f 
1_--~~I.3IJ..U.ING EXPERIMENTING FARMERS FOR 
* 
** 
PARTICIPATION IN AGRICUl TURAl 
RESEARCH AND TECHNOlOGY 
DEVElOPMENT 
025589 
2 3 ~t I lSS 
(\v 
Jacquehne A{I Ashby 
Teresa Gracia 
i\ Y 
Mana del Pilar Guerrero 
Carlos Arturo QUlros 'ro--r~ 
Jose IgnacIo Roa 
Jorge Alonso Beltra n G 
Deputy Director General, IFPRI 
HllIsldes AgroEcosystem Program, IPRA ProJect, CIAT 
1 
INTRODUCTION 
Farmers who expenment wlth new ways of farmmg are an Important resource 
helpmg rural communltles to solve thelr farmmg problems Yet these experlmentmg 
farmers are generally u nrecogn Ized, unsupported, and dlsconnected from the often 
substantlal mvestment m formal agrlcultural research Experlmentlng farmers are a 
neglected resource because conventlonal approaches to agncultural technology 
generatlon are top-down Technology IS deslgned by sClentlsts who make declslons 
about what to recommend to farmers wlthout glvlng farmers any dlrect say In thls 
process The conventlonal approach IS hke a doctor-pallent relallonshlp The • 
researeher and extenslomst (hke the doctor) are supposed lo formulate a prescnptlon 
to cure the farmer-patlent's IlIs But when the doctor or sClentlst cannot dlagnose 
enough problems eorrectly nor formulate appropnate prescnptlons because the needs 
are so many and dlverse then thls approach breaks down Developmg teehnology 
whlch IS sUlted to the particular, locatlon-speclfic needs and problems of the 1 5 bllllon 
people who depend on complex dlverse, nsk-prone agrlculture reqUlres a dlfferent 
approach (Chambers, 1994) 
One solutlon could be the establishment of a commumty-based capaclty tor 
carrymg out adaptlve research wlth the partlclpatlon of farmers m Identlfymg problems 
and In Implementlng technology testmg There are an mcreaslng number of 
expenences mvolvlng orgamzmg groups of farmers, or workmg wlth eXlstlng farmer 
orgamsatlons to Implement farmer partlclpatlon (see for example Maltee and Lasalle, 
1994, Muchagata et al 1994, Mushlta, 1993, Dnnkwater, 1994, Hemrlch et al , 1991) 
The strategy of orgamzmg groups 
offarmers to partlclpate m adapltve technology tesllng IS In part a response to concerns 
about how to reduce the costs of Involvlng farmers In research when thls makes heavy 
demands on the time of salaned professlonals (researchers or extenslomsts) 1I al50 
addresses the need to 'sea le up' farmer partlclpatlon In research and exlenslon so tha! 
technology testlng can be carned out m numerous, dlverse mlcro-envlronments wlthout 
Incurnng excesslve expenses and compromlsmg the quahty of partlclpabon (Okah el al , 
1993 AShby, 1991, Bebblngton el al, 1994) 
JAA 01 I rmrClAL. WIJ 
2 
Many q uestlons have been ralsed about the vlabllity of Insbtutlonalizlng an 
adaptlve research role for farmers and the constralnts such efforts are likely to face 
(Bebblngton et al 1994) Cntlcs argue that farmers' tradltlonal or folk expenmentabon 
IS a form of knowledge generatlon supenor to western sClence The strength of folk 
expenmentatlon IS maklng contlngent, sequentlal adJustments over time to changlng 
clrcumstances whlch are unpredlctable (Rlchards, 1989, Scoones and Thompson 1994, 
Dnnkwater 1994) Thls Indlgenous form of knowledge generabon does not readily fit 
wlthln models of controlled expenmentatlon used by western sClence 
A more useful analysls draws on understandlng the nature of folk 
expenmentabon For example, farmers compare "treatments", but the check or control 
may be "In the farmers head" because farmers compare thls year's performance wlth 
last year's Another companson used by farmers contrasts results In a dlstant fleld wlth 
a nearby one or the results of addlng a little bit more fertllizer to one furrow compared 
to the remalnder of the field Folk expenmentatlon Involves replicatlon, but thls IS 
typlcally malnly replicabon over time, In contrast to replicatlon In space and bme 
(charactenstlc of the sClentlfic method) Moreover, farmers recognlze confoundlng 
effects In folk expenmentatlon For example, a small amount of seed of a new vanety 
IS typlcally nurtured and multlplied up In the more fertlle home garden, then the next 
plantlng moves the new vanety to testlng In dlfferent types of sOII, testlng the genotype 
x envlronment Interactlon Only once performance IS assessed In a vanety of 
envlronments, are concluslons drawn about the likely performance of new germplasm 
In the farmers' envlronment 
Expenence shows that farmers' knowledge generatlon can draw on both the 
sClentlflc method of controlled companson and folk expenmentabon It IS not an elther-
or dlchotomy (see for example Uphoff 1992 282-3, Hardon and de Boef, 199367, Berg, 
1993 Llghtfoot 1987) 
Another Issue IS whether the creatlon of a speclal group bUllds on eXlstlng 
authonty structures or creates a parallel non-tradlbonal structure and the extent to 
whlch such groups can represent the research agendas of dlfferent Interest groups 
wlthln the communlty or may Indeed exclude particular groups (Bebblngton et al 1994) 
JAA 01 rrmrC1AL WB 
3 
Expenence wlth on-farm research shows that when formal cntena for selectlng farmers 
to partlclpate m research were not used, the resultant partlclpants were usually more 
wealthy and politlcally active farmers (Mernll Sands et al , 1991 303) 
Expenmenters, or Innovators who can afford to expenment, are likely to be the 
relatlvely better-off farmers, who have the Skllls, and resources (mcludlng power) to 
devote to a particular kmd of knowledge-generalion There IS some expenence wlth 
workmg wlth research-mmded farmers, that suggests It IS deslrable to purposlvely 
select mnovators who have the time and Interest for expenmentatlon (see for example 
Abedln & Chowdry 1989, clted m Mernll-Sands et al, 1991, Ashby et al, 1987) 
The reallssue IS whether expenmentlng farmers who represent the local capaclty 
for research m rural commUnltles can be harnessed to a research agenda, defmed at 
the communlty level, whlch IS also useful to the very poor or to other mterest groups, 
such as women who may have dlfferent pnonlies from the relalively better-off who 
carry out the local expenmentatlon 
Could thls local capaclty If linked effeclively to research agencies, share the 
costs and expand the coverage of adaptlve research, whlle ensunng that thls IS relevant 
to local farmers? How IS communlty-based partlclpatory research to achleve broad 
coverage whlch IS cost-effectlve? Can a self-sustalnlng capaclty and responslbllity for 
promotmg farmer partlclpatlon be created m rural communltles? How can linkages 
among these dlfferent actors be managed wlthout Increasmg the transactlon costs to 
an unwleldy extent? Llttle systematlc work has been done on the costs of creatmg 
organlsalions at the communlty-Ievel to fill thls functlon, nor has there been much 
emplncal assessment of the extent to whlch such organlsatlons can mcrease coverage 
and Improve targetlng of adaptlve research m a way whlch IS self-sustalnlng (Axmn, 
1994) 
Thls paper reports on an effort to provlde emplncal data on some of these Issues 
from actlon-research carrled out In 1990-1994 by the IPRA proJect of the Internalional 
Center for Tropical Agnculture (CIAT) wlth support from the W K Kellogg Foundatlon 
The proJect alms to assess the potentlal for Instltutlonalizlng a communlty-based 
capaclty for mvolvmg farmers In carrymg out adaptlve research Thls paper reports 
JM 01 FrmrCJAL WB 
4 
results obtalned on the devolutlon of adaptlve research responslblhtles to commlttees 
of expenmentlng farmers, the effects of scahng-up thls approach to achleve broad 
coverage, and ItS cosl 
The paper IS organlzed as follows The followlng sectlon describes the 
procedures used for formlng farmer commlttees and thelr actlVltles Then results are 
presented, In relallon lo Ihe evolutlon of the farmers commlttees over the four year 
penod 1990-1994 The paper ends by pOlntmg out Issues that these ralse for future 
appllcallon of lhls approach 
METHODOLOGY 
The project's strategy IS to nnplemcnt partlclpatory research methods for adaptlve 
technology te&tmg, by fonnmg commlttecs of farmcrs based m !Ural commumUcs to carry out 
technology testmg togcthcr wlth pubhc sector agncultural research and cxtenslOn agenclcs, and 
Illtcrmcdlatc orgamsatlons (NOO' s and farmer coo[leratlves) Developmcnt of trallllllg coursc, 
and matenals used for trdmmg farmers and 8taff 01 pubhc sector ,md mtermcdldte orgamsatlons 
for ti liS purpose IS mtegral to the strategy 
The purpose of the farmers' lesearch commlttecs (Conlltes de Invcstlgaclon AgropecuarIa 
Local or CIAL) IS to mobl]¡ze local lcadershlp among farmers to lake responSlblhty for 
cxpellmentmg wIIh technologles not known m thelr commulJIty In thls way, the proJect allns 
to create "demand-pull" by chenls of pubhc sector and mtennedldte orgamsatlOns, on agrIcultural 
research and extenSlOn, dlverslfymg Ihe type of technologles avaIlable, and mcreasmg the 
number dnd rate of flow of technologlcs to resource-poor farmers, so Improvmg ddoptlon, fanl1 
mcomes and welfare Expenence shows thal new technology selected wlth farmer parllclpallon 
methods IS better adapted locally than that recommended by resean"heIs workmg on thelr own 
(Sperhng, 1993, Worede and Mekblb, 1993) 
• 
I he proje\.t WdS mltlated m a pllOI ,Irea m Cauea Department, m southern Colombia -
Cauea IS one of Ihe poorest, lowest-wage departments m lhe eountry The pIlot arca IS 
charactenzed by hIlly terram, poor mfrastructure of roads and markets, and small farms 
averagmg S ha 111 slze (average cultlVated arca IS less than 3 ha) All farms engage m a mlx of 
Wlmnll CI,II ,lI1d ~Ub,IM("Illl plOdultlOn Thl& 1, ,1 m,¡rgllMI \.OflL\. prodUUIOIl ,u l.J, Wllh mili tll!.. 
JAA 01 FrmrClAL Wl1 
5 
aCld sods, otten badly eroded Most farmers cultlvale cotfee, together wlth cassava as a cash 
Lrop, sorne maIze and clllubmg beans are grown tradltlonally for subslslence Llvestock are 
scalce (only 13% of farms have any cattle) 
The proJect began the formatlon of CrALs m flve commumtles (veredas), In 1990, tIJe 
l1umber mcreased to 18 commumtle, m late 1991, to 32 m 1992-3 and tIJen to 55 commumt1es 
by 1994 A further 30 CIALs WhlCh welC formed m Bolivia, Ecuador, Peru and Hondura_ by 
mternatlonal tramees m the method, broughl Ihe total lo 85 so fal ThlS papel reporls 
mfOrmdtlOn oblamed trom momtonng Ihe 48 CrAL, formed betwecn 1990 lo mld-1994 In the 
pllot alea In ColombIa, these CrALs cover an area of approxull,ltely 1605 Km2, mvolvlllg an 
estlmated 50.000 fam¡]les, and dlrect contact Wlth over 4,000 fafmers, of whom 220 partlclpated 
m tramlllg as members of Ihe CIAL or research commlltees 
Each CIAL IS formed wlth four tarmers elected at a commumty meetmg, wluch meets 
regul,lrly dunng the hrst Irammg cycle (or experImental perIod, usually eqUlvalent (O a croppmg 
season of about SIX monlhs) The fml trammg cycle mvolves up to ten tral11l11g VISltS by a 
support~farmer who has bad al least one year of pnor expenence m CIAL Over the next cycle 
or croppmg season tbese VISltS are proglesslvely reduced 111 number, dS Ihe CIAL gdms 
expenence and cames out expenments Wlth mcreasl11g autonomy (Box 1) 
The supporl-fdrmer 15 backed up by an agronomlst who provldes mpuI to statlstlcal deslgn 
of CIAL expenments and the analysls of data taken by the CIAL members At present 111 the 
proJect alea, Ihe 48 CIALs are attended by three support-fdrmers, backed up pllm.lpally by one 
tramer-aglollomlst 
RESULTS 
• 
Tllls sectlon ol' lhe PdPer revlews lhe results oblJl11ed durmg 1990-1994 from lhe __ 
orgamzatlon of 48 CIALs or farmers' research comm!ltees m ¡he ptlot arca m Cauca, ColombIa 
lhe plO~edulL 101 tOlllllng CIAL; \VdS developed 111 d pllot phdM .. 1101ll 1990~1991 m wluLh hve 
farmers' research commlttees were establtshed and tramed m teehmques fOl parllclpatory 
d¡,¡gnosls, planmng and estabhshmg rep]¡cated on-fdrm tndls, partlclpatory evaludtlon oí 
lechnology, analysls and mterpretatlOn of results, budget analysls oí the tolal cost of Ihe tnal and 
of lhe mdlVldual treatmenls Plannmg and presentmg a short oral leport on Ihe ICSUltS lo thc 
JAA o/ TrlllrCI!Il. WB 
Box 1 Procedure for format~on of a new CIA~ 
1 On-farm research and extens~on staff of the host ~nst~tutlon 
receLve tra~n~ng ~n the CIAL methodology and select cornmun~tles 
or respond to requests from communltles to form a erAL The host 
lnst~tut~on may be the state agency or an NGO or farmer 
coóperat~ve 
2 The host ~nst1.tutlon calla a communl.ty meetlng l.n whl.ch farmers 
make a group analysls of what l.t means to experl.ment: wl.th new 
agrl.cultural practl.ces~ of local experlence w1.th experl.mentatl.on 
and lts results and of the purpose of a local research comm:tLLee 
3 rf the commun:tty dec:tdea to establlah a erAL lt electa a four-
member comml.ttee Di farmers recogn1.zed locally as experl.menLers, 
wlth leadershlp qualltles def:tned together wlth the communlty, 
before the elect:ton 
4 The erAL conducts a dl.agnosl.s 1n one or more 
at WhlCh a tOplC for the erAL experlment (eg 
pract~ce fertll~2er use) lB pr1or~tlzed 
commun1ty meetl.ngs 
a crop, cultural 
5 In a plann1ng meetl.ng w1th thel.r host l.Ilst1tutl.on s agronom1st 
the erAL deflnes the ab)8ctlve of thelr experlment the treatments 
and the check cr~ter1a for s~te selectl.on , tl.ffil.ng l.nputs, data 
needed ta draw canclus10ns fram the tr1al, responsl.b111.tl.es for 
dlfferent tasks In the fl.rst tralnlng cycle, a support farmer 
V1Sl.ts the eIAL on a regular hasla, as these tasks are 
lmplemented 
6 Once the experlment lS planned, the erAL carrles out the 
actlvltles lnvolved from plantlng to harvest, managlng the 
communlty' s erAL fund ThlS 18 a collectlve rotatlng fund ln 
wh:tch each erAL has a share In Colombla the erAL fund amounts to 
less than 50% of the value of a head of llvestock ln the p:tlot 
area (US$375 per erAL at current exchange rates) 
7 Once the experlment has be en harvested the eIAL meets wl.th the 
agronornl.st ta draw concluslons freID the data they have taken on 
thelr experlment, and plana the commun~ty meetlng at WhlCh the 
erAL w:tll present lts resulta 
B The communlty meets to hear an oral report by the erAL of lte 
act~v1tles results and fl.nanc1al status If appropr1ate the 
d~agnosls ~s repeated to orlent the erAL s actlvlt~es for the next 
season 
9 In the second and subsequent cycles of experlmentat~on two or 
three mon1torl.ng Vl.Slts are conducted by the support-farmer 
6 
Ílve CIALs wIm.J¡ mel d~ .i glOUp, dmI to e,u..!! commuIIIty Wd& pdll ol lh", plOl-eb& Alter the 
ftrst tra10mg cycle (01 cropp1Og season), the agronomlsts m the IPRA proJect team began to 
gradually hand over each operatlOn m ¡he process to the farmers Momtofll¡g VISlts by a 
lAA 01 FrlllrCIAL WB 
7 
soelologlsl were made regularly lo assess how well Ihe farmers were able lo mandge eaeh 
operatlon, dnd to detecl when follow-up Irdmmg was reqUlred 
On lhe bdSIS ot thls expenem.e, Irammg matenals m lhe lonu ot twelve CIAL hdndbooh 
were prepared usmg dlscusslons wlth the farmers mvolved, who helped to prepare the text and 
Illustratlons (see Appendlx 1) 
At the end of 1991, the second phase of formatlOn of CIALs was mltlated The plOJect 
used the t1ammg malt-nab to teaLh a cour~e wlth NGO'~ m the pllot <lILd, lo plcpale 
agronomlsls on Ihelr staff to eslabhsh CIALs As a result of the coursc, a furlher 13 CIALs 
were establlshed usmg Ihc Irammg handbooks MOllllonng by Ihe proJcct of Ihls second phase 
now eovered 18 CIALs, and mcluded revlslon of!he trammg handbooks as Ihese were used m 
practlce by the NGO tramees and thelr CIALs Bdsed on thls expenence, the trdnllng hdndbook~ 
were fmahzed, and m 1992 the proJect began to teach a regular course on the CIAL method to 
NGO tramees (who are ulllverslty students domg a slx-month agncultural extenslOn practlcum 
m thL lUral areas wlth the NGO), slale extcnslon agcnts and local commulllty ICdders 
In a thud phase ,1 turther 28 CIALs were formed m response to rcquests hom commullltle~ 
and talmer assoclatlons In 1993 tramees m the course on the CIAL method mcluded three 
farmers who were members of CIALs formed m the second phase Thesc support-farmers werc 
contraLted (one by an NGO, one by a tarmer cooperatlve, one by the proJect) to form the CIALs 
m the tlmd phase 
DevolutlOn of respons¡b¡llty for adaptlve testlng 
One of the most Important questlons for the proJect IS "what typLS ot lespOIlSlblllty lor 
locatlon-speclflc technology testmg can be successfully taken on by expernuentmg tdnners 
orgalllzed m aCIAL?" 
The 48 CIALs formed m the pllot area have been estabhshed successlvely by tramees 
eommg fresh lo the methodology TIlIs has dllowed the proJect to evaluate the trammg 
I eqUlrements for settmg up new CIALs, the rate at wlllch CIALs can be progresslvely 
"detached" from then Iramer, and the rate at WhlCh they can take over the responslblhty for 
cdrrymg out expenments m the absence of a tramer 
JAA 01 FrmrCIAL WB 
8 
Table 1 summanzes concluslOns on the type of mslitutlOnal ;UppOlt requllLd by a fully-
tramed CIAL m the form of trammg and momtormg VISltS to carry out a clOp-related On-IalIl1 
expenmcnt Qur expenence demonstrates that farmers' commiltees workmg on thelr own, 
confldentIy and accurately record results for separate treatments wlthout confusmg these when 
they are mterested m the results When m the plannmg meetmg, farmers have defmed data they 
want to take, m measurement unlts that make sense to them, they are able to analyze these dat,! 
to compare treatments, to as ses s germmatlOn rates and crop development For example helght 
of malze plants was evaluated as too low (dogs can reach and stcal cobs), medlum, wlll 
wlthstand wmd and leslst lodgmg (deSlfed), too tall (susceptible to lodg111g) Yleld data IS 
commonly processed by larmers 111 terms of yleld per umt 01 seed, because they do not 
customallly use measures of area, although experimental plots ale mc,lsured and staked out 
Momtonng of the 48 CIALs showed that of thc 15 actlVltles detallcd 111 1able 1, a fully-
tramed CIAL (wlth two cycles of expenence) reqUlred trammg support m four actlVltles at most 
In phases I dnd II, two of these reqUlred the presence of a tlamed agronOllllst plannmg, 
mcludmg the statlstlcal deslgn, and analysls of results Two VISltS by a support-farmer were 
Identlhed dS deslrable to check that plot selectlOn IS conslstent wIlh the cxpenmentdl obJedlves, 
and to ensure that data at mld-term or harvest evaludtlon IS taken accurately Momtonng VISltS 
loutmely mvolve a VISlt by a support-farmer to the commumty diagnOSIs dnd comlllul1lty ICpOlt 
Illeetmgs By phase III the support- farmers were begmmng to take ovcr rcsponslblltty fOI 
;UppOlt of plannmg the CIALs tnals, and the analysls of results, by bnngmg the plans and later 
the results to a meetmg for thls purpose wlth one of the host mStltutlOns' agronomlsts 
Mos! 01 the CIALs' resedrch questlOns can be addrcssed by smgle fdctor expenmcnts (eg 
6-10 valletles supenmposed on local cultural practlces, or 3-4 fertlltzcr or pest contlOl 
treatlllents) Thls Illakes support of tnal plannmg and analysls feaslble for pdraplOlesslOnal 
falmers, together wlth a check of the deslgn and mterpretatlon provlded by an agronollllst (In 
practlce trammg was reqUlred as Illuch to remforce skllls of tramee agronollllsts to deslgn and 
analyze on farlll tnals, as to teach these ski lis to farlllers) 
The research agenda def111ed by the CIALs 15 evolvmg from pnnclpally germplaslll-based 
strategles (the search for new crops and vanetles) at the beglnnmg, to an mtelest m cultural 
practlces once a viable new crop or locally adapted vanetles ha ve been selectcd by thelr 
JM DI TrmrCIAL WB 
9 
exp\..nments Most recent expenence shows thdt the paraprofes~londl fdrmers cdn support CIALs 
m the deslgn of a two-factor expenment plantmg denslty x ferulIzer dosage IS one example ot 
a deslgn set up by aCIAL wlthout the mterventIon of an agronomlst 
Tdbh.. 2 ~hows the rdte of mcredse m dctlvltles carned out by CIALs mdcpendently ot 
mstltutlOnal support m the form of trammg by the agronomlst or support-farmer The ddta on 
the flrst cycle show how the number of trammg VISltS reqUlred has gone down trom 17 needed 
to develop the method and the trdmmg maten al s m Phase 1 to 10 trammg VIsIls m Phdse III 
Trammg sesslons folIow the actIvltIes outlmed m Table 2 In practIce, lhe number of tI allllng 
VISltS has been reduced, because some actIvltIes IIsted m Table 2 such as obtammg mputs, dnd 
repeated actIvIlles IIke observatlOn or evaluatlOn of the expenment, can be carned out 
mdependently by lhe fanners even dunng lhe first trammg cycle 
T dble 2 shows that the newest CIALs formed m Phase III are operatmg by cycle 2 wlth an 
average of 4 VISltS tor trammg and support 1he 5 pllot CIALs formed m Phdse 1 hdv\.. 
conlmued to mcrease thelr dutonomy by the last cycle, pllot CIALs were operatmg wIlh only 
2 ~upport VIsIls (one for plallllmg lile expenment, one tor analysl~ of lesults) by the 
paraplOfesslOnal farmer 
In concluslOn, the proJcct's expenence demonstrates that the trammg ot tarmer research 
comnllttees can be accomphshed m two cycles (le dunng two expenments), and that the fulIy-
tramcd CIAL can tdke over responMblhty for the maJonty ot the aclIvltlc~ Icqull\..d for 1,llmcI~ 
to llnplement on-fann expenments 
Expenence m 1993-4, m phase III, mdlcates that support-farmer~ (wlth two cycles ot 
expenence as aCIAL member) who are contracted as support fdrmers can plOvlde almost alI 
of the trammg and momtonng support reqUlred for formatlOn and mamtendnce of the CIALs' 
expenmentatIon m the form of sImple on-fann tnals TIllS IS permIltmg the agronOITIlsts 
mvolved m the proJect to delegate the plannIng and analysls needed to routmely support fulIy-
tIamed CIALs, as well as the process of CIAL formatlon I 
The next Issue of Importance IS the quahty of the resedrch carned out by CIAL' s operatmg 
wIlh thls degree of autonomy 
JAA-OJ hmrCJAL WB 
10 
Quallty 01 research cOllducted by CIALs 
-
EvaJuatmg Ihe quahty of fanner,' researeb 15 relaled lo Ibe 15sues dlscussed earher, oflhe 
usefulness lO fanners of lhe sClentltk method as compared wlth folk expenmentallon The 
proJect's strnlegy 15 to combme both approache5 a formal expcnment lS planned and eslabhshed, 
bul lÍ farmel s decide 10 make changes m trealments or alter the expenment along the way lO !he 
slyle of folk expenmentatlOn, !hen Ihe only reqUlrement 15 thal t111S 15 a deCI510n made m 
~olOlOlllee by lhe coll.tbor JllOg [,muer; 
crAL expellmenls have been estabhshed wlth a mlOlmum of lhree rephcatlOlls (tarmers), 
and on OCCdSlons are al so rephcaled wlthm each slte, lf an agronomlst judged t1us to be 
ddvlsable Trlals ha ve beell estabhshed on land belollgmg to members of the crAL, on 
LOmmulldl Idnd, or on ¡,md belongmg lo other fmmers, and h,lve mcluded Icnlal or shale-
cloppmg arrdngemenls common m !he commumty m questlon Slte seIectlon IS ,1 decIslon made 
by the erAL, wlth a follow-up VISlt to check that Ihe proposed sltes are conslstent wlth the 
experimental ObjcCllves Idenllfled m the plannmg actlVlty Consequently, the tUdls are earncd 
out wllh a vanety of collaborallve arrangements mvolvmg on occaslOns a group of commulllty 
membels who fOf example, donate labor for the tnal, or an mdlVldual who shdfccrops, 
contnbutlllg Iand or ].tbor, and who gets a share of the harvest Observatlon and evaluatlOn 01 
the tll<ll~' progn .. s~ Ill<ly lIIvoJve ~evcldl ~xpenmentlllg 1,lrm~r~, IdLlI[¡II~u by lh~ CIAL 
commIttee members as knowledgeable experts III the ¡Ople ehosenlll the commulllty's dlagnostlc 
meetlllg, who tben take part m plannlllg and Implementmg the rep]¡catlollS Tllere eXIsts 
therefore, scope for farmers to llltervene and to combme folk expenmentatlOn WIlh the formal 
experimental deslgn 
The project has momtored the quahty of CIAL's rescareh wlth respect to three cntena (1) 
l~ lhe expenmenl Illterpretdble by farmers and also statlstleally dndIyzabIe ¡ (2) were farmers 
wll sdtlshed thJt they couId drdw useful concluslOlls from the í..xperunent, even Ir M,¡tIMI;..,¡lIy 
ulJanalysable? (3) dld fanners conclude !hat ¡hey could not draw useful mfonnatlon fraln ¡he 
expenment? The evaluatlon asks therefore, If fanners pcrcelve the 
experullents as useful for gencratlllg IIlfonnatlon, and as well, do lhe expenments have Ihe 
potennal to supply useful mformatlon to formal research and extenslon systems 
1AA 01 FrlllrC1AL IVlJ 
....--.--
11 
T dble 3 presents the reóults 01 tlus eVdluatlOn Of the 273 lnal plols mdndgcd by CIALó 
durmg 1991-4, the percent 01 plotó (replIcates) that could be used for stattstllal analysls has 
averaged 75 percent In phase 1, 91 percent of plots were Judged mterpretable by farmers, 
although fewer (84 percent) were stattstIcally analyzable In Phase 1I, only 62 percent were 
statlsttcally analyzable, although farmers stIll Judged 89 percent to be mterpletable for thelr 
pUlposes The reasons for thló drop m pelcent statlsttcdlly analyzable welc detcLted m the self-
evaluatmn exerclse conducted wlth each CIAL In the second phase, CIALs were hnked to 
trdmee extenslOn agents who werc managmg the supply of mputs fOl the CIAL c..xpenmcnts 
togcthel wlth those lor thelr NGO's credlt progfdln The credlt plOgrdm WdS pldgued by dc..ldyó 
m obtammg the funds for purchasmg mputs glven m kmd to parttclpatmg tarmers, whlch 
rcsulted m delayed plantmg and subsequent loss of tnal plots from the CIAL expenments The 
CIALs requested that thcy mdnage lhe pelly cash fund for purchasmg cxpcrnncntal mpuls, and 
once thls was put mto operatton, the capaclty of the CIALs to Implement thelr llldls III ,1 tunely 
fashlon Improved slgmflcantly In phase 1Il, the number of plots lost to analy~ls due to late 
plantmg decreascd to three (managed by one CIAL), the remalllder werc lost due to other 
factOls 
In óum, the avelage success rdte by statlstIcal c..ntena IS 75 percent or by larmeró' cntend 
IS 90 percent, m terms of carrymg out tnals Judged locally useful for knowledge genelatIOn 
Reasolls why devolutlOlI succeeds or fmls 
Why IS thls degree of responslblhty and accuracy m conductmg adapttvc tndls achleved by 
reSOUI ce-poor farmers who are very busy people, strugghng to cope wlth runmng thelr own plots 
and fatms? Thls IS especlally puzzlmg III vlew 01 the huge resources devoted m the past to 
trammg and equlppmg teams of on-farm (or fdrmmg systcm) researchers tor whom obtdmmg 
larmer collaboratton or parttclpatlOn III formal expenments was a maJor source of frustratton 
(LIghltoot and Barkel, 1988) Analysls of the success and fallure of CIALs over the penod 
1991-1994, dunng whlch tIme 5 CIALs have become mactIve (representmg 11 pcrcent of the 
total number formed) suggesls tlldt there are several detenmnants of the degree to wluch aCIAL 
makes a COl11mltment to runmng Its expenments Wlth a mllllmum of mstttutIOnal support 
JAA 01 frmrCIAL Wil 
12 
Flrst, the CIAL's trammg must successfully Impdrt the pnnclple that the commlttee's 
obJecllve IS to expenment, to generate knowledge and to dlsprove or dlscredlt uruehabk 
recommendallons If thls obJecllve IS not clear, the CIAL members expenence loss of purpose 
If an expenment shows that local practlce IS m fact the best aVdll,¡ble alternallve to the 
mnovallon bemg tested Our expenence shows truly Impresslve perslstence of ~ome CIALs m 
the face of several expenment~ whlch dld not Identlty a promlsmg mnovdllon compated to 10Cdl 
practIce In thls respect, con(act dmong CIALs IS an Important mgrcdlent ot success, onL CIAL 
beneht~ from the others' expenmentatIon and IS motIvated by It 2 
Second, lt IS ObvlOusly useful for aCIAL to mclude one hterate member who can lead the 
CIAL handbooks aloud to the other members and who can keep recOlds, and tally dLcounts 
becausc thl~ fdclhtates the managcmcnt process However, wntten lecords have plOved 
Important mamly to the host mstItutlOn whlch IS collatmg data from several CIALs Non-hterate 
farmers lecall comphcated vanetal code-names, the lay-out of treatments, dnd the dlfferences 
among treatments wlth amazmg faclllty If (hey percelve the mtormdtlon as ImpOl tdnt dnd useful 
to them Our Lxpenence suggests that htclacy may not be a prelequlsltc fOl farmers (o carry 
out the CIAL's adapllve research responslblhues, but It does mean that support by the 
paraprofesslOnal may have to be more mtenslve over an extended penod Nor does tunctlonal 
Ilhteracy prevent farmers from exchangmg results wlth each other, smce the mal Iladltlon IS 
strong m these communltles 
More cntIcal to success, IS IdentlflcaUon m the group dmgnosls of d problem or questIon 
for the CIAL'~ expenment whlch the farmers concerned want to answer, and whlch IS ot mterest 
to the commumty Thls IS why the momtonng VISltS at diagnosIs and report-tllne are Important 
to ensure that the commlttee feels accountable to lts commumty, and al the same time, gets 
Lncouragement from the mterest shown m ItS results The sense of commumty servlce and 
responslblhty to the group welfare created and remforced m these meetmgs IS posslbly the smgle 
most Important determmant of successful compleuon of the expenment by aCIAL Momtonng 
shows that contllct m the commumty and/or contllct m the CIAL IS convcrsely, most hkely to 
result ltl an ltlaCllve crAL For thls reason, the approach ltlcludes use of a techmque tor 
penodlL eVdluatlOn by the crAL of how ItS members feel about each other and thelr relallonshlp 
JAA 01 FrmrClAL WB 
13 
to thclr commumty, wlllch can be used by the committee wlth or wlthout the SUppOlt fdlmel 's 
presence 
Another motlvatmg factor, IS tlldt mdnagmg aCIAL and expenmenls WllllJl commdnd Ihe 
respecl of "oulslders" has proved to be a useful 1001 for "pullmg m" the altenllon, and re~ources 
of mstltutlons external to the commumty One CIAL has, for example, successfully negotmted 
a grant of land for a communal farm from the state land reform agency, on the strength of 
demonstrated management capaclty and teamwork, others have attracted marketmg arrangements 
wlth Imddlemen who plevlously would nol Journey ro a dlstant vlllage, but dre now attracted by 
lh<.. qUdllty dnd quanllly of produce re:,ulllllg trom expellmcntdllOn, yLt olhcls lldve pLl;uadLd 
Ihe NGO's to mtroduce results of crAL expellmenls mIo thelr credlt proglams fhe motlvdtlon 
lo run an expenment autonomously IS as much relaled to ItS orgamzatlonal funcllOn dS an 
mterface wlth externdl orgamsatlons as ItS usefulness as a method of knowledge generatlon 
Impact 01 eIAL tnals 
The results reported so fm show that the 48 CIALs m the pllot area have, wlth a decreasmg 
amount of mstltutlOnal support carned out a large and mcreasmg number of on-fmm tnals wlllch 
tarmers' conslder useful for knowJedge-generatlon and whlch are to a very Jalge exlent, 
:,tatlstlcdlly analyzable Thls sectlOn exammes the Impact ot these tnals 
A rapld ,lpprdlsdl of thc CI¡\L~' Impact ~howed lhat 111 75 percent of the partlclpatmg 
commumtles there Wd; a percclvcd bcneflt from thelr crAL 111 lhe fOlm of new sLcd, nLW 
cultural practlces or IllfOrmatlon about whlch recommendatlOns to follow Of those CrALs wlth 
no percelved beneflt, aH but two were formed m phase III and are therefore newcomers wlth less 
hkehhood of Impact so far For example, one commumty asked ItS resealch commlttee to 
compdre the state agency's lecommendatlOn to cover the ground under frult bushes wlth black 
plastlc, and a local practlce for controlhng nematodes The CIAL expenment shows to date, 
dt least that the local practlce IS more effectlve under falmer managemcnt 
Stdte mstltutlons 111 the pllot arca set resedrch and extenslOn pnontlcs on the b,ISIS of dlCd 
devoled lo dlfferenl crops m Ihe mumclpahly Ihus 111 Ihe pllol area, Ihe pnonlles are cassava, 
pastUles, sugar cane and coffee Small farmers' parllclpalmg m Ihe CIALs' dlagnostlc meelmgs 
IIdd dlllelLnt pIIOlllle;, as sllown by the ClOpS selected 10l CIAL expenments hsled 111 Table 4 
JAA 01 Trl1lrClAL WB 
- • 
14 
It IS apparent Ihal Ihe commumlles Idenllfled a much more dlyerse research agendd Ihdn Ihe 
m~lllullOns For eXdmple, nol one commumly pllonllzed CdSSdYd m Ihur glOup dldgnosls, 
dlthough oyer 4000 falmers have parllclpaled m commumty meetll1gs to! thls pUl pose Dlvelslty 
111 Ihe CIAL agenda reflects farmers' obJecllves to Idenllfy alterndllves to tradlllOnal cash crops 
(coffee and cassava) and to mcrease thelr food-sufflclency by growmg staples, such as potatoes, 
bedns (a substllute for meal 111 lhe rurdl dlet) and malze (u~ed for feedll1g cluckcns, an Important 
soulce of locally-produced prolell1, and as well an Important II1gredlent of trddltlOnal staple 
dlshes) Cauca department Imports these staples from other parts of the country to meet lis food 
reqUlrements (SAG, 1989), so the local food self-sutflclency agenda reflects d regIOnal problem 
Al pfl.~(.nt, the CIAL dre hdVll1g somc ~uccess 111 exerlll1g dl.l11dnd pull ,md IInplOvmg Ih" 
dlverslty of technologles on offer vla state and NGO programs For example, m the NGO credll 
dnd techmcal asslstance programs, malze and peas were II1troduced as a result of CIALs' 
expenments, and beans were glven more Importance The mumclpal credlt and techmcal 
asslstance agencies (UMATA) .lIso began to respond to farmers' pnontle~, especlally once sOl11e 
of Ihe CIALs started producll1g seed of vanetles they had selected 111 thelr tllals, whlch the 
UMAT A recommended and dlstnbuted to other farmers One UMA T A recently began to use 
the CIALs' results lor formulatmg recommendatlOns to tdrmers partlclpatll1g III Its crcdll 
proglam Another responded to a CIAL's request for hclp III IdentlÍyll1g peanut vanelle~ tor 
testlllg, by obtall1lllg a selectlOn of new vanelles from ICRISAT through the nallonal agncultural 
research dgency 
An lI11portant development was the evolutlOn of some of the CIALs whlch hdd successfully 
&elected new locally-adapted crop vanetles, 11110 small seed-productlOn enlerpnses dehvenng 
seed of these and of 10Cdl vanelles to farmers 111 the area To date SIX CIAL have begun to 
produce seed from SIX yanetal tnals (wlth 23 rephcatlOns) conducted ovel three years Ihat have 
progres~ed to commerclal-scale plots, for whlch they recelve addltlOnal trall1l11g 111 simple seed 
[llOductlon plOccssmg and quahty-control techmques frol11 the agron0l11lst-trall1er Tlm seed 
can be &old wllh statc approval, when VISltS from the natlOnal agency lesponslble fOl seed 
cerllhcdtlOn are mdde, under ¡he cdtegory of "falIller-lmplOved seed " 
T dble 5 shows the dmount of sced produced by the SIX CIAL seed enterpnses The CIAL 
&eed IS dlstnbuted locally 111 the vlllage stores and weekend markets An estlmdted 281 ha of 
JAA DI TrmrCJAL WB 
15 
malze, 3064 ha of beans and 3 5 ha of fleld peas (an entlrely new crop mtroduced by erAL 
experlmentallOn mIo Ule pllol dred) llave been planled wllh eIAL seed More Ihdn 10,000 
farmels llave purchased erAL seed, whlch over one plantmg season Ih eslmuled to hdve 
produced gram lo a glOSS value of over US$2 mlll!on 
Based on Ihe yleld dlfferentlal between locally ava¡].¡ble vanelles and those selecled by Ihe 
CIALs for seed productlon, Ihls producllOn represents an addll!onal US$76S ,000 of gros;, mcome 
lo 10Cdl farmers from malze and beans, and a ncwly mtroduced mcome sOUlce worth over 
US$8000 10 dale from peas On a per caplta basls, thls represents an IlIcrement worlh aboul one 
monlh of wage mcome m one plantmg season, to Ihe farmers who purchased CIAL seed 3 
TI1(. ;,eed enterpnses also gencI,lIe employmenl smce they must hne addltlOnal Idbor lo 
plant, harvest sort, clcan and pack Ihe seed mito 5 kg sacks, abo made locally by women 
The bean seed enlerpnses have for example generated dn average of 20,000 ],¡bor days of 
employmenl locally over flve seasons worlh an esllmaled US$ 85,000 at current wage rate;, over 
lhe 5 years of oper,ltIon 
ThlS Impact has been achleved by SIX CIALs formed early m the proJcIA There 1S 110 
guaranlee of course Iha! Ihe l1ewer CIALs wllI repeal Ihls expenence by Idenufymg l1ew 
pracllces or germplasm wllh comparable Impact The SIX erAL Whlch have developed mIo seed 
enterpnsCb JUay hdve aheady cdplured Ihe besl opporlumly, and Ihe wmdldll prohts hOll1 
parllelpalory breedmg and sced prodUCtlOl1 The illlpact of Ihe newer CIALs may be more 
dlffleult lo rea]¡ze especlally as Table 4 mdlcates, fuelf research agenda IS shlftmg emphasls 
from grdll1~ lo penshables On the other hand, the recent mtroductlon ol held peas, Vid crAL 
expenll1entallon, suggests tlldt Ihere could be scope tor d slgmtJcanl mClease m llllpad 110m 
CrAL expenmentallon wllh hlgh value CIOp~ 
Scalmg-up and the costs 01 tlle e/AL program 
Tllt .. IL'lIl1~ plLsenlul lo 11m pom! show Ih lt " fully-lnmLd crAL L 1I1 ! IkL Ic~pon~lt)lllly 
fOf executmg mosl of fue actIvltles mvolved m Ihe managemenl of Ihe kmds of adapttve research 
trmls reqUlred for Ihe research agenda Idenl![¡ed m fue 48 parttclpalmg comnmmtles The 
expenmentdl resulls have been uselul for knowledge-generatlOn, and more bpeclfl(..ally llave 
conlllbulcd lo mcreasmg ¡he dlverslty of lechnology tested as well as lI11provmg Ihe rale of llow 
JAA 01 FrlllrCfAL WB 
16 
of teclmologles to the partIClpatmg commul1ltIes, wlth slzeable monetary beneflts m the speclflc 
Cdse ot CIAL ~eed purchasers 
One of the most Importdnt questlons thls reseaIch dlmed to dddrl.~s Wd~ to wh,lt extent 11m 
type of farmer partlclpatlon m research could be scaled up to achleve broad coverage, and at 
what cost Before partIClpatory research became fashlOnable, cntlcs often quelled whelher llus 
approach was an expenslve luxury, dltractIve on a case-by-cdse basls dnd when supporled by 
lughly skllled professlOnals, but not dffordable for workmg wlth ldrge members of farmers 
(Farnngton and Martm, 1988) 
In thls sectIon we present mformatlon on the potentIal of the CIAL melhod to IIICledse lhe 
elhclency 01 salaned personnel workmg m on-farm adaptIve resedrch dnd extenslOn, by 
decredsmg the amount of tIme reqUlred for them to carry out on-farrn tnal~ We ,liso examme 
the operdtmg costs of the CIAL corporatIon, a second-order orgal1lzatIon formed by lhe CIALs 
m Cauca, to provlde sorne mSlghts mto the feaslblhty and costs of creatmg self-sustammg 
CIALs 
One way of assessmg the potentml of the CIAL method to mcrease the efflclency of pubhc 
sector or NGO programs carrymg out adaptIve research, IS to compare the amount of the lIme 
reqUlred to conduct an on-farm tnal wIlh and wlthout aCIAL Table 6 presents estImates 01 
the mandays requlred and cost of manpower for an on-farrn tnal run by an extenslon agent, an 
on-fdrm tndl run by a new CIAL m the fIrst cycle of trammg, dnd by a fully tramcd CIAL The 
dnalysls IS based on the actlVltIes m Table 1, for whlch we e~tImate thal dn t.xtenslon agent 
would reqUlre 8 mandays for a tnal wlth up to three rephcates (sltes) A new CIAL reqUlres 
an average of 10 trammg VISltS by the support farrner plus one manday 01 extenslon agent mput 
to do the same Job A fully tramed CIAL can carry out a tnal wlth 4 mandays of support 
trammg from the support farmer and a fractlOn of the mput from the extenslOmst, conservatIvely 
costed here at one day EstImates of the dlfferent manpower costs show that even trammg a new 
CIAL to carry out an on-farm tnalls less costly than runnmg a tnal wIlh a salaned professlOnal, 
glven tlle pdy dlfferentIdls for the pllot alea More Import,mt, devolvmg dn on-lallll tll,tI to d 
tully tramed CIAL costs 60 percent less m manpower costs than runnmg d tnal usmg an 
extenslOn agent 
JAA 01 I 'IIIrClAL \VD 
17 
Dne of Ihe Imphcallons of Ihls figure IS Ih.!t dddptlVe rcscarch progr,uns could pOlcntlally 
reduce thelr manpowcr costs for on-farm leótmg slgmficantly (by up to 60 pelcent), by wOlklllg 
wllh CIALs Alternatlvely, a glven amount of professlonal manpower can be expected to al Icast 
double liS coverage, lhal IS to mcrease Ihe number of on-farm tnals and farmer groups attended, 
by workmg wJth CIALs 
Impoll,ml vanables WhlCh dlleet the etftcleney dnd eovel,lge 01 <lddptlVe reSLdlLh .lIt. tht.. 
vdndblhly of mlcro-,¡groecologlcdl reglmes, the denslty of lhe populatlon ,1IId Ihe typt- ot lerram 
wluch dffc<.t the amount of time requlred ror slte VIS lIs Before gomg on lo examme lhe currenl 
oper,lUng t-OóIS 01 Ihe CIAL corpordllon, lhe soclO-geographlt-dl conlt-xl 1Il wlut-h COVLJ,lgr. 
obldllled by lhe eXlslmg CIALs has been developed, needs lo be descnbt-d 
Today, m early 1995, there eXIsI a total of 55 CIALs scattered m mne mumclplos m Ihe 
Departmenl of Cauca m Southern Colombia (see FIg 1), whlch logelher compose an arca of 6648 
Kml , wlth an average populallon dcnslty of 40 persons/Km2 The commumlles of smal! farmcrs 
partlClpatmg m Ihe proJect represent an arca of mfluencc of approxlmaldy 1605 Km2 1ll WllllJl 
lhe populatlon IS concenlrated at a much hlgher denslty farm-Ievel surveys show an average 
of 132 per~ons/Kml, when Ihe f.xlenslvc cattle and fores! holdmg, ,1I"e nOI I!lduded l 
Commumtles are eh,lfactenzed by a land use of O 25 ha ot cropland per caplta, d tIgure 
eompmable 10 estlmatcs for BollVla (O 33 ha eropland per caplta), EcuadO! (O 25 ha per cap Ita) 
01" Pero (O 17 ha cropland per cap Ita) (Pacluco el di , 1994) 
SlIlce 1991, the CIALs m the proJect arca have deelded to mee! on an armual bdSIS to 
exch,mge Icsults In tl115 one or two ddy meetmg, fmanced by I,Uó1llg money m thell 
commumtles for transportatlon and lodgmg, CIALs glvc oral rcports on thelr expenmenls 
exehange seed, swap notes aboul Ihelr host msl!tul!ons, and formulate recommendd110nS on how 
10 Improve lhelr performance on gOdls Ihey themselves estabhsh 1ll edeh mef.lllIg 11115 
expencnce prompted the electlon of a central eoordmallllg C01l1mlttee (¡unta) m 1993, ,1Ild then 
m 1994 led 10 Ihe CIALs' declslon lo mcorporale legally al Ihe rec01l1mendatlOn of Ihe JI/lita 
Donatlons were obldmed whlch enabled Ihe crAL corporatlon to estab]¡~h an llIvestmenl fUlId, 
ftom wlllch Ihe COrpoldllon Cdn dldW up to 70 per cent ofthe mlel'cst (the remamder gomg bdCk 
mIo lhe capllal) for operalmg expenses Thls put lhe CIALs on a sclf-su"tammg fmanCldl basls 
In addlllon the SUppOl1 larmers have begun lO glve courses lo the 1l1UIlIClp,ll exlemlon ~erVlce<¡ 
¡AA 01 FrmrCIAL WB 
• 
18 
(UMATA~), WhlCh hdve wntracted them lo form small numbers of pllol CIALs dsewhere, 
paymg up to 50 per cent of Ihelr salary, and so generatmg addltlOna1 mcome for Ihe corporallon 
The followmg data on Cosls are tdken from the CIAL corporatlon's annua1 operalmg budget, 
ltself den ved from the proJect's dala on the cosls of runnmg lhe CIALs m 1994 
There Me very j ew publIshed data on lhe costs of domg dd,lptlVe research wllh groups 01 
farmels wIlh whlch lO compare ¡he fIgures m Table 7 These show the tOldl operalmg cost pel 
CIAL le per commumty per yedr dt US$502, and .In annual pcr Lapltd w~ll,mglllg hom $125 
11 we only consHler Ihe 220 farmers who are commltlee membcrs, lo undel $1 ba~ed onlhe totdl 
pOpUIJtlO11 m lhe ,lle.l of mfluence, to $65 per caplta If we a~,umc IIMt only ,1 Illird 01' lhe 
populatlon m the CIALs' commumtlcs actu,¡lIy rccclve any conlaCI wlth thelr CIAL's adapuve 
testmg B.lsed on the estlmated number of purchasers of CIAL seed, lhe cost per LapIta would 
be approxlmately $3 The tolal annual operatmg budget of lhe CIAL corporatlon currently 
,unounts lo the eqUlvalent of about two agronomlsts salanes at natlOnal program rates The,e 
figures compare favor,lbly wllh costs clted by NImios and Savage (1991) of $36 per c.lplta and 
$2664 per commul1lty dnnually for an extellslOll program uSlI1g vllldge-level support tdrmers m 
ecuador Also m Eluador, Romanoff (1993 clted 111 Bebbll1gton el ,11 , 1994) rl-pOIlS Ihe CO&t 
oj formmg groups ot 10-30 members usmg farmer-Io-fanner trall1mg mechanlsm~ at around 
$3000 However, Ihese groups were farmer aSSOCldllons for processmg dnd markelmg CdssaVd, 
much larger and more cOll1plex than the CIALs 5 
Wuh respect lo coverage, figures caed by Schwartz (1994 ll-l2) rdnge from 100 lo 300 
tdrnlers per extellSlOn dgent (pnvdte sector) to 3,000 per extenslOn agen! (pubhc &ectOl) flom 
Cdse sludles m NIgen,I, Kenya and Thalldnd Comparable figures for the CIAL wrporatlon can 
be e&tllllated dI between 66 (duect conlacl wlth comlllltlee membCI&) to ,Hound 3,000 (populdtlon 
01 Ihe CIALs' commumUe~ or seed purchasers for example) pcr salaned 
p,lraprofcsslonal/agronomlst lIowever, Slllce Ihe CIALs do not dI thlS tlIne have ,1 complctL 
l.XIl-INOIl hIllCl!on, hUI ,1 p.!rlMl research/exlenslon funcUon f,lcllIt11l1lg Ihe 'Id tpldllon 01 
lechnology, IhlS compJnson 1& not completely eqUlvalent 
One teason why Ihe cosl of tOlmlllg ,llld runlllng CIALs 15 relauvLly &0 low m,¡y be tlMt 
Ihe procedures for crcatmg these groups werc formal!zed falrly early on m the process, mIo 
Ir,unmg m,\I{,f1dls Ih.!t were wnllcn together wllh ¡,llmers and dre easlly u"ed by 1.llmu~ U\l 
JM 01 FrmrCIAL \VD 
19 
of these matenals means that support farmers wllh praclIcal expenence m the procedures, who 
represent very low-cost manpower, can form and run CrALs wlth mmlll1al external ~llppOrl 
[xpenence m BolIvia for example, ~ugge~ls lhal lhe crAL handbooh can be leadlly u~ed lo 
form and run CrAL~ by exlenSIOI1l~ls wllhoul pnor lrammg m Ihe method (Sona pel ~onal 
commul1lcallon 1995) 
Nonelheless, lhe cost dala pre~enled here should not be vlewed ,IS conclu~lve, lO lhe extent 
that further testmg 01 the approach wlthollt Ihe mtervenlIon of the ongmators (the rPRA proJecl 
lea m) I~ Ilndel\vay, and wIll pellllll assessment oí how IObust and replIcable lhe melhod IS m 
dlíferent envlronments, wlth vanant cost structures 
Eqmty 
An Issue related to as~essll1g the effecllveness of the crAL melhod tm hlOadenmg lhe 
coverage ot adaptlve lesearch, IS the que~tlOn of how eqult,lble IS lhe dlstnbullon 01 belleflt; 
fhe proJecl has yet to conduct a comprehenslve analysls to address thls questlOlI but survey dala 
ale av,lllable on a sub-populalIon of 11 commUl1llles whlch provlde so me lI1slghl As dl~cussed 
earlIer, ;elecllon of crAL members IS predlcated on the assumpllon that expenmentmg fanners 
,Ire IIkely lo be relalIvely better-off members of the rural commul1lty Moreover, the CIALs are 
not deslgned to mvolve a large populatlOn m resealch, the commlttee mobllIzes a e,lpaelty to tesl 
technology wlthm Ihe commul1lty, on the basls of a IImlted partlclpallon m conductmg the actual 
1 esearch TherefOle, dlstllbutlon of knowledge about a CIAL's aetlvllle~ ratller thall 
palllclpallon mil, IS a more Imporlant lest of Ihe exlenl and nalure ot lhe coverage adlleved 
A companson of three social slrala dlfferenllaled on an mdex of well-bell1g (R,lvnborg 
1994) shows that of lhe 64 farmers aClIvely parlIclpalmg m eleven CrALs, 39% come from the 
upper stralum compaled wllh 22 % who come from Ihe lowesl stralum (CIll Squdle p = O 046) 
Among lhe very poor, only 8 % partlclpale 111 CIALs compared wlth 17% of the upper slratum 
lIowever, the commul1lty populallon IS essenlIally one ot small farmers dnd m tlllS sub-~,lll1ple 
there IS no slgmflcant dlfference 111 farm slze belween Ihose who parllclpate 111 lhe COll1ll1lttees 
(average farm slze IS 4 4 ha) and those who do nol (aver,lge farm ~Ize 3 5, probabllIly 01 I = 
() 14R4) Knowlcdge of the CJ¡\J ~ I~ more evenly dl~tnhuled ~2 percenl 01 Ihe popullllon 
JAA 01 FrmrCIAL WB 
20 
surveyed knows of lhe CIALs, and there IS no slgnlflcant dltference bctween the ploporuon of 
very poor people (49%) and Ihe remauung two better-off strata (53 %) who have Ihls knowlcdge 
(Clu Square p = 0491) 
The key 155ue 15 lO what extenl specJaI mterest groups m ¡he commullIty are ahle lO gel 
Ihen pnonlIes onlo Ihe agenda dehned m Ihe commulllly diagnOSIs whlch decides Ihe problems 
on whlch ¡he CIALs do research MOnllonng by Ihe proJect has detecled Ihe m.¡rked tendem:y 
tor tew women lo dUend ¡hese meetmgs, ,md Ihose who do attend often propose research 
problems whlch are nOI pnonlIzed In order lO address Ihls need, lhe proJect eSI,lbhshcd ,1 
~ep.¡rdle tund lor COll1mullltle~ lo sel up d womcll's CIAL 11 .t group 01 10 01 1ll01l. WOml.ll 
lequesled lo do so 1I0wevcr, only IwO women's groups lormed CIAu" .tnd tour 01111.1 S .¡ddl.u 
women lo Ihe commlltees Women sull only represent 7 percenl of COlllmlttee membcrs The 
mam ¡CdSOn Cor thls appears ro be Ihe dlfflculty women have m devolmg tune lo regular 
rneetll1gs thal lake Ihem oul of Ihe home For speclal Inlcresl groups Ilke women, or Ihe seml-
landless labollng poor, Ihe teseaJch commtttee may nol be .In appropnatc lIlstrumenl fOl 
addresslIlg thelr spectal research agenda Several OpllOnS have yet to be explored for exarnple, 
sepdrdte diagnOSIs wltll spectal ll1teresl groups to tdenllfy pnonucs wlltch then ale ll1duded ,l~ 
!reatl11t.ll!S 111 !f1dls l.dmed out by CIAL members Thts ralses Ihe queslton howevl.f, of tlu. 
dcgrec 01 m011 Vdllon of CIAL members lo carry out Indl, on 10plCS of s(,.wnuat y llupOltdnu. lo 
thcm, dnd !he more powerful membcrs oC Ihe commumty Ano!her OpUOll 15 10 havc specJaI 
mlerest groups evaluale Ihe tnals so Ihal Ihelr cnlena Cor whal 15 a deslrable Innov.ttJon are 
ll1cludcd 111 Ihe analysls and rccommendattons drawn Crom CIALs' research It may be however, 
Ihat ll1creasll1g Ihe eqUlly of coverage by adapltve research has ro be acluevcd by targellllg the 
very poor, wtth Ihe "slack" research/extenslon capaClty oC lIltenncdtary orgamSdtlons crealcd by 
devolvmg part oC !he research agenda lo CIALs These lssues are toples for IUlther empmc,11 
research, now 51artlllg m Ihe proJee! 
CONCLUSIONS 
Formmg and mOllllorlllg tlle evoluUon of Ihe CIALs 15 an ongolllg expenrncnt lo assess Ihe 
feaslhllIty, and the lIupllCaltOnS of devolvmg the responstbtlttles for adapttve agrlcultural 
tl.~l.<trdl (o f,t1ll1cr~ I hc CIJ\L~ wen .. formcd 10 IIlve~llg,lle to wlmt cxtent Ihe l11cthod~ fOI" 
¡AA 01 TrmrClAL WB 
21 
p,lrllClpútory dl,lgno,ls únu probJem ddlmllon, plannmg únd eVdlu,llIOll, dUU U[lllllúldy 
momtollng ot adaptlve technology tesllllg, cOllld be handLu over to commumty-Ievel 
Olgalllzatlons, to generale "uemand-pull" on formal research and extenslOn systems, and to 
lmprove the <lceess of resouree-poor farmers to an adapllve technology testlllg scrVlce, at a 
redsondble LO,t Our CXpenL!lCe so I~lf, wggesls tllclt 11 15 po;,slblL to llIWlullOI!.lhzc lIm, 
lesponslblhty wlth falmcrs, that It IS !lot u!lrcahSllc to expect "hard-data" from farmer-managcd 
ddaptlve research, and that tlllS demonslraUon of farmers' capablhty wms respeí.t fOf farmers 
wlllch 15 calalyzmg a gradual reonenlatlon of bureaucratlc mStltutlons' pnOfltles ReslIlts show 
Ihal CUlren! cosls and coverage compare quue fdvorably wuh sorne slale or pnvale seLlO! 
systems, although Ihe basls fO! compansolllS very hmlled A favorable cosl structure 15 clea!ly 
lelated to Ihe demonstraled effectlveness of paraprofesslOnals lIke SUppOlt farmers for scalmg 
up, dnu lor d¡,hlevmg ucvolutlon 
rhe proJcct 15 enterlllg a ncw phdse wlth an IllternatIonal tr,unmg progrdm and l11ol1ltorlllg 
of new CIALs, wluch are uIssemmatlllg m wldely-contrdstlllg socio-cultural envlrorunents as 
dlstant as Brazll and Honduras Many questlOns remam about the long-run vlablhty of lhe 
CIAL~ dS <111 dpproach to m;,lltullonalIzlllg farmer partlclpdtlOn III dgllL.ullur,11 lc;,edll.h Bul 
thele are already some slgnals -- lIke CIAL MIske m BoliVIa WhlCh IS reachmg out to serve 22 
commullllles, Of the CIALs (rechnstened CALITS) m Peru WhlCh orgdlllzed as a group to 
campalgn fOf 5uppOrt from lhelr regIOnal stdte expenment statlOn -- that a qUlet lcvolutton may 
be unuerw,¡y, to bung tlle cdpaclty 01 farmers as ¡esearchers to fuller lcwgmllon 
JAA 01 TrmrC/AL WO 
rabie 1 ActlVlttes carrtcd out by a CIAL In a crop-relatcd experunent, 3nd 
llIstttuttonal support reqUIred 
T¡,llmng & 1\lrd- MOlllLoIIllg by 
Type of dctlVlty Number of asslstance by profcs- para-
acllvltIes agronomlst slonal professlOlldl 
farmer 
GIOUp DIagnosIs 1 (1)' 
Pldnnmg 1 1-
J ~-------=r( 
• - -,.; I"lfJO-~-:' 
, 
Plot selecllon 1 1 ~, -', 
Land prcparatlon 1 
Obt,lIn mputs 1 ~ 
" '- ,-:'«:'1'/ 
Est"bh~h expenment 2 
Check germlllatlOn 2 
Crop management Vauable 
Mld-term ev,¡luatlon 2 1 
Ha! vest evaluallon 1 
Andlysls ot lesult~ 1 1 
Commumty lepor! 2 1 
1 atal 15 2 2 2 
Infrcqucnt 
JAA (JI I rlllrCIAI_ WB 
22 
Table 2 Rate uf merease 1Il mdependence of CIALs from mstItutlOnal SUppOl t 
1991-1994 
Phase of 
CIAL 
Mean Ilumber of VISlts per trammg cycle 
fOlmatlO1l 
No 
of 
CIALs Cycle 1 Cycle 2 Cycle 3 Cycle 4 Cycle 5 
Phase 1 (ptlot) 5 17 10 4 2 2 
Phase II 13 12 7 3 3 
Phasc III 30 10 4 4 3 
JAA DI FrmrCIAL \VD 
23 
Tab1e 3 Quahty of on farm tnals conducted by farmers' Commlttees 
(CIAL) Jalluary 1991 - August 1994 
Pllase of CIAL 
formatlon 
Plldse 1 (P ¡Iol) 
Phase II 
Phase 1II 
Totdl 
1M 01 I rmrC/¡1L \VB 
No of 
plots 
42 
85 
146 
273 
Percent of 
plots 
~ldll~l!L.dlly 
analyzable and 
IIIlupretablc 
by 
fUlmers 
84 
62 
78 
75 
StatlstJcally 
ulldnalysable 
bul IJIterplLtdbk 
by farmcrs 
7 
27 
12 
15 
LosI lo 
aIMlysls 
9 
11 
10 
!O 
24 
25 
'rabIe 4 Expenments oC Local Agncultural Research Commlttees 
NUMBER 01" [XP[RIMENTS' 
TOPIC 
Phase l Phase 11 Phase III 
Pe,¡s ,¡na rel,¡tea cultur di pracllce~ 1 3 5 
Patato 6 5 3 
Malze ano related practlces 7 8 7 
PCdnut 1 3 4 
FrUlts ano related fertt]¡zer 
dosages, pest control 3 6 11 
Bcans 6 4 O 
Snap-beans O 1 1 
Tomato 1 1 1 
Soya bean 1 2 1 
Sugar cane 1 1 
Vegetables 4 7 6 
Chlcken feea nuxes 1 1 O 
Fordge grasses 1 1 
Cover crops (green manure) O O 1 
Gumea plgs O O 1 
101AL 32 42 43 
, 
Sums to less Ihan the number of CIALs because no! aH CIALs are esldbhshmg new 
expenments 
JAA 01 FrmrCIAL \VD 
26 
Tablc 5 Secd prodllctlOn by SIX CIALs and Its c'ltlmatcd Impact over one plantmg season 
Cror 
Mal7C 
Peas 
Total 
lAA O/ I nllTerAL WII 
No 
CIALs 
2 
2 
2 
6 
Total ,eed 
productIon 
(Kg) 
147,080 
8430 
136 
155,646 
EstImated 
area planted 
(ha) 
3064 
281 
7 
3352 
Estlmated 
productlon 
(ton8) 
3064 
1124 
3 5 
4191 5 
Farm gate 
pnce 
(US$/ton) 
683 
488 
2439 
Gross W 
value pi 
(US$OOO) )'il 
2,093 e 
549 ;j 
8 (E 
2650 3 
27 
Table 6 Companson of manpower reqlllfements of an on-farm managed by CIAL 
and by extensJOn 
TRlAL MANAGEMENT MANDAYS TOTAL COST or 
REQUIRED1 SALARlED 
(N) MANPOWER2 
BY EXTENSION RESEARCH 8 62 
NEW CIAL (CYCLE 1) 11 46 
FULL y TRAINED CIAL 5 23 
Notes 
Excludes crop management afler tnal establIshment WhlCh 15 varIable dependmg 
on Ihe crop, and 11111131 diagnOSIs 
¡ Support farmer' S tlme costed al lUuumum wage, extenslOIl agcnt costed al 
2 x mmllllum wage, agroIlOllllst tllne coslcd al average salary current In Ihe p¡]ot 
area 
/AA 01 FrmrClAL WB 
• 
Table 7 Annual opcratmg costs of the CIAL corporatlOn fOl 55 CIAL 
ANNUAL COSTS 
Plrsonnel costs per CIAV 
CasI of expenments pcr CIAU 
OthCI operdtIonal t.Osts per CIAL' 
TOTAL COST PCR CIAL 
Cost per captta 
fotal populatIon (50,000) 
CIAL commumtIcs (12,900) 
33 % of CIAL commumlIes (4260) 
Sced purcha~ers (10,500) 
CIAL CommIttee members (220) 
US $ 
290 O 
90 O 
122 O 
502 O 
$ per caplta per year 
055 
2 1 
65 
26 
1255 
Notcs 
2 
3 
Ineludes AgronOlTIlst (O 33), farmer coordlOator (1 O), support tarmers (2 O) 
Average of costs per CIAL charged agalOst CIAL funds m 1994 
Average of transportalIo11, supphes, and capItal depIecIalIon 011 4 motorcyeles 
/tiA-O/ /IIIIIUAL WIJ 
28 
APPENDIX 1 
IPRA PROJECT TRAINING MATERIALS" 
Ashby, Jacquelme A , Evaluatlng Technology wlth Farmers, A Handbook, CIAT, Palmna, 
1990,95 pages AIso m Spamsh, Portuguese and French 
29 
Guerrero, M P , The lpra Method VIdeo Study GUlde, CIAT, Palmlra, 1991 AIso m Spdmsh 
Guerrero, M P ,J A~hby, and T GracI,I, Farmer Evaluatwn of Technology Preferellce 
Ranking lnstructwnal Umt No 2, CIAT, Pdlmlra, 1993 Forthcommg m Spamsh 
QUlroS, CA, T Grdcld, J A Ashby, Farmer Evaluatwm of Technology Methodology for 
open-ended evaluatwn lllstructwnal Una No 1 , CIAT, Palmna, 1991 AIso m Spamsh 
Roa, JI, L IIernandez and J A Ashby, EvaluacIOnes de tecnologw con productore~ Anallm 
de Informacwn Umdad de lnstruccwn No 3 (m press) Engh,h torthcommg 
THE 1PRA METIIOD (vIdeo), dVdlldblc m Engh,h and Spam,h 
CIAL HANDBOOKS 
IPRA, CIAT, El Ensayo (The Expenment'), CartIlla No 1, 1993, 43 pJgcs 
IPRA, CIAT, Los Conutes de InvestlgacLOn Agropecuaria Local (Locdl AgllcullUIJI Rl.se,lllh 
CommIttees'), CartIlla No 2, 1993, 35 pages 
IPRA, CIAT, El Dwgnosllco (The DIagnosIs'), CartIlla No 3, 1993,29 pages 
IPRA, CIAT, El ObjeflVO del Ensayo (The ObJectIve of the Expenment'), Cartilla No 4, 1993, 
27 pJgl.S 
IPRA, CIAT, La Planeacwn del Ensayo (Plannmg the Expenment'), Cawlla No 5, 1993, 
44 pdges 
IPRA, CIAT, La Evaluacwn del Ensayo (Evaluatmg the Expenment'), CartIlla No 6, 1993,41 
pages 
IPRA, CIAT, Cosas que pueden pasar (Tlungs that can go Wrong'), CartIlla No 7, 1993,43 
pages 
IPRA, CIAT, Compartullos los resultados de nuestro ensayo (Let's shale the results of our 
Expenment'), CartIlla No 8, 1993, 25 pJges 
11'1U\, UA 1, Un w~o leal (A I{LJI-LIIL CI;L ::'luuy ), Culllld Nu Y, IYY3, 37 IMgl.; 
JAA-OI Irn/lClAL WE 
30 
IPRA, CIAI, Como manejar lo:, gaMos del emayo (How to mdlldge w,[, 01 (h" [xp,,¡!!l¡l-nl), 
Cdrtllld No 10, 10 press 
IPRA, CIAT, Las experiencias tamblen cuentan (We Learn flom our Expencnc<.s), Cartilla 
No 11, 10 press 
IPRA, CIAT, SabIendo a tiempo SI vamos bIen (Let's fmd out If we'¡e domg t1ungs nght) , 
Cartilla No 12, ln press 
.. Avallable from IPRA Project (AUn T GarCIa), CIAT, Apartado Aereo 6713, Cah, 
Colombia, South Amenca 
/AA-OI rllnrUAL WB 
31 
REFERENCE~ 
Ashby, J A and L Sperlmg 1994 "Inslltullonahzmg parllelpaloly chcnt- dnvcn 
researeh and Icdmology devclopment 111 agnculture" Agricultura! Adlllllll~trallon 
(Research and ExtenslOn) Network Paper 49 London 001 
Ashby, J, CA QUlfOS ,md Y Rlvcr.l 1987 'Tarmer P.!rttclpillton III On-ral11l V,HILlJI 
Ina1s" Agncultural AdmullslratlOll Nctwork, DIseuSSlon P.lpcrNo 21, London 001 
AXII111, G 1994 RC&POll~C& lo N(.w~lcllcr 29 m Agneulluril! Rcscilrdl .lnd CxtLllMOll Nuwol k, 
News!clter 30 18, London 001 
Bcbbmgton, A J, D Mernll-Sands and J ranlllgton 1994 'Tarmcr and COll1ll1Ul1lty 
orgal1lsatlons m Agncultura! Researeh and CxtcllslOn funetlOns, Impaels ¡¡nd qucsttons" 
AgrIcultura! Admullsl¡atIon CResearch and Exlenslon) Nclwork paper 47 London 001 
Berg, r "The SClcncc of plant brccdmg --support or altCll1atlvc to 
In W de Boef el al, 72-77 
lradltton'll pl.lUICCS" 
Blrkhacusí.T, D, R [venson and G rcder 199\ "The eeonomlc Ill1pact of Agncultural 
[xlcllslOn A Revlcw" [cononuc Devclopment and Cultura! Ch,mgc (Apnl) 607-650 
Bunch, R 1985 Two Ears of Corn A gUldc to pcoplc ccnten"d ,lgnLulluldl IlllDIOVLlllcnl 
Oklahonhl, Wurld NLlghbor~ 
Chambw>, R 1994 'Torcword" to SLOuncs I and J Thompson (cd~) Bcyond r lrlllU rlr'it 
London Intermedmte Tec1mology PubhcatIons LId 
Colhnson, M 1986 "On farm rescarch and agllcullural rcsearch and cxtenslon IIlsl1tullons" 
AgrIcultural Adnllmstratton (Research and ExtcllslOn) Nctwork Pclpcr No 17, LOlldon 001 
Dnnkwaler, M 1994 "Knowledgc, consclousncss .1Ild prcJudlce" In Scoones ¡¡nd Thompson 
(~ds) 32-41 
Drlllkwaler, M 1994 "Devc1optng mtcrdLtlon and undcrstandlllg RRA and I,¡rmer r\"~carch 
groups 111 Zambla" In ~coonc; .lnd I hompson (cds) 133-139 
Dugue P 1993 "fhe Scncgalese [nsUlulL for Agncultural Research (ISRA) ,ll1d lhe ratlck 
RLglon r..tllllU~ fh,uu,¡UOll" In K Wt.lI,nd ¡¡nd J Cop~;t<lke (~d,) NGO, ,md Ih,," ~I,¡I\" 111 
Alnca, London, Routlcdge 
DUIIL Itl, A 1993 ''rnnnung ag! Icultur.11 ~LfV!(.C" l1l sub-S,¡h,¡nn Afnw" ("ood Pohcy 18(6) 
433-465 
JAA 01 ¡ rmrCIAL !Ve 
32 
Farrmgton J and A Martm 1988 "Farmer partlclpatlon III agncultural researeh a revlew 01 
conccpts and praellces" Agrlcultura1 AdmllllstratlOl1 Umt OecaslOnal papcr No 9 London 
001 
LlfflllgtOI1 J, A Bebbmgton, K Welldrd und O J Lewls 1993 Rcluctant Pdrtncrs? Non-
governmcntal orgamza(lOIlS, (he sta(e and sustamable agncultural developl!lcnt London 
Routlcdge 
Gral: W, J Voss and P Nyabycnda 
Rwanda" In R Tnpp (cd) 39-62 
1991 "Chmblllg bcun mtroduLtlon m ::'outhLIIl 
Ilardon, J J und W S de Bocf 1993 "Lmkmg farmcrs and breeders 1Il lo¡,ul crop 
development", m W de Boef, K Amanor and K Wellard Cultlvatmg kl!QwIcdge Genetl¡, 
cllver'illy, Idrmcr eX7JCnmelltatlOn und eroD research lnlcrmcdmtc Tcdmology PubhUl!lons 
LId ,London 64-71 
Helllfleh G, F Worman alld C Kokelso 1991 "Inlegratlllg rPR wllh convcnllOll,lI 011 farm 
researeh programs an example from Botswana Journal fOl" Farnung Systcms Rescarch-
2 1 -15 
Jansscn W, N RUlZ de LondOlio, J A Beltran and J Woolley 1991 "On-Jdrm leseareh III 
support ofvanetal dlffuslOn Bean producllon In CaJamarea, Peru" In R Tnpp (ed) 191-214 
LlghlfoOI, e 1987 "Indlgcnous rc~c.1rch and on-tarm lnab" AgIIGllltUl,¡J Adlllllmll,llloll 
llnd ExlcllSlO11 2479-89 
Llghttoot e and R Barker 1988 "On furm tnals A survey of mcthods" Agncultural 
Adl11ll1lsllallOn und ExtensJOI1 3015-3023 
Maltee, A Z und T Lasalle 1994 "Dlverse and Lumled r¡¡rmers' OrgalllsatIons 111 
1 anzama" Agncullural Admm¡slrahon fRc:,careh and ExtenslOn) Nclwork Pap¡,r 50 London 
001 
Mernll-Sdnds O, S O Blggs, R J Bmgen, P T Ewell, J L MeAlhster ,md SR Poals 1991 
"lntcgratll1g On-farm Resedrch ll110 Nalmnal Agneultural Researeh Syslcms Lessons for 
Researeh Polley, Orgalllzallon and Managemcnt", III R Tnpp (cd) 287-316 
Muchdgala M G V de Reyndl and 1 P Verga Jr 1994 "BuIldmg a Omlogue bctwccn 
ICbLdl\"IJLI~ dnd ~Ill •. ¡(l l,llnl<.l~ [hc lo¡',lI1ll1lb AglOLwlogy CLnltL (CA 1) In l3¡.JLll" 
Agrlcultural AdmllllstratlOn fResearch and ExtenSlOn) Paper SOd, London ODI 
Muslllta, T A "Sllcngthenlllg Ihe IIlformal seed system III communal area; of 21mbabwe" 111 
W de Roef el al (eds) 85-88 
JAA-Ol íIlJIICJAL!VB 
33 
Nunlos 1 J and R F Savdge 1991 "Successfu! SOl! conbcrvallOn III lhe Lcuadoflan 
Ihghlands" Journal of SOII and Water Conservabon Septcmber-October, p 341-343 
Okah C, V Sumbcrg and J Farnngton 1994 Battleficlds and tnal plotb Rhetonc and 
j\callty offarmer partlcloatory rc~eareh London Intermedla!L tcchnology puhhc.ltlon~ 
PdclllCO, D dnd J A Ashby !983 "SlagLs III leehnology Dlifuslon Among Small LUl11cr 
s BlOloglcal and Managcl11ent Screenmg oI a Ncw Rice Vallety m Nepal" AgllLulturnl 
Admml'itrdtlOn 13 (1983) 23-37 
l'acluco D , J A Ashby, and L R Samnl 1994 "Natural Resource and Agncllllural Plospccts 
lor Ihe Hillsldes of Latm Amerlca" CIAT I-hllsldes Program, Anl1llal Rcport, 1993-1994 p 
283-321 
Ravnborg H 1994 "Mcthodology for Idcnlifymg dlffcrent ¡cvels ot wcll-bcmg, ,md theB 
dlstnbullon", mtcrnal dlseusblon papcr, CIAT Hlllsldes Program, CIAT, Pd[nurd 
I{¡eh.uds, P 
fhlUpp, (eds) 
1989 "Agllculture as a performance" lfl Chambers, R, A Pacey and LA 
fanner rlrst rarmer lfll10vatlons and Agncllltma[ Rcsearch Londol1 TTr 
Rogers, [M 1983 DII[u;1011 of Innovatlon (tlurd edltlOn) N Y [he 1 rce Ptess, Glcneoc 
Rudde!l E D 1994 Documentabon oC Fanncr LXPenmCl1ts A Key Strategy for AcllIevll1g 
J'ood Sewntv on a Su;tall1able Basls, Santiago, Clllle, V CC1I10S Mundldles 
5AO, 1989 el Cauea Tlcrras de Progreso SAO, Popayan, Cauea 
5,,[.15, M A 1994 "1 he tcchl11u,ll1~ only bLheve lfl ~uencc dnd wnnot ICad Ihe sky" IhL 
cultura[ dnncn;lOn of lhe knowledgL confllct 1I1 Ihe AntiLs" In SCOOIlCb alld 1 hOIllP;on (cd) 
57-69 
Schwartz, LA 1994 "The role ol' the pnvate sector 111 Agncultural Extensloll eConollllc 
Analysls and case studJes" Agrlcultural Admllllstratloll (Rcscilrch and exlenslOn) Nclwork 
Papu 48, LondolJ OD¡ 
Shah, P 1994 "VllIagc-mandgcd cxtcnslOn systems lfl Indw Il1lphLdllon:. 101 pohey Jnd 
practIcc In Seoones and Thompson (eds) 248-252 
5tHI", e 1995 P"I~Oll.l[ "OIllIllUlllc,¡lIon wllh J A:.hby <1m! 1 GldLJ.l ,¡boul Bohvwn CIAL, 
dunng 1993-4 
Spcrllllg L , M Loevl11sohn, and B N1 abomvura 1993 "Rethlllkmg the farmers role 111 plant 
brccdl11g local bean expcrts and on-cxtenslOn sclcclIon 111 RWJnda" L"pcrunLllt,¡1 Agllcu[llIrL 
¡AA-O/ TWIICfAL Wli 
34 
Spcrlmg L, U Scheldcgger, R Buruchara, P Nyabyenda and S Munyanesa 1994 
"lntenslfymg ProductlOn among smallholder farmers The 11l1pact of Improved chmbmg beans 
m Rwanda" CIA1 Network on Bean Rcsearch m Afnca, OccaslOnal PubhcalIons, No 12, 
CrA T/RE5APAC, Butare, Rwanda 
~tolzenbach, A 1994 "Learnmg by ImprovIsatlOn farmers' expenmentatlOn m Mah" In 
Scoonc; and Thompson (eds) 155-160 
Tllpp, R 1991 
Chlchester, UK 
ed Planned Lhange 111 rarm1l1g Systems Progress 111 On-rdnll Research, 
John Wlley & Sons 
Van CroWdCI, L 1991 "[xtLI1;lOn for profit" Human Orgdl1lZdlIon 50(1) 39-42 
WOlcde, M and H Mckblb 1993 "Lll1kmg genelIc reSOlllce conservatlon to f.lflllcrs 111 
CthlOpm" In W de Boef et al (eds) 78-84 
IAA 01 ,ml/elAL WB 
\ 
35 
END NOTES 
1 Thls however, places heavlcr reqUlrements on the paraprofesslOnals' skill;, pdrtl< .. uldrly 
wlth lespect to the trdmmg lmked to remforcmg the CIALs management capacllies At present, 
the mo;t effeclive paraprofesslOnals m tlus respect are two young farmers wlth pnor expenence 
m commumty commlttees, who llave the self-confldence to speak m lront of a group, and who 
llave tduglJt III the rural schools dlthough they have no liallllllg as tedcher;, they have cOlllplekd 
secondary school 
2 The followlllg example Illustrates how CIALs work together In 1991, Loma Corta 
pnonlized fleld peas m thelr commumty diagnOSIs, wlth the obJectlve o[ tllldmg a shOlt-se.lson 
crop, easy to cultlvate and wlth a stable pnce, useful for consumptlon as well as for sale, and 
Ld;y lo mdrkct ricio ped; were d cOlllpletely new crop 111 lhe CIAL l-XpU IIncnl; TIIi; ll-glOlI 
IS cOII;loereo mdrgm,tl for [Iclo peds dno SO the crop IS not lecolllmenoeo offlClally Lom.¡ COI 1.1 
planted a vanetal tndl wlth four vanelies, obtamed from an expenment stalion 111 another 
department by the paraprofesslonal In another expenment Loma Corta compared three systems 
of support for fleld peas posts wlth stnng, the techmcal recommendatlon, bamboo stakes 
collecled from the local glOves, bamboo stakes wlth one-tlmd ot the dlllount ot stllng 
recommended CIAL Loma Corta dlscarded the techmcal recommendatlOn w¡th posts and sU IIIg 
(thlS costs $5 for the stnng and the other systems use only local matellals, or 60% less stnng 
wlth the bamboo-stdke/stnllg support) Thelr budget analysls showed that the local SUppOlt 
systems I eqUlre more labor but less cash outlay 
At the annual meetmg of CIALs (the "Encuentro e/AL"), CIAL Betama learned about 
the results of Loma Corta wlth peas, then 1Il1tS second cycle (parcela de colllprobaclOll) 
As d result CIAL Betdllld planted the two vanetles dnd the support ;y;telll sekcted by 
CrAL Loma Cortd 
crAL Esperdnza, a colder chmate commulllty, repeated the vanetal tnal wlth the two 
vanetles selected by CrAL Loma Corta to see If they were adaptl-d, wlth two plantlllg 
system; (lme plantmg and "caJuela" le thelr tradltlOnal system of pldntmg holes) After 
oetelmllllllg that Ime plantlllg was preferred, because of the Illghel plant denslty obtamed 
m a ;mdll plot close to the home garden, CrAL Esperanzd pldnted a sewnd expelllnent 
to test the two support systems wllh bamboo stakes 
Loma Corta lost one year (2 cycles) waltmg for an dgronomlst who had prOllllsed to 
obtam more f¡eld ped vanetles WlllCh never matenahzed Notwllhstanomg thls 
demorahzlllg expenence, Loma Corta then went back to expellmentmg wlth the two 
vaneUes, after observmg the progress crAL Betallla was maklllg Wlth peas 
IAA 01 FIIIII ClAL /VB 
36 
CIAL Betama, havmg learned from Loma Corta that the support system of bamboo 
stakes wIlh stnng was preferred, had selected one vanety whlch had the best commelclal 
quahty and whlch the women ~elected fOI ItS large slze (Plqumegra) They planted a 
producuon plot, and began to sell the produce They declded to sell part íresh, and 
dnother part as seed (worth 120% more than the fresh peas) to other farmers m Betamd 
dnd m the reglOn 
On the basls of Betama's expenence, Loma Corta planted Plqumegra wlth the bamboo 
~tdkes dnd strmg, and went Immedlately to productlon plot dnd seed muluphcaUon 
EspeIanza, havmg tested vanety, pldntmg system and support to ItS satlsfactlon, scaled 
up to plOductlOn plot and seed multIpllcatIon, buymg seed from Betama Now fIeld peas 
are begmmng to appear m monocrop and m assocJatlon wlth other crops m farmers' 
flelds, after Betama took It~ ~eed íor dlstllbutlOn to the "Encuentro C/AL" 
3 An analysls by Jansen el al , (1991 195-211) provldes some data on the llnpaLt oí on-farm 
re~edrch wIlh beans carned out m CajdmarCa, Pcru, from 1982-89 the tcsedrLh LdlllLd out by 
on-fdrm researchers and extLmlOmsts mvolved 10 on-farm tnals m 1982, 30 demonstratlon plots 
m 1983 and dn unspeclfled member of demonstratIons and on-farm vancty tnals m 7 reglOns m 
1984, 363 mtervlews wlth farmers m 1985, and 51 expenments, subsequently 53 tnals were 
planted m 1986 EstllnatlOn of the llnpact oí thls on-farm research determmed thdt about 70% 
of the totdl Impact was mduced by the research program, resultmg m addltlOnal bean plOductlon 
worth between US$130,000 - US$265,000 over an addItIonal 5000 ha of bean~ 
4 1 hc 55 CIAL~ were not dlstnbuted geogrdplucdlly JCro~~ tlm JICJ on IhL b,IM~ 01 lO! 
example, the requlrements of parhculJr agroecologlcal mches or chutt glOUpS for 
locdtlOn-speclflc adaptIve testmg The project's pnmary objccUve was methodology 
development, and secondanly to assess how dlfferent mstItutlonal hnkages affec! íarmers' 
capacIty to do pdrtIClpatory research m agnculture, so that CIALs wele locatcd 
geogl aphlcally to faclhtate companson among mstItutlOnal settmgs A~ d result, the 
eXIstmg CIAL~ are probably more lughly concentrated than warranted by the degree oí 
10catIon-speclflc dlverslty m the reglOn 
5 Grdf et al (1991 56-57) reportmg an on-farm research program m Rw,mdd found that 
farmer partlclpatlOn m on-farm research, and a system of group meetmgs reduced the 
costs of on-farm vanetal testmg These were esUmated at US$27,225 between 1986-90 
for wOlkmg wlth about 79 farmers m four commumtIes, later reduced to about 40 
fatl11els m two COl11l11ullltles The research ared covered about 31,500 hd dnd 20,000 
JAA 01 FnlllUAL WIi 
37 
falmhes Our calculallOns based on the costs reported only tor expenmentauon m Ihls 
~tudy (ofwhu.Jl 80% wcrc covcn.d by Ihe mllon..¡1 program ISAR logclhcl w¡th CIAr, 
and fOf WhlCh salanes of researchers were cosled at the level of nal!onal program 
salanes) range from about $63 to $34 per farmer per year ¡f we COfiSlder only the 40-79 
farmers reported as aclually mvolved m the tnals Smce Ih¡s study, lhe chmbmg be..¡ns 
tcstcd huve d¡tuscd dram,1I1cally, so thal lhe per caplta costs oí lhe on-farm researdl 
based on lhe pOpUlatlOIl covered musl have gone down very slgmÍlcantly (Spellmg el al , 
1994) 
IAA-OI FIIIlIUAL WB