THE ENVIRONMENTAL INDEX:
A METHODOLOGY FOR ANALYSING FARM GENERATED DATA
Project Proposal Submitted by
North Florida Farming Systems Research and Extension
October 29, 1982
One of the salient characteristics of FSR/E is that of performing
on-farm experimentation that allows the farmer to directly contribute
to the evaluation and development of new technology. This process
considers the social, economic and physical constraints that farmers
face, and allows for the development of technology within these
It is known that some technologies are satisfactory for some
farmers, yet others find it undesirable, i.e. they are constrained
or:unable to adopt it for some reason. Determination of the constraints
which hinder a technology from successful adoption is central to the
FSR/E process. Once the factors that limit the adoption of a certain
technology are identified, researchers are in a better position to
alter the technology and correct any deficiencies.
In recent years,attempts at defining the constraints facing farmers
have become complicated and cumbersome and have shown only modest
results. Examples might include whole farm linear programming models
or models involving farmer decision making trees. In each case, the
attempt has been made to delineate constraining factors of production
(e.g. land, labor capital, physical environment,tastes and preferences,
etc.) for a farmer or group of farmers and thus define an environment
that may or may not be suitable for certain technologies. In effect,
such methods must detect and define all possible constraints that affect
the adoption of new technology before an attempt is made to place
specific technology in a specific recommendation domain. Determining
constraints is difficult in this manner even by highly skilled
social scientists who have an intimate knowledge of the culture
and environment in which they are operating.
An exciting method of defining farmer constraints, delineating
recommendation domains and evaluating the economic aspects of new
technology has recently come to light(Appendix I1). Based on a method
of evaluating varietal stability by plant breeders, Environmental
Indexing allows researchers to evaluate technology over a large range
of environments (as influenced by physical, social and economic factors)
without the necessity of defining constraints before-hand.
The method basically consists of comparing new and old technologies
under a wide range of on-farm environments. The performance of these
tehcnologies are then studied relative to the environments they were
subjected to. Because most technologies are developed or evolve under
specific environments (i.e. on experiment station research plots or
subsistence farmers' fields) one would expect their performance to vary
under different environments. With environmental indexing it is
possible to group farmers according to the performance of technologies
on their farms. Those farms that show similar responses to various
technologies can then be classified into specific recommendation domains-
and can be assumed to possess similar production constraints be they
physical or socioeconomic. From this potnt, determination of specific
constraints is simply a matter of observing appropriate characteristics
that are common to farmers in the same domain.
Environmental indexing has been used to analyse on-farm data
drawn from recent Farming Systems research in Malawi. The results
are encouraging and indicate that environmental indexing may be a
powerful method of analysing new technologies. However, further
experimentation and experience are needed for adequate verification
of the method.
Environmental indexing canbe successfully utilized to evaluate
on-farm trials and help to determine and define recommendation domains
and farmer constraints.
The objective of this research project is to obtain enough on-farm
research data to adequately test the environmental indexing methodology.
Thi, will be accomplished by augmenting existing on-farm research projects
of the North Florida FSR/E program.
The North Florida FSR/E has determined that wheat may be an
important new crop to farmers in North Florida. However, many questions
pertaining to wheat production practices remain unanswered to researchers
and farmers alike. It is in these areas that additional on-farm research
is being conducted by the North Florida FSR/E team and will be augmented
to test the environmental indexing methodology. Inithis way resources
used to evaluate thepotential of environmental indexing will also
supplement the on-farm research efforts of the North Florida project.
The following three research projects will be enhanced with
additional on-farm trials that,could include farms in Madison,
Hamilton, Suwannee, Columbia, and Lafayette counties.
1. Response of Wheat to Tillage Method (Appendix 2)
Originally planned to include five farms, this project
will be expanded to fifteen on-farm trials with one supporting
2. Florida 301 Wheat Response to K Fertilization (Appendix 3)
This experiment as originally planned contains 11 treatments
and will be replicated 4 times on 3 farms. With the addition
of the Environmental Indexing study, the number of farms will
be increased by 10 with one replication per farm. Collaborating
with J. Kidder, Soil Science.
3. Wheat Response to Six Fertilization Programs and Effects on
Residual Potassium Availability to soybeans. (Appendix 4).
Earlier FSR/E research plans included three on-farm tests
with six treatments replicated four times. Expansion under
the new plan would include 10 more on-farm trials, each
replicated once per farm. Collaborating with C. Hiebech,
Bruce Dehm, a M.S. candidate in FRE has agreed to move to Live Oak
to take on major responsibility in carrying out the additional on-farm
trials. He is presently on a 1/3 assistantship under the Cooperative
Agreement with USDA/OICD. During the time Mr. Dehm is working full-
time on this project, he will receive a full-time assistantship from
these same funds.
Additional funding from a Strengthening Grant will be needed
and is requested as follows:
Fertilizer and misc. supplies $ 4,000
TOTAL O.E. $ 7,000
Total funding requested $10,000
Increasing interest in the Farming Systems Research and Extension
(FSR/E) approach to technology generation, evaluation and promotion is
focusing interest in on-farm research. Described is a form of research
design and analysis that explicitly incorporates variation in farmer .
management as well as in soils and climate, to help agronomists evaluate
responses to treatments and partition farmers into recommendation domains.
Only simple pre-programmed calculators are necessary. Mean treatment
yields at each location are used as an "environmental index". Treatment
results are regressed on environmental index to determine the differential
effects of each to environment.' A frequently distribution of confidence
intervals within partitioned groups helps in final decisions for selection
of superior treatments.
Data from an unreplicated trial on 14 farms in 2 villages in.Malawi
are analyzed. Design is a 2 X 2 factorial with 2 maize (Zea mays L.)
cultivars and 2 fertilizer treatments (0 and 30 kg N/ha). Results show
that in the poorer maize environments, the local flint cultivars are superior
to an improved semi-flint composite, with or without fertilizer. The
composite yields more than the local material with or without fertilizer
in the better environments. In all cases there is a marked and significant
response to fertilizer.
RESPONSE OF WHEAT TO TILLAGE METHOD
A study (Wright, 1982) recently completed shows that moldboard
plowing significantly increased grain yield over yields obtained
from disked plots. These trials were performed where a till pan
of high resistance (over 400 psi) exists within 8" inches of the
soil surface (Wright, 1982). It is not known that the same yield
response will occur where soil resistance is less. Further, it is
unclear that moldboard plowing will be effective from an economic
viewpoint, given the low average yields that have been achieved in
the study areas.
The objective of the trial is to determine the yield response
of wheat, particularly, Florida 301, to tillage methods. The economic
benefits or loss that results from each practice will be determined.
The trials will be conducted on-farm, with a supporting on-station
trial. Each participating farmer will be asked to prepare four strips
by moldboard plowing and four strips by disking A completely randomized
design with four replications will be used.
FLORIDA 301 WHEAT RESPONSE TO K FERTILIZATION
Potassium fertilization recommendations for wheat in Florida
have been determined, based largely on work at the Quincy ARC. Soils
in the FSR/E study area are generally poorer than those farther west,
and the 60 bu/ac wheat yields achieved at Quincy may not be achievable
in the FSR/E study area. During the 1981-82 growing season, for example,
wheat yields on the 14 farms (Suwannee and Columbia Counties) monitored
by the FSR/E team averaged only 12 bu/ac.
The 1981-82 growing season was particularly poor. Nonetheless,
farmers, extensionists, and FSR/E personnel all feel that average yields
of about 30 bu/ac may be the norm in the study area.
Potassium fertilization at the levels currently recommended may
therefore be inappropriate in the study area. Enterprise records on the
14 farms show that fertilizer was a major variable cost in producing
wheat for these farmers, often the most costly single input. Economically,
lower potassium fertilizer use may be optimal. Further, plant response
to high potassium fertilizer level may be limited by other' factors such
as soil moisture and disease. The deep sands common in the study area do.
not retain moisture well. None of the farmers who kept records during
the 1981-82 growing season sprayed for disease control, and probably less
than half of all wheat growers in the study area did so.
The objective of the trial is to determine wheat response, particularly
Florida 301, to five potassium fertilizer levels. Three common soil types
in the study area, which vary in clay content in the upper 6-10" inches
of the profile, will be included.
The trials will be performed on-farm with a split plot design
with four replications Five potassium fertilizer
levels will be employed: (1) 0 lbs/ac K20, (2) 40 Ibs/ac K20, (3)
80 Ibs/ac K20, (4) 120 Ibs/ac K20, and (5) 160 Ibs/ac K20. At each
potassium level, two nitrogen and phosphorus fertilization practices will
be followed. They are: (1) 70 Ibs/ac N in two applications, one pre-
plant and one post emergence and 60 Ibs/at P205 in one pre-plant appli-
cation and (2) the particular farmer's practice. The latter nitrogen
treatment will vary from farm to farm.
Data taken will include grain yield and initial and final soil test
potassium levels. Enterprise records will be maintained. Plant response
will be correlated with applied potassium and initial soil test potassium
Collaborator: J. Kidder, Soil Science
10 sands only
3 full 4 reps
7-12 2 rep only
WHEAT RESPONSE TO SIX FERTILIZATION PROGRAMS AND EFFECTS
ON RESIDUAL POTASSIUM AVAILABILITY
Many wheat growers, especially those who double crop, do not
follow standard university recommendations when fertilizing wheat. Of
9 farmers who received soil test results prior to planting wheat in
the fall of 1981 for example, only one grower applied suggested rates
of N, P, and K. Similarly, many growers prefer to make only one, late
fertilizer application. They follow this practice because they fear
losses due to leaching of N and K applied prior to planting and because
they feel that late K application increases the residual K available
for the following crop.
The objective of this trial is to compare wheat grain yields,
particularly Florida 301 yields, under several fertilization programs
and to determine the effects of those programs on residual potassium
The trials will be conducted on-farm with a supporting trial at
the Live Oak ARC. A randomized complete block design will be used, with
four replications. The following treatments will be used:
(1) application of 35,lbs/a N, 60 Ibs/a P205 and 120 Ibs/a K20
followed by a post-emergence application of 35 Ibs/a N.
(2) a single, late (early Feb.) application of 70 Ibs/a N, 60 Ibs/a
K205 and 120 Ibs/a K20
(3) a pre-plant application of 26 Ibs/A N, 45 Ibs/A P205,
and 90 Ibs/A K20 and a post-emergence application of 26 Ibs/A N.
(4) a single, late application of 52 Ibs/A N, 45 Ibs/A P205,
and 90 Ibs/A K20.
(5) a pre-plant application of 15 lbs/A N, 30 Ibs/A P205, and
60 Ibs/A K20 followed by a post-emergence application of 20 Ibs/A N.
(6) a single, late, application of 35 Ibs/A N, 30 Ibs/A P205,
and 60 Ibs/A K20.
Data collected will include wheat grain yield. Soil potassium
levels will be determined at harvest. Providing that there are
significant differences between residual potassium levels between
treatments, the same area will then be used as a trial involving
soybean response to residual and applied potassium.