• TABLE OF CONTENTS
HIDE
 Front Cover
 Title Page
 Preface
 Table of Contents
 List of Figures
 Introduction
 The farm trial design process
 Source of diagnostic data: Informal...
 Technical design issues
 Political (or institutional) design...
 Specific design issues in domain...
 Specific design issues in domain...
 Specific design issues in domain...
 Recommendations
 Conclusion
 Reference






Group Title: Networking paper ;, no. 12
Title: The process of on-farm trial design : : The Honduran experience of 1978
CITATION PAGE IMAGE ZOOMABLE PAGE TEXT
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00054306/00001
 Material Information
Title: The process of on-farm trial design : : The Honduran experience of 1978
Series Title: Networking paper ;, no. 12
Physical Description: 25 p. : maps ; 28 cm.
Language: English
Creator: Galt, Daniel L.
Farming Systems Support Project.
Publisher: Farming Systems Support Project, International Programs, Institute of Food and Agricultural Sciences, University of Florida,
Publication Date: 1985.
 Subjects
Subject: Agricultural systems -- Research -- Honduras.
Agricultural extension work -- Research -- Honduras.
Farming   ( lcsh )
Agriculture   ( lcsh )
Farm life   ( lcsh )
Caribbean   ( lcsh )
Spatial Coverage: Caribbean
 Notes
General Note: "September 27, 1985."
General Note: Includes bibliographical references (p. 25).
Funding: Electronic resources created as part of a prototype UF Institutional Repository and Faculty Papers project by the University of Florida.
 Record Information
Bibliographic ID: UF00054306
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: notis - ocm6896

Table of Contents
    Front Cover
        Front Cover
    Title Page
        Title Page
    Preface
        Page i
    Table of Contents
        Page ii
    List of Figures
        Page iii
        Page iv
    Introduction
        Page 1
    The farm trial design process
        Page 1
        Page 2
        Page 3
        Page 4
    Source of diagnostic data: Informal versus formal surveys
        Page 5
    Technical design issues
        Page 6
    Political (or institutional) design issues
        Page 7
    Specific design issues in domain two
        Page 8
        Maize variety trial
            Page 8
            Page 9
            Page 10
        (Maize and sorghum) spatial arrangement trial
            Page 11
            Page 12
        Sorghum trial
            Page 13
        Bean trials
            Page 13
        Use of the experiment station for support/backup trials
            Page 14
    Specific design issues in domain three
        Page 14
        (Maize and insecticide) trial
            Page 15
            Page 16
        Sorghum variety trial
            Page 17
        Use of tailored, follow-up questionnaires
            Page 17
    Specific design issues in domain four
        Page 17
        Page 18
        Rice variety trials
            Page 19
        Rice insecticide trials
            Page 20
        Rice weed management trials
            Page 20
    Recommendations
        Page 21
        Farm number: Statistical representation versus work overload
            Page 21
        Trial replication
            Page 22
        Plot size
            Page 22
        Division of the FSR unit into sub-teams by domain
            Page 22
        Logistics and the division of labor
            Page 22
            Page 23
    Conclusion
        Page 24
    Reference
        Page 25
        Page 26
Full Text

4~30 2


THE PROCESS OF ON-FARM TRIAL DESIGN:
THE HONDURAN EXPERIENCE OF 1978


Farming Systems Support Project


International Programs
Institute of Food and
Agricultural Sciences
University of Florida
Gainesville, Florida 32611


Office of Agriculture and
Office of Multisectoral Development
Bureau for Science and Technology
Agency for International Development
Washington, D.C. 20523


NETWORKING PAPER No. 12






















THE PROCESS OF ON-FARM TRIAL DESIGN:
THE HONDURAN EXPERIENCE OF 1978


Daniel Gait
Associate Director, FSSP
International Programs
University of Florida
3028 McCarty Hall
Gainesville, FL 32611


September 27, 1985

















PREFACE


For all the progress made in FSR/E in the past decade, the actual process
of on-farm design remains one of the largest "black boxes" in the FSR/E
sequence. This paper represents an attempt by the FSSP to stimulate
submissions from FSR/E practitioners of current and relevant processes of
on-farm trial design experiences. While the material contained in this
Network Paper is dated, reflects a single person's biases, and is not meant as
a guide to the best method of designing on-farm trials, it does represent a
real field experience in trial design.

Since it is difficult for practitioners to find sufficient time for
documentation, and because there has been a general reluctance on the part of
current FSR/E practitioners to exchange their methods and processes of on-farm
trial design, the FSSP feels that publication of something in this general
area is better than publication of nothing at all. This is especially true
since the design and analysis phases of FSR/E are those about which most first
time practitioners exhibit the most concern. Futhermore, Dr. Hubert Zandstra
of IDRC has been instrumental in encouraging the publication of this type of
information, as well as lending his specific encouragement to Dr. Galt to draft
and revise this paper.

Please read this Network Paper with the above thoughts in mind. We
encourage as many of you as possible to document the process which your team
uses during on-farm trial design. The FSSP welcomes submissions on the farm
trial design process, and will do its best to publish as many as possible in
the Network Paper Series.

Send your submissions to:

Steve Kearl, FSSP editor
IFAS/International Programs
3028 McCarty Hall
University of Florida
Gainesville, FL 32611

We need to continue to learn from one another. Let us hear from you.

Steve Kearl
January, 1986











-ii-


TABLE OF CONTENTS

Section Page

PREFACE .................................................... ........ i

LIST OF FIGURES....................................................... iii

INTRODUCTION .......................................................... 1

THE FARM TRIAL DESIGN PROCESS........................ .................. 1

Source Of Diagnostic Data: Informal Versus Formal Surveys............. 5

Technical Design Issues.............................................. 6

Political (Or Institutional) Design Issues............................. 7

Specific Design Issues In Domain Two ................................. 8

Maize variety trial............................................... 8
(Maize + sorghum) spatial arrangement trial....................... 11
Sorghum trial..................................................... 13
Bean trials.................................................... ... 13
Use of the experiment station for support/backup trials........... 14

Specific Design Issues In Domain Three................................. 14

(Maize + insecticide) trial ...................................... 15
Sorghum variety trial............................................ 17
Use of tailored, follow-up questionnaires......................... 17

Specific Design Issues in Domain Four...................... ......... 17

Rice variety trials............................................. 19
Rice insecticide trials............................... ...... 20
Rice weed management trials....................................... 20

RECOMMENDTIONS...................................................... 21

Farm number: Statistical representation versus work overload..... 21
Trial replication............................................ 22
Plot size.................................................. ..... 22
Division of the FSR unit into sub-teams by domain................. 22
Logistics and the division of labor ............................... 22

CONCLUSIONS............ .... ....... .... ..... ...... ..... ............. .. 24

REFERENCES....................................................... 25











-iii-


LIST OF FIGURES

Number Title Page

1 Administrative Regions of Honduras ............................ 2

2 Recommendation Domains in the Comayagua Region of Honduras,
1978-80 .............................. ................ ....... 3

3 Characterization of FSR Zones, Comayagua Region of Honduras,
1978 ........................ ............................. 4

4 Types of FSR Trials Conducted in the Canayagua Region, by
Zone, 1978-80........................................ 9

5 Most Frequently Cited Farmer Problems by Crop, Recommendation
Domain 2, La Paz............................................ 10

6 Spacial Arrangement Details of the (Maize + Sorghum) System.... 12

7 Most Frequently Cited Farmer Problems by Crop, Recommendation
Domain 3, El Rosario....................................... 16

8 Most Frequently Cited Farmer Problems by Crop, Recommendation
Domain 4, San Jeronimo....................................... 18










INTRODUCTION

In Honduras, FSR began in early 1978 with an informal survey to help the
farming systems unit focus on the potential research areas of the Comayagua
Valley (Galt, et al., 1982). The Comayagua Valley is located in Region 2, the
Central-Wst Region (FIGURE 1). The results of this survey were combined with
a "pilot" survey in La Esperanza to design a formal, six-page farmer
questionnaire. The formal questionnaire was administered in four different
but relatively homogeneous zones in the Cmaayagua Region during the week of
March 30-April 3, 1978. Surveys were carried out in the first and second
homogeneous areas -- Flores and La Paz -- on Monday and Tuesday. The final
formal surveys were completed in the third and fourth homogeneous areas -- El
Rosario and San Jeronimo -- by Friday afternoon (FIGURE 2).

Formal analysis of the data collected was not performed by the survey
teams. Instead, an agronomist/ecologist and an economist took the 111 surveys
home on the weekend of April 4-5 and completed a partial analysis on those
variables thought to be most valuable in assisting the farming systems team
design farm trials. The analysis process consisted of sunnarizing all vari-
ables of most value for trial design per recommendation domain on spread
sheets, deriving ranges and means for these variables using hand-held calcu-
lators, noting the frequencies of systems per domain, summarizing the planting
details of the basic grains of the systems on the spread sheets, and sumnari-
zing this information in tabular form.

The process of team farm trial design began the following Monday and
occupied the FSR unit for three days. Eight researchers representing the six
disciplines of plant breeding, agronomy, entomology, agricultural economics,
plant pathology and ecology participated. The whole process was quite dynamic
and included a great deal of discussion among disciplines and researchers.
Technical considerations were only a part of the relevant design criteria.
Other design criteria included logistics, cultural practices, politics,
estimated costs, and personal relationships.

THE FARM TRIAL DESIGN PROCESS

The trial design meeting began by the team economist summarizing the
findings of the "pre-analysis" exercise which had been carried out over the
weekend. The parameters covered in the pre-analysis summary included a total
accounting of the number of farming systems encountered in each recommendation
domain, a summary of the three or four most common systems, the means and
ranges of the yields of the crops in these most common systems, the number of
total farmer-reported problems by crop within the systems, and a summary list
of the three most commonly mentioned farmer problems from the previous total
list. FIGURE 3 provides a summary of some of these parameters by homogeneous
area (or recommendation domain) and component crop.

In addition to these system and component crop parameters, the details
of the cropping patterns by crop and system were also summarized for each of
the major systems within each recommendation domain. These summaries included
(1) common name of farmer variety, (2) seed color (for maize and beans), (3)
type of planting (broadcast, row or hole), (4) average number of seed planted
per hole or row, or seeding rate, (5) distance between holes within rows (if
applicable), (6) distance between rows (if applicable), and (7) depth of seed










INTRODUCTION

In Honduras, FSR began in early 1978 with an informal survey to help the
farming systems unit focus on the potential research areas of the Comayagua
Valley (Galt, et al., 1982). The Comayagua Valley is located in Region 2, the
Central-Wst Region (FIGURE 1). The results of this survey were combined with
a "pilot" survey in La Esperanza to design a formal, six-page farmer
questionnaire. The formal questionnaire was administered in four different
but relatively homogeneous zones in the Cmaayagua Region during the week of
March 30-April 3, 1978. Surveys were carried out in the first and second
homogeneous areas -- Flores and La Paz -- on Monday and Tuesday. The final
formal surveys were completed in the third and fourth homogeneous areas -- El
Rosario and San Jeronimo -- by Friday afternoon (FIGURE 2).

Formal analysis of the data collected was not performed by the survey
teams. Instead, an agronomist/ecologist and an economist took the 111 surveys
home on the weekend of April 4-5 and completed a partial analysis on those
variables thought to be most valuable in assisting the farming systems team
design farm trials. The analysis process consisted of sunnarizing all vari-
ables of most value for trial design per recommendation domain on spread
sheets, deriving ranges and means for these variables using hand-held calcu-
lators, noting the frequencies of systems per domain, summarizing the planting
details of the basic grains of the systems on the spread sheets, and sumnari-
zing this information in tabular form.

The process of team farm trial design began the following Monday and
occupied the FSR unit for three days. Eight researchers representing the six
disciplines of plant breeding, agronomy, entomology, agricultural economics,
plant pathology and ecology participated. The whole process was quite dynamic
and included a great deal of discussion among disciplines and researchers.
Technical considerations were only a part of the relevant design criteria.
Other design criteria included logistics, cultural practices, politics,
estimated costs, and personal relationships.

THE FARM TRIAL DESIGN PROCESS

The trial design meeting began by the team economist summarizing the
findings of the "pre-analysis" exercise which had been carried out over the
weekend. The parameters covered in the pre-analysis summary included a total
accounting of the number of farming systems encountered in each recommendation
domain, a summary of the three or four most common systems, the means and
ranges of the yields of the crops in these most common systems, the number of
total farmer-reported problems by crop within the systems, and a summary list
of the three most commonly mentioned farmer problems from the previous total
list. FIGURE 3 provides a summary of some of these parameters by homogeneous
area (or recommendation domain) and component crop.

In addition to these system and component crop parameters, the details
of the cropping patterns by crop and system were also summarized for each of
the major systems within each recommendation domain. These summaries included
(1) common name of farmer variety, (2) seed color (for maize and beans), (3)
type of planting (broadcast, row or hole), (4) average number of seed planted
per hole or row, or seeding rate, (5) distance between holes within rows (if
applicable), (6) distance between rows (if applicable), and (7) depth of seed









FIGURE 1: ADMINISTRATIVE REGIONS OF HONDURAS


\ REGION 4 LITORALL ATLANTIC)
LA CEIBA ,, _

REGION 3 (NORTH)
S SAN PEDRO SULA
,-, r REGION 5 (EAST)
( ," ,- .... OLANCHO
REGION 7 ....,- / -
(WEST) REGION 2
SANTA REGION 2 \
SANTA \ (CENTRAL-WEST)J -
OSA DE \ COMAYAGUA ,
COPAN REGION 6
(CENTRAL-EAST)
DANLI

REGION 1
(SOUTH)
CHOLUTEC






3
FIGURE 2 : RECOMMENDATION DOMAINS IN THE COMAYAGUA REGION OF HONDURAS, 1978-80


S





To San Pedro Sula




















RD #2:
La Paz


RD #3:
El Rose

0" -
c ) ,


I---

-S


Experiment Station


-
p


S RD #4:
C) San Jeronimo


I--

irio )
/-o
6
/
I
I














Idirt road
/
i










dirt road






-----dirt road


RD #1:
Flores


To Tegucigalpa




















Characterization of FSR Zones, Comayagua Region of Honduras, 1978


Average No. ofa
Is Irri- Percent Farm Existing In Order of Importance, In Order of Importance, the
nation a Average Size, Cropping the 3 Predominant Cropping 3 Most Important Problems
Zone Possibility? Slope Ha. Systems Systems Identified Were: Listed by the Farmers Were:


Flores Year-round 0-2 5.0 25 1) Corn in spring followed 1) Fall armyworm in corn
by corn intercropped 2) Bird damage in corn
with beans In fall 3) Slugs (Babosa) in beans
2) Tomatoes in spring
3) Three tied for 3rd:
a) Corn monoculture
b) Rice monoculture
c) Corn in spring followed
by tomatoes in fall

La Paz Supplemental 0-2 10.6 19 1) Corn intercropped with 1) Fall armyworm in corn
sorghum in the spring 2) Bird damage in sorghum
2) Corn monoculture 3) Bird damage in corn
3) Cassava monoculture

ill Rosario No 25-30 3.6 3 1) Corn in spring followed 1) Slugs (Babosa) in beans
by beans in fall 2) Two tied:
2) Corn and sorghum Inter- a) Fall armyworm in corn
cropped in spring b) Bird damage in sorghum
3) Corn monoculture

San 3eronimo Yes, but cur- 0-2 4.5 10 1) Rice monoculture I) Fall armyworm in corn
rently there 2) Corn monoculture 2) Two tied:
is almost none 3) Corn Intercropped with a) Bird damage in corn
beans b) Carapacho in rice


aAs identified from 28, 28, 28, and 27 questionnaires for Flores, La Paz, El Rosario, and San Jeronlmo, respectively.


FIGURE 3.







5


planting.

Several additional parameters were added to the summary forms but rarely
summarized for the group designing the on-farm trials. The parameters in this
category included (1) farm size, (2) distance between fields and farm house-
hold, (3) age of farmer, (4) availability of supplemental irrigation, (5) type
and average length of grain storage in-home, (6) seed source, (7) average
price received for crop and (8) months) of peak labor need. These parameters
were summarized by means only.

After examining the total number of systems by recommendation domain, the
team retained domains two, three and four, but dropped domain one (Flores)
from trial design consideration. The decision to eliminate domain one was
based on three reasons: (1) there were too many systems (25 systems grown by
one or more of the 28 farmers interviewed), (2) there was no predominant
system based on basic grains, and (3) the typical planting time for the
majority of crops -- vegetables -- was approximately two weeks away. The team
unanimously agreed that there was not sufficient time to design trials, obtain
seeds and necessary inputs, and contact collaborating farmers in Flores.

The design process began by listing the systems in the second reccamenda-
tion domain (La Paz). In this domain, the 28 farmers questioned listed a
total of 29 distinct systems (FIGURE 3). However, unlike the situation
encountered in Flores where no one system was predominant, in La Paz about
75% of the farmers followed one system: (maize + sorghum), while a second
system, (maize beans), was also common (FIGURE 3).

Source Of Diagnostic Data: Informal Versus Formal Surveys

The design process for trials in domain two was quite wide-ranging and
discussion tended to be lively. This was due in part to differences between
the results of the formal questionnaire as administered during the proceeding
week and the rapid rural appraisal (formally named "sondeo" during March,
1978, by researchers in Guatemala) that the team had initially carried out in
the region some two months earlier. We did not know it at the time, but much
of the debate concerning reliability of data in a sondeo versus a formal
survey would continue to be a major issue, off and on again, for at least the
next seven years. However, acceptance of the rapid rural appraisal technique
as one characterized as an efficient, accurate and valid scientific tool has
been steadily increasing (Chambers, 1985; Conway, 1985a and 1985b; Galt, 1985;
Rhoades, 1982). Indeed, some have suggested that rapid rural appraisal is
more than a catchy tool, but represents an on-going paradigm shift in social
science diagnostic methodology (Jamieson, 1985).

In some crucial aspects, the former informal survey was more reliable
than the formal survey. This was particularly true for some details of the
cropping systems which, the group felt, were better accessed by the informal
questioning technique. The informal technique consisted of standing (or
squatting) with farmers in their fields, and drawing their particular planting
details on the ground with the help of a stick or pocketknife. In the more
formal setting of the formal questionnaire process, questions were asked and
answered in the more abstract environment of a farmer's home or store.

The observation that the informal information was often more accurate










than that collected formally lead the Honduran farming systems unit to agree
that future surveys would be one of two types:

(1) "formal" sondeos utilizing a fixed, but flexible, set of ques-
tions to zero in on farmer's actual field practices, and

(2) follow-up formal surveys to take place after farm trials were in
place, designed to elicit answers from farm households on specific details or
processes within the given farm system which need further clarification.
(Examples of such tailored, formal follow-up surveys administered by the unit
in 1978 or 1979 included (a) greater detail on land preparation (including
crop arrangements within systems), (b) soil fertility/soil degregation issues,
(c) sesame cropping systems, and (d) bean cropping systems in two additional
Regions of Honduras.

The contradictions between the informal survey activity and the formal
questionnaire process were compounded by three additional facts:

(1) the entire process of administering the formal questionnaire was
done with only one relatively experienced practitioner,

(2) most of those questioning farmers had little to no experience in
social science interview techniques, and

(3) little or no orientation proceeded the interviewing process.

These factors lead to the introduction of sane unknown amounts of bias into
either the asking of, or recording the responses to, certain questions. Some
of this occurred even though the composition of each two-person interview team
changed.from day to day.

Finally, the farming systems team used a combination of the results from
both the informal and the formal surveys during trial design. In the final
analysis of the activity, common sense and group consensus often meant as much
as "hard" means and ranges.

Technical Design Issues

Quite a bit of discussion centered around definition of the real systems
in farmer's fields. Sane examples follow:

(1) Did beans really follow the maize in field x, meaning that the
system was (maize beans), or were beans planted in only a portion of field
x, meaning there were two systems: (maize beans) and (maize), or were the
beans actually planted in field y instead?

(2) Did farmers in domain two really plant (maize + sorghum) in
three different spacial arrangements, and were they really incapable of
telling us why these three distinct patterns existed, or were we not asking
the right questions to find out?

(3) When the (maize + sorghum) was planted any other way but
"casado" (both seed types together in the same hole), were the two grain types
actually planted at the same time? Or was one planted later? If so, which











one was planted first? How much later was the second crop planted?

In addition to listing the most frequently-reported systems per domain,
and attempting to agree upon what they actually were, frequencies of the most
prevelant problems per system were listed in defending order. Thus, for
example, in the (maize + sorghum) system in domain two, the problem most
frequently mentioned by farmers was foliar insect damage. (Cogollero, or fall
armyworm (Spodoptera frugiperda), was the most frequently-mentioned specific
insect pest causing such damage.) In addition, in the (maize + sorghum)
system in domain three, farmers again identified foliar insect damage as their
most frequent problem. For this reason, it was decided to include an insect
control trial in at least these two domains.

Political (Or Institutional) Design Issues

Another objective of the farming systems team during trial design was to
encourage and build upon interactions with national commodity researchers.
One way in which to do this is to specifically incorporate their most promis-
ing improved cultivars in on-farm trials. For this reason, another early
design decision was to include variety trials in each domain.

A lot of team discussion followed from this decision. Questions included
the following:

(1) What is the maximum number of improved varieties the farming
systems team should request from the heads of the ccnnodities we were mandated
to work with (rice, maize, beans, and sorghum)?

(2) What is the minimum number of new varieties we should try to
involve in any variety trial?

(3) Was sufficient seed available for all of the improved varieties
we were going to ask for and, even if this amount of seed was initially
available during first year trials, where would farmers obtain seed if they
were to demand it the following year? (This question was especially trouble-
some for those varieties which were open-pollinated, such as maize and
sorghum.)

(4) Should the testigo (or check plot) variety be

(a) that of each individual farmer,

(b) a "representative" variety bought from one of the collabo-
rating farmers or from a store in the village near the center of the domain,
or

(c) a composite formed by combining small samples of grain from
each collaborating farmer?

(5) Should farmer practice, or recommended practice, be followed for
plant and row spacings, or should a combination of the two be included in each
trial (farmer spacing for farmer varieties; recommended spacing for improved
varieties)? and, finally,










(6) In the (Maize + sorghum) system, should improved varieties of
both maize and sorghum be combined in a single variety trial, or should the
farmer's variety of maize be used with all improved varieties of sorghum and
vice-versa with farmer's sorghum and improved maize varieties?

Specific Design Issues In Domain Two

In domain two, La Paz, the predominant systems actually were (1) (maize +
sorghum) in the primera (first) season and (2) (maize beans), covering both
the primera and the postrera (second) seasons. From the results of the formal
survey, the "beans" were not considered a part of the (maize beans) system
(FIGURE 3). Again, this is a problem which may arise from strict application
and interpretation of formal surveys.

The major problems reported by farmers by basic grain crop are provided
in FIGURE 5. Trials designed included (FIGURE 4):

(1) Maize varieties (5 improved cultivars + farmer's variety);

(2) [(Maize + sorghum) + soil insecticide] x 4 spacial arrangements;

(3) Maize + nitrogen fertilizer (two farms only for purposes of
observation); and

(4) Legume varieties (5 improved cultivars + farmer's variety).

These researcher-planted, farmer-managed trials were planted on eight farms.
Six of the host farms were privately owned, while two were farmer collectives
(land reform groups, or asentamientos). The legume variety trial, which
contained three improved varieties of common beans and one variety each of
mungbean and cowpea, plus the farmer's variety, were planted in the maize
variety trials in relay (to utilize the maize stalks as support) in the second
(fall, or postrera) season.

On these eight farms, the two-man team assigned to La Paz was responsible
for a total of 364 plots in 34 trials (counting the bean trials of the second
season as eight distinct trials instead of as a part of the (maize bean)
system, which they really were). Plot size ranged from approximately 3.5m x
5m to 10m x 10m, depending on the trial and the method of planting employed by
the individual farmer or farmer group. In general, these trials were super-
imposed researcher/farmer-planted, researcher-managed trials, although all
non-treatment variables were managed by the farmer at planting, as were
certain treatments. For example, the farmer planted all varieties in the
maize and sorghum variety trials, so as to reduce the researcher introduced
management bias at planting to a minimum.

Part of the reason there were so many plots was that, after a long
discussion, the team decided that two replicates per site was the minimum
statistically acceptable for the on-farm trials. Thus, each trial on each
farm contained two replicates.

Maize variety trial

In designing maize trials, the team agreed to contact the head of the










(6) In the (Maize + sorghum) system, should improved varieties of
both maize and sorghum be combined in a single variety trial, or should the
farmer's variety of maize be used with all improved varieties of sorghum and
vice-versa with farmer's sorghum and improved maize varieties?

Specific Design Issues In Domain Two

In domain two, La Paz, the predominant systems actually were (1) (maize +
sorghum) in the primera (first) season and (2) (maize beans), covering both
the primera and the postrera (second) seasons. From the results of the formal
survey, the "beans" were not considered a part of the (maize beans) system
(FIGURE 3). Again, this is a problem which may arise from strict application
and interpretation of formal surveys.

The major problems reported by farmers by basic grain crop are provided
in FIGURE 5. Trials designed included (FIGURE 4):

(1) Maize varieties (5 improved cultivars + farmer's variety);

(2) [(Maize + sorghum) + soil insecticide] x 4 spacial arrangements;

(3) Maize + nitrogen fertilizer (two farms only for purposes of
observation); and

(4) Legume varieties (5 improved cultivars + farmer's variety).

These researcher-planted, farmer-managed trials were planted on eight farms.
Six of the host farms were privately owned, while two were farmer collectives
(land reform groups, or asentamientos). The legume variety trial, which
contained three improved varieties of common beans and one variety each of
mungbean and cowpea, plus the farmer's variety, were planted in the maize
variety trials in relay (to utilize the maize stalks as support) in the second
(fall, or postrera) season.

On these eight farms, the two-man team assigned to La Paz was responsible
for a total of 364 plots in 34 trials (counting the bean trials of the second
season as eight distinct trials instead of as a part of the (maize bean)
system, which they really were). Plot size ranged from approximately 3.5m x
5m to 10m x 10m, depending on the trial and the method of planting employed by
the individual farmer or farmer group. In general, these trials were super-
imposed researcher/farmer-planted, researcher-managed trials, although all
non-treatment variables were managed by the farmer at planting, as were
certain treatments. For example, the farmer planted all varieties in the
maize and sorghum variety trials, so as to reduce the researcher introduced
management bias at planting to a minimum.

Part of the reason there were so many plots was that, after a long
discussion, the team decided that two replicates per site was the minimum
statistically acceptable for the on-farm trials. Thus, each trial on each
farm contained two replicates.

Maize variety trial

In designing maize trials, the team agreed to contact the head of the









Types of FSR Trials Conducted in the Comayagua Region, By Zone, 1978-80


Year: 1978

Zones: La Paz El Rosario San Jeronimo

Corn varieties Corn varieties Rice varieties
((Corn + sorghum) + soil insecticide) Sorghum varieties Rice demonstration lots
(Corn + fertilizer) (Corn + soil insecticide) (Rice + weed control)
Legume varieties Legume varieties (Rice + fertilizer)
Soil insect control

Year: 1979

Corn varieties (Corn + fetilizer) (Herbicides + dosage + timing)
(3 corn varieties + 2 bean varieties) Corn varieties: 3 maturities (Fertilizer + dosage + timing)
(Local corn+local sorghum) versus Bean varieties Rice variety trials
(Local corn + improved sorghum) Soil conservation
((Corn+sorghum) + fertilizer)
(Corn varieties + fertilizer)
(Sorthum + spacing + fertilizer +
management)
Bean varieties

Year: 1980

Livestock trials, including: Corn varieties Corn varieties
-fodder from sugar cane Fertility Rice varieties
-livestock survey (Density + variety + fertilizer) Weed control
-mixed farming systems (Corn + sorghum) Date of planting
Corn varieties
Maturity: corn versus bean
Fertility


FIGURE 4.






















FIGURE 5: MOST FREQUENTLY CITED FARMER PROBLEMS BY CROP,
RECOMMENDATION DOMAIN 2, LA PAZ


Number of Farms
Growing the Crop
Crop (Total=28 Farms)


Number of Dis-
tinct Problems


Maize


Sorghun


Bean


Cassava


Three Most Frequently Mentioned
Problems and their Frequencies

Fall armyworm (includes "worm") =18
Drought = 9
Bird damage = 5

Bird damage = 10
Drought = 9
Gallina Ciega (soil grub) = 2

Babosa (soil slug) = 7
Drought = 3
"Diseases" = 2

Drought = 3
Babosa (soil slug) = 2










national maize program to obtain sufficient seed of the top five varieties
recacmended for the Region. These varieties became five of the six treatments
included in a randomized complete block (RCB) design. The sixth treatment was
the farmer's variety. The improved cultivars included materials from
Guatemala as well as Honduras; white as well as yellow grain types. (The
farmer survey questionnaire indicated that farm households raised and consumed
both color types).

(Maize + sorghum) spatial arrangement trial

This major trial contained the three spacial planting systems followed by
the farmers of La Paz. A fourth treatment, suggested by the ecologist, was
the introduction of a spacial arrangement from an similar ecological area in
Nicaragua for (Maize + Sorghum). The details of the four spacial arrangements
are provided in FIGURE 6. The hypothesis for adding the fourth treatment was
that double rows of maize and sorghun, in contrast to either single rows or
mixtures of the two crops within rows, would lessen interspecies plant ccm-
petition and lead to higher crop yields.

To check the economic importance of soil insects, a soil insect treatment
was added to this trial. This was done to try to control Gallina Ciega (a
white soil grub, or larvae), which was the third most common problem reported
in sorghum in La Paz (FIGURE 5). The trial ended up as a RCB design, with
soil insecticide (Aldrin at planting) versus no soil insecticide being the
additional treatments.

Based upon extensive team discussion, control of foliar insects during
the first year on farm was ruled out, even though this was the most
frequently-mentioned farmer problem in La Paz (FIGURE 5). No treatment to
control Cogollero, or fall armyworm, was included because two of the re-
searchers had experience working with foliar maize insects in Mexico, and
unanimously recommended that more research be conducted on-station before
exposing farmers to farm trials. Discussion on this issue revolved around the
following:

(1) the most efficacious insecticides for control of foliar maize
insects were judged to be either too expensive to recommend for farmer's use
without first conducting timing trials on-station to establish the best time
of application, or were locally unavailable, and

(2) the less expensive insecticides for use on maize were priced so
competatively because they were being dumped on Honduras by their primary
producers since having been banned for use in the U.S. by the Environmental
Protection Agency (EPA).

The farming systems unit then decided that no farmers should be encour-
aged to purchase these lower priced but less efficacious and environmentally
suspect insecticides in the first year of on-farm research. When the team did
include one of these insecticides (Aldrin) in La Paz for control of soil
insects, the team itself applied the insecticide to avoid any potential
health-related problems with farmer households. This method of application
did not resolve the dilemma of what would be done at the recommendation phase
if such a treatment proved to be an economic improvement over farmer practice,
although the team agreed to test a substitute insecticide at some future point




















FIGURE 6: SPACIAL ARRANGEMENT DETAILS OF THE (MAIZE + SORGHUM) SYSTEM


Arrangement

x o


1 (Honduras)

x o

x o

x o

x o


Arrangement

x o


2 (Honduras)

x o


o x 0

x 0 X

o x 0


Arrangement


XO XO
XO XO

XO XO

XO XO


3 (Honduras)

xo xo

XO XO

XO XO

XO XO


Arrangement

x x

x x

x x

x x


4 (Nicaragua)

o o

o o

o o

o o


Key to crops:

x = Maize

o = Sorghum











in the farm trial phases.

A second team discussion lead to another decision which affected the
method of providing recommended inputs to collaborating farmers. With the
exception of the two collaborating cooperative farm groups in the domain, the
nsall producers had no chance of obtaining credit through official channels.
Since the team was working with the poorest farmers in each domain, it was
decided that outlays for on-farm inputs in any given trial would be limited to
a maximum of L 50.00 (U.S. $25.00) per farm household in subsequent years.
During the first season of exploratory trials, any needed inputs were provided
by the farming systems unit. However, never were more than two inputs pro-
vided (seed and insecticide or fertilizer) per trial.

Sorghum trial

In this domain (FIGURE 5), drought is the problem ranked second in
importance by farmers growing all three of the major grain crops listed. The
farming systems team decided that, in the short run, there was no readily
apparent policy solution (such as more, or better distributed, irrigation
water) to this problem. However, from the rapid rural appraisal survey, the
team knew that farmers' use of sorghum in the (maize + sorghum) system was a
risk-aversion tactic employed to reduce the potentially adverse effects of
drought. The team agreed to try to obtain and use some of the sorghum culti-
vars recommended by the head of the national sorghum breeding program for
drought conditions. This was another example of positive interaction between
the farming systems unit and commodity researchers, and marked the first time
in Honduras that these sorghum varieties had been subjected to testing under
natural rainfall on-farm.

Finally, the most frequently mentioned problem farmers indicated for
their sorghum was grain damage by birds (FIGURE 5). To address this issue,
the head of the sorghum program was contacted again for advice, and agreed to
provide sufficient seeds of two distinct sorghum head types:

(1) the open, long-panicled type, and

(2) the closed, short-panicled type.

Use of these two types in farm trials allowed on-farm examination of which
head type suffered the least loss from natural bird damage in the region.
This is a positive example of how research priorities of the team may be
identical to those of a major commodity program. The head of the sorghum
program welcomed the chance to expose different sorghum head types to
significant bird damage under farmers' conditions. The head of the sorghum
program also assisted the team in developing guidelines for quantifying bird
damage to heads of the sorghum plants during the growing season and at
harvest. The sorghum variety trials were all of the RCB design.

Bean trials

Design of bean trials to follow maize in relay in the second (fall)
season included both red and black beans. Red and black beans were included
since both types were grown and consumed in the domain. The major differences
between the two color types are yield and preference. Preferred red bean











in the farm trial phases.

A second team discussion lead to another decision which affected the
method of providing recommended inputs to collaborating farmers. With the
exception of the two collaborating cooperative farm groups in the domain, the
nsall producers had no chance of obtaining credit through official channels.
Since the team was working with the poorest farmers in each domain, it was
decided that outlays for on-farm inputs in any given trial would be limited to
a maximum of L 50.00 (U.S. $25.00) per farm household in subsequent years.
During the first season of exploratory trials, any needed inputs were provided
by the farming systems unit. However, never were more than two inputs pro-
vided (seed and insecticide or fertilizer) per trial.

Sorghum trial

In this domain (FIGURE 5), drought is the problem ranked second in
importance by farmers growing all three of the major grain crops listed. The
farming systems team decided that, in the short run, there was no readily
apparent policy solution (such as more, or better distributed, irrigation
water) to this problem. However, from the rapid rural appraisal survey, the
team knew that farmers' use of sorghum in the (maize + sorghum) system was a
risk-aversion tactic employed to reduce the potentially adverse effects of
drought. The team agreed to try to obtain and use some of the sorghum culti-
vars recommended by the head of the national sorghum breeding program for
drought conditions. This was another example of positive interaction between
the farming systems unit and commodity researchers, and marked the first time
in Honduras that these sorghum varieties had been subjected to testing under
natural rainfall on-farm.

Finally, the most frequently mentioned problem farmers indicated for
their sorghum was grain damage by birds (FIGURE 5). To address this issue,
the head of the sorghum program was contacted again for advice, and agreed to
provide sufficient seeds of two distinct sorghum head types:

(1) the open, long-panicled type, and

(2) the closed, short-panicled type.

Use of these two types in farm trials allowed on-farm examination of which
head type suffered the least loss from natural bird damage in the region.
This is a positive example of how research priorities of the team may be
identical to those of a major commodity program. The head of the sorghum
program welcomed the chance to expose different sorghum head types to
significant bird damage under farmers' conditions. The head of the sorghum
program also assisted the team in developing guidelines for quantifying bird
damage to heads of the sorghum plants during the growing season and at
harvest. The sorghum variety trials were all of the RCB design.

Bean trials

Design of bean trials to follow maize in relay in the second (fall)
season included both red and black beans. Red and black beans were included
since both types were grown and consumed in the domain. The major differences
between the two color types are yield and preference. Preferred red bean











yields are about one-half those of black beans, whereas prices of black beans
are approximately one-half those of reds. In addition to these two traditional
bean types, the decision was made to incorporate two non-traditional types in
the trials: (1) a mungbean variety and (2) a cowpea variety. Thus, the bean
variety trial contained a mixture of traditional and non-traditional beans,
including each farmer's variety. These trials were all RCB in design and were
planted in relay after the maize harvest.

Use of the experiment station for support/backup trials

During the discussion of La Paz trials, the question of station trials
arose: how could such trials complement and reflect on-farm investigation?
At this time, a healthy debate sprang up around the general issue of input
availability. The specific question was the recommended use of the herbicide
glyphosate (Roundup) to control nutsedge (Cyperus esculentus L.), the most
noxious weed on-station. The two sides of the argument were:

(1) glyphosate should be used because it was the most efficacious
herbicide for the job, and

(2) glyphosate should not be used because it was not available in
Honduras.

Further discussion centered on the fact that trials using this herbicide would
have long-run implications, but that no useful short-run conclusions could be
drawn for farmer collaborating in on-farm trials. The decision was finally
made that no unavailable inputs would be used in the farming systems trials,
even for those conducted on-station.

Station trials were designed later during the growing season to address
the problem of control of babosa (Vaginulus plebejus Fisher, or Limax maximus
L.), a soil slug now acknowledged by CIAT bean entomologists as the most
serious economic pest of common beans in Central America. Babosas were
identified by farmers as the most severe problem in their beans (FIGURE 5).
As soil slugs were just beginning to emerge as serious problems in common
beans in various parts of Central America at this time, not enough was known
about control options to recommend anything on-farm.

This concluded the design process for trials in the domain of La Paz.
Combined analysis of experiment station trials with farm trials using modified
stability analysis (Hildebrand, 1984; Hildebrand and Poey, 1985) was not done,
although each farm trial type was replicated exactly on-station.

Specific Design Issues In Domain Three

In domain three, El Rosario, only three distinct systems were mentioned
by the 28 farmers interviewed during the formal questionnaire survey process
(FIGURE 3). This is in marked contrast to the 10 systems in domain 4 and the
29 systems in domain 2 (FIGURE 3).

The three systems identified in El Rosario were:


(1) (maize beans) (as in La Paz),











yields are about one-half those of black beans, whereas prices of black beans
are approximately one-half those of reds. In addition to these two traditional
bean types, the decision was made to incorporate two non-traditional types in
the trials: (1) a mungbean variety and (2) a cowpea variety. Thus, the bean
variety trial contained a mixture of traditional and non-traditional beans,
including each farmer's variety. These trials were all RCB in design and were
planted in relay after the maize harvest.

Use of the experiment station for support/backup trials

During the discussion of La Paz trials, the question of station trials
arose: how could such trials complement and reflect on-farm investigation?
At this time, a healthy debate sprang up around the general issue of input
availability. The specific question was the recommended use of the herbicide
glyphosate (Roundup) to control nutsedge (Cyperus esculentus L.), the most
noxious weed on-station. The two sides of the argument were:

(1) glyphosate should be used because it was the most efficacious
herbicide for the job, and

(2) glyphosate should not be used because it was not available in
Honduras.

Further discussion centered on the fact that trials using this herbicide would
have long-run implications, but that no useful short-run conclusions could be
drawn for farmer collaborating in on-farm trials. The decision was finally
made that no unavailable inputs would be used in the farming systems trials,
even for those conducted on-station.

Station trials were designed later during the growing season to address
the problem of control of babosa (Vaginulus plebejus Fisher, or Limax maximus
L.), a soil slug now acknowledged by CIAT bean entomologists as the most
serious economic pest of common beans in Central America. Babosas were
identified by farmers as the most severe problem in their beans (FIGURE 5).
As soil slugs were just beginning to emerge as serious problems in common
beans in various parts of Central America at this time, not enough was known
about control options to recommend anything on-farm.

This concluded the design process for trials in the domain of La Paz.
Combined analysis of experiment station trials with farm trials using modified
stability analysis (Hildebrand, 1984; Hildebrand and Poey, 1985) was not done,
although each farm trial type was replicated exactly on-station.

Specific Design Issues In Domain Three

In domain three, El Rosario, only three distinct systems were mentioned
by the 28 farmers interviewed during the formal questionnaire survey process
(FIGURE 3). This is in marked contrast to the 10 systems in domain 4 and the
29 systems in domain 2 (FIGURE 3).

The three systems identified in El Rosario were:


(1) (maize beans) (as in La Paz),











(2) (maize + sorghum) (as in La Paz), and

(3) Maize sole cropped in the first (spring) season.

(Maize beans) was identified as a system used by 100% of the farms surveyed,
while (maize + sorghum) appeared on 61% of the farms surveyed and maize as a
sole crop was on only one surveyed farm.

Design of trials for this domain naturally revolved around maize, sorghum
and beans. The most common problems in maize, according to the farmers
surveyed, were (1) fall armyworm damage, (2) bird damage and (3) (tied with
bird damage in frequency) medidor (another foliar maize insect pest) (FIGURE
7). The most ccmaon problem identified by farmers for beans was babosa, the
soil slug also identified by farmers in La Paz as the major problem in their
beans. Finally, the most frequently-mentioned problem in sorghum was bird
damage (FIGURE 7).

During discussion of possible farm trials for the domain, it was gener-
ally agreed that this domain was closest to the "edge" of subsistence, and
that risk aversion was a high priority of the typical farm household. Water
was scarcest in this domain, with rainfall being most problematic of the three
domains, and supplementary irrigation an impossibility. The average family
farm slope ranged from gently rolling to above 55%. For these reasons,
simple, low-cost farm trials were agreed upon as a necessity for the farmers
of El Rosario.

Again, the decision to access promising varieties from national commodity
programs led to two variety trials out of a total of three trial types. The
trials finally designed for the domain were:

(1) improved maize varieties (four improved varieties + the farmer's
variety) in a RCB arrangement. This trial also hosted the improved beans
trial during the second (fall) season (refer back to the bean trials of La Paz
for details, as they were identical),

(2) improved sorghum varieties (four improved varieties + the
farmer's variety) also in a RCB arrangement, and

(3) A split plot trial imposed on the (maize + sorghum) system.
This trial included an application of the insecticide Aldrin to the soil (in
each hole dug by a planting stick) at planting time versus no insecticide, and
included both the farmer's variety of sorghum and an improved variety of
sorghum grown together with the farmer's variety of maize.

The two-man team assigned to this domain ended up supervising 21
individual trials on eight farms, consisting of 196 separate treatments or
plots. Again, as in La Paz, each trial was replicated twice. Also, these
trials generally represented superimposed, researcher/farmer-planted,
researcher-managed trials (see previous discussion on La Paz for planting
routines followed), with farm family labor being used as much as possible to
reduce research-introduced management confounding to a minimum.


(Maize + insecticide) trial
























FIGURE 7: MOST FREQUENTLY CITED FARMER PROBLEMS BY CROP,
RECOMMENDATION DOMAIN 3, EL ROSARIO


Number of Farms
Growing the Crop
Crop (Total=28 Farms)


Number of Dis-
tinct Problems


Maize


Bean


Sorghun


Three Most Frequently Mentioned
Problems and their Frequencies

Fall armyworm = 9
Medidor (worm) = 6
Bird damage = 6

Babosa (soil slug) = 10
"Escarabajo" (insect) = 7
Medidor (worm) = 6

Bird damage = 9
"Ice" (brown plant disease) = 2
"Bulbo" (unknown cause) = 2










The third trial type represented a classic example of a farm trial
designed in the interdisciplinary mode. Since the farmer survey indicated
that insects were a problem in maize, the main plots in the split plot design
contained either an application of Aldrin to the soil at planting, or no
insecticide. Aldrin was chosen only for its low cost and general wide
availability to farmers in the domain, not because the team was comfortable
recommending a pesticide banned by EPA in the U.S. Use of Aldrin as the soil
insecticide represented the classic trade-off between (availability + low
price) versus (efficacy + safety).

Sorghum variety trial

One of the improved varieties of sorghum recommended for this trial was
thought to have sane degree of drought tolerance by the head of the sorghum
program. It was hoped that the improved sorghum would out-yield the farmer's
variety in a direct comparison, especially under low or sporadic rainfall.
This drought tolerance was considered to be of high importance in a zone where
drought was so endemic that it was not mentioned by the farmers as a problem
during the formal survey process. Again, this points out a major difference
between a formal and informal diagnostic approach. During the rapid rural
appraisal earlier in the year, farmers explained that planting was absolutely
determined by the first spring rains, and that this timing was the most
critical aspect of management of their cropping systems. However, during the
formal survey, the problem of absolute dependency upon the start of rainfall
- and its corollary, drought -- were not mentioned as problems in the re-
gion. Most probably this is because the start of the rains and the possibi-
lity of drought are integral parts of the system in farmer's minds, and are
not viewed as "problems" to be mentioned during a survey.

Use of tailored, follow-up questionnaires

After the farm trials were planted in this domain, a second formal
questionnaire was administered to farmers. This questionnaire focused on soil
conservation issues with which the farming systems unit became increasingly
concerned during the cropping season. The analysis of this questionnaire
resulted in further design and implementation of several ad hoc soil conser-
vation trials. Such trials consisted of comparing two types of live contour
plantings to control erosion and start the slow but necessary process of
building up contours in sane of the more steeply-sloping farmer's fields. The
two types of live plant materials used to construct these contours were
perennial sorghum and pineapples.

Pineapples were added to these trials only after a lengthy discussion,
characterized by the presentation of several eloquent arguments and counter
arguments, and only after the team's agronomist promised to "obtain the
pineapples by himself, at his own expense, and without use of program vehic-
les." It can be seen that even in a relatively smoothly-running interdiscip-
linary team, certain decisions must be made unilaterally.

Specific Design Issues In Dcnain Four

In domain four, San Jeronimo, a total of 10 cropping systems were identi-
fied (FIGURE 8). In defending order of importance, the major systems identi-
fied were:










The third trial type represented a classic example of a farm trial
designed in the interdisciplinary mode. Since the farmer survey indicated
that insects were a problem in maize, the main plots in the split plot design
contained either an application of Aldrin to the soil at planting, or no
insecticide. Aldrin was chosen only for its low cost and general wide
availability to farmers in the domain, not because the team was comfortable
recommending a pesticide banned by EPA in the U.S. Use of Aldrin as the soil
insecticide represented the classic trade-off between (availability + low
price) versus (efficacy + safety).

Sorghum variety trial

One of the improved varieties of sorghum recommended for this trial was
thought to have sane degree of drought tolerance by the head of the sorghum
program. It was hoped that the improved sorghum would out-yield the farmer's
variety in a direct comparison, especially under low or sporadic rainfall.
This drought tolerance was considered to be of high importance in a zone where
drought was so endemic that it was not mentioned by the farmers as a problem
during the formal survey process. Again, this points out a major difference
between a formal and informal diagnostic approach. During the rapid rural
appraisal earlier in the year, farmers explained that planting was absolutely
determined by the first spring rains, and that this timing was the most
critical aspect of management of their cropping systems. However, during the
formal survey, the problem of absolute dependency upon the start of rainfall
- and its corollary, drought -- were not mentioned as problems in the re-
gion. Most probably this is because the start of the rains and the possibi-
lity of drought are integral parts of the system in farmer's minds, and are
not viewed as "problems" to be mentioned during a survey.

Use of tailored, follow-up questionnaires

After the farm trials were planted in this domain, a second formal
questionnaire was administered to farmers. This questionnaire focused on soil
conservation issues with which the farming systems unit became increasingly
concerned during the cropping season. The analysis of this questionnaire
resulted in further design and implementation of several ad hoc soil conser-
vation trials. Such trials consisted of comparing two types of live contour
plantings to control erosion and start the slow but necessary process of
building up contours in sane of the more steeply-sloping farmer's fields. The
two types of live plant materials used to construct these contours were
perennial sorghum and pineapples.

Pineapples were added to these trials only after a lengthy discussion,
characterized by the presentation of several eloquent arguments and counter
arguments, and only after the team's agronomist promised to "obtain the
pineapples by himself, at his own expense, and without use of program vehic-
les." It can be seen that even in a relatively smoothly-running interdiscip-
linary team, certain decisions must be made unilaterally.

Specific Design Issues In Dcnain Four

In domain four, San Jeronimo, a total of 10 cropping systems were identi-
fied (FIGURE 8). In defending order of importance, the major systems identi-
fied were:










The third trial type represented a classic example of a farm trial
designed in the interdisciplinary mode. Since the farmer survey indicated
that insects were a problem in maize, the main plots in the split plot design
contained either an application of Aldrin to the soil at planting, or no
insecticide. Aldrin was chosen only for its low cost and general wide
availability to farmers in the domain, not because the team was comfortable
recommending a pesticide banned by EPA in the U.S. Use of Aldrin as the soil
insecticide represented the classic trade-off between (availability + low
price) versus (efficacy + safety).

Sorghum variety trial

One of the improved varieties of sorghum recommended for this trial was
thought to have sane degree of drought tolerance by the head of the sorghum
program. It was hoped that the improved sorghum would out-yield the farmer's
variety in a direct comparison, especially under low or sporadic rainfall.
This drought tolerance was considered to be of high importance in a zone where
drought was so endemic that it was not mentioned by the farmers as a problem
during the formal survey process. Again, this points out a major difference
between a formal and informal diagnostic approach. During the rapid rural
appraisal earlier in the year, farmers explained that planting was absolutely
determined by the first spring rains, and that this timing was the most
critical aspect of management of their cropping systems. However, during the
formal survey, the problem of absolute dependency upon the start of rainfall
- and its corollary, drought -- were not mentioned as problems in the re-
gion. Most probably this is because the start of the rains and the possibi-
lity of drought are integral parts of the system in farmer's minds, and are
not viewed as "problems" to be mentioned during a survey.

Use of tailored, follow-up questionnaires

After the farm trials were planted in this domain, a second formal
questionnaire was administered to farmers. This questionnaire focused on soil
conservation issues with which the farming systems unit became increasingly
concerned during the cropping season. The analysis of this questionnaire
resulted in further design and implementation of several ad hoc soil conser-
vation trials. Such trials consisted of comparing two types of live contour
plantings to control erosion and start the slow but necessary process of
building up contours in sane of the more steeply-sloping farmer's fields. The
two types of live plant materials used to construct these contours were
perennial sorghum and pineapples.

Pineapples were added to these trials only after a lengthy discussion,
characterized by the presentation of several eloquent arguments and counter
arguments, and only after the team's agronomist promised to "obtain the
pineapples by himself, at his own expense, and without use of program vehic-
les." It can be seen that even in a relatively smoothly-running interdiscip-
linary team, certain decisions must be made unilaterally.

Specific Design Issues In Dcnain Four

In domain four, San Jeronimo, a total of 10 cropping systems were identi-
fied (FIGURE 8). In defending order of importance, the major systems identi-
fied were:



















FIGURE 8: MOST FREQUENTLY CITED FARMER PROBLEMS BY CROP,
RECOMMENDATION DOMAIN 4, SAN JERONIMO



Number of Farms
Growing the Crop Number of Dis- Three Most Frequently Mentioned
Crop (Total=27 Farms) tinct Problems Problems and their Frequencies

Rice 20 11 Drought = 9
Carapacho (stan insect) = 5
two tied:
Bird damage = 2
"Diseases" = 2

Maize 24 11 Fall armyworm = 7
Bird damage = 5
two tied:
Medidor (worm) = 4
Gallina Ciega (soil grub) = 4

Bean 9 7 all tied:
Thrips = 1
Babosa (soil slug) = 1
"Ice" (brown plant disease) = 1
Medidor (worm) = 1
Drought = 1
Gusano (worm) = 1
Dibrotica (Diabrotica spp) = 1










(1) rice planted in monoculture in the spring,

(2) maize planted in monoculture in the spring,

(3) coffee (grown in the mountain holdings of the farmers who lived
in the river valley), and

(4) (maize beans) (FIGURE 3 lists 1, 2 and 4).

Note that coffee is not listed in FIGURE 3. This was because the farming
systems unit was bound by the Honduran government's five-year plan to focus on
research in basic grains. Research on coffee came under the mandate of an
autonomous Honduran research entity, ICAFE. For this reason, team trial
design discussions centered on the problems of the predominant basic grain
grown in the domain: rice.

In fact, only after a great deal of discussion was it decided to design
only trials which involves rice during the first year's research. Neither
maize nor beans were researched during the first season. Maize was ignored
because (1) conducting systematic maize trials might have meant sampling a
second group of farmers to act as collaborators, as rice farmers were not
necessarily maize farmers and vice-versa, and (2) the team assigned to this
domain began with only one individual whose specialty happened to be rainfed
rice research. The group decided it would be too much work to assign a
heavier farm trial load to only one researcher. In fact, this one researcher
was soon joined by a Honduran counterpart but, during trial design, it was not
at all sure that a two-person team would be available for research in this
domain. Beans were ignored for both of the above reasons, as well as for the
fact that maize in monoculture appeared to be about three times as common as
the (maize bean) system.

Farmers surveyed identified the following as being the most frequent
problems in rice: (1) drought, (2) carapacho damage (an insect pest), and (3)
(two tied) bird damage and diseases (FIGURE 8)). Almost all rice in San
Jeronimo was already planted with fertilizer (either 12-24-12 or 15-15-15),
but no other inputs were used routinely. Thus, fertilizer became a part of
the check plot in each trial.

Rice variety trials

The team discussed the problem of drought in the domain. Some farmers
and cooperative groups of farmers used pumps to supplement rainfall, while
others did not. It was finally decided that the best way to address the
problem of drought was through the rice variety trial. Including the most
promising varieties from the national rice breeding program was one of the
goals of the first season's on-farm trials. Thus, the decision was made to
include extra rice varieties in the trial which had showed signs of being
drought tolerant in station research trials. So, instead of five improved and
one farmer's variety, these trials consisted of nine improved varieties and
one farmer's variety, for a total of ten treatments. The agronomist assigned
to this domain was given the task of making the final variety selections in
consultation with the head of the national rice research program.

Actually, during the trial design process, the final number of treatments











in the rice variety trial was unknown. While the farming systems unit real-
ized that more than five varieties would be involved, an upper limit of no
more than 12-15 varieties was agreed upon. The final number of nine improved
varieties was agreed upon later the team agronomist had consulted the national
head of the rice breeding program. In addition, availability of seed forced
the team to eliminate four of the improved rice varieties from trials on
several farmer's fields when the trials were actually planted in the following
weeks. However, the same five improved varieties were always used in these
smaller trials. The rice variety trials were RDB designs with two replicates.

Rice insecticide trials

To address the problem of carapacho damage on the rice crop, a separate
insecticide trial was designed. This trial consisted of two treatments: the
farmer's variety, which was usually (CICA 6 + fertilizer), with and without
insecticide -- carbofuran (Furadan) 5% granuals applied to the soil in rows
at time of planting. This is an example of adopting an available shelf
technology to initial farm trials. Once again, the team entomologist played a
key in the design and planning of this trial. Limitations of available land
also meant that there was no room for the insecticide trial with two of the
collaborating farmers with the smallest fields.

Rice weed management trials

Finally, the formal farmer survey in the domain lead the farming systems
unit to believe that the most labor-intensive activity during the rice season
was weeding. Such activity not only required much hand labor, it also re-
quired more oxen teams for cultivation than were in the region.* Thus, more
efficient weed control appeared to be a high priority of the domain. This
specific problem was not explicitly mentioned by the farmers questioned during
the formal survey process.

To address this issue, a RCB design weed control trial with five
treatments was developed. The treatments selected were: (1) farmer check
(farmer weed control practice), (2) application of Stam F-34 to replace a hand
weeding, (3) application of 2,4-D to replace a hand weeding, (4) a combination
application of (Stam F-34 + 2,4-D) to replace a hand weeding, and (5) no weed
control.

The last treatment, to enable the team to assess the economic impact of
general weediness on yield, was added only after another protracted discus-
sion. On the one hand, it was correctly pointed out that none of the farmers
practiced zero weed control and that this treatment was less than meaningless
to them. Indeed, it represented a practice the farmers would think of as
extremely stupid. On the other hand, while those arguing for the inclusion of
this treatment agreed with this general argument, they pointed out that the
unit could agree to compensate the farmers for the rice yield lost by inclu-
ding this treatment. Those assigned to work in San Jeronimo could explain the
rational of the treatment to the collaborating farmers. This rationale was
that this treatment represented a low-cost method of obtaining a reliable
field estimate of the average economic impact of weeds in the domain upon
yield.

The team placed greatest emphasis on the weed control trial, the next











in the rice variety trial was unknown. While the farming systems unit real-
ized that more than five varieties would be involved, an upper limit of no
more than 12-15 varieties was agreed upon. The final number of nine improved
varieties was agreed upon later the team agronomist had consulted the national
head of the rice breeding program. In addition, availability of seed forced
the team to eliminate four of the improved rice varieties from trials on
several farmer's fields when the trials were actually planted in the following
weeks. However, the same five improved varieties were always used in these
smaller trials. The rice variety trials were RDB designs with two replicates.

Rice insecticide trials

To address the problem of carapacho damage on the rice crop, a separate
insecticide trial was designed. This trial consisted of two treatments: the
farmer's variety, which was usually (CICA 6 + fertilizer), with and without
insecticide -- carbofuran (Furadan) 5% granuals applied to the soil in rows
at time of planting. This is an example of adopting an available shelf
technology to initial farm trials. Once again, the team entomologist played a
key in the design and planning of this trial. Limitations of available land
also meant that there was no room for the insecticide trial with two of the
collaborating farmers with the smallest fields.

Rice weed management trials

Finally, the formal farmer survey in the domain lead the farming systems
unit to believe that the most labor-intensive activity during the rice season
was weeding. Such activity not only required much hand labor, it also re-
quired more oxen teams for cultivation than were in the region.* Thus, more
efficient weed control appeared to be a high priority of the domain. This
specific problem was not explicitly mentioned by the farmers questioned during
the formal survey process.

To address this issue, a RCB design weed control trial with five
treatments was developed. The treatments selected were: (1) farmer check
(farmer weed control practice), (2) application of Stam F-34 to replace a hand
weeding, (3) application of 2,4-D to replace a hand weeding, (4) a combination
application of (Stam F-34 + 2,4-D) to replace a hand weeding, and (5) no weed
control.

The last treatment, to enable the team to assess the economic impact of
general weediness on yield, was added only after another protracted discus-
sion. On the one hand, it was correctly pointed out that none of the farmers
practiced zero weed control and that this treatment was less than meaningless
to them. Indeed, it represented a practice the farmers would think of as
extremely stupid. On the other hand, while those arguing for the inclusion of
this treatment agreed with this general argument, they pointed out that the
unit could agree to compensate the farmers for the rice yield lost by inclu-
ding this treatment. Those assigned to work in San Jeronimo could explain the
rational of the treatment to the collaborating farmers. This rationale was
that this treatment represented a low-cost method of obtaining a reliable
field estimate of the average economic impact of weeds in the domain upon
yield.

The team placed greatest emphasis on the weed control trial, the next










most emphasis on the variety trial, and the least emphasis on the insecticide
trial during the first season in this domain, even though drought and cara-
pacho damage were cited by farmers as their most severe problems in rice. The
two-person team covering the San Jeronimo area ended up with 14 trials on six
farms consisting of 172 total treatments or plots. Again, each trial was
replicated twice. These trials were superimposed, researcher/farmer planted,
researcher-managed trials. Again, farmer management at planting was maximized
to minimize researcher-introduced management confounding.


RECOMMENDATIONS

Several interdisciplinary issues were important in the design of trials
in each recommendation domain. These issues generally arose each time a trial
was proposed and discussed. These topics included the following:

(1) the number of farms which would constitute a minimum represen-
tative sample of the farm households in a given domain for each system being
researched,

(2) replication of trials at each farm site,

(3) size of any treatment plot within a given trial,

(4) division of the farming systems unit into two-person teams (one
per recommendation domain), and

(5) overall logistics, division of labor, and vehicle assignment and
availability.

Each issue will be discussed in turn briefly below.

Farm number: Statistical representation versus work overload

There were two basic issues to be addressed regarding farm numbers.
These were

(1) the desire to include as many farms as possible to insure the
representativeness of the sample from the given domain, and

(2) the desire to include less farms, but to make sure that a
tighter sample of collaborators was drawn to represent more closely a chosen
sub-set of clientele.

The dilemma of adequate sample size always canes up in farm trial design
situations. Usually the underlying variability of the domain is unknown,
making it literally impossible to know with certainty how many farm households
should be selected to collaborate with trials in the domain. Estimates in
this team setting ranged from five to 25 farms. To compromise, eight to ten
host farms were chosen for the major trials in each domain. Lately,
Hildebrand has shown how on-farm research teams can explicitly include on-
station trials in evaluation of technology (Hildebrand, 1984; Hildebrand and
Poey, 1985). However, this issue of proper station/farm research interface
was never resolved by the taming systems research unit in Honduras.










most emphasis on the variety trial, and the least emphasis on the insecticide
trial during the first season in this domain, even though drought and cara-
pacho damage were cited by farmers as their most severe problems in rice. The
two-person team covering the San Jeronimo area ended up with 14 trials on six
farms consisting of 172 total treatments or plots. Again, each trial was
replicated twice. These trials were superimposed, researcher/farmer planted,
researcher-managed trials. Again, farmer management at planting was maximized
to minimize researcher-introduced management confounding.


RECOMMENDATIONS

Several interdisciplinary issues were important in the design of trials
in each recommendation domain. These issues generally arose each time a trial
was proposed and discussed. These topics included the following:

(1) the number of farms which would constitute a minimum represen-
tative sample of the farm households in a given domain for each system being
researched,

(2) replication of trials at each farm site,

(3) size of any treatment plot within a given trial,

(4) division of the farming systems unit into two-person teams (one
per recommendation domain), and

(5) overall logistics, division of labor, and vehicle assignment and
availability.

Each issue will be discussed in turn briefly below.

Farm number: Statistical representation versus work overload

There were two basic issues to be addressed regarding farm numbers.
These were

(1) the desire to include as many farms as possible to insure the
representativeness of the sample from the given domain, and

(2) the desire to include less farms, but to make sure that a
tighter sample of collaborators was drawn to represent more closely a chosen
sub-set of clientele.

The dilemma of adequate sample size always canes up in farm trial design
situations. Usually the underlying variability of the domain is unknown,
making it literally impossible to know with certainty how many farm households
should be selected to collaborate with trials in the domain. Estimates in
this team setting ranged from five to 25 farms. To compromise, eight to ten
host farms were chosen for the major trials in each domain. Lately,
Hildebrand has shown how on-farm research teams can explicitly include on-
station trials in evaluation of technology (Hildebrand, 1984; Hildebrand and
Poey, 1985). However, this issue of proper station/farm research interface
was never resolved by the taming systems research unit in Honduras.











Trial replication

The number of replications per trial and the size of each plot per trial
are related issues which received considerable attention. (Plot size is
considered in the next sub-section.) The team decided to compromise between
no replicates (or the use of farm site as the only source of replication) and
the four replicates traditionally used by agronomists in their work on experi-
ment stations.

Initially, the FSR unit hoped that three replicates could be used for
certain trials, but after all trials were designed and two-person teams had
been assigned to each domain, the group soon realized that the size of the
overall work load precluded including a third replicate for any of the trials.

Plot size

Plot size as an issue was settled in nearly the same manner as number of
replicates. In general, the team began by trying to decide what the smallest
acceptable plot size would be, by crop (or by system when more than one crop
was to be grown in the plot per year) and by trial type. Once such a minimum
plot size was agreed upon, the group considered larger plots. However, the
size of the workload prevented the team fran expanding plot sizes in any trial
type.

Generally speaking, the plot sizes were four or five rows in width by 5m
in length for maize, sorghum and beans, and 5m x 5m for rice. The only time
larger plots were used was in the spacial arrangement trials in domain two (La
Paz). In this case, each plot averaged nearly 10m x 10m. However, since the
team had little experience in spacial arrangement trials, there was no "guar-
antee" that these 10m x 10m plots were of adequate size.

Division of the FSR unit into sub-teams by domain

A fair amount of time was spent discussing how the farming systems unit
would divide personnel responsibilities between the three domains. Since
three of the four expatriates in the unit were not native speakers of Spanish,
each was initially assigned to a different domain. Then the native Spanish
speakers were assigned to regions to complement the expatriates.

However, assignment of personnel to domains was not a random process.
The individual with expertise in upland rice research was assigned to domain
four (San Jeronimo) where rice trials formed the backbone of research on
farmer's fields. The rest of team personnel chose the recommendation domain
of their own preference (subject to the language ability constraint noted
above). Unfortunately, during the early weeks of farm trial implementation,
such an assignment of personnel meant that the expatriate with the least
amount of working Spanish ended up alone in domain four. This situation was
soon remedied by adding a Honduran counterpart to that domain, but other teams
should be aware of language constraints when assigning research (or extension)
personnel to various regions of any given country.

Logistics and the division of labor

Finally, major discussions of logistics, potential labor bottlenecks and











Trial replication

The number of replications per trial and the size of each plot per trial
are related issues which received considerable attention. (Plot size is
considered in the next sub-section.) The team decided to compromise between
no replicates (or the use of farm site as the only source of replication) and
the four replicates traditionally used by agronomists in their work on experi-
ment stations.

Initially, the FSR unit hoped that three replicates could be used for
certain trials, but after all trials were designed and two-person teams had
been assigned to each domain, the group soon realized that the size of the
overall work load precluded including a third replicate for any of the trials.

Plot size

Plot size as an issue was settled in nearly the same manner as number of
replicates. In general, the team began by trying to decide what the smallest
acceptable plot size would be, by crop (or by system when more than one crop
was to be grown in the plot per year) and by trial type. Once such a minimum
plot size was agreed upon, the group considered larger plots. However, the
size of the workload prevented the team fran expanding plot sizes in any trial
type.

Generally speaking, the plot sizes were four or five rows in width by 5m
in length for maize, sorghum and beans, and 5m x 5m for rice. The only time
larger plots were used was in the spacial arrangement trials in domain two (La
Paz). In this case, each plot averaged nearly 10m x 10m. However, since the
team had little experience in spacial arrangement trials, there was no "guar-
antee" that these 10m x 10m plots were of adequate size.

Division of the FSR unit into sub-teams by domain

A fair amount of time was spent discussing how the farming systems unit
would divide personnel responsibilities between the three domains. Since
three of the four expatriates in the unit were not native speakers of Spanish,
each was initially assigned to a different domain. Then the native Spanish
speakers were assigned to regions to complement the expatriates.

However, assignment of personnel to domains was not a random process.
The individual with expertise in upland rice research was assigned to domain
four (San Jeronimo) where rice trials formed the backbone of research on
farmer's fields. The rest of team personnel chose the recommendation domain
of their own preference (subject to the language ability constraint noted
above). Unfortunately, during the early weeks of farm trial implementation,
such an assignment of personnel meant that the expatriate with the least
amount of working Spanish ended up alone in domain four. This situation was
soon remedied by adding a Honduran counterpart to that domain, but other teams
should be aware of language constraints when assigning research (or extension)
personnel to various regions of any given country.

Logistics and the division of labor

Finally, major discussions of logistics, potential labor bottlenecks and











Trial replication

The number of replications per trial and the size of each plot per trial
are related issues which received considerable attention. (Plot size is
considered in the next sub-section.) The team decided to compromise between
no replicates (or the use of farm site as the only source of replication) and
the four replicates traditionally used by agronomists in their work on experi-
ment stations.

Initially, the FSR unit hoped that three replicates could be used for
certain trials, but after all trials were designed and two-person teams had
been assigned to each domain, the group soon realized that the size of the
overall work load precluded including a third replicate for any of the trials.

Plot size

Plot size as an issue was settled in nearly the same manner as number of
replicates. In general, the team began by trying to decide what the smallest
acceptable plot size would be, by crop (or by system when more than one crop
was to be grown in the plot per year) and by trial type. Once such a minimum
plot size was agreed upon, the group considered larger plots. However, the
size of the workload prevented the team fran expanding plot sizes in any trial
type.

Generally speaking, the plot sizes were four or five rows in width by 5m
in length for maize, sorghum and beans, and 5m x 5m for rice. The only time
larger plots were used was in the spacial arrangement trials in domain two (La
Paz). In this case, each plot averaged nearly 10m x 10m. However, since the
team had little experience in spacial arrangement trials, there was no "guar-
antee" that these 10m x 10m plots were of adequate size.

Division of the FSR unit into sub-teams by domain

A fair amount of time was spent discussing how the farming systems unit
would divide personnel responsibilities between the three domains. Since
three of the four expatriates in the unit were not native speakers of Spanish,
each was initially assigned to a different domain. Then the native Spanish
speakers were assigned to regions to complement the expatriates.

However, assignment of personnel to domains was not a random process.
The individual with expertise in upland rice research was assigned to domain
four (San Jeronimo) where rice trials formed the backbone of research on
farmer's fields. The rest of team personnel chose the recommendation domain
of their own preference (subject to the language ability constraint noted
above). Unfortunately, during the early weeks of farm trial implementation,
such an assignment of personnel meant that the expatriate with the least
amount of working Spanish ended up alone in domain four. This situation was
soon remedied by adding a Honduran counterpart to that domain, but other teams
should be aware of language constraints when assigning research (or extension)
personnel to various regions of any given country.

Logistics and the division of labor

Finally, major discussions of logistics, potential labor bottlenecks and











Trial replication

The number of replications per trial and the size of each plot per trial
are related issues which received considerable attention. (Plot size is
considered in the next sub-section.) The team decided to compromise between
no replicates (or the use of farm site as the only source of replication) and
the four replicates traditionally used by agronomists in their work on experi-
ment stations.

Initially, the FSR unit hoped that three replicates could be used for
certain trials, but after all trials were designed and two-person teams had
been assigned to each domain, the group soon realized that the size of the
overall work load precluded including a third replicate for any of the trials.

Plot size

Plot size as an issue was settled in nearly the same manner as number of
replicates. In general, the team began by trying to decide what the smallest
acceptable plot size would be, by crop (or by system when more than one crop
was to be grown in the plot per year) and by trial type. Once such a minimum
plot size was agreed upon, the group considered larger plots. However, the
size of the workload prevented the team fran expanding plot sizes in any trial
type.

Generally speaking, the plot sizes were four or five rows in width by 5m
in length for maize, sorghum and beans, and 5m x 5m for rice. The only time
larger plots were used was in the spacial arrangement trials in domain two (La
Paz). In this case, each plot averaged nearly 10m x 10m. However, since the
team had little experience in spacial arrangement trials, there was no "guar-
antee" that these 10m x 10m plots were of adequate size.

Division of the FSR unit into sub-teams by domain

A fair amount of time was spent discussing how the farming systems unit
would divide personnel responsibilities between the three domains. Since
three of the four expatriates in the unit were not native speakers of Spanish,
each was initially assigned to a different domain. Then the native Spanish
speakers were assigned to regions to complement the expatriates.

However, assignment of personnel to domains was not a random process.
The individual with expertise in upland rice research was assigned to domain
four (San Jeronimo) where rice trials formed the backbone of research on
farmer's fields. The rest of team personnel chose the recommendation domain
of their own preference (subject to the language ability constraint noted
above). Unfortunately, during the early weeks of farm trial implementation,
such an assignment of personnel meant that the expatriate with the least
amount of working Spanish ended up alone in domain four. This situation was
soon remedied by adding a Honduran counterpart to that domain, but other teams
should be aware of language constraints when assigning research (or extension)
personnel to various regions of any given country.

Logistics and the division of labor

Finally, major discussions of logistics, potential labor bottlenecks and










vehicular requirements took place toward the end of the three-day design
session. It was pointed out that at trial design time, the unit had only one
assigned vehicle, but needed a minimum of three to work effectively in the
domains (FIGURE 2). These domains were separated by a distance of between 45
minutes and one hour and 15 minutes. The unit agreed to seek assistance from
the donor sponsoring one of the expatriates in the unit for continued
vehicular support until the two work vehicles on order arrived. Permission
was obtained, and this expatriate contributed his personal and work vehicle to
the early efforts of the team.

In addition, the CATIE outreach ecologist stationed in Honduras loaned
his vehicle to the unit upon his departure to CATIE headquarters in Turrialba,
Costa Rica. These two ex-oficio vehicles, plus the one Toyota Land Cruiser
seconded to the unit from the Honduran Coffee Rust Campaign, enabled the unit
to carry out its necessary field work during the first year's trials.

The issue of labor bottlenecks within the interdisciplinary unit was also
discussed. The issue involved the anticipated planting and harvesting bottle-
necks which always arise during any cropping season. It was agreed by all in
the trial planning meeting that the following policy would be adopted:
whenever a two-person team did not have top priority field work to do, and the
team in another domain needed extra hands for planting or harvesting, the team
with slack time would assist the other team in planting, monitoring, or
harvesting activities. Thus, field work took priority over both analysis and
office work.

The same general policy applied to vehicle assignments. Each two-person
team was assigned a work vehicle. Sharing of vehicles between domains occur-
red when and if the need arose.

In actual practice, these policies worked out pretty well. The only
exception was that the first domain planted received the most assistance from
other two-person teams, while the last domain planted received the least
assistance from others. However, there is no substitute for guaranteed
mobility to the field during critical stages of cropping and/or livestock
trials. Two ways to assure such mobility are:

(1) assigning a work vehicle to each domain, or

(2) physically locating sub-teams in the domains (or villages) where
their farm trials are located.










CONCLUSIONS

Anyone can design crop trials. Agronomists, breeders, plant protection
specialists, biometricians and statisticians do so all the time. However,
many factors other than technical ones must be considered during any process
of farm trial design.

Farm trials in FSR/E are designed by groups or teams with certain in-
terests and goals in common, but the team will almost never be in total
agreement. Compromise is the key to a successful team trial design effort.

The disciplinary composition and age distribution of team members are
also important factors. The political mandate of the research organization,
and of the FSR/E research team, must be considered. The institutional setting
and the interpersonal relationships between traditional researchers and the
FSR/E team researchers must also be considered. Existing research results,
directions, goals and objectives, and breeding pipelines, must all be examined
for relevant ideas, materials and thrusts. Researchers in charge of each
commodity and disciplinary program should be included in the decision-making
process whenever their disciplines or commodities will play major roles in the
farm trials designed by the FSR/E team.

In addition, the total number of important, researchable problems or
constraints will always exceed (1) the amount of available technology which
can be brought forward to address the needs, and (2) the capacity of the FSR/E
team to implement, monitor, harvest, analyze and interpret results within
these problem areas. Therefore, the trial design process must always include
a great deal of prioritization. Top farmer-identified problems, given the
greatest amount of emphasis in design of on-farm trials, may need to be
shifted.in light of existing knowledge concerning solutions to them. Other
lesser priorities may be elevated to farm research status level, while reality
dictates that some of the farmers' more pressing needs must be researched
first at the experiment station level.










REFERENCES

Chambers, Robert. 1985. "Agricultural Research for Resource-Poor
Farmers: The Farmer-First-and-Last Model." Agricultural
Administration (20): 1-30.

Conway, Gordon R. 1985a. "Agroecosystem Analysis." Agricultural
Administration (20): 31-55.

Conway, Gordon R. 1985b. "Rapid Rural Appraisal and Agroecosystem
Analysis: A Case Study from Northern Pakistan." Paper presented at
the International Conference on Rapid Rural Appraisal, Khon Kaen,
Thailand, 2-5 September.

Galt, Daniel L. 1985. "How Rapid Rural Appraisal and Other Socio-
Economic Diagnostic Techniques Fit into the Cyclic FSR/E Process."
Paper presented at the International Conference on Rapid Rural
Appraisal, Khon Kaen, Thailand, 2-5 September.

Galt, Daniel, Alvaro Diaz, Mario Contreras, Frank Peairs, Joshua Posner
and Franklin Rosales. 1982. "Farming Systems Research (FSR) in
Honduras, 1977-81: A Case Study." Department of Agricultural
Economics, Michigan State University. MSU International Development
Papers, Working Paper No. 1. East Lansing, MI 48824-1039.

Grandstaff, Terry B. and Scmluckrat W. Grandstaff. 1985. "A Conceptual
Basis for Methodological Development in Rapid Rural Appraisal."
Paper presented at the International Conference on Rapid Rural
Appraisal, Khon Kaen, Thailand, 2-5 September.

Hildebrand, Peter E. 1984. "Modified Stability Analysis of Farmer
Managed, On-Farm Trials." Agronomy Journal (76): 271-274.

Hildebrand, Peter E. and Federico Poey. 1985. On-Farm Agronomic Trials
in Farming Systems Research and Extension. Lynne Rienner Publishers,
Inc. Boulder, CO.

Jamieson, Neil. 1985. "The Paradigmatic Significance of Rapid Rural
Appraisal." Paper presented at the International Conference on Rapid
Rural Appraisal, Khon Kaen, Thailand, 2-5 September.

Rhoades, Robert E. 1982. "The Art of the Informal Agricultural Survey."
Social Science Department, Training Document 1982-2. International
Potato Center, Lima, Peru.




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

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