R&Pujblique du Niger
Miniist&-re du Dr:veloppernerit. Rural
Institute National de Recherches Agrornomiques du Niger
D6part.erment de Recherches en Ecornomie Rurale
Department de Recherches Agricoles
Drparterent. de Recherches Ecol ogi ques
The Evolution of Farming Systems Research
at. the National Institute of Agricultural
Research of Niger
S. Ly, R. Deuson, M. GoubI,
G. Numra, C. Reddy, et S. Swinrton
on Farming Systems
held at Kansas
at the Symposium
Research and Extension
October 13-16, 1985.
Document. No. 10E
The authors are listed in alphabetical order after Samba Ly:,
Head of the Department of Research in Rural Economics (DECOR) of
INRAN. Robert Deuson is Assistant Professor in the Department of
Agri cul tural Economics at Purdue University (West Lafayette,
Ind-iana) and Seni or Economist assigned to DECOR. Moussa Gi:oubl: -is
Head of the Soil Chemnistry and Fertility Section in the Depar-
tment of Ecological Research (DRE). Gilbert Numa is an agricul -
tural economist with the Fonds d'Assistance et Cooperation (FAC)
assigned to DECOR. Chandra Reddy is Assistant Professor in the
Department of Agronomy at Alabama A & M Uniiversity (Normal,
Alabama) and assigned to the General Agroniomy Section of the
Department of Agronomic Research (DRA). Scot.t Swinton is Inst-
ructor in the Department of Agricultural Economics at Purdue
University assigned to DECOR. Deuson, Reddy and Swinton work on
the Niger Cereals Research Project (NCRP) fina-nced by USAID under
cont ract no. 683-0225-820.
The authors wish to thank Issa Mahamane and Maliki Kadi for
the constructive comments and corrections which they cont ributed
on the early drafts of this Paper.
THE EVOLUTION OF FARMING SYSTEMS RESEARCH AT THE
NATIONAL INSTITUTE OF AGRONOMIC RESEARCH IN NIGER
The authors summarize the history of farming systems re-
search at the National Institute of Agronomic Research of Niger.
In 1983, INRAN undertook a Farming Systems Research Program with
assistance from a USAID-funded project in collaboration with
Purdue University and Alabama A & M University. The program
seeks to identify farm-level constraints by agro-climatic zone
and to provide on-station researchers with farming systems infor-
mation that will allow them to direct their research accordingly.
A dual methodology uses both long-term village-level data collec-
tion and rapid reconnaissance surveys by mul tidiscipl inary teams
of INRAN researchers. A working group composed of biological and
socio -economic researchers is develop ng onr-stat-i on research
programs to address the identified constraints. Promising tech-
nologies are subsequently tested on farmers' fields in distinct
agro-cl i matic regions. The participation of INRAN administrators
in the working group is helping to institute onal ize this emergent
L'EVOLUTION D'UN PROGRAMME DE RECHERCHE SUR SYSTEMS DE PRODUCTION
A L'INSTITUT NATIONAL DE RECHERCHES AGRONOMIQUES DU NIGER
Les auteurs rsument 'histori-iue de la recherche sur sys-
temes de production agricole entreprise A 1 Instit ut. National de
Recherches Agronomiques du Niger (INRAN). En 1983, 1 INRAN a
entrepris un programme de recherches sur syst.mes de production
agricole dans le cadre d'un project financed par 'USAID en colla-
boration avec PFurdue University et Alabama A & M University
(U.S.A.). Le programme tente d'cidenti fier les contraintes au
niveau de 1'exploitation agricole par zone agro-climatique afin
de donner aux chercheurs sur station des information ns sur les
syst.mes de production quli leur permettr-oant de modeler leur
recherche en consequence. Urne mrnetc thodolog:ie double utilise et la
collect de donnres dans le temps au niveau des villages, et des
sondages rapids effects par les chercheurs de 'INRAN gro.up&s
en aquipes pluridiscipl inaires. Un group de travail, compose de
chercheurs danesles sciences biologiques et dans les sciences
sociales et aconomiques developpent ensemble des programmes de
recherche sur stati on qui s'adressent aux:: contrai ntes -identi -
fiies. Les technologies prometteuses sont ensuite testes en
milieu rnel dans diffrentes zones agro-climatiques. La partici-
pati on des administrateurs de 'INRAN au group de travail permit
1 insti. tuti ornal i sati onr de cette nouve lle mrthocdologi e de
recherche sur syst.rmes de production.
I. Introduction ........................... ........... .... .... ... 1
2. An Introduction to I.N.R.A.N .............................. 2
2.1 History of the Creation of I.N.R.A.N. ................. 2
2.2 Research Directions at I.N.R.A.N ..................... 2
2.3 Research Infrastructure .............................. 2
3. From Research to Extension ................................. 3
3. 1 The Traditional Approach ............................... 3
3.1.1 Otn-Station Research and the "E::xtension
Recommendations Themes" .......................... 3
3.1.2 Agricultural Extension and the
Multi locati onal Trials ......................... 3
3.2 The Development of the Farming Systems
Approach at I.N.R.A.N. ................................4
3.2.1 Monitoring the "Test Farmer" at Tarna .........5
3.2.2 Monitoring Two Graduates of the Farmer
Training Centers ............. .... ............5
3.2.3 The "Experimental Agricultural
Production Units" Operation ................. 6
3.2.4 The Conception and Birth of the
Current Approach ....................... .... 8
4. The "Farming Systems Research" Working Group ...............9
5. Current Research on Farming Systems at I.N.R.A.N. ........10
5. 1 The Multidisciplinary Recornaissance
Survey (sondeo) ..................................... 10
5.2 On-Stat-i on Agronomic Research ....................... 12
5.3 On-Farm Research .................................... 13
5.3. 1 Socio-Economnic Research ....................... 13
5.3.1. 1 Objectives .................................. 13
5.3. 1.2 Methodology ........................... 14
5.3. 1.3 Results ....................... ..... 15
188.8.131.52 Feedback .............................. 16
5.3.2 On-Farm Trials ..........................17
184.108.40.206 Objectives .............. ........ 17
220.127.116.11 Types of On-Farm Trials ......... 18
18.104.22.168 Trials Adapted to Local
Conditions .................... 18
22.214.171.124 Analysis of On-Farm
Trials ............................. 19
6. Conclusions ...............................................20
Bibliography ............... ................ ............ 23
Table 1. Characteristics of the recommendati ons
domains ........................................ 24
Table 2. Salient characteristics of surveyed farms
in 1984 by recommendations domain .................. 25
Figure 1. Location of the study zones in Niger ............... 26
Niger is a country of 1,267,000 square kilometers situated
in the heart of West Africa. It is bordered on the east by Chad,
on the west by Mali and Burkina Faso, on the north by Algeria and
Libya, and on the south by Nigeria. Besides being very large,
Niger is a desert country (approximately four-fifths of its area
being covered by the Sahara). Climatic variability among other
factors exerts a strong influence over the country's economic and
Agriculture, which employs 80% of the population of six
million, is li mited by fickle rainfall and high temperatures. The
rainy season lasts only four months, from June to September.
Average temperatures from 27 to 29 degrees centegrade are typi-
cal, with maxima in May-June and September-October.
Given such conditions, people practice manual, extensive
agriculture centered around millet, sorghum and cowpea, which are
often grown in association (Reddy and Gonda 1985 and Swinton,
Nurna and Ly 1985).
However, chronic droughts and soil deterioration have broken the
cropping cycle and undermined traditional practices. Deteriora-
tion was partly the result of of monocropped export crops and of
demographic growth that was accompanied by an expansion of land
under cult-ivat.ion and the subdivision of fields, with a corollary
disappearance of fallows.
In order to meet expanded demand, particularly for cereals,
Nigerien agriculture needs to move towards an evolved intensive
system of crops and livestock, and hence one that relies more on
a base provided by agricultural research.
2. AN INTRODUCTION TO I.N.R.A.N.
2.1 History of the creation of I.jN..A.N.
Since the recent years of drought., the Nigerien countryside
has encountered numerous and complex::: problems. Aware that agri-
culture and livestock raising, the pillars of economic develop-
ment. in Niger, cannot progress without the solid support of
agricultural research, the public authorities decided in January
1975 to created the Institut National de Recherches Agronomiques
du Niger. INRAN pulled together and took over the various French
research organizations which had hitherto been responsible for
carrying out agricultural research.
2.2 Researcgh Di2reticton at I.N.R.. A._N.
Niger's national policy of food self-sufficiency determines
the nature of agricultural research at INRAN. Hence, the or-ien-
tat-ion is toward subsistence crops (millet,. sorghum, cowpea and
groundnut), with emphasis being placed on varietal improvements
and cultural practices. In contrast with the former approach,
the strategy is primarily directed toward applied research taking
account of environmental conditions and the real needs of the
rural world. To carry this out INRAN is divided into five tech-
nical departments responsible for research in the following
areas: agronomy, ecology, animal and veterinary sciences, fores-
try and economics.
2.3 Research Infrastruct-ure
In operating its numerous research programs, INRAN utilizes
58 university-educated researchers, including 25 Nigerieris. Most.
research is carried out on the two principal research stations
(Tarna and Kolo), which cover a total of 308 hectares. These
stations are supported by a nation-wide network of sub-stat.i ons
and research support sites. During the 198:3-84 fis-cal year, the
budget totaled 633,330,000 C.F.A. francs (US$ 1,583,325 at an
exchange rate of 400 FCFA per dollar). Of this sum, about 16;
went into the financing of non-salary research activities (INRAN
1984, page A XX).
3. FROM RESEARCr Ti .-"-TEISIIN
3. 1 The Traditi onal reproach
3. 1 1 IrnStati on Research and tthe "E
In the past, agricultural researchers would synthesize re-
search station results in to "recipes" for cropping practices for
the agricultural extension service. These were called "extension
and recommendations themes" (INRAN 1982). The research approach
was thoroughly specialized and particular stick.
The results coming off the research stations and sub-sta-
t.ions were only rarely checked in the peasant farming environ-
ment, and the differences between the two were at times striking.
Recommended practices tended to focus almost exclusively upon
technical or technological aspects of farmers' problems, without
taking into account the structures and systems of production and
paying g insufficient attention to socio-economic repercussions on
the lives of rural people. As a result, ex::tended practices were
rarely adopted by small farmers.
:3. 1.2 .Aricltura1 c::-:tens.ri onr and th-.e Mult-i 2ocatj..onal Tr-i als
Such a research approach did not. facilitate the e::::xtension of
the agricultural recrnmmandatiors. Moreover., collaboration bet--
ween researchers and e::xt.ensionists was limited to contacts at the
annual meetings. Indeed, from 1979 to 1984, INRAN was part of
the Ministry of Higher Education and Research, while the exten-
sion services were located -in the Ministry of Rural Development.
(The two are now joined within the latter Mini story .
Toward the end of the 1970's, an initial effort was begun to
link research and e::t.ension. At the suggestion oif some ex:::t.ension
agents, INRAN undertook a program of rulti locational trials which
are carried out by extension agents under experimental designs
provided by INRAN. They are placed across the country at ext::en-
sion sites. For the first time recognize rng officially the need
to distinguish among different agricultural zones, INRAN divided
the rmultilocat.ional trials between a low-ra infa'll zone (aninua"i
mean under 400 mm) and a moderate rainfall zone (over 4o00 mm).
For the varietal trials, a second dist.inct-ion was make between
dune (sandy) and valley (sandy-clay) soils.
The goal of the rmu ltil ocatl ional t.ria'ls is to evaluate vai-
ties and fertilizer levels before extending them. Results of
these trials have been very uneven during the initial years.
These variations were due not only to differences inherent to the
physical environment., but also to a shortage of adequate super-
vision of the trials. Since 1970, the rural development projects
have contributed to the improvement of the management of these
trials. Since 1984, INRAN has simple field the experimental de-
signs of the varietal trials entrusted to the extension service,
leaving the more complex designs on the interaction of varieties
with fertility levels to the applied research sections of the
rural development projects.
3.2 The 'evelFmrnent of the Farm ing Syst.es Aeeroyach at I..=_NR-eA.._N.
After the creation of INRAN and the reorientation of re-
search toward development needs, it rapidly becomes apparent that
agricultural research should be placed in its true context and
should take account of rural realities. There have been several
attempts to do this, beginning in 1977. From 1977 to 1983, the
approach pursued was to carry out case studies on agricultural
recommendations already being e::::ended to farmers. There were
three different versions of this approach.
3.2.1 Monitori i the "Test FarEmer at Tarna
At first, an artificial farm was established on an existing
research station in order to study the application of the agri-
cultural recommendations package under the best possible condi-
tions and to evaluate their profitability. From 1977 to 1983,
the monitoring of the "test farmer" (E.arsD e.ilote) at Tarna
allowed a better understanding of how the extended agricultural
recommendations performed in an environment with unconstrained
access to capital (chiefly in the form of fertilizers, improved
seed varieties and animal traction equipment) (Roesch 1982, pp.
17-24). Overall, the study showed that the recommendations pac-
kage could provide attractive returns, but the profitability of
individual recommended practices and the returns to capital
investments were never evaluated.
3.2.2 Mornitoring Two !.-.t_ of the Farmer TraKi ni r Cet-.e2rs
A second attempt at evaluating the extended agricultural
recommendations studied the behavior of two farmers trained at
the farmer training centers (Ce ntres de Promontio2 Rural) of t.he
Maradi Rural Development Project (Roesch 1982b). The objectives
of this study were "to test the validity of the techniq ues
taught, the fashion in which these techniques wee adapted in the
field, the problems and deficiencies in the training, and the
social and economic constraints which an e::-trainee was likely to
encounter... For INRAN, interest focused on the study and
improvement of cultural techniques and the identificati on of
peasant needs in agricultural research." (Roesch 1982a, p. 25).
To carry out the study, two training graduates were chosen, one
from a village on dune soils and the other from a vi-llage on
sandy-clay soils. The study took place from 1979 to 1981 and was
expanded in 1981 to cover 28 farms located near the two farms
that were being studied in depth.
Moni tor-ing the farms of the two training- center graduates
from the Maradi Project and the survey of the 28 farms
surrounding g them a allowed feedback not only to the ex:tensionr
service, but also to INRAN (Roesch 1982a, p.43). It indicated
that certain hi therto neglected areas of research should be
considered, notably the feeding and train-ing of draught animals,
making the agr-icultural recommendations more flexible in the face
of climatic and economic uncertainty es, int ercroppi ng marnagi ng
organic matter and addressing the high cost of the equipment
needed to practice, the recommended techniques.
3.2.3 The "E:::Eer mental Ar0I,-cultural Produ,-tin n-i ts" OE2 elra. o02
The third approach adopted for the evaluation of the agr-icu-
Stural recomrnmedat-i ons, which continues to this day, ai ms at
studying the recommendations on sample farms managed by peasants
who have received no previous training. These farms have been
named "Ex:perimental Agricultural Production Units" (Unit.es de
FC'-d:uctJin Ag.1, ,:e E:erimental es) or UPAE. The objectives of
the UPAE study are : 1) the ident.ificati on of real constraints
encountered in the application of the recommendations, and 2) the
search for solution ns that will relax: or remove the observed
constraints through comb-ining all factors of production, includ-
ing soc-io-economic parameters which have long been neglected
(DECOR 1983a, p.14).
This program has evolved through several stages since its
beginning in 1980. A survey of 113 farms in Filingue Arrond-isse-
ment made it possible to choose six: for an in-depth study. Six
farms are split between two Zarma villages, one located on dune
soils and the other in a large dry valley. After one year of
monitor-ing the farms without any researcher -interventi on, INRAN
began in 1982 to introduce certa-in~ recommendations on a part of
each UPIAE. The recormmenrdati ons in clude:
1) rmnorcropped millet and cowpea,
2) improved varieties of millet and cowpea,
4) higher planting density (millet at 10,000 hills per
hectare; cowpea at 55,000 plants per hectare),
5) mineral ferti izers (urea and superhosphate,
6) pest management, and
7) poultry farming with improved species.
In 1984, the sample was expanded to include five UPAE in
each of the two villages as well as fifteen farms (without resea-
rcher intervention at first) in a third village on dune soils,
but made up of ethnic Hausas.
Farmer reactions to the introduction of these recommenda-
tions have provided lessons to researchers (DECOR 1984, pp. 4-5):
1) Use of the line tracer (to guarantee recommended
planting density) is difficult,
2) Mixing urea into the soil around each hill of cereal
crops considerably increases labor time required,
3) A millet variety which tillers better than the
recommended one (HKP) should be developed,
4) A cowpea variety that tastes better than the
recommended variety (TN 88-63) should be developed.
THe UPAE program is allowing the identify cation of both
favorable and constraining factors affecting peasant adopting of
a.gromonic practices developed on the research station and recomn-
mended to the extension service. Moreover, the results of this
program have emphasized the importance of elaborating on-station
research programs based a prior on farmers' objectives and the
actual constraints imposed by their environment.
3.2.4 The Conce pti on and Birth of the Current Aer2oach
The results of the three programs aimed at evaluating the
application of the agricultural recommendations confirmed a pre-
vious belief by agencies involved in extension and rural develop-
ment that on-station research and agricultural extension both
need a reorientation. The monitoring of the two farmer training
center graduates from the Maradi Project. in particular focused
on the lack of recommenedati ons for intercropping and the homo::ge-
neity of recommendations that does not take into account the
variety of farmers' ecological and economic conditions. Recom-
mendations requiring greater investments than farmers can afford
are particularly unlikely to succeed. The UPAE operations further
establ i shed the importance of developing recommendations which
meet the specific needs of farmers in different agro-c i matic
zones of the country.
These discoveries have led INRAN to reflect upon how to assure
that the products of agronomic research will better meet the
needs of its clientele. It is clearly inadequate simply to
evaluate agricultural recommendation ons after they have already
gone into extension or to conduct mult.ilocational trials managed
by extension agents to assure the appropr i ateness of these
recommendations. If we accept that the peasant farmer is the
final user of our research products, then a profound reori enta-
tion of our scientific approach is necessary. It becomes impera-
tive to understand his objectives, his needs, his resources and
the constraints he faces before elaborating research programs.
From the outset, he must be invited to join INRAN researchers in
formulating research objectives.
4. THE "FARMING SYSTEMS RESEARCH" WORKING GROUP
In 1984, INRAN established several technical groups called
"working groups", among which was the working group on "Farming
Systems Research". The groups objectives are as follows:
1) To establish with INRAN a mul tidisciplinary team composed of
researchers from different fields (biologists and economists)
and extension agents (developers).
2) To formulate research programs on farming systems and, after
their execution by the multidiscipl inary team, to evaluate
the results and generate feed-back for future programs,
3) To inform INRAN administrators of the progress achieved in
these programs and to insure administrative support for them.
This working group, which is coordinated by the head of the
Rural Economics Department assisted by two biological and two
economic researchers, counts among its members production agrono-
mists, plant breeders, entomologists, plant pathologists, veteri-
narians, soil scientists, agricultural chemists, agricultural
economists and extension agents.
So far, the working group has accomplished the following
1) Formation of a multidisciplinary team,
2) Organization of a rapid reconnaissance survey (sondeo),
3) Elaboration of an on-farm trial program.
5. CURRENT RESEARCH ON FARMING SYSTEMS AT I.N.R.A.N.
At present, INRAN has adopted a multidisciplinary approach
oriented to satisfying the specific needs of farmers from diffe-
rent agr-iclimatic zones of Niger. The development of this ap--
proach englobes the following steps:
1) the establishment of multidisc-iplinary teams,
2) a rapid reconnaissance survey (sondeo),
3) on-stat.ion agrronomic research,
4) an in-depth soc-io-econiomic survey,
5) on-farm agronomic trials.
5 As one of its animal field visits for oSuther INRAN resear-e
As one of its annual field visits for other INRAN resear-
chers, the Rural Economics Department organized a trip for the
multidisciplinary team to the Experimental Agricultural Produc-
tion Units at Tashi, Lagare and Kouka in F-il ingue Arronndissemernt.
The team, made up of three soil scientists, a production agr-ono-
mist, three economists, a plant breeder and a statistician,
evaluated the work underway in order to see what contribution the
other INRAN research departments could make to these programs.
In this particular case, the principal difference vis-a-vis the
approach followed by Hi debrand (1979) in Guatemala was that the
rapid reconnaissance survey verified a program already underway
rather than an area where research had not yet begun.
Upon arriving at the villages, the team found that after
three years of on-farm demonstrations, farmers on the UPAE still
limited the application of the agricultural recommendations to
the INRAN test fields. When asked "Why don't you apply the
recommended techniques to your other fields?" they gave two
principal reasons. First, they do not like mono-cropping, belie-
ving that it does not make rational use of space and that it
increases the danger of a poor harvest when rains are poor.
Second, agricultural inputs (chiefly m-inreral fertilizers) were
unavailable to farmers, as much for financial reasons as because
they were logistically difficult to obtain (the farmer had to:.
travel 17 km by donkey). They found the recommended application
technique for urea fertilizer (localized around the hi 1I of
millet) to be too time-consuming:, since it coincides with the
weeding period. As farmers saw it, "if ferti lizer- is not
available, it is unwise to plant at the high density recommended
by INRAN (10, 000 hills of millet per hectare).
Important lessons for on-station research came out of the
rultidi sciplinary reconnaissance survey. Farmers' unwi 1llingness
to practice mono-cropping once again convinced researchers of the
importance of developing intercropping recommendations. More-
over, the unacceptabi lity of the recornrnendations package as a
whole suggested a need either to simplify the recommendations or
introduce them in stages while adapting them to local conditions
in the target zones. These lessons have brought three changes to
INRAN: the renewal of research on intercropping, on-farm verifi-
cati on of techniques developed on the research stati on before
extension to farmers, and the acceptance, not only by researchers
but also by administrators, of the importance of multi-
disciplinary approach in on-farm research programs. The partici-
pation of administrators (the heads of three research depar-
tments) in the team's activities facilitated the rapid adoption
of these new directions.
5.2 On--S0t0ati Aon .s.CDSoiE Researtch
In order to resolve some of the problems mentioned above,
the General Agronomy Sect.ion undertook a cropping systems re-
search program at Tarna and on the different INRAN research sub-
stations early in 1984. The program is in two parts: 1) research
on cultural practices that can be adopted immediately by the
farmers, and 2) long-term research aimed at reaching a better
understanding of intercropping systems and their interaction with
ecological factors (e ._ rainfall) and factors influencing farmer
access to production resources (such as animal traction equipment
and fertilizers). Under the first part of the research program,
trials on varieties, planting density and geometry, and land
preparation techniques were initiated in 1984 at Tarna and the
sub-stations and continued in 1985. The same is true for the
second part of the research program, which encompasses experi-
ments on plant physiology, water utilization, -insolation, plant
nutrients, insect incidence, plant diseases, weeds and crop rota-
Aside from the agronomic experimentation, results from
socio-economic surveys carried out by the Rural Economics Depar-
tment as well as contacts with other researchers and extension-
i sts are utilized to -identify the objective es, resources and
constraints faced by farmers who are the targets of on-station
The on-stat'ion research aimed at finding technical solutions
to production problems is conducted by the General Agronomy
Section with the collaborate on of researchers from other depart-
ments. The on-farm trials were initiated on a small scale by the
Rural Economics Department in 1984 in order to learn to manage
the logistics of such trials. They have been continued in 1985 in
conjunction with the Genreral Agronomy Section. These trials are
based on research results for on-station ex:peri mentation in 1984
at Tarna and on the sub-stations. Eventually, these on- farm
trials will lead to definitive recommendations that will be
transmitted to rural people on a large scale by the agricultural
5.3 On-Farm Research
On-farm research is carried out -in order to study farmers'
needs and to evaluate techniques and new varieties that have
per-formed well in a controlled environment. The focus is on
surveying farmers' needs and orienting agronomic research in
order best to meet those needs.
The identification of farmers' needs is done in two stages.
The first, already described, is the multidiscipl inary rapid
reconnaissance survey. This permits a general assessment of
needs as felt by farmers and observed by researchers inr the
course of a single visit. In order to acquire a more complete
understanding of farmers' needs and to be able to prior tize
these different needs, it. is necessary to monitor closely the
management of their farms.
5.3.1 Socio-Eonormic Research
In order to monitor farm management, a socio-economic sur-
vey is carried out with the following objectives:
- In the short run:
1. To establish a typology of farms in order to group
farms together by recommendations domain.
2. To identify farmers' objectives.
3. To identify the constraints and the factors that favor
the realisation of these objectives.
In the medium run:
1. To help establish priorities for agronomic research
given farmer constraints and the likelihood that
research can find solutions for them.
2. To help guide national agricultural policy in order
to relax economic constraints (for e::ample, by
improving access to modern inputs and by bringing
product prices in to equilibrium).
3. To evaluate the results of agronomic and economic
126.96.36.199 Methodol ogy
A research program in a given zone of the country begins
with the collection of baseline data on the region. This data
collection can be done either through library research or, fail-
ing that, by a survey of the entire zone. For two programs, a
lack of information forced INRAN to undertake baseline surveys.
In both cases, a large sample of farmers was drawn at random for
an ernt.ire arrondi ssement (113 farms -in Fi 1 ing ue inr 19 80, 348
farms in Madarounfa in 1982). The analysis of data collected
during the baseline survey furnished the necessary in formation to
establish recommendations domains. In the case of the study Con
the interaction between rainfed and irrigated agriculture along
the River Niger, the baseline information necessary was already
available from the National Office of Agricultural Water Manage-
The second stage is an in-depth survey of a subsample of
villages intentionally y chosen to represent each recommendat-ions
domain. At this stage, peasant objectives are studied, as well
as farm resources and agricultural activities. The goal of these
studies is to be able to:
1. Quantify the agricultural resources of typical farms.
2. Establish partial budgets for different cropping
systems and technology levels.
3. Integrate these into a model of a typical farm
The initial results of the socio-economic surveys have made
it possible to establish recomrnendations domains, and thus to
choose sample villages for in-depth studies. Each study zone
represents a distinct region, particularly as regards rainfall,
the principal constraint to agricultural production in Nigcer.
Hence, as shown in Table 1, the north-west and center-south
agricultural zones were chosen for their significant differences
in rainfall, and the irrigation perimeters along the River Niger
for their reliable supply of water on irrigated fields. Within
each zone, the baseline information made it possible to choose
vi llages having different characteristics of topography, soil
texture and predominant ethnic group (this last being linked to
important differences in agricultural techniques.)
The second analytical stage is to describe quantitatively
the farms' agricultural resources and the u.tilisation of these
resources in different farming systems. It is not surprising to
see in Table 2 that there ex-ist quite substantial differences in
farm resources according to the recommendations domain. For
example in the north-west of the rainfed agricultural zone, farms
in the valley are smaller than those on the plateau. Resource
utilization also varies by zone, and so, consequently, do cropp-
ing systems. Sorghum, for example, is present in all major crop-
ping systems in the center-south of the agricultural zone of
Niger (Madarounfa) while it is rarely encountered in the north-
west (Fi'lingue, except for the valleys).
Analyzing the data on labor time and the management of
different cropping systems will allow the establishment of of
I/ For an example of the methodology. y used to determine a
recommendations domain, see Swinton and Ly 1984.
partial budgets by level of technology (animal traction and
manual farming) and the cropping system. The utili station of farm
resource inforrat.ionr and the partial budget results wit.hini an
input-output mathematical model wi ll permit the identi fi cat.ion of
the principal constraints to agricultural production for each
type of farm and the classification according to the value of
marginal product which could be obtained as each constraint is
The creation of input-output mathernmt.i cal models offers
several important possible -ities to agronomic researchers, as well
as economists. Besides quanti flying constraints, these models
make i t. possible to simulate the impact of a new technology
already tested on the research station when such a technology is
introduced into an existing farming system on a typical farm.
The possibility of evaluating the farm-level profitability
of an investment in the development of a mew technology allows a
more efficient management of agricultural research. Moreover,
the choice of a profitable technology increases the probability
that it will be accepted by the farmer in the course of on-farm
Obviously, a mathematical model is not something fixed and
immutable. It lends itself to constant mod-ifi cations and correc-
tions so that. it better approaches reality. For example, al-
though at present INRAN's farming systems research program pays
little attention to livestock production, this does not prevent
it from being -incorporated into farm models int the future.
The feedback between socio-economic and agronomic research
does not stop there. Agronornic research offers the possibili ty
to ver-ify the i nterpretati on of economic research results. For
example, in 1984, intercropping trials were installed in the
there representative villages of Madarounfa Arrondissernent to
verify the hypothesis that agronomic conditions (chiefly soi
texture) vary sig ni ficantly from one recommendations domain to
another. Although the drought prevented the crops from reaching
maturity, the results of soil tests showed that soils in two of
the villages were not different, which led us to assume that the
differences in cropping systems between the two domains were due
to the use of different agricultural techniques.
Eventually, the feedback between research and the'reality of
the peasant world (via agricultural extension as well as on-farm
trials) should play a primary role. Only lessons learned in the
field can give a true evaluation of research results and the
efficacity of communication between e::tension agents and farmers.
However, we are still in the process of perfecting recommended
technologies according to the needs of farmers in different zones
of the country. At the current stage, the role of extension is
limited to fixing research directions and evaluating the on-farm
trials. Once certain recommendations have been transmitted :to
the extension service, its role in helping to evaluate the recom-
mendationrs on a large scale will expand enormously.
5.3.3 Onr-Far TrMials
The objectives of the on-farm trials are as follows:
1) To determine whether technologies proposed by
on-station research are ready for extensionn.
2) To determine whether the proposed technologies
should be re-evaluated by on-station researchers
in order to take into account biologic and economic:
constraints encountered at the farm level.
3) To determine to what extent the extension and input
supply services should prepare for the diffusion of
these new technologies.
5.33.32 TyEes of OnzFarm Trials
Two types of on-farm trials are distinguished at INRAN: 1)
researcher-managed trials and 2) farmer-managed trials.
The first type is used above all for developing new techno-
logies which take into account actual farming conditi ons. The
trials are very simi'liar to those carried out on-stat.i on, since
the researcher controls all the e::perimental variables. However,
these trials differ from those done on-stat-ion in that they are
conducted on farmers' fields, under ecological conditions more
varied than those on the research station, and taking into ac-
count the reactions of the farmer.
The second type of trials, those under farmer management, is
used to study how and why the farmer adopts a new technology,
wholly or partially, or rejects it. In this method, the farmer
himself tries to adopt and adapt the new technology proposed
within the context of his biological environment and his soc-io-
economic c circumstances.
188.8.131.52. Trials Adapted to Local Con dit.ons
Farmer-managed trials frequently have coefficients of varia-
tion from 30 to 50 percent, thus considerably higher than those
generally registered in on-station trials (Shaner et a1. 1982,
p. 119) .
In order to reduce or eliminate these variations, on-farm
trials at INRAN are conducted on relatively large parcels (1000
m ), with sufficient repetitions (25 per zone) and are designed
according to local conditions. E::xperimental blocks must be cho-
sen according t.o predominant agro-climratic conditions in the
region and treatments must be adapted t.o the region. For exam-
ple, since rainfall is the determining factor in Niger, a single
variety of millet cannot, be recommended for several regions
having different levels of rainfall. Hence, an early variety
(HKP) was used for trials in low rainfall regions (mean of 300-
450 mm/year) and another variety (CIVT) for higher rainfall
regions (mean of 450-600 mm/year).
Given the high variability of rainfall and soils in Niger,
INRAN's 1985 on-farm trials were divided among three distinct
agro- climatic zones (Filingue, Madarounfa and Kolo). These
zones were chosen based on the recommendations domairns previously
defined. In addition, ICRISAT installed the INRAN on-farm trial
expert mental design in one of its village research sites.
184.108.40.206 AnlZysis of OnzEarm rj..als
INRAN's analytical methods are essentially centered around
the desire to integrate the different disciplinary understandings
of the rural environment brought by the team members.
First. of all, agronomists determine the biologic and techni-
cal feasibility of the recommendations proposed by finding out if
there exist statistically significant differences between yields
obtained using the proposed recommendations and those gotten
using traditional practices. To do this, analysis of variance
(ANOVA) is used. However, in contrast with on-stat-i on experimen-
tal analysis, the ANOVA can be done at higher levels of accep-
table error (for example, 10% instead of 1 -or 5%). In effect,
the farmer may be willing to accept a higher risk of error than
the researcher normally would (Perrin et al. 1976).
Secondly, the economists establish the economic profitabi l
ity of the proposed technology. Two methods can be used to do
this: analysis of the entire farm, or the construction of partial
budgets. The former method is appropriate for cases where the
proposed new technology requires a profound transformation of
traditional farming practices, either by demanding considerable
capital investments or by strongly tipping the balance among
existing production systems (the case of "heavyweight recommenda-
tions"). Whole-farm analysis requires a broad base of detailed
in formation and sophisticated analytical tools (linear program-
ming, for example). However, since INRAN's on-farm trials so far
involve only simple techniques, this form of analysis is not yet
The second method, the construction of partial budgets, is
appropriate when the proposed recommendations represent a margi-
nal change in traditional methods (the case of "1 ightwei ght
recommendations"). Such a change necessitates neither fundamen-
tal reorganization of farm management, nor great capital invest-
ment. Partial budgets make it possible to quantify the contribu-
tion of a recommendation to the farmer's net income. When seve-
ral alternative recommendations are proposed, it is best to
-identify the one which will contribute the most to met revenue
for its variable cost. Dominance analysis is used to do this.
Such an analysis has been used at INRAN (Kennedy et al.,1984).
At its birth in 1975, INRAN's role consisted of developing
agricultural recommendations without thoroughly understanding the
farming systems of Nigerien peasants. At. that t.i me, agronomic
research, confined exclusively to the research station, was foc-
used on ronocropping and transmitted recommendations to the ext-
ension service in the form of recipes without. having tested them
under farmers' conditions. Having realized that. certain recom-
rendations were not accepted by Nigerien farmers, INRAN set up
two new programs to evaluate the recommended techniques: 1) the
multi locational trials and 2) the monitoring and evaluation stu-
dies of farmers using the recommended techniques (the monitoring
of the "Test. Farmer"' at Tarna, that of the two farmer training
center graduates from the Maradi Project, and the E:xperi mental
Agricultural Production Units of Filin argue. The multi locat.i onal
trials chiefly aim to insure that the cultural practices and new
varieties developed on the two principal research stations behave
well agronomnically elsewhere. The socio-economic studies of the
use of the extended recommendations aimed to identify the factors
which facilitated or impeded their adoption and to suggest, means
of rel ax::i rg the constraints.
But the fact. that certain recomrmendati ons hold inherent. con-
straints to their adopt.i on (excessive labor time, high cost)
brought INRAN to opt for a new approach. From the outset, it.
seeks to discover the needs of Nigerien farmers and to develop
agricultural recommendations applicable to zones with suffici ern-
tly homogenous characteristics. Currently, INRAN's Farming
Systems Research Program is a multidisciplinary program with
scientists in both the social and biologic realms. Once identi-
fied for research, an intervention zone is rapidly surveyed by a
multi -discipi inary team to determine its agronomic_ possibilities
and the needs of its farmers. Sim ultaneously, a socio-economic:
survey is begun in order thoroughly to understand farming systems
in the zone and to quantify their parameters. The findings of
this survey are exarnined jointly with the agronomists. Based on
the reconnaissance survey and the socio-economic study, the odds
for the adoption of a new technology are evaluated within the
context of the traditional farming systems. All promi sing." tech-
nolo:gies are then tested in on-farm trials. At first, these
trials are carried out under researcher management. Those which
do well are then repeated under farmer management.
Several years will be necessary before this approach can
yield e::tendable recommendations. However, it has al ready
brought about certain promising new research directions. The
focus on intecropping research will doubtless lead to the develo-
pment of recommendations more acceptable to farmers. Economic
research will prevent unprofitable recommendations from being
extended. The participation of Farmers and extension agents in
agricultural research will permit other factors liable to be
overlooked by researchers to be taken into account. These new
directions insure that the products of agricultural research will
better respond to the needs of Niger's farmers.
D.E.C.O.R. 1984. Recherche sur les systmres de production agricole.
Campagne 1983: Deulxi me annee d'intervent.ion dans les unites de
Production Agricole Ex:pri mentales (U.P.A.E.) des villages de
Lagard et de Tashi (Arrondissement de Fil-ingu.) .. INRAN, Niamey.
1983a. Etude sur les syst&mes de production "Unit.es de
product on agricole experimentales" premiere annae d' intervent.iorl
dans 1'Arrondissement de Filingue. INRAN, Niamey.
1983b. Enqute socio-aconomri que par sondage et choice:
d'Unites de Producti on Agricoles Ex:perimentales dans la region de
Filingu6 (Rapubli que du Niger). INRAN, Niarey.
HI-ildebrand, P.E. 1979. Generatit'ng technology for traditional farmers
the Guatemalan experience. PFresented at the 9th Int. Cong. of
Plant Prot. 5-11 Aug. 1979. Washington, D.C. Soc:i oeconomrnia Rural,
Sector Publico Agricola, ICTA, Guaterala.
I.N.R.A.N. 1984. Note information sur 1'INRAN. Niamey.
1982. Fiche signal]tique des variotrs vulgarisees. Niamey.
Kennedy L., A.S. Ly et R.R. Deuson. 1985. Rapport de Recherche: Etude.
des cas portant sur des essais en mi lieu r.el effects dans
"I'Arrondissement de Kolo en 1984. DECOR, INRAN, Niamey (*).
Ly A.S. et R.R. D:euson. 1985. Pro:fi de tro-is amrnagements du Fleuve
Niger et inventaire pr"l iminaire des resources de 58 e:::ploitations-
achantillons. DECOR, INRAN, Niarey.
Numa G. 1985. Les Uni ts de Production Agricaoles E:xparimentales
(U.P.A.E.) dans Arrondissement de Filingu,: RFsuiltats de a
Campagne 1984. DECOR, INRAN, Niamey.
Perrin R.K., D.L. Winkelmann, E.RF. Moscardi, and ...R. Anderson. 1976.
From agronomric data to farmer re:cornmmendations: an economics training
manual. Inf. Bull. 27. CIMMYT, El Bat.an, Mexico.
Reddy K.C. and J. Gonda. 1985. Recherche en cultures associaes au
Niger. Liaiiso Sahel no. 3. Institute du Sahel. Bamako.
pp. 109-125 (*).
Roesch M. 1982a. Rapport Provisoire Agronornmie Gneral e 1981. INRAN,
1982b. Su ivi de deux anciens stagiaires des Centres de
Promotion Rural e du Projet de Developpement Rural de Maradi et des
28 exploit stations de leur environnement dans les villages de
Atchitakofoto et. :Djiratawa. INRAN, Maradi.
Shaner W.W., P.F. Phili pp and W.R. Schrehl. 1982. Farming iSystems
Research and :Development. Guidelines for DevelopingC Coruntries. The
Consorti urn for Internatio nal Developmenet. Westview Press, Inc.
Bou older :, Colo radio, U.S.A.
Swinton S. 1985. Les resources agricoles des exploitation dans trois
vi ages repr'sentatifs de Arrond-issement de Madarounfa. D:ECOR:,
et. Ly S.A. 1984. Definition des dornaines de recommandati ons
agricoles dans le Centre-Sud dui Niger: Arrondissement de Madarounfa.
DECOR, INRAN, Niamey. Document pr-sentW lors du Symposium de
Recherches sur syst.mes de Production Agricole tenu A Kansas State
University, A Manhat.t.an, Kansas, du 7 au 10 octobre 1984 (*).
SNuma G. et S.A. Ly. 1985. Les cultures associes en milieu
paysan dans deux r ,gions du Niger: Filingur et Madarounfa. Liaison
Sahel no. 3. Institute du Sahel. Bamako. pp. 183-194 (*).
(*) Available in English from the authors.
Table 1 : Characteristics of the recommendations domains
Crop Zone rainfall Topography Soil Dominant. Number of
1971-84 1/ Te>::t..ure Et.hn ic GroupF: Farms
Ko 1 0
1 In order to obtain an adequate time series, the figures given come
from the following sites: Fi lingue from Fil-ingue, Madarounfa from
Maradi :, Tillabery from Tillabery and Kolo from Niarney airport.
Table 2 : Salient characteristics of farms in 1984 by recommendation domains
Recommendati on Livest.ock
domains by crop Mean farm Mean farm -------------- Cropping
zone size population Bovine Ovine/ system
.a E-j 'l e.L'.S r h d
Valley Zarma 6.7 8.6 MC.MSCNM2
Plateau Zarma 15.8 10.8 M
PF"lateau Haoussa 12.3 8.6 M.MC
Valley 5.5 8.0 10.2 10.0
Plateau P:eulh 4.1 5.5 3.2 5.6 MSMSC
Valley Haoussa 4.0 5.4 1. 0 5.1 M, MSC, MS, MC
Plateau Haoussa 5 9.0 1.8 14.6 MSCMC, MS MC
Ri ver -north
Zarma/Sonrai 3.9 4.4 5.2 1.4 MCR
Soinkee 1.5 3.7 1.9 0.8 M, R
Zarma 4.3 4.7 2.1 2.9 MC,R
(1) Covering over 10% of the surveyed land aea.
(2) By order of imrnportance decreasing from left. to right :
M = Millet S = Sorghum C = Cowpea R = Rice
S'yrcgs: Numa 1985, pp. 20, 35; DECOR 1983b, p. 1383 Swintonr 1985,
pp. 6, 21, 24; Ly et. Eeuson 1985, pp. 22, 36, 40.