in mixed farming systems
RESEARCH METHODS AND PRIORITIES
International Livestock Center for Africa,
FP m Farming Systems Support Project
Network Report No. 2
FSSP NETWORK REPORT NO. 2
"LIVESTOCK IN MIXED FARMING SYSTEMS:
RESEARCH METHODOLOGIES AND PRIORITIES"
Proceedings of a workshop held at the
International Livestock Center for Africa
Addis Ababa, Ethiopia
24-27 June, 1985
Steve Kearl, Editor
a collaborative effort of the
Farming Systems Support Project, University of Florida
International Livestock Center for Africa
The livestock component of small farms remains a topic of great
concern, not only to ILCA but to farming systems researchers and
practitioners throughout the world. In the context of a systems
approach to agricultural research and extension, the livestock
component of small farms has until recently been somewhat neglected, or
left out of the movement altogether. In part this seems to be due to a
number of perceptions: the complexity of the animal situation; the
traditionalism of animal researchers; the multiple nature of livestock
use; the perceived dominance of the importance of crops; and the
greater difficulty of on-farm research with animals as compared with
Soon after the Farming Systems Support Project (FSSP) was organized by
the University of Florida staff, a Livestock Task Force was appointed
to bring together. the facts on the importance of livestock in mixed
crop/livestock farming systems. The Task Force was charged with
reporting on the state of the knowledge relative to on-farm and
on-station farming systems research with an animal component, and to
assess the needs that should be addressed in future research activities
within the farming systems research context. One of the motives for
formulating the Task Force was to improve the understanding of the
livestock component of limited resource farms and to stimulate
consideration if not inclusion of livestock in the systems approach.
Among a number of recommendations emanating from the Livestock Task
Force, one addressed the need for regional workshops that would bring
together researchers and practitioners to share their experiences with
research methodologies and techniques being used to conduct on-station
and on-farm trials involving livestock. The workshops were to be
interdisciplinary, including scientists from different disciplines such
as animal nutrition, forage and crop agronomy, agricultural economics,
rural sociology, anthropology and veterinary science. The common
ground for workshop participants would be experience in mixed farming
systems including a livestock component.
"Livestock in Mixed Farming Systems: Research Methods and Priorities"
is the second in a series of workshops that the Farming Systems Support
Project has promoted to address these various concerns. This workshop
was a collaborative effort between the FSSP and ILCA, whose hosting,
participation and contribution are gratefully acknowledged.
Objectives of the workshop were to present, discuss and evaluate
research methods and design techniques used in on-farm and on-station
trials in terms of their usefulness and application to farming systems
research in crop/livestock systems. Participants were expected to
identify major constraints in such mixed farming systems and place
priorities on them according to their potential economic importance and
amenability to research. In addition, participants were asked to begin
to develop guidelines and recommendations for conducting research on
livestock in farming systems that would be useful to project planners,
implementors and evaluators associated with national research programs,
universities, international agricultural research centers and bilateral
It is hoped that these proceedings will contribute in a meaningful way
to the continuing evolution of research priorities and the nature of
the research thrust with livestock in mixed farming systems.
TABLE OF CONTENTS
1. The ILCA/NLPU On-Farm Alley Farming Project:
Considerations on Data Collection and Evaluation..........1
2. Station Work with Traction Animals, An Example
3. Difficulties in Evaluating On-Farm Experiments
with Livestock: Examples from the Ethiopian Highlands..27
4. On the Alleged Difficulties of On-Farm Research
5. Socioeconomic Methodologies in the Dual-Purpose
Goat Farming Systems Research in Western Kenya..........71
6. Dairy Production Systems in Small Holdings in
Gezira Area of Sudan....................................98
7. Use and Issues in On-Farm Research to Improve
Fodder for Livestock of Agro-Pastoralists in
the Nigerian Subhumid Zone............................117
8. Matching the Impositions of the Farming and Research
Systems as a Prelude to Farming Systems Research........136
9. Mineral Research Techniques for Ruminants .............163
10. Analysis and Interpretation of Data from Farms.........195
Summary of Group One Discussion On-Station
Summary of Group Two Discussion On-Farm
LIST OF PARTICIPANTS...........................................217
THE ILCA/NLPU ON-FARM ALLEY FARMING PROJECT:
CONSIDERATIONS ON DATA COLLECTION AND EVALUATION
Small Ruminant Programme
International Livestock Centre for Africa
(ILCA)/Nigerian Livestock Project Unit
The prime objective of agricultural research can be seen as the
improvement of production inputs and technologies for the increased
production of food and other agricultural products. It is therefore
essential that the results of agricultural research get to the farmer
who is the ultimate user of the improved technology. This being so, it
is equally essential that the socio-cultural environment of the farmer
is taken into account in the development of technologies, to ensure
that the end-product of research is feasible and viable under farmer
This question of feasibility and viability under farmer conditions has
often been cited or inferred as the main reason for moving research
from station to farmers' farms (Kirkby et al, 1981; Steiner, 1982).
On-farm research (OFR), is thus seen as serving an extremely important
role in translating promising experiment station results into
practically relevant and economically sound recommendations for farmers
Several types of OFR can be distinguished on the basis of a number of
.factors, two of which are (i) the level of farmer involvement and (ii)
the complexity of the trial. These two factors together influence the
type of data that can be obtained from OFR. A distinction should
therefore be made between OFR designed specifically for the provision
of on-farm data on biological and technical parameters, and those
established for the development and assessment of the relevance and
acceptability of proposed interventions for the farmer.
This latter type of OFR, which has been described as pre-extension OFR
(Stoop, 1982) has its major emphasis in the farmers' maximum control
and responsibility in the interventions being tested. Use is made of
extension methodologies to gain the required farmer understanding and
involvement and evaluation is through assessment of farmer's reactions
and responses, rather than through precise measurements in specific
parameters. An example of this type of OFR is the on-farm alley
farming research of the International Livestock Center for Africa
(ILCA), at Ibadan, Nigeria.
ILCA's Alley Farming Package
The Humid Zone Programme of ILCA, based in Ibadan, Nigeria, has, since
1980, been working on the development of an integrated production
system for the improvement of crop and small ruminant production in the
humid tropics. The system, alley farming, is based on the alley
cropping concept earlier developed by the International Institute of
Tropical Agriculture (IITA) and is the major component in ILCA's
intervention in the farming systems of the humid zone. In alley
farming, food crops are grown between alleys formed by rows of fast
growing leguminous browse species which are pruned frequently to
prevent shading of the food crops. Being legumes, these trees are
capable of fixing nitrogen as well as siphoning soil nutrients from
deeper layers of soil and releasing them to the top soil through
mulching of their nitrogen-rich foliage. The prunings can also be used
as feed for small ruminant. Through alley farming therefore soil
fertility is maintained to allow for continuous arable crop production
at sustained yield levels, and small ruminants production is improved
through availability of high-protein fodder.
A second component of the ILCA intervention is the annual vaccination
of small ruminants against Pestes des Petits Ruminants (PPR) a viral
disease responsible for major mortalities in the small ruminant stock.
The disease strikes in epidemic levels and whole flocks may be lost as
a result of its occurence. ILCA's research has shown that PPR can be
prevented by annual vaccinations of Tissue Culture Rinderpest vaccine
(TCRV). In village level (on-farm) trials to test the effectiveness of
the vaccine, mortalities were reduced by over 70% with vaccination
(ILCA, 1983). These two components (alley farming and TCRV
vaccination) together address the major constraints to small ruminant
production in the zone disease and feed and are seen as having high
potential for improvement of small ruminant production without major
additions to current inputs.
On-farm Alley Farming Research
The development and testing of ILCA's alley farming intervention is
carried out both on-station and on-farm. Figure 1 gives the evolution
of ILCA's on-farm alley farming research from 1980 (with a single
farmer) to its culmination in a pilot research/development project
(involving over 60 farmers) in 1984. Direct researcher involvement in
the establishment and management of these trials has changed markedly
from very high in the initial years to an almost zero involvement in
the pilot project. It is argued that such zero involvement is
necessary to enable an accurate assessment of farmer response to a new
technology prior to its incorporation in a national development
The Pilot Project
The pilot project was initiated in 1983 by the Nigerian Livestock
Project Unit (NLPU), a livestock development agency under the Nigerian
Ministry of Agriculture and Natural Resources. The project is based
largely on ILCA's small ruminant improvement package (SRIP) which has
two components (i) establishment of leguminious fodder trees (alley
farming) and (ii) annual vaccinations of sheep and goats against PPR.
The objective of the project was to test the relevance, and
acceptability of the alley farming intervention and assess its
potential for adoption and spread amongst small scale farmers in the
humid zone of southwest Nigeria.
Owu-Ile and Iwo-Ate, two adjacent villages in the Ejigbo Local
Government Area of Oyo State were selected for the project. These
villages are just about 2 km apart, and together have about 500
permanently resident adults living in 201 houses (Okali, 1984). The
criteria for selection of these villages were active arable crop
farming and interest in small ruminant production.
Background methodologies and different kinds of activities carried out
in this project are described by Atta-Krah (1985) and are therefore not
detailed in this paper. These activities included village meetings,
farm tours, and demonstrations, on principles and practices of alley
farming. Leguminous tree seed for the project was supplied free of
charge by ILCA. Farmers receiving seed were required to establish
alley farms on one of their farm lands. The maximum size suggested for
an alley farm was 0.3 ha per farmer. On average, small-scale farmers
in the area cultivate 3-6 fields per farmer with a total area of about
2 ha. Only one of a farmer's fields and in some cases only a portion
of one field is used for the establishment of the alley farm.
All but four of the 67 farmers receiving tree seeds from ILCA planted
alley farms. Planting was done entirely by the farmers with no
supervision nor enforced uniformity practises amongst the various
In the following sections of this paper, monitoring and evaluation
issues in the ILCA/NLPU pilot alley farming project will be reviewed
against a background of unlimited farmer control in the establishment
and management of alley farms.
PROBLEMS OF DATA COLLECTION AND ANALYSIS
In order to fully appreciate the data collection and analysis issues in
this type of OFR, it is important to realise that farms under study are
actual farmers' farms in which farmers have full control, ownership and
responsibility. These are not "research-managed" research plots in
It must also be borne in mind that the major objective of this research
approach is the development and assessment of the acceptability of an
intervention to the farmer. Assessment of the system, whether on
biological, economic or social grounds, is therefore, best made by
taking the farmer's own view point into consideration. This is because
a farmer's decision to either adopt or abandon a tested intervention is
more likely to be based on his own personal experiences with the system
rather than on the "potential optimum benefits" that may be claimed
for the system.
In this type of OFR, as a result of the usually large and fluctuating
number of farmers participating, as well as the absolute control
enjoyed by the farmers in the management of the trials, it is often
difficult to obtain realistic quantitative productivity data. This is
because such operations as weeding, tree pruning and crop harvesting
are done by farmers on an ad hoc basis, in most cases in the absence of
the researcher. The only information that can be recorded in such
instances is whether or not a particular activity was carried out on a
particular farm, rather than the exact time and quantities involved in
the various activities. For this reason,. it is necessary to keep such
measurements to a minimum, putting greater emphasis on the qualitative
assessments, and using adoption and dropout analysis as a reflection of
the workability of the system.
Some quantitative data in fixed parameters such as. farm size, tree
spacing etc. could be collected to provide information for the
development of specific case studies or to establish the degree of
variability that exists in specified characters. Examples of this type
of data from the pilot project are shown in Tables 1 and 2, for farm
size, and tree intra-row spacing. Table 1 shows that when given the
option, farmers testing a new intervention will prefer to do it on a
small- rather than large-scale. Only about 15% of 'farmers in both
villages established alley farms up to the 0.3 ha recommended by ILCA
as the starting hecterage. Table 2 also shows that trees in most farms
were established with within-row spacings of 40-50 cm, as against the
25 cm spacing recommended by ILCA. Most farmers explained that it was
difficult sowing at the close spacings, and even more difficult
maintaining the closely-spaced seedlings under farm conditions.
As a result of the immense and complex web of variabilities arising
from the experimental design (or lack of it) is is often difficult to
make direct comparisons between farms on the basis of any single
factor. The relevance of these variability issues in the pilot alley
farming project are discussed under the various headings:
The decision as to which of'a farmer's farm lands should be used for
alley farming is made entirely by the farmer. While some farmers
planted trees on reasonably 'new' and fertile land, others chose to
plant on 'old' depleted land which was due for fallow. There are
situations where the land used was so heavily depleted that even
cassava could not grow satisfactorily. Under such a range of soil
fertility conditions comparisons between farms on the basis of other
characters can be misleading.
Variability in Cropping Patterns and Rotations
ILCA did suggest an establishment crop of maize (at least for the first
crop season in the trees' establishment) but farmers were free to plant
whatever crop they preferred (see Table 3). This was in line with the
philosophy of minimum interference in farmers' decisions with respect
to the establishment of the alley farms.
Consequently alley farms were established with a variety of crops -
maize, cassava, yam, cocoyam, pepper and melon, usually in mixed stands
of 2 to 5 crops, and in complex relays and rotations. The different
mix of crops, in widely different proportions, on various farms makes
it impossible to use the alley crops as source of classification of the
farms, or to use crop yields as basis of comparison between farms.
This issue of crop yields is further complicated by the ad hoc nature
by which harvesting is carried out on farms.
Variability in management
Perhaps the most important source of variability in this project is
that due to management of the farms. Farm operations are carried out
independently by the farmers, in most cases, in the absence of the
researcher. There is, therefore, immense variability arising from such
management practices as land preparation, date of planting, species and
spatial arrangements of trees and food crops, crop production and farm
maintenance (weeding). Correlations between these various factors and
tree and crop yields (if obtainable) are not expected to yield any
significant relationship on account of the uncontrolled variability in
the various factors.
Problems of Labour Estimation
The question of labour requirement for alley farming is probably best
treated as part of a full economic analysis of the system. This could
be done through economic modeling, using figures from field surveys and
OFR. This paper argues that labour is not a critical issue in the
establishment of an alley farm. This question is addressed below under
various farm activities so far encountered in the establishment and
management of alley farms and considerations are given for the
assessment of additional labour demands in alley farming.
a) Land clearing and preparation
In alley farming, the tree hedgerows are established in the farmers'
food-crop farms and not by themselves as in intensive fodder tree
cultivation systems. Clearing and land preparation costs are therefore
irrelevant as all these would have to be done with or without planting
of the trees. The situation is the same for other land preparation
activities such as ridging and mounding. No specific ridges or mounds
are required for the trees; in fact in situations where a farm is
unridged, the trees are seeded on the flat. No extra labour is
expended in land preparation as a result of converting a farmer's food
crop farm into an alley farm.
b) Planting and management
Tree planting and related activities such as thinning and row-filling
can be regarded as the first activities which introduce an additional
labour input into farm operations. The issue, at stake in this,
however, is the magnitude or significance of the extra labour required
and the returns, from the farmer's point of view, of that labour. Tree
planting to establish alley farms in the humid tropics does not involve
nursing of seedlings, digging of holes in the field and transplanting,
that is required in the establishment of alley cropping farms in the
drier areas (Hoekstra, 1985).
Planting is done by direct seeding of tree seeds along every fourth or
fifth prepared ridge which would be 4 or 5 metres apart. In all
situations where farmers have established alley farms, planting was
done by the farmers themselves, usually aided by their children or some
other members of the family. In no case was labour hired for the
purpose, as is commonly done for land clearing, ridging, and
preparation of mounds.
Even though family labour for tree planting represents a cost, (to the
farmer?) this is believed to be minimal and not sufficient to
constitute a limitation to the farmer in establishing an alley farm.
Furthermore, in all cases, the trees were seeded only after the farmer
had completed the planting of his food crops, thus minimizing a
possible conflict with critical planting time.
Weeding represents the major management requirement in the
establishment of the trees. This is especially critical in the first
three months of seedling growth. In some cases, however, no special
weeding is done for the trees during this period. They benefit from
routine weedings carried out for the food crops. It is recognized that
a farmer who is either not serious or committed enough to keep the food
crops in his farm weeded will not find time to weed and maintain the
trees. In such situations survivability is low and establishment may
During the dry season, and also in situations where a farmer's alley
farm is carrying an old stand of cassava, farms are usually left
unweeded. It has been observed that some farmers make deliberate
efforts to keep the tree rows weeded while not doing general weeding on
their farms. This type of weeding represents an additional labour
input into farm operations and should be taken into account in the
over-all assessment of the economics of the system. This labour
demand however occurs outside the peak period.
c) Tree pruning and utilization
The major management activity following the establishment of the trees
is the frequent pruning of the trees to prevent shading of the food
crops, and for use as mulch or feed for small ruminants. On small
farms, such as exist in the pilot project, and given the freedom and
control exercised by the farmers in the management of the farms, it is
virtually impossible to obtain quantitative data on this regard.
Farmers prune the trees when they feel they have to be pruned, or when
they need the foliage for mulching or for feeding.
In most situations, except at planting of the food crops, pruning is
seldom done in a uniform manner as is the case in on-station alley
cropping/farming work. On farms, pruning is a continuous and staggered
exercise this is especially so when foliage is being used both as
feed and mulch. It is therefore very difficult under conditions in the
pilot project, to establish the relative quantities used as feed and as
mulch or the labour input that goes into pruning for either mulch or
feed. These quantities can however be estimated from work carried out
on station and in controlled researcher-managed on-farm trials.
More of the prunings are used as fodder during the dry season than
during the wet (cropping) season, when they are also used as mulch.
The extra labour involved in cutting for fodder is suspected to be
minimal. This is because the fodder is cut and carried home usually
after the day's work on the farm. During the dry season, farmers
sometimes go to the farms specifically to cut fodder, but the
alternative to this might be a longer walk into the bush to get lower
MONITORING AND EVALUATION PROCEDURES IN PILOT PROJECT
The objectives of this on-farm research/development project have been
met mainly through continuous observations and qualitative assessments
and descriptions of what is happening on various farms, and also
through the use of short and simple questionnaires for obtaining
specific information at different stages of the project. The major
tool of this approach has been observation of responses and reactions
of the farmer as an indication of his own assessment of the system.
The approach has also allowed the identification of problems and
opportunities arising out of the exercise of farmer's initiative, and
helped establish the remaining programme of research to be done on
The alley farms are evaluated periodically by ILCA personnel. In the
evaluation each farm is scored on the basis of the establishment of the
fodder and their management. The scoring scheme used was as follows:
Poor establishemnt, describes farms which have had very low tree
survival (less than 25%) with trees showing signs of stunted. growth as
a result of intense competition with weeds. Farms rated excellent
usually have moderate-to-good survival (50-70%) with trees showing
satisfactory growth, and the entire farm is generally well maintained.
A summary of the trend of the results is shown in Table 4 and suggests
that in general over 50% of the alley farms were well established and
managed. An analysis of the adoption/drop out trend (Table 5) gives a
good indication of'the overall assessment by the farmers, and the
potential acceptability of the intervention.
Farmers' assessments, impressions and experiences gained from the
system are further probed by the use of a simple 'post-establishment'
questionnaire which was designed to find answers to such questions as:
Why did adopters adopt?
Why did non-adopters not adopt?
Why did abandoning adopters give up?
What use is made of the trees?
What problems were encountered with the system?
What solutions do farmers propose for these problems?
For the purpose of this survey, farmer were classified into 3
1. Farmers with poor/abandoned alley farms
2. Farmers with fair/good alley farms
3. Farmers with excellent alley farms
Ten farmers were randomly sampled from each farmer category for the
study. This has enabled some information to be obtained, as to why and
how, from the farmer's own point of view, certain farms had excellent
establishment while others were poorly established. Such information
could then be compared with observations made independently by
researchers on the farms. Details of this comparison are, however, not
reported in this paper.
The alley farming package offers two benefits to livestock. These are
the vaccination against PPR and the increased availability of
high-protein fodder for supplementary feeding. Vaccination is open to
all livestock in the village (Atta-Krah, 1985); it should therefore not
be considered a treatment factor for the purpose of showing the benefit
of vaccination. Response of sheep and goats to vaccination against PPR
has been adequately shown in earlier on-farm work set up specifically
for that purpose (ILCA,1983).
The benefit of browse supplementation, is also not achieved until after
the first year of tree growth. No attempt was therefore made to
monitor farmers' animals during this first year. In January 1985,
following the commencement of tree pruning and browse feeding by
farmers, a monthly inventory was initiated to enable primary
indications of the effect of the alley farming package on flock sizes.
For this purpose ten "Alley farming" and ten "non-alley farming"
households were randomly selected from each of the two villages. There
are three categories of animals in these households: (i) those
belonging to alley farmers, (ii) those belonging to non-alley farmers
living in "alley farming households" and those belonging to non-alley
farmers living in "non-alley farming households". Animals belonging to
all individuals in the sampled households are monitored monthly. A
summary of the inventories is given in Tables 6 and 7 for sheep and
goats respectively. These are only preliminary data and no detailed
analyses have as yet been done. The trials have not run long enough to
enable productivity and flock size differences to be adequately picked
up. A major limitation also in showing animal response to browse
feeding under village conditions in the project, has to do with the
The animals are free-roaming, and therefore graze and scavenge together
in small village flocks. Under such conditions it is difficult to
ensure that browse meant for alley farmers' animals are eaten only by
that group. This is even more so as there are animals belonging to
both alley and non-alley farmers in most households.
While it may be difficult to detect actual changes between animals of
alley and non-alley farmers especially in the short-term, the overall
picture of small ruminant development over time will give an indication
of whether the alley farming intervention has had some effect in the
area as a whole. So far indications are that flock size of alley
farmers' goats is increasing (Table 7). It is not yet clear to what
extent this is a direct result of alley farming package (reduced
mortalities from PPR control, increased productivity from browse
feeding) or an indirect effect through renewed interest and increased
investment in small ruminants following an improvement in the health
and feed situation as a result of alley farming.
SUMMARY AND CONCLUSION
The monitoring methodology and type of data collected in any on-farm
trial should be determined from the objectives of the trial as well as
other site-specific factors and circumstances. In situations where the
major objective is to "measure biological and technical parameters of
proposed interventions on farms", adequate researcher control is
required to enable reliable quantitative data to be obtained. On the
other hand, where socio-cultural compatibility and farmer acceptability
of a new intervention is being assessed, as in the case of the pilot
project, it is often adequate to rely on adoption analysis as an
indication of acceptance of interventions for farmers.
The results obtained so far have been very encouraging with over 60% of
farmers who planted alley farms in 1984 actively managing their farms
and about 45 new farmers planting alley farms in 1985. What has been
seen by some as the major weakness in the approach i.e. lack of
analyzable quantitative data is in fact its strength, as it has
enabled the required involvement and control of the farmer in the
Researchers should be aware of the tremendous potential of qualitative
data for measuring the relevance of proposed interventions for farmers.
Increased emphasis should be placed on evaluation of farmers' own
responses which should be seen as the sum total of all
considerations-technical, economic and socio-cultural.
Atta-Krah, A.N., 1985. A developmental approach to on-farm research:
A pilot project for improving small ruminant production in Humid
West Africa. Paper presented at ICARDA/IDRC Regional Workshop on
Research Methodology for Livestock On-farm Trials, ICARDA/Aleppo,
25 28 March, 1985.
Hoekstra, D.A., 1985. The u.
Council for Research in
se of economics in diagnosis and design of
Working Paper No. 29. International
Agroforestry, Nairobi, Kenya.
ILCA, 1983. Annual Report. 1983. Humid Zone Programme, International
Livestock Centre for Africa, Ibadan, Nigeria.
Kirkby, R., Gallegos, P., and Cornick. T., 1981. On-farm research
methods: a comparative approach. Cornell International
Agriculture Mimeograph 91, New York State College of Agriculture
and Life Sciences, Ithaca, New york.
Okali, C., 1984. Community response to a pilot alley farming project.
Unpublished Programme Document, Humid Zone Programme,
International Livestock Centre for Africa, Ibada, Nigeria.
Steiner, K.G., 1982. On-farm experimentation-some proposals for the
implementation. Paper prepared for the Workshop on "On-farm
Experimentation for Farming Systems Research", 31 May 4 June,
1982, IITA, Ibadan, Nigeria.
Stoop, W.A., 1982. Some aspects and examples of on-farm experiments.
Paper presented at the Workshop on "On-farm Experimentation for
Farming Systems Research", 31 May 4 June, 1982, IITA, Ibadan,
Q. Have you increased productivity and if so, how much and in what
sectors? Without these data, how can you approach a ministry or a
donor to justify large-scale extension of the project?
A. Farmers would not adopt the practice if it were not more
productive. Rate of adoption indicates success and more farmers
want to join the program. Indirect measures such as stock numbers
on trial farms, indicate higher income. Ministries and donors
have been interested and supportive in spite of the lack of this
type of data.
Alley farming has a long experimental history which shows it is a
profitable innovation. Economic modelling has been done with
mixed farm and station parameters; so we are confident that the
system has at least a good chance.
ILCA at Ibadan, Nigeria, has converted a group feeding trial with
browse supplementation to an individual feeding trial to measure
intake more accurately and secure a better estimate of
performance. The problems with on-farm trials and the measurement
of intake of supplementary feed are far more difficult to
Expanding on the above answers, ILCA is not ignoring an essential
element of FSR by supposing that good results in on-station trials are
necessarily achieved when a new technology is transferred on-farms.
Surely FSR requires a feedback process from on-farm trials to assess
the degree of achievement on-farm of the benefits found from on-station
work. Where there is a serious discrepancy in achievement, further
research to explain this is necessary.
We have not taken detailed measures of biological productivity on farms
with alley cropping, because of the difficulty of taking these
measures, and the difficulty of interpretation of necessarily highly
variable data. We have taken the rate of adoption of the technique by
farmers as an empirical indication of its usefulness and benefit to
small farms. We are also monitoring changes in herd sizes of farms who
have taken up the technique.
It is agreed that feedback from FSR is necessary in the development of
technologies. Not only is this realized through the use of
quantitative data, but also through the identification of problems and
opportunities arising from on-farm work through participant observation
and informed interviews of farmers as well as through the use of simple
questionnaires. Much on-station research has been generated this way,
all of which enables us to continue to improve and develop the system.
Variability among farms is not reason enough for failing to compile
quantitative data for comparisons and measuring trial performance among
farms. Farms can be stratified on the basis of the major factors of
variability such as soil, climate,' firm size, household size, wealth,
etc. and comparisons made.
This may be possible for some international projects but not in the
case of the project in question. This is because one can do
stratification on the basis of only one factor at a time, leaving all
the other factors uncontrolled. In situations where farmers have
maximum control, the issue is one of a "complex web of variabilities"
rather than variability in a particular factor.
In response to a query of obtaining economically valid data from
on-farm trials with alley cropping, one can say that researcher
managed-farmer executed trials would be performed in collaboration with
IITA to obtain answers to some of the points raised in the Question.
However, the farmers would be forced, by the trial conditions, to
modify their farming practices, and thus the validity of the results
with respect to normal practices of the farmers will remain
questionable. The variability of the farms will remain problematical
in the absence of the researcher managed trials.
ON-STATION RESEARCH ON DRAUGHT ANIMALS:
AN EXAMPLE FROM NIGER
M. S. Dicko
Animal Nutritionist, ILCA/ICRISAT
Animal traction, an important example of livestock integration into
agriculture, is practiced in many African countries. In Kenya, 12 per
cent of all cultivated land is ploughed by draught oxen (Eicher and
Baker, 1982). In Burkina Faso and in Mali, 20 to 30 per cent of
farmers include animal traction in their cultural practices (Muzinger
Less hard work and a greater productivity are among the incentives
which have led the farmers to adopt animal traction (Casse 1965; Hasif
1978). Currently, the increasing cost of fossil fuel encourages
draught technology development to the detriment of motorized equipment.
In spite of the growing importance of animal traction, knowledge on the
specific needs of draught animals is much more limited than that of
livestock raised for its meat and dairy production (Smith 1981).
Research at the station has, however, covered many related subjects,
improvement of the yoke and harness
development of ploughing instruments adapted to animal traction
development of cultural practices using draught power
studies on power output of different draught animals
breeding of hybrid animals adapted to traction
nutritional needs of animals during different work load and
effects of work on reproduction and lactation capacities
On-farm research has concentrated on testing results from station
studies and examining the socio-economic impact of animal traction
introduced to small farms.
CIPEA-NIGER ANIMAL TRACTION RESEARCH PROJECT PROGRAM
CIPEA-NIGER program objectives
The CIPEA-Niger Program is part of the FSR teamwork of ICRISAT2
Sahelian Center in charge of:
(a) studying the role and contribution of livestock in the economy
of agro-pastoral farming systems in semi-arid regions of Niger.
(b) research and testing of appropriate technology to increase
animal production and the small farmer's income.
Four villages were chosen as study sites in a region where the mean
annual rainfall varies between 400 and 600 mm.
Animal Traction Research Project
The goal of the project executed at the experimental station is to
determine the most suitable ways to introduce animal traction at the
study sites. "Ex ante" analysis of study sites has shown that:
the Sahelian climatic conditions 'are characterized by
soils are predominantly sandy
millet is the main crop
draught power is not used much and the high cost of agricultural
tools and animals is a major obstacle to its introduction
Methodology consists of comparing manual cultural practices to
different types of animal draught practices used in the cultivation of
millet (harrowing, plowing, ridging). Various types of hitch (e.g. one
ox, one donkey, a team of cows or oxen) and of agricultural tools
(hoe-sine, plough, canadian cultivator) are tested. A profitability
study of the various practices is also planned.
An initial trial done in 1984 (a low rainfall period of only 260 mm)
has shown that the use of donkey and a pair of oxen in soil preparation
saved respectively 14 and 40-50% of the time needed to prepare the soil
manually (Table 1). As for the overall work time, only ridging showed
a significant economy of time, 26% of the total time used for the
manual practice (Table 2). The low rainfall affected the yields which
showed no significant difference between the various cultural
APPROACH METHODOLOGY AND CONSTRAINTS TO ON-STATION RESEARCH ON ANIMAL
The on-station research seeks to evaluate intervention packages (team
hitch tools cultural techniques) which must then be tested, in the
field for the most part, and later disseminated through extension.
The approach methodology consists, therefore, of the following:
1. defining research priorities taking into account the needs of
target groups and national policies in order to facilitate the adoption
of the new technologies.
Those research priorities must also include the study of the
nutritional needs of the animals in question. Graph 1, which traces
the curves of draught oxen mean weight changes, indicates that, at the
end of the dry season and during plowing, the animals show a
considerable weight loss reducing their performance.
The study of the different work capacities of the various teams and the
research on their specific needs, if not interfering in the farm work,
should be intensified.
2. establishing research protocols keeping in mind the food
resources and the animals available on the small farms.
4. economic analysis of data, knowing that the small farmer, the
ultimate user of the developed technology, perceives all technologies
in terms of profit.
On-station research can also have constraints:
1. of a financial nature: particularly for some national
institutions which, for lack of funds, are forced to stop their
2. involving human nature: research on animal traction is still
fragmentary because it was initiated by foreign researchers whose
contract is usually short term.
3. of a technical nature: it is often difficult to extrapolate the
results of research done at the station due to the small size of their
The importance of the role of animal traction in the agro-pastoral
farming systems justifies the research done at the station aimed at
developing cultural materials and techniques, breeding and nutrition of
draught animals, and evaluation of draught power. The example of the
studies undertaken in Niger has attempted to define the modes of
introduction of animal traction in a climatically-risky region.
However, in a general manner, the definition of on-station research
priorities and the establishment of protocols must take into account
the needs of the small farmer and the available resources of the target
1. See FAO 1970; Gryseels 1980; Goe 1983; Monnia 1965; Nourrissat
1965; Smith 1981.
2. International Crops Research Institute for Semi-Arid Tropics
Casse, Dumas, Garin, 1965. "Bilan de experiences de culture attelee en
Afrique occidentale d'expression franchise", Guinee exceptee
BDPA/IEMVT Maison Alfort, Paris, France.
FAO, 1970. "L'outillage agricole pour les regions arides et tropicales"
par H.C. Hopgen, Rome, Collection FAO: Progres et mise en valeur -
Agriculture no. 91.
Eicher, C. K., Baker, D. C., 1982. "Research on agricultural
development in Sug-Sahara Africa, an initial survey". MSU
International Development paper no. 1, pp 335.
Goe, M. R., 1983. "Etat actuel des recherches sur la traction animal"
Revue mondiale de zootechnie, no. 45, pp 2-17.
Gryseels, G., 1980. "Improving livestock and farm productivity in the
Ethiopian Highlands = initial results". Addis Ababa, CIPEA,
Systems Study draft.
Hasif, E., 1978. "L'emploi de la traction animal dans les
exploitations agricoles' text presented at the CILSS/IER meeting
Monnier, J., 1965. "Contribution a l'etude de la traction bovine au
Senegal" I. Machinisme agric. trop. 10: 3-25.
Muzinger P., 1982. La traction animal en Afrique, GTZ, Edition
Eschborn, 522 pages.
Smith, A.*, 1981. "Recherches sur 1'energie animal, un domaine
delaisse", Revue mondiale de zootechnie, no 40, pp 43-48.
GRAPH 1. Study of animal traction in small farms in Mall.
Curves of draft oxen means weight changes.
350. group 2 supplemented nutrition
....... group 1 not supplemented
300 group 2
250. *** group 1
Supplementation period /
24 31 29 2 23
APRIL MAY JUNE AUG. AUG.
TABLE 1. Time for soil preparation (in manhours and % of time taken in
Type of soil Working time
preparation MH % of manual time
Manual 28 100
Harrowing donkey 25 14
Harrowing team of oxen 15** 50
Ploughing team of oxen 14** 46
Ridging team of oxen 16** 42
(~) standard deviation
** P = 0.01
tests of variance between draft time and manual time: degree of
TABLE 2. Types of soil preparation.
% of manual
- (~) standard devation
- ** P = 0.05
- *** P = 0.01
DIFFICULTIES IN EVALUATING
ON-FARM EXPERIMENTS WITH LIVESTOCK: EXAMPLES
FROM THE ETHIOPIAN HIGHLANDS
International Livestock Centre
for Africa (ILCA)
The basic objective of ILCA's Highlands Programme is to study ways and
means of improving the overall productivity of mixed smallholder farms
by increasing the technical and economic efficiency of livestock
enterprises. Particular emphasis is given to enhancing the
complementarity of the livestock and crop components in these mixed
farming systems (Gryseels and Anderson, 1983a).
Although to date field research has been undertaken only in the
Ethiopian highlands, it is anticipated that the results and experience
of this research will, in many cases, have direct relevance to other
highland smallholder situations in sub-Saharan Africa. In addition
to research undertaken at ILCA's headquarters in Addis Ababa, field
activities of the Highlands Programme focus on two study areas:
around Debre Zeit, located 50 km south of Addis Ababa at an altitude
of 1850 m, and around Debre Berhan, 120 km northeast of Addis Ababa
at an altitude of 2850 m.
The programme has adopted the farming systems approach to research.
This integrated and problem-oriented approach stresses on-farm
technology testing and approval, complemented by relevant station
research on individual components in cases where greater experimental
control is advantageous (Gryseels et al., 1984). The evaluation of
technology is on a whole-farm basis, though farmer-managed tests are a
crucial part of the approach.
A major part of the activities of the Highlands Programme are
conducted outside the station. Baseline surveys in each of the study
areas have been followed by continuing studies of 'control' farmers in
the traditional farming system of the local Peasants Association (PA)
surrounding both ILCA sites.
These studies of the traditional farming system have been complemented
by on-farm trials of various innovations. On-farm trials have been
initiated on a range of topics, including the use of crossbred cows
and forages for smallholder dairy production, the use of single oxen
rather than the conventional pair for land cultivation, the
construction of ponds and dams using oxen-drawn metal scoops, and the
use of crossbred cows as draught animals in addition to their primary
role as milk producers.
These on-farm trials are important to assess the feasibility of the
new or improved technology under farmers conditions, and in monitoring
and evaluating adoption problems. Because of the special nature of
livestock enterprises, such on-farm trials do not give statistically
adequate information to assess the longer-term impact of new
technology on livestock productivity. As a result, on-farm research
with livestock cannot be a substitute for station-based research.
This paper illustrates this problem using results of on-farm research
at Debre Zeit as a case study.
SPECIAL CONSIDERATIONS WITH ANIMALS IN ON-FARM EXPERIMENTATION
Various authors (Bernsten, 1982; Gryseels, 1983; de Haan, 1983;
Bernsten et al., 1983; Zandstra 1983 and Zandstra, 1985) have
described the difficulties in conducting and evaluating on-farm
livestock research. These problem areas are summarized in Table 1.
Bernsten et al. (1983) list them as problems related to the following:
Mobility of Livestock
Mobility makes it difficult to describe environment-livestock
interactions, to measure and control factors not included as
treatments, and to organize data collections.
Life Cycle Duration
While grain crops typically mature in a few months, the reproductive
cycle of ruminant livestock extends over at least a year. This
increases the timeframe and cost of experimentation, as well as the
risk that experimental animals may die or be sold before the trial is
Life Cycle Synchronization
Crops of similar varieties are planted and harvested more or less at
the same time. Animal production, however, is not synchronized and
occurs at different times and intervals. This makes it difficult to
find animals of the same production categories and in the same
Animals produce several outputs of economic value. These outputs
include milk, meat, manure, draught power and hides. Many of these
animal outputs are intermediate products and are used as inputs in the
crop enterprise. This makes it difficult to measure the impact of
treatments, to evaluate the economic impact of an intervention and to
assess the constraints in the farming system.
Non-market Inputs and Outputs
Smallholder livestock production systems depend on inputs such as
children, crop residues, semen, and water which are difficult to
value, and produce outputs such as draught power and manure for which
there is no ready market. The value of other functions of animals
such as capital accumulation, risk management and ceremonial functions
are difficult to measure.
Size of Experimental Unit
Smallholder farmers have only a few large ruminants. Exposing these
to trials and treatments in the research exposes the producer to
substantial risks. Moreover, a large number of farmers will have to
participate in trials in order to achieve statistical significance.
This increases the cost of on-farm experimentation substantially.
Also, farm comparisons with a control group are rarely possible and
the researcher needs to resort to cross-farm comparisons.
Livestock are subject to various religious and cultural taboos which
makes it difficult to cull, castrate and earmark them.
The management of livestock includes a large number of critical
decisions (feeding, watering, milking, breeding, animal health
control, etc.) which need to be made regularly, often daily, over a
long production cycle. The variability of this management makes it
difficult to attribute the effects of certain treatments given the
number of experimental units.
Number of Observation Units
Livestock performance is measured as production per animal, and as
small farms tend to have only few animals, the statistical variability
of performance within treatment groups tends to be large.
Other factors that could be added to this classification are problems
Ownership of Animals
Many animals are inherited, or managed by people other than the owner
in a benefit-sharing agreement. This makes individual decisions
related to management difficult.
Livestock will often graze on land which is communally owned. This
tenure problem severely limits the scope of on-farm pasture
Livestock are managed by more than one person. The role of women and
children is particularly important. This complicates the organisation
of management of livestock experimentation.
ON-FARM LIVESTOCK RESEARCH AT DEBRE ZEIT
ILCA's initial field research activities in the Ethiopian highlands
centred on a 160 ha site in the Ada Wereda of the Yerer Kereyu Awradja
near Debre Zeit, 50 km south of Addis Abeba. The area was considered
representative of the medium altitude highlands of Ethiopia. A
baseline survey of the Ada Wereda was carried out first, in order to
provide a basic understanding of the traditional farming system. This
baseline survey was complemented by socio-economic studies in the
local Peasants Associations surrounding the ILCA site. Results of
these studies are summarized in T. Makonnen and G. Assamenew (1978)
and Gryseels and Anderson (1983).1
Around Debre Zeit, smallholder mixed farming is the dominant mode of
production. Details of this cultivation system can be found in
Gryseels and Anderson (1983) and Gryseels et al. (1984). Most farm
produce is kept for family subsistence consumption and average cash
incomes are between US$ 150-200 per annum. Rainfall averages 845 mm
of which 70% falls in the main rainy season between July and
September. Farm sizes averaged 2.5 ha around 1980 but with increasing
population pressures have since been declining and now average around
2 ha of cropland. The area is intensively cultivated and virtually no
arable land is left fallow. Teff (Eragrostis teff) is the principal
cereal grown. About two-thirds of the cultivated land is sown to
cereals, with most of the remainder sown to pulses. The main crops
grown are teff, wheat, barley, maize, sorghum, horse beans, chick peas
and field peas. Net grain yields (after deducting seed) average
around 800 kg/ha.
Most farmers own livestock and a typical farm inventory includes two
oxen, a cow and young stock, a few sheep or goats and a donkey.
Cattle are kept mainly as a source of draught power and for manure
which is dried and used as fuel. Productivity of all livestock is
low, reflecting an underexploited genetic resource and generally
inadequate nutrition, particularly during the extended dry season of
up to seven months each year. For example, milk offtake from
indigenous cows kept under traditional management rarely exceeds 400
Major constraints limiting smallholder productivity around Debre Zeit
are low soil fertility on slopes; poor drainage of fertile
bottomlands, limiting grain production; limited wood supplies,
necessitating use of manure as fuel; dry season feed shortages,
causing production losses in livestock; low cash incomes, limiting
investment; variation in work oxen ownership, causing differences in
area cultivated, cropping patterns, and income; and marketing
constraints for livestock products.
Station research at Debre Zeit therefore deals with topics related to:
draught animal utilisation; forage, legume and crop agronomy; soil
fertility; dairy technology; animal nutrition; and valley bottom
development through cheap methods of surface water control.
On-farm trials at Debre Zeit focus on improved dairy husbandry through
the introduction of crossbred cows and forages, the use of these cows
as draught animals in addition to their role as milk producers, and
the use of oxen as singles rather than in pairs for land cultivation.
The results of these trials follow.
Introduction of Crossbred Cows and Forages for Dairy Production
Thirty-four families were originally resident on the land granted by
the Ethiopian Government to ILCA at Debre Zeit. Of these, 18
volunteering farmers were the initial focus of the on-farm trial of
crossbred dairy cows which started in 1978. Each farmer bought a
crossbred (Friesian x Boran) dairy cow, cultivated an average 2.5 ha
of arable land, and had access to a 30 ha communal pasture. The
crossbred cow was to be fed mainly on a special purpose forage mixture
of oats (Avena sativa) and vetch (Vicia Dasycarpa). The productivity
of the farming system was to be further improved by the use of
improved seed and chemical fertilizer on subsistence food crops, and
generally improved farm management.
The 18 'test'farmers adopted this dairy husbandry package at their own
expense and risk. They themselves could decide whether or not to
accept the technology and ILCA's management recommendations. In
return for agreeing to participate in the research programme, these
farmers received long-term (five years) credits for the purchase of
the cow (which cost US$300) and the construction of a shed. Extension
inputs were provided by ILCA until early 1981 and then gradually
reduced to correspond to conditions which would normally apply in a
The performance of these test farmers was monitored by ILCA through
regular visits by a 12th grade enumerator. Initially these visits
were weekly, but after 1981 they were reduced to once every two weeks
to avoid excessive ILCA influence. Literate farmers participated in a
self-recording scheme for milk production. Originally it was planned
to have over 30 dairy test farmers, but this target was not achieved
because insufficient crossbred cows were available from the government
During 1981, the ILCA programme was instead expanded with 13
additional farmers who bought their crossbred cows through other
channels, mainly the Extension and Project Implementation Division
(EPID) of the Ministry of Agriculture. These farmers are referred to
in this paper as EPID farmers. Their farms are located within a 20 km
radius of the ILCA station.
For breeding purposes, ILCA relied initially on the artificial
insemination (AI) service of the Ethiopian Government. When this
proved unsatisfactory, ILCA purchased a purebred Friesian bull for the
station needs, and to serve the cows of the test farmers. As a result
of using a purebred Friesian bull, the calves had 75% exotic blood.
Although only few smallholder farmers have the feed resources and
management ability to handle 75% crossbreds, the available half bred
bulls lacked libido and proved unsuitable for breeding purposes.
There were significant differences in the results obtained before
1981, when ILCA was still providing extension services (weekly visit
of an extension officer, supply of forage seeds and feed concentrates,
animal health care, etc.) and when ILCA stopped providing these
services, limiting its activities to monitoring. There were also
significant differences in the performance of 'test' and EPID farmers
(because of ILCA vicinity) and between 50% and 75% crosses. The
managerial factor was of overwhelming importance in performance
evaluation of the crosses.
Results on average milk production are indicated in Table 4. EPID
farm results monitored from 1981 only refer to the fourth to seventh
lactation as previous data were not available. This illustrates the
problem of synchronisation of experiments. Average values of
lactation yields and lengths for crossbred cows of 'test' farms before
and after 1981 are indicated in Table 5. The differences in
productivity are due mostly to a reduction in the use of feed
concentrate and to increasing age of the cows.
From 1978 to 1984, overall average milk production for all farms per
lactation was 1969 kg (cv 22%), average production for the first 305
days was 1775 kg (cv 19%) and average adjusted annual milk yield
(AAMY) was 1667 kg (cv 23%). An analysis of variance indicates that
these differences are caused by differences between test and EPID
farms, individual farms, and the number and year of calvings. (Table
2). The differences between 50% and 75% cows can almost be completely
explained by year effects. All lactations of 75% cows date indeed of
the period after 1981. An analysis of variance of this period did not
give indications of significant differences between crosses.
High yields of the first lactation are due to the long lactation
period which was caused by the absence of breeding services and the
delay in purchasing an ILCA bull. Farmers kept milking until the cow
stopped producing milk. Overall average daily milk yield was 5.8 kg
(cv 18%) while milk yield per metabolic unit was 20.9 kg (cv 22%).
Lactation Length, Age at First Calving, and Calving Interval
Average lactation length for all cows was 339 days (cv 21%) and the
dry period 122 days (cv 79%). The third lactation of test farms
was substantially shorter than the first and the second (291 days vs
439 and 319 days). This is due to a shortage of feed concentrates
but other factors (forage and pasture production) could have played a
role as well. The dry period of 75% crosses was almost twice as long
as for 50% crosses, even though there was no significant difference
in lactation length. The lack of feed may have caused delays *in the
post partum oeustrus cycle, but this is only a hypothesis.
Average age at first calving was 968 days for 50% crosses and 1016
for 75% crosses. The CV of age at first calving was generally less
than ten percent. On average two services were necessary before
successful conception, though there was high variation in this. Around
36% of cows were pregnant after one service while 68% were pregnant
Average calving interval was 460 days (cv 24%). No significant
differences were observed between test and EPID farms, between farms,
between 50% and 75% crosses, number of calvings or season of
Overall calf mortality up to the age of two years, including
abortions, was 42% for 75% calves and 53% for 87.5% calves. On
test farms abortion was responsible for 21% of mortality: 15% died
during the first 30 days of life; 8% between one and three months of
age; 21% between 3 and 6 months; 12% between 6 and 12 months; and 23%
between one and two years of age. Mortality was slightly higher for
male calves than for female calves. Until 1981 average weaning age on
test farms was 124 days for female calves who consumed an average of
328 kg of milk, and 88 days for male calves who consumed an average of
328 kg of milk. Calf mortality increased dramatically after 1981.
Birth weights of calves averaged 30 kg for 75% crosses, compared to
only 25.5 kg for 87.5% crosses. This difference in birth weights
between both crosses was statistically significant at the 0.1% level.
Mortality of adult cows was only 12%. The reasons for death are
unclear, but liverfluke, skin and tick diseases and diarrhea are
common. Since the enumerators were 12th grade students without
veterinary expertise, the precise cause of death could not be
There is an indirect correlation between rainfall and milk yield. As
the rainy season starts, pasture growth is stimulated and feed
resources improve dramatically. The peak of milk production is about
one month before the peak of the rainy season. Conception increases
dramatically also. Increasing rainfall also increases mortality,
probably because of the higher incidence of ticks, insects and
parasites. (Table 3)
Impact on Socioeconomic Welfare
Gryseels and Anderson (1983b) have evaluated the introduction of
crossbreds on family welfare and found that cash income per farm
increased on average by 227%, although large variations were observed.
These results deal with the period before 1981, and the results of the
later period are presently being analysed.
It was difficult to determine the profitability of the dairy
enterprise. For example, it is almost impossible to determine the
value of a female heifer which is not yet pregnant. The majority of
crossbred animals are supplied by the government breeding ranch and as
yet there is no free market for them. This also complicates assigning
a value to the cow which is sold by the government at a subsidized
Other socioeconomic and welfare factors include the following.
Forages are grown on arable land and there is an opportunity cost in
terms of growing subsistence crops versus forages. This opportunity
cost of the land has to be taken into account. Cattle dung is dried
and used as a fuel in the household. Although there is a market price
for dung, the capacity of the market to absorb supplies is limited.
Most Ethiopians belong to the Coptic Orthodox church which prohibits
the consumption of veal. This makes the valuation of male calves
difficult. Feed inputs are difficult to value. There is no
commercial forage seed production; pastures are communal; straw is a
by-product of the crop enterprise; and the supply of feed concentrates
is irregular. Health care and breeding services are not available
either commercially or from the government, and difficult to value.
Followers of the Ethiopian Orthodox church also observe around 140
fasting days per year, during which the consumption of animal protein
is prohibited. This severely limits the marketing of milk during
these periods; milk prices will then fluctuate widely. Evaluation of
the crossbred cow enterprise is further complicated by its
interactions with the crop enterprise. For example, farmers who grow
forages replace crops with the lowest gross margin. Mostly these are
pulses which make a valuable contribution to soil fertility.
Competition for labour between crop and livestock enterprises has also
been a constraint.
The objectives of the on-farm trial with crossbred cows were to study
the productivity of these cows under farm conditions, to appraise
smallholder adoption problems and to evaluate the impact on the socio-
economic welfare of the farmer and his family. The on-farm trial has
clearly indicated that a genetic treatment (through crossbreeding)
gives significant effects. There are also significant effects of
another treatment, i.e. concentrate feeds. A third is that there are
major interactions between these two treatments. More statistical
work is needed to quantify differences of performance within
treatments. Results are confounded and sometimes difficult to
interpret. The differences in management are large between the
different farmers and it is not easy to attribute the effects of
treatment related to fodder, health, watering, housing or general
management. For example, feed concentrates have a dramatic impact on
milk yields but their supply is irregular. As they are cheap, every
farmer will use as much as possible as long as they can obtain them.
Some farmers have easier access to feed concentrates than others and
this explains the large variation in their use. Out of 31 farmers,
16 grew forages in 1984, while 15 did not because with rapidly
increasing teff prices, they thought it was more profitable to grow
cereals than to produce extra milk. The feed problem is accentuated
by the fact that most farmers still keep cows of local breed for
investment purposes and for supply of oxen.
During the dry season only four farmers watered their cows three times
a day; 17 watered two times a day and eight farmers watered once a
day. Roughly half the farmers never had health problems with the cow.
The other farmers reported only minor health problems. Farmers
identified their most important problems as being the lack of feed,
breeding (lack of suitable bull) and lack of a market for fluid milk.
Out of 28 farmers, 16 let their calves graze on home compound only, 10
grazed calves on the home compound and outside the farm compound,
while only two had a zero grazing system. All of these differences in
management make it difficult to attribute management effects on
The impact of management on the productivity of cows can be further
illustrated through ILCA's experience with a cooperative farm unit. A
20 ha cooperative farm was established in 1978 and was mis-managed by
10 farmers nominated by the local PA. A production package similar to
the one for dairy test farmers was introduced. It included a dairy
enterprise based on 12 Arsi x Friesian crossbred dairy cows and forage
Results of this cooperative venture are indicated in Table 6. These
results clearly indicate the bottom level of productivity of cows when
badly managed. The cooperative farmers had no previous dairy farming
experience and lacked motivation. The venture was discontinued after
Given the reliance of farms on purchased concentrates, one wonders
what the effect would be on milk production and reproductive
performance without concentrates, using only a feed regime of grazing
and straw At the moment the supply of concentrates is irregular and
insufficient data are available to estimate a production function.
Given the high CV's of the various productivity parameters in the on-
farm trial, and the impossibility of attributing the effects of
certain management variables, these technical relations are to be
established in on-station research where greater experimental control
Use of Single Oxen
Ploughing in the Ethiopian highland is traditionally done using paired
oxen, but surveys showed that half of the smallholders in the
highlands owned fewer than two oxen During 1983 ILCA developed a
single ox yoke and harness, and a modified version of the wooden
plough, the Maresha, suitable for use by a single ox of local breed.
On-station testing showed that an adequately fed ox could cultivate
singly in a day 60% to 70% of the area normally ploughed by a pair.
Field days were organized for local farmers from the Peasant
Association around ILCA's research stations. After the field days,
farmers were invited to try the adaptation of the traditional method
at their own farms, at their own risk and expense. ILCA provided
assistance in retraining oxen to work as singles and in teaching the
farmers how to modify the plough.
At Debre Zeit more farmers had two or more oxen than the national
average but around 25% of the smallholders had fewer than two. The
number of oxen owned by farmers at Debre Zeit strongly influences the
area cultivated and the cropping pattern. This also affects farm
incomes substantially. A farmer owning fewer than two oxen has
various ways of overcoming the problem of inadequate draught power.
These ways are described in Gryseels et al. (1984) but generally the
farmer can opt for either renting an ox, exchanging his labour for an
ox or share his ox with another farmer who also has one ox.
During 1983, 12 Debre Zeit-farmers volunteered to try the single ox
system. During 1984, twenty-four additional farmers also started
ploughing with a single ox, while two farmers of the first year left
the trial for reasons not related to the research. Of the 34 farmers
who participated in the trial during 1984, 30 were individual farmers
from 3 different Peasant Associations, and four worked in producers'
Results of the 1984 Single Ox Trial
This section focuses on the results of trials involving 30 test
farmers during the 1984 cropping season. Results of the previous
season are reported in Gryseels et al. (1984).
The average farm'size was 2.3 ha (CV 33%) which was virtually all
cultivated; pastures were on communal land. Of the 30 farmers, two
cultivated less than one ha of land (group I); 6 had between one and
1.9 ha (group II); 17 between two and 2.9 ha (group III); and five had
more than three ha (group IV). The number of oxen owned by these
farmers was almost proportional to the area cultivated. One farmer
had no ox; 12 had one ox; ten had two; and seven had three or more
oxen. All farmers used a combination of single and paired oxen for
the land cultivation. Of those farmers who only had one or no ox, 60%
had a 'minda' (renting) agreement, and 40% a 'Mekanajo' (pairing up
with somebody else's ox) arrangement.
This combination of using both paired and single oxen reflects a
natural caution about the new technique. Around Debre Zeit, because
soils are heavy (black cracking clay-soils, vertisols) and rainfall is
essentially unimodal, cultivation for the main cereal crops has to be
done within a very limited period. Land preparation on the vertisols
begins in mid-June and must be completed by the end of July. It is
very difficult for a farmer to finish all the necessary cultivations
(five) and planting within that period, because each farmer has to
work one or two days each week on communal activities organized by the
PA. In addition, religious restrictions limit the available days for
field work to two or three per week.
Of the mean holding of 2.3 ha, 0.45 ha (or 20%) had been ploughed with
a single ox, while 1.85 ha (80%) was ploughed using the conventional
paired oxen. The area cultivated per farm with the one ox system
ranged from 0.1 ha to 1.1 ha with a CV of 62%. Farmers with less land
ploughed proportionally more with one ox than larger farmers.
However, in each group the CV remained very high (CV=59% for farmers
cultivating less than 1.9 ha, 65% for farmers with between two and 2.9
ha, and 49% for farmers cultivating between three and 3.9 ha).
The area cultivated seemed proportional to ownership of draught oxen
and and family size. Farmers cultivating less than 1.9 ha had one ox
each and an average family size of 4.7; those cultivating more than
three ha had an average of 3.6 oxen and an average family size of
With a single ox it took an average 166 hours to cultivate one ha of
land, or 19% more time than the 139 hours required by a pair of oxen.
This is partly explained by the higher fraction of light soil ploughed
by single oxen, which, surprisingly, took more time than cultivating
the heavy black soil. Table 7 summarizes the results in cultivation
time for both single and paired oxen. An analysis of variance has
indicated that the differences in cultivation time between single oxen
and ox-pairs were statistically significant at the 10% level for the
light soil, but that the differences were statistically not
significant on the black soil.
The average cultivation depth obtained with single ox ploughing was
12.2 cm on the first pass (11.3 cm on the light and 13.1 cm on the
black soil), rising to 15.4 cm on the fourth cultivation (15.3 cm on
the light soil and 125.5 cm on the black). The main crops cultivated
were teff (59% of the area cultivated), other cereals such as wheat,
sorghum and maize (19%), pulses mainly chick peas, rough peas and
horse beans (18%) and vegetables (3%). Only 1% of the land was left
fallow. There was no significant difference in the cropping pattern
of land ploughed by single oxen to that ploughed by a pair. Neither
was there a significant difference in yield for the two cultivation
systems. [Table 8]
The on-farm trial using oxen as singles has shown that the technology
is feasible on a small amount of land. Only a few technical problems
have arisen and the greatest constraint to an expanded use of single
oxen is that farmers have to participate in communal activities two
days a week. These activities consist of ploughing on cooperative
land from the PA, or on land that belongs to the infirm or to members
of the army.
Farmers are not allowed to use a single ox for communal activities as
the Peasants' Association fears that they will not cover their
assignments with only one ox. On-station research is therefore
necessary to determine the maximum potential of a single ox on various
soil types in the Debre Zeit area. Farmers also complained about the
shortage of feed. The performance of oxen under nutritional stress is
as yet poorly understood and these trials are too limited to provide
the necessary technical information. On-station trials need to be
done to determine work output of single oxen of local breeds under
different levels of nutrition.
Cows for Draught
Because of the importance of oxen power in Ethiopia, a high fraction
of the bovine biomass kept by smallholders is accounted for by
oxen. Farmers attempt to produce their own replacements and to this
end, they keep on four or five other cattle. If cows could be used for
draught without a serious loss of performance, then the opportunity
would exist for a substantial restructuring of holdings.
ILCA's research on the use of crossbred cows to improve smallholder
dairy production, has led to increases in livestock holdings. Farmers
were still keeping local cows in addition to the crossbreds for milk
and for the provision of draught. If successful, this multipurpose
use of cows would allow farmers to make more efficient use of feed and
to sell much of their unproductive stock.
During 1984, nine farmers used crossbreds for cultivation. These
farmers were on average 40 years old and had an average family size of
nine. The average size of their holding was 3.5 ha with a range of
1.9 to 5.3 ha (cv 28%). On average 82% of the land was sown to
cerals (mainly teff), 12% to forages (oats and vetch), and 6% to
pulses (horse beans, chick peas). Regionally, more than 30% of the
cultivated land is sown with pulses. Cereals require around 60% more
draught power and labour inputs per hectare than do pulses. Using
cows for traction greatly increases the availability of draught,
hence the greater fraction of cereals.
The test farmers had relatively large livestock holdings. They owned
between two and three crossbred cows and three crossbred young stock,
two local cows, one or two local oxen, and four other local cattle of
mixed ages, a few sheep or goats, two donkeys and four chickens.
These holdings are substantially higher than the regional average. On
average the value of livestock holdings of test farmers was $ 2200 or
more than three times that of a traditional holding.
Six of the farmers did almost all of their cultivation (more than 80%
of the land) with the crossbred cows. The remaining three received
their additional cows late in the season and as a result they used
them to cultivate only 40 % of the total of the land. On average,
around 450 animal-hours were necessary for land preparation and
planting. Cows were worked four hours per day for the seedbed
preparation, but six hours per day during the seeding period.
The crossbred cows were approximately 40% faster than the traditional
cows in cultivating. Farmers achieved a greater ploughing depth ( 17
cm vs 15 cm) and did a larger number of cultivations than with the
traditional method (five or six vs four). Teff yields of cow traction
farmers averaged 1320 kg/ha, against 1000 kg/ha for other farmers in
the area. This, however, is also attributable to their greater use of
chemical fertilizer. Farmers with crossbred cows have significantly
higher cash incomes than traditional farmers and a substantial
fraction of this income is used to purchase additional farm inputs
such as fertilizers.
The cows needed some training before they could be used for draught.
This took ten to 15 days. One method was to pair them with an ox.
Another one was to have them tied and led by a second handler. One
cow was too wild and could not be trained. The number of working
days averaged 40 (ploughing only). Adjusted average milk yields of
working cows over a 305 day period were 1230 kg (cv 32%). On working
days the penalty on milk yield was around one liter of milk, or 15-20%
of daily production. In the cultivation season of 1984, farmers
stopped working their cows only a few days before calving, and worked
them again in some cases the day after. They liked the idea of one
animal providing simultaneously milk, meat, manure, and draught power.
The on-farm trial has shown that the use of crossbred cows for
traction and milk is feasible, and that it has minimal effects on milk
production. Farm income can increase dramatically through an increase
in area cultivated, a higher fraction of land sown to cereals and the
opportunity of greater livestock sales. Farmers could also finish
their cultivation earlier, and use the saved time for other
enterprises. The sample size of the on-farm trials was, however, too
limited to capture the effects on cow fertility. Also, the draught
cows had much lower milk yields than the dairy test farmers and EPID
farmers reported in the first section of this paper. This was because
they were not good cows to begin with, and not due to their draught
On-farm trials gave the researcher clear indications on how farmers
relax certain conditions considered by researchers to be ideal.
Researchers assumed the farmer would stop working the cows 30 days
before calving and for 40 days after calving. In reality, farmers
accepted much smaller time periods. It is also not yet known what the
performance of such cows is under greater nutritional stress.
The practical problems reported by farmers arising with the use of
crossbred cows for draught are mainly related to training the animals.
Although in most cases it took only ten to 15 days, two farmers
complained that it had taken them two to three months before the cow
was fully adjusted to work. Farmers were also looking for practical
ways to plan seasonal breeding to avoid calving during the cultivation
Other problems were not related to using cows for draught but to the
management of crossbred cows in general. These include issues related
to breeding, marketing, fodder production, calf rearing, and animal
Differences in management between individual farmers complicated the
evaluation of using cows for draught. It also seems particularly
difficult to find ways for smallholders to synchronize calving with
period of low work demand. Most importantly, given the small sample
size it is impossible to investigate and. appraise the technical
tradeoffs between work output, milk production, and fertility. This
important gap in knowledge will have to be addressed in a formal on-
On-farm trials with livestock are most useful to assess adoption rates
of new technology, to appraise farmer adoption problems, to evaluate
the impact on productivity and income, and to identify areas for
further station research. On-farm trials are not suitable to
establish technical relations and technical trade-offs between various
variables and treatments. Most sample sizes do not allow for an
evaluation of the impact of certain treatments on long-term
On-farm livestock researchers face a continuous struggle with what is
theoretically desirable, and what is practically possible for these
aspects of sample size. The only practical way out of this difficulty
is for the researchers to lower objectives so that the information
obtained is still useful.
The experience of ILCA's Highland's Programme at Debre Zeit indicates
that the most relevant criterion for success of on-farm trials is
whether the farmer uses the technology and how he modifies it to suit
his multiple objectives. The limited period during which ILCA's
trials have been conducted does not yet allow for an ex-ante
assessment of the impact of these technologies in a larger development
setting without ILCA's direct involvement. Additional experience, is
being gained through a development project jointly set up by ILCA's
Famine Relief Committee and the Ethiopian Ministry of Agriculture. The
project involves 650 farmers who have been supplied with 1 ox, seeds,
and single ox ploughing equipment. Results of this experience will be
of crucial importance for a further in-depth evaluation of the single
1. Within each Ethiopian province the administrative subdivisions are
Awradjas and Weredas. Several Weredas comprise each Awradja.
2. The author is grateful to Tesfaye G. Hanna and Aklilu Assefa for
data collection and to Wagnew Ayelneh and Kristien de Boodt for
3. Average adjusted annual milk yield (AAMY) = (total lactation
yield/calving interval) 365.
4. MW = LW
whereby MW = metabolic weight and LW = liveweight. MW is a unit
which on a comparative basis better reflects the use of energy for
maintenance by an animal.
Bernsten, R. H., 1982. Analytical approaches to farming systems
research with an emphasis on animal production, in J. C. Fine
and R. G. Lattimore (eds) Livestock in Asia: Issues and Policies.
IDRC. Ottawa, Canada.
Bernsten, R. H., Fitzhugh H. A., and Knipscheer, H. C., 1983. Livestock
in Farming Systems Research. Paper presented at the Farming
Systems Research Symposium "Animals in the Farming System."
October 31-November 2,1983. Kansas State University.
DeHaan, C., 1983. Towards a framework for pastoral systems research.
In pastoral systems research in sub-Saharan Africa. Proceedings of
the Workshop held at ILCA, 21-24 March, 1983. Addis Ababa,
Gryseels, G., 1983. Livestock in farming systems research for
smallholder agriculture: experiences of ILCA's Highlands
Programme. Paper presented at the Seminar on Agricultural
Research in Rwanda. Kigali. 5-12 February, 1983.
Gryseels, G., and F. M. Anderson,
livestock productivity in the
Initial Results, 1977-1980. ILCA
Gryseels, G., and F. M.
Newsletter 1 (3):
1983b. Research on farm and
Central Ethiopian highlands:
Research Report No. 4. Addis
Anderson, 1983a. Farming Systems Research in
Programme. Farming Systems Support Project
7-10. University of Florida. Gainesville,
Gryseels, G., and F. M., Anderson 1985. The use of crossbred dairy
cows as draught animals: experiences from the Ethiopian Highlands.
Paper presented at the ICARDA/IDRC Regional Workshop on Research
Methodology for Livestock On-Farm Trials. ICARDA. Aleppo, Syria.
Gryseels, G., Abiye Astatke, F. M. Anderson, and Getachew
1984. The use of single oxen for crop cultivation in
ILCA Bulletin No 18. ILCA. Addis Ababa, Ethiopia.
McDowell, R. E., 1972. Improvement of livestock production in warm
climates. Freeman & Company. San Francisco, California.
Zandstra, H. G., 1982. Experiences with research on crop-animal
systems. In Research on crop-animal systems. Proceedings of a
workshop held at CATIE, Costa Rica.
Zandstra, H. G., 1985. The design and testing of
technology for mixed farms. Discussion paper
FSSP Networkshop number 1 "Animal Traction in
Perspective." Lama Kara, Togo.
prepared for the
a Farming Systems
Table 1. Comparison of characteristics of crops and livestock and
implications for on-farm testing
Situation with respect to
Nonmarket inputs Few
Difficult to measure
and control non-
Increases costs, and
likelihood of losing
Difficult to find
Difficult to measure/
Difficult to value
Increases cost, risk
Difficult to cull,
Difficult to isolate
from Bernsten et al (1983)
Table 2: ANOVA of Average lactation yield (A), 305 day yield (B),
AAMY (C), milk production per day (D), lactation length (E),
and milk production per metabolic unit (F)
A B C D E F
Between EPID and
1 0.0001 0.0001 0.0027
Between individual 27 0.0001
0.0001 0.0001 0.0001
Number of calving 6 0.0067 0.0031 0.02
Year of calving
1 0.0001 0.0001 0.0005 0.0001
R2=0.83 R2=0.84 R2=0.8 R2=0.83
Table 3: Monthly milk production, conception, calving and calf mortality
on test farms, and monthly rainfall (Average 1978 1984)
Month Rainfall (mm) Milk No. No. Calf
production conceptions calvings mortality
January 13 183 5 11 1
February 11 158 9 16 4
March 72 187 16 10 2
April 42 187 11 17 4
May 69 197 16 9 2
June 105 220 10 5 3
July 217 209 17 1 3
August 194 176 9 4 6
September 91 156 5 4 7
October 24 155 1 5 1
November 149 4 9 1
December 169 4 16 4
Table 4. Average lactation yields and lengths of 50% and
crossbred cows of 'test' farms and 'EPID' farms
Parameter Test farms EPID farms
50% 75% 50% 75%
average yield, kg 2219 1552 1741 1576
305 day yield, kg 1990 1397 1603 1450
annual yield, kg 1858 1212 1538 1174
yield/day, kg 6.5 4.5 5.5 5.1
lactation length, days 347 340 320 310
dry period, days 102 98 95 188
Table 5. Average lactation yields and lengths before 1981 on 'test'
Parameter Period before 1981 Period after 1981
average yield, kg 2433 1536
305 day yield, kg 2175 1389
annual yield, kg 1987 1284
yield/day, kg 7.0 4.6
lactation length, days 355 332
dry period, days 101 157
Concentrate use, kg/year 1086 339
Table 6. Productivity of crossbred cows under.cooperative management
Parameter Mean value
yield/lactation, kg 785
305 day yield. kg 781
annual yield, kg 776
calving interval, days 419
lactation length, days 221
dry period, days 198
yield/day, kg 3.5
mortality of 75 %
calves to 1 year 60
Table 7. Area cultivated and time need for ploughing by single and
paired oxen at Debre Zeit, 1984
Number of oxen soil type area no of time needed cultivation
measured for cultivated passes (hours) time
work (m2) (hours/hectare)
one black 38374 5 543 142
light 72671 7 1305 180
Total 111045 1848 166
two black 483179 8 6456 134
light 113258 8 1815 160
Total 596437 8271 139
Table 8. Cropping pattern and gross crop yields of single ox farmers
at Debre Zeit, 1984
Crop cultivated area Yield
(% of total) grain straw
teff 59 984 1646
wheat 14 1083 1701
sorghum 3 405 362
barley 2 1435 2292
rough peas 4 475 496
chick peas 11 609 619
horse beans 3 597 1073
vegetables 3 na na
fallow 1 na na
Q. Can you elaborate on the use of cows for traction and the effects
on milk production?
A. Nine crossbred cows were used for traction-six were used to plow
all the farmers' land, three came late and only plowed 40% of the
land. The animals worked four hours per day on seedbed
preparation and six hours per day during planting. The crossbred
animals plowed deeper and carried out more passes over the land
than local oxen. Milk production was 1230 kilograms over 305 days
compared to 1990 kilograms per non-working crossbred cows. Local
cows produced 400 kilos. Animals were worked up to calving and
returned to work only one to two days after calving.
Q. Please explain the reasons for the high rate of calf mortality
(42%) for up to two years of age? When did the majority of deaths
A. Most deaths occurred before one year.
Q. In Asia and parts of Africa, oxen having gone through the dry
season are too weak to plow the land and thus planting is delayed
as much as one month. How important is this problem in most
A. Very important, but less so in Ethiopia because the dry season is
Q. Due to small size of sample, rigorous statistical analysis was not
possible. Were accounting techniques which do not require
statistical rigor used for some quantitative analyses?
A. Yes, partial budgeting and programming techniques have been used
in analysis of data.
Q. Do you have any empirical data on adoption rate of the crossbred
cows around the Debra Zeit site? The test has been going on since
1978, and given the high milk production of the crossbred cows,
the farmers should adopt if they have found it acceptable.
A. No adoption survey has been conducted. Although farmers are
interested in buying crossbred cows, the supply has been the
Q. I am puzzled by the statement that farmers' selection of shallow
soils for use of single ox plows is a result of uncertainty over
their capabilities. Do they make their decision based on
short-term observations, or do they want to see long-term effects
on the animals? If the former, why are they still uncertain?
A. Farmers are mostly concerned with short-term capabilities of the
technology. However, in any introduction of a technology, it
takes some time for farmers to satisfy themselves as to how it
should be used.
Q. Is it appropriate to include 21% aborted calves in calf mortality
figures? Calf mortality should be based on live births.
A. It is generally agreed that the abortion figure should not be
included in calf mortality.
We need to be-careful in using crossbreeds of stock for on-farm trials.
Such cross breeds may still need a higher level of managerial input
beyond the target farmers as symptomized by high mortality .rates.
There may be need to further cross the stock to get a breed that is
more environmentally adaptable. It's still more productive than the
local stock, that requires a lower level of management than the first
crossbreeds or it might be better to simply select from local breeds to
The chairman asked me to comment on the issue of "where research ends
and extension begins." It is my feeling that researchers usually end
their research too early. It should be the responsibility of the
researcher to carry the research up to the point where it is clearly
shown that a so called "improved intervention" is workable and
practicable for the farmer who is supposed to use that intervention.
This phase should be carried out on farms with close collaboration of
the farmer and the extension agency. It is only after that phase is
successfully completed that the job of research is successfully
Pastoralists in Nigeria (sub-humid zone) do not keep animals for
plowing. The castrates are sold to farmers since the pastoralists see
no purpose in fattening animals for plowing. The farmers keep one to
three castrates, having sufficient resources to feed them. Because the
castrates are well fed, pastoralists in turn hire the oxen from the
farmers for their own farming. Farmers prefer the local breeds to
crosses for management ease.
Experiments conducted in Niger show that one can know, with sufficient
precision, the quantities of feed that are fed. But the animals also
ingest feed coming from the household as well as from pastures. In
such cases, the nutrients are hard to quantify. All this makes for an
appreciation of the impact of feed supplementation of on-farm animals,
and some of the difficulties involved.
ON THE ALLEGED DIFFICULTIES OF ON-FARM
RESEARCH WITH LIVESTOCK
Livestock Economics Unit
International Livestock Centre for Africa
Addis Ababa, Ethiopa
Several papers (Bernsten, 1982; Gryseels, 1985) argue that livestock
on-farm trials (LOFT) are inherently more difficult than crops on
farm trials (COFT). They adduce various reasons for this, such as
the costs of monitoring a continuous production process and the
extended period over which animal production occurs. The increased
costs are said to result in lower levels of statistical confidence
about the inferences made from the research, because of the smaller
samples which can be measured at a given cost.
This argument confounds two-issues. The first issue is the
comparison of a crops research program to a livestock research
program. The second is the comparison of experimental treatments in a
livestock trial. The first issue is of choices in a research
portfolio, once the decision to undertake on-farm research (OFR)
has been made. It is improbable that a system would ever be faced
with a strict choice since the sequences of events leading to OFR
in crops or livestock would not often coincide. The second is
relevant only when the decision to undertake LOFT has been made.
Then there are genuine problems in deciding which methods to use and in
executing the field work and the analysis. But these problems are
different from those in choosing between COFT and LOFT. The
distinction between the two issues is necessary for appreciating the
real magnitudes of the difficulties in this kind of research.
The argument also ignores an important characteristic of research
costs- the marginal costs of adding sample units to a survey or to a
trial are low. I have shown previously (McIntire 1984) that the share
of internationally recruited professionals and of capital (e.g.,
vehicles and computers) in research costs per sample unit in farm
management studies in West Africa is almost two-thirds of the total.
Since such costs would not increase with reasonable changes in sample
size, only one-third of the total would be affected by changes in
sample sizes needed to accommodate livestock trials. Costs affected by
increasing sample sizes are those of low-paid enumerators and
inexpensive materials, such as seeds, fertilizers, and bags.
The argument has two other shortcomings. It misses the point that
much of what is relevant to improved animal production is crops
research. For example, some of the papers presented in the recent
ICARDA/IDRC workshop on this topic were reports of forage trials in
which animals were not involved or in which they were not the only
component. Second, it is also true that emphasis on the costs of
non-treatment error ignores the possibility of different benefits
from COFT and from LOFT. Relative benefits are a different issue, but
they do need to be considered.
PROBLEMS IN LOFT
The following is a brief analysis of difficulties noted in the
literature, in particular of 11 problems cited by Bernsten and by
Gryseels. It refers to examples of the problems, to actual
solutions from the literature, and to logical solutions.
Gryseels argues that this "makes it difficult to describe
environment-livestock interactions, [and] to measure and control
factors not included as treatments." Animal mobility is a source of
non-treatment error. Such error is not unique in livestock research,
because crops research has similar sources of non-treatment error.
It is only if such error is greater in LOFT than in COFT that its
costs would affect the choice of research program (the first issue).
Relevant to the second issue-of the analysis of LOFT treatments-
animal mobility introduces random error into the dependent variable,
and it increases the costs of monitoring animal behavior. The
principal factor related to mobility is pasture quality. This is
an issue of pastoral systems in which animals travel long distances
over pastures of different quality. In high population density areas
animals are less mobile; therefore, the costs of monitoring them are
lower and potential variation is reduced. This error is a source of
inefficiency in the estimate, but it is not a source of bias
unless the error caused by mobility is correlated with one of the
Life Cycle Duration
This is the problem of the duration of animal production, and of the
temporal pattern of reproduction. In choosing between COFT and
LOFT, it is important because the costs of monitoring herds or
individual producers for many years are high. The place to resolve
this problem is not on farms, but on research stations where the
long-term costs can be reduced. If the research problem is to
estimate the reproductivity effects of a treatment, then this can be
done with experimental herds more accurately than with producer
herds. If there are sources of farm variation which are hypothesized
to affect reproductivity, then these can be analyzed with simulation
models (von Kaufmann, 1984) showing the potential effects of
treatments and guiding the decision maker about the potential value
of LOFT and the research methods necessary to realise it.
Specific sources of life cycle error in LOFT cited by Gryseels are
the "risk that animals may die or be sold before the trial is
completed". There are two kinds of risks: covariant and
specific. Covariant risks affect the whole sample: for example, an
epizootic. Specific risks affect sample units: for example, an
accident to an animal. These risks are greater than with station
research involving animals, but one should not exaggerate their
importance. Mortality can be estimated by surveys before LOFT are
done, so that the sample size can be adjusted accordingly.
Adult mortality in African cattle tends to be under ten percent per
year. A covariant risk in LOFT, such as a disease outbreak, has
analogies in COFT, such as a drought. Second, covariant risks are not
correlated with the treatments, so that they do not introduce bias
into inferences made from the results (the second issue). They may
destroy the experiment, but the risk of this is not necessarily
higher than in crops (the first issue). It is only if there is some
specific risk related to the sample unit or to the treatment that bias
is introduced and this can be discovered by the scientist. If it is a
specific risk (e.g., a farmer sells an experimental animal), then
it is probably unrelated to the treatment and simply results in
the loss of an observation.
Life Cycle Synchronization
Gryseels argues that "crops of similar varieties are planted and
harvested more or less at the same time [but] animal production is
not synchronized..." The result is that it is "difficult to find
comparable units." On the first issue, crop synchronicity is not
necessarily a cost advantage. The asynchronicity of animal production
spreads out the work load-there is not the peak involved in setting
out yield plots and harvesting them as there is with crops work.
There are also pronounced planting date effects in crops which are
less reversible than in animals. If a crop is planted late then it
cannot catch up; whereas if an animal loses condition in the dry
season, then it can regain it in the wet season.
On the second issue, asynchronicity is mainly a problem with cattle,
since the reproductive cycles of small ruminants are short. (For
example, one productivity index for ILCA goat data is based on kid's
weight at 150 days). Even for cattle, this difficulty is somewhat
exaggerated. There are many studies of cattle reproduction in
Africa (see Otchere, 1984) with large sample sizes across many
cohorts and birth seasons. For example, if milk production is the
dependent variable, then the sample size in one year is limited
by the calving rate. Even in pastoral conditions, this is close
to 50 %, so that there would be sufficient pregnant females to
measure treatment effects within comparable groups. One study in
India used 27 calves and achieved statistically significant results
by "blocking" supplemented animals by age and by weight (Agarwal
and Verma, 1982). It is also true that many interventions are
designed to alleviate peak production constraints, such as dry or
plowing season feed shortages. Since such constraints are binding in
briefer periods, then the asynchronicity problem is less
It is argued that multiple outputs make economic evaluation of
animal research more difficult. Such outputs include milk, meat,
manure, draft power, and hides. This problem is a real one, but it is
not like the others, since it is not a problem of detecting treatment
effects but one of assigning values to the treatment coefficients.
It is hard to imagine that this difficulty would induce a research
manager to choose a crops research program over a livestock program
and it is therefore largely relevant to the second issue.
It is also a problem whose impact on the second issue is somewhat
exaggerated. Meat and hides are available only once per animal in
well known markets, so there is no problem in measuring their values
or in monitoring their changes over a long period. Milk and manure
are available continuously and manure is not generally traded, so
that there are monitoring and valuation problems. However, since
the principal goal in LOFT is to test the marginal effects of
improved technologies, the main dependent variable with some
measurement difficulty is milk, since measuring marginal changes in
manure output due to some treatment would not be a major research goal.
(This is not to say that manure is not an important output-it is to
say that its value is low compared to that of milk or meat, and
that its value is probably not differentially affected by an
experiment.) The valuation problem would not apply to milk since it
is a traded good; admittedly some milk is consumed at home, and does
pose a valuation problem, but the marginal quantity produced is
probably going to be sold. The effect of draft power is not
impossible to measure and to value, but some of the technical
difficulties in doing so (e.g., cultivation depth) are such that
station research is probably required.
Gryseels argues that smallholderr livestock production systems depend
on inputs such as child labor, [and] crop residues [...] which are
difficult to value". Many of these inputs, especially child labor,
are also used in crop production, so that the first issue would not
always be raised by the valuation problem for non-market inputs. It
is again hard to believe that difficulties in evaluating such inputs
would make a research manager reject a LOFT program.
Under the second issue, since the problem is evaluating different
livestock production technologies, we are still mainly concerned with
marginal effects. Unless use of crop residues or of child labor is
significantly affected by different treatments, it can be taken as a
constant. If one evaluates feeding crop residues against mulching,
then there is a valuation problem. However, because feeding residues,
as opposed to grazing them, is more appropriate to high population
density areas, then fodder markets can develop. (This was found in
ICRISAT's sorghum work in India). Where such markets have not
developed, then it is possible to survey current uses and to decide
how to value them. Where residues are left in the field, then it is
legitimate to assume that they have no alternative value.
Size of Experimental Unit
Gryseels argues that "Smallholder farmers have only a few large
ruminants, and exposing these to trials and treatments exposes the
producer to substantial risks in participating in the research."
This is the problem with block (e.g., farmer) effects. Similar block
effects in crops work-e.g., a disease affects one farmer's plots,
and not another's-are estimated by replicates within a block. Such
replication is said to be more difficult in animal work. There are
two problems here. One is the causes of block effects, and these can
be estimated by measuring management variables,
not purely random. If one has some real measure of the block (eg,
watering frequency) then it is not necessary to have within-block
replicates if the real cause is measured across farms. If one has
only a qualitative measure of the block -such as a dummy variable
for farmer-then within-block replicates are necessary. The choice
between the real measure and the dummy variable approach is one of
choosing between an approach which requires measuring more variables
for a given sample size and one which requires more replicates of a
given set of variables. These are very different approaches, and the
former does not necessarily involve increasing sample size. The
second problem is the risk of confounding treatment with block
effects, leading to inefficient estimators of treatment effects.
This again reduces the confidence of inferences made. While this is
a genuine problem, it can be alleviated by replication and by
stratification (as done by Agarwal and Verma), and it is mainly a
problem of cattle. As noted above in the discussion of life cycle
synchronization, the cycle of small ruminants is short and farmers
might have sufficient animals to enable within-block replication
(Tully et al., 1985). This is less likely to be true for cattle.
However, one study in Lesotho used within-farm replicates in a study of
a winter lick for oxen (Molapo, et al., 1982).
Different farmers have different resources, some of which (e.g., water,
crop residues, small children, or cash) allow better nutrition or
management. This fact creates two problems. One is sample
heterogeneity, and the other is sample size. In a random sample,
heterogeneity might be troubling, since the sample may include units
drawn from more than one population, and increasing size might do more
harm than good. In a volunteer sample, it is probably less important
(and this hypothesis is supported by the nine cow traction farmers
discussed in Gryseels 1985) because of selectivity bias in getting
farmers with similar resources. In this case, there will be no
treatment effect because of lack of variation in the treatment, not
because of too much variation in the dependent variable.
if the block effect is
It is argued that livestock producers have attitudes toward their
animals which differ from the attitudes of crop producers toward their
crops and that these attitudes make some treatments difficult or
impossible in LOFT. The examples cited by Bernsten are of eartagging,
culling, and castrating experimental animals. Although I have only
done a superficial review of the literature, I have found no examples
of the latter two being proposed as treatments in smallholder
LOFT in developing countries. Eartagging has posed practical
difficulties, but my experience in Niger showed no farmer resistance
in principle if the technique was explained clearly. This does
not mean that mistakes in animal identification do not occur, but
that it is possible to reduce them. These attitudes are genuine
difficulties, but one should not exaggerate their importance.
It is argued that there is significant management variability in
animal work; the tacit argument is made that this variability is
greater in livestock research than in crops research. This
argument has been discussed above in the sections about mobility, life
cycle synchronization, and size of experimental unit. It is agreed
that daily care must be monitored. However, since the two principal
treatments in LOFT are health and nutrition, it is difficult to
believe that there would be such negligence in the application of
one, that the effect of the latter would be masked. If, for
example, concentrate feeds are being tried then presumably farmers know
that water is essential. If some health intervention is tried,
then farmers know that nutrition is important. This knowledge
would not eliminate all variation in the other variable, but it should
eliminate extreme variation, or variation due to farmers' execution
errors in the application of treatments. This does not eliminate
the problem of different resource bases, leading to different
management practices. That problem is not specific to LOFT, and it
can be resolved by choosing homogeneous strata in the sample
so as to have sufficient numbers of observations in each stratum.
Ownership of Animals
It is argued that joint ownership and therefore joint management of
animals constitute a difficulty. This is mainly relevant to the
first issue, because owner consent is a condition for participation in
LOFT. Once consent is gained then ownership problems should not affect
incentives to apply treatments correctly. Consent presumably depends
on prior resolution of the ownership problem within the management
unit. The example cited by Gryseels of joint ownership is
essentially that of animal entrustment, which is more common in
pastoral areas than in smallholder systems. There are incentive
problems in technology adoption by managers who are not owners,
but they are not problems in estimating the returns to
technology, only in analysing the distribution of those returns.
On the second issue, if it were hypothesized that ownership
characteristics, such as animal tenure, affected the outcome of the
trial, then it is possible to test that hypothesis explicitly by
including different tenure classes among the participants.
Gryseels argues that joint ownership of resources, such as
pastures, "reduces motivation." It is unclear if it is meant that
the resource attribute problem complicates the interpretation of the
results because of poorly applied treatments, or whether it
affects the motivation for farmers to participate in the first place.
If the latter is meant, then this is again an incentive problem,
which is relevant to the first issue. If there are common property
questions, such as communally held pastures, then one would have
to question the wisdom of doing LOFT on such properties. This
is a well-recognized fact in the economic literature, and it only
means that on-farm pasture experiments must be just that-carried out
on farms which are privately held. If, for example, pastures are
communal, then farmers or herders would be unable to set aside
for pasture improvement trials or they might
the worst areas, figuring that such areas are
This is a particular version of the animal ownership problem. The
difficulty is that target populations in a trial are hard to identify
because of joint ownership of animals. It is partly a research
resource allocation problem, and partly an incentive problem.
Presumably the issue of the target population is determined by
the needs and priorities of the research institute. While it is true
that "livestock are managed by more than one person" (Gryseels,
1985), individual animals are usually owned by one person alone,
which makes identification of the target population less complicated.
Bernsten's original list of the difficulties in LOFT was adduced as a
set of reasons why research programs have been unwilling to use a
valuable technique. Though I do not find those reasons convincing,
they still stand as reasons why more of this kind of work has not
been done. That is to say, research managers believed in them
misguidedly. What can be done about this ? First, there should be a
review of loft in smallholder conditions. I have only made brief
reference to such work and someone who knows the literature could do
better. Second, explicit cost functions for research need to be
specified. Just as I previously argued that the cost advantage of
rapid appraisal methods of rural economics research was
exaggerated, I think that the cost disadvantage of LOFT vis-a-vis COFT
is also exaggerated. (This is only a hypothesis-I have not proved
this.) Third, the justification for LOFT, as opposed to livestock
research on station always needs to be made very clearly. In the
stampede to do on-farm work, it is possible that higher quality work,
better done on stations, will be trampled.
Agarwal, I.S., and Verma, M.L., 1982. Experiences in on-farm
research and application of by-product use for animal feeding in
Asia. in Kiflewahid, B., Potts, G. R., and Drysdale, R.M. By-
product utilization for animal production. Proceedings of a
Workshop on Applied Research held in Nairobi, Kenya, 26-30
September, 1982. Ottawa, Ontario. IDRC.
Bernsten, R. H., 1982. Analytical approaches to farming systems
research with an emphasis on animal production. in Fine, J.C.,
and Lattimore, R.G. (eds) Livestock in Asia: Issues and
policies. Ottawa, Ontario. IDRC.
G., 1985. Difficulties in evaluating on-farm experiments
livestock: Examples from the Ethiopian Highlands.
presented at this Workshop.
McIntire, J., 1984
P. J. (ed).
in the development
and rural economics
Coming Full Circle:
of technology. Ottawa,
Molapo, M., Motjope, L., and Klosterman, E.W., 1982. Value
lick fed to cattle under Lesotho farmer conditions.
Networkshop on Draft Power and Livestock Feeding
and Southern Africa. Nairobi, Kenya.
Otchere, E. 0., 1984. Traditional cattle production in
zone of Nigeria. Paper presented at the ILCA/NAPRI
livestock production in the subhmid zone of Nigeria.
of a winter
Tully, D., Thomson, E.F., Jaubert, R., and Nordblom, T., 1985. On-farm
trials in northwestern Syria: Testing the feasibility of annual
forage legumes as grazing and as conserved feed. Paper presented
at ICARDA/IDRC Regional Workshop on "Research Methodologies for
Livestock On-Farm Trials." Aleppo, Syria.
von Kaufmann, R., 1984.
The expected impact and future of the ILCA
Paper presented at ILCA/NAPRI Symposium on
in the Subhumid Zone of Nigeria. Kaduna,
Crop on-farm trials (COFT) can be done without much attention to
livestock on-farm trials (LOFT) but not the other way around, which
makes LOFT necessarily complex.
We need to understand the complementarity that may exist between crops
and livestock before LOFT is undertaken.
Concerning the need to have substantial amounts of capital tied up in
LOFT, I would argue that a case properly made in the overall research
process, most researchers concerned with LOFT should be able to secure
animals from national stations. Most, if not all, African states have
large numbers of animals which could be used for LOFT.
Q. On designing on-farm trials, do we need to think as farmers or
researchers? And do we look for convincing results or
statistically significant results? This is a question of
conceptual approach to on-farm trials?
A. There are biases in subjective measurements. When one goes for
"convincing results" one seldom can go for objective measurement.
In this case the design of on-farm trials should be done in such a
way that statistical results are obtained.
I would agree that the difficulties with LOFT are sometimes
exaggerated. The problem cited can be solved, and has been solved, by
some practitioners already. However, I do not think sample size should
be increased without increasing the scientist's time commitment. If
the additional farmers are only contacted by enumerators, a major
benefit of all on-farm trials, that is scientist-farmer contact for
increased understanding, is not increased.
I believe it would be more profitable to discuss the differences rather
than trying to claim LOFT is more difficult than COFT or vice versa.
Milk is different from sorghum-it can be fed to calves or humans. Its
value rises according to family:herd size ratio. A big herd vs. a
small family means that more milk is left for calves or is sold as
butter. Milk is sold in combination with grain. The price varies
seasonally. This is compounded by dilution variations. Also milk is
Sorghum also has such complications as to labor arrangement. Payment
in kind is common for sorghum. It can be used for consumption, for
sale or for labor.
That is my point, they are very different, not necessarily more
To get livestock owners to cooperate in on-farm studies will depend on
the type of study. At the descriptive stage, their cooperation is a
function of their confidence in the researcher. But when the studies
involve the application of methods which (according to the owners)
might lead to damages to their animals, then their cooperation becomes
more difficult. For instance, we have had no problem in Niger because
we are at the descriptive state. However, in Mali, in order to use the
fecal collection bags, we had to buy the animals from the owners with
whom we had good relations.
SOCIOECONOMIC METHODOLOGIES IN THE DUAL-PURPOSE
GOAT FARMING SYSTEMS RESEARCH IN WESTERN KENYA
A. W. Mukhebi
and A. J. DeBoer
Small Ruminant Collaborative Research Support Program
Ministry of Agriculture and Livestock Development
P.O. Box 252, Maseno, Kenya
Morrilton, Ar 72110-9537, U.S.A.
Department of Rural Sociology
University of Missouri
Columbia, MO 65211, U.S.A.
This paper discusses socioeconomic methodologies applied in the farming
systems research of the Small Ruminant Collaborative Research Support
Program (SR-CRSP) in western Kenya. The SR-CRSP is concerned with
research on types of goats with good milk- and meat-yielding
characteristics suitable for small-scale farmers and their households.
Such "dual-purpose goats" (DPGs) would provide extra milk and meat to
enhance family diets and provide cash income in the form of off-farm
SR-CRSP is supported by the Kenya Ministry of Agriculture and Livestock
Development (MALD), the United States Agency for International
Development, Winrock International, the University of Missouri,
Washington State University, Texas A&M and the University of
A farming systems research (FSR) approach is applied by a
multidisciplinary team of scientists in pursuit of program objectives.
Team members collaborate with one another through the particular
projects they represent. These projects include: Nutrition and
Management, Feed Resources, Animal Health, Animal Breeding, Systems
Analysis, Agricultural Economics, and Rural Sociology. Each project is
headed by a resident scientist and staffed with counterpart scientists
from MALD who acquire practical laboratory and field training as a
prelude to further local and overseas training for advanced degrees in
their areas of specialization. The emphasis on training serves another
important objective for the SR-CRSP, which is to develop the
institutional capability of the MALD for FSR.
For logistical reasons most of the Breeding Project work is conducted
at 01 Mogogo, a MALD station close to Naivasha (northwest of Nairobi in
central Kenya). Similarly, research in the areas of systems analysis
and animal health is based in Kabete, at the facilities of the National
Veterinary Laboratories just outside of Nairobi. Otherwise, a
substantial amount of the Program's activities are located in western
Kenya, in that part of the Lake Victoria basin covering Kakamega
District in Western Province, and Siaya and Kisumu Districts in Nyanza
Those working in western Kenya include scientists representing the
Nutrition and Management, Feed Resources, Economics, and Rural
Sociology projects. Site work is also provided for under the Health
Project (disease monitoring). Headquarters for the Program's western
Kenya activities is the Maseno Veterinary Station, which lies a short
distance northwest of Kisumu Town, on the Lake. Offices, a small
laboratory, and stock barns are located at Maseno, as well as a flock
of crossbred (Toggenburg X East African) DPGs kept for research and
multipication purposes. The station is also the site of extensive
forage experiment fields. On-farm work takes place in smallholder
communities representing different physical and socioeconomic
environments in the area surrounding Maseno.
The Economics and Sociology projects were built into the Kenya SR-CRSP
from its beginning in 1978 and have operated continuously since then.
Socioeconomic contributions to the overall research effort include:
initial characterization of the western .Kenya study area;
identification of constraints to potential adoption by small-scale
farmers; of a DPG enterprise collaborative work with biological
scientists in the design and implementation of station and on-farm
trials of DPG technologies; and regular evaluation of the latter.
In this paper, socioeconomic methodologies are discussed under three
sub-headings, namely: small farm systems surveys, station and on-farm
trials, and short term studies.
SMALL FARM SYSTEMS SURVEYS
Agricultural economists and socioligists were involved in small farm
systems surveys of the target area during the period November 1980 to
October 1981 in collaboration with agronomists, veterinarians, and
animal scientists (Matthewman and DeBoer, 1982; Nolan, 1982). Overall
objectives of the surveys were to characterize basic features of the
target area and to identify principle constraints to DPG production on
small-scale farm as a basis for designing and implementing relevant FSR
directed at alleviating these constraints. The surveys were
administered in two stages: baseline survey came first, followed by a
long-term monitoring survey.
The sample frame used for the small farm systems surveys (and which is
still being used for SR-CRSP on-farm trails) is part of a nationwide
stratified sample previously used by the Kenya Ministry of Finance and
Planning in its first four Integrated Rural Surveys. This frame
stratified the entire smallholder area of the country into various
agroecological zones. Further stages of stratification led to the
identification of "cluster" groups of approximately 200 households.
Finally, twenty households were randomly selected to represent each
cluster (Sands et al, 1982).
The SR-CRSP surveys made use of four clusters, containing 80 farm
households. Two clusters are in the agriculturally high potential zone
of Kakamega District; the second pair is situated in the medium
potential zone of Siaya District. (SR-CRSP activities, however, were
subsequently discontinued from one cluster in Siaya district due to a
problematic enumerator who had damaged rapport with respondent
Arrangements were made with the Central Bureau of Statistics (Ministry
of Finance and Planning) to provide enumerators to the Program and post
them to each cluster. The enumerators were selected on the basis of
their fluency in local languages and they took up residence within or
close to their assigned clusters. Before assignment, they underwent a
short training course at Maseno in order to become familiar with the
survey objectives and methods. These same enumerators continue to
serve the Program and have done so in subsequent surveys and other
activities in the clusters.
The baseline survey was conducted using a single visit questionnaire
composed of several modules. The household module provided information
on household composition, land and capital resources, cropping
patterns, and livestock types and numbers. The labour module yielded
data on household labour supply and demand. The sociology module
identified general outlook and attitudes of the people surveyed as well
as local beliefs and practices relating to goat keeping and goat
products. Finally, the animal management and health module provided
data on existing livestock husbandry, health practices and problems.
The monitoring survey was based upon information derived from the
household baseline survey. It lasted a complete year divided into
thirteen 28-day lunar cycles. According to Sands et al (1982), the
lunar cycles have a number of advantages over the more traditional
calendar months. First, all cycles start on the same day of the week
and the time between individual farm visits is theoretically the exact
same length. Second, biases resulting from always visiting the same
household at the beginning or end of the month are eliminated. Third,
it was possible to arrange a household visit schedule for each
enumerator that remained constant over the entire period of the survey.
Thus, every enumerator knew which questionnaire module to administer to
which household on any given day.
The monitoring survey recorded changes on each farm since the previous
28 days. Information recorded included: changes in household
composition, land and capital resources, livestock inventory and
disease incidence, status of field crops and individual animals,
purchases of farm inputs, labour use in various farming activities,
animal feeding and milk production, quality of livestock feeds, and
prices at local livestock markets.
Principal Constraints to DPG Production
Data from the small farm systems surveys were analyzed by the SR-CRSP
team of researchers including economists and sociologists.
Socioeconomic constraints to the production of DPGs on smallholder
farms in western Kenya were identified as follows: land, labour, and
capital scarcities; inadequate livestock marketing infrastructure; and
prejudices against goats and goat products. Biological constraints
were cited as: lack of adequate quality feeds on a year-round basis;
susceptibility of goats to diseases and parasites; and lack of a
suitable goat genotype capable of producing surplus milk (beyond kid
requirements) for human consumption.
Lessons from the Surveys
A number of lessons were noted from the small farm systems survey.
First, it did not necessarily have to take a complete year to delineate
principle constraints to DPG production by small scale farmers in
western Kenya. A rapid appraisal approach might have achieved the same
results in shorter time at lower cost. However, and second, the
relatively long time taken for the initial survey provided ample time
and opportunity for several scientists with different academic
backgrounds to interact and appreciate each other's special
contribution to the entire SR-CRSP program more fully. This may have
facilitated the subsequent design and implementation of station and
on-farm trials in a more collaborative interdisciplinary manner which
has been the hallmark of SR-CRSP activities to date. Third, the
multidisciplinary approach led to a more complete identification of
technical (biological) as well as socioeconomic constraints and more
thorough characterization of existing farming systems in study areas
than it would have been possible with a mono-disciplinary approach.
STATION AND ON-FARM TRIALS
Component (disciplinary) research designed to resolve the identified
constraints to DPG production has been conducted both on-station at
Maseno and on-farm in clusters since 1982 following the small farm
systems surveys. The socioeconomic input has consisted of development
of analytical models for economic analysis of technical biological data
from component trials of prototype DPG technologies, economic analysis
of the data, and sociological assessment of farmers' reactions and
views to the trial DPG technologies which include DPGs, new feeds, and
management and health practices.
Participation in Biological Experiment Design and Data Analysis
Economists collaborate with biological scientists in preparing research
protocols for biological station and on-farm trials. Participation by
economists ensures that relevant variables are included in the design
of biological trials to facilitate subsequent economic analysis of the
technical data that they generate. When the trials are completed, the
Economics Project extracts the required data for economic analysis.
The Economics Project has developed and is developing quantitative
models for economic analyses of whole-farm systems in cluster ares as
well as individual component research trials.
Linear Programming Model
A whole-farm linear programming (LP) model was developed from the small
farm systems survey data and updated with data generated from station
and on-farm trials (DeBoer et al, 1982; 1983). The LP model was
constructed for various farm-size labour categories. As is typical of
most LP models, the model maximizes gross margin from specified farm
enterprise activities subject to various constraints. It attempts to
capture the more important elements of the farm's decision environment.
The basic structure of the LP model includes farm enterprise
activities, constraints, and input-output coefficients. Several crops
and livestock production, consumption purchasing and selling activities
are specified. The constraints include monthly labour supplies, land,
livestock feed supplies, human nutritional requirements, goat
nutritional requirements, and cash availability. Input coefficients
reflecting output productivity, or input requirement levels for each
activity in the model, are also specified.
The important feature of the LP model is the embodied concept that a
study of all parts of a farm system is necessary for understanding the
complete system, or solving problems that stand in the way of its
improvements. The model recognizes that a change in one component of
the farm system may cause or require a change in one or more of the
other system components as they compete for limited farm resources at
The LP model was used to assess and simulate optimal resource
allocation under existing farming systems in cluster areas and under
systems with DPG technology and government policy intervention. The
impact of DPG technology on farm incomes and welfare was simulated by
incorporating various DPG.genotype-management practice combinations of
activities in the model. Most promising enterprises were identified as
basic activities in optimal LP model solutions.
The Economics Project has formulated two accounting models for
quantifying costs and benefits of various component DPG technologies
and management practices, namely the gross margin model and the partial
The gross margin (GM) model has been used to determine relative
profitability of various farm enterprises. It involves quantifying
products of each farm enterprise and their farm-gate prices, as well as
variable inputs used by the enterprise and their farm-gate costs.
Enterprise GM is then calculated as the difference between the value of
enterprise products and cost of enterprise variable inputs. Enterprise
GMs are computed per unit of land, labour and capital inputs as
measures of short-run enterprise resource returns.
The GM model has been used by the Economics Project in developing
input-output coefficients for the LP model, and for preliminary
comparisons of the profitability of the DPG enterprise with other
competing farm enterprises in clusters (Mukhebi, 1984).
The partial budget model has been used for economic analyses of
selected component DPG technologies and management practices. It
involves quantifying incremental costs and benefits of a test
technology or practice over a conventional one, and determining the
marginal net benefit or benefit-cost ratio (absolute marginal rate of
return) of the test technology or practice. The benefit cost ratio or
marginal rate of return is then used as a choice indicator for ranking
and identifying economically most promising DPG technologies or
management practices. Examples of partial budget analyses undertaken
are summarized below.
a) Economics of Intercropping Forage Crops with Maize in Kaimosi
Cluster (Mukhebi and Onim, 1985). This analysis quantifies
economic costs and benefits of intercropping selected forage
crops with maize assesses the impact of applying fertilizer
to maize both in pure stand and intercropped with forage
crops, and identifies economically most promising
maize/forage intercrops for the target area.
b) Economics of Pre-weaning Kid Feeding Practices at Maseno
Research Station (Mukhebi and Sidahmed, 1985). This
analysis quantifies economic costs and benefits of various
pre-weaning kid feeding practices, assesses the impact of
supplementing milk with forage in such practices, and
identifies economically most promising practices.
c) Economics of Internal Parasite Control Among Goats (Mukhebi
et al, 1985). The analysis quantifies economic costs and
benefits of controlling worm infestation among goats and
assesses the economic performance of different breeds of
goats under both infested and uninfested conditions.
d) Economics of Simulated Pre-weaning Kid Rearing Practices for
the East African Goats in Hamisi, Western Kenya (Mukhebi et
al, 1985). This analysis quantifies economic costs and
benefits of various simulated kid rearing practices, assesses
the impact of improved forage in such practices and
identifies the economically most promising practices.
Lessons from Economic Modelling and Analysis of Biological Trials
Formulation of economic models and their subsequent applications to the
analysis of results of biological trials has yielded several notable
lessons. First, the input from other collaborating disciplines is
indispensable in specifying input-output (technology) coefficient
levels that reflect more accurately both the real world and simulated
farming systems performance. However, and as a second point, other
scientists appear to be somewhat apprehensive of economic evaluations
of the results of their research trials. They seem to fear that
economic analysis of their results will reveal them to be
"uneconomical". As one researcher put it, "just analyse my data but
for heaven's sake don't tell me my work is uneconomical".
Nevertheless, when these scientists see results of economic analysis
and their interpretation, they appreciate economic input more, and
become more cooperative and less fearful of the "butcher", as one
colleague characterized the economist. A third point bears on some
discontinuity in the Economics Project work due to staff turnover. The
researcher who constructed the initial LP model simply went away with
it when he ceased working with SR-CRSP. This meant that another LP
model had to be formulated from scratch to continue with the whole-farm
evaluations and simulations of the evolving DPG technologies and
management practices. The lesson to be drawn here is that as staff
turnover is a regular feature of FSR, careful documentation of all
research activities is vital for program continuity. This would save
time, effort and funds that would otherwise be wasted in repetition of
Sociological Participation in On-Farm Trials
The Rural Sociology Project seeks to contribute to the SR-CRSP on-farm
work over the long-term by using both formal and informal observational
techniques to monitor research activities in the clusters, including
farmers attitudes and opinions towards DPG trials and associated work
in feed development, management, and health practices.
Participant Observation by Field Enumerators
The Rural Sociology Project has been a component of the SR-CRSP in
Kenya from the beginning, but there was no long-term resident scientist
posted until late 1982. Since that time, the project has worked very
closely with the Program's field enumerators.
In the process of living and working in their respective cluster areas,
the enumerators have become known and accepted by the local residents.
Regular, long-term interaction with these people has given enumerators
an intimate view of community life. Communication is no problem since
the enumerators are native speakers of the languages common to their
areas. All of these attributes combine to make enumerators ideally
suited for the role of "participant observers" in the various study
"Participant observation" is an approach commonly employed by social
researchers, especially social anthropologists, when involved in
long-term community studies. The technique is self-explanatory: the
investigator serves at once as an "actor" and a "witness" of community
life. A participant observer operates informally to gather
"qualitative" (as opposed to more formal and "quantitative")
information on various aspects of people's behaviour and beliefs. The
approach is open-ended, and frankly impressionistic and opportunistic
in nature: casual conversations, chance remarks, visits to
respondents' homes and farms, encounters at market centres or
neighbourhood meetings, etc., are all potential sources of interesting
and revealing anecdotes about the way people live and think about
themselves and their environments. Although it may sound very
elementary and perhaps trivial (after all, people everywhere are in a
sense "participant observers" of their social settings), participant
observation as a research tool demands a fair measure of discipline:
careful records of observations need to be made, and doubts and
questions that arise need to be followed up. Moreover, one must
develop a strong degree of sensitivity and attentiveness in order to
pick up relevant "data". In time, even though participant observation
is part and parcel of everyday life, it can also be refined as a
conscious and deliberate component of fieldwork endeavours.
SR-CRSP field enumerators have been encouraged to develop their skills
as participant observers by maintaining field journals for recording
their experiences and impressions of, and thoughts and comments on
people's behaviour and attitudes. Of particular concern have been such
matters as gender roles and age gradations in the division of household
and farm labour, livestock and crop management practices, land use,
*off-farm employment, and the stock and forage activities SR-CRSP
researchers carry out in the study areas. By reviewing their notes
with Sociology team members from time-to-time, and participating in
evaluation and write-ups of observations collected, the enumerators
have been able to enlarge the appreciation of local people's
priorities, problems, and needs an appreciation vital to successful
development of alternative smallholder production strategies.
Participant observation and regular journal recordings by the
enumerators complement the periodic use of formal interviews with
farmers in sociological investigations of ongoing work in the clusters.
To cite one example, there is ongoing work being done in conjunction
with the feed resources project in testing different forage producing
alternatives in farmer's fields. Sociological input to the study is
based on routine observations by the enumerators as well as results of
formal questionnaires administered in the communities. In addition to
monitoring in a more structured way farmers' reactions and preferences
vis-a-vis the Feed Resources trials and interventions, data are
collected through the questionnaires on past and continuing methods of
feed/forage production in the study communities, and reasons underlying
changes in people's patterns and preferences are identified. By
placing the Feed Resources trial alternatives in the context of
established practices, the Sociology Project may be able to offer
better informed commentary on the feasibility of these alternatives to
the Feed Resources team.
Lessons from Sociology Participation in On-farm Trials
The involvement of sociologists in technical on-farm work in the
clusters has been of mixed issue. Of special concern are the
difficulties of effectively communicating to non-sociologist colleagues
ideas and recommendations derived from observational work based on
"soft" evidence derived from open-ended interviews or the recording of
casual events, visual and mental impressions, etc. Partly this can be
attributed to differing styles and modes of data collection which exist
between the social and biological or technical sciences. Technical
scientists are, by virtue of their training and research orientation,
most comfortable with "hard" facts and figures derived from closely
controlled experiments and statistical inference. Thus, experience in
the Program has shown that when the presentation of sociological
findings is couched in less discursive, more formal terms, where highly
structured data collection techniques and numerical measures are
emphasized, the problem of communication and acceptance by technical
colleagues is considerably eased.
But it seems that the problems of effective communication go deeper
than this, judging by some of the reactions to sociological inquiry and
commentary that have been forthcoming. There have been times when
biological scientists have been positively irritated with sociologists,
regarding the letters' work as meddlesome and a violation of
professional "turf". Observations and suggestions which are intended
to be helpful may be taken in just the opposite fashion, even when the
attempt is made to present them tactfully. One may be met with extreme
defensiveness, with the attitude of what do you know about these
things anyway?" The role of the sociologist therefore has not been a
particularly easy one to fulfill, especially on those occasions when
one bears tidings that technical scientists are not pleased to hear -
perhaps not even wanting to hear in the first place. In retrospect, it
would have been wise to have spent more time with colleagues in the
early stages of the Program in order to better acquaint them with the
procedures and style of sociological work, and its attendant
SHORT-TERM SOCIOECONOMIC STUDIES
The SR-CRSP Economics and Sociology Projects have conducted a number of
studies in clusters and surrounding communities aimed at generating
socioeconomic data to supplement data from station and on-farm
biological trials. These studies provide a better base for assessing
potential farmer adoption of DPGs.
Socioeconomic data are collected from primary and secondary sources.
Primary sources include surveys and case studies of target farmers,
local agricultural extension officers, local businesses and other
relevant respondents such as community leaders. Secondary sources
include farm records and public documents, especially from the Ministry
of Agriculture and Livestock Development.
Prior to data collection, a research protocol for a new study is
prepared in collaboration with the other scientists. The protocol
includes the experiment number, title, place of study, names of
principle and participating scientists and support staff (including an
estimate of each individual's time input), the justification for the
study, study objectives and hypotheses, procedures to be followed
(including variables about which data are to be collected), an activity
chart, and an itemized financial budget.
For each study, a questionnaire or an interview guideline is prepared
for data collection. The questionnaire or interview guideline is
designed in a way that enables all relevant data to be collected in as
short a time as possible.
Trained enumerators are employed to assist with data collection. They
are made to clearly understand the purpose of the study, what data are
to be collected, the design of the questionnaire or interview
guidelines, and how to ask or interpret the questions to respondents.
Before primary data sources are used, as much of the required data as
possible is extracted from secondary sources using the questionnaire to
save time and cost. The partly pre-filled questionnaires are then used
by the scientists and enumerators to collect data from the respondents.
For some studies, one day's visit to a respondent is adequate to
collect data based upon the respondent's memory and available records.
For other studies, enumerators may visit respondents two or three times
weekly, for an extended period of several months or through a complete
crop year, with less-frequent visits by principle scientists for
supervisory purposes. Examples of socioeconomic studies undertaken are
a) Some Relations Between Farm Characteristics and Herd/Flock
Health Indicators on a Sample Farm in Western Kenya (DeBoer
and Mortimer, 1983). A socioeconomic assessment was carried
out on factors influencing current types and severity of
animal health problems and measures, and animal management
and herd/flock health. The aim of this study was to assist
in the planning of an effective animal health program for the
A farm survey was conducted in Kakamega and Siaya districts
in mid-October 1980 and late January 1981. Three locations
were randomly selected from each district; three
sub-locations were randomly selected from each location. A
list of farmers who owned goats was constructed for each
sub-location with the assistance of local veterinary
officers, animal health assistants and assistant chiefs.
Farmers were then visited until the pre-selected sample size
for each sub-location was filled. A socioeconomic
questionnaire was administered in conjunction with regular
sampling of animals carried out by the Animal Health Project.
b) Consumer Acceptability of Goat Milk Products in Western Kenya
(Boor et al, 1983). The aim of this study was to determine
whether consumer acceptability of goat milk products would be
a constraint to the establishment of a DPG production system
in western Kenya. Seventy secondary school students were
requested to sample four fresh dairy products: goat milk, cow
milk, goat cheese, and cow cheese. Following tasting, the
students assigned each product a score between 1 and 5 (1 =
excellent, 5 = distasteful). No significant difference was
found between the scores for goat and cow dairy products.
c) Problems and Prospects for Small Scale Tick Control Method:
Some Observations on the SR-CRSP Dipping Trials in Kakamega
and Siaya Districts (Reynolds et al, 1983). Since March
1982, the SR-CRSP Nutrition and Management Project has been
conducting field trials with regular dipping of sheep and
goats as a low investment method of tick control in cluster
areas. A number of crossbred DPGs had been introduced on a
trial basis in the clusters and the exercise in tick control
was partly to ensure that the DPGs were kept free of
tick-borne diseases. It was also designed to test the
potential of dipping as a component of small ruminant
husbandry practices. A sociologist, an economist and an
animal scientist visited dipping sites in each cluster with
the assistance of an enumerator. They collected relevant
data through discussions with farmers participating in the
dipping trials and by their own observations of the dipping
d) Analysis of Household Consumption in Western Kenya During the
Short and Long Rains with a Comparison of On-farm Producing
Food and Off-Fam Food Purchases (Nyaribo, 1983; Nyaribo et
al, 1984). The objective of this study was to compare
consumption of on-farm produced foods with off-farm purchased
commodities, and assess the impact of season on the
consumption pattern. A household consumption survey
questionnaire was administered by enumerators to a random
sample of 20 respondent households in each cluster.
Data were collected in April and June 1981, October 1982, and
in April, June and October 1983. Each household was visited
during one week in the relevant months. Respondents were
asked to recall types and quantities of food consumed over
the previous seven days. They were also asked the cost of
purchased foods consumed, or if it was farm produced, to
value it. A range of market prices was collected from
selected markets in cluster areas for verifying farmer values
of farm-produced products.
e) Implications of Family Labour Use for Adoption of the DPG
Enterprise on Small Farms in W. Kenya (Nyaribo et al, 1984).
The goal of this study was to quantify labour supply and
demand on farms with DPGs in clusters and compare labour
requirements between the DPG enterprise and other farm
enterprises on these farms. Data were collected for six
months, March to September 1983, through farmer interviews
using a questionnaire, and were verified by some field
observations of farming activities.
f) A Socioeconomic Evaluation of the DPG Enterprise in Small
Scale Farming Systems in Western Kenya (Mukhebi et al, 1984).
This study provides a preliminary socioeconomic evaluation of
the DPG enterprise in cluster areas. It assesses the
economic benefits and costs accruing to farmers through gross
marginal analysis, and farmers' attitudes and reactions in
regard to the enterprise.
Input-output and attitudinal data were collected from 26
cluster farmers who had been given trial DPGs. Recall data
were collected through single-visit farmer interviews and
conversations with the farmers by economists, sociologists
and enumerators. Gross margins were computed for major farm
enterprises including the DPG enterprise and compared on a
per unit input basis for land, labour and capital resources.
g) Seeking the "Udder" Truth about Goats in Western Kenya:
Conversations with Community Elders (Reynolds, 1984). This
study, as with b) above, assesses consumer acceptability of
goat milk, this time from the point of view of community
elders, who served as key informants. They were identified
on the basis of their long association with the communities
they lived in, the rich stock of personal experience and oral
history they possessed, and their willingness to submit to
interview. The SR-CRSP enumerators assisted the sociologists
in identifying the elders. Fifteen men and women aged 65 to
80 years were interviewed. Formal questionnaires were not
used for the interviews. Instead, a set of interview
guidelines was used to structure conversations. Short notes
were taken during the interviews and later transcribed into
more complete form for review and discussion among the
h) Extension Services and the Smallholder: A Report of
Preliminary Findings (Reynolds, 1984). The aim of this study
was to assess the structure and performance of existing
livestock extension service and determine its potential
capability for delivering DPG technologies to small-scale
farmers in western Kenya. Using questionnaires designed with
the collaboration of the Economics and Animal Health
Projects, the Sociology Project interviewed all livestock
extension and technical personnel in five administrative
division where SR-CRSP had existing or planned study sites.
The personnel interviewed included Veterinary Officers,
Assistant Animal Production Officers, Dairy Development
Officers, Animal Health Assistants, and Junior Animal Health
Assistants. A total of 62 persons were interviewed through
single day visits. Technical data related to disease
incidence and livestock census were collected through
questionnaires left with officers to be returned later. A
second part of the study focused on farmers' experiences
with extension services. A sample of about 100 farmers was
randomly selected from villages in the same five
administrative divisions covered in the first part of the
study. Each respondent farmer was visited at least once, and
follow-up visits were made where necessary. Basic farm
profiles were collected and respondents were asked to relate
their views on the availability of extension services, and
the positive and negative experiences they have encountered.
Lessons from Short-Run Socioeconomic Studies
The short-run socioeconomic studies discussed above have' tended to
cover a wider area than the biological trials which are confined to
cluster areas. One result of this has been increased awareness of
SR-CRSP activities amongst a larger audience of farmers, which in turn
has led to increased requests for DPG trials to be conducted in other
areas. While such a development can be read as a positive
demonstration of general farmer interest in DPG production, it has also
created problems in that farmers develop expectations about
participating in the trials which cannot be satisfied, due to lack of
enough experimental animals for distribution, and lack of personnel to
conduct more widespread monitoring.
With studies conducted in established cluster areas, it has become
apparent that repeated interviewing of the same farmers over a long
period of time has tended to wear down the patience and cooperation of
some of the farmers. Cases of "respondent fatigue" are not widespread,
but it does seem that certain farmers have begun to feel a bit "stuck"
with the research program, particularly amongst those who have been
involved from the early days.
Cases of enumeratorr fatigue" also bear mention. The core of four
enumerators occasionally expressed dissatisfaction with their
circumstances. The enumerators have worked for the program in western
Kenya since its beginning without any opportunities offered to them for
advancement of their careers. This has led to a feeling of "let down",
as they have witnessed more technically qualified MALD staff come and
go on advanced training, locally and overseas. It has also been noted
that when left to themselves in the field, some enumerators arrogate to
themselves more responsibilities in the name of the research program
than their contracts stipulate. This has damaged rapport with some
respondents and led to their non-cooperation with further research
SUMMARY AND CONCLUSIONS
Economists and sociologists have applied a variety of methodologies
that have yielded a number of instructive outcomes. Experience with
the initial small farm systems surveys of the target area indicated
that while similar data could probably have been assembled more
efficiently (with less time at lower cost) using a rapid appraisal
approach, there were nonetheless some advantages gained through the
year-long exercise. The long-term interaction of scientists from
different disciplines and backgrounds provided an opportunity for them
to better appreciate one another's contribution to the entire research
program. Their combined efforts fostered more thorough constraint
identification, and set the stage for collaborative and
well-thought-out design and implementation of the subsequent station
and on-farm trials.
Socioeconomic involvement in the design and implementation of
biological trials ensured generation of more complete technical data
ameneable to socioeconomic evaluation. Mathematical programming and
accounting models, as well as informal and formal social impact
assessment techniques, have been employed. These evaluations
quantified benefits and costs, and assessed farmer acceptability of
alternative DPG technologies and management practices. Short-term
socioeconomic studies generated supplementary data to augment station
and on-farm biological data and provided a better base for assessing
potential farmer adoption of DPGs.
Various lessons are drawn from the experiences of economists and
sociologists with station and on-farm trials and short-term studies.
First, the input of biological scientists is indispensable in
generating more realistic input-output physical data for socioeconomic
evaluations. Second, as staff turnover is a regular feature of FSR,
careful documentation of research activities is vital for program
continuity in order to avoid repetition of work and save time, effort
and funds. Third, generation of farmer awareness of on-farm trials
beyond sample farmers creates a demand for the extension of such trials
that might not be met with available resources. Fourth, prolonged use
of the same respondent farmers and enumerators tends to wear out their
patience and cooperation.
A fifth lesson is that socioeconomic involvement in the FSR has not
always been gladly received by some of the other scientists. Reactions
of apprehension and defensiveness have been noted, due in part to
differences of style and procedure existing between social scientists
and their biological or technical colleagues, and in part to the
protective feelings that are aroused when professional "outsiders"
begin to offer judgements, sometimes not laudatory, about the worth of
one's research efforts. However, over time, as colleagues have
acquired greater familiarity with socioeconomic interpretations and
their significance, collaboration has been facilitated. Still, looking
back on the experiences, it seems that a more deliberate effort should
have been made to allay any doubts and fears harboured by colleagues
about the conduct and motives of their socioeconomic counterparts.
Boor, K. J. H. A. Fitzhugh and E. Ochieng', 1983. "Consumer
Acceptability of Goat Milk Products in Western Kenya". Proceedings
of the Second Small Ruminants CRSP Kenya Workshop, Naivasha,
February 22, 1983, pp. 35 39.
DeBoer, A. J.,
M. Job and R. W. Matthewman, 1982. "Economic Modelling
Farms in Western Kenya". Proceedings of the Small
CRSP Workshop, Nairobi, Kenya, March 15, 1982, pp.91-96.
DeBoer, A. J., M. O. Job and B. A. McCarl, 1983. "Economic Factors
Influencing the Simulated Adoption of Dual Purpose Goats in
Representative Small Farms in Western Kenya". Proceedings of the
Second Small Ruminants CRSP Kenya Workshop, Naivasha, Kenya,
February 22, 1983, pp. 40-48.
DeBoer, A. J. and D. L. Mortimer, 1983. "Some Relationships Between
Farm Characteritic and Herd/Flock Health Indicators on a Sample of
Farms in Western Kenya". Proceedings of the Second Small Ruminant
CRSP Kenya Workshop, Naivasha, Kenya, February 22, 1983, pp.49-63.
Mathewman, R. W. and A. J. DeBoer, 1982. "Economic Constraints on
Purpose Goat Production in Western Kenya". Proceedings of
Small Ruminant CRSP Workshop, Nairobi, Kenya, March 15, 1982,
Mukhebi, A. W., F.
B. Nyaribo, R. H. Bernsten, J. E. ]
"A Socioeconomic Evaluation of the
Small Scale Farming Systems in
f the Third Small Ruminant CRSP
March 5-6, 1984, pp. 95-108.
Reynolds and A. N.
Dual Purpose Goat
Mukhebi, A. W. and J.
F. M. Onim,
the Fourth Sn
Economics of Inl
Ruminant CRSP Kenya
Mukhebi, A. W. and A. Sidahmed, 1985. "Economics of Pre-weaning Kid
Feeding Practices at Maseno Research Station." Proceedings of the
Fourth Small Ruminant CRSP Kenya Workshop, Kakamega, March 11-12,
Mukhebi, A. W., R. R. Chavulimo, F. Ruvuna and F.Rurangirwa, 1985.
"Economics of Internal Parasite Control Among Goats." Proceedings
of the Fourth Small Ruminant CRSP Kenya Workshop, Kakamega, March
Mukhebi, A. W., F. Ruvuna and H. Blackburn, 1985. "Economics of
Simulated Pre-weaning Kid Rearing Practices for the East African
Goats in Hamisi, Western Kenya." Proceedings of the Fourth Small