itself the purpose of the research
but serves as an explanatory element of peasant practices and economic
dynamics, in addition to other factors. The analysis must include a diachronic
and spatial dimension, necessary for shedding light on the present conditions
of access to resources and, more generally, the ambient economic system. The
processes of economic differentiation are stressed; rural society is considered
neither stable nor homogeneous. This heterogeneity proves to be a determin-
ing factor for the understanding of the diversity of peasant practices, even at
local or regional levels. Starting with empirical questions, the research has to
provide a framework for the understanding of a specific, localized reality. This
viewpoint tends to distend the connection with established theoretical bodies
and with a "hard" disciplinary approach.
In the first part of the paper, I sketch a link between fieldwork and
explanations. In the second part, I illustrate this point through the presenta-
tion of two research experiences. In the Ivory Coast, it was possible to give
causal explanations of the dynamics of the agricultural production systems. In
Mexico, the APS were roughly described and some typologies and correlations
produced, but what I would consider satisfactory explanatory models were not
reached. The focus is on the objective of each program and the fieldwork
techniques used. For further information regarding the differences in the types
of knowledge produced, see Colin, 1992.

PRODUCTION OF KNOWLEDGE AND
DATA COLLECTION SYSTEM


Regarding Explanation
The deductive-nomological model of explanation proposed by Hempel and
Oppenheim (1948) is generally considered as the model of scientific explana-
tion. It defines a valid explanation as composed of two parts, an explanandum
(description of the event to be explained) and an explanans (including a list of
antecedent conditions and general laws), the former being a logical conse-


Vol. 4, No. 3, 1994






COLIN


quence of the latter. However, it is sometimes objected that this model of
explanation is too restrictive, especially in the social sciences, in which other
models may be more appropriate (Piaget, 1970; Grawitz, 1981; Caldwell 1982):
historical explanation (stressing the singularity of a historical cause); genetic
explanation (looking for the conditions for the occurrence of the events); or
motivational or functional teleological explanations (explanation by reference to
ends or purposes). The specificity of causality in social sciences is underlined by
Grawitz (1981): the purpose of the analysis is less to find the generating fact than
to discover the dynamics of interdependent facts. Another specificity of the
analysis of human behavior is that we have to understand the motivations and
reasons that induced the actors' decisions; this Weberian Verstehen implies the
understanding of the (subjective) meaning of the actions from the actor's
viewpoint (Aron, 1967).
Another model of explanation, the pattern model-considered the typical
institutionalist mode of explanation (Wilber and Harrison, 1978; Ramstad,
1986)-describes quite well the kind of procedure followed in the Ivory Coast. I
will follow Diesing (1971:142-167) in his description of the steps leading to the
construciton of such a model. Prior to fieldwork, the researcher builds a checklist
of "things to look for" on the basis of theoretical issues of empirical questions. The
first step consists in the socialization of the researcher-participant observers, which
allows them to be impressed by recurrent themes. The next step is to interpret
these themes, looking for their significance. These interpretations are tested
through contextual validation4 by cross-checking several types of evidence (e.g.,
data obtained through informant statements, documents, observation, etc.). The
final step is to build the model by connecting the themes in a network or pattern.
Themes and the linkages between them are thus explained byspecifying their place
in the pattern.
Diesing underlines some major differences between the pattern model of
explanation and the deductive model. In the deductive model, the anplanas is
always a general law; in the pattern model, both the explanandum and the
explanans are specific to the system studies, without reference to a general law. The
symetry between prediction and explanation, so crucial in the deductive model,
does not appear in the pattern model. Finally, a pattern model can never be
considered as complete and definitive, due to information constraints and also
because human systems are always changing.

Fieldwork and Explanation
It is my view that the main aim ofAPSRis not to put forward a general theory
or to embellish an existing theoretical edifice, but rather to provide explicative
models that are valid locally-that is, partial (as opposed to general) theories.5

4 As Wilber and Harrison point out, "this technique of contextual validation can never produce
the rigorous certainty espoused by logical positivists; it can only indicate varying degrees of
plausibility" (1987:76).
5 In short, as a sociologist, I favor Weber over Durkheim.


Journal for Farming Systems Research-Extension






PRODUCTION OF KNOWLEDGE


These theories-which always remain conjectural (Popper, 1990)-can be built
through a combination of historical, genetic, teleological, pattern model, or
even deductive-nomological modes of explanation. But the main point is that
constructing these locally valid models requires immersion in the local reality
in order to gain a many-sided perception of the rural society and accumulate
knowledge that is specific to that society. As Ramstad stated (1986:1075),
... one needs a theory capable of saying a great deal about a few cases, rather than
very little about all cases ... to develop 'practitioner's knowledge', that is,
knowledge directed to the understanding and control of the specific case. This is
in sharp contrast to the formalist's preoccupations with the development of
knowledge applicable to aggregates even if it is of limited applicability to
individual cases.
This practitioner's knowledge has to be soundly grounded in a researcher's
personal field experience.
Couty (1991:4) observes that "in social sciences, experience requires
personal, sincere and durable involvement in the historicity and singularity of
the situation under study. Without this we are threatened by mathematical
formalization or by verbiage, whichever one likes." We pointed out elsewhere
(Colin and Losch, 1992) how direct implication of the researcher in informa-
tion gathering has epistemological effects. The nearness of the realities of the
field (which are not given, but must be constructed from a paradigmatic
framework) in all their complexity makes one sensitive to the interrelations
between the economic, technical, and social dimensions of the problems.
Awareness of local is an excellent antidote against the reductive oversimplifi-
cations of the great theoretical constructions with universal claim, especially
when the purpose is to understand peasant practices in a specific environment.
This tradition also allows one not to sink in what Hirschman calls the syndrome
of the economist on an assignment: "(the) habit of giving peremptory opinions
and prescriptions while invoking economic principles and remedies of univer-
sal values ... after having barely got to know the 'patient'" (1984:76).
I disagree with Heady's (1952) description of information collection as
simple routine. Instead, I see this phase as a determining component of
research justifying on-site investment of the researcher, even with a PhD. Here
I join Parsons, an institutional economist who, writing as early as 1949,
stressed that data collection constitutes an integral part ofresearch-and one of
the most difficult.
In FSR methodology, the main researcher's direct participation in data
collection is generally limited to "sondeos," or "exploratory surveys." The job
has to be done quickly-"Social science research methods need to be flexible,
relatively simple, well focused and rapid"-as opposed to the use of "long and
tedious baseline questionnaires to obtain information about all aspects of the
system" (Byerlee and Tripp 1988:147). However, the question remains as to
whether these are the terms of the alternative; has the well-founded rejection
of long and tedious questionnaires led systematically and normatively to rapid


Vol. 4, No. 3, 1994






COLIN


rural appraisal? Defining the purpose of information collection is essential to
answering this question. I feel strongly that if one aims to understand and
explain farmers' socioeconomic and technical practices, rapid information
collection with a low researcher involvement may not be the best methodol-
ogy.
An implicit assumption of FSR methodology is that it is possible to define
the problems and discover their causes through rapid data collection. This may
sometimes be the case, but is surely not always true-especially regarding
socioeconomic issues. In FSR, as Baker remarked (1991:46), data quality is
often
pragmatically sacrificed in favour of quickly obtaining results. Unfortunately, in
many cases, "time- and cost-efficient" methods were not actually so efficient
since inadequate understanding was generated on the important factors influenc-
ing farmer behaviour and farming systems performance.
The risks of misunderstanding farmers' socioeconomic and technical prac-
tices and environment are real. Let me give just four illustrations.
Motivational explanation may encourage an easy construction of ex-post
facto accounts. The lower the researcher's personal first-hand knowledge of
the local society, the higher the risk.
As emphasized by Malinovski in anthropology (quoted by Salamone,
1979), or more recently by Milleville (1987) in agronomy, what people say
about what they do has to be distinguished from what they actually do. In
order to distinguish between the norm and the practice, it is questionable
whether rapid information collection is the most appropriate field technique.
In some contexts, the definition of socioeconomic categories such as
production, consumption, residence, and accumulation groups (Gastellu,
1980) may need some time.
If the analysis is too superficial, the theoretical and practical risk of causal
explanation is to over-reduce the conditions (causes) of what has to be
explained. Using the deductive-nomological model form, "ifL L1, L,... Ln, and
if C1, C2, ... C., then E",6 the risk is to put forward "if C1, then E," where it
would have been necessary to write "if C1, C2, ... C., and especially C1, then
E" (Mingat et al., 1985). One way to limit this risk is to develop models of
explanation that are as complete as possible-that is to say, to have the best
possible understanding of the situation under study.
This brings up to the key question of how far one has to go in empirical
observation to avoid these pitfalls. Or, to put it differently, how can we know
that we have reached a satisfactory explanation? It is always possible to build
a coherent explanatory model, whatever our knowledge of the situation-even
if it remains very superficial. But it is also possible to progress toward more
satisfactory (complete) explanations-under the stimulus of better empirical
knowledge, and under the pressure of external criticism. Unfortunately, in our
6 With L for law, C for conditions (explanans), and E for what has to be explained (explanan-
dum).


Journal for Farming Systems Research-Extension






PRODUCTION OF KNOWLEDGE


field of study characterized by locally specific research, little can be expected
from peers' criticisms when the researcher handles (consciously or not) the art
of rhetoric, except when those peers have sufficient knowledge of the situation
under study to question a weak but coherent explanation.
Thus our question remains unanswered, for there are no norms, no recipes
allowing one to define ex ante or to evaluate ex post the appropriateness of a
researcher's fieldwork investment-apart from an ex post subjective evaluation
of the researcher him/herself. On the basis of the two research experiences
related in this paper, I feel that I succeeded at least partially in building
explanatory (qualitative) models in the Ivorian case, and that I would have
failed in the Mexican case had this been the purpose of the research.
The origin of this difference in the type of knowledge produced using the
same conceptual framework can be traced to differences in the objectives of the
research programs, and in the way the fieldwork was conceived and organized.
Let me turn now to the presentation of the two research experiences to
illustrate this point in concrete terms.

FROM INTENSIVE TO EXTENSIVE INFORMATION
COLLECTION METHODS


The Place of APS Analysis in the Research Programs
In the Ivory Coast, research was concerned with the dynamics of a
smallholder plantation economy. This economy can be characterized by land
abundance, labor scarcity, and extensive coffee and cocoa production. Two
dominant factors of change act upon this model: an increasing land shortage
and, mainly in the Lower Coast, diversification of farming systems led by
agroindustrial parastatals that are developing smallholder contractual farming.
Understanding the evolution of the peasant plantation economy in this new
and relatively specific context was the purpose of the study. Two fundamen-
tally related topics were addressed by the research: (1) the sources and features
of technical and institutional changes, and their incidence on the plantation
economy; and (2) the production strategies adopted by farmers, according to
their different opportunities, resource availabilities, and objectives. The re-
search required a holistic, empirical analysis encompassing cropping systems
(food and cash crop relations, place of new crops), input combinations (toward
intensification), and the social organization of production (evolution of the
land tenure system, evolution of labor relationships, analysis of peasants-
parastatals relations).
In Mexico, the analysis of the agricultural production systems was only the
first stage of an economic program that was itself part of a multidisciplinary
study of the production and commercialization of potatoes in the central area
of the country. This diagnostic (not problem solving) study was initiated at the


Vol. 4, No. 3, 1994





COLIN


request of the Veracruz State authorities, who were facing an agricultural crisis
in the Sierra (Cofre de Perote region). The crisis was linked to agroecological
and economic problems facing potato production-a near monoculture in this
region. An agronomic program was concerned with the study of potato yield
and producers' technical practices. The economic program, implemented by
four researchers, had various components beyond the conception of the APS
research: the study of sharecropping systems, the labor market, and the potato
subsector. The author of this paper had responsibility for the APS analysis and
for the study of sharecropping. The purpose of the APS study was to draw a
general picture of the socioeconomics of potato production in the area
delimited by the Pico de Orizaba (Puebla State) and the Cofre de Perote
(Veracruz State). This was done to provide an overview for the other compo-
nents of the research program, and to prepare the analysis of the sharecropping
system.
The features of potato production (a cash crop with great price variations
and high production costs), the economic context (breaking-down of the
prices since 1989, bank credit rationing), and the socioeconomic production
(differentiated production structures) led to the following questions: Is there
an economic efficiency difference between small- and large-scale farmers,
between ejidatarios and pequeros proprietarios,7 between the Sierra (moun-
tains) and the Altiplano (mountain plateau)? What is the impact of the potato
price crisis, and what are farmers' reactions?
The two APS studies differed in the following ways:
In the Ivory Coast, the economic APS study was the objective of the
research, conducted by only one economist. In Mexico, the purpose was to
draw a panorama, a context in which to situate other economic or agronomic
studies; in other words, the research served as a simple component of broader
multidisciplinary research.
In the Ivory Coast, the approach was open to all the diversity of
agricultural production systems at a local level. In Mexico, it was restricted,
focusing on a specific crop. Because potato production was almost the only
enterprise on the farms studies, the "system" perspective came mainly from the
analysis of farm and off-farm activities.
Thus the researcher's direct participation in data gathering, the length of
the data collection stage, the complexity of the information collection system,
and the quality of the data were intensive in Ivory Coast and extensive in
Mexico."
7 Ejidatarios are producers who own land in common; pcquejos proprietarios are smallholders
who own their own land.
8 For analysis of data collection methodology, see the abundant literature produced by the
AMIRA group (Improvement of investigation methods in African rural areas), e.g., Couty and
Winter (1983). The AMIRA network has played a driving role in producing and publishing
documents on as varied themes as the comparison of random and non-random sampling, the
connection between the qualitative and the quantitative, the complementary of statistical
survey and monographic studies, the definition of economic units, and the problem of scale
of analysis.


Journal for Farming Systems Research-Extension






PRODUCTION OF KNOWLEDGE


From the Researcher-Observer to the Researcher-Supervisor
In the three-year research project in the Ivory Coast,9 the researcher was a
constant presence in the field and participated directly in the information
collection, with the help of one permanent high-school level enumerator, and
temporarily with a team of topographers. In Mexico, the information collec-
tion stage for the APS study was limited to five months, and largely delegated
to enumerators (four agronomists and two graduates in history and geogra-
phy). In the Ivory Coast, the researcher used data collection techniques close
to the fieldwork tradition of anthropologists-but common in the practice of
French Africanist rural economics (see Colin and Losch, 1992). In Mexico, he
took care of research design, logistics, and data analysis, but had little direct
involvement in data gathering.

From a Simple Village Study to a Multi-Localized Survey
An essential difference between the two programs came from the choice of
a single-site study in the Ivory Coast (mainly a single village) and multi-
localized sites in Mexico.
In the Ivory Coast, the research was an in-depth analysis of the economy of
agricultural production at the level ofa Lower Coast village (Djimini-Koffikro,
in Adiak6 sous-prifecture). Study of a simple village afforded various advantag-
es. The village is often an optimal level at which to observe the relationships
between production units, and the diversity of producers' constraints, strate-
gies, and practices. Here it was considered a life-size laboratory in which to
observe changes in the plantation economy. Another advantage was that the
crosschecks permitted by the researcher's immersion in everyday village life
greatly improved the quality of data.
In Mexico, the APS analysis had to be carried out quickly and it had to
ensure a regional perspective. For these reasons, the study was conceived as a
classical questionnaire survey, with the collection of information on a sample
of production units in various villages.

The Question of Representativeness
In neither case were sample sites chosen at random. This choice merits some
discussion.
The major criticism of approach chosen in the Ivory Coast is the lack of
representativeness of a village study, and consequently the impossibility of
inferring rigorously from the conclusions. But first, let us recall that orthodox
statistical approaches may and often do hide, by the "scientificity" of the
figures, the fundamental problem of the quality of field data on which analysis
is based.'1 Furthermore, there is a significant risk of reductionism, as these
9 Followed by one year in France to complete data analysis and write the report.
10As stated Salamone (1979:57), "I argue that no matter how sophisticated are methodological
analyses or how elegant our theory construction, if our primary data-gathering techniques are
faulty, then our theories and methodologies will be but elegant exercises in futility."


Vol. 4, No. 3, 1994





COLIN


approaches presuppose the collection of information on the basis of a concep-
tual framework that may not reflect the complexity of the real world. However,
the lack of representativeness criticism is well-founded if the study pretends to
build a general model of refined applicability. Such was not the case in the
Ivory Coast research experience. The village was conducted was not intended
to be representative of the villages in the area. Djimini-Koffikro was deliber-
ately chosen for its specific characteristics of land shortage, and its wide range
of cropping opportunities related to the simultaneous intervention of several
parastatals. These features, especially the presence of cropping opportunities,
were expected to give rise to a range of production strategies and behavioral
patterns. Our objective was to build models of behavior of production units
taken from a life-size laboratory, which under the same economic and
institutional context faced different constraints in land, labor, etc.
Data was gathered for some variables in all the farms of the village, or in the
farms for which the variable under study had meaning, and for other variables
the information was collected through purposive sampling (informal quota
method)." Complementary surveys were carried out in four other villages, all
farmers being interviewed, to test whether the behavioral models identified in
Djimini-Koffikro could adequately explain the strategies adopted by farmers
operating under different circumstances. These villages were chosen from the
same region to be as different as possible from Djimini-Koffikro, on the basis
of two variables, easily discernible in a pre-survey: the type of tree crops
planted, and ethnic composition of the village.
In the Mexican case, the village was chosen as the first sampling unit, on the
hypothesis of the existence of a village specificity regarding potato production
and marketing conditions-the existence of this village-effect was, in fact,
proven by the investigation. After a pre-survey (visit to all villages, and
collection of data regarding the number of potato growers and potato
cropping cycles), the villages were chosen on the basis of their location
(distribution along the Sierra, and between Sierra and Altiplano) and the
number of producers, in order to include systematically the most important
centers of the potato production zone.
Contrary to the initial plan, the farms to be surveyed in each village were not
chosen from probability sampling; first, because of the lack of sampling frame
and the unfeasibility of constructing one for what was primarily an exploratory
survey-some of the villages having around 500 production units; second,
because of a serious reluctance of the producers to participate in surveys. This
reluctance comes, on the one hand, from governmental taxation and manda-
11 Purposive sampling is a method inn which the selection of the units to be surveyed is subject
to conscious purpose. The informal quota method is an informal method involving non-
probability sampling (e.g., a procedure of selection of the units which does not allow th cause
of sampling theory and the estimation of the sampling error. Quota refers to the non-
probability selection of produciton units, aiming to include a given number of units belonging
to specific groups offarmers (i.e., the sample was realized after the construction of a qualitative
typology of production units, and on the basis of this typology). See Casley and Kumar, 1988.


Journal for Farming Systems Research-Extension






PRODUCTION OF KNOWLEDGE


tory permits to grow potatoes and, on the other, from the lack of understand-
ing of a study that did not pretend to benefit the interviewed farmers directly.
Proceeding with a random sample would have required a great deal of time
spent in explanations to producers as to why they were registered on a list, and
how their names were chosen. An agreement dragged from them would not
have guaranteed a true willingness to participate, and the risk of getting
voluntarily-biased information was too high. Therefore, interviews were first
focused on the volunteers who registered themselves after the meetings that
were organized in each village to explain the research. Then, to eliminate major
selection biases, we sought to cover the complete range of intra-village
variations in terms of acreage under potato, through an informal quota
sampling. We tried to interview the largeholders systematically; because they
were rarely present at the meetings, they required a researcher's specific
"diplomatic investment." As recognized in a FSR handbook, "Ideally, strati-
fied sampling probably would be best. (But) practical realities make compro-
mises necessary" (Worman et al., 1990:139).
The position advocated here is that, in studies of production economics in
LDC, rigor does not always consist in using formal random sampling proce-
dures, with biased sample frames and answers-generally detected by the
researcher but overshadowed in the final report. It consists of recognizing the
almost impossibility of using such a tool in some situations, and trying to put
as much rigor as possible in more "informal" sampling techniques-knowing
that this choice would certainly upset the supporters of "scientific" formalism
and will not permit the publication of the study's results from most academic
journals.

From Complex to Simple Data Collection Systems
In the Ivorian case, different data collection systems (DCS) were used in
Djimini-Koffikro and in the other villages. In Djimini, the collection of
information was based on a mixture of various techniques, from qualitative to
structured surveys; in the other villages, a simple, one-shot, formal question-
naire survey was administered. In Djimini-Koffikro, the topic of the study
required the collection of two kinds of information:
Historical information sketching the history of migrations, the first land
occupation and exploitation, land tenure system changes, etc. This informa-
tion was collected through topic-focused interviews with old farmers because
the use of the colonial archives turned out to be disappointing.
Information regarding the current context and conditions of agricultural
production. In addition to direct observation allowed by the researcher's
lengthy stay in the village, the collection of this data required three techniques:
(1) qualitative interviews; (2) topic-focused interviews to grasp, in rather
qualitative terms, the logic of the functioning of the agricultural production
systems; and (3) measurements and questionnaire surveys to systematize the
data collection and, when necessary, to quantify them.


Vol. 4, No. 3, 1994





COLIN


The basic DCS framework was provided by an initial agricultural and
demographic census, by the measurement and mapping of the plots of all the
village farms (as there was no previous cadastral survey),12 and a weekly 34-
farm sample survey over a one-year period. Furthermore, a set of specific topic
semistructured or structured surveys included a review of land transactions
since the end of the pioneer era for all land owned by farmers in the village, off-
farm activities, migration itineraries, labor requirements per hectare for each
crop, crop yield measurements, production strategies, labor availabilities,
agricultural product sales for all farms not sampled for the weekly survey,
technoeconomic analyses of local processing activities, etc. (see Colin, 1990).
Thus the information was collected at various levels, according to the
variables under study:
All 180 production units of the village. Basic history of the farm; land
tenure; area of cultivated crops; agricultural product sales (excluding market
gardening); cash-crop yields (estimated on the basis of product sales, assuming
no or insignificant autoconsumption); quantitative analysis of permanent
labor force availabilities; qualitative analysis of temporary labor force em-
ployed; and one year agricultural net income (excluding temporarily hired
labor cost and market gardening income).
34-farm sample (weekly-visit survey). Quantitative analysis of labor use
and social division of labor; quantification of market gardening; off-farm
income; and budget data.
Ad hoc samples. Food-crop yields; labor requirements per hectare,
processing activities, etc.
The frequency of information collection varied according to the variables:
single-visit surveys to grasp structural data; multiple-visit surveys to establish
flows of labor, products, and money. Multiple-visit surveys included the
weekly visit of the 34-farm sample, as ad hoc visit surveys; for example, the
collection of marketing information was organized for each of the 180 farms
on the basis of the anticipated harvesting time calendar, specific for each plot
for those crops (as cassava) that did not present a common cropping cycle
(planted and harvested all year long). Therefore, the combination of tech-
niques used during the study of this village economy borrowed various types
of data collection techniques, from qualitative surveys to structured question-
naires, and from micro-approaches (in-depth study on a limited sample) to
macro-approaches (collection of rough quantitative or qualitative data on a
large number of production units).
In the four other villages to which the research was later extended, a single-
visit questionnaire-built on the basis of the knowledge produced by the former
baseline study-was administered to all farms in each village (128 production
units in total), without plot measurement or direct observations. Information
was collected on variables such as farm structure, historical processes (migra-
12To obtain reliable data on land availability and use, and to provide a solid baseline for future
studies of the evolution of this village economy.


Journal for Farming Systems Research-Extension






PRODUCTION OF KNOWLEDGE


tions or evolution of access to land), and behavioral variables (such as the
determinants of crop choices).
In Mexico, the DCS was a single-visit questionnaire survey of a 245-farm
sample selected from seven villages. Information was collected on: (1) farm
structure (area owned and cultivated, land tenure system, crops cultivated,
permanent labor force availabilities, machinery, and agricultural financing);
and (2) potato production (recent evolution ofacreage, and for each plot sown
in 1990: variety, cropping cycle, production, and sales). Detailed production
and marketing costs for 1990 were also collected for one potato plot per
farm." Because producers' estimates of acreage were not reliable, we planned
to measure the plots to improve the quality of the quantitative economic
analysis; however, it was possible to realize this time-consuming measurement
for all farms. A general questionnaire was also filled out for each village, during
interviews with key informants, to collect historical information regarding the
village, potato production and marketing, land access, etc.
A common weakness of both the Ivorian and Mexican programs has to be
stressed: over-investment during what was initially devised as only the first
stage of the research. In the Ivory Coast, extension of the research to other
villages was excessively subordinate, in comparison with the simple village
study, and should have received a more significant time investment. In Mexico,
what was planned initially as a rapid baseline survey turned into a rather heavy
study, requiring major fieldwork and data analysis that was probably not
justified by the quality of the data. Thus between these two remote poles of the
methodological continent, the equilibrium point is still to be found-assuming
it exists.

The Search for Data Quality: From Purism to Compromise
The differences in DCS led to differences in data quality between the two
research programs. In the Ivory Coast, the DCS monographic stage involved
a search for information as reliable and precise as possible. This search used a
set of measurements (acreage, yields), the duration and localized character of
the information collection (allowing acquaintance with farmers, progressive
rectification of errors in structural data, crosschecks of the information), the
multiple-visit system for flows-information collection, and the researcher's
existing contextual knowledge.
In the Mexican experience, the objectives and conditions of the APS study
precluded such data quality. However, the intent ofsome elements of the DCS
was to limit the risk of collecting excessively questionable information: the
selection of highly qualified enumerators; the one-month stay of the enumer-
ators in each village, which offered the possibility of building up confidence
relationships with farmers and the progressive subjective evaluation of the
quality of data collected; and the fact that questionnaires were filled out only
13Using only 159 of the 245 questionnaires, after eliminating questionnaires of doubtful
reliability or, more frequently, presenting incomplete information.


Vol. 4, No. 3, 1994






COLIN


with volunteers. The decision to measure one plot per farm to provide a sound
basis for quantitative economic analysis is an example of excessive concern for
data quality. Seen a posteriori, the decision was not justified for this study, the
other data remaining of a relatively rough quality. However, even these
attempts to reduce the risk of collecting low-quality information were not
sufficient to overcome the limits on information collected through a single-
visit questionnaire survey.

CONCLUSION

This account of the two research methodologies provides an opportunity to
examine the kind of knowledge produced by each. Of course, the objective of
the research can never be forgotten.
The method followed in the Mexican research program is probably well
justified if the research aims at sketching the main structural characteristics of
a given situation, or answering a few simple questions on the basis of an already
available database. This kind ofknowledge is no doubt useful, but will provide
less an explanation of a situation than a set of correlations.1
If, however, the objective is to understand peasant decision-making and its
institutional setting, to go beyond the 'what they are doing' to reach the 'why
they are doing it', to build explanatory models, then such a method is
inadequate. It constitutes only a pre-survey- the first stage of research that has
to be followed by a personal field-investment of the researcher (without
spending necessarily three years!). To take a concrete example, the first
insights for the analysis of sharecropping systems in potato production, in the
Mexican case, came from the APS analysis, but subsequently required specific
fieldwork carried out directly by the researcher and a graduate student who
lived for several months in two villages.
Ultimately, the problem does not lie in the choice of method, but in the
purpose ofits use, the bias-is it exceptional in the vast field of farming systems
research?-occurring when one pretends to complete the second type of
objective using the first type of method.

REFERENCES

Aron, R. 1967. Les tapes de la pensie sociologique. Paris: Gallimard. 663 pages.
Badouin, R 1987. L'analyse 6conomique du system productifen agriculture. Cahiers
des Sciences Humaines 23(3-4):357-375.
Baker, D. 1991. Economic research in farming systems research: Steps in methodol-
ogy. Agricultural Systems in Africa 1(1):45-51.


14For example, it has been possible to answer the questions regarding economic efficiency
differences between farmers, but the analysis could not offer a coherent and well documented
explanation of these differences.


Journal for Farming Systems Research-Extension






PRODUCTION OF KNOWLEDGE


Byerlee, D., and R. Tripp. 1988. Strengthening linkages in agricultural research
through a farming systems perspective: The role of social scientists. Experimental
Agriculture 24:137-151.
Caldwell, B. 1982. Beyond positivism: Economic methodology in the twentieth century.
London: George Alien & Unwin. 277 pages.
Casley, D., and K Kumar. 1988. The collection, analysis and use of monitoring and
evaluation data. Baltimore and London: The Johns Hopkins University Press. 174
pages.
CIMMYT Economics Staff. 1984. The farming systems perspective and farmer
participation in the development of appropriate technology. Pages 362-377 in C.
Eicher, J. Staatz, eds., Agricultural Development in the Third World. Baltimore and
London: The Johns Hopkins University Press.
Colin, J.-P. 1990. La mutation d'une iconomie de plantation en base Cdte d'Ivoire.
Paris: ORSTOM. 284 pages.
Colin, J.-P. 1992. La production de la pomme de terre dans les Etats de Puebla et de
Veracruz (Mexique). Elements d'analyse 6conomique. Unpublished work, Colegio
de Postgraduados (Centro de Economia Agricola), Texcoco, Mexico. Paris: ORS-
TOM.
Colin, J.-P., and B. Losch. 1992. But where on earth has Mamadou hidden his
production function? French Africanist rural economics and institutionalism. To be
published in J. Acheson, ed., Anthropology and Institutional Economics. The
University Press of America.
Collinson, M. 1987. Farming Systems Research: Procedures for Technology Develop-
ment. Experimental Agriculture 23:365-386.
Couty, P. 1991. Editorial. Chroniques du Sud (ORSTOM) 4:3-5.
Couty, P., and G. Winter. 1983. Qualitatifet quantitatif. Deux modes d'investigation
compl6mentaires. R6flexions a partir des recherches de I'ORSTOM en milieu rural
africain. Paris, INSEE-ORSTOM, Note AMIRA # 43.
Crawford, E. 1981. Farming systems research and agricultural economics. Michigan
State University, Farming Systems Research Group, Working Paper # 1.
Crawford, E., and D. Baker, 1993. Farming systems research and economics. Pages 75-
81 in H. Navarro, J.-P. Colin, and P. Milleville, eds., Sistemas de producci6n y
desarrollo agricola. Texcoco, Mexico: condition ORSTOM-CONACYT-Colegio
de Postgraduados en Ciencias Agricolas.
Diesing, P. 1971. Patterns of discovery in the social sciences. Chicago: Aldine-Atherton.
Dillon, J. 1976. The economics of systems research. Agricultural Systems 1(1):5-22.
Gastellu, J.-M. 1980. Mais oui sont done ces unites dconomiques que nos amis
cherchent tant enAfrique? Cahiers ORSTOM des Sciences Humaines28(1-2):3-11.
Gilbert, E., D. Norman, and F. Winch. 1980. Farming systems research: A critical
appraisal. MSU Rural Development Paper 6, Department ofAgricultural Econom-
ics, Michigan State University.
Grawitz, M. 1981. Mithodes des sciences sociales. Paris: Dalloz. 1102 pages.
Heady, E. 1952. Economics of agricultural production and resource use. New York:
Prenctice-Hall.
Hempel, C., and P. Oppenheim. 1948. Studies in the Logic of Explanation. Philosophy
of Science 15:135-175.
Hirschman, A. 1984. L'economie comme science morale et politique. Paris: Hautes
Etudes, Gallimard, Le Seuil. 111 pages.


Vol. 4, No. 3, 1994






COLIN


Johnson, G. 1986. Research methodology for economists. Philosophy and Practice. New
York: Macmillan Publishing Company. 252 pages.
Maxwell, S. 1986. The social scientist in farming systems research. Journal of
Agricultural Economics 37:25-35.
Milleville, P. 1987. Recherches sur les pratiques des agriculteurs. Cahiers de la
Recherche-Developpement 16:3-7.
Mingat, A., P. Salmon, and A. Wolfelsperger. 1985. Mithodologie iconomique. Paris:
Presses Universitaires de France. 576 pages.
Norman, D. 1980. The farming systems approach: Relevancy for the small farmer. Rural
Development Paper # 5, Department of Agricultural Economics, Michigan State
University.
Parsons, K. 1949. The logical foundations of economic research. Journal of Farm
Economics 31(4):656-686.
Piaget, J. 1970. Epistimologie des sciences de l'homme. Paris: Gallimard. 380 pp.
Popper, K. 1990. Le rialisme et la science. Paris: Hermann. 427 pp. (Translated from
The Postcript to the Logic of Scientific Discovery. I Realism and the Aim of Science.
London: Hutchinson. 1983).
Porteres, R. 1950. La recherche agronomique dans les pays chauds. Lemon inaugural
du Cours d'Agronomie Tropicale. Revue Internationale de Botanique Appliquie &
d'Agriculture Tropicale 331 and 332:241-263 (two-volume series presented as one
volume).
Ramstad, Y. 1986. A pragmatist's quest for holistic knowledge: The scientific method
of John R. Commons. Journal of Economic Issues 20(4):1067-1105.
Salamone, F. 1979. Epistemological implications offieldwork and their consequences.
American Anthropologist 81(1):46-60.
Sdbillotte, M. 1974. Agronomie et agriculture. Essai d'analyse des teaches de l'agronome.
Cahiers ORSTOM, sirie Biologic 24:3-25.
Sebillotte, M. 1987. Les d6marches de l'agronome en situations paysannes. Elements
de r6flexion. Paper presented at the 4th Thailand National Farming Systems
Seminar, Prince of Songkla University, Haad Yai, Thailand, April 7-10, 1987.
Tripp, R., P. Anandajayasekeram, D. Byerlee, and L. Harrington. 1990. Farming
systems research revisited. Pages 384-399 in C. Eicher, and J. Staatz, eds.,
Agricultural Development in the Third World. Baltimore and London: The Johns
Hopkins University Press.
Wilber, C., and R. Harrison. 1978. The methodological basis of institutional econom-
ics: Pattern model, storytelling, and holism. Journal of Economic Issues 7(1):61-89.
Worman, F., D. Norman, and J. Ware-Snyder. 1990. Farming systems research
handbook for Botswana. ATIP, Department of Agricultural Research, Ministry of
Agriculture, Botswana.


Journal for Farming Systems Research-Extension








Evolving Crop-Livestock Farming Systems
in the Humid Zone of West Africa:
Potential and Research Needs '

M.A. Jabbar2


INTRODUCTION
Livestock is an integral part of the economies of most Sub-Saharan African
countries. For the region as a whole, livestock constituted eight percent of
total Gross Domestic Product (GDP) and 25 percent of agricultural GDP in
1988. If the values of intermediate products such as traction and manure were
included, livestock's share of agricultural GDP might be as high as 35 percent
(ILCA, 1987; Winrock, 1992).
The incidence, functions, and relative importance of different types of
livestock vary across countries and agroecological zones. The main focus of
this paper is the humid zone consisting of rain forests and derived savannas
located mainly in Central and West Africa. The purpose is to assess the status
and potential of livestock development in the zone and determine research
needs and priorities.

PREVALENCE OF LIVESTOCK IN THE HUMID ZONE
The incidence of tsetse flies and trypanosomiasis has been the single most
important determinant of the distribution of livestock across ecological zones.
The humid zone has generally been considered unsuitable for livestock
production due to high tsetse-fly infestation (Stenning, 1959). However,
transhumant pastoralists from the semi-arid and sub-humid zones visit the
derived savannas during the dry season, when the tsetse challenge is reduced,
in search of feed and water. In fact, transhumant pastoralism made cattle
production viable in the given ecological stratification in West Africa.
Where the tsetse challenge allowed and/or where an acceptable degree of
tolerance developed in the livestock, there has been a tendency among
pastoralists to remain in the more humid areas. Over a long period, this process
has led to a degree of adaptation, facilitating permanent exposure of livestock
to light tsetse challenge (Ford, 1971; Fricke, 1979). There are also breeds of
cattle, goats, and sheep that have developed trypanotolerance through long
1 Presented at the 12th Annual Association for Farming Systems Research-Extension Sympo-
sium, Michigan State University, East Lansing, Michigan, 13-18 September 1992.
2 International Livestock Centre for Africa, Humid Zone Programme, PMB 5320,
Ibadan, Nigeria.






JABBAR


periods of exposure to high tsetse challenge. Such animals are kept by
pastoralists in some regions with high tsetse challenge and by crop farmers in
the humid zone.
More recently, changing climatic patterns, land and bush clearance for
agriculture due to population pressure, and specialized tsetse-fly control
programs have contributed to expand the tsetse-free areas. By 1978, over
300,000 km2 were freed from tsetse flies in Tropical Africa, of which over
200,000 km2 was in Nigeria alone (Ford, 1971; Jahnke, 1982). Some parts of
the derived savannas were covered directly by the organized control programs
(Putt et al.,1980; Bourn, 1983), whereas other parts were indirectly affected
by the spillover effects of the programs in adjoining sub-humid areas; as the
borderline of the tsetse-free area moved southward into the savannas, the
tsetse challenge in the adjoining humid areas became lighter. The degree of
tsetse challenge continued to decline due to increased incidence and severity
of bush fires as well as increased human settlement in cleared areas. Such areas
attracted pastoralists for seasonal grazing and subsequently for more perma-
nent settlement (Bourn, 1983).
The overall consequences are that the humid zone now has several times
more cattle and small ruminants than, say, two decades ago, and an increasing
number of these cattle are trypanosuceptible Zebus. At present there are 8.8
million cattle and 19.8 million small ruminants in the humid zone (Table 1).
In the late 1970s, an estimated 7.6 million trypanotolerant cattle and 26.7
trypanotolerant sheep and goats were raised in the humid and subhumid zones
(Jahnke, 1982). Thus it can be reasonably assumed that a significant propor-
tion of the cattle in the humid zone are trypanosusceptible Zebus.
Such generalization can be substantiated on the basis of more specific
information from specific areas. For example, the number of cattle in the
humid zone of Southwest Nigeria increased from 43,000 in 1950 to over
200,000 in the early 1980s (Table 2). Most of the increased cattle are Zebus.
On the basis of aerial survey RIM (1988) estimated that over 100,000 cattle
owned by transhumant Fulanis were seasonal migrants from the north into the
derived savanna but a population of nearly 200,000 cattle were stationary in
the zone. Cattle densities were higher near cropping areas, open woodlands,
and good sources of water. Between four and six thousand semi-permanent
rugas (group of households) were located along the cattle-grazing routes.
These are indications of a tendency toward sedentarization among some cattle
owners.
Cattle movement into the derived savanna of southeast Nigeria is a
relatively recent phenomenum compared to that in the southwest, yet esti-
mates of Akinwumi and Ikpi (1985) and of di Domenico (1989) suggest a
rapidly increasing process ofsedentarization. Such increases in cattle numbers
can be explained by significantly low incidence of trypanosomiasis and its
vector, tsetse flies. Field studies in the Southwest (Ikede et al., 1987; Reynolds
and Opasina, 19.87) and in the southeast (Nwanta, 1988) recorded low


Journal for Farming Systems Research-Extension






CROP-LIVESTOCK FARMING SYSTEMS


pastoralists and then agropastoralists by adopting crop cultivation. Evidence
of increasing investment in cattle by crop farmers, traders, and retired civil
servants was also observed.

Table 1. Some characteristics of agroclimatic zones in Sub-Saharan Africa, 1989.
HUMID SUB-HUMID SEMI-ARID HIGHLANDS
MAIN REGION CENTRAL WEST + WEST EAST
+ WEST CENTRAL
Arable land and
permanent crops (mil ha) 19.8 16.1 38.5 18.8
Permanent pasture (mil ha) 38.1 43.5 318.1 50.1
Tsetse free area (percent) 10.3 31.8 49.7 80.3
Agric in total GDP (percent) 29 33 31 38
Livestock in Agric
GDP (percent) 6 7 37 27
Value of net milk trade,
mil$ 56 78 145 33
Value of net meat trade,
mil$ 138 40 -70 -3
Ruminant Livestock (mil)
Cattle 8.8 32.8 45.5 29.0
Goats 11.6 20.3 33.2 11.9
Sheep 8.2 14.2 23.1 21.4
Main animal feed Natural Grazing natural Harvested
vegetation vegetation and residues crop residues
Main source of Fallowing, Fallowing, Manure, Manure,
soil fertility fertilizer fertilizer, fertilizer fertilizer
manure
Animal traction Absent Emerging Present Present
Source: McIntire et al., 1992.


Table 2. Estimated cattle population in Southern Nigeria.
YEAR AREA CATTLE NUMBERS BY TYPE
ZEBU TRYPANOTOLERANT TOTAL
1950 Western Region 43300
1954 Western Region 70000
1959 Western Region 65000
(Oyo Province) (32000)
1982 Derived savannah Southwest
Wet season 302000
Dry season 187000
1984 Total South 240000 67000 307000
Southwest 161000 38000 199000
(Oyo State) (115000) (25000) (140000)
Southeast 79000 29000 108000
1989 North Anambra (SE) 53000 14000 67000
Source: Shaw and Colvile, 1950; IMF, 1954; Akinwumi and Ikpi, 1985; RIM, 1988; di Do-
menico, 1989.


Vol. 4, No. 3, 1994





JABBAR


incidence and infection rates among sedentary cattle. These studies concluded
that trypanosomiasis was not a major problem for Zebu cattle production in
the derived savanna zone, though protection against the disease was necessary.
Moreover, with good husbandry and feeding, Zebu cattle were found to
acquire, through natural selection, some degree of trypanotolerance.
Recent reconnaissance surveys by ILCA in Accra Plains and Central Ghana,
in the Maritime and Plateau zones in Togo, the Coastal belt in the Republic
of Benin and the Northwest Province of Cameroon have indicated an ongoing
process of sedentarization whereby nomadic pastoralists become sedentary.

CATTLE PRODUCTION SYSTEMS

Although general descriptions of different cattle production systems are
available, information on proportions of cattle reared under different systems
is scarce. It is reported that over 80 percent of cattle in West Africa are reared
under some form of crop-livestock system (World Bank, 1987). Such systems
include sole livestock farming practised in proximity to and functional associ-
ation with sole crop farming, and crop-livestock mixed farming under the same
management with various degrees of mixes (Jahnke, 1982). The incidence of
more integrated crop-livestock farming is higher in the drier zone than in the
humid zone. For example, about 40 percent of the cattle and small ruminants
in the Semi-arid (Sahelian) zone are owned by crop farmers (World Bank,
1987). Similarly, most of the trypanotolerant cattle in the humid zone are
owned by crop farmers and most of the Zebu cattle are owned by sedentary
pastoralists.
Crop-cattle farming in this zone is a recently evolving phenomenum. Recent
surveys in Southern Nigeria show that cropcattle farming is emerging from two
directions. The dominant line is that of nomadic Fulani pastoralists becoming
sedentary pastoralists and eventually agropastoralists. A minor but perceptible
line is that of crop farmer becoming mixed farmer by purchasing cattle and first
giving them to Fulani herdsmen for management or hiring Fulani herdsmen for
management, then taking up management to themselves after gaining experi-
ence. For example, in a sample of 66 cattle farmers in southwest Nigeria, 53 (80
percent) were Fulani and 13 (20 percent) were indigenous Yoruba. Eight out of
the 13 Yoruba cattle owners managed cattle by themselves and five others hired
Fulani herdsmen for management or on a caretaking sharecroppingg) basis
(Jabbar et al., 1992). In another sample of 52 Fulani agropastoralists, 30 (28
percent) managed their own cattle and 22 (42 percent) managed their own as
well as those given to them for caretaking by 64 different Yoruba people. There
were about three owners per caretaker. Nineteen percent of the total cattle
managed by the sample farms were owned by the absentee Yorubas. Among 64
Yoruba cattle owners, 30 percent were crop farmers, 15 percent livestock
traders/buchers, and 55 percent had trading or government service as an
occupation in addition to farming (Mohammed, 1990).


Journal for Farming Systems Research-Extension






CROP-LIVESTOCK FARMING SYSTEMS


Cattle farmers were found to settle in the supply hinterland of small and
large cities (in the peri-urban areas) in order to gain access to a regular market
for their products (mainly dairy). Crops farmers' investment in cattle for
fattening is also a sign of commercialization.
On the basis of aerial survey, RIM (1988) found an average grazing unit (a
proxy for herd size) of 50 in the region. Mohammed (1990) found average
herd size of 27 among sample agropastoralists. Jabbar et al. (1992) found an
average herd size of 65 and 38 for Fulani and Yoruba owners, respectively. The
Fulani sample included some sedentary pastoralists. Generally, herd sizes were
larger among recent settlers (sedentary pastoralists); however, with longer
duration of settlement and with cattle rearers' involvement in crop produc-
tion, the herds became less mobile between seasons and herd sizes decreased.
Increased competition for labor between cropping and herding may contrib-
ute to reduced mobility and herd size. Akinwumi and Ikpi (1985) observed
smaller herd sizes in the southwest compared to the southeast, which could be
explained by the fact that farmers in the southeast were more recent settlers.
With longer duration of settlement, the proportion of farms with mixed
Zebu/trypanotolerant cattle herds increased (de Jode, 1989; Mohammed
1990; Jabbar et al., 1992). This was a risk management strategy given the fact
that some degree of trypanosomiasis challenge still existed and veterinary
services in the area were not easily accessible.
With increased duration of settlement, a progression from no cropping to
single or multiple cropping, but sole cropping to mixed cropping was observed
(Jabbar et al., 1992). Mohammed (1990) found an average farm size of 3.33
ha with the size slightly increasing with longer period of settlement. Land use
intensity was 67 percent (i.e., for two ha under crop, one ha was under fallow).
This land-use intensity is more than twice that of local Yoruba crop farmers,
made possible through intensive manuring by the Fulani farmers. Although
these animals generally graze away from the homestead during the day, night
kraaling helps to collect adequate manure for crop fields, which are generally
located around the homestead.
Mohammed (1990) also found that 24 percent of the crop area was devoted
to cassava, yam, maize, sorghum and other sole crops, and 76 percent to
various mixes of these crops (Table 3). Taking sole and mixed crops together,
93 percent of the crop area was devoted to different cereals, and 77 percent to
roots and tubers. The share of cereals in a local Yoruba farmers' land allocation
would be much less.
Jabbar et al. (1992) found that 23 percent of the sample farmers used crop
residues as feed, and most of them grazed residue in local crop farmers' fields.
Mohammed (1990) found that 96 percent of the farmers grazed maize
residues in local crop farmers' fields, 17 percent used sorghum residues mostly
from their own sources, 27 percent used cassava leaves mainly from their own
sources, and 100 used tree browse from the range. In another survey, de Jode
(1989) found that residues of wet-season maize were not always used due to


Vol. 4, No. 3, 1994






JABBAR


Table 3. Farm size and cropping pattern of agropastoralists, southwest Nigeria.
Average farm size ha 3.33
Percent under fallow 33.00
Percent crop area devoted to
Sole crops 24.05
Cassava 2.70
Yam 4.32
Maize 7.30
Sorghum 2.43
Others 7.30
Mixed crops 75.95
Maize Cassava 48.11
Maize Sorghum 3.24
Yam- Sorghum 22.16
Other mixes 2.43
All roots and tubers 77.29
All cereals and grains 92.97
Source: Mohammed, 1990.

the availability of better quality grass and the possibility of damage to second
season crops; however, second-season residues were more widely used because
available grass was then of lower quality and quantity. Mohammed (1990)
mentioned mixed cropping as a deterrent to using residues as feed. However,
with more land pressure, farmers may move to more sole cropping in order to
use residues as feed because most residue users said that even under present
conditions, crop residues were important to them as a source of feed.
de Jode (1989) found a calf mortality rate of 29.5 percent: 36.5 percent for
White Fulani, 24.2 percent for Fulani x Ndama crosses, and 2.25 percent for
Ndama. Worms and diarrhea were mentioned as reasons for calf mortality by
86 and 60 percent of farmers. Mohammed (1990) found calf and adult
mortality of 23.1 and 6.4 percent, respectively. Diarrhea, worms, and sand
eating were mentioned as major reasons for calf mortality by 94, 86, and 50
percent of the farmers, respectively. Calf mortalities were generally higher in
the wet season, whereas 78 percent of the adult cattle death occurred in the dry
season. Sand eating was the major reason for adult cattle death in both dry and
wet seasons. Incidence of sand eating in dry and wet seasons were mentioned
by 91 percent and 78 percent of farmers. Incidence of trypanosomiasis was
mentioned by only six percent farmers.
Mohammed (1990) found the following reproductive rates: age at first
calving: 47.1 months; pregnancy rate: 47.6 percent of cows; calving rate: 45.6
percent of cows; calving interval: 18.7 months.
de Jode (1989) reported similar rates for White Fulani, and more favorable
rates for Ndama and Fulani x Ndama crosses. Overall, these rates were
favorably comparable to those found in the subhumid zone of Nigeria (Nuru
and Dennis, 1976; Oyedipe et al., 1982; Egbunike, 1984; Pullin, 1984) and
in the humid tropics in general (Mugrewa, 1989).


Journal for Farming Systems Research-Extension






CROP-LIVESTOCK FARMING SYSTEMS


With longer periods of settlement, changes in diet and sources of income
were also observed. Among the agropastoralists in Mohammed's (1990)
sample, 50 percent produced adequate crops for home consumption, roughly
20 percent produced some surplus, and about 30 percent had a deficit. On the
other hand, about 20 percent of their daily milk output was consumed and 80
percent was processed into wara (soft cheese) for sale to local farmers and
traders who in turn sold them in the nearby towns. Propensity to sell tended
to increase with the longer period of settlement. Farmers reported increased
cash needs as the main reason for higher propensity to sell milk (Mohammed,
1990; Gherzi, 1991). Total offtake (percent of total cattle sold) rate was eight
percent, whereas effective offtake rate (percent of weaned calf crop sold) was
25 percent. These rates were higher than those generally found in the drier
zones.
Dairy consumption studies in southern Nigeria showed a good market
potential. Average daily per capital consumption among indigenous popula-
tions was 45gm LME which was comparable to west African averages, and
consumption was generally higher in the urban areas. Where consumers had
access to local products from peri-urban producers, up to 80 percent of the
households consumed such products and 45 percent of consumption came
from local sources (Jabbar and di Domenico, 1992).

POTENTIAL FOR DEVELOPMENT
In a major assessment of animal agriculture in sub-Saharan Africa, rapid
intensification of agriculture and development of crop-livestock farming has
been predicted on the basis of projected high population growth and high rate
of urbanization, and consequent changes in food production, marketing, and
consumption patterns. Sub-humid zones and the higher rainfall part of the
semi-arid zone of West Africa have been identified as the high-potential areas
for crop-livestock farming. Arid and humid zones have been identified as
second-priority areas (Winrock, 1992).
Several authors have postulated population pressure as the prime mover for
agricultural intensification and development of crop-livestock farming (Bose-
rup, 1965, 1981; Ruthenberg, 1980; Pingali et al., 1987; McIntire et al.,
1992). Market access, presence of cash crops, dominance of cereals in the
cropping pattern, and relative prices have been mentioned as additional factors
fostering crop-livestock interaction in specific situations (de Wilde, 1967;
McIntire et al., 1992). If these criteria are applied to judge the potential for
crop-livestock development in the humid zone, there are good reasons to be
optimistic. First, the humid zone is generally lightly settled, but in parts of
West Africa and around major cities, the densities are high and comparable to
those in the drier zones. Most major cities are located in this zone and they are
expected to grow rapidly with migration from within and outside the zone.
Population pressure in other zones may also push migration into rural areas in


Vol. 4, No. 3, 1994






JABBAR


the humid zone. Second, two major dimensions of market access are size and
links between producers and consumers. The humid zone is better endowed
than the other zones in this respect because of the presence of big cities well
connected with rural areas. Several times more animal (and some crop)
products are consumed in this zone than are produced (Table 1), and the size
of the market will expand rapidly, as indicated earlier. Third, although cereals
may not immediately become dominant in the cropping systems, their impor-
tance is increasing steadily. With the commercialization of agriculture and
increased income of consumers, the importance of cereal will increase further
and there will be a move toward producing cereals as sole crops.
These robust arguments may sound like mere assertions in view of the rather
rudimentary level of crop-livestock farming in the zone. However, if the
elements of this rudimentary development are superimposed on a long-term
view of the history of agricultural development around the world, the asser-
tions become realistic and plausible. Two thousand years ago, few or no crops
were grown in Western Europe; there were no domesticated grazing animals
and hardly any crop production in the entire American hemisphere, Australia
and New Zealand, or the tropical forests of Africa and Asia. Crop growing was
concentrated in tropical and semi-tropical river valleys such as Egypt's Nile,
Mesopotamia's Tigris-Euphrates, India's Indus-Ganges, South-East Asia's
Irrawady-Mekong, China's Yangtze-Yellow river. Part of Japan, Java, and
Mezzo-America are also ancient crop-growing locations. Outside this envi-
ronmental belt, crop growing in the Mediterranean littoral (Greece and
Southern Italy) was a remarkable exception. While Asian crop growers had
been using animal energy for crop production for a few centuries, the Graeco-
Roman crop-growing empire was expanding through slave-based production
methods. It appears that the relationships between people, crops, and livestock
has changed very little in Asia to this day, whereas the collapse of the Graeco-
Roman empire in the fifth century, after over-extending into pastoral North-
Western Europe, paved the way for a course leading to the world today where
crop and livestock production in the developed countries are highly specialized
and mechanized (Crotty, 1980).
The history of human civilization is marked with people's changing rela-
tionship with plants and animals. Methods of crop and livestock exploitation
evolved through ages, in varied and complex ways, to suit specific environ-
ments and meet varying needs of the societies. Institutions and customs
governing the exploitation of these resources also evolved to serve specific
needs of the society. For example, beef eating was banned, making it a religious
taboo by the Hindus in India when saving draft animal became crucial for
expanding crop production to less fertile areas. Centuries later, the Christian
church in Europe prohibited meat eating during Lent for economic consider-
ations (i.e., to dissuade peasants from slaughtering and eating their oxen
during springtime when they were most needed for draft power). At an
opportune moment this partial ban was lifted in Europe, but in Ethiopia meat


Journalfor Farming Systems Research-Extension






CROP-LIVESTOCK FARMING SYSTEMS


is still prohibited during Lenten fasting. In both India and Ethiopia, it is
unclear whether conditions have changed such that these bans could be lifted.
People have adapted animals and plants to their needs, and they themselves
have adapted to the plants grown and animals domesticated. For example,
adult lactose tolerance has been selected for among pastoralists, and the trait
has been diffused among other populations. On the other hand, East Asians
and people in the high tsetse infested areas in Africa have remained beyond the
reach of the pastoralists in the past and retain the normal characteristic of adult
lactose intolerance. However, with tsetse control and decline in the tsetse
challenge, cattle movement into humid zones has become possible. For
example, cattle population in the Central African Republic increased from
700,000 to two million in the 15 years following tsetse control (World Bank,
1987). As a consequence, people hitherto unaccustomed to dairy products
have developed tastes for some dairy products, though not necessarily for fresh
liquid milk. Throughout the coastal belt of West Africa, nomadic and seden-
tary pastoralists produce soft cheese and yoghurt for indigenous populations.
Through experience they found that many people did not like fresh milk,
probably because they could not digest it, but that cheese and yoghurt were
readily acceptable. Areas classified in the 1940s as non-cattle raising are now
inhabitated by cattle, and areas classified as non-milk producing are now
producing and consuming milk (Stenning, 1959). Trypanotolerant Muturu
are not milked by local farmers in southern Nigeria but rather by Fulani
owners/rearers of such cattle (Ferguson, 1967; Grandin, 1980; Mohammed,
1990). About 40 percent of Muturu owners in southeast Nigeria have
expressed interest in milking their cattle, if taught for commercial reasons (di
Domenico, 1989). More recently, with aid, relief, and the commercial import
of dairy products, hitherto non-consuming populations are consuming as
many dairy products as in conventional milk-producing areas (Jabbar and di
Domenico, 1992).
After the demise of Graeco-Roman Empire, crop production in Europe
relied on oxen rather than slaves. In the middle ages, oxen were replaced by
horses, a more powerful animal. Slavery was reintroduced in North America by
European colonizers more than a thousand years after its demise in Europe.
Defeated European pastoralists were turned into slaves by Roman crop
growers. Defeated African pastoralists and crop growers were turned into
slaves by colonial crop growers in North America, where slaves were later
replaced by horses. Asian crop growers still depend on cattle, buffalo, and
mules for power, sources they probably started using three thousand years ago.
In Europe, North America, and Australia, crop and livestock competed for
land, so the transition from horse to engine power released land for extra crop/
livestock production. In Asia, people competed for land and the competition
has increased over time. In much of Africa, a predominantly pastoral system is
evolving into a mixed farming system. The late arrival of cattle in the humid
zone does not necessarily mean that crop-livestock farming development in the


Vol. 4, No. 3, 1994





JABBAR


zone should follow the same pattern. Indeed, there is no natural sequence or
uniform evolutionary process of technical and system change that one can
follow as a blueprint or rule. Historically, crop farmers-being more intensive
and stationary land users-pushed pastoralists into marginal lands. But a time
came when neither could push the other further without affecting each others'
survival; the ensuing conflict was resolved through the adoption of mixed
farming. Crop farmers needed manure and traction to cultivate difficult and
less fertile soil, and pastoralists needed better feed for animals and themselves.
Mixed farming increased productivity of both crops and livestock, and en-
hanced the human supporting capacity of land.
Frequent conflicts between crop farmers and pastoralists in present day
humid zones are replications of such phenomena. These conflicts and other
external forces, such as frequent droughts in the drier areas and increasing
restriction on movement across political boundaries, are likely to hasten the
process of pastoral settlement and development ofcrop-livestock farming. The
settlement history of agropastoralists in southwestern Nigeria indicates a
continuous process of settlement, but there are clear peaks immediately after
the 1972/73 and 1984-86 droughts in the Sahel. Herd build-up and low
offtake are major hindrances to livestock productivity in Africa. The evidence
that herd size declines and offtake increases with settlement and mixed farming
are indications that mixed farming is the proper avenue for natural solution to
competing needs.

RESEARCH NEEDS AND PRIORITIES
Alongside tsetse control programs, research initiatives were taken in several
West African countries to evaluate the feasibility of cattle production in the
humid zones, particularly in the derived savannas. Notable among these are
Fashola Stock Farm, Mokwa Station and the University ofIbadan in Southwest
Nigeria; the Research and Animal Husbandry Centre, Avetonou, Togo;
Bunake in C6te d'Ivoire; Bamenda in northwestern Cameroon; and several
other university and research farms in the region. Two main foci and experi-
ences of these initiatives were:
Breeding: In some places, adaptability of pure breeds of temperate climates
and their crosses with locals were assessed for dairy production. Local breeds
were sometimes used as controls. In other places, trypanotolerant breeds were
used as the basic material mainly for beef production with traction as a
secondary objective. Despite their overall higher productivity as compared to
local breeds, pure temperate breeds and their crosses were found to be largely
unsuitable for dairy production in the humid zone due to high mortality,
intensive management requirements, and high costs.
Feeding: Intensive pasture and purchased concentrates were used as princi-
pal feeding strategies for both dairy and beef production. Beef production on
intensive pasture was marginally profitable and risky because of the too high


Journal for Farming Systems Research-Extension






CROP-LIVESTOCK FARMING SYSTEMS


cost of pasture development. Provision of uniform feed throughout the year
was a major problem causing dry season weight loss and affecting profitability
(Ruthenberg, 1974; Doppler, 1980).
The breeds, feed, and management strategies in these research initiatives
were meant for large-scale, specialized commercial producers. Small-scale
crop-livestock farmers were not the target for such research. For example,
grass-legume mixtures were studied for maintaining soil fertility but the role
of legumes in crop/pasture rotation has not been addressed. Without crops,
crop residues were also absent from these systems (Ahlgren and Adegbola,
1959; FAO, 1966).
Because crop-livestock farming systems are evolving and will presumably
become the dominant system in the derived savanna and sub-humid zones,
research and development strategies should be aimed at
supporting, accelerating and helping to direct the natural forces of intensifica-
tion of agriculture and the evolution and maturation of mixed crop-livestocks
farming systems that will make agriculture more productive and sustainable,
while at the same time improving the social and economic conditions of people
(Winrock 1992).
It is envisaged that such a strategy should adopt a resource-management
approach using land as the most critical resource.
In the humid zone, soils are low in organic matter content, fragile, and
easily degraded when the vegetative cover is lost. These problems have been
aggravated by agricultural intensification due to exponential growth of human
populations and the need for more agricultural land. Tsetse-fly control or
eradication programs have.further exacerbated these problems in two ways.
Tsetse-fly control through insecticide application has destroyed many non-
target insects and animals, thus creating ecological imbalances. Areas freed
from tsetse flies attracted increased human settlement and livestock, and this
has led to land-use intensification, bush burning, and damage to vegetation
(de Vos, 1975; Ormerod, 1976). Whether tsetse control programs should
have been pursued in more fragile areas is debatable, but without increased
human settlement and cultivation, tsetse-cleared areas would be taken over by
the flies again, and the purpose of eradication would be lost. So intensive land
use may be considered a necessary adjuncts for the consolidation of tsetse
eradication (Putt et al., 1980; Eicher and Baker, 1982).
There is no evidence that rain forests are cleared to expand livestock
production per se, but forest is regularly cleared to expand crop production.
Then cattle may move in, particularly trypanotolerant breeds in the beginning.
Unless constrained by external forces, this process is likely to expand. There-
fore, it is essential to develop controlled, integrated land-use systems for
sustainable crop and livestock development. In the absence of such systems,
tsetse-fly control and crop-cattle production in the more fragile areas should
be positively discouraged.


Vol. 4, No. 3, 1994






JABBAR


The main objective of research will be to modify existing resource manage-
ment practices and/or design new ones. Major strategic research issues in the
resource management area are, among others:
1. Determination of optimum livestock densities for crop-livestock system
sustainability and resource conservation under varying crop/pasture rotation
systems;
2. Determination of the environmental impact of animals as against the
impact of other sources such as climatic change, cultivation intensity, and bush
burning;
3. Role and potential of animal manure in system sustainability and
productivity;
4. Role of legumes (forage, crop, and tree) in soil fertility, weed manage-
ment, crop productivity, and feed supply;
5. Efficient use of crop residues in the system and residue quality and
quantity as criteria in crop breeding; and
6. Potential use of animals for traction.
Dry-season feeding is a major problem, and research on strategies for
production, conservation, and utilization of feeds to solve this problem is a
priority. Low milk production potential is a problem for the adoption of
intensive management. Breeding for milk production and disease resistance is
necessary to solve this problem. It is assumed that the system will be dairy
oriented because of its regular cash-generating potential, although beefwill be
an important component of the system. Research will be needed to develop
appropriate milk processing and preservation options for small- and medium-
scale producers. In the past, unfavorable policy environments kept the dairy
sector undeveloped in West Africa. Development of dairy-based crop-livestock
systems will depend on the existence of proper policy environments; thus
research on various policy issues will be needed.

REFERENCES

Ahlgren, G.H., andA. Adegbola. 1959. Development of grasslands in the Western Region
of Nigeria. Ibadan: Ministry ofAgriculture and Natural Resources, Western Region of
Nigeria.
Akinwumi, J.A., and J.E. Ikpi. 1985. Trypanotolerant cattle production in Southern
Nigeria. Report submitted to the International Livestock Centre for Africa, Humid
Zone Programme, Ibadan, Nigeria.
Bourn, D. 1983. Tsetse control, agricultural expansion, and environmental change in
Nigeria. Ph.D thesis, Christ Church College, Oxford.
Boserup, E. 1965. The conditions of agricultural growth: The economics of agrarian
change under population Pressure. New York: Aldine.
Boserup, E. 1981. Population and technological change: A study in long-term trends.
Chicago: University of Chicago Press.Crotty, R 1980. Cattle, economics and devel-
opment. Farnham: Commonwealth Agricultural Bureaux.


Journal for Farming Systems Research-Extension






CROP-LIVESTOCK FARMING SYSTEMS


de Jode, Andrew. 1989. A preliminary study of cattle production parameters in Oyo State,
Southwest Nigeria. M.S. Thesis. The Centre for Tropical Veterinary Medicine,
University of Edinburg, UK.
de Vos, A. 1975. Africa, the devastated continent? Man's impact on the ecology ofAfrica.
The Hague: Junk.
de Wilde, J.C., et al. 1967. Agricultural development in tropical Africa. Two volumes.
Baltimore: The Johns Hopkins University
di Domenico, C.M. 1989. Counting the cost: A study of dairy consumption, marketing
and production in southern Nigeria. Consultancy Report to ILCA, Humid Zone
Programme, Ibadan, Nigeria.
Doppler, W. 1980. The economics of pasture improvement and beef production in semi-
humid west Africa. Eschborn: GTZ.
Egbunike, G.N. 1984. Reproductive performance of Ndama (N), German Brown (GB)
and NGB cows in a humid tropical environment. Paper presented at the 10th
International Congress in Animal Reproduction and Artificial Insemination, Univer-
sity of Illinois, Urbana Champaign, 10-14 June.
Eicher, C.K, and D.C. Baker. 1982. Research on agricultural development in sub-Saharan
Africa: A critical survey. MSU International Paper No 1. East Lansing: Michigan State
University.
FAO. 1966. Agricultural development in Nigeria 1965-80. Rome: FAO.
Ferguson, W. 1967. Muturu cattle of Western Nigeria: Census, distribution, husbandry
and behaviour of village herds. Journal of the West African Science Association 12(1).
Ford, J. 1971. The role of the trypanosomiasis in African ecology: A study of the tsetse fly
problem. Oxford: Clarendon Press.
Fricke, W. 1979. Cattle husbandry in Nigeria: A study of its ecological conditions and
social-geographical differentiations. Heidelberger Geographische Arbeiten Heft 52.
Heidelberg: University of Heidelberg.
Gherzi, Edoardo. 1991. Production, processing and marketing of dairyproducts by settled
Fulani in Southwest Nigeria. LCA, Humid Zone Programme. Nigeria.
Grandin, B.E. 1980. Small cows, big money: Wealth and dwarf cattle production in
southwestern Nigeria. PhD dissertation, Stanford University.
Ikede, B.O., L. Reynolds, A.O. Ogunsanmi, M.K Fawumi, J.O Ekwuruke, and V.O.
Taiwo. 1987. The epizootiology of bovine trypanosomiasis in the derived savannah
zone of Nigeria A preliminary report. Paper presented at the 19th meeting of the
ISTRC, Lome, Togo, March 30 to April 3.
ILCA (International Livestock Centre for Africa). 1987. ILCA Annual Report 1986.
Addis Ababa, Ethiopia: ILCA.
IMF (International Monetary Fund). 1954. The economic development of Nigeria.
Washington, DC: International Bank for Reconstruction and Development.
Jabbar, M.A., L. Reynolds, and P.A. Francis. 1992. Sedenterization of cattle farmers in the
derived savannah region of southwest Nigeria: Results of a survey. African Livestock
Research.
Jahnke, Hans E. 1982. Livestock production systems and livestock development in tropical
Africa. Kiel: Kieler Wissenschaftsverlag Vauk.
McIntire, John, Daniel Bourzat, and Prabhu Pingali. 1992. Crop-livestock interactions in
sub-saharan Africa. Washington, DC: World Bank.
Mohammed, T.A. 1990. A study of peri-urban cattle pastoralism in the derived savanna
of Oyo State, Southwest Nigeria. ILCA Humid Zone Programme, Ibadan, Nigeria.


Vol. 4, No. 3, 1994






JABBAR


Mukasa-Mugerwa, E. 1989. A review of reproductive performance of female Bos Indicus
(Zebu) cattle, Monograph No. 6. Addis Ababa: ILCA.
Nuru, S., and S.M. Dennis, 1976. Abortion and reproductive performance of cattle in
northern Nigeria. Tropical Animal Health and Production 8: 213-219.
Nwanta, J.A. 1988. Survey of incidence of trypanosomes in settled Fulani cattle in
Anambra State. Unpublished DVM thesis, University of Nigeria, Nsukka.
Ockerman, H.W., and H.A. Agboola. 1991. A review of Nigerian livestock production
with a particular reference to the meat processing and utilization common among the
Yorubas of south-western states. World Review ofAnimal Production 26(2):69-79.
Ormerod, W.E. 1976. Ecological effect of control of African Trypanosomiasis. Science
815-821.
Oyedipe, E.O., V. Buvanendran, and L.O. Eduvie. 1982. Some factors affecting the
reproductive performance of White Fulani (Bunagi) cattle. Tropical Agriculture 59:
231-234.
Pingali, P., Y. Bigot, and H.P. Binswanger. 1987. Agricultural mechanization and the
evolution of farming systems in subsaharan Africa. Washington, DC: World Bank.
Pullin, N.B. 1979. Productivity ofWhite Fulani cattle on the Jos Plateau, Nigeria. I: Herd
Structure and reproductive performance. TropicalAnimal Health and Production 11:
231-238.
Putt, S.N.H., A.P.N. Shaw, T.W. Mathewman, D. Bourn, M. Underwood, A.D. James,
J. Hallam, and P.R. Ellis. 1980. The Social and economic implications of trypanoso-
miasis. Study No 25. Reading: University of Reading.
Reynolds, L., and B. Opasina. 1987. Trypanosomes and other blood parasites in slaughter
cattle at Ibadan, Nigeria. International Livestock Centre for Africa, Humid Zone
Programme, Ibadan, Mimeo.
RIM. 1988. Cattle, cultivation and human settlement in the derived savannah zone of
southwestern Nigeria. Report prepared for ILCA Humid Zone Programme, Ibadan.
Jersey, UK- Resource Inventory and Management.
Ruthenberg, H. 1974. Artificial pastures and their utilization in the southern guinea
savanna and the derived savanna of West Africa. Zeitschrift fur Auslandische Land-
wirtschaft 13(3): 216-231 and 13(4): 312-330.
Ruthenberg,H. 1980. Farming Systems in the tropics, 3rd Edn. Oxford: Clarendon press.
Shaw, T., and S. Colville. 1950. Report of Nigerian Livestock Mission. Colonial No. 266.
London: Her Majesty's Stationary office.
Stenning, D.J. 1959. Savanna nomads, London: Oxford University press.
Winrock. 1992. Assessment of animal agriculture in Sub-saharan Africa. Morrilton,
Arkansas: Winrock International.
World Bank. 1987. West African agricultural research review. Washington DC: World
Bank.


Journal for Farming Systems Research-Extension









Growth, Forage Content, and Biomass

Yield Response to On-Farm Fodder Tree
Species Trials in Nepal'

Madhav B. Karki2 and Michael A. Gold3



ABSTRACT
Over a three-year period, analysis of growth performance, nutritive value, and
lopping management was carried out on six native species located on farmers'
fields at three sites in Nepal. A concurrent study conducted on-station verified
on-farm results, showing that the best performing on-farm trial species were
also among the best performers on-station. A. lakoocha and Ficus semicordata
(Buchattam. ex Sm.) were found to perform best in terms of both height and
diameter growth. In general, farmers preferred as fodder trees the species with
high crude protein (CP) and organic matter (OM) contents both on their
fields and in their forests. The results indicated that native fodder tree species
of Nepal, if carefully selected and planted on the farmers' fields, have the
potential to improve the poor fodder stands commonly found throughout the
Middle Hills.

INTRODUCTION

Animal husbandry is a major economic activity, second only to crop farming
in the Middle Hills, and is the sole occupation of the people in the Mountains4
in Nepal. The most recent livestock population estimates are 6.5 million cattle,
3.2 million buffalo, 5.6 million goats, and 0.8 million sheep (Ministry of
Forests and Soil Conservation, 1988). The hilly regions are experiencing
extreme pressure from: (1) the high ratio of population to cultivated land,
approximately 16 persons per hectare (Wyatt-Smith, 1982); (2) high rate of
both human (2.6 percent per annum) and livestock (1.0 percent per annum)
population growth; (3) a very large population of ruminant animals; and (4)
shrinking supply of food, fodder, and fuelwood. These pressures have ex-

1 Paper presented at the Twelfth Annual Association for Farming Systems Research-ExtensionSym-
posium, Michigan State University, East Lansing, MI, USA. September 13-18, 1992.
2 Associate Dean, Institute of Forestry, PO Box 43, Pokhara, Nepal.
3 Assistant Professor, Department of Forestry, Michigan State University, East Lansing,
Michigan, 40024 USA.
4 Nepal's land area distribution is as follows: Terai or Southern Plains 17 percent; Middle Hills 76
percent; and Mountains 7 percent (CBS, 1984).






KARKIAND GOLD


pressed themselves in the form of rapid deforestation for firewood and fodder,
overgrazing, and clearing of steep slopes for cultivation. Resultant soil erosion
rates on poorly managed sloping terraces and degraded rangelands are estimat-
ed at 20-100 and 40-200 tons/ha/year, respectively (Carson, 1985). Contin-
uous decline in agricultural productivity is causing extreme poverty. This cycle
has led to serious environmental degradation as well as to deterioration in the
quality of peoples' lives.

Shortage of Animal Feed
It is estimated that an average household in Nepal's Middle Hills owns six
animal units and 10 multipurpose trees (Campbell and Bhattarai, 1984; Kafle
and Karki, 1988). The average cultivated land per household is one hectare
(ha). Currently only about 64 percent of the total feed requirements of the
livestock are met in the Middle Hills (Agricultural Projects Services Center,
1986). One of the major reasons for the shortage is the continuing shrinkage
in grazing and forest land. Good quality green fodder in adequate quantities
is generally available only from June to November (the wet season). Thus, the
major problem is the lack of green fodder during the long dry season
(December-May). Farmers have traditionally attempted to overcome this
problem by growing fodder trees on their farms and extracting tree fodder
from the forests. However, due to increasing population pressure, the latter
source is shrinking. As a solution to this problem, experts (Panday, 1982;
Wyatt-Smith, 1982; Karki, 1992) have recommended stall feeding of live-
stock, the production of more fodder on individual farm holdings, and
creation of more community forests composed of multipurpose tree species.
A study in western Nepal indicated that each family of five to six on an
average farm area of 1.25 ha required, with present agricultural and forestry
management practices, 3.5 ha of land for fodder, 0.3 to 0.6 ha for fuelwood,
and 0.4 ha for timber to sustain current levels of household activities (Wyatt-
Smith, 1982). However, the gross per capital availability of forest land is only
0.14 ha (Food and Agriculture Organization, 1979). Therefore, there is a need
to increase the number of trees on private land. But since there are biological
and physical limitations to planting significant numbers of additional trees,
particularly on private farms, a more viable alternative is to improve the
management of existing as well as future tree resources and increase their
productivity (Rusten and Gold, 1991).

Research Project
As a part of the Fodder Tree Management Project at the Institute of
Forestry, Nepal, fodder tree species (FTS)5 trials were established at four on-
station and three on-farm sites in three ecological zones of Nepal, i.e., the
Terai, the Inner Terai, and the Middle Hills (Table 1, Table 2; Figure 1).
5 A tree grown deliberately or kept and managed primarily for fodder production in agroforestry
systems belonging to mainly subsistence farmers.


Journal for Farming Systems Research-Extension






FODDER TREE SPECIES TRIALS


Table 1 Sites and Fodder Tree Species Included in On-Farm Trials.
SPP. SITE
HETAUDA TANAHU POKHARA
A.lakoocha x -
B.variegata x x x
F.semicordata x x x
F.auriculata x x -
G.pinnata x
L.monopetala x


Table 2 Selected Fodder Tree Species: On-station Research Trials.
S.NO. COMMON NAME SCIENTIFIC NAME SITE CODE
1 Khanayo Ficus semicordata K, H, R, P
2 Pakhuri Ficusglaberrima K, H, P
3 Nemaro Ficus aunculata K,H,R,P
4 Badahar Artocarpus lakoocha K, H, R, P
5 Koiralo Bauhinia variegata K, H, R, P
6 Kutmiro Litsea monopetala K, H, R, P
Site Code: K- Karmaiya; H Hctauda; R Rampur; P Pokhara

Of a total often species tested over all sites, six important species, included
in the on-farm trials, are discussed in detail (Table 1). The sites selected
represented the upland farm (bari) land found in the Middle Hills and the
Inner Terai regions of Nepal. Marginal lands including noncultivated inclu-
sions are also commonly found growing FTS in the Middle Hills, which the
selected sites also covered.
Studies of growth performance, lopping management, and nutrient analysis
were carried out on native species located on farmers' fields at three sites. On-
farm trees were selected from locations near the research-station trials at
Hetauda, Tanahu, and Pokhara (Figure 1). Most of the farms in Nepal are less
than one hectare in size. Within and between the individual farms, site
variability is high due to the interaction of factors including altitude, aspect,
soil type, soil depth, and management practices adopted by the respective
farmers. Equally important is the type of species grown to fulfill the specific
needs of the farmers concerned. It has been shown (Rusten, 1989) that an
average farmer in the Middle Hills requires a range of species in order to meet
a variety of needs fulfilled by tree products for a large part of the year at a
minimum risk. Thus, the Midhill farmers grow a large number of species of
fodder trees. On-farm FTS trials must take these factors into consideration.
The diversity of tree species extant on private farms is a valuable resource for
yield, production, and management studies, provided that appropriate re-
search design and inventory can be established.


Vol. 4, No. 3, 1994







Figure 1.






FODDER TREE SPECIES TRIALS


On-farm trees of known age, species, owner, and site were leased from
farmers for a period of five years. Treatment variables and levels were deter-
mined based on discussions held with the farmers. Major variables measured
were: phenology, height, diameter at breast height (dbh), tree age, regener-
ation method, age of first lopping, lopping intensity, average yield per tree,
preferred fodder species, and foliar nutrient contents. These data were used to
complement the results obtained from the on-station FTS evaluation trials
(Karki, 1992).
Nutrient analysis was carried out on fodder samples collected from the trial
sites. The purpose was to evaluate site variations in nutrient contents of
recommended fodder tree species.

LITERATURE REVIEW

The objective in testing multipurpose species is to assess their adaptability to
new sites in terms of vigor and productivity. Response to management
interventions is the most important variable in ensuring a sustained supply of
products such as animal fodder (Stewart, 1990). Species choice also depends
upon nutritive values. The chemical composition of fodder foliage is known to
vary widely not only by species, but also by location, age, method of cultiva-
tion, climatic conditions, soils, and season of the year (Ivory, 1989). This is
particularly true of protein and mineral contents, which reflect the qualities of
fodder.
The traditional approach to such research is to design on-station trials and
make inferences. However, research-station experiments inevitably have an
element of artificiality due to the high level of control and the large amount
of resources that can be applied. On-farm research generally suffers from the
constraints of weaker statistical analysis due to fewer experimental plots, less
choice in plot selection, and less control over the trials (Mead, 1991).
However, research design methods and statistical techniques are being devel-
oped to meet the special requirements of on-farm agroforestry research
(Hocking, 1991; MacDicken et al., 1991; Mead, 1991).

STUDY OBJECTIVES

The overall objective of the trial was to test on-farm fodder tree species in order
to verify the species evaluation results from on-station trials. The specific study
objectives included: (1) an evaluation of species performance in terms of
height and diameter growth, foliage and wood biomass yield, and total dry
matter yield across the three sites; (2) an evaluation of the effect of different
lopping intensity on the final foliage yield of the selected fodder trees; and (3)
comparison and contrast of the on-station and on-farm results, especially
results related to growth performance, nutrient content, and lopping treat-
ment.


Vol. 4, No. 3, 1994






KARKI AND GOLD


METHODOLOGY

A single tree or a group of three trees were treated as an experimental unit and
three to nine such trees as three replicated experimental units depending on
the availability of suitable number of trees. A limited number of trees was
selected for two reasons: (1) undisturbed trees were difficult to find; and (2)
farmers were unwilling to lease more than a few trees for a three-year period.
These three to nine trees formed a replicated block at each site for a particular
species. This design was considered appropriate since trees were found
growing far apart in a condition similar to a randomized situation. Trees of
approximately uniform age and size were selected.
Sample trees were marked prior to felling. Heights, diameters, and crown
dimensions were measured. Each sample tree was cut down and measurements
were recorded as key variables: tree height (length), foliage weight, biomass
dry weight, and a foliage nutrient analysis.
Oven-dried foliage samples were ground and were submitted for nutritional
analysis in the laboratories of the Central Livestock Development Center,
Kathmandu. The samples were analyzed for total nitrogen (N), crude protein
(CP), total ash (TA), organic matter (OM), cellulose, hemicellulose, and
lignin contents using the procedure of Goering and Van Soest (1970). Each
value estimated was the average of three determinations.
The species data were tested through linear parameters. Analysis of variance
(ANOVA) using species, sites, and interactions as levels were carried out for
each site, and F statistics were computed to test hypotheses. Pairwise mean
comparison was used to compare the specific means. Multiple comparison
procedures were used to compare the means of the six species at all sites. In
order to compare and contrast the six species included in the trial, multiple
comparison of the means was carried out using the Waller-Duncan Bayes LSD
(BLSD) method (Peterson, 1985).6 Factorial ANOVA was used to partition
the growth variations among sites, blocks, and species (Zar, 1984).

RESULTS

Height, dbh, crown diameter growth, and crown height growth of four
common FTS trees-F. semicordata, Bauhinia variegata, Artocarpus lakoocha,
and Ficus auriculata-were measured for three years. The average age of these

6 In this method, the standard error of a difference between means is computed and the outcome
is multiplied by an appropriate minimum average risk (t tB) to obtain the BLSD value. Then
a risk ratio comparable to the significance levels of other procedures is selected. Duncan (1955)
showed that a k ratio of 100 is equivalent to an a of 0.05, while a k ratio of 500 is equivalent
to an a of 0.01. The following analysis has been obtained using BLSD method at k = 100.
Peterson (1985) recommended the use of BLSD for two reasons: a) it gives a single value which
is large when the sample F indicates that the means are homogeneous, and small when the
means appear to be heterogeneous; and b) its power to detect real differences does not depend
on the number of means being compared. BLSD was used to compare different combinations
of means.


Journalfor Farming Systems Research-Extension






FODDER TREE SPECIES TRIALS


trees was 10 years. The evaluation criteria were to compare the net increment
in height and dbh growth during the three-year period (Table 3). The mean
height and dbh growth differences among common on-farm FTS are given in
Table 4. To compare growth rates, an ANOVAwas carried out on mean height
and diameter increments was carried out (Tables 5 and 6). Since the species
were significantly different both in height and diameter growth, multiple
comparisons of the mean values were carried out using the BLSD method. In
terms of height growth, A. lakoocha was significantly different from others. F.
semicordata, B. variegata, and L. monopetala were not significantly different
from each other. Based on the diameter increment values, A. lakoocha was
significantly different from F. semicordata. F. semicordata and F. auriculata
were also significantly different from the remaining three FTS (Table 4).
To evaluate the effect of different lopping regimes practiced by the villagers,
24, 12, six and six randomly selected on-farm trees (F. semicordata, B.
variegata, F. auriculata, and A. lakoocha) respectively were given four treat-
ments: 25 percent lopping, 50 percent lopping, 75 percent lopping, and 100
percent lopping based on the crown length. At the end of the third year, all the
trees were fully lopped and dry foliage and branch weight were compared.
The results for F. semicordata (Table 7) indicated no significant differences
in the dry weight of the four treatments. The effect of the three different sites
-Hetauda, Tanahu, and Pokhara-was also not significant for total foliage
weight. The other FTS were not analyzed due to the small number of
observations per species.


Table 3. Mean Height, Dbh, and Three-Year Changes in Crown Diameter and-
Crown Length of On-Farm Trees at Age Ten.
SITE HEIGHT(M) DBH(CM) 3 YR. HT DIFF. 3 YR. DBH DIFF
MEAN(SD)
Hetauda 7.3(1.3) 23.0(4.9) 2.0(0.41 ) 4.4(3.06)
Tanahu 9.8(1.3) 26.9(4.8) 1.3(0.28) 1.9(0.58)
Pokhara 6.5(1.1 ) 1.9(2.2) 1.5(0.38) 2.1(0.66)
SD = standard deviation; DBH = diameter at breast height, 1.4 meters above the ground.

Table 4. Height, Dbh, and Mean Growth Differences' of On-Farm FTS.
SPECIES HT87 HT90 MEAN DBH87 DBH90 MEAN
SAMPLE (M) (M) DIFF.(M) (CM) (CM) DIFF.(CM ) NO.
A.lakoocha 10.1 12.7 2.6a 25.7 34.8 9.2a 12
F. semicordata 8.1 10.0 1.9b 23.1 27.6 4.5b 24
G.pinnata 8.2 9.8 1.6c 19.0 21.5 2.5c 6
L.monopetala 10.0 11.7 1.7bc 27.3 29.2 1.9cd 12
B.variegata 7.7 9.4 1.7bc 19.3 21.2 1.9cd 12
F.auriculata 7.9 9.3 1.4c 23.1 26.9 3.8b 6
Multiple mean comparisons were carried out via the Waller-Duncan Bayes LSD (BLSD)
method.


Vol. 4, No. 3, 1994






KARKI AND GOLD


Collection of foliage samples from all the study sites and their chemical
analysis indicated high crude protein (CP) levels in all species with the
exception ofF. semicordata (Table 8). These FTS were reported by the farmers
as highly preferred trees. The farmers' preference was based on the milk
yielding capacity of the fodder leaves when fed to the lactating buffalos.
Presentation of in vitro proximate analysis data is not intended to be an
accurate indicator of the actual nutritive value to an animal in a feeding trial.
However, FTS that were preferred by the farmers were found to contain
relatively higher CP and lower cellulose contents. The preferred species also
had high DM and low lignin contents (with the exception of L. monopetala),
other indicators of good quality feed.
Nutritive values are known to vary according to the phenology or stage of
leaf maturity of a particular FTS. Detailed farm-level indigenous knowledge of
individual FTS nutritive values and their relation to season, phenology, and
stage of leaf maturity has been documented (Rusten, 1989). FTS nutritive
values also vary by species and site. In this study, the samples were all collected
during the same season. However, due to altitudinal differences of the three
sites, the growth stages were not similar. None of the nutrient variables was
significantly different among sites (Table 9). Ficus hispida had the highest
percentage of CP content at all three sites, followed by A. lakoocha. Dry matter
(DM) and organic matter (OM) were not significantly different (p = 0.05)
among species. Crude protein (CP), lignin and cellulose contents were
significantly different among species.



Table 5. Analysis of Variance-Mean Height Increment (m) (On-Farm).
SOURCE SUM OF SQUARES DF MEAN SQUARE F-RATIO
Spp. 2.271 3 0.757 16.99**
Site 3.262 2 1.631 36.62**
Spp.*Site 0.996 6 0.166 3.73**
Error 0.463 24 0.045 -


Table 6. Analysis of Variance-Mean Dbh Increments (cm) (On-Farm).
SOURCE SUM OF SQUARES DF MEAN SQUARE F-RATIO
Spp. 47.885 3 23.943 185.15**
Site 52.841 2 17.614 136.21**
Spp.*Site 5.777 6 9.296 71.89**
Error 3.104 24 0.129 -
NS = non-significant; **- significant at p=.01


Journal for Farming Systems Research-Extension






FODDER TREE SPECIES TRIALS


Table 7. Analysis of Variance-Mean Foliage Dry Weight Based on Lopping
Intensity, F. semicordata (On-farm).
SOURCE SUM OF SQUARES DF MEAN SQUARE F-RATIO
Site 81.70 2 40.85 1.17NS
Lopping treatment 34.24 3 11.41 0.33NS
Site*Treatment 95.43 6 15.91 0.46NS
Error 418.33 12 34.86 -


Table 8. Nutrient Contents (Mean Values) of Common On-Farm Fodder
Tree Species.
SPECIES % DM % OM % CP % LIGNIN % CELLULOSE
A. lakoocha 29.0 91.4 20.3 21.6 28.3
B. varitgata 30.9 92.1 20.2 14.1 22.3
F. hispida 27.7 87.5 21.2 15.0 23.3
F. semicordata 30.4 92.2 14.0 26.3 25.3
G. pinnata 29.7 93.1 20.4 16.1 20.7
L. monopetala 29.0 93.0 19.2 35.6 20.3
DM dry matter; OM organic matter; CP crude protein


Table 9. Analysis ofVariance-On-Farm FTS Nutrient Contents.
SOURCE SUM OF SQUARES DF MEAN SQUARE F-RATIO
VARIABLE CRUDE PROTEIN
Site 1.66 2 0.83 0.64NS
Spp. 05.16 6 17.53 13.66**
Error 15.51 12 1.29
VARIABLE LIGNIN
Site 42.54 2 21.27 0.82NS
Spp. 1 082.79 6 180.46 6.92**
Error 313.012 12 26.08
VARIABLE DRY MATTER
Site 138.32 2 69.160 2.571NS
Spp. 21.67 6 3.611 0.134NS
Error 322.747 12 26.896
VARIABLE ORGANIC MATTER
Site 2.19 2 1.10 0.217NS
Spp. 66.93 6 11.15 2.210NS
Error 60.57 12 5.05
VARIABLE CELLULOSE
Site 105.32 2 52.66 3.36NS
Spp. 352.43 6 58.74 3.75*
Error 188.05 12 15.67
NS nonsignificant; *- significant at p=0.05; **- highly significant at p=0 01


Vol. 4, No. 3, 1994






KARKI AND GOLD


DISCUSSION AND IMPLICATIONS

The research trial results have presented an approach for combining on-farm and
onstation FTS evaluation methods. Although only six species were discussed in the
on-farm trials due to the unavailability of suitable trees, the results have provided
important complementary information to interpret on-station results. This study
emphasizes the following points:
The conventional practice of selecting FTS based solely on their geographic
location and native habitat is not necessary
Widely adaptive high yielding species should be identified and tested outside
their area of local origin
FTS selection should be based on farmer preferences as well as on silvicultural
considerations and foliar nutritional properties. For example, A. lakoocha is very
difficult to propagate, and while this species is highly regarded as a fodder tree,
there are equivalent species available which are easier to propagate and more
widely adapted to a range of locations
Farmers should be made aware of trade-offs between nutritional values and
total biomass. A particular FTS rich in crude protein (CP) may yield so little foliage
that the ultimate nutrition gained will be quite small whereas nutritively poor
species may be able, in total, to provide higher fodder protein yield. A case in point
is the comparison between F. semicordata and A. lakoocha. Although the former
has a higher percent age of CP, the total digestive nutrient derived from a single
tree of the latter species is quite high
Traditional methods of screening FTS soley based on height and dbh are not
suitable
Tree and land tenure, gender, and other socio-economic factors are also
important in selecting FTS and prescribing their management.
Traditional research methods should be complemented by other suitable
methods, based on: (1) fodder (green or dry) weight; (2) crown diameter; (3)
fodder to wood ratio; and (4) percent CP content. Ifa single criterion is to be used,
crown volume or dry fodder weight is recommended.
Ongoing government and private forestry programs are not expected to solve
Nepal's current fodder and fuelwood crisis. The major problem is that there is not
enough land to plant trees even if there were resources to carry out massive
aforestation programs. Under such tremendous pressures, it is evident that the
research activities must be carried outwithin the constraints of the present farming
systems. Selection of high-yielding, multipurpose tree species, of which the FTS
are an important part, should be a high research priority. An ideal tree would grow
well on poor sites; would be available to, known and desired by farmers and their
animals; and would yield much fodder and fuel.
This study has the following implications with regard to future FTS research
and development:
SF. semicordata, B. variegata, and Litsea monopetala are potential FTS for the
Terai and Inner Terai regions of Nepal, where the fodder situation is critical and
fodder trees are not cultivated as widely as in the Middle Hills.
Journal for Farming Systems Research-Extension






FODDER TREE SPECIES TRIALS


Variation in individual growth performance by the FTS due to site factors
is not significant. Nitrogen is the most commonly found limiting element,
followed by low pH. Application of organic manure and lime, both of which
are easily available in Nepal, can remedy these limitations
*Some of the FTS currently growing in farmers' fields are not the ideal trees
suitable for the farmers or their animals. Broadly speaking, farmers prefer
medium trees with dense crowns. Gender differences exist in choice of FTS and
must be taken into account (Gold and Rusten, 1993). Animals require trees
which produce succulent, protein-rich foliage during the dry season. Many
species currently found on farms are the residual regeneration of the natural
vegetation, which farmers allowed to grow as fodder trees, apparently without
selecting for fodder value. Therefore, there is a need for fodder stand
improvement within the framework of whole farm-system development and
management
Finally, as with all field research (especially in developing countries), this
study has been carried out within numerous limitations. They include: (1)
difficulty in finding ideal trial sites with uniform soil conditions; (2) difficulty
in establishing stringent control of exogenous factors on sites; (3) lack of
standard on-farm research methodology; (4) lack of trained research staff; and
(5) insufficient time and funds. Therefore, wide application of the reported
results should be done with caution. Further verification of results should be
carried out. However, for general extension and planning purposes the results
are expected to be useful.

CONCLUSION

This paper has presented a methodology for evaluating a large number of
fodder tree species in several locations. This is the beginning of one of the first
such long-term systematic on-farm studies in Nepal. Results indicate that
conventional thinking for example, that all FTS growing in the Middle Hills
will not do well in the plains (Terai), needs reexamination. This has 14
important practical implications in that government agencies, driven by this
belief, are trying to plant exotic fodder trees in the Terai farming systems.
However, the vast number of FTS found in the Middle Hills, more than 195
in Kathmandu valley alone (Bajracharya etal., 1985), can provide the necessary
fodder germplasm for the Terai. What is needed is more thorough testing and
development of germplasm for the diverse range of farming systems found in
the country.

ACKNOWLEDGMENTS

The authors acknowledge the assistance of USAID and Winrock International
under the Forestry/Fuelwood Research and Development (F/FRED) Doc-
toral Fellowship program, and the Institute of Forestry, Pokhara.


Vol. 4, No. 3, 1994






KARKI AND GOLD


REFERENCES

Agricultural Projects Services Center. 1986. Perspective Land Use Plan (1986-2005).
Ramshah Path, Kathmandu, Nepal. 140 pages and annexes.
Bajracharya, D., T.B. Bhattarai, M.R. Dhakal and T.N. Mandal. 1985. Nutritive
Values of Fodder Plants from Nepal. Journal of Economic Botany, Botany Instruc-
tion Committee, Tribhuvan University, Kathmandu, Nepal.
Campbell, G.J., and T.N. Bhattarai. 1984. People and Forests in Hill Nepal. Prelimi-
nary Presentation of Findings of Community Forestry Household and Ward Leader
Survey. Project Paper No. 10. HMG/UNDP/FAO, CFDP, Kathmandu, Nepal.
Carson, B. 1985. Erosion and Sedimentation Processes in the Nepalese Himalaya.
International Center for Integrated Mountain Development (ICIMOD), Paper
No. 1, Kathmandu, Nepal, 39 pages.
Central Bureau of Statistics. 1984. Statistical Pocket Book-Nepal-Kathmandu.
Duncan, D.B. 1955. Multiple Range and Multiple F Tests. Biometrics 11:1-42.
Food and Agriculture Organization of the United Nations. 1979. Forestry for Rural
Communities. Forestry Department. Rome, Italy. 56 pages.
Goering, H.K. and Van Soest. 1970. Forage Fiber Analysis. Agriculture Handbook No.
379. Agricultural Research Service, United States Department of Agriculture,
Washington, DC.
Gold, M.A. and E.P. Rusten. 1993. Learning From the Farmers: A Case Study from
Nepal. Agroforestry Today. 5(1):10-12.
Hocking, D. 1991. Orienting agroforestry research toward social objectives. Pages 21-
37 in M.E. Avery, M.G.R. Cannell and C. Ong, eds., Biophysical Research for Asian
Agroforestry. Winrock International and South Asia Books, USA.
Ivory, D.A. 1989. Major Characteristics, Agronomic Features, Nutritional Values of
Shrubs and Tree Fodders. In C. Devendra, ed. Shrubs and tree fodders for farm
animals. Proceedings of a workshop in Denpasar, Indonesia. Ottawa, Canada:
International Development Research Center.
Kafle, N.R. and M.B. Karki. 1988. Fodder tree management practices in Tanahu
district. Unpublished project paper. Institute of Forestry, Pokhara, Nepal. 32 pp.
Karki, M.B. 1992. Improved Fodder Tree Management in the Agroforestry Systems
of Central and Western Nepal. Ph.D. Dissertation, Michigan State University, E.
Lansing, MI USA. 244 pages.
MacDicken, K.G., G.V. Wolf and C.B. Briscoe, eds. 1991. Standard Research Methods
for Multipurpose Trees and Shrubs. Winrock International. 92 pages.
Mead, R. 1991. Designing Experiments for Agroforestry Research. Pages 3-20 in M.E.
Avery, M.G.R. Cannell, and C. Ong, eds., Biophysical Research for Asian Agrofor-
estry. Winrock International and South Asia Books, USA.
Ministry of Forestry and Soil Conservation.1988. Master Plan for Forestry Project,
Babar Mahal, Kathmandu, Nepal. 168 pp.
Panday, K.K. 1982. Fodder trees and tree fodder in Nepal. Swiss Development
Corporation, Berne and Swiss Federal Institute of Forestry Research, Birmensdorf,
Switzerland. 107 pages.
Peterson, RG. 1985. Design and Analysis ofExperiments. New York and Basel. Marcel
Dekker Inc. 429 pages.


Journal for Farming Systems Research-Extension


72






FODDER TREE SPECIES TRIALS


Rusten, E. 1989. An Investigation of an Indigenous Knowledge System and Manage-
ment Practices of Tree Fodder Resources in the Middle Hills of Central Nepal.
Ph.D. Dissertation, Michigan State University, East Lansing. 282 pages.
Rusten, E.P. and M.A. Gold. 1991. Understanding an Indigenous Knowledge System
for Tree Fodder via a Multi-method On-farm Research Approach. Agroforestry
Systems 15:139-165.
Stewart, J. 1990. International Trial of Central American Dry Zone Hardwood
Species. -Evaluation Manual, Oxford, UK: Oxford Forestry Institute, 55 pages.
Wyatt-Smith, J. 1982. The Agricultural System in the Hills of Nepal: The Ratio of
Agricultural to Forest Land and the Problem ofAnimalFodder. APROSC Occasion-
al Papers 1. Agricultural Projects Services Center, Kathmandu, Nepal. 17 pages.
Zar, J.H. 1984. BiostatisticalAnalysis. Prentice-Hall, Inc., Englewood Cliffs, N.J. 718
pages.


Vol. 4, No. 3, 1994








Survival and Sustainability in the

Midwestern Hills of Nepal 1

Ashok K. Vaidya and David Gibbon 2


INTRODUCTION

Sustainability is difficult to achieve in the mountain agroecosystem context.
There has been a clear degradation of the ecological integrity of the system in
the hills of Nepal, which Jodha (1989) characterizes as a fragile, marginal,
diverse niche in his discussion of the adaptation mechanisms of mountain
people.
The cause of the problem lies in rapid population growth, which has
changed the pattern of usage of natural resources. Hill farming is characterized
by scarcity of arable land, diversified farming, few employment opportunities,
marketing difficulties, and weak institutional support. There is also a lack of
environmental awareness by rural actors.
The array of forces that are leading toward environmental degradation are
deforestation, soil erosion, lack of fodder and fuel wood, decreasing livestock
population, reduction in organic matter collection, and a decline in crop
productivity, with an increase in subfertile areas under lengthening fallow
periods. An inventory of such objectively verifiable changes could be listed,
which in total have directly affected food and income provision and which the
hill people call into question their survival.
This article discusses the development of approaches in the context of
highly interactive and interlinked farming systems in the hills by means of a
case study. The role of microsocial structures in this process and the sustain-
ability issues of the approach itself are also discussed. It shows how Lumle
Regional Agricultural Research Centre (LRARC) has evolved some techniques
of approaching these problems through the experience of providing extension
and research services for some 15,000 farming households over the last 15
years.
Lumle is a British-funded project established in 1968 as a resettlement
program to train retired British Gurkhas in improved farming. In 1975 it was
developed into an agricultural extension, research, and training center serving
the people of its Extension Command Area (ECA) in 25 Village Development
Committees (the smallest political unit in Nepal) in parts of the Hill Districts

1 Paper presented at the Eleventh Annual Association for Farming Systems Research-Extension
Symposium, Michigan State University, East Lansing, Michigan, USA, October 5-10, 1991.
2 Lumle Regional Agricultural Research Centre, c/o BTCO, PO Box 106, Kathmandu, Nepal;
and School of Development Studies, University of East Anglia, NR4 7TJ, UK.






VAIDYA AND GIBBON


of Parbat, Myagdi, and Kaski within the Western Development Region.
Through being given additional responsibilities, and by working in conjunc-
tion with other Agricultural Extension and Research Projects, the center has
progressively integrated into the Government's Research System, and now has
agricultural research responsibilities for eleven Hill Districts.
This background has enabled scientist and extensionists to develop a
comprehensive understanding of the needs, aspirations, potentials, and re-
source constraints of the clients in the command area. These experiences and
responsibilities have demonstrated that an agricultural extension system based
upon single discipline approaches have only a limited impact. A model has
therefore been developed in which cross-disciplinary and farmer participatory
on-farm research has been institutionalized. Of all trials and studies carried out
by Lumle, between 65 and 75 percent are carried out on farmers' fields, using
farmers' techniques and labour (Pound et al.,1990).

RESEARCH STRATEGY

Agricultural research is carried out at three types of location: (1) On-Station,
(2) Off-Station Research (OSR) Sites, and (3) Research Outreach (RO) Sites.
On-station research is essentially basic in nature. The developed technolo-
gies are tested at OSR sites in specific research domains. OSRsites are also used
for testing technologies and problems related to the site and its domain.
Technologies generated or verified from the OSR sites are then extended to
RO sites for further verification and dissemination under farmer conditions. At
present, there are six OSR sites and 29 RO sites in operation in the Research
Command Area (RCA) and ECA.

INTERDEPENDENCE

Within the Hill Districts of Nepal, variation in altitude of a few hundred
meters, either up or down, means changes in climate, soil type, cropping
pattern, and ethnic group. This variation in topography and socioeconomic
genre has given rise to numerous microproduction localities, and this charac-
teristic combined with inaccessibility and fragility, results in highly interactive
subsistence and semisubsistence farming systems, dependent upon crops,
livestock, and forest resources (Figure 1). It is apparent from the above that
a complete understanding of the interrelationships between these various
component parts with respect to socioeconomic factors, all ofwhich influence
the pattern of use of the resource base of a system, is needed. The persistently
negative changes previously detailed have contributed to a series of problems
in the hill farming systems. However, no problem can be solved in isolation and
a global approach must be adopted. A multifaceted problem requires a
multifaceted response (Khadka and Gibbon, 1988).


Journal for Farming Systems Research-Extension






SURVIVAL AND SUSTAINABILITY


Lumle has developed a research methodology based on two farming systems
approaches, and has been an integral part of this institutional development
(Figure 2). These are now described.


Figure 1. Conceptual model of Nepal hill farm production system.
Source: Thapa (1987).


Samuhik Bhraman
The literal meaning of Samuhik Bhraman is 'traveling together,' and it is a
form of rapid rural appraisal developed in response to the Nepalese field
conditions. It involves staff from all the technical sections at the LRARC,
representatives of the District Offices of Agriculture, Livestock and Forestry,
National Commodity scientists and others as appropriate.
This kind of multidisciplinary trek enables researchers to understand and
interactwith the real problems facing the farming community, and to establish
priorities in research. It also promotes interaction and linkages with the
individuals of other organizations on the same topic and provides adequate
time for informal discussion and interaction to take place between interdisci-
plinary staff during the seven to ten day trek.


Vol. 4, No. 3, 1994






VAIDYA AND GIBBON


Research "Thrusts"
These are cross-disciplinary teams that plan and carry out research programs
into identified field problems (Joshi et al., 1990). After a series of Samuhik
Bhramans, it was realized that research into the following major problems was
essential to improve sustainability in the hill farming systems.
Income-Generation Thrust. To identify areas of low cost technology within
the farming systems that generate income. (Low income is often cited as the
reason for the inability to take up new technology).
Fodder Thrust. To improve fodder availability by identifying suitable fodder
tree species and improving animal nutrition. Due to deforestation, degrading
of natural resources, and seasonal variation of fodder availability, the farmers
themselves have emphasized the need to grow fodder trees and grasses.
Soil Fertility Thrust. To investigate cost effective and sustainable methods
of maintaining and upgrading soil fertility with optimum emphasis on the use
of local and biological resources.
Apart from the LRARC itself, multidisciplinarity is reflected in the various
mechanisms developed to address the situations arising from the complex
nature of the farming systems.

Representative Off-station Research (OSR) Site
To implement these approaches, especially in the diverse socioeconomic
and agroecological context of the command areas, a selection of representative
sites has proved to be successful and cost effective in generating appropriate
technology. It has facilitated good field management, minimized area of
supervision, and above all, provided researchers with the rare opportunity of
living with a farming community in order to gain a better understanding of the
problems as perceived by the farmers themselves.
Delineation of'Recommendation Domains' is carried out in order to select
a representative site to serve a defined geographic region with similar agrocli-
matic and socioeconomic conditions, for which generated technology could
be verified. Each OSR site represents a specified number of Farming Systems
Research and Extension (FSRE) villages for verification purposes.
To date, Lumle has opened eight OSRsites. OSRsites are managed without
permanent infrastructure, and are staffed by a graduate site coordinator, junior
technical staff and locally employed field recorders. This approach offers great
flexibility when the time comes to withdraw again after completing the initial
objectives.
The OSR site has proved instrumental in achieving the LRARC's technical
objectives to generate, verify, and disseminate appropriate and sustainable
technologies, and to explore the potential oflocal resources. It operates farmer
participatory adaptive research in the farmers' field.


Journal for Farming Systems Research-Extension






SURVIVAL AND SUSTAINABILITY


Technical Meetings
Technical meetings are held every month and attended by most of the
technical staff, including socioeconomists, extensionists, and managerial offi-
cials. Site coordinators and field staff from OSR arrive at the LRARC from as
far away as two days walk. The meeting is chaired by the project director or
research advisor. It is marked by a session in which a monthly report of each
site is presented. Practical problems in the field, on-going and completed
trials, and programs of all the disciplines, including technical policy matters,
are discussed. Farmers' views on the trials and their ranking of the treatments
are given close consideration. All trial protocols are presented in this meeting
and have to be thoroughly discussed in an FSR perspective and passed before
implementation. This encourages quality proposals and avoids discipline
biases. Similarly, technical papers for publication also have to be approved in
the meeting.
Publication of Prabidhi Sangalo (a technical newsletter for extension
agents) and various types of papers, FSR review meetings and farmer work-
shops are also crucial components that have contributed in reinforcing newly
developed approaches.

OSR SITE, TAPU: A CASE STUDY

Tapu village was chosen as the first Lumle OSRsite during a Samuhik Bhraman
in September 1984. The site has since been used as a representative site to carry
out multidisciplinary research on various problems faced by similar villages
inhabited by the predominantly Brahmin3 communities in the Western midhill
Region of Nepal (800-1500 m. asl.). The other ethnic groups are Chhetri,
Gurung, Damai, and Kami. The problems identified during the Samuhik
Bhraman were mainly related to an acute fodder shortage for livestock during
the dry season, fuelwood shortage, limited grazing land, lack of irrigation, and
nonavailability of farm inputs. The site has been used extensively to identify
suitable varieties for food and horticultural crops, fodder tree species, forage
and grasses, and livestock breeds.
A holistic approach in research with cross-disciplinary analysis of the
problem has been adopted, providing technical staff and social scientists an
opportunity to interact. This has been possible through coordination and
interaction among various technical sections, supported by a favourable
institutional set-up.
In Tapu, research activities have been accompanied by an extension service,
and the presence of local field staff has helped in promoting farmer participa-
tion. Farmer participation is sought in the planning, implementation, and
evaluation stages ofresearch. It has been further reinforced by developing new
microinstitutions and strengthening the old ones.

3 One of the major farming ethnic communities in the hills of Nepal.


Vol. 4, No. 3, 1994






VAIDYA AND GIBBON


Utilization of local resources, cost-benefit analysis of the developed tech-
nology, and testing of varieties under low-input and farmers' conditions are
also emphasised.

Biophysical Changes
After the problems identified during the Samuhik Bhraman were studied,
a plan of action was prepared, and research activities initiated. Research
broadly aimed to improve food security and investigate the possibility of
increasing cash income, as the farmers were entirely dependent upon agricul-
ture for their livelihood. After more than six years of work in Tapu, the major
objectives were achieved in terms of more intensive cropping patterns with the
introduction of high yield and improved varieties as well as new crops,
introduction of cross-bred animals, and significant improvements in the
availability of fodder. All these factors have resulted in increased food security,
and in addition, several income-generating activities have been identified
within the farming systems.

Diversification
The OSR concept has helped in understanding the interactions between the
components of agricultural systems in the hills and in formulating a program
that has promoted a systemic approach to identifying crops, varieties, live-
stock, fodder species, and grasses with enhanced potential and wider adaptabil-
ity. The cropping patterns before and after the intervention indicate that
diversity has been achieved, with the introduction of vegetables and food and
cash crops (Table 1).
Advantage: The varieties adopted by the farmers are a consequence of both
basic and adaptive research carried out in farmers' fields at various stages over
a period of time. Farmers emphasize maintaining a wide genetic diversity of
crop varieties to counter unforeseen breakdown of a variety due to insect and/
or disease outbreaks. Local varieties are therefore always included in screening
processes for new varieties.
Tapu farmers have adopted more than 25 varieties of staple food/cash crops
(rice, wheat, maize, finger millet, potato, lentil, mustard, early rice, and
soybean), more than 22 varieties of vegetables (cabbage, cauliflower, broad
leafmustard, butter bean, brinjal, cucumber, capsicum, tomato and zucchini),
and about 15 types of fruits (mandarin, banana, lime, lemon, guava, sweet
orange, peach, plum, grape, and apple).
Similarly, more than 7,000 saplings of 33 species of different fodder, fuel,
and spice value trees have been planted. Farmers have also responded by
planting ground cover grasses and legumes such as Setaria, Hybrid Napier, and
Desmodium, on the terrace risers and wasteland.
Efforts were also made to improve livestock performance, but the village
failed to capitalize on the developed technologies due to its own site-specific


Journal for Farming Systems Research-Extension






SURVIVAL AND SUSTAINABILITY


problems, such as frequent exchange and sale of animals, lack of fodder, and
lack of community land for grazing. However, there are more than 20 Murrah
buffaloes (an improved riverine breed animal native to Indian plains) in the
village. Aulo and Khari goats, the local selections from other parts of the
country, were found more successful than the exotic breeds because of their
ability to produce twins and triplets.

Table 1. Changes in Cropping Pattern by Land Type at Tapu.


BEFORE INTERVENTION
Khetland* (21 ha)
Rice**-fallow
Rice-maize
Rice-wheat

Rice-maize-beans (local)
Rice-mustard-maize
Rice-potatoes-maize








Bariland*** (3 ha)
Fingermillet-fallow
Maize/fingermillet-fallow


AFTER INTERVENTION

Rice-wheat (4.5-5 ha)
Rice-fallow-maize+beans (4-4.5 ha)
Rice-radish seed multiplication-maize+beans (2.5
ha)
Rice-potatoes-maize+beans+cowpeas (1.5 ha)
Rice-fallow (1.5-2 ha)
Rice-lentil (0.75-1 ha)
Rice-wheat-Sesbania sps. green manuring (0.75-1
ha)
Rice-fallow-off season summer vegetable (0.75-1
ha)
Rice-fallow-potatoes+maize (0.75 ha)
Rice-mustard-maize (0.5 ha)
Rice-vegetable-maize+beans+cowpeas (0.5 ha)


Maize/fingermillet-vegetable (0.05-0.1 ha)
Maize-radish nursery (0.1-0.15 ha)
Fingermillet-fallow (1.75-2 ha)
Fingermillet-butterbeans (0.1-0.15 ha)
Fingermillet-mustard (0.2-0.25 ha)
Maize/fingermillet-fallow (0.65-0.7)


* Irrigated lowland with bunds (rice field); ** Generally Blackgram (vigna mungo) is also
grown on the rice bunds, and in a few cases, soybean is grown.; *** Unirrigated upland
without bunds.

Problems: There were problems related to the intervention itself. The
introduction of radish seed multiplication programs shortened the crop
duration available for spring maize and decreased maize production signifi-
cantly. Farmers returned to local varieties because improved varieties did not
integrate into the cropping pattern. As a result, farmers have to buy maize from
other villages a cost they can afford from an increased earning from seed
production. Research has been focused on identifying short duration rice and
spring maize varieties to adjust the cropping pattern. The problems of insect
pests and diseases, as well as soil fertility are, however, now also more
pronounced.


Vol. 4, No. 3, 1994






VAIDYA AND GIBBON


Production Systems and Income Generation. A study carried out by Lumle
(Gurung et al., 1990) indicated that the area under all cereal crops has
increased by 4.5 percent. The area under local varieties of cereals decreased by
18.6 percent and area under improved varieties increased nearly four fold. The
food balance has decreased, indicating increase in consumption (Table 2).
Farmers can now afford to buy grain from the cash income earned from
vegetable farming and seed-multiplication enterprises. The village earned
257,884 rupees (1 US $= Rs. 42.6 in 1992) annually from fresh vegetables,
vegetable seed, and fruit farming as against 12,510 rupees from fruit farming
only before the intervention (Table 3). The number of improved livestock also
increased from none to 44, and the total animal population decreased by 40.5
percent indicating removal of unproductive stock (Table 4).
The FSR program has identified several cash-generating enterprises within
the system such as seed multiplication of radish, potato, legumes, and cereals,
cross-bred animals, and a wide range of seasonal as well as off-season vegetable
cultivation, and fruit farming.
Table 2. Food Balance Status Before and After Intervention at Tapu.
BEFORE INTERVENTION AFTER INTERVENTION
Production (tons) 84.5 95.8 (+13.4)*
Deficit (tons) 2.0 2.9 (45.0)
Surplus (tons) 15.5 15.6 (+0.6)
Net balance (tons) 13.5 12.7 (-5.9)
* Figures in parentheses indicate percentage increase (+) or decrease (-).; Source: Impact
study survey, 1989.

Table 3. Annual Cash Income (rupees)* From Vegetable and Fruit Farming at
Tapu Before and After Intervention.
BEFORE INTERVENTION AFTER INTERVENTION
Vegetable seed 354 129,084
Vegetable seed sale 0 86,785
Fruit sale 12,510 42,015
Total 12,885 257,884
* 1 US $ = Rs. 44 (1991); Source: Impact study group, 1989.

Table 4. Number and Type of Animal Breed Before and After Intervention at
Tapu.
BEFORE INTERVENTION AFTER INTERVENTION
100% Local breed 351 165 (53)*
Improved breed
Cross 0 36
100% improved* 0 8
Total 0 44
Total 351 209 (-40.5)
* Figures in parentheses indicate increase (+) or decrease (-) over the base year figure.; **
100% improved breeds are limited to small livestock species (chickens and rabbits).;
Source: Impact study survey, 1989.

Journal for Farming Systems Research-Extension






SURVIVAL AND SUSTAINABILITY


Socioeconomic Changes
Sustainable economic development is directly concerned with increasing
the material standard of living of the poor at the grassroots level (Barbier,
1987). The OSR activities at Tapu have increased the cash income for these
Brahmin farmers, whose earnings were formerly less than the minimum
required to buy basic essential goods. This can be verified by several indicators
such as increases in food consumption and nutrition level, polythene piped
drinking and irrigation water, thatched houses replaced by slate roofs, in-
creased numbers of children attending school, etc. As a result, farmers have
developed confidence in new technology and are willing to bear risks. This has
been employed to the mutual benefit of the LRARC and the farmers.
Migration from the hills to the Terai plains is a consequence of unproduc-
tive farming. Villagers had to go to Indian cities to seek employment in order
to survive. Several people have now returned to the village because of the new
opportunities, and others now give more careful consideration before deciding
to leave.

Farmer Participation
In Tapu, farmers are involved at all stages ofresearch, providing the land, labour,
and other management inputs. Farmers are involved in the following activities.
On-farm trials. Several types of trials are carried out, ranging from simple
demonstrational types to complicated, replicated ones. Farmers' Field Trials
(FFT) and Pre Production Verification Trials (PPVT) are common for staple
food and cash crops and, similarly, FFT's and Production-Cum Verification
Trials (PCVT) for horticultural crops. Livestock studies have focused on breed
improvement and introduction of new animals. Forestry and pasture activities
have involved identification of suitable species of fodder and grasses.
Farmer Feedback and Preference Ranking. Farmer feedback plays a key role
in understanding their criteria of adoption, as it varies across the regions and
communities, and must be considered when making recommendations, espe-
ciallyin the context of the extreme agroecological and socioeconomic diversity
in the Command Area.
At Tapu, near the time of crop maturity, a Farmers' Field Day is organized
in consultation with the Village Committee, and farmers are asked to rank the
treatments of Farmers' Field Trials in the field. Further assessment is carried
out later to assess other criteria of selection, such as taste, cooking quality,
milling percentage, etc. of new varieties. General agreement among the
farmers as to acceptability and correlation between biological yield and
farmers' ranking helps to evaluate a new variety. It is often found that the high-
yield variety is not necessarily the farmers' choice.
Seed Multiplication. Several new varieties in different crops have been
identified as a result of on-farm research. However, the supply of improved
seed has always been a problem. First, limitations of Lumle to undertake seed
production of a wide range of crops to meet the demands of farmers of the


Vol. 4, No. 3, 1994






VAIDYA AND GIBBON


Extension as well as the Research Command Area is a constraint. A second
factor is the cost involved in transportation. Emphasis was therefore placed on
village- based seed production.
Tapu has produced and supplied quality seed of different cereals, legumes,
potato, and vegetables to the region. This has been achieved through technical
supervision, careful planning, and farmers' cooperation at the site.
Bringing potato seeds from the higher hills involved a risk of importing the
bacterial wilt disease (Pseudomonassolanacearum) into the area, in addition
to the transportation costs. Tapu farmers responded to Lumle's request to
launch a seed multiplication program using a 'seed plot technique.' With this
experience, Tapu has become a major seed potato growing area in the region
and the seed potato a major source of cash income for farmers.
In addition to these, development of microinstitutions (Village Develop-
ment Committee, Seed Growers' Committee, and Forest Development Com-
mittee), farmers' field days, study tours, and participation of Tapu farmers in
"Farmers as Trainer" program are further evidence of fruitful cooperation.

Farming Systems Perspective
Soil Fertility and Fodder Scarcity. Soil fertility is considered the main
constraint in hill farming systems. The general trend of crop production in the
hills is declining yields, according to several reports including the Agricultural
Development Bank (ADB)/ His Majesty's Government, Nepal (HMGN)
(1982). An impact study has revealed that the total production of most crops
at the site has increased after the intervention. This is partly due to crop
intensification and use of farm inputs, which Harrington et al., (1990)
perceived as confounding factors in understanding long-term productivity
problems and masking the productivity trend. Nonetheless, application of
organic manure, a traditional method, continues to be the principle method
of maintaining soil fertility at Tapu. The others are inclusion oflegumes in crop
rotation, use of a short fallow, and slicing of terrace risers.
As implied earlier, the problems of soil fertility and fodder scarcity are
interrelated, and the respective thrusts have carried out research on the topic.
Soil Fertility Thrust has initiated several research activities to maintain and
improve soil fertility at Tapu, especially against the background of declining
livestock population, crop intensification, and declining soil fertility condi-
tions.
The contraction of the fodder base has not led to decline in livestock
numbers, and Nepal has the highest livestock population per unit of cultivated
area of any country in the world (ADB/HMGN, 1982). However, under
similar conditions, livestock population in Tapu has decreased significantly
mainly due to the farmers' understanding of the need to ensure adequate
fodder for a smaller number of productive animals. The adoption of improved
and cross bred animals that demand better management and nutrition required
a community decision to enforce a stall-feeding system. Strict control by the


Journal for Farming Systems Research-Extension






SURVIVAL AND SUSTAINABILITY


Village Forest Development Committee over the area and resources has also
contributed to reducing animal numbers through removal of the unproductive
ones. These factors have encouraged fodder planting in private lands, better
regeneration of existing forests, and better collection of urine and dung for
composting.
Measures Taken. Several research activities were initiated to improve and
maintain soil fertility. More than five fodder tree species were identified. Lentil
was introduced in the rice-based cropping pattern to encourage symbiotic
nitrogen fixation in the soil. Comparative performances of different exotic and
indigenous green manuring plants were studied. Research on bio-ertilizers for
cereal and legume crops was initiated and, as a result, farmers have found
Rhizobium (Rhizobiumjaponicum) effective for lentil, especially in soils where
the crop is grown for the first time. Cattle urine topdressing of vegetables was
found encouraging, but the urine loss in the traditional animal shed is
considerable. Therefore, a livestock housing improvement program to facili-
tate collection of urine and sanitation has been initiated.

Micro-institutional Development
Existing Infrastructure. Several types of traditional institutions prevail in
Nepalese hill society and enforce regulations to maximize social welfare, protect
natural resources, and maintain their cultural integrity by the local people
themselves. In Tapu, the "Village Forest Committee" has been operating from
time immemorial, and has successfully protected the Sal (Shorea robusta) forest, a
high-value timber tree species, even during the period that forests in Nepal were
nationalized and controlled by regionally-based officials. The general scenario has
seen the disappearance of the forest simply because local people were excluded in
the resource utilization process, and through centralization of legislative power.
However, the villagers of Tapu resisted officially permitted contractors from
cutting down the forest, knowing the consequences and importance ofthe canopy
that almost encircles the village over a very steep topography. The committee
undertakes the following activities:
Monthly open village meetings;
Opening the forest to the villagers for a certain period to allow them to collect
dead wood and other forest products;
Appointment of a forest watchman, who is paid both in cash and kinds from
the committee fund;
Provision of construction timber and fuelwood to the needy at an agreed rate
of extraction. Ground grasses are also sold to the highest bidder to raise funds;
Imposition of fines on offenders. Cattle grazing and leaf litter collection are
also prohibited within the forest; and
Encouragement ofprivate plantations. The committee has a unique tradition
of allowing villagers to plant bamboo anywhere on community land, even inside
the forest, and ownership is guaranteed.


Vol. 4, No. 3, 1994





VAIDYA AND GIBBON


OSR Involvement. The OSRForestry and Pasture program is closely coordinat-
ed with the Committee and technical matters are discussed regularly. With the
assurance that all government forests will be devolved back to the local level, the
villagers have now applied for Community Forest status at the instigation ofLumle
LRARC. Although at present, the Committee has no legal authority to act in the
way it functions, Lumle has helped survey the forest and has drafted a management
plan. The use of improved stoves is also promoted, because of their more efficient
use of fuelwood and smoke-free operation. Impressive regeneration of the forest
has occurred as the result of terminating the open grazing system of cattle. The
approach of strengthening the traditional microinstitution, with the help of
greater institutional support, has promoted environmental stability and encour-
aged desirable changes in the system.
Village Development Committee. Tapu is a representative research site. The
problem of positive farmer participation arose, as a natural process, when several
income-generating activities were identified within the system and farmers
became production oriented. As a consequence, there is now a declining interest
on the farmers' part in testing technologies in their fields, and lack of cooperation
is becoming apparent in carrying out trials.
Subsequently, the need for an educational approach to explain to farmers the
importance of trials in their fields was felt. In 1989, a "Village Development
Committee" was formed with the objective of developing a forum in which overall
development of the village could be discussed. In practice, it is difficult to isolate
the agricultural objectives entirely from others. LRARC has often been obliged to
contribute toward village development activities such as construction of school
facilities and toilets, supply of furniture, irrigation and water supply, etc. in the
form of kind, cash, and technical expertise. Moreover, the villagers expect the
technical staff to be in a position to comment upon matters such as sanitation,
health, education, etc. This demonstrates the flexibility and integrated approach
to appropriate institutional development.
The finding ofthis committee has been noteworthy in several aspects, including
the creation of a good understanding between Lumle staff and the villagers, and
promotion of a cooperative atmosphere for the OSR program.
Seed Growers' Committee. Encouraged by the impressive potato seed multipli-
cation program that helped overcome the traditional view of the potato as a high
hill crop, a "Potato Development Committee" was launched to take care of new
challenges such as strict quarantine to prevent diseases, maintain quality seed,
ensure timely availability of inputs, and to develop marketing opportunities.
Thus, microinstitutional development has increased in importance in the
Lumle farming systems program but is often unpredictable and takes its own
course. Its nature has varied from place to place due to the richly heterogeneous
nature of the hills and their peoples. It promotes farmer-participatory research and
better cooperation in the field. The committee has been the interface between the
farmers and the research organization (Figure 2).


Journal for Farming Systems Research-Extension






SURVIVAL AND SUSTAINABILITY


Plantation/Improved stoves


Figure 2. Micro-institutional linkages with research organizations
at the OSR site, Tapu.


SUSTAINABILITY ISSUES OF THE APPROACH

Lumle is a regionally based autonomous research organization with a limited
regional mandate, and sufficient financial and managerial flexibility that allows
it to adopt radical measures to operationalize its program. The project has been
funded by the British Government since 1968, and has developed a strong,
competent technical workforce during the project period. Though recently the
LRARC has been incorporated into the national agricultural research system,


Vol. 4, No. 3, 1994






VAIDYA AND GIBBON


it has maintained its autonomous character. However, the policy is to gradu-
ally adopt the Government practices and eventually integrate it into the
national research system. This is the point at which every activity will have to
be examined.
Sustainable agriculture is more than a sustainable farming system (Wein-
schenck, 1989). Therefore, a critical analysis of the approach will obviously
induce some reservations that need be considered in refining the concept and
in understanding the existence of a relationship between production systems
and the environment with a systems perspective.

Content of the Research Program
The conventional perception that multidisciplinarity is a holistic approach
is narrow in the sense that a number of factors that influence the objective have
failed to come under the purview of the research mandate. The FSR priorities
are demarcated "too tightly around the farm field" (Anderson, 1990) and have
not been broadly based. These narrow research mandates may present an
institution successful in terms of output per unit, but disadvantaged target
groups may remain worse off. Other components such as stable prices,
marketing arrangements, credit, and input availability are also crucial factors.
Thus, production maximization alone cannot improve social welfare.
The LRARC's FSR and FSRE program have often resulted in increased
production without much market opportunity, only to discourage farmers
from growing the next year. This is apparently due to lack of information about
consumer choice and market demand. The dependence of Resource Poor
Farmers on nonagricultural activities cannot be ignored. Similarly, and most
importantly, the emphasis on environmental issues in the hill farming context
is of paramount importance. The problem of soil conservation is as important
as, if not more so, than increasing productivity. This calls for a more complete
perspective approach to foster an integrated effort to achieve sustainable
agricultural growth in the longer run.

Multidisciplinarity
The practical problems in a heterogeneous social and cultural context, and
the problem of fitting this approach neatly into the current bureaucratic nature
of the organization of Governmental agricultural research in Nepal, are
discussed adequately in Sthapit et al..(1988). There is no effective mechanism
to coordinate different departments under the Ministry ofAgriculture, and any
technology developed through an interdisciplinary approach will not fit in the
present system.
FSR is an expensive approach in itself. The multidisciplinary team and
Samuhik Bhramans involve high costs in terms of movement in the hills,
handling logistics, field allowances, and report writing. The research method-
ology adopted seems unviable given the limitations of Government resources


Journal for Farming Systems Research-Extension






SURVIVAL AND SUSTAINABILITY


and capability. The selection of sites involves a series of exploratory surveys,
formal surveys, and finally impact surveys, followed by massive data analysis
that also incurs high costs.

OSR Sites
The OSR site has emerged as a major tool for generating technologies for
the farmers in its domain. However, despite a few successful examples, the sites
fail to serve the long-term needs of research. The concentration of activity in
the OSRsites, involving the considerable support offarmers, has weakened the
"representative cell" role of the sites, for it is difficult to envisage such a focus
of resources and expertise spread over a wider area. This leaves the site
appearing atypical rather than representative.
A more sustainable approach would require pragmatism in research activity,
so that resource and expertise concentration is more in tune with conditions
in other locations under the domain. Operating within domain constraints
may imply less impressive results at the site, but these results would more
closely reflect the performance capabilities of other areas under the site's
auspices.

Farmer Cooperation
The OSRsites have served as more than a FSRsite which often involved the
basic and complicated trials and did not attract farmers as their results did not
instantly benefit them. This is simply because the LRARC does not represent
its diverse command area. Only the provision of compensation, not the
participation, sustained such efforts. However, compensation is not an integral
feature of the Government's research systems. The researchers' emphasis on
neatly managed and statistically valid trials has imposed an unauthentic
environment of farmer participation. The design of simple and problem-
oriented trials may help to overcome these problems. Nonetheless, the need
for varietal screening for specific agroclimatic zones cannot be underestimat-
ed. It is a matter of concern that even the FFTs and PPVTs, which are
conducted in the farmers' field and are nominally under their management,
also involve considerable amounts of researcher control. This inevitably
implies a trade-off between positive farmer participation and so-called valid
research. Nevertheless, one must add that technology, however technically
sound it may be, has no meaning unless it is adopted by farmers.
The final arbiters of change are the farmers themselves. Although there have
been noteworthy quantitative gains at the site, the impact upon farmers in
terms of social welfare and their own personal reactions define the quality of
these gains. In gauging the replicability of the technology, it is essential to
consider the variables of social response. The social acceptability of technology
determines whether it will succeed.


Vol. 4, No. 3, 1994






VAIDYA AND GIBBON


CONCLUSION

Sustainable agricultural and social development of the hills lies entirely on the
pattern of its resource use and the present gloomy picture can be attributed to
the growing pressure on it. However, the use pattern of the resource is an
expression of deep political, economic, and cultural structure (Eckholm,
1976) and cannot be changed easily. Biophysical, social, economic, and
national policies are the factors involved to meet this end. The will to
implement policies that improve sustainability is often hampered because small
hill farmers are less receptive to change. The development of mechanisms to
involve the microsocial structures in the research and extension program are,
therefore, essential in addition to any strategy aimed at increasing the biophys-
ical productivity of a system.
The Lumle approach to hill farming systems has contributed to developing
a methodology based on a consideration of most of the components in the
system, including the social and environmental, of which they are a part. An
emphasis on farmer participatory research and microinstitutional development
has not only facilitated a process of generating appropriate technology for
farmers, but has also improved their environmental awareness.
Low external input farming systems in these environments can be both
sustainable and productive given the right conditions of local control in
decision making regarding the management ofresources and that research and
extension agencies play an appropriately given supportive role.

ACKNOWLEDGEMENTS

The Lumle Regional Agricultural Research Centre (LRARC) is funded by the
Overseas Development Administration of the British Government and works
in cooperation with His Majesty's Government of Nepal. The support of both
Governments is gratefully acknowledged.
We wish to thank John Abington and Bhuwon Sthapit for their valuable
comments on the earlier draft of this paper. Finally, the support of the AFSRE
Symposium Organizers in allowing the first author to attend this meeting is
also gratefully acknowledged.

REFERENCES
ADB/HMGN. 1982. Nepal Agricultural Sector Strategy Study, Vol. II, Detailed
Sector Review and Analysis, Asian Development Bank and His Majesty's Govern-
ment of Nepal, Kathmandu.
Anderson, J.R.1990. FSRE Impact Inquisition: Investor Issues, Paper Prepared for the
1990 Asia Farming Systems Research and Extension Symposium, sponsored by the
Asian Institute of Technology Bangkok: 6-8.


Journal for Farming Systems Research-Extension






SURVIVAL AND SUSTAINABILITY


Barbier, E.B. 1987. The Concept ofSustainable Economic Development. Offprint Series
No. EC1, pages 101-110. Reprinted from Environmental Conservation 14(2).
International Institute for Environment and Development.
Eckholm, E.P. 1976. Losing Ground: Environmental Stress and Food Prospects, 24.
W.W. Norton & Co., New York: 166-167
Gurung, H.B., P.K. Shrestha, and L.K. Amatya. 1990. Impact ofLumle Agricultural
Research LRARC'sFSR/E Programme: A Case Study ofFour Hill Farming Commu-
nities in Western Nepal. Lumle Agricultural Centre (LRARC) Seminar Paper
No.37.
Harrington, P. Hobbs, T. Pokhrel, B. Shrarma, S. Fujisaka, and C. Lightfoot. 1990.
The Rice-Wheat Pattern in the Nepal Terai: Issues in the Identification and
Definition of Sustainability Problems. Journal of Farming Systems Research-Exten-
sion: 1-27
Jodha, N.S. 1989. Mountain Agriculture: Search for Sustainability. Paper presented at
International Symposium on Impacts of Farming Systems Research and Extension
on Sustainable Agriculture, University of Arkansas, October 9-12, 1989.
Joshi, K.D., B.R. Sthapit, B.Pound, and J. Gurung. 1990. Research Thrust: A
Multidisciplinary Research Approach to Generate Sustainable Technologies, LAC
Technical Paper No. 26.
Khadka, R, and D. Gibbon. 1988. The Contribution of Pakhribas Agricultural
LRARC to the Development of Farming Systems in the Eastern Hills of Nepal.
Paper presented at the Eighth Annual Association for Farming Systems Research-
Extension Symposium, Arkansas, USA, October 9-12th.
Pound, B., K. Budathoki, and B.R1 Joshi. 1990. An approach to Mountain Agricultur-
al Development: The Lumle Model. LAC Seminar Paper No. 23.
Sthapit, B.R., P.A.K. Balogun, and J.A. Seeley. 1988. Multidisciplinary Farming
Systems Research In Practice: The Experience ofLumle Agricultural LRARC, LAC
Technical paper No. 88/23, Nepal.
Thapa, B.R. 1987. The Potential of Agroforestry in the Middle Hills of Nepal. MSc
dissertation, University College of North Wales, Bangor, UK.
Weinschenck, G. 1989. From Sustainable Household to Sustainable Farming Environ-
ment Systems, Quarterly Journal of International Agriculture 28(3/4):242-253.


Vol. 4, No. 3, 1994








Circumstances of Rapid Spread of
Cultivation of Improved Cassava Varieties
in Nigeria

Felix I. Nweke, S.K. Hahn, and B. 0. Ugwu


INTRODUCTION
The International Institute of Tropical Agriculture (IITA, 1988) defines a
staple food crop as that which accounts for more than 200 calories per day in
the diet of an individual. The Food and Agriculture Organization of the
United Nations (FAO, 1970) estimated that from 1964-66, cassava accounted
for more than 300 calories per day in the diet of an average individual in
Nigeria as a whole. In the moist areas of the country, however, cassava
contributes much more than 300 calories per day. Nweke et al. (1992a)
estimate that cassava accounted for more than 800 calories per day in the diet
of an individual among some population groups of southeastern Nigeria. In
addition, since the 1964-66 period, cassava production is reported to have
increased in 90 percent of villages that represent cassava-producing areas of
Nigeria, displacing yams and, in some cases, grains as well. Cassava is also a
major source of cash income for rural household producers. An average of
nearly 50 percent of total production is marketed, and some smallholders who
produce almost entirely for market sell close to 90 percent of total production
(COSCA Nigeria, 1993). In addition, cassava is a reliable staple food in areas
where the risk of drought is high. Cassava's well known tolerance to drought,
poor soils, and irregular labor requirement are clearly advantages in such areas
(Berry, 1993).
In 1971, IITA commenced breeding for resistance against African cassava
mosaic virus (ACMV) and cassava bacterial blight (CBB), which was devastat-
ing cassava crops at the time, and for high root yield. In 1974, IITA and Shell-
BP Petroleum Development Company of Nigeria Limited (Shell-BP)
commenced on-farm testing and distribution of promising clones selected for
resistance to the diseases and for high yield. The Collaborative Study of
Cassava in Africa (COSCA) showed that by 1989 the improved cassava
varieties were available in 90 percent of cassava-growing villages throughout
Nigeria, and many or most farmers planted the improved varieties in 60
percent of such villages in the same year. In 1991, another COSCA study
showed that the improved cassava varieties covered 60 percent of cassava fields

1 International Institute of Tropical Agriculture (IITA), c/o L.W. Lambourn & Co., Carolyn
House, 26 Dingwall Road, Croydon CR9 3EE, England.






NWEKE ET AL.


in the humid and 40 percent outside the humid climate areas of the country
(Nweke et al., 1992b).
This paper attempts to assess reasons for the rapid spread of the improved
cassava varieties in Nigeria. The hypothesis is that the wide adoption of
improved cassava varieties by Nigerian farmers was rapid primarily because
such varieties are superior to alternative local varieties in most of the attributes
of interest to the farmers. The rapid adoption was also due to the prevalence
of peace and political stability over a relatively long time period and the
availability of a stable revenue source-petroleum-which enabled Nigerian
Governments to put in place relevant market and processing infrastructure and
to carry out extension programs. This paper is based on information provided
from the COSCA study. After a brief discussion of the method of the COSCA
study, the circumstances of the wide spread of the improved cassava varieties
are presented in four parts. The firstis an assessment of the performance of the
improved cassava varieties in comparison to local alternatives; the second is a
discussion of the effect of political stability and a stable revenue source; the
third focuses on the government's contribution to widespread adoption; and
the fourth describes the contribution of private and other nongovernmental
organizations to the rapid spread of the improved cassava varieties.

METHOD OF THE COSCA STUDY

The COSCA study was funded by the Rockefeller Foundation. The objective
was to collect authoritative information over a wide area on production and
processing methods, market prospects, and consumption patterns in order to
guide research on the cassava crop. The study was conducted in six coun-
tries-Cote d'Ivoire, Ghana, Uganda, Tanzania, Zaire, and Nigeria-which
account for more the 70 percent of cassava production in Africa and provide
wide variability in conditions that influence cassava production (Carter and
Jones, 1989). It was executed in phases. The first phase was a broad character-
ization survey of cassava-producing areas of Africa conducted in 1989. The
second phase was a detailed production survey conducted in 1991. The third
phase was a detailed survey for information on a wide range of post-harvest
issues (Nweke, 1988). All three survey phases were conducted on the same
households in the same village.
Climate, demographic pressure, and market access infrastructure formed
the bases for sampling for the COSCA study (Carter and Jones, 1989). Areas
with a growing season daily temperature mean of above 22 degrees Celsius,
range of less than 10 degrees, and less than four months of dry season are
classified as lowland humid climate zone. A month of dry season is that with a
total rainfall of not more than 60 mm. Areas with a growing season daily
temperature mean of less than 22 degrees, range of less than 10 degrees, and
less than four months of dry season are classified as highland humid climate
zone. Areas with a growing season daily temperature mean of above 22 degrees,


Journal for Farming Systems Research-Extension






IMPROVED CASSAVA VARIETIES


range of 10 degrees, and four to six months of dry season are classified as sub-
humid. Other areas with a growing season temperature mean of above 22
degrees, range of above 10 degrees, and four to nine months of dry season are
classified as non-humid.
All-weather roads, railways, and navigable rivers were derived from the
1987 Michelin travel maps and used to create a market access infrastructure
map of Africa. This map was divided into good and poor zones according to
the density of roads, railways, or navigable waterways. Population data from
the US Census Bureau (unpublished data), projected forward to 1990, were
used to calculate population densities to create a population map that was
divided into high and low demographic pressure zones, the former comprising
areas with 50 or more persons per square kilometer.
The three maps ofclimate, population density, and market access infrastruc-
ture were overlaid to create zones with homogenous climatic, demographic,
and market access infrastructure conditions of the cassava-producing areas.
Each climate/population density/market access infrastructure zone with less
than 10,000 ha of cassava in each country was excluded as unrepresentative of


Figure 1. Location of COSCA survey sites in Nigeria.


Vol. 4, No. 3, 1994






NWEKE ET AL.


cassava-growing areas. The remaining areas, which formed the potential
survey regions, were divided into grids of cells 12' latitude by 12' longitude
to form the sample frame for sites selection. Sixty-five grid cells distributed
among the climate/population density/market access infrastructure zones in
proportion to the zone size were also selected by a random method in Nigeria.
One village was selected, by a random method, within each of the grid cells
(Figure 1).
In each selected village, key informants assisted in compiling a list of farm
households and grouping them into "large," medium," and "small" small-
holder units. Farm units that cultivated 10 ha or more of all crops were
excluded. One farm unit was selected from each stratum by a random method.
Information was taken from all the fields, including cassava and noncassava
fields, of such selected farmers.
This sampling procedure was selected in order to make the observations on
the spread and performance of improved cassava varieties in Nigeria very
objective; the sampling design is general for cassava-growing areas and not
specifically aimed at the assessment of the spread of improved varieties.
For the Phase 1 survey, a rapid rural appraisal technique was employed in
which farmer groups, consisting of men and women representing a wide age
range, were constituted and interviewed in each village. Hence, in Phase I of
the COSCA study, the unit of analysis was the village site. The groups in each
village were interviewed with structured instruments for qualitative informa-
tion on aspects including: (1) production practices: cassava varieties grown at
present and in the past. For varieties only grown in the past, why is the variety
no longer grown, and what year was it last grown. For varieties currently
grown, what year was the variety first grown, and what are its advantages and
disadvantages. Following the interview meetings, the investigators went to the
cassava fields, identified the varieties, and classified them as local or improved;
(2) cassava processing methods, including cassava products processed; and (3)
cassava marketing, including cassava products marketed.
Phase II of the COSCA study aimed at detailed characterization of the
cassava production methods; the information was collected at the field level as
the unit of analysis. The information taken included field history, cassava
varieties grown, cassava rootyield, HCN level, as well as some other agronom-
ic yield components and field size, among other information. Yield estimation
was made for fields that were nine months old and above, except when the
farmer harvested at less than that age. The estimation was based on a
representative sample plot of 40 square meters, except when the field was too
small, in which case a 20 square meter plot was used. There were one or two
plots per field, depending on the size and heterogeneity of the field in terms
of soil and toposequence. Cassava stands within the sample plot were separated
by variety, counted, and then harvested. Both the roots and the tops were
weighed separately and the roots counted, again by variety. The estimates of
the level of potential HCN in the roots were based on simple field techniques


Journal for 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