Front Cover
 Title Page
 Table of Contents
 Expectations and realities in On-Farm...
 Chimanga Cha Makolo, hybrids and...
 Response of four maize varieties...
 On-farm evaluation of short-season...
 Support project for eastern and...

Group Title: Farming systems bulletin
Title: Farming systems bulletin : Eastern and Southern Africa
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00096264/00001
 Material Information
Title: Farming systems bulletin : Eastern and Southern Africa
Series Title: Bulletin - International Maize and Wheat Improvement Center ; 11
Physical Description: v. : ill. ; 30 cm.
Language: English
Creator: International Maize and Wheat Improvement Center
Donor: unknown ( endowment ) ( endowment )
Publisher: CIMMYT Eastern and Southern Africa Regional Office
Place of Publication: Mount Pleasant, Harare. Zimbabwe
Publication Date: 1992
Frequency: quarterly
Subject: Agricultural systems -- Periodicals -- Africa, Eastern   ( lcsh )
Agricultural systems -- Periodicals -- Africa, Southern   ( lcsh )
Genre: bibliography   ( marcgt )
periodical   ( marcgt )
Bibliography: Includes bibliographical references.
Dates or Sequential Designation: No. 1 (Jan./Mar. 1989)-
General Note: Description based on: No. 1 (Jan./Mar. 1989); title from cover.
 Record Information
Bibliographic ID: UF00096264
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 27691521
issn - 0187-828X
 Related Items
Preceded by: Farming systems newsletter

Table of Contents
    Front Cover
        Front Cover
    Title Page
        Title Page 1
        Title Page 2
    Table of Contents
        Page i
        Page i-a
        Page ii
        Page iii
    Expectations and realities in On-Farm research
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
    Chimanga Cha Makolo, hybrids and composites: Farmers' adoption of maize technology in Malawi, 1989-90
        Page 14
        Page 15
        Page 16
        Page 17
    Response of four maize varieties to nitrogen at Lisungwi extension planning area, Mwanza, Malawi
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
    On-farm evaluation of short-season sorghum and fertilizer for smallholder farmers in a semi-arid region of Zimbabwe
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
    Support project for eastern and Southern African training notes/manuals, networkshop reports, other reports and papers published by project staff
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
Full Text

/ 5- JLq

farming systems

Eastern and Southern Africa

No. 11
July-December 1992


farming systems

Eastern and Southern Africa

No. 11
July-December 1992

/5-- 149

farming systems

Eastern and Southern Africa

No. 11
July-December 1992

Dear Readers 111

/Expectations and Realities in On-Farm Research
SRobert Tripp 1

-' / Chimanga Cha Makolo, Hybrids, and Composites: Farmers'
3. Adoption of Maize Technology in Malawi, 1989-91
.-.i' Melinda Smale, with Z.H.W. Kaunda, H.L. Makina,
M.M.M.K. Mkandawire, M.N.S. Msowoya, D.J.E.K. Mwale, and P.W. Heisey 14

.3 I Response of Four Maize Varieties to Nitrogen at
5. ip7Lisungwi Extension Planning Area, Mwanza, Malawi
Wickson Kawonga and Paul W. Heisey 18

09 On-Farm Evaluation of Short-Season Sorghum and Fertilizer
.yfor Small-Holder Farmers in a Semi-Arid Region of Zimbabwe
Cornelius Chiduza, StephenR. Waddington and Mandivamba Rukuni 25

CIMMYT FSR/OFR Support Project for Eastern and Southern Africa: Training
Manuals/Notes, Networkshop Reports, Other Reports, and Papers Published by Project Staff -- 34

A book about on-farm research:
On-Farm Research in Theory and Practice, H.J.W. Mutsaers and P. Walker (eds.) 41


Addresses of CIMMYT Regional Offices in Eastern and Southern Africa:

c/o ILCA
P 0 Box 5689
Addis Ababa
Telex: 21207 ILCA ET

P 0 Box 30727
Lilongwe 3
Tel.: 734118
Telex: 43055 MI

PO Box 25171
Cable: CENCIMMYT, Nairobi
Telex: 22040 ILRAD

PO Box MP 154 or 163
Mount Pleasant
Tel. : 35534
Telex: 22462 ZW

Farming systems bulletin

Dear Readers:

Welcome to this our last Farming Systems Bulletin for Eastern and Southern Africa. The-Bulletin is part of the
CIMMYT FSR/OFR Support Project for Eastern and Southern Africa which closed on 30 September 1992.
We take this opportunity to thank all the agricultural researchers that have contributed to the Bulletin over the
last four years, and those that contributed to the Farming Systems Newsletter before it. Best wishes in your
future work. We also thank USAID for their financial support, without which the Bulletin would not have been

We begin this last issue with a review of the state of the art of on-farm research by Robert Tripp, CIMMYT
Economics Program, Mexico. The review was the lead paper in the 'Workshop on the Impacts of On-Farm
Research' held in Harare recently. Next, Melinda Smale and others from Malawi present findings from a
comprehensive study of the adoption of hybrid and composite maize by smallholders in Malawi. Again in
Malawi, Wickson Kawonga and Paul Heisey examine the agronomy and economics of maize response to
fertilizer in the Mwanza district of the country. Following from diagnostic work presented in FS Bulletin 10,
Cornelius Chiduza and others look at the role of short-season pearly white grain sorghum and fertilizer for
smallholders in Siabuwa, Zimbabwe.

Finally, with the closure of the CIMMYT FSR/OFR Project, we decided to include a list of the many training
documents, networkshop reports, and published papers produced by the project over the years. Single copies
of these are available on written request from our Nairobi office.

With very many thanks,

Stephen Waddington and Paul Heisey
(Technical Editors)

Expectations and Realities in On-Farm Research*

Robert Tripp, Assistant Director, Economics Program, CIMMYT, Mexico


While most national agricultural research budgets are
under increasing pressure, questions are being
asked about how much should be invested in
adaptive on-farm research. At the same time,
national and international agricultural research
organizations and donor agencies have a widening
agenda, including everything from natural resource
conservation to biotechnology, that competes for
attention'and funds.

It is thus appropriate that we take some time to look
back at what OFR has accomplished in the past
decade and a half, to judge the degree to which our
current position matches our o, ginal expectations,
and to look to the future. This paper attempts to
make a contribution to those ends. It begins by
reviewing some of the major expectations that were
held for OFR. It then examines some of the problems
that made the establishment of OFR more difficult
than originally expected. This is followed by a review
of some of the major accomplishments of OFR. The
final sections focus on what we have learned along
the way and the implications for the future.

The Expectations for OFR

There is no need, in 1992, to give a lengthy definition
of OFR. It is widely recognized as a term connoting
adaptive research that is based on an understanding
of farmers' problems and priorities and that aims to
generate technology for well-defined groups of
farmers. It is distinguished by a set of diagnostic and
experimental methods that directs most research to
farmers' fields. But we should recognize that this
definition encompasses a wide variety of approaches
and institutional forms; that it is part of a larger
movement in farming systems research (FSR) that
occupied much donor attention beginning in the
1970s; and that this movement itself grew out of a

long history of other activities related to
understanding farming practices and delivering
adequate technology. The analysis presented in this
paper will focus on OFR, but it should be understood
that the comments apply in most instances to the
wider FSR movement.

The fact that OFR is now a well-accepted part of our
vocabulary is due in part to its ability to address, in a
coherent fashion, a number of common concerns
about agricultural research. A review of some of the
expectations that motivated the establishment of
OFR will help place OFR in context and will allow us
to assess its current position.

Agricultural technology-One of the motivating
factors behind OFR is a belief in the ability of
agricultural technology to contribute to the alleviation
of rural poverty. A major wellspring for this faith was
the success of the Green Revolution, particularly the
spread of new wheat and rice technology in Asia and
other parts of the world. Although the subject of
considerable controversy and debate, particularly in
the early years, the results of the Green Revolution
have clearly demonstrated the potential of new
varieties and new techniques to contribute to
improving the productivity of resource-poor farmers
(Lipton with Longhurst 1989). To a certain degree,
OFR has been an attempt to emulate that success
over a broader range of environments.

Farmer knowledge and initiative-This wide
variety of farming conditions contributed to another
expectation for OFR. The heterogeneity of farming
systems in the developing world would obviously
require research methods that could generate
location-specific technologies. It was also foreseen
that a crucial contribution to technology development
would come from farmers themselves. It was farmers
who were responsible for the elaboration of these
varied systems and who were most aware of their
characteristics and potential. It was expected that

* This paper was first presented at the workshop, 'Impacts of On-Farm Research',. 23-26 June, 1992. Harare. Zimbabwe. The views are those of the author
and should not necessarily be attributed to CIMMYT.

OFR would not only provide recognition for the
creativity of the farmers who had developed these
systems but could also enlist their participation in the
technology generation process. In Norman's (1980)
phrase, OFR was to 'give voice' to farmers.

Strengthening public research and extension
services-Farmers' contribution to technology
generation was to be matched by a reorientation of
government research and extension services. In its
various forms, OFR has always stressed the
necessity of organizing research and extension in
order to efficiently address the needs of a varied
clientele. It was hoped that OFR would be
accommodated in national research and extension
systems in such a way that the flow of information
between research stations and farmers would be
enhanced, in both directions. OFR expressed an
expectation that government services could be
organized to respond effectively to meet the needs of
their clients. There was particular hope that the
approach would make extension systems more

The systems perspective-A fourth factor that
helped to motivate the development of OFR was the
challenge provided by a broad vision of agricultural
development. OFR drew upon many concepts from
systems theory, which provided one of the principal
sources of intellectual excitement as OFR developed
(Brush and Turner 1987). On a more practical level,
many people were motivated by the prospect of
effective interdisciplinary collaboration in pursuit of
improvements in rural welfare. OFR attracted
researchers who believed in looking at agricultural
technology generation as a complex process and
who were eager to break through traditional
disciplinary boundaries.

Rural development movements-These four
expectations (related to new agricultural technology,
rural people's knowledge and skills, the role of
government research and extension services, and
the efficacy of a systems perspective for rural
development) served as motivating forces during the
conception and development of OFR. But it is
important to realize that none of these expectations
is at all unique or original to OFR. Indeed, it is
possible to look at various movements related to
rural development in the past few decades and to
discern the interplay of these same factors.

Figure 1 attempts to summarize the variable
emphasis accorded these factors in a number of
movements related to rural development. The
movements are listed in rough chronological order.
Such a summary does not pretend to be an
absolutely accurate or complete assessment of these
movements, but it does show that the concerns that
motivated the growth of OFR are common in one
degree or another to many other efforts. The
changing emphasis among the factors reflects
changes in fashion and in strategy among donor
agencies, as well as reactions in which an earlier
overemphasis on a particular factor is
counterbalanced in a subsequent movement. It is not
easy to summarize the varying fortunes of these
factors, although a clear trend in the past decade
away from support to government initiatives, as well
as an uncertainty about technology, may be

This analysis in no way assumes that foreign
assistance or development programs are wholly or
even largely determined by the innocent
combinations of such 'pure' motivations as the ones
listed in Figure 1. There are many alternative ways of
looking at these movements. Individual political
motives, as academic or administrative careers are
being shaped, are certainly important in directing new
movements, for instance. On a more general level,
some commentators see foreign assistance as
determined by an 'aid regime', a set of principles and
rules that helps donors maintain control over
recipient countries (Wood 1986). Most of the
movements in Figure 1 would then be seen as either
expressions of the regime (e.g., structural
adjustment) or as brief and largely unsuccessful
subterfuges by more liberal elements in donor
countries for expressing discontent with this regime
(e.g., basic human needs) (Hawkins and Buttel
1989). This paper cannot address the various
elements that go into determining the direction and
character of donor assistance. But we will proceed
on the assumption that it is legitimate and useful to
analyze the expectations of the people who
participate in rural development efforts like OFR.

These are certainly not the only parameters one
could use to describe the various movements in rural
development, but they are important elements in ',he
motivation of individuals and institutions. Each one of
them seems to play an independent role. Although

government and local level initiatives would seem to
be opposites, for instance, government programs
may try to stimulate local activity, as in the case of
community development, while a de-emphasis on
government may reflect faith in local initiatives (as
with NGOs) or merely a faith in the invisible hand (as

with privatization). Similarly, faith in new technology
is tied to a fairly ngrrow'vision of development in the
Green Revolution but is associated with a much
broader view in the Appropriate Technology

Degree to which movement is based upon:

Developing Broad
new Local Government vision of
Movement technology initiative initiative development

Community Development

Green Revolution

Appropriate Technology

Integrated Rural Development

On-Farm Research/
Farming System Research __ 0

Basic Human Needs

Training and Visit Extension

Structural Adjustment

Farmrrer Participation/
Non-Government Organization


Sustainable Agriculture

Note: The size of the circle indicates the relative importance of the factors-n which each movement is based.

Figure 1. Characteristics of rural development movements.

The remarkable thing is that, as interest in these
various factors ebbs and flows, a particular
combination occasionally crystallizes into a
movement that provides organization, energy, and
direction to donor efforts. This is what happened with
OFR/FSR. One thing that distinguishes OFR/FSR in
Figure 1 is that these interests and motivations are
quite well balanced compared to those of most of the
other movements. Moderation is not always a virtue,
of course, but this analysis shows OFR as a
reasonable and responsible approach that expects
contributions from both the government and the
grass-roots level, and places faith in the possibilities
of technology and in the advantages of a broad
vision of rural development. It will be useful to bear in
mind this analysis of OFR as we examine some of its
failings and its accomplishments and consider its
future position.

Unrealized Expectations:
Excesses and Omissions

There is no doubt that the expectations for OFR were
at times unrealistic. The excesses and omissions
characteristic of any movement, combined with
competition from continual positioning for donor
attention and support, have meant that the FSR
movement has been the subject of various doubts
and criticism. It is true that the FSR movement has
been a 'bandwagon' (Simmonds 1991), and it is well
known that bandwagons attract both followers and
detractors. The criticism of OFR ranges from good,
tough questions (Herdt 1987) to sour grapes (Biggs
and Farrington 1991). A recent book on agricultural
policy comes to the following conclusion about the
FSR movement:

FSR became overblown in the 1980s due to the
almost reckless enthusiasm of aid donors for the
concept. Its productivity gains in relation to the
resources committed are unproven, and its
outcome in terms of increased understanding of
farming systems is disappointing. While
enormous energy has been expended in
preparing FSR manuals, methodologies,
guidelines, and newsletters remarkably little
material has been produced on FSR results,
insights, impacts, or lessons (Ellis 1992:235).

This critique is itself overblown, but it represents a
current of opinion regarding the status and
accomplishments of OFR. We need to be able to
acknowledge and respond to such concerns. Early
commentators on FSR foresaw this type of criticism:

Expectations are running high. FSR is regarded
by some as a panacea. But FSR clearly is not a
panacea for solving all the problems facing small
farmers. The hope is that sufficient progress can
be made to sustain FSR's credibility while it
grows, in the face of inevitable disappointments
(Gilbert et al. 1980:83).

Mistaken assumptions about technology--OFR
places considerable emphasis on the ability of new
technology to make a contribution to improving the
welfare of the rural poor. At least three problems
related to this assumption have led to results that are
less than expected. The first is that technology
development for many farming systems has proven
to be more difficult than envisioned. Considerable
research skill is needed to provide technological
change that proves to be consistently superior under
the difficult conditions of many tropical farming
systems. Research in Zambia, for instance, has
shown significant advantages for improved weed
control, yet the labor requirements of extra weeding,
the costs and management requirements of
herbicides, and the variability in field conditions are
such that no technological alternatives have yet
proven feasible (Vernon and Parker 1983). The
research investment required to improve many
farming systems has certainly been underestimated.

A second, related, difficulty has been in providing
technology that makes enough of a difference to
farmer productivity to warrant the investment of
farmers' time and money in acquiring the new
technique. A great proportion of resources in OFR
has been devoted to tinkering with things like
fertilizer or seeding rates whose results are either
modest enough to escape the attention and interest
of most farmers or require more concerted extension

A third difficulty in relying on technological change
has been inadequate attention to targeting. Most
technological change requires an investment in
inputs or the acquisition of additional skills and

knowledge, rather than the mere rearrangement of
household resources. Households will invest in new
agricultural technology only if that investment will
yield a significant return. Households for whom
agricultural production represents a small proportion
of income, or who have access only to uncertain
product or input markets, will not be likely to risk
investing in new technology. Some OFR has been
carried out in areas where dependence on
agricultural production is minimal, or in countries
where the policy environment is not conducive to any
sort of investment in farming technology. This is not
to say that OFR should not be directed towards the
poorest sectors of the rural population, but to
highlight the necessity for a clear targeting policy and
coordination with other government entities
responsible for improving rural welfare, so that
agricultural technology development can be most
effectively directed. A combination of rural
development rhetoric and inconsistent targeting
serves no one.

Overlooking the farmer-OFR also places
considerable expectations on the abilities of farmers
themselves to define problems and to r*ognize and
test solutions. This has led to both excesses and
omissions. In the first instance, there has been a
tendency to believe that a complete description of a
farming system will lead inevitably to the
identification of ways of improving it. In a number of
instances too much has been invested in
descriptions and diagnostic work. This has served to
increase our admiration for the complexity and
rationality of farmers' strategies, but has often not
resulted in any tangible improvement.

On the omission side, practitioners of OFR have
shown a disappointing tendency to pay only lip
service to the role of the farmer in the research
process. OFR, like any other set of research
methods, is subject to being carried out in a
mechanical and unimaginative fashion. This has
been one of the concerns of a considerable literature
on farmer participatory research (e.g., Chambers et
al. 1989) which in many cases has provided very
accurate criticisms of the conduct of OFR.

Institutional capacity-From its beginning, OFR
has been seen as a way of improving the
effectiveness of national agricultural research and
extension programs. Expectations regarding the

responsiveness of government organizations have
often proven to be unrealistic, however. The
weakness of the civil service in many countries,
combined with severely reduced budgets and almost
total dependence on donor funding for agricultural
research, has proven to be a less than ideal
environment for the organizational changes implied
by OFR. Too much time has been spent
institutionalizing OFR as a separate entity, rather
than seeing how it can contribute to the more
effective functioning of the entire research
organization. Shiny new departments and projects do
little good if basic systems for managing research
planning and reporting are not in place. In addition,
location-specific research needs to be organized so it
contributes to developing a critical mass of research
resources aimed at particular problems.

The desirable versus the feasible-Finally, there
have been instances where the breadth of vision of
OFR has perhaps interfered with the pursuit of more
limited but practical objectives. Complete
descriptions of farming systems have often been
accompanied by long lists of recommendations
directed to an unspecified audience of researchers or
policy makers. The systems perspective has at times
led to a confusion of the desirable with the feasible
(Johnston and Clark 1982:15). There is a great
distance between the recognition of a problem and
the development of a practical solution. OFR
practitioners have at times seen themselves as pivot
points for complex policy recommendations, rather
than as important players in the slow and detailed
process of technology generation.

The Accomplishments of OFR

OFR has grown and matured sufficiently so that in
1992 it need not feel unduly threatened by shifts in
donor interest. OFR is a part of many national
research programs in Asia, Africa, and Latin America
(Merrill-Sands et al. 1989). Many extension services
have been able to profit from an OFR approach
(Merrill-Sands and Kaimowitz 1989). OFR is firmly
established in research and extension programs in
Eastern and Southern Africa (Anandajayasekeram
and Rukuni n.d.). The methods and perspective of
OFR are well known in the literature of agricultural
development (e.g., Caldwell 1987, Upton 1987).

OFR is now being asked to pay more attention to
impact assessment. It is certainly true that much
work in agricultural research, including OFR, has not
paid enough attention to monitoring and assessing
outcomes. But as we think about how the conduct of
OFR can be improved by assessing its impact, we
should concentrate on the real value of such an
exercise. We should not be overly impressed by the
current posturing about impact among donors and
others, which often represents no more than an
excuse for shifting their directions to accommodate
the latest political fashion, rather than an attempt to
examine past assumptions, to learn from mistakes,
or to modify and strengthen strategies.

Anyone can ask tough questions about impact. The
only ones who gain, however, are those who are
willing to examine their own expectations and
commitments. There are a large number of
researchers and extension agents who have long
experience with OFR and who have devoted portions
of their careers to improving the effectiveness of
agricultural research. Keeping their expectations and
commitment in mind, it will be useful to provide a
brief review of the major accomplishments of OFR.

Technology generation-OFR has been
responsible for bringing new technologies to
resource-poor farmers. A recent review (Tripp 1991),
for instance, shows that a wide variety of institutions
in Africa, Asia, and Latin America have experience in
generating useful technology through OFR. There is
no question that OFR is able to put new technology
into the hands of farmers. It is unfortunate that we
have not placed sufficient emphasis on monitoring
and reporting the results of OFR programs.

Such reporting is important for at least two reasons.
First, faced with increasing pressure on budgets and
a myriad of alternative proposals on how agricultural
research resources should be invested, national and
international research institutions need to
demonstrate what they have accomplished. OFR
has not done a good enough job of analyzing and
presenting its results, and it is paying the price for
this lack of attention in current budgetary battles.

A second, and perhaps more important, reason for
emphasizing the follow-up of OFR activities is to
make those activities more efficient. OFR defines
research priorities based on an analysis of the

conditions and problems of specific groups of
farmers and it proposes a research agenda to
address those priorities. OFR's advantage over much
conventional research is this ability to define and
justify objectives. But this is only useful if we are able
to come to closure on these objectives: to pursue
them towards location-specific technology
generation; to apply them to the formulation and
pursuit of larger research goals; or to uncover
inadequacies in the original assumptions and thus
reorient research efforts. If the issues of follow-up
and reporting are overlooked, OFR loses much of its

Knowledge of farming systems-Although it is
correct to focus attention on OFR as a method of
adaptive, location-specific research, we should not
lose sight of the fact that OFR has also greatly
enhanced our understand of farming systems and
has contributed towards reorienting some broader
goals of agricultural research. OFR's impact is
greater than the sum of instances of location-specific
technology generation. A few examples follow.

OFR practitioners have long been interested in the
way that farmers view crop varieties of different
maturities (Haugerud and Collinson 1990). Crop
maturity is of course an important variable for plant
breeders, but in a number of instances OFR has
proven to be the necessary link in identifying actual
opportunities. In Tanzania, an early maturing maize
was accommodated in the minor rainy season and
not only addressed food shortages but allowed a
more efficient use of the major rains for commercial
crop production (Ringia 1991). In Ethiopia, an early
maturing maize was used mostly for alleviating food
shortages, but also proved to be useful in new
intercropping combinations and crop rotations
(Negassa et al. 1991). The impact of this work goes
well beyond the specific examples; information from
OFR has made an important contribution to the way
that plant breeders marshall information for setting

Another contribution of OFR has been to show how
plant breeding priorities must take account of the
entire cropping system. In Pakistan, wheat scientists
have demonstrated how crop rotation in rice-wheat
and cotton-wheat systems influences the choice of
technology for the wheat crop. This work has led to a
significant shift away from selecting wheat varieties
for ideal planting dates and toward selecting optimum

varieties for the planting dates dictated by the
management of the entire farming system (Byerlee et
al. 1986).

OFR has also served to broaden researchers' views
of the farming system to include crop-livestock
interactions. Experience from locations as varied as
Ecuador (Cornick and Kirkby 1981), Egypt (Fitch
1983), and Pakistan (Byerlee et al. 1991) has shown
how maize research often must address farmers'
needs for adequate fodder supplies and recognize
that management of the maize crop is partly
determined by the fact that livestock are an integral
part of the farming system.

Other characteristics of farming systems, such as the
rationale for intercropping (Norman 1977), the
importance of labor constraints (Low 1986), and the
role of seasonal food shortages in cropping decisions
(Collinson 1972) are now better understood by a
wide range of researchers. This understanding has
already been responsible for the development of new
technology through OFR, and, perhaps more
important, it forms part of a knowledge base that
informs a wide range of research. OFR cannot be
seen as an isolated research enterprise, and the
information it develops will often be most useful for
long-term plant breeding or crop management
research strategies that go well beyond location-
specific adaptive research.

OFR methods-Another important contribution of
OFR has been the development of improved
research methods. The comment is sometimes
heard that researchers were talking to farmers and
experimenting on their fields long before OFR was
'invented'. That argument entirely misses the point
that OFR has helped provide much more effective
research tools for diagnosis and for experimentation.
OFR is not merely talking to farmers or planting a
demonstration on farm. OFR is a well-integrated set
of research methods. It involves carefully eliciting
information and motivating farmers' contributions for
an analysis of production constraints. The techniques
for collecting and organizing that information have
been improved over the past decade and a half. OFR
also involves doing high quality experimentation
under farmers' conditions. Whether farmers or
researchers take major responsibility for managing
the experiments, decisions on design, location, and
method of analysis involve many difficult
considerations which have become clearer thanks to

the dedicated work of a large number of agronomists
and others conducting OFR. Examples of innovative
methodological advances developed for OFR in
Africa include Anandajayasekeram (1985), Worman
et al. (1990), Neeley et al. (1991), and Mutsaers and
Walker (1991).

What We Learned along the Way

OFR is here to stay. It has proven its worth as a
means of generating technology, as a source of
useful knowledge about farming conditions, and as a
stimulus for effective research methods. These will
remain, no matter what the latest development
fashion, buzzword, or donor demand happens to be.
The future will not be easy, however, and
researchers will want to take advantage of what has
been learned along the way. This section outlines a
few points that seem to be particularly important.

The complexity of OFR-If one reviews cases
where OFR has been successful in providing new
technology to farmers, one striking commonality is
the complexity of the research path. OFR is often
represented in various training manuals as a short
series of steps, starting with diagnosis and ending
with technology delivery, adorned perhaps with a few
feedback loops. The reality turns out to be much
more complicated. There are two separate issues
that deserve attention. The first involves the quality
of the research. Most farmers operate under very
difficult circumstances and the skills required to
develop adequate technological alternatives are
considerable. OFR provides an effective means for
organizing a research program, but it does not
substitute for well-trained agronomists and social
scientists. Table 1 provides a summary of the
technical issues that were examined in a series of
cases where OFR was able to generate new

The second issue is the flexibility required for
successful OFR. In almost no case are the original
priorities and hypotheses of a successful OFR
program maintained through to technology
development. Researchers must be prepared to
adjust their program based on the outcomes of
successive research cycles. One of the principal
sources of such changes is the farmers themselves.
In the majority of cases where OFR has successfully

brought technology to farmers, the technology Table 2 shows how the research programs of
development (not to mention the identification of the successful OFR were continually adjusted during the
research themes and the testing of alternatives) has course of technology generation.
profited from strong farmer participation. The
technologies originally proposed to address particular OFR as part of a larger research strategy-
problems have often been modified or adapted by Another point that needs to be addressed is the
farmers before widespread adoption is possible. OFR recognition of a well-defined role for OFR. OFR will
must be able to encourage this type of participation, be effective only if it is seen as part of a larger

Table 1. Examples of technical Issues in on-farm research

Case Technology Main types of on-farm experiments

Climbing beans for farmers growing
bush beans.

Improved maize varieties, row planting,
improved plant spacing, fertilizer use.

Alley farming of Leucaena and Gliricidia spp.
in yam/maize and cassava/maize fields; use
of tree foliage for feeding small ruminants.

Dry seeding early maturing rice varieties
to intensify cropping patterns.

Insect control and improved plant stand
management for maize planted in rotation
with maize or upland rice.

Improved maize variety.

For beans intercropped with maize: varietal
dissemination, change from broadcasting to
row planting, increased plant density,
fertilizer use.

For maize in a maize-bean rotation: chemical
weed control and/or zero tillage, row planting
and improved plant spacing, elimination of
use of inappropriate fertilizers.

Diffused light storage of seed potatoes.

Diagnostic ('minus-one) trials on production constraints; on-farm
variety trials (farmer-managed); demonstration of basic technology
for growing climbing beans; in second stage, further variety, fertility,
and intercropping trials, and agro-forestry trials (farmer-managed).

Exploratory factorial experiments; screening maize varieties;
density and spatial arrangements; NPK response; timing and
method of fertilizer applications; verification/demonstrations.

Farmer-managed experiments with alley farming; on-farm
experiments on management of alley farms; on-farm animal feeding
experiments (on-station trials used to get information for technology
design early in research process and later to investigate farmers'
adaptations to the technology).

Adaptive trials (extension-managed) on variety, establishment,
fertilizer, weed and insect control; package demonstrated
on large plots in farmers' fields.

Exploratory and verification experiments on plant stand
management; plant protection with insecticide included in
exploratory, insect control, and verification experiments; experiments
on N dose and timing, P dose and variety.

Farmer-managed verification experiment to examine new variety,
phosphorus application and early thinning; experiments on
phosphorus response; variety experiments (on-station experiments
on variety x plant density).

Variety trials; variety x planting systems trials; exploratory trials to
investigate establishment problems; fertilizer and
Rhizobium inoculation.

Exploratory factorial experiments on density, weed control, N and P;
types of chemical weed control; demonstrations on improved plant
density and spatial arrangements, chemical weed control, and zero

Small demonstration and test stores in highlands; research/
demonstration stores in coastal locations.

Source: Tripp (1991).










Table 2. Examples of modifications in the process of on-farm research

Identification and Farmer modifications
Case modification of priorities to technology

Rwanda Initial research interest in fertility and disease control;
climbing beans performed well in variety trials and
attracted farmer interest; special interest paid to staking
materials for climbing beans, leading to agroforestry

Ghana Best-bet maize varieties identified rapidly; emphasis
placed on developing practical recommendations for
planting and fertilizer application; weed control research
postponed; fertilizer recommendations adjusted
according to changing prices.

Nigeria Soil fertility initially found to be a more important concern
for farmers than fodder production; women as important
clients of research.

Philippines Early cropping systems experimentation identified wet
seeding as the most appropriate technology for most of
Iloilo, but subsequent on-farm tests showed dry seeding
to be superior in certain environments.

Initial interest in low adoption of improved maize varieties,
but no clear differences between old and new
recommended maize varieties; plant population and
nutrition observed to be problems; hypotheses that high
planting densities were caused by fodder needs or poor
seed quality rejected in favor of importance of shootfly.

Farmers' plant population management, and weeding
and thinning practices thought to be inadequate, but
subsequent analysis showed them to be efficient for
managing maize for grain and fodder; densities varied
for local and improved maize.

Initial interest in disease resistance; more efficient
plant type, yield, and establishment; the need to row
plant some trials led to discovery that row planting was
a feasible technology.

Interviews and observations showed main problem was
weed control; plant spacing and density had to be
improved concomitantly with weed control; improved
weed control could be linked to reduced tillage to address
erosion problems.

Initial interest in potato storage changed to focus on
importance of seed potato storage, particularly for new
varieties; if farmers on coast could store own seed
potatoes, costs of production lowered.

Farmers chose varieties not only on basis of grain yield
but also for grain cooking qualities and production and
palatability of leaves; farmers experimented with various
intercrops with climbing beans, particularly sweet
potatoes and bananas; farmers did not accept row
planting of climbing beans.

Sighting poles found to be more practical than string for
line planting; effectiveness of farmer practice of applying
basal fertilizer after germination confirmed in

Management of hedgerows (spacing, structure, pruning)
modified by farmers; farmers experimented with alley
farms in a wide variety of crops; farmers used hedgerow
trees for purposes beyond those originally envisioned.

Farmers used less fertilizer than recommended; only
farmers with light soils used pre-emergent herbicides.

Shallower planting of new variety; development of market
potential for new variety; adoption of fungicides as
commercial opportunities increased; change in planting

Light tractor harrowing substituted for manual chopping
of crop residues in zero tillage by farmers with labour
shortage; homemade shield devised by farmers to
help direct application of contact herbicide.

Farmers adopted principle of diffused light storage
and used local materials to construct a variety
of seed stores or to improve other storage techniques.

Source: Tripp (1991).






research strategy (Byerlee et al. 1991). The methods
and perspectives of OFR will play a leading role in
the organization of adaptive research and will also
contribute to priority setting for longer-term research.
This priority setting needs to include identification of
the relative strengths of the public sector and the
private sector in agricultural research. It also needs
to be more articulate regarding the role of agricultural
technology in meeting national goals for rural
development. Agricultural research cannot be
expected to bear the entire burden, and national
research programs should be very clear in identifying
their potential contributions.

Sustainability and bottom-up planning-The
growing interest in sustainable agriculture is a
reminder that agricultural research needs to address
both farmers' and society's needs. There is no doubt
that little progress will be made in developing
resource-conserving technology acceptable to
farmers without following the lead and the lessons of
OFR (Harrington 1992). !n this context, the image
held by some of OFR as simply a way of polling
farmers' opinions and summing them up needs to be
examined. OFR has placed much emphasis on
'bottom-up' planning, and this remains an important
objective. But both researchers and farmers are
participants. 'Top-down' technologies have been
successfully introduced through the use of OFR
techniques. Farmers may not immediately see the
rationale for something like alley cropping, for
instance, but the use of OFR to discuss, test, and
adjust this technology has led to its successful
introduction (Reynolds et al. 1991). OFR is not
distinguished by exclusive reliance on grass-roots
technology development, but rather by a respect for
grass-roots knowledge and aspirations. Ideas for
technological change may come from anyone; there
is an open forum. The touchstone of the OFR
perspective is whether or not the research proceeds
with a consideration for the ideas of the farmers and
an acknowledgement of the tradeoffs involved in
adapting technology to their conditions.

Institutional development-Finally, a particularly
important lesson we have learned along the way is
that OFR cannot substitute for strong national
institutions. OFR has both been part of the solution
and part of the problem. OFR has been part of the
solution in offering an effective way to organize
adaptive research. There are many instances where
national research programs are more effective at

targeting technology generation and more efficient at
utilizing field-level data because of OFR. The
management of OFR has proven to be less than
straightforward, however, and it is only recently that
a comprehensive set of guidelines for the
organization of OFR has become available (Merrill-
Sands et al. 1991).

But OFR has also been part of the problem in the
sense that-its presence in national research
programs is often due to special donor projects. The
proliferation of individual donor projects undermines
the capacity of national institutions to take decisions
or define overall strategies (Morss 1984). In Zambia,
for example, 20 different donors fund over 150
projects in the Ministry of Agriculture (Kean and
Wood 1992). It is recognized by many that a long-
term commitment to institution-building in Africa is
necessary (Eicher 1989). It is currently fashionable to
be quite critical of the performance of national
agricultural research programs, but current donor
strategies, although often presented in terms of
'capacity building', may weaken rather than support
national programs' ability to adapt the methods and
perspectives of OFR to their own goals. OFR has a
much better chance for survival in a stable
institutional environment with long-term funding and
career prospects, rather than as part of a short-term
special project.


In considering the future of OFR, it may be best to
return to the four expectations that contributed to its
growth. We pointed out that the role of technology,
support for government institutions, a faith in
people's own skills and knowledge, and a sense of
awe regarding the complex systems that farmers
manage, were four strong motivations for the
development of OFR. We also saw that these factors
play an important role in many other rural
development movements. To the extent possible, it
would be sensible for those involved in rural
development to focus on those motivations, which
appear to be widely shared, rather than on the
particular movements, which are often a source of
dissention and controversy.

But it is probably inevitable that development
programs and donor strategies will continue to be
organized around these types of movements ('new

directions' or 'development fads', depending on one's
point of view.) The ideal situation would be to take
advantage of the missionary zeal provided by such
movements, without getting involved in the
missionary self-righteousness that usually
accompanies them. Not only are many of the
motivations stable through time, but the basic
problems to be addressed are constant: limited funds
and limited institutions must devise ways of
improving the productivity and welfare of rural people
whose poverty is embedded in complex and
challenging agricultural systems.

OFR is now at a point where it can stop competing
as a movement and concentrate on the factors that
have motivated its development. Agricultural
technology is far from being a complete answer to
rural poverty, but we have enough examples of
success and of further potential that we believe the
investment is well justified. Such technology must be
developed in close collaboration with the farmers
who will use it, and OFR offers a strategy for
appropriate types of research. The current
movement toward privatization will undoubtedly
contribute to improving the agricultural economies of
many countries, and will hopefully make public sector
institutions leaner and more efficient. But the invisible
hand will not make its presence felt everywhere, and
there will be a continuing need for government to
address many of the basic problems of resource-
poor farming populations. OFR was conceived to
help meet those needs.

OFR can make a strong contribution to technology
generation and to strengthening local institutions. In
doing so it must pay increased attention to the
demands of administrators and donors for more
precise planning and more competent impact
measurement. These are reasonable requests, and
will contribute to improving the efficiency of OFR and
to ensuring that OFR leads to increasing farm
productivity. But attention must also be paid to two
other motivating factors for OFR. We cannot forget
that OFR is also built on a respect for farmers and
their knowledge and aspirations, as well as on a
sense of wonder at the complexity of their farming
systems. In our rush to measure impact we should
not overlook the role that this more human side of
OFR plays in the effective management and flow of
information in a research organization. It provides
much of the motivation and the spirit for the people
- researchers, farmers, and extension agents -

who participate in OFR. It has been part of the
balance of OFR, and indeed of any successful effort
at rural development. Administrators or donors who
ignore these expectations and overlook the spirit and
the beliefs of the people carrying out the work will
negate most of the benefits that improved impact
measurement systems can provide.


Useful comments on an earlier draft were provided
by R.J. Bingen, D. Buckles, D. Byerlee, P. Heisey,
M. L6pez, M. Morris, M. van Nieuwkoop, and S.
Waddington, none of whom are responsible for errors
remaining in the final product.


Anandajayasekeram, P. 1985. Teaching Notes on the
Diagnostic Phase of OFR/FSP: Concepts, Principles and
Procedures. Nairobi: CIMMYT Eastern African
Economics Program.

Anandajayasekeram, P., and M. Rukuni. n.d.
Institutionalization of On-Farm Research With Farming
Systems Perspectives (OFR/FSP) in Eastern and
Southern African Region: Achievements and Future
Direction. Unpublished manuscript.

Biggs, S.D., and J. Farrington. 1991. Assessing the effects of
farming systems research: Time for the reintroduction of
a political and institutional perspective. Journal of the
Asian Farming Systems Association 1: 113-131.

Brush, S.B., and B.L. Turner II. 1987. The nature of farming
systems and views of their change. In B.L. Turner and
S.B. Brush, eds., Comparative Farming Systems. New
York: Guilford Press.

Byerlee, D., K. Khan, and M. Saleem. 1991. Revealing the
rationality of farmers' strategies: on-farm maize research
in the Swat Valley, Northern Pakistan. In R. Tripp, ed.,
Planned Change in Farming Systems: Progress in On-
Farm Research. Chichester: John Wiley. Pp. 169-190.

Byerlece, D., P. Hobbs, B. Khan, A. Majid, M.R. Akhtar, and N.
Hashmi. 1986. Increasing Wheat Productivity in the
Context of Pakistan's Irrigated Cropping Systems: A
View from the Farmer's Field. PARC/CIMMYT Paper
86-7. Islamabad, Pakistan: Pakistan Agricultural
Research Council.

Caldwell, J.S. 1987. An Overview of Farming Systems
Research and Development: Origins, Applications, and
Issues. In J.P. Gittinger, J. Leslip, and C. Hoisington,
eds., Food Policy: Integrating Supply, Distribution, and
Consumption. Baltimore: Johns Hopkins. Pp. 165-178.

Chambers, R., A. Pacey, and L.A. Thrupp. 1989. Farmer First:
Farmer Innovation and Agricultural Research. London:
Intermediate Technology Publications.

Collinson, M. 1972. Farm Management in Peasant Agriculture.
New York: Praeger.

Cornick, T.R., and R. A. Kirkby. 1981. Interactions of Crops
and Livestock Production in the Generation of
Technology in Sloped Areas. Cornell International
Agriculture Mimeograph 90. Ithaca: Cornell University.

Eicher, C.K. 1989. Sustainable Institutions for African
Agricultural Development. ISNAR Working Paper No.
19. The Hague: International Service for National
Agricultural Research.

Ellis, F. 1992. Agricultural Policies in Developing Countries.
Cambridge: Cambridge University Press.

Fitch, J.B. 1983. Maize Production Practices and Problems in
Egypt: Results of Three Farmer Surveys. CIMMYT
Economics Working Paper. Mexico, D.F.: CIMMYT.

Gilbert, E.H., D.W. Norman, and F.E. Winch. 1980. Farming
Systems Research: A Critical Appraisal. MSU Rural
Development Paper No. 6. East Lansing: Michigan State

Harrington, L.W. 1992 Sustainability in Perspective: A
Discussion of International Center Contributions
towards Sustainable Agricultural Development.
CIMMYT Economics Working Paper. Mexico, D.F.:
CIMMYT (forthcoming).

Haugerud, M., and M. Collinson. 1990. Plants, genes, and
people: Improving the relevance of plant breeding in
Africa. ExperimentalAgriculture 26:341-362.

Hawkins, A.P., and F.II. Buttel. 1989. The political economy of
'sustainable development'. Paper prepared for the annual
meeting of the American Sociological Association,
August 1989, San Francisco.

Herdt, R. 1987. Whither farming systems? In Hlow Systems
Work, Proceedings of the Farming Systems Research
Symposium, 1987. Little Rock: University of Arkansas
and Winrock International Institute for Agricultural

Johnston, B.F., and W.C. Clark. 1982. Redesigning Rural
Development: A Strategic Perspective. Baltimore: Johns

Kean, S.A., and A.P. Wood. 1992. Agricultural policy reform
in Zambia: The dynamics of policy formulation in the
Second Republic. Food Policy 17(1): 65-74.

Lipton, M., with R. Longhurst. 1989. New Seeds and Poor
People. London: Unwin Hyman.

Low, A. 1986. Farm-Household Economics and the Food
Crisis: Agricultural Development in Southern Africa.
London: James Currey.

Merrill-Sands, D., and D. Kaimowitz, with K. Sayce and S.
Chater. 1989. The Technology Triangle: Linking
Farmers, Technology Transfer Agents, andAgricultural
Researchers. Summary Report of an International
Workshop held at ISNAR, The Hague, 20 25 November
1989. The Hague: International Service for National
Agricultural Research.

Merrill-Sands, D., P. Ewell, S. Biggs, and J. McAllister. 1989.
Issues in institutionalizing on-farm client-oriented
research: A review of experiences from nine national
agricultural research systems. Quarterly Journal of
InternationalAgriculture 28(3/4): 279-300.

Merrill-Sands, D.M., S.D. Biggs, R.J. Bingen, P.T. Ewell, J.L.
McAllister and S.V. Poats. 1991. Institutional
considerations in strengthening on-farm client-oriented
research in national agricultural research systems:
Lessons from a nine-country study. Experimental
Agriculture 27: 343-373.

Morss, E.R. 1984. Institutional destruction resulting from donor
and project proliferation in Sub-Saharan African
countries. World Development 12(4): 465-470.

Mutsaers, H.J.W., and P. Walker (eds). 1989. On-Farm
Research in Theory and Practice. Proceedings of a
workshop on design and analysis of on-farm trials, 27
February to 3 March 1989, IITA. Ibadan, Nigeria:
International Institute of Tropical Agriculture.

Neeley, D., J. Matata, and F. Kisyombe. 1991. Statistical
Techniques for the Analysis of On-Farm Crop Trials: A
Manual. Nairobi: CIMMYT, Eastern and Southern

Negassa, A., B. Tolessa, S. Franzel, G. Gedeno, and L. Dadi.
1991. The introduction of an early maturing maize (Zea
mays) variety to a mid-altitude farming system in
Ethiopia. ExperimentalAgriculture 27: 375-383.

Norman, D. 1977. Economic rationality of traditional Ilausa
dryland farmers in the North of Nigeria. In R.D. Stevens,
ed., Tradition and Dynamics in Small-Farm Agriculture.
Ames: Iowa State University Press.

Norman, D. 1980. The Farming Systems Approach: Relevancy
for Small Farmers. MSU Rural Development Paper No.
3. East Lansing: Michigan State University.

Reynolds, L., C. di Domenico, A.N. Atta-Krah, and J. Cobbina.
1991. Alley farming in South-Western Nigeria: The role
of Farming Systems Research in technology
development. In R. Tripp, ed., Planned Change in
Farming Systems: Progress in On-Farm Research.
Chichester: John Wiley. Pp. 85-108.

Ringia, 0. 1991. Diffusion of Kito/Cotton/Cowpea
Intercropping Technology Among Farmers in Kilosa,
Tanzania: Policy Implications. Paper presented at the
FSR/E Symposium, 5-10 October 1991, Michigan State
University, East Lansing.

Simmonds, N.W. 1991. Bandwagons I have known. Tropical
Agriculture Association Newsletter, December.-Pp 7-10.

Tripp, R. (ed.) 1991. Planned Change in Farming Systems:.
Progress in On-Farm Research. Chichester: John Wiley.

Upton, M4. 1987. African Farm Management. Cambridge:
Cambridge University Press.

Vernon, R., and J.M.N. Parker. 1983. Maize/weed competition
experiments: Implications for tropical small-farm weed
control research. Experimental Agriculture 19: 341-47.

Wood, R.E. 1986. From Marshall Plan to Debt Crisis: Foreign
Aid and Development Choices in the World Economy.
Berkeley: University of California Press.

Worman, F., D. Norman, and J. Ware-Snyder (eds). 1990.
Farming Systems Research Handbook for Botswana
ATIP RP 3. Agricultural Technology Improvement
Project (ATIP), Ministry of Agriculture, Gaborone,

Chimanga Cha Makolo, Hybrids and Composites:
Farmers'Adoption of Maize Ibechnology in Malawi, 1989-90*

Melinda Smale, with Z.H.W. Kaunda, H.L. Makina, M.M.M.K. Mkandawire, M.N.S. Msowoya,
D.J.E.K. Mwale, and P.W. Heisey


During the 1989-90 and 1990-91 cropping seasons,
researchers from CIMMYT and the Evaluation Units
of the Malawian Ministry of Agriculture (MOA)
collected data to profile farmers' adoption behavior
and varietal choices, and to elucidate how farm
household factors influence farmers' selection of
maize varieties in Malawi. Survey areas are shown in
Map 1. This paper is a brief summary of selected
findings for the 1989-90 season. More complete data
and interpretation can be found in CIMMYT
Economics Working Paper 91/04.

Selected Findings

Selected statistics on maize technology and varietal
adoption in Malawi are presented in the Table 1. The
most striking finding revealed by the data is the
complexity of farmers' adoption patterns. Farmers'
adoption decisions consist of several interrelated but
distinguishable choices. The first choice adoption
- is the decision to adopt or not to adopt the
recommended variety and related practices, and in
what combination. The second choice extent of
adoption is how much land to allocate to the new
and old techniques. The third choice intensity of
adoption is the level per hectare, or rate of
application, of inputs such as fertilizer.

The percentage of farmers adopting hybrid maize
seed (adoption) appears to vary sharply by
agroeconomic zone. Only 14% of sample farmers in
Blantyre, nearly 40% in Mzuzu, and about 33% in
Kasungu sowed hybrid seed in the 1989-90 season.
Few farmers in any zone grew composite maize. The
proportion of farmers who used fertilizer on their local
maize was similar across survey zones, although
Blantyre farmers were slightly less likely than
Kasungu and Mzuzu farmers to apply fertilizer to

local maize. Over all zones, roughly half of the
sample farmers fertilized local maize. Almost all
farmers used fertilizer on hybrid maize, but compared
to the Mzuzu and Kasungu farmers, Blantyre farmers
were less likely to adopt both seed and fertilizer in
the 1989-90 season.

Map 1. Location of study areas in Blantyre, Kasuwgu,
and Mzuzu Agricultural Development Divisions,
Malawi, 1989-90.

* This paper is a brief summary of Smale etl al.. 'Chimanga Cha Makolo'. Hybrids, and Composites: An Analysis of Farmers'Adoption of Maize Technology in
Malawi, 1989-91, CIMMYT Economics Working Paper 91/04 (Mexico. D.F.: CIMMYT. 1991).


Kasungu ADD / .:,


Blantyre Al

Study zone *'.
Agricultural Development Division
(ADD) boundary

The average proportion of farmers' maize area
allocated to hybrid maize (extent of adoption) varied
little by agroeconomic zone. Farmers universally
grew local maize for home consumption, and even
when they chose to adopt hybrid maize, they tended
to devote over 60% of their maize area to local

varieties. This feature of maize varietal choice,
although not unique to Malawi, appears to be more
pronounced in Malawi than in other nations that have
experienced similar seed-fettilizer transformations.
The similarity of this pattern across agroeconomic
zones also suggests that the factors affecting the

Table 1. Selected statistics on maize technology and varietal adoption, 1989-90

Economic stratum8
Characteristic Blantyre Kasungu Mzuzu strata

Cultivated area (ha) 0.8* 1.4 1.5 1.2
Maize as percent of farm area 98 84 85 89

Adoption of seed
Percent of farmers growing:
Local maize 97 99 97 98
Hybrid maize 14* 33 38 27
Recycled hybrid 4* 7 9 6
Composite variety 4 4 5 4

Percent of farm maize area sown to
hybrid maize by adopters 30* 35 42 34

Percent of aggregate maize area in:
Local maize 91 84 74 85
Hybrid maize 6 13 22 12
Recycled hybrid 1 2 2 2
Composite variety 1 1 2 1

Hybrid maize as percent of
aggregate maize output 18 44 47 35

Adoption of fertilizer
Percent of local maize growers apply 44 52 58 50
Percent of hybrid maize growers apply 71 97 97 87

Application, rate of adopters (kg N ha-')
Local maize 48* 37 37 41
Hybrid maize 64* 86* 111 82

Other agronomic practices
Percent of aggregate area intercropped
Local maize 32 2 16 15
Hybrid maize 31 3 1 13

Percent of aggregate area weeded twice
Local maize 77 54 56 63
Hybrid maize 90 68 66 76

Yield (t/ha)
Local maize, unfertilized 0.7 0.9 0,6 0.8
Local maize, fertilized 1.2 1.4 1.2 1.3
Hybrid maize, fertilized 2.2* 3.0 2.9 2.7

Source: Maize Variety and Technology Adoption Survey, CIMMYT/MOA, 1989-90.
a Strata correspond to higher potential maize-producing zones in Blantyre, Mzuzu, and Kasungu Agricultural Development Divisions.
Combined figures are weighted by probability of selection. Strata n=140. Total N=420.
* Indicates statistically significant differences between strata (5%), pairwise Chi-square or t-test.

land allocation decision are not necessarily the same
factors that affect farmers' initial choice to sow new

The third aspect of the technology adoption decision
(intensity of adoption) is illustrated by farmers' choice
of fertilizer application rates. The mean and modal
nitrogen application rates in all three zones were
near recommended levels for local maize, and in
Kasungu and Mzuzu the same was true for hybrid
maize. Relative to the Mzuzu and Kasungu farmers,
Blantyre farmers reveal a propensity to apply higher
rates of fertilizer to local maize and lower rates to
hybrid maize.

Because hybrid maize adopters continue to allocate
a large proportion of their maize area to local
varieties, aggregate maize area sown to hybrid
varieties, recycled hybrids, and composites remains
fairly low, or about 15% of the maize area in the
combined survey zones. Although hybrid maize may
represent a fairly small percentage of aggregate
maize area, aggregate hybrid maize output
constitutes a sizeable percentage of total maize
output (slightly under 20% in Blantyre and over 40%
in both Kasungu and Mzuzu). The fact that hybrid
maize represents a relatively large proportion of
aggregate output, and the evidence that many food-
deficit farm households must and do rely on their
own or purchased hybrid maize to bridge annual
consumption needs, underscore the growing role of
hybrid maize in both national and household food

Selected agronomic practices associated with local
and hybrid maize varieties are also presented in the
summary table. The percentage of maize area that is
intercropped is highest, for either local or hybrid
maize, among the Blantyre farmers. Mzuzu and
Kasungu farmers, who tend to have larger land
areas, are more likely to bring fallow land into maize
cultivation, especially for hybrid varieties.
Corresponding to general rainfall patterns, the
frequency of plots planted after mid-December is
higher the more northern the location, but planting is
also shifted to later dates for hybrid maize in all
zones. Blantyre farmers, with their limited land areas,
are more likely to weed either their local or hybrid
maize plots twice.

The divergent patterns found among the survey
zones probably reflect different objectives and
constraints faced by farmers, some of which are
hypothesized in greater detail in CIMMYT Economics
Working Paper 91/04 and tested elsewhere, although
a brief summary will be presented below.1 For
example, most of the Blantyre hybrid maize growers
learned about hybrid maize on their own or from
neighboring farmers, rather than from extension
agents, and purchased their inputs with cash earned
from off-farm employment rather than through the
formal credit system. The combination of practices
these self-styled hybrid maize growers select is
distinctly different from the practices found among
the full-time, larger maize producers of the Mzuzu
and Kasungu zones, who are more likely to use
credit and obtain extension advice.


The data suggest that no single variable determines
a farmer's adoption choice in Malawi. To increase
aggregate area sown to hybrid maize, the
Government of Malawi can choose from several sets
of options, including actions that affect the choice of
whether to grow hybrid maize or not (adoption) and
actions that can influence farmers' choice of land
allocation to local and hybrid maize (extent of
adoption). Pursuing a goal of increasing aggregate
hybrid maize production may involve a third set of
options designed to improve the efficiency of hybrid
maize production among adopters by shifting their
yield or net returns distributions toward higher

Each set of options is associated with distinct
national welfare and distributional consequences. A
combination of options may be more likely to produce
measurable and desirable long-term results. For
example, actions that affect adoption opportunities
may be primarily institutional. Factors such as credit,
timely provision of appropriate seed for a particular
locality at the proper planting time, and provision of
fertilizer and seed to markets to enable farmers to
obtain inputs with cash rather than on credit, can
result in a larger number of farmers growing at least
some hybrid maize. The social welfare
considerations associated with these actions include

1 See also M. Smale, Risk, Disaster Avoidance, and Farmer Experimentation: The Microeconomics of HYV Adoption in Malawi, unpublished Ph.D. dissertation
(College Park, Maryland: University of Maryland. 1992).

household food security, since higher yield potential
on some plots can improve the conditions of even the
smallest farmers.

Aggregate hybrid maize output and area can also be
increased through actions that affect the area hybrid
adopters allocate to hybrid maize (that is, the extent
of adoption). Such actions include current efforts to
breed and diffuse flintier hybrids and to educate
farmers about storage and processing alternatives
for dentier hybrids. Farmers appear to plan their
production according to the belief that local maize
continues to be superior in processing and storage,
which suggests that a flint hybrid is likely to have
broader appeal than a dent hybrid. Those who are
now adopting may allocate larger portions of their
maize area to hybrid maize if that hybrid is more
substitutable in consumption. Those who are able to
adopt but have not adopted may be more willing to
grow a flint hybrid maize than a dent variety.

Because these more technical options primarily
affect farmers who are already capable of adoption,
they do not have the same distributional welfare
implications as the institutional options cited above,
although they can improve national welfare. For
example, flintier hybrids cannot relieve underlying
expenditure constraints or inability to qualify for
credit. Even those farmers who can afford to
purchase inputs cannot be expected to relinquish

their local sources of seed until they can rely on
marketing institutions for timely, certain delivery of
quality seed meeting their own specifications. Even
then, comparison of net returns distributions
indicates that varietal diversification may remain an
objective that is consistent with reducing total
economic risk.

A third set of policy actions involves improving
technical and economic efficiency among hybrid
producers through developing recommendations that
are more closely tailored to their various operating
conditions and concentrating on agronomic practices.
As long as overall adoption rates are modest,
investments of this type can improve the welfare of
only a limited proportion of farmers, although they
could increase aggregate output.

Finally, efforts to breed more varieties with other
desirable traits, such as early maturity, and to
provide farmers with seed-fertilizer packages of
varying composition and a wider range of sizes, are
likely to improve adoption rates. The data show that
farmers are willing to try diverse combinations of
technological options and may find them to be
consistent with their objectives. The data at this
stage are sufficient to confirm that the overall
adoption pattern in Malawi does not express farmers'
rejection of new technology, but constrained

Response of Four Maize Varieties to Nitrogen at Lisungwi
Extension Planning Area, Mwanza, Malawi

W.T. Kawonga and P.W. Heisey


Lisungwi Extension Planning Area (EPA) in Mwanza
Rural Development Project (RDP) has insufficient
rainfall for optimum maize growth (Figure 1). The
soils are generally low in organic matter, nitrogen,
phosphorus, and magnesium. Farmers in this area
grow cotton for sale and use the money realized from
cotton sales to buy maize. In the past, Blantyre
Agricultural Development Division (ADD)
Management Unit offered no credit package for
maize in this area, as it was thought that farmers
would not produce enough additional maize to pay
back the loan. On the initiative of the Blantyre ADD
adaptive research team, a survey was conducted to
monitor the farming systems of the area and assess
farmers' opinions of producing maize as a staple

The survey confirmed the ADD Management Unit's
findings that farmers grow cotton for sale. However,
it was also found that many farmers from areas such
as Thyolo, Blantyre Central, Blantyre North, and
Chiradzulu have migrated to Lisungwi. These
farmers have been growing maize for a long time and
were interested in a maize credit package.


This study was undertaken to evaluate the
performance of four maize varieties and determine
the response of these varieties to increasing levels of

Materials and Methods

There were four experimental sites in Lisungwi EPA,
located at Kandoje and Waiyatsa (1986-87 season),
Kamwamba (1987 88 season), and Symon (1988-89
season) At these sites the soils were slightly alkaline
to neutral in reaction. Texture ranged from sands to
loamy sands. Organic matter, nitrogen, phosphorus,
and magnesium were low. Potassium was medium

whereas calcium was moderately high (Table Al,
Appendix). The design of the experiment was a
randomized block laid as a 4 x 4 factorial replicated
two times at Kandoje, Waiyatsa, and Kamwamba
and three times at Symon. Plots consisted of 4
ridges 4.5 m long as gross and 2 middle ridges 3.5 m
long as net. Treatments consisted of four maize
varieties fertilized at four rates of nitrogen. The maize
varieties were local, NSCM 41, Tuxpeno, and MH 15,
which were fertilized at 0, 30, 60, and 90 kg of
nitrogen ha1.

Statistical (ANOVA) and agronomic evaluations of
the trial were done. In addition, the relationship
between treatments and environmental index was
also investigated. Finally, economic analysis of
nitrogen response was done using both discrete yield
data and continuous response functions obtained by
regression analysis. Economic assumptions in both
the discrete and continuous cases were the same.

Rainfall (mm)






3: /|

Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep

Figure 1. Monthly rainfall, Lisungwi, Malawi, 1986-87.


Tables 1 and 2 show the results of the first statistical
analysis. It should be noted that in analysis of
variance of the data pooled over all sites, though the
variety x nitrogen interaction was not significant, both
variety x site and nitrogen x site interactions
appeared to be significant.

Maize yields as affected by maize varieties-
Table 1 shows that at two locations, Waiyatsa and
Symon, there were no significant differences in
maize yields among the maize varieties. Though
statistically insignificant according to Student-
Newman-Keul's test based on means, MH 15 was
the highest yielder at these locations, and NSCM 41
also yielded relatively well at Waiyatsa. At Kandoje,

(text continued below)

Table 1. Usable maize yield (kg ha-1) at 12.5% moisture content

Year and site

1986-87 1987-88 1988-89
Kandoje Waiyatsa Kamwamba Symon Pooled

Maize variety
Local 3,866 b 3,220 2,243 1,873 2,697
NSCM 41 6,885 a 3,987 3,953 a 2,388 4,090
Tuxpeno 4,913 b 3,212 2,295 b 2,491 3,146
MH 15 5,350 b 4,296 3,095 ab 2,770 3,744

Nitrogen (kg ha-1)
0 2,317 b 1,790 c 1,545 b 1,993 b 1,921 b
30 5,710 a 3,598 b 2,829 a 2,078 b 3,390 a
60 6,296 a 4,403 ab 3,463 a 2,623 ab 4,021 a
90 6,690 a 4,874 a 3,747 a 2,828 a 4,345 a

Variety ** NS NS *
Nitrogen ** *** *** ***

Variety 239 160 335 239 243
Nitrogen 239 160 335 239 243
VxN 478 320 670 478 486

CV (%) 26 25 33 35 30

Note: Means followed by the same letter are not significantly different from each other according to the Student-Newman-Keul's test

Table 2. Interaction effects in the variety x fertilizer
level trial

Maize variety and yield (kg ha-')

Local NSCM 41 Tuxpeno MH 15 Means

N (kg ha"')
0 1,457 2,435 1,912 1,878 1,921
30 2,631 4,056 3,097 3,775 3,390
60 2,937 4,710 3,784 4,655 4,021
90 3,764 5,160 3,790 4,617 4,345

Means 2,697 4,090 3,146 3,744

NSCM 41 yielded significantly better than the other
three maize varieties. No significant yield differences
existed among local maize, Tuxpeno, and MH 15 at
this location, although both MH 15 and Tuxpenio
yielded 1 t ha-1 more than local maize. At
Kamwamba, NSCM 41 was still the highest yielding
variety. However, no statistical differences existed
between NSCM 41 and MH 15. The local variety and
Tuxpefo were the lowest yielding maize varieties at
this site. No significant yield differences were
observed among local maize, Tuxpero, and MH 15.

Pooled analysis across locations and years indicated
no significant differences in maize yields between
NSCM 41 and MH 15, MH 15 and Tuxpenio, and
Tuxpefio and local maize. Nonetheless NSCM 41
and MH 15 outyielded Tuxpeio by 600-900 kg ha-1
and Tuxpeio outyielded the local variety by
450 kg ha-1.

Maize yields as affected by nitrogen rates-
Maize yields from fertilized plots at Kandoje and
Kamwamba were significantly higher than those from
unfertilized plots. No significant yield differences on
the basis of simple means comparison were
observed by increasing nitrogen rates from 30 to
90 kg ha1 at these sites, although at both sites yields
did appear to increase somewhat with increasing
application of nitrogen. The same pattern of
increasing yields with increasing nitrogen application
was also observed at Waiyatsa and Symon.
Nonetheless, at Waiyatsa, means comparison
indicated that fertilization gave significantly higher
yields than no fertilization. No significant differences
appeared to exist between 30 and 60 kg ha' at
Waiyatsa, and again between 60 and 90 kg ha1. At
Symon, means separation failed to determine any
differences between application rates up to
60 kg ha1-, and again between 60 and 90 kg ha'.
Oddly enough, however, at Symon maize yields
appeared to increase more between 30 and
60 kg ha' than between any other two adjacent
nitrogen treatments. Increased yields with successive
increases in nitrogen application rates at Symon
thus did not appear to be either linear or concave,
the two most commonly expected forms of response.

Pooled analysis across locations indicated that maize
yields from unfertilized plots were significantly lower
than from fertilized plots. Though yields increased
with increasing application levels from 30 to
90 kg ha'-, no significant yield differences were
observed among the three rates.

Response of maize varieties to environment at
Llsungwi-A linear regression of the mean yield of
each maize variety at each location on environmental
index (mean yield of the four maize varieties
evaluated per location/year) was run as a measure of
varietal stability. Theoretically, varieties with above
average phenotypic stability (b < 1) are relatively
insensitive to environmental changes and do not
show large changes in yield. Such varieties may be
more productive under poor management. Varieties
with below average phenotypic stability (b > 1) are

relatively more productive under high management
and are very sensitive to environmental changes
(Finlay and Wilkinson 1963).

Figure 2 shows the response of maize varieties to
environment at Lisungwi. The regression coefficients
(R2) indicated very good fits for the regression lines.
Below an e-value of 1.5 t ha'-, local maize performs
better than Tuxpenio and vice versa as management
becomes better. The response of NSCM 41 was
steep, indicating response to higher levels of
management by the Finlay-Wilkinson criterion.
Regression lines for Tuxpeio and MH 15 were
almost parallel, indicating that the response of these
varieties to environment was similar.

Despite the fact that by the Finlay-Wilkinson criterion
local maize had above average stability (b < 1), MH
15 had average stability (b approximately equal to 1),
and NSCM 41 was below average (b > 1), it should
be noted that even at the lowest value of e (e = 0),
both hybrids performed better than local maize.

Response of nitrogen rates to environment at
Lisungwi-Figure 3 shows the effect of environment
on different nitrogen application rates at Lisungwi.
Maize yields for each of the nitrogen rates were
related to environment by a simple linear regression

Maize yield (t/ha)
10 --

0 2 i4 6
Environmental index (e), (t/ha)

* Local Y = 316 + 0.7e
- Tuxpeno Y = -37 + 0.9e

-W NSCM 41 Y = 842 + 1.5e
*A MH 15 Y = 600 + 0.9e

Figure 2. Response of maize to the environment at
Lisungwi, Malawi.

Maize yield (t/ha)
10 T

0 2 4
Environmental index (e), (t/ha)

-0- N =0 Y =1,261 + 0.1lBe
-0 N =60 Y =124 +1.23e

-* N = 30 Y = -725 + 1.22
- N = 90 Y = 400 + 1.37

Note: N rates are in kg/ha.

Figure 3. Response of nitrogen to the environment at
Lisungwi, Malawi.

Y = a + bx (Hildebrand and Poey 1985). The analyst
indicated that applying fertilizer resulted in below
average stability, and by not applying fertilizer, the
result was above average stability. Maize yields
increased with increasing nitrogen rates and
management except for control plots, where maize
yields were fairly stable around 1.6 t ha1 across the
environment. Coefficients of determination showed
good fits for the regression lines for fertilized plots.
Unfertilized plots showed a relatively poor fit for the
regression line (R2 = 0.43).

Economic Analysis

Assumptions-The following assumptions were
used in the economic analysis. The field price of
maize was assumed to be MK .278 kg-1, reflecting
the selling price to ADMARC minus adjustments for
harvesting and transport. The field price of nitrogen
was assumed to be MK 2.02 per kg N, based on the
price of urea plus an adjustment for transport. In the
discrete (partial budget) analysis based on actual
yield data, the labour cost of a single fertilizer
application was assumed to be MK 4.30 ha1. In the
continuous analysis, the field price of nitrogen was

adjusted upward to MK 2.12 per kg N, to reflect the
labour costs of application as well as the urea price
and transport costs. When seed costs were also
considered in the partial budget analysis, seed of
NSCM 41 and MH 15 was valued at MK 2.50 kg-',
A seed of Tuxpeio at MK 2.00 kg-1, and local seed at
the grain price.

In both partial budget and continuous analysis, yields
were adjusted downwards by 25%. The minimum
acceptable marginal rate of return was set at 100%,
partially because fertilizer use appears to be fairly
risky in Lisungwi.

Procedures-Initially, partial budget analysis
(CIMMYT 1988) was applied to each variety at each
location, and to each variety in the pooled data, to
find the desired level of nitrogen application. In
addition, the partial budget analysis also looked at all
2e varieties at each location, and in the pooled data,
e including seed costs, to find the overall
recommended treatment (e.g., variety plus nitrogen

Initially regression analysis was based on the finding
of the pooled analysis of variance that there was no
variety x nitrogen interaction. Response curves were
is fitted at each site along with dummy variables that
maintained the hypothesis that variety only shifted
the intercept, but not the shape, of the response
curve. In the pooled data, response curves were
fitted along with dummy variables maintaining the
hypotheses that both variety and location shifted
intercepts, but again not the shapes of the response
curve. Both at each location, and in the pooled data,
the optimum nitrogen application rate was found by
solving for N in the equation:

PN(1 + R)

where f'(N) was the first derivative of the yield
response function f(N); PN was the field price of
nitrogen; pM was the field price of maize; A was the
yield adjustment factor in decimal form (100% 25%
= 75% = 0.75); and R was the minimum acceptable
marginal rate of return in decimal form (100% = 1)
(Jauregui and Sain 1992).

The initial partial budget analysis, the initial
continuous analysis, and reconsideration of the
actual yield data all suggested two things. First, the

response at Symon seemed quite atypical of the
response at the other three locations for the following
reasons. At most locations the hybrids NSCM 41 and
MH 15 appeared, as expected, to be more
responsive to nitrogen than Tuxpefio or local maize.
However, at Symon the reverse appeared to be true
with the hybrids exhibiting very little response while
local maize and Tuxpefio did appear to show
response to increasing rates of nitrogen.
Furthermore, the response of local maize at Symon
appeared very unusual, with the largest increase in
yield occurring between 60 and 90 kg ha-'. Finally,
fitting of response curves at Symon was fairly
difficult, with very low measures of goodness of fit.
Therefore, in the economic analysis presented
below, results are presented for the individual sites
and for pooled data excluding-Symon (i.e., for
Kandoje, Waiyatsa, and Kamwamba only).

Second, though statistical analysis indicated no
variety x nitrogen interaction, perusal of the initial
economic analysis as well as the raw data suggested
that the large differences between sites may have
obscured differences in the ways in which the
different varieties responded to increasing nitrogen.
Therefore, continuous response functions were re-
estimated under the assumption that response
slopes for local maize, Tuxpeio (a composite), and
the hybrids were all different. The response slopes
for NSCM 41 and MH 15 were assumed to be the
same, but intercepts (i.e., yield at N = 0) were
assumed to vary between the two hybrids.

In all cases of continuous response estimation, two
functional forms were used. The first was the
quadratic response curve,

Y = bo + b,N + bN2.

This is probably the most commonly fitted functional
form in response analysis and is easy to estimate
with standard statistical packages. However, it may
not always be consistent with the response predicted
by agronomic theory, and economic optima
suggested by the quadratic function may be
somewhat overstated.

Alternatively, the Mitscherlich-Baule response

Y = bo{1 exp[b,(b2 + N)]},

was also fitted to the response data. A functional
form such as the Mitscherlich-Baule may be

appropriate in situations in which yield response to
additional increments of nitrogen application may
approach a plateau, as is suggested by agronomic
theory (Tronstad and Taylor 1989; Frank, Beattie,
and Embleton 1990). A major disadvantage of the
Mitscherlich-Baule functional form is that it must be
estimated by non-linear regression and therefore
requires a specialized statistical package. This may
make it impractical in many instances.

Results of the Economic Analysis

The results of the economic analysis are shown in
Table 3.

Most results will be discussed by variety. Before
beginning this discussion, it should be noted that
even when the anomalous site (Symon) is excluded
from the analysis, there appear to be substantial
differences in response by location. (When Symon is
included, the significance level for the location x
nitrogen interaction is .17, which is non-significant at
standard levels but still suggests there may be
differences in response by site.) Responses may
differ because of differences in other components of
soil fertility or because of differences in the type of
season, since the trial was run over three years. This
in turn means it may be somewhat difficult to obtain
universal recommendations for Lisungwi based on
these data.

Local maize-Economic optima based on standard
partial budgeting and on the Mitscherlich-Baule
response function are fairly similar at Kandoje and
Kamwamba. At Kandoje no economic response is
observed beyond about 30 kg N ha-', and at
Kamwamba beyond 50 to 60 kg ha-1. The quadratic
function probably overestimates the economic
optimum at Kandoje. At Waiyatsa both response
functions appear to substantially overestimate the
economic optimum. This is in part because of the
strange pattern of response, with no difference in
yield at 30 and 60 kg N ha-' but an increase of 1 t
between 60 and 90 kg ha-'. Waiyatsa also affects the
pooled analysis based on continuous response
functions. As a fairly conservative recommendation,
the optimal application rate for local maize in
Lisungwi may be somewhere around 30 kg N ha'

Tuxpefo-Responses for the composite Tuxpeio
vary widely from site to site. The partial budget
analysis selects any application rate from 0 to

Table 3. economicc optimum nitrogen application rates
(kg ha-') fbr three sitesA separately and pooled, Malawi


Method of analysis:
Variety Partial budget Quadratic Baule

Local 30 56 3t5
Tuxpero 90 107 134
MH15 60
NSCM 41 90
Hybrid .. 56 36
Overall optimum treatment, partial budget:
NSCM 41 at 90 kg ha '


Method of analysis:
Variety Partial budget Quadratic Baule

Local 30 77 65
Tuxpefo 30 41 NC
MH 15 60
NSCM 41 90
Hybrid .. 83 100
Overall optimum treatment, partial budget:
MH 15 at 60 kg ha'


Method of analysis:
Variety Partial budget Quadratic Baule

Local 60 53 49
Tuxpeao 0 LR NC
MH15 90
NSCM 41 60
Hybrid .. 53 42
Overall optimum treatment, partial budget:
NSCM 41 at 60 kg ha1"

Pooled (excluding Symon)

Method of analysis:
Variety Partial budget Quadratic Baule

Local 30 56 46
Tuxpeno 30 57 NC
MH 15 60
NSCM 41 60
Hybrid .. 60 53
Overall optimum treatment, partial budget:
NSCM 41 at 60 kg ha:'

Note: NC = non-convergent (estimation of non-linear Mitscherlich-
Baule equation failed to converge). LR = linear response
(estimation of quadratic function suggested response was in
fact linear rather than quadratic).

90 kg ha' as optimal, depending on the location.
Neither functional form used in the continuous
analysis appears to fit very well from location to
location or in the pooled data. Furthermore, the
application rates selected as optimal by continuous
analysis at Kandoje a4e greater than 90 kg ha-'; in
other words they lie outside the range of the
experiment and therefore are not reliable. It does not
appear possible to draw even tentative conclusions
about optimum application rates for Tuxpenro.

Hybrid maize-Optimum nitrogen application rates
for hybrid maize as determined by either partial
budget or continuous response analysis appear to be
relatively consistent at Kandoje, Waiyatsa, and
Kamwamba. Nonetheless the purported optimum
based on the Mitscherlich-Baule function appears to
understate the optimum at Kandoje and Kamwamba.
This may be because yields of MH 15 at Kandoje
decline at 90 kg N ha-1, and the same result is true
for NSCM 41 at Kamwamba. In other words, at these
two locations the quadratic function appears to give a
better fit for one hybrid and the Mitscherlich-Baule
function to give a better fit for the other. In the pooled
data, both response functions and partial budget
analysis give fairly similar results. Tentatively,
optimal nitrogen application rates for hybrid maize in
Lisungwi may be around 50 to 60 kg ha-'.


Maize varieties-Based on the agronomic and
economic analysis, farmers can grow either
NSCM 41 or MH 15. High yields can be achieved
under high management. Though yields of NSCM 41
appear to be more variable over different
environmental conditions, both hybrids still appear to
yield better than local maize in poor environments.
The use of MH 15 might be constrained by seed
availability, and recent semi-flint releases might be
preferable to both of these hybrids by virtue of grain
texture, assuming the response of the new releases
to nitrogen at Lisungwi is similar to that of the two
hybrids tested here.

Fertilizer use-It is not possible to make firm
nitrogen application recommendations for Lisungwi.
Though statistically over the entire data set response
to nitrogen rates over 30 kg ha-' appears to be
minimal, further economic analysis suggests that the
optimum rate for local maize may indeed be around

30 kg ha-', but the optimum rate for hybrid maize
may be somewhat higher, perhaps in the range of 50
to 60 kg N ha-1.

Further work-Firmer fertilizer recommendations for
Lisungwi may await either further analysis, further
experimentation, or both. Further experimentation
may require the use of a single hybrid and local
maize as a control, since in the present experiment
use of local maize, an improved composite, Tuxpefio,
and two hybrids, all with somewhat different
expected responses, considerably complicated the
analysis. Furthermore, the assumption of equal
variances used both in the ANOVA and regression
analysis may not be warranted. Differences in sites
might be examined to determine to what extent they
result from differences in soils and to what extent
from differences in seasonal rainfall patterns. For
example, is the apparently higher overall fertility at
Kandoje related to the higher levels of soil P and/or
Mg at that location (Table Al, Appendix)? Risk
analysis might also be applied to the results of future
fertilizer response experiments, since it appears that
the use of any nitrogen, which has a real costto.the
farmer, also increases the variability of results.


CIMMYT. 1988. From Agronomic Data to Farmer
Recommendations: An Economics Training Manual.
Completely revised edition. Mexico, D.F.: CIMMYT.

Finlay, K.W., and G.N. Wilkinson. 1963. The analysis of
adaptation in a plant breeding programme. Australian
Journal ofAgricultural Research 14: 742-754.

Frank, M.D., B.R. Beattie, and M.E. Embleton. 1990. A
comparison of alternative crop response models.
American Journal ofAgricultural Economics 72: 597-603.

HIildebrand, P.E., and F. Poey. 1985. On-Farm Agronomic Trials
in Farming Systems Research and Extension. Boulder,
Colorado: Lynne Rienner Publishers, Inc.

Jauregui, M.A., and G.E. Sain. 1992. Continuous Economic
Analysis of Crop Response to Fertilizer in On-Farm
Research. CIMMYT Economics Paper No. 3. Mexico,

Tronstad, R., and C.R. Taylor. 1989. An economic and
statistical evaluation of functional forms and estimation
procedures for crop yield response to primary plant
nutrients. Unpublished MS.

Appendix A
Soil and Rainfall Data

Table Al. Soil analysis data

Location Depth pH C Texture %C %N C/N P Na K Mg Ca

Kandoje Top 7.0 10 LS 0.78 0.07 11 12 7.4 2.9 25.8 156.3
Bottom 6.2 16 SL 0.51 0.04 13 12 6.3 1.6 17.9 123.8
Waiyatsa Top 6.6 6 Sand 0.66 0.06 11 3 6.9 3.3 11.6 106.8
Bottom 6.4 11 LS 0.40 0.04 10 3 6.3 1.6 13.3 115.5
Kamwamba Top 6.3 6 Sand 0.43 0.04 11 4 6.0 2.6 7.9 74.8
Bottom 6.0 8 Sand 0.31 0.03 10 2 6.0 1.2 12.9 82.3
Symon Top 6.2 10 LS 0.81 0.07 12 2 6.3 3.6 18.3 123.3
Bottom 6.3 12 LS 0.81 0.07 12 2 6.6 2.9 19.1 123.3

Table A2. Rainfall analysis data

Number of years analysed
Mean annual rainfall total
Mean start of rainy season
Moan end of rainy season
Mean net season length
Quality of season (%)
Mean season index
Rainfall reliability (%)
Number of pentades with > 67% chance of rain occurring

746 mm
17 November
22 March
132 days

On-Farm Evaluation of Short-Season Sorghum and Fertilizer for
Smallholder Farmers in a Semi-Arid Region of Zimbabwe

Cornelius Chiduza, Stephen R. Waddington, and Mandivamba Rukuni, Senior Research
Fellow, Department of Soil Science and Agricultural Engineering, Agronomist, CIMMYT, and
Senior Lecturer, Department of Agricultural Economics and Extension respectively,
University of Zimbabwe, P 0 Box MP 167, Mount Pleasant, Harare, Zimbabwe


Three on-farm experiments were conducted during
1984-85, 1985-86, and 1986-87 in Siabuwa,
Zimbabwe to test 1) earlier maturing food sorghums
with pearly white grain and 2) fertilizer application as
possible solutions to production problems faced by
smallholder farmers growing sorghum.

Replacement of current long-season sorghums by
improved short-season white grain open pollinated
varieties, such as SV-2 and 321CR, consistently
increased grain yield (by an average of 62%) in the
three seasons at farmers' levels of inputs and
management. Farmer evaluation of the taste of
porridge from sorghum grain showed a high
preference for the pearly white short-season varieties
compared to the long-season local, Balala.

Fertilizer (50 kg N ha'-, 12 kg P ha'-) significantly
raised the grain yield of the long-season local
sorghum and shorter season varieties on vertisols,
siallitic clays, and sandy loam soils in average and
above average rainfall years (1984-85 by 37% and
1985-86 by 60%) but not in 1986-87, a very dry year.
Use of fertilizer was not economic in any of the

Pearly white grain, short-season food sorghums are
useful additions to the portfolio of sorghum varieties
available to farmers in Siabuwa. These varieties
should help increase the productivity of the important
sorghum enterprise without requiring any major
modification of current production practices and
without the use of fertilizer in the short term.


Based on diagnostic studies conducted during 1984
with smallholder farmers in Siabuwa Communal
Area, northwestern Zimbabwe, an experimental
agenda was developed to address several production

problems identified in sorghum (Sorghum bicolor L.,
Moench), the dominant staple food crop. These
problems were described in Chiduza et al. (1992).

In this paper we describe the high priority on-farm
experiments implemented starting in 1984-85 to
examine early maturing varieties and fertilizer as
solutions to sorghum production problems. The
experiments were accompanied by porridge taste

Materials and Methods

Variety x fertilizer trial--The sorghum variety x
fertilizer trial was planted to assess grain yields
possible with the new short-season sorghums and to
compare their potential with yields obtained from
current long-season local sorghum. This trial was
planted in Siabuwa at six sites in the 1984-85 and
1985-86 seasons. A randomized complete block
design (RCBD) with a factorial combination of five
varieties and two fertilizer levels was used, replicated
three times at each site. The five varieties were: the
farmer's local long-season variety (Balala); two
intermediate maturity improved varieties, i.e., Serena
(red grain) and M36172 (white grain); and two short-
season improved varieties, i.e., SV-2 (white grain)
and 321 CR (white grain). Fertilizer levels were: no
fertilizer; and 50 kg N ha-1, 12 kg P ha1, and 8 kg
K ha1. All of the P and K, and 15.5 kg N ha-1, were
applied as an initial dressing of compound fertilizer at
planting. The remaining 34.5 kg N ha'- was applied
as an ammonium nitrate top dressing six weeks after
planting. The plot size was 5 m x 4.5 m with 1 m
between plots. Row width was 75 cm, with one plant
per 20 cm of row.

Soil fertility levels in Siabuwa are known to vary
greatly with soil type (Hungwe 1985), and so sites
were selected to cover the three main soil types in
the area, the vertisols, the siallitic clays, and the
sandy loams. The trial was established at two sites of

each soil type during both seasons. The same six
sites were used in the two seasons, but a different
location within the same field was selected during the
second season. At two sites land was prepared using
a tractor-drawn mouldboard plough; the other four
sites were not ploughed.

The trial was dry planted (reflecting farmers' practice)
at all sites on 16 November during 1984-85 and on
28 November during 1985-86. The researcher was
responsible for planting, thinning, applying the
fertilizer treatments, and harvesting. Other non-
experimental inputs and practices were at the farmer
level and managed by the farmer. In particular, host
farmers were responsible for weeding (generally two
hoe weedings) and scaring birds.

At harvest two crop rows were discarded on each
side of an individual plot and a 50-cm length of plants
was removed from both ends of the plot. The number
of plants, number of panicles, and fresh weight of
panicles were recorded for the remaining 4 m x 1.5 m
nett plot.

A sub-sample of 10% of the nett plot fresh weight
was collected from each plot and dried to constant
weight at 80C for 72 h and then threshed. Grain
yield per hectare was calculated from these sub-
samples, and reported at 12.5% moisture.

An analysis of variance was first done for each site in
each year, then across sites and across years, and
was followed by means separation tests. A partial
budget analysis (CIMMYT 1988) was done for the
local sorghum, SV-2, and 321CR, with and without
fertilizer. The latter two varieties are short-season
sorghums which taste tests have shown to produce
acceptable-tasting porridge. The sorghum grain
yields from the trial plots were reduced by 20% in the
partial budget to reflect yields that farmers would
expect to get in their fields (CIMMYT 1988). The
1986 Grain Marketing Board price of Z$ 180/t was
used to value sorghum grain and farm gate costs of
fertilizer and labour for applying the fertilizer and
harvesting the extra yield were used. These were
Z$ 356.00/t for compound D fertilizer and Z$ 406.00/t
for ammonium nitrate fertilizer. Labour costs were
Z$ 4.50/ha for applying both compound D and
ammonium nitrate fertilizer, and Z$ 3.60/ha for
harvesting the extra grain yield.

Fertilizer levels trial-The fertilizer levels trial was
established in 1985-86 and 1986-87 to examine the
response of the short-season sorghum, SV-2, to a

range of low levels of inorganic fertilizer. The
experimental design was a split plot. Levels of initial
dressing of compound fertilizer comprised the main
plots, and levels of top dressing of ammonium nitrate
the sub-plots. There were three replicates. A total of
16 experimental treatments were included as a
factorial of the following: 1) initial fertilizer at 0:0:0,
10:7.3:4.6, 20:14.6:9.2, and 30:21.9:13.8 kg
NPK ha1; and 2) topdress fertilizer at 0, 28, 56, and
84 kg N ha-1. Gross plot sizes were 20 m x 3 m for
the mai0 plot and 5 m x 3 m for the subplot.

Three Sites, one on each of the three main soil types,
were selected for 1985-86 and for 1986-87, but
different farms were used each year. The trial was
planted on 30 November during 1985-86. In 1986-87
two sites were planted on 28 November and the third
site on 10 December. Management of this trial,
including planting, thinning, weeding, and harvesting,
was by the researcher. Harvest methods and data
gathered were similar to those described above.

Data for the 1985-86 season were available from
only one site (crops at two sites were destroyed by
drought); data for the 1986-87 season were available
from all three sites. A grouping of sites by soil type
resulted in homogeneous variances for data from
sites located on sandy loam soils and for sites on
clay or sandy clay loam soils. An analysis of variance
was then carried out on the two sets of data.

As previously described, partial budgets were
calculated from the combined data from the sandy
loam sites. No economic analysis was done for sites
on the clays and sandy clay loams because fertilizer
effects on grain yields were not significant with those

Variety trial-This trial examined whether shorter
season food sorghums have yield advantages over
other types currently grown by farmers in Siabuwa.

The experimental design was a RCBD replicated
three times. Ten sorghum varieties were used:
MR730, SV-1, SV-2, Segaolane, Chisumbanje,
MR748, MR726, MR703, Serena, and the farmers'
local variety (Balala).

The trial was planted during 1986-87 at seven sites
across all soil types in Siabuwa. Plot size and
management responsibilities were similar to those in
the variety x fertilizer trial. Since most farmers do not
use fertilizer on sorghum, fertilizer was not applied in

this trial. The data gathered and methods used at
harvest were also similar to those in the variety x
fertilizer trial.

Farmer and site selection for the experiments-In
1984-85 farmers had to be selected largely for their
willingness to co-operate, but many farmers offered
land for the next season of trials, and so it was
possible to choose farmers with representative
resource levels and soil types for the 1985-86 and
1986-87 seasons.

Several farmers could not provide a sufficiently large
piece of land with a similar cropping history on which
to place a trial. Where necessary, trials were blocked
such that each owner managed one replicate.

Some difficulties were experienced as the trial
programme was implemented. Trials at several sites
showed high coefficients of variation. Results from
those sites have been reported because they reflect
variations that occurred in farmers' fields. Some of
the varieties tested, especially M36172 and to a
lesser extent 321CR and SV-2, had approximately
77% of the numbers of plants/m2 that were present
for the long-season local and another variety,
Serena. Plant population density was in part reduced
by birds, particularly partridges (Perdix perdix), which
dug up seeds and seedlings.

Farmer evaluation of sorghum porridge-Two
tests to evaluate the taste of stiif porridge (the main
food prepared from sorghum grain), a multiple
comparison preference test and a hedonic (or degree
of liking) test (Larmond 1982), were administered
during the 1986 season to 21 panelists randomly
selected from a gathering of farmers. Hot porridge
made from four sorghum varieties (Balala, SV-2,
321CR, and M36172) was served to farmers. Balala,
the local variety most commonly grown by farmers,
was used as the standard. The detailed conduct of
these tests followed the guidelines given by Larmond

In the preference test, farmers were asked to decide
whether each coded sample of SV-2, 321CR, and
M36172 had a better taste than, was comparable to,
or was poorer than the standard, Balala. Farmers
were then asked the amount of difference that
existed between each of the coded samples and the
standard. Scores were assigned to the responses: a
score of 9 indicated a very high preference and a
score of 1 an extreme dislike.

The multiple comparison preference test can only tell
if one sample is preferred to another and cannot
show whether two samples are highly preferred or
not. Accordingly, the hedonic test was carried out to
ascertain how much people really liked the samples.
The standard and test varieties were recorded for this
test. Again a scale of 1 ('dislike extremely') to 9 ('like
extremely') was used for the responses.

An analysis of variance of results from the tests was
carried out and Tukeys test used to determine which
varieties were significantly different from each other.


Rainfall during the three seasons of on-farm trials in
Siabuwa was 775 mm in 1984-85, 720 mm in 1985-
86, and 355 mm in 1986-87. A graph comparing
monthly rainfall for the three seasons with 34 years
of rainfall data from the Siabuwa meteorological
station is shown in Figure 1. The first two seasons,

Rainfall per month (rnm)
300 .

Oct Nov

Dec Jan Feb Mar Apr
//////////////////////A I

SV-2 Y//////////////// I

Figure 1. Relationship between the development cycle
of Balala and SV-2 sorghum and rainfall per month for
the 1984-85, 1985-86, and 1986-87 growing seasons in
Siabuwa, northwestern Zimbabwe. P = planting, A =
50% anthesis, and PP = physiological maturity.

* Calculated from 1951-52 to 1986-87 data.

1984-85 and 1985-86, were among the 47% of years
in which rainfall was above the mean of 676 mm,
whilst 1986-87 had the third lowest rainfall total in the
34-year record. Responses to experimental
treatments, particularly fertilizer, in the trials very
much depended on the pattern and amount of rainfall
in each season.

Variety x fertilizer trial--Variety x season (p =
0.01), variety x location (soil) (p = 0.05), and fertilizer
x season x location (soil) (p = 0.05) interactions for
grain yield were significant and are examined

The grain yield of all varieties was lower during 1985-
86 compared to 1984-85 (Table 1), but in both
seasons all the introduced shorter season varieties
yielded better than the local long-season variety,

Response of variety with season depended on the
cropping season's rainfall pattern and the number of
days needed for a particular variety to reach
physiological maturity. Although rainfall totals for both
1984-85 and 1985-86 were above the 34-year
average (775 mm and 720 mm respectively) the
distribution of rainfall in relation to planting date
favoured the 1984-85 season. In particular, in 1985-
86 there was less rainfall shortly after planting and in
March (during grain filling) (Figure 1). The low

Table 1. Grain yield, number of shoots ha-'1, and
number of panicles ha-'1 at maturity for five varieties
of sorghum, averaged over two fertilizer regimes,
Siabuwa, northwestern Zimbabwe, 1984-85 and

Season and Grain yield Number of Number of
variety (kg ha-1) shoots ha' panicles ha'

M36172 2,583 128,600 127,400
321 CR 2,499 125,700 131,500
SV-2 2,511 100,400 108,600
Serena 3,640 160,300 132,800
Long-season local" 2,223 160,000 141,800

M36172 892 49,000 41,100
321CR 1,636 73,800 69,000
SV-2 1,824 69,100 66,700
Serena 1,244 77,400 70,200
Long-season local" 702 77,400 70,200

SED (season x variety) 123 6,420 7,770

a Balala.

numbers of shoots and panicles per hectare
achieved by all varieties in 1985-86 (Table 1) are
attributable to the low rainfall received just before
and soon after planting. The interaction arose
because the two intermediate maturity materials
(Serena and M36172: 126 and 121 days to
physiological maturity, respectively) and the long-
season local (around 131 days to physiological
maturity) experienced a longer period of water deficit
during grain filling than did 321CR and SV-2 (106
and 94 days to physiological maturity, respectively) in
1985-86 and consequently achieved much lower
grain yields (Table 1, Figure 1).

Relative performance of varieties did differ by site but
there was no consistent link with either the type of
soil at a site or with whether the site was ploughed
(Figure 2). For example, at the two sites on the
sandy clay loam soil, variety SV-2 shifted from being
the highest yielding at one site to being the lowest
yielding at the other. Except for one site on the clay
soil, Serena (intermediate type) was consistently the
best or equally best yielding variety at all sites. The
varieties 321CR and M36172 also yielded
consistently higher than the local, whereas SV-2 was

Grain yield (kg ha"1)
3,500LSD (P = 0.05)
LSD (P = 0.05)

3,000 -

2,500 -

2,000 -

1,500 -



Local variety s
Serena 0
SV-2 A
321CR 0
o M36172 A

0 2


Sandy clay
loam soil

Sandy loam Clay soil

Location (soil)


Figure 2. Response of five varieties of sorghum grown
under ploughed or unploughed conditions on three soil
types in Slabuwa, northwestern Zimbabwe, 1984-86:
variety x fertilizer trial. P = ploughed, NP = not

less consistent but generally yielded more than the
local. Ploughing appeared important for high yield
among the new short-season varieties at sites where
the soil can cap (sandy clay loam) and on the clay
soils (Figure 2).

All varieties, including the local, responded to
fertilizer, but the level of response depended on the
crop season (rainfall distribution) and the location
(soil type) (Figure 3). Grain yield of local with fertilizer
was generally less than the yield of the improved
varieties without fertilizer, especially in 1985-86 when
the long-season local was affected by a water deficit
during grain filling and so could not benefit from the
fertilizer applied (Figure 3). In 1985-86 there was a
consistently higher grain yield (approximately 60%
increase) with fertilizer across all soil types compared
to approximately 37% in 1984-85. For the new
varieties fertilizer effects on the sandy clay loams
and sandy loams tended to be greater at sites that
had been ploughed (Figure 3). Response to fertilizer
did not differ greatly with the different soil types in the
two years (Figure 3), which reflects the similar N and
P status of the soils (Chiduza et al. 1992).

Variety trial-As noted earlier, rainfall during 1986-
87 was the third lowest rainfall in the 34-year record.
Little rain fell in January and none in February
(Figure 1), and the resulting grain yields of the
varieties examined was low (Table 2). Farmers did
not harvest any grain from most of their longer
season sorghum crops in Siabuwa in 1986-87. In the
trial Serena (intermediate) and two later maturing
varieties, Chisumbanje and Balala (the farmers'
local), did not reach physiological maturity because
the growing season ended early. These three
varieties yielded no grain at almost all sites.
However, the earlier maturing varieties (SV-1, SV-2,
Segaolane, MR748, MR703, MR730, and MR726)
yielded some grain. Days to 50% anthesis explain
well the grain yields obtained. Serena, Chisumbanje,
and Balala are 10-18 days later to anthesis
compared to the other varieties (Table 2), and this
meant that the early part of the grain filling phase
coincided with the period of low rainfall (Figure 1).

Segaolane produced above average yields at all sites
while SV-1 was above average at six of the seven
sites. The varieties MR730 and MR748 had above
average yields at some sites but yielded below
average at others. Variety SV-2, which yielded well in
the previous cropping seasons, again yielded well
above average (at six of the seven sites) (Table 2).

Fertilizer levels trial-There was no effect on grain
yield of either the initial dressing or the topdressing of
fertilizer on either the sandy clay loam or the clay
soils in 1985-86 and 1986-87. Relatively high levels
of NPK present in these soils and insufficient
moisture (especially in 1986-87) explain the low
response to fertilizer.

The response to fertilizer during 1986-87 at the two
sites with sandy loam soils was inconsistent (Table
3). Site 2 showed significant yield increases with
increasing levels of both topdressing and the initial

Grain yield (kg ha-')









1 4 LSD (P= 0.05)

0 IF A
e LF A


o o
A o
I Improved varieties with fertilizer D
0 Improved varieties without fertilizer A
1985-86 Local variety with fertilizer *
Local variety without fertilizer 0
LSD (P = 0.05)



o LNF A 8

Sandy clay Sandy loam
soil soil
Location (soil)

Clay soil


Figure 3. Response to applied fertilizer of local and
Improved varieties of sorghum grown on three soil
types in Siabuwa, northwestern Zimbabwe, 1984-86:
variety x fertilizer trial. P = ploughed, NP = not

dressing up to 20:15:9 kg NPK ha-1 basal and 56 kg
N ha-1 topdress, but little response was seen at the
other site.

The different response to fertilizer at the two sandy
loam sites was probably due to the differences in
planting date and rainfall distribution following
establishment of the trial at the two sites. A
prolonged dry spell was experienced soon after
establishment at the non-responsive site and
resulted in little effect of fertilizer on grain yield.

Farmer evaluation of taste of porridge from
sorghum grain-The multiple comparison
preference test showed that farmers preferred the
taste of thick porridge prepared from the grain of the
early maturing varieties SV-2, 321CR, and M36172
significantly more (p = 0.05) than the taste of the
farmers' current variety, Balala. Farmers had the
highest preference for 321CR, followed by SV-2 and
then M36172, compared to Balala.

In the hedonic test there were significant differences
(P=0.01) in how much thick porridge prepared from
the varieties SV-2, 321CR, M36172, and Balala was
liked. Preferences agreed with those from the
multiple comparison test. Porridge made from variety
321CR was liked the most, receiving a score of 8.3
out of 9 (thus it was 'liked very much'). SV-2 was
also 'liked very much' (score 7.8), whereas farmers

had a 'moderate preference' for M36172 (score 7.2).
Balala received a neutral score of 5.0 from farmers,
meaning that it was neither liked nor disliked. The
preference for variety M36172, the least preferred of
the three introduced sorghum varieties, was
nevertheless significantly higher (p = 0.05) than that
of Balala. Thus all three new white early maturing
sorghums were preferred more or equally to varieties
currently grown by the farmers for thick porridge.

Table 3. Effect of initial and topdress fertilizer on
sorghum grain yield (kg ha-1) for sandy loam soils,
Siabuwa, northwestern Zimbabwe, 1986-87

Site 1 Site 2

Initial dress, NPK (kg ha-')
0:0:0 732 1,253
10:7:5 553 1,709
20:15:9 574 2,651
30:22:14 532 2,427
SE 124 220

Topdress, N (kg ha,)
0 427 1,478
28 617 1,722
56 670 2,313
84 677 2,528
SE 105 151

Table 2. Grain yield of 10 sorghum varieties at seven sites, Siabuwa, northwestern Zimbabwe, 1986-87, and
number of days from emergence to 50% anthesis and to physiological maturity for the same varieties grown
under irrigation at Chiredzi, southern Zimbabwe (a location with temperatures similar to those in Slabuwa)

Grain yield (kg ha') Days from
Variety emergence to:
Site mean over 50% Physiological
Variety 1 2 3 4 5 6 7 sites anthesis maturity*

MR730 825 859 523 741 333 288 327 557 76 112
SV-2 1,313 1,763 1,626 1,276 755 150 684 1,081 67 109
Segaolane 1,031 1,444 769 718 774 768 359 838 67 107
SV-1 1,077 1,195 839 789 822 213 908 835 75 111
MR703 947 939 946 645 360 224 294 622 73 121
MR726 517 728 120 323 0 324 268 326 75 122
MR748 731 731 305 646 780 115 522 547 74 123
Serena 0 0 0 0 0 19 0 3 84 126
Chisumbanje 0 0 0 0 0 9 0 1 84 128
Local 444 0 0 0 0 156 0 86 90 131

SE 63.4 99.8 354.0 216.1 202.7 221.4 187.5 90.2 na na

a Physiological maturity determined by appearance of dark hilum.
b Balala.
na = not applicable.


Grain yield of short-season sorghum varieties-
Introduced pearly white grain sorghum varieties with
a short plant development cycle that better fits the
available rainfall season yielded consistently better
(by an average of 62% in three trials) than current
local long-season varieties in the three years of
experimentation. These sorghums sampled a good
(greater than 700 mm of rainfall, well distributed
during the growing season), an average, and a poor
rainfall season in Siabuwa. Varieties such as SV-2
and 321CR'performed well in all the seasons in
which they were tested, while varieties such as
M36172 showed greater potential to take advantage
of a good rainfall season. Given the highly variable
rainfall pattern in Siabuwa, the stability of the yield
improvement offered by the new short-season
sorghum varieties over the different types of season
is important. Seasons with good rainfall occur only
about one year in three in Siabuwa.

Serena, an intermediate maturing improved variety,
yielded better than the short-season varieties SV-2
and 321CR in 1984-85 (when rainfall was good and
well distributed), but in 1985-86 and 1986-87 most
short-season varieties yielded better than Serena.
Data from the variety trial conducted during 1986-87,
which had a total of just 355 mm of rainfall, highlight
the advantage that farmers can obtain by growing the
short-season sorghum varieties. The long-season
local (Balala) yielded 86 kg ha-1 compared with an
average 828 kg ha-' for introduced short-season
varieties under farmer management in that extremely
dry year (Table 2).

Thus in almost all seasons in Siabuwa, any of the
three short-season improved varieties (SV-2,
M36172, and 321CR) would be preferable to the
current local longer season varieties under present
farmer management, including no use of fertilizer.

Farmer food preference for short-season
sorghum varieties-The priority for most farmers in
Siabuwa is to meet their subsistence food
requirements. The crops and varieties of those crops
they produce are dictated by, amongst other things,
food preferences. At the moment one of the main
reasons why farmers grow long-season sorghum
varieties is because they like the taste of porridge
prepared from them. Taste preference has been
shown elsewhere to be important in the adoption of
sorghums by subsistence farmers, e.g., in West

Africa (ICRISAT 1980-83). Results from the food
evaluation tests indicated that farmers in Siabuwa
preferred the taste of porridge from the new
improved short-season sorghum varieties compared
to the current long-season variety, Balala.

If data on grain yield only are considered, then in the
wetter years the intermediate variety Serena had the
best advantage over the long-season local compared
to other varieties. However, no farmer was observed
to grow Serena even though it had been available in
Siabuwa since the early 1980s, while in 1985-86 and
1986-87 a few farmers grew three other improved
shorter season sorghum varieties, SV-2, 321CR, and
M36172, straight after their release. The difference
between Serena and the other varieties, apart from
maturity length, is grain colour and taste. Serena is a
bitter red-grained brewing type of sorghum, whilst the
other three are pearly white grained varieties
specifically bred as food sorghums. Unsuitability for
porridge may explain the low adoption of brewing
sorghum varieties reported by Reid (1982) in Kariba

Adoption potential of short-season sorghum-
The combination of higher, more stable grain yield
with a preferred taste should quickly lead Siabuwa
farmers to adopt the new short-season sorghums.
The new short-season sorghums are open pollinated
varieties, and seed was made available free to
farmers who expressed interest in growing them
during the 1987-88 and 1988-89 seasons. Thus little
extra cost was associated with the use of these
materials. Even if seed has to be purchased, the
change of variety will be extremely economic given
the estimated marginal rate of return of 1,232% in
moving from the farmer's local to SV-2 (Table 4).

Nevertheless it is anticipated that for two main
reasons farmers will continue to plant a proportion of
their sorghum area to longer season sorghums,
especially as ratoon crops on the vertisols. First,
these experiments indicate that long- or intermediate-
season sorghums with grain acceptable for making
porridge and higher yield potential may have a role in
the system to take full advantage of good rainfall
years. However these materials have still to be
developed. Second, a shortcoming of the new short-
season sorghums appears to be their greater
difficulty in establishing a good plant stand on the
vertisols. This was observed at trial sites, especially
in 1985-86 when rainfall was low just after planting,
and the resulting yield reductions require further

study in relation to tillage. Only Serena and the local
achieved good stand establishment on all soil types,
whether ploughed or not. It seems that the short-
season varieties were more responsive to ploughing
and may therefore establish themselves better as
land preparation improves in Siabuwa.

Fertilizer use-Because of the small number of sites
and seasons sampled, the conclusions on fertilizer
use are preliminary.

In very dry seasons similar to 1986-87, fertilizer
generally does not increase yield and so an
economic disadvantage would accrue from applying
fertilizer. Similar dry seasons with less than 450 mm
of rainfall, in which little or no response to fertilizer is
likely, occur about one year in seven in Siabuwa.

During above average or average rainfall seasons,
such as 1984-85 and 1985-86, use of 50 kg N, 12 kg
P, and 8 kg K ha-1 fertilizer should show a yield
increase of about 44% across all soil types,
compared to no fertilizer. Although in these trials the
local variety produced the highest percentage
response to fertilizer, the absolute grain yield for the
local with fertilizer was only slightly higher than that
of the short-season improved varieties without

Table 4 shows net benefits and the marginal rate of
return to cash for applying fertilizer to the farmers'
local, 321CR, and SV-2 sorghum during 1984-85 and
1985-86, the years when fertilizer consistently raised
grain yield. Marginal rates of return in moving from
no fertilizer to 50 kg N, 16 kg P, and 7.6 kg K ha1
were 75% for the farmers' local variety and 96% for
SV-2. Farmers in Siabuwa are risk averse and not
accustomed to purchasing high levels of inputs. As a
general rule, in an environment like Siabuwa farmers
will be unlikely to make an investment unless the
average rate of return is at least 100% per crop per
season (Perrin et al. 1976, CIMMYT 1988).
Consequently the cash returns to fertilizer use are
probably unattractive to farmers in Siabuwa even in
seasons when rainfall is above average.

Thus, preliminary results show the application of
fertilizer to either current long-season local or
improved short-season sorghums in Siabuwa to be
very risky and uneconomic. Fertilizers were not
generally used by local farmers on sorghum at the
time of the trials and these findings support current
farmer practice. However the issue of fertilizer use x
soil type x season interaction does need further
investigation. Lower levels of N fertilizer (20-40 kg N
ha'1) applied as a cheaper straight N source shortly
after emergence (rather than mixture of compound
basal and straight N topdress) may be economic in
years when rainfall is average or better than average,
especially on ploughed land.

Table 4. Costs that vary and net benefits from the use of fertilizer (50 kg N, 16 kg P, and 8 kg K ha-1 split
dressing) with farmer's local, 321CR, and SV-2 varieties, Slabuwa, northwestern Zimbabwe, 1984-85 and 1985-

Total costs Net Marginal rate
that vary benefits of return
Treatment (Z$/ha)' (Z$/ha)b (%)C

Farmers' variety, no fertilizer 0.0 156.1 1
321CR, no fertilizer 6.7 229.7 1,232% 75%
SV-2, no fertilizer 7.0 242.2 J
Farmers' variety, with fertilizer 145.4 265.1
321CR, with fertilizer 152.7 365.8 96%
SV-2, with fertilizer 153.1 382.2

a The post of $0 given to the farmers' variety biases the marginal rate of return (MRR) in favour of this variety. If a seed cost were
attached to the farmers' variety, the MRR in moving to the improved variety would be much higher.
b Yields reduced by 20% to approximate farmers' yields.
c The MRR is calculated only in moving to the best yielding improved variety, SV-2.


In conclusion, results from the experiments reported
here indicate that:

* Use of improved white grain, open pollinated
short-season sorghum varieties acceptable for
porridge, such as SV-2 and 321CR, would
increase grain yield per hectare (and yield
stability) over almost all rainfall seasons under
farmers' current management without use of
fertilizer in the short term. Such short-season
sorghums are useful additions to the portfolio of
sorghum varieties available to farmers in Siabuwa.

* Use of fertilizer (at 50 kg N, 16 kg P, and 8 kg K
ha-1) is at present uneconomic for most seasons
on all the main soil types in Siabuwa, either with
current local long-season sorghums or the new
short-season varieties. Lower rates of fertilizer
need testing.


The authors thank Greg Edmeades, Ren6e Lafitte,
and Compton Paul for helpful comments. The
research reported in this paper was funded by Ford
Foundation grant number 830-0683.

Chiduza, C., S.R. Waddington, and M. Rukuni. 1992.
Production practices, problems, and research
opportunities for smallholder sorghum production in
Siabuwa, Zimbabwe. Farming Systems Bulletin, Eastern
and Southern Africa 10: 11-18.

CIMMYT. 1988. From Agronomic Data to Farmer
Recommendations: An Economics Training Manual.
Completely revised edition. Mexico, D.F.: CIMMYT.

Hungwe, A. 1985. A Study of the Soils of Siabuwa Valley,
Sebungwe Region, and Their Management for Dry Land
Peasant Agriculture. Department of Soil Science,
University of Zimbabwe. Mimeo.

ICRISAT. 1980-83. Rapport annuel. Programme cooperatif.
Ouagadougou, Burkina Faso: International Crops
Research Institute for the Semi-Arid Tropics.

Larmond, E. 1982. Laboratory Methods for Sensory Evaluation
of Foods. Publication 1637. Ottawa, Ontario, Canada:
Food Research Institute.

Perrin, R.K., D.L. Winkelmann, E.R. Moscardi, and T.R.
Anderson. 1976. From Agronomic Data to Farmer
Recommendations: An Economics Training Manual.
Information Bulletin 27. Mexico, D.F.: CIMMYT.

Reid, M.G. 1982. Crops appropriate to the Sebungwe region. In
Sebungwe Regional Study, Proceedings of a Workshop,
February 1982, Hwange Safari Lodge, Zimbabwe.


Support Project for Eastern and Southern Africa:
Training Notes/Manuals, Networkshop
Reports, Other Reports, and Papers Published by Project Staff

Training Notes

No. 1. Case Study of a Diagnostic Survey of a Farming
System in Zambia. These notes follow a low-cost
approach to understanding a farming system in order to
plan relevent adaptive agricultural experiments. July

No. 2. Some Notes on the Sequence of Procedures in a Farmer
Survey: From Formulating Objectives to Using the
Data. August 1979.

No. 3. Some Notes on Development Questionnaires for a
Farmer Survey. August 1979.

No. 4. Formulation of Data Requirements for a Farmer
Survey. November 1979.

No. 5. Some Issues in Off-Station Experimentation.
November 1979.

No. 6. Adaptive Sampling in East Africa. January 1980.

No. 7. An Example of an Enumerator's Reference Manual for
Farm Surveys. February 1980.

No. 8. Design and Management of Survey Research: A Guide
for Agricultural Researchers. March 1980.

No. 9. Guidelines for Enumerator Training for a Single-Visit
Diagnostic Survey. December 1980.

No. 10. How Knowledge of Farmers' Circumstances can be
Useful for Planning Experimental Content in
Agronomy. August 1982.

No. 11. On-Farm Experimentation Concepts and Principles.
November 1985.

No. 12. On-Farm Experimentation-Evaluation of On-Farm
Trials: Statistical Evaluation and Interpretation. August

No. 13. Guidelines for Using OFE Methodology in Crops,
Livestock, and Agroforestry Experimentation.
November 1985.

No. 14. Teaching Notes on the Diagnostic Phase of OFR/FSP
Concepts, Principles, and Procedures. November 1985.

No. 15. Application of the MSTAT Microcomputer Statistical
Program to the Analysis of On-Farm Trials: Tutorial
Manual. April 1987.

No. 16. Introduction to Experimental Evaluation Economic

No. 17. A Guide to the Statistical Analysis of On-farm Trials
for E and S Africa.

Training Manuals

From Agronomic Data to Farmer Recommendations: An
Economics Training Manual. Completely revised edition.
1988. Mexico, D.F.: CIMMYT.

The Planning Stage of On-Farm Research: Identifying Factors
for Experimentation (1989). R. Tripp and J. Woolley.
Mexico, D.F., and Cali, Colombia: CIMMYT and CIAT.

Planning Technologies Appropriate to Farmers Concepts and
Procedures. 1982. Mexico, D.F.: CIMMYT.

Statistical Techniques for the Analysis of On-Farm Crop Trials.
1991. By D. Neeley, J. Matata, And F. Kisyombe. Harare
and Nairobi: CIMMYT.

Networkshop and Networking Reports


Report No. 1 Seminar for Senior Agricultural Research
Administrators from Eastern and Southern
Africa. 18-20 April, 1983, Nairobi, Kenya.

Report No. 2 Report of a Networkshop on Draught Power and
Animal Feeding in Eastern and Southern Africa.
4-6 October, 1983, Ezulwini, Swaziland. June

Report No. 3 Draught Power Problems and Related Research
in Botswana, Lesotho, and Swaziland.
J. Stephen.

Report No. 4 Networkshop on Extension Methods and
Research/Extension Linkages. 10-16 June, 1984,
Eldoret, Kenya.

Report No. 5 Networkshop on Issues in On-Farm
Experimentation, 24-28 May 1984, Lilongwe,

Report No. 6 ARPT/CIMMYT Networkshop on The Role of
Rural Sociology and Anthropology in Farming
Systems Research and Extension.

Report No. 7 Networkshop on the Role of Socio-Economists
and Microcomputers in FSR. 1-4 July, 1985,
Gaberone, Botswana.

Report No. 8 Networkshop of Eastern and Southern African
Senior Agricultural Administrators on Issues in
Systems Based On Farm Research. 24-28
November, 1985, Maseru, Lesotho. July 1986.

Report No. 9 Second On-Farm Research Field Review
Networking Workshop. 12-15 May, 1986,
Malkerns, Swaziland. September 1987.

Report No. 10

Report No. 11

Report No. 12

Report No. 13

Report No. 14

Report No. 15

Report No. 16

Report No. 17

Report No. 18

Report on a Networkshop on Household Issues
and Farming Systems Research. 27-30 April,
1987, Lusaka, Zambia. December 1987.

Research and Extension Administrators
Networkshop on OFR Issues. 5-8 May, 1987,
Lilongwe, Malawi. December 1987.

Review of On-Farm Research in Swaziland:
1982-1988. Ed. J.J. Curry and C. Seubert. 1-4
March 1988, Nhlangano, Swaziland. August

Networkshop on Linkage between On-Farm
Research and Technical Component Research.
21-25 September, 1987, Nairobi, Kenya.

On-Farm Research in Arid and Semi Arid
Regions of IGADD Countries, 23-26 May,
1988, IGADD, Djibouti.

Crop-Livestock Interactions in On-farm
Research, 27-30 June, 1988, Harare, Zimbabwe.
May 1989.

Eastern and Southern Africa Research Extension
Administrators' Workshop. 14-17 November,
1988, Nairobi, Kenya.

Research Methods for Cercal/Iegume
Intercropping. Ed. S.R. Waddington, A.F.E.
Palmer, and O.T. Edje. 23-27 January, 1989,
Lilongwe, Malawi. December 1990.

Issuis Related to Planning, Management, and
Evaluation of On-Farm Experiments. 3-8
December, 1989, Tanzania.

Report No. 19

Report No. 20

Report No. 21

Report No. 22



On-Farm Research and Extension Linkages:
Experience from Swaziland. C. Seubert. October

On-Farm Research and Extension Linkages:
Experience from Zambia's Eastern Province,
1982-1988. J. Waterworth. December 1989.

Farming Systems Research in University
Curricula: A Final Year Practical Option,
University of Zambia. 0. Lungu. February 1990.

Tillage, Past and Future. A Workshop held at the
Institute of Agricultural Engineering, 14 and 15
November 1989, Hatcliffe, Harare, Zimbabwe.
CIMMYT and GTZ. March 1991.

Research-Extension Administrators' Workshop -
Eastern and Southern Africa. 8-12 April, 1991,
Nairobi, Kenya.

Impacts of On-farm Research. A Technical
Networkshop. 23-26 June 1992, Harare,


OFR Program Review Workshop. 16-22 May, 1987, Rwanda.

Workshop on Institutionalization of OFR/FSR Training in
National Training Institutions, 23-25 May, 1989.

National FSR Programme Review/Orientation Workshop. 12-17
November, 1989, Uganda.

National Workshop on Institutionalization of OFR/FARM
training in Tanzania. 4-9 November, 1990.

Middle Level Policy Makers Workshop. 22-24 October, 1990,
Njoro, Kenya.

High Level Policy Makers FSA R/E/T Orientation Workshop.
25-26 October, 1990, Naivasha, Kenya.

Other Meetings that Received
Support from the FSR Project

The Second East, Central, and Southern African Regional Maize
Workshop. 15-21 March, 1987, Harare, Zimbabwe.

The Third Eastern and Southern Africa Regional Maize
Workshop. 17-23 September, 1989, Nairobi, Kenya.

Regional Wheat Workshop. 2-6 October, 1989, Ethiopia.

Informal Meeting of the CG and Non CG Economists in the
Region. 11 June, 1990, Nairobi, Kenya

Workshop on Coordination of IARCs' Training Activities in
Africa. 26-28 November, 1990.

CG and Non-CG Social Scientists Workshop. 12-13 April,
1991, Nairobi, Kenya.

Regional Wheat Workshop. 16-19 September, 1991, Kenya.

Papers and Reports Published
by Project Staff, 1987-1992


Anandajayasekeram, P. 1987. CIMMYT's mode of
collaboration in OFR training for national research
programmes. In Research and Extension Administrators
Networkshop on OFR Issues (Networkshop Report No.
11). Pp. 125-143.

Anandajayasekeram, P. 1987. Training at CIMMYT. Presented
at the Regional Workshop on Training Needs for
Agricultural Research in Eastern and Southern Africa,
All India Congress Committee, 20-24 July, 1987,

Collinson, M.P. 1987. Farming systems research procedures for
technology development. ExperimentalAgriculture 23:

Low, A. 1987. The place and functional relationship of FSR in
research and extension systems in Eastern and Southern
Africa. In Improving Food Crop Production on Small
Farms in Africa. Harare, Zimbabwe: FAO/SIDA. Pp.

Low, A. 1987. Diagnostic training for extension staff. Fifth
Annual Agricultural Extension and Research Conference,
15-18 June, 1987, Kariba.

Whingwiri, E., Natarajan, G. Makombe, D. Mataruka, A. Low.
T. Cameron, and P. Kunjeku. 1987. The package for
higher maize yields in the semiarid zones (Natural
Regions III and IV): Is it economical? In Vol. 2 of
Cropping in the Semiarid Areas of Zimbabwe. Harare,
Zimbabwe: AGRITEX, Department of Research and
Specialist Services, and GTZ.


Anandajayasekeram, P., and A.F.E. Palmer. 1988. Evaluation
of on-farm trials for deriving recommendations.
Presented at the Tanzania National Maize Workshop, 6-
11 June, 1988, Arusha, Tanzania.

Low, A. 1988. Farm household economics and the design and
impact of biological research in Southern Africa.
Agricultural Administration and Extension 29: 23-34.

Low, A., and P. Kunjeku. 1988. The perspective and prospect
for technological changes in SADDC food production,
with particular reference to the rate of such changes.
Prepared for SADDC Early Warning Seminar, 21 June -
29 July, 1988.

Low, A., D.F. Mataruka, and G. Makombe. 1988. The use of
economic analysis in researcher managed trials.
Economic Analysis in Research Results: Part 1. Harare,
Zimbawe: Department of Research and Specialist

Mataruka, D.F., G. Makombe, and A. Low. 1988. Example of a
maize nitrogen x phosphorus trial in Zimbabwe.
Economic Analysis in Research Results: Part 2. Harare,
Zimbabwe: Department of Research and Specialist'

Palmer, A.F.E., P.,Anandajayasekeram, and J.K. Ransom.
1988. Defining and establishing research priorities.
Presented at the Tanzanian National Maize Workshop, 6-
11 June, 1988, Arusha, Tanzania.

Ransom, J.K., P. Anandajayasekeram, and A.F.E. Palmer.
1988. A "call" system approach for teaching on-farm
research methodologies. Agronomy Abstracts 66.

Ransom, J.K., S. Waddington, and P. Kunjeku. 1988. Potential
technology and research needed for rainfed cereal
production in drought prone environments of Eastern and
Southern Africa. In Forum on On-Farm Research inArid
and Semi Arid Regions ofIGADD (Inter-Governmental
Authority on Drought and Development) Countries.
Djibouti. Pp. 189-208.

Waddington, S.R. 1988. Crop systems research: Overview of
the CIMMYT approach. Presented at the Networkshop
on Crop-Livestock Interactions in On-Farm Research,
June, 1988, Harare.

Waddington, S., and A. Low. 1988. CIMMYT's on-farm
research support program for eastern and southern
Africa. In B. Gelaw (ed.), Towards SelfSufficiency:
Proceedings of the Second Eastern, Central, and
Southern Africa Regional Maize Workshop. Harare,
Zimbabwe: CIMMYT. Pp. 160-167.


Anandajayasekeram, P., G. Durr, and G.M. Karanja. 1989.
Challenges of the regional research centers in the
generation, testing, and dissemination of appropriate
technologies for small-holder farmers in Kenya.
Presented at the Annual Scientific Conference of the
Kenya Agricultural Research Institute, 14-16 August,
1989, Nairobi, Kenya.

Anandajayasekeram, P., and C.N. Muriithi. 1989. The role and
place of on-farm research with the farming systems
perspective (OFR/FSR) in the National Agricultural
Research Project (NARP) and its implications. Presented
at the Kenya Agricultural Research Institute National and
Regional Research Programme Development Workshop,
26 November 1 December, 1989, Nairobi, Kenya.

Anandajayasekeram, P., and M. Rukuni. 1989.
Institutionalization of on-farm research with farming
systems perspectives (OFR/FSP) in eastern and southern
African region: Achievements and future direction.
Zimbabwe Journal ofAgricultural Research 27:67-81.

Dereje, D., D. Tanner, and W. Mwangi. 1989. On-farm
verification of four bread wheat varieties under high and
farmers' weeding management levels. 1989.

Durr, G., and P. Anandajayasekeram. 1989. The role of social
scientists in on-farm experimentation. Paper presented at
the Regional Technical Networkshop on Issues Related
to Planning, Management, and Evaluation of On-Farm
Experiments, 4-8 December, 1989, Arusha, Tanzania.

Low, A., and G. Makombe. 1989. Targeting recommendations
for small farmers: The role of economic factors in
Zimbabwe and Swaziland. Farming Systems Bulletin,
Eastern and Southern Africa 4:22-26.

Palmer, A.F.E., and P. Anandajayasekeram. 1989. The
importance of site descriptors and farmer management
practices in on-farm trial evaluation. Presented-at.he
Regional Technical Networkshop on Issues Related to
Planning, Management and Evaluation of On-Farm
Experiments, 4-8 December, 1989, Arusha, Tanzania.

Ransom, J.K., P. Anandajayasekeram, and A.F.E. Palmer 1989.
A call system approach for training on-farm research
methodologies. Presented at the American Society of
Agronomy Meeting, 29 November 3 December, 1989,
Anaheim, California.

Shumba, E.M., S.R. Waddington, and M. Rukuni. 1989.
Delayed maize plantings in a smallholder farming area of
Zimbabwe: Problem diagnosis. Zimbabwe Journal of
Agricultural Research 27: 103-112.

Waddington, S.R. 1989. CIMMYT Southern Africa Regional
Maize Agronomy Program. In "First Meeting of
CIMMYT Maize Agronomists, Collected Papers."
Mexico, D.F.: CIMMYT. Pp. A66-91.

Waddington, S.R., and P. Kunjeku. 1989. Potential technology
and research needs for rainfed maize production in
drought prone environments of southern Africa. Farming
Systems Bulletin, Eastern and Southern Africa 3: 28-41.

Waddington, S. 198 On-Farm Experimentation with
Intercrops: A Summary of Issues from a Workshop Hleld
in Lilongwe, Malawi, Janhy 1989. Presented at the
Regional Technical IMetworkshop on Issues Related.to
Planning, Management and Evaluation of On-Farm
Experiments, 4-8 December, 1989, Arusha, Tanzania.


Anandajayasekeram, P. 1990. Some reflections on agricultural
economics research, its integration with post-graduate
training, and institutional support from IARCs in Eastern
and Southern Africa. Presented at the Agricultural
Economics Subject Matter Workshop, 15-18 October,
1990, Harare, Zimbabwe.

Anandajayasekeram, P. 1990. Institutionalization of on-farm
research with farming systems perspectives (OFR/FSP)
in Eastern and Southern African Region: Achievements
and future directions. Presented at the'National
Workshop on Institutionalization of OFR/FSP Training
in Tanzania, 4-9 November, 1990, Morogoro, Tanzania.

Anandajayasekeram, P., A. Low, and G. Durr. 1990. Economic
evaluation and interpretation of intercropping trials. In
S.R. Waddington, A.F.E. Palmer, and O.T. Edje (eds.),
Research Methods for Cereal/Legume Intercropping.
Mexico, D.F.: CIMMYT and CIAT. Pp. 147-153.

Anandajayasekeram, P., B. Grisley, and P. Ewell. 1990.
Summary Report of Informal Meeting of the CG and
Non-CG Economists. Held at ILRAD, 11 June, 1990.

Anandajayasekeram, P., et al. (1990). Regional Research
Centre Review Mission Report. Volume 1, Highlights of
General Issues Across Centres. Volume 2, National Plant
Breeding Research Centre, Njoro. Volume 3, National
Agricultural Research Centre, Kitale. Volume 4,
Regional Research Centre, Mtwapa. Volume 5, Regional
Research Centre, Embu. Volume 6, Regional Research
Centre, Kakamega. Reports submitted to the Kenya
Agricultural Research Institute, 1990.

Dadi, L., W. Mwangi, and S. Franzel. 1990. Socioeconomic
constraints to maize production in Ethiopia. In B.
Gebrekidan (ed.), Maize Improvement, Production, and
Protection in Eastern and Southern Africa: Proceedings
of the Third Eastern and Southern Africa Regional Maize
Workshop, Nairobi, Kenya. CIMMYT. Pp. 436-449.

Ilailu, B., W.M. Mwangi, and W. Negatu. 1990. Review of
research on wheat production constraints in Ethiopia.
Presented at the National Wheat Research Review
Workshop, Addis Ababa, Ethiopia.

Ilailu, B., S. Franzel, and W.M. Mwangi. 1990. Constraints to
increasing wheat production in Ethiopia's sm:dl-holdcr
sector. In Sixth Regional Wheat Workshop for Eastern,
Central, and Southern Africa. Mexico, D.F.: CIMMYT.

Heisey, P.W. 1990. Maize research in Malawi: A comment.
Journal of International Development 2(2): 243-253.

Low, A. 1990. Overview of procedures for diagnosis. In S.R.
Waddington, A.F.E. Palmer, and O.T. Edje (eds.),
Research Methods for Cereal/Legume Intercropping.
Mexico, D.F.: CIMMYT and CIAT.

Low, A.C., and S.R. Waddington. 1990. On-farm research on
maize production technologies for smallholder farmers in
southern Africa: Current achievements and future
prospects. In B. Gebrekidan (ed.), Maize Improvement,
Production, and Protection in Eastern and Southern
Africa: Proceedings of the Third Eastern and Southern
Africa Regional Maize Workshop, Nairobi, Kenya.
CIMMYT. Pp. 491-511.

Low, A., and S. Waddington. 1990. Maize adaptive research:
Achievements and prospects in southern Africa. Farming
Systems Bulletin, Eastern and Southern Africa 6: 29-41.

Mataruka, D.F., G. Makombe, and A. Low. 1990. The
contribution of economic analysis in developing
promising research agendas Example of a maize
nitrogen x phosphorus trial in Zimbabwe. Farming
Systems Bulletin, Eastern and Southern Africa 5: 1-5.

Mudhara, M., and A. Low. 1990. Comparative economics of
maize production in risky environments in Zimbabwe.
Farming Systems Bulletin, Eastern and Southern Africa

Mudhara, M., and S. Waddington. 1990. Study of.maize
planting dates in Natural Region 2 and 3 Communal
Areas of Zimbabwe, 1989-90 season. Preliminary
Report, June, 1990.

Mwangi, W.M. 1990. Fertilizer demand functions under
varying levels of prices and operating capital in Kenya.
Eastern Africa Journal of Rural Development 16: 78-89.

Smale, M., and P. Heisey. 1990. New evidence on varietal
adoption and classification. Report presented to the
Malawi Maize Project (Rockefeller Foundation) Phase 2
Meeting, 9-12 July, 1990, Mangochi.

Tanner, D.F., M. van Ginkel, and W. Mwangi (eds.). 1990.
Sixth Regional Wheat Workshop for Eastern, Central,
and Southern Africa. Mexico, D.F.: CIMMYT.

Tripp, R., and P. Anandajayasekeram. 1990. Building
institutional capacity for on-farm research: Call system
training courses. Journal of Agronomic Education 19(2):

Tripp, R., P. Anandajayasekeram, and G. Sain. 1990. The
Design and Management of Call System Training in On-
Farm Research. CIMMYT Economics Working Paper
90/01. Mexico, D.F.: CIMMYT.

Tripp, R., P. Anandajayasekeram, D. Byerlee, and L.W.
Harrington. 1990. Farming systems research revisited. In
C.K. Eicher and J.M. Staatz (eds.),Agricultural
Development in the Third World. Baltimore, Maryland:
Johns Hopkins. Pp. 384-399.

Waddington, S.R. 1990. On-farm experimentation with
intercrops: A summary of issues from a workshop held in
Malawi. Farming Systems Bulletin, Eastern and Southern
Africa 7: 20-26.

Waddington, S.R., A.F.E. Palmer, and O.T. Edje (eds.). 1990.
Research Methods for Cereal/Legume Intercropping.
Proceedings of a Workshop on Research Methods for
Cereal/Legume Intercropping in Eastern and Southern
Africa. Mexico, D.F.: CIMMYT and CIAT.

Waddington, S.R., A. Low, and P. Kunjeku. 1990. Planning of
on-farm intercrop trials. In S.R. Waddington, A.F.E.
Palmer, and O.T. Edje (eds.), Research Methods for
Cereal/Legume Intercropping. Mexico, D.F.: CIMMYT
and CIAT. Pp. 89-93.


Anandajayasekeram, P., and S. Waddington. 1991. The linkage
issues with respect to on-farm research with a farming
systems perspective (OFR/FSP). Presented at the Eastern
and Southern Africa Regional Research and Extension
Administrators' Workshop, 8-12 April, 1991, Nairobi,

Anandajayasekeram, P., and J.K. Ransom. 1991. CIMMYT's
approach to sustainable soil and water management for
small scale farmers. Presented at the World Association
of Soil and Water Conservation / International Institute
for Environment and Development Workshop on Soil
and Water Management for Sustainable Small Holder

Anandajayasekeram, P., J.B.W. Matata, G. Angwenyi, M.W.
Oggema, and E.O. Wandera. 1991. National guideline on
farming systems approach to research, extension, and

Ensermu, R., A. Kefyalew, and W.M. Mwangi. 1991. On-farm
verification of improved bread wheat varieties and
management practices in Gojam region of Ethiopia.
Presented at the Seventh Regional Wheat Workshop for
Eastern, Central, and Southern Africa,: 16-19 September,
1991, Nakuru, Kenya.

Grisley, W., W. Mwangi, and G. Degu. 1991. Food self-
sufficiency, fertilizer use, and access to formal credit: A
test of the relationships on small farms in Ethiopia.
Poster paper presented at the 21st International
Conference of Agricultural Economists. 22-29 August,
1991, Tokyo, Japan.

Hassan, R., and W.M. Mwangi. 1991. Wheat production
technologies in Kenya: A diagnostic analysis of the
major characteristics and constraints to productivity
growth. Presented at the Seventh Regional Wheat
Workshop for Eastern, Central, and Southern Africa. 16-
19 September, 1991, Nakuru, Kenya.

Heisey, P.W. 1991. Building social science capacity within
national agricultural research systems. Paper presented at
the Research/Extension Administrators' Networkshop, 8-
12 April, 1991, Nairobi, Kenya.

Heisey, P.W. 1991. Economic analysis of the 1989-90 MOA/
FAO fertilizer demonstrations. July, 1991. Lilongwe,
Malawi: CIMMYT.

Heisey, P.W., and J.P. Brennan. 1991. An analytical model of
farmers' demand for replacement seed. American
Journal ofAgricultural Economics 73: 1044-1052.

Low, A., and S.R. Waddington. 1991. Farming systems
adaptive research: Achievements and prospects in
southern Africa. Experimental Agriculture 27: 115-125.

Low, A.R.C., S.R. Waddington, and E.M. Shumba. 1991. On-
farm research in southern Africa: The prospects for
achieving greater impact. In R. Tripp (ed.), Planned
Change in Farming Systems: Progress in On-Farm
Research. Chichester, England: John Wiley. Pp. 257-272.

Low, A., C. Seubert, and J. Waterworth. 1991. Extension of On-
Farm Research Findings: Issues from Experience in
Southern Africa. CIMMYT Economics Working Paper
91/03. Mexico, D.F.: CIMMYT.

Mudhara, M., and S. Waddington. 1991. A short report on a
survey of maize plantings in Zambia, 1989/90 season.
Harare: CIMMYT.

Negatu, W., W.M. Mwangi, and T. Tesemma. 1991. Farmers'
varietal preferences for durum wheat in Ada, Lume, and
Gimbichu Woredas of Ethiopia. Presented at the Seventh
Regional Wheat Workshop for Eastern, Central, and
Southern Africa, 16-19 September, 1991, Nakuru,

Negatu, W., and W.M. Mwangi. 1991. On-farm economics of
herbicide use in durum wheat in Ada and Akaki Woredas
of Ethiopia. Presented at the Seventh Regional Wheat
Workshop for Eastern, Central, and Southern Africa, 16-
19 September, 1991, Nakuru, Kenya.

Shumba, E.M., S. Waddington, and L.A. Navarro (eds.). 1991.
Research and Extension Linkages for Small Holder
Agriculture in Zimbabwe. Proceedings of a Workshop on
Assessing the Performance of the Committee for On-
Farm Research and Extension (COFRE), 7-9 May 1990,
Kadoma, Zimbabwe. Harare, Zimbabwe: Department of
Research and Specialist Services, AGRITEX, and
International Development Research Centre.

Smale, M., with S.H.W. Kaunda, H.L. Makina, M.M.M.K.
Mkandawire, M.N.S. Msowoya, D.J.E.K. Mwale, and
P.W. Heisey. 1991. 'Chimanga Cha Makolo', Hybrids,
and Composites: An Analysis of Farmers' Adoption of
Maize Technology in Malawi, 1989-91. CIMMYT
Economics Working Paper 91/04. Mexico, D.F.:

Smale, M., P.W. Heisey, and H.D. Leathers. 1991. Maize of the
ancestors and modern varieties: The microeconomics of
HYV adoption in Malawi. Paper submitted for journal

Smale, M., and P.W. Heisey. 1991. Seminar to present results
of Malawi maize technology and varietal adoption study.
Conducted by M. Smale and Malawi Ministry of
Agriculture for personnel from Malawi Department of
Agricultural Research (research) and Malawi Department
of Agriculture (extension).

Tanner, D.G., and W.M. Mwangi. 1991. Current issues in
wheat research and production in Eastern, Central, and
Southern Africa: Constraints and achievements. Paper
presented at the Seventh Regional Wheat Workshop for
Eastern Central, and Southern Africa, 16-19 September,
1991, Nakuru, Kenya.

Waddington, S.R. 1991. Agronomic problems related to tillage
in smallholder cereal production, and some research
opportunities. In Tillage, Past and Future. Proceedings
of a workshop held at the Institute of Agricultural
Engineering, 14 November, 1989, Hatcliffe, Harare,
Zimbabwe. CIMMYT FSR Networkshop Report No. 22.
Harare, Zimbabwe: GTZ and CIMMYT. Pp. 46-57.

Waddington, S.R. 1991. CIMMYT on-farm research support
project for southern and eastern Africa. FSR-E Southern
Africa Newsletter, June 1991. Pp. 5-9.

Waddington, S.R. 1991. CIMMYT maize crop management
research and regional activities in southern Africa.
Presented at the CIMMYT Maize Presentations, 7-9
March, 1991, Harare, Zimbabwe.

Waddington, S. 1991. Maize development and growth x
planting date study, Harare, 1990-91 season. Report of
an experiment done at Harare Station. Harare,
Zimbabwe: CIMMYT.

Waddington, S.R., and P.W. Hleisey. 1991. Adaptive research
achievements and failures, and its focus for the future in
southern Africa. In Vol. 2. of ARPT in the 1990s. Report
of the 1991 ARPT Half-Yearly Meeting, 19-21 February,
1991, Siavonga, Zambia. Lusaka, Zambia: Research
Branch, Department of Agriculture. Pp. 104-115.

Waddington, S.R., and M. Hlatshwayo. 1991. Agronomic
monitoring to determine causes of low yield in later
planted maize, Mangwende communal area, Zimbabwe
1987-88 and 1988-89 seasons. CIMMYT Maize Program
Report. Harare, Zimbabwe: CIMMYT.

Waddington, S.R., M. Mudhara, M. Illatshwayo, and P.
Kunjeku. 1-9.91. Extent and causes of low yield in maize
planted late by smallholder farmers in subhumid areas of
Zimbabwe. Farming Systems Bulletin, Eastern and
Southern Africa 9: 15-31.


Chiduza, C., S.R. Waddington, and M. Rukuni. 1992.
Production practices, problems, and research
opportunities for smallholder sorghum production in
Siabuwak,Zimbabwe. Farming Systems Bulletin, Eastern
and Southern Africa 10: 11-18.

Chiduza, C., S.R. Waddington, and M. Rukuni. 1992. On-farm
evaluation of short season sorghum and fertilizer for
smallholder farmers in a semi-arid region of Zimbabwe.
Farming Systems Bulletin, Eastern and Southern Africa

Heisey, P., and W. Mwangi. 1992. Overview of methods of
measuring research impacts. Paper presented at the
CIMMYT Networkshop on Impacts of On-farm
Research, 23-26 June 1992, Harare, Zimbabwe.

Kawonga, W., and P.W. IHeiscy. 1992. Response of four maize
varieties to nitrogen at Lisungwi extension planning area,
Mwanza, Malawi. Farming Systems Bulletin, Eastern and
Southern Africa 11:

Negassa, A., W. Mwangi, and II. Beyene. 1992. Impacts of on-
farm research on farmers and their adoption of
technologies around Bako, Ethiopia: Some selected
cases. Presented at the CIMMYT Networkshop on
Impacts of On-farm Research, 23-26 June 1992, Harare,

Ransom, J., K. Short, and S. Waddington. 1992. Improving
productivity of maize in stress environments. Presented
at the Maize Research Review Meeting, April, 1992,

Seyoum, K., H. Beyene, and W. Mwangi. 1992. The impact of
on-farm research in the Institute of Agricultural Research
of Ethiopia. Paper presented at the CIMMYT4
Networkshop on Impacts of On-farm Research, 23-26
June 1992, Harare, Zimbabwe.

Shumba, E.M., and S.R. Waddington. 1992. Zimbabwe's
experience with agricultural research for smallholder
farmers. Presented at the South African Society of Crop
Production Congress, 21-23 January, 1992,
Bloemfontein,_South Africa.

Shumba, E.M., and S.R. Waddington. 1992. Links between
research and extension for smallholder agriculture: The
Zimbabwe case. Invited review submitted to the
Zimbabwe Journal ofAgricultural Research.

Shumba, E.M., S.R. Waddington, and M. Rukuni. 1992. Tine
tillage, with atrazine weed control, to permit earlier
planting of maize by smallholder farmers in Zimbabwe.
Experimental Agriculture (forthcoming).

Shumba, E.M., R.H. Bernsten, and S.R. Waddington. 1992.
Maize and groundnut yield gap analysis for research
priority setting in the smallholder sector of Zimbabwe.
Zimbabwe Journal ofAgricultural Research

Smale, M., and P.W. Heisey. 1992. Simultaneous estimation of
seed-fertilizer adoption decisions: An application to
hybrid maize in Malawi. Paper submitted for journal

Waddington, S.R. 1992. The future for on-farm crop
experimentation in southern Africa. Presented at the
CIMMYT Networkshop on Impacts of On-farm
Research, 23-26 June 1992, Harare, Zimbabwe.

Waddington, S.R., and E.M. Shumba. 1992. Late planted
maize: A source of low productivity for farmers in
subhumid areas of Zimbabwe. In Looking Back -
Planning Ahead, Proceedings of the 6th Australian
Agronomy Conference. 10-14 February 1992, Armidale,
NSW, Australia. Australian Society of Agronomy.


IITA have sent us information about thp following book:

Title: On-Farm Research in!Theory and Practice

Editors: H.J.W. Mutsaers (agronomist) and P. Walker (biometrician)

Publisher: International Institute of Tropical Agriculture (IITA), P.M.B. 5320, Ibadan, Nigeria

The book On-Farm Research in Theory and Practice
is the proceedings of a workshop on the design and
analysis of on-farm trials held at IITA, 27 February to
3 March 1989.

This book is divided into two parts; the first is a
collection of papers on the elements of on-farm
research methodology, while the second part
concentrates on case studies.

The authors of papers in the first part of the book
suggest various approaches or methodologies to
carry out field surveys, on-farm trials, and data
analysis, depending on the issues involved. The book
calls for a balance between diagnosis and technology
testing, adding that the majority of OFR programs
rely on an informal exploratory survey for diagnosing
constraints, but in most cases attention is soon
turned to technology testing. The authors say that
more emphasis is needed on diagnosis that
combines socioeconomic and agronomic data.

The authors recommend that whatever the results of
surveys and the available technologies, farmers
should be consulted more on the type of technology
to be tested and on methods of testing, irrespective
of the type of trial.

In addition, the book suggests emphasis on
participatory trials, in which farmers are involved in
the design of experiments.

With regard to trial design, the authors explain that
researcher-managed on-farm trials are not different
in principle from station trials, apart from limitations
on number of plots per site.

Data collection from trials is another important area
considered in the book.

The complementarity between economic analysis
and farmer evaluation of technologies is recognized.
Economic analysis without farmer assessment is
likely to overlook factors which are not immediately
obvious to researchers, but which need to be

To complement the discussions in the first part of the
book, case studies were collated and published in the
second part. Part 2 contains 13 papers describing
personal field experiences of OFR scientists in West
and Central African countries, including Benin,
Cameroon, C6te d'lvoire, Nigeria, and Zaire. Some of
the papers describe the yield potential of improved
varieties in comparison with farmer's varieties, others
discuss the effect of chemical weed control, fertilizer
application, and general crop management practices.

Agronomists and extension agents, particularly those
involved in the adaptation and transfer of
technologies in food crop production, will find this
book useful. The book is also relevant to the work of
agricultural economists and socio-economists who
interact with farmers to assess the profitability of new
technologies recommended by research institutes.

To request a copy of this book, please contact:

International Institute of Tropical Agriculture
c/o L W Lambourn and Co.
Carolyn House
26 Dingwall Road

Tel. Ibadan, Nigeria 400300-400314
Telex TDS IBA NG 20311 (BOX 015) or

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

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