A DECADE OF ON-FARM RESEARCH IN LOWLAND RICE-BASED FARMING SYSTEMS: SOME LESSONS
Farming Systems Support Project
International Programs Office of Agriculture and
Institute of Food and Office of Multisectoral Development
Agricultural Sciences Bureau for Science and Technology
University of Florida Agency for International Development
Gainesville, Florida 32611 Washington, D.C. 20523
NETWORKING PAPER NO. 6
A DECADE OF ON-FARM RESEARCH IN LOWLAND
RICE-BASED FARMING SYSTEMS: SOME LESSONS
This paper was presented at the 4th
Annual Conference on Farming Systems Research
at Kansas State University, Manhattan, Kansas, October 1984
Richard A. Morris Agronomist and Head Multiple Cropping Department The International Rice Research Institute P.O. Box 933 Manila, Philippines.
Networking Papers are intended to inform colleagues about farming systems research and extension work in progress, and to facilitate the
timely distribution of information of interest to farming systems practitioners throughout the world. The series is also intended to invite response from the farming systems network to help advance FSR/E
knowledge. Comments, suggestions and differing points of view are invited by the author or authors. Names and addresses of the author or authors are given on the title page of each Networking Paper.
Networking Papers do not necessarily present the viewpoints or
opinions of the FSSP or its affiliates. Readers wishing to submit materials to be considered for inclusion in the Networking Paper series are encouraged to send typewritten, single-spaced manuscript,
ready for publication. The FSSP does not edit Networking Papers, but simply reproduces the author's work and distributes it to a targeted
audience. Distribution is determined by geographic and subject matter considerations to select a sub-group from the FSSP mailing list to receive each Networking, Paper.
A DECADE OF ON-FARM RESEARCH IN LOWLAND
RICE-BASED FARMING SYSTEMS: SOME LESSONS
In tropical Asia, rice is central to most farming
systems because it is a crop well adapted to the monsoon climate. Moreover, governments and households place high priority on rice production. Although rice cultivation must remain central to farming systems, increased cropping frequency, by growing other species in sequence with rice, is regarded as a means to increase food production and employment opportunities for rural populations.
When implemented with linkages to conventional
agricultural research programs, on-farm cropping systems research projects complement experiments station research. Innovations tested in on-farm cropping systems research projects are those perceived as having a potential to increase cropping frequency. Innovations are tested on-farm to determine the acceptability of these innovations to farmers.
Experience with on-farm research methods and agricultural innovations in rained rice farming environments has yielded lessons. Methodological and organizational aspects of on-farm research are discussed. Among these are the role of on-farm research, phases of on-farm research, and project relationships with farmers, with research station-oriented scientists who are responsible for generating appropriate technology, with extension officials, and with community leadership. Technical aspects of increased cropping frequency in lowland rained rice-based sequences are also discussed.
Although the International Rice Research Institute (IRRI) is a single commodity international agricultural research center, it recognizes that the well-being of rice farmers and regional food production are not tied only to the rice crop. That rice cultivation is central to cropping system, however, does have important implications for research on both rice and associated crops. The IRRI Cropping Systems Program has emphasized the incorporation of technical innovations into farming systems to increase frequency of crop harvests per year rather than increase yield per crop.
This paper contains four sections. The first section pulls together several disparate threads in a section entitled "Background". The second section discusses the on-farm systems research methods used by IRRI. The third summarizes agronomic research findings from three lowland rice environments. The fourth discusses issues important to on-farm cropping systems research which appear to require a few years of experience before they are recognized.
Why rice-based? During the tropical Asian wet season, few crops other than rice can be successfully grown on extensive areas of heavy alluvial, coastal marine, and lacustrine soils. Rice is a highly esteemed staple food which stores easily in the humid tropics. Governments in the region recognize the economic and political nature of rice and, therefore, support rice production by price policies, institutional credit, and input subsidies. Recognizing these factors, rice cultivation remains central to most Asian cropping systems research programs.
Why crop intensification? The focus of rice-based
cropping systems research, as applied at IRRI, arises from increased pressure on land. In Asia, land for agricultural expansion is limited, but population is expected to increase 138% before it stabilizes in 2090 (Dudal 1982). Furthermore, as incomes rise, land must produce more food per person. An expanding food and fiber demand offers opportunities to farm families to increase income by increasing output. With limited arable land but expanding population, multiple cropping (i.e., growing two or more crops on the same field in one year) is regarded as an important strategy to increase production.
Since the mid 1960s, IRRI has had a multiple cropping research component to investigate the potentials and problems of increasing harvest frequency .The on-farm
cropping systems research activities started in the early 1970s. At IRRI, on-farm cropping systems research complements experiment station research. About 25% of IRRI's cropping systems program is devoted to on-farm cropping systems research. The remainder is devoted to conventional experiment station research and to the analysis of factors that influence crop productivity and resource allocation.
Why on-farm research? or put another way: What
information can an on-farm research project generate which experiment stations cannot more efficiently generate? Answers to the following are sought from on-farm research:
o Agronomic adaptation. How fully are crop physical requirements met under farm conditions?
o Managerial feasibility. Does the farmer have
sufficient resources and skills to cope with necessary cultural practices?
o Economic viability. Can a new cropping pattern compete with current use of the land?
Valid answers to these questions cannot be obtained on research stations.
Of what use is information generated by on-farm
research? First technical innovations necessary for crop intensification that are manageable within the resource limitations of small farmers can be recognized and popularized rapidly. Second, because innovations may be widely adopted only after management skills and markets are improved and resources are augmented, specific resource constraints, weak markets, or managerial deficiencies that limit adoption are identified. This information is directed at agencies responsible for development projects and extension activities. Third, problems for which technology does not now offer a solution or which cannot be overcome by a program of resource augmentation, market development or support, or extension activities are identified. Solutions to such problems require thorough research and therefore,
information of this type is directed at agricultural research institutions.
Why involve farmers? Farmers are important components of the system. Most farmers are skillful crop husbandry men, keen observers of physical, biotic, and socioeconomic variables, and shrewd businessmen. An innovation used by a farmer will be tested under conditions in which it ultimately must be effective. Finally, farmers (and their partners and advisors in the farm-household complex) quickly exploit suitable technology and expose unsuitable technology. observational units must include farmer.
Why interdisciplinary? Although a technical innovation may-be the ultimate factor contributing to the insertion of
*The Cropping Sstems Program is one of ten research program areas at i'RRI.
another crop in an annual sequence, the incorporation of that innovation invariably has ramifications for many cropping activities. For example, substituting an early-maturing, high-yielding, photoperiod- insensitive rice variety for a late-maturing, low-yielding, traditional variety will create slack field time. But to grow a second crop to exploit that field time, it will be necessary to 'harvest, thresh, and store the first crop before the field can be prepared for the second crop. Realistic management ttechniques to secure the first crop and to plant the second crop must be identified so that valuable growing time is not ,lost. Furthermore, a newly introduced second (or third) crop must be protected and therefore crop protection research nust be initiated. Recognizing that no single discipline has the expertise required to address all potential problems, cropping systems research projects are commonly interdisciplinary.
The preceding discussion suggests that an introduction of an innovation is necessary to increase cropping intensity. Therefore on-farm methods must provide a mechanism for the syntheses and tests of cropping patterns which incorporate innovations. The methods must also yield conclusive results regarding agronomic adaptation, managerial feasibility, and economic viability. Furthermore, the methods must encourage several disciplines to focus on a common goal.
ON-FARM CROPPING SYSTEMS RESEARCH METHODS
Methods for on-farm cropping systems research outlined .at a Cropping Systems Working Group meeting in 1975 were strongly influenced by applied systems research theory. For example, Figure 1 shows the similarity between the applied systems research method described by Witz (1973) and the cropping systems research method outlined by the Cropping Systems Working Group (CSWG 1975). Proponents of systems methods (Witz 1973, Ebersohn 1976, and Wright 1971) often stress three advantages of a systems research approach: increased systems comprehension, refined research focus, and improved interdisciplinary communications. A fourth advantage, emphasized by Dillon (1976), is that systems research programs are goal-seeking, and therefore, oriented toward problem solving.
Fortmuate JPerftorwene Cr aea Control
Define 51o09 Ora
Cooperct Zcn' Ccoo
ConstrUct Vo e. bles
Models ECluoh.on Mcdel
( ythess) Dessgn A ternctee
# Croppeg S1tens
System CDeut,on Test A terc' e and Controi Systern cad
a Ol World 'ese P o
Sy tem Es.erarentol Test c Re-e
Prototye System Ad;ustm.nt P e *
PseIt t change
Fig. 1. Similarities of applied research methods described by Witz (left) and cropping systems research methods outlined by the Cropping Systems Uorking Group (right). (Witz 1973, CSUG 1975).
Four phases are followed in the on-farm research methods used by IRRI (Zandstra et al 1981).
1. Target area delineation/Site description. The target area is an extensive tract for which cropping patterns are synthesized and tested. Land in the target area is classified on soil and terrain features. Target areas are usually typified by repeating patterns of farms which operate similar enterprise mixes. The repeating patterns arise from repeating terrain patterns, and from similarities in climate and patterns of social and economic organization.
A site, usually I to 5 villages, is selected to
represent typical conditions within the larger target area. The physical, biotic, and socioeconomic features of the site area are characterized for two purposes: first, patterns are designed to fit within the physical, biotic, and socioeconomic limitations of typical farms. Second, these site characteristics are subsequently used to identify locations within the gross target area having physical, biotic, and socioeconomic conditions similar to those of the site.
2. Designing alternative improved cropping patterns/ Determining management variables to test. Cropping patterns are normally designed to fit crops into a cropping system without exceeding environmental or resource limitations. in the monsoon tropics to which rice is naturally adapted, the length of the hydrologic growing year must be considered. Although a hydrologic growing year (defined as the period of the year during which water is sufficient to produce crops with values exceeding production costs) depends primarily on rainfall, it is influenced by local terrain and soil features. Other factors that will affect agronomic feasibility (e.g., traction sources) and economic viability (e.g., product prices) are considered during the design phase.
For each crop in a pattern, many practices must be specified. Those are classified into three broad categories: crop establishment, crop care, and crop harvesting. Some practices can be specified based on local knowledge and general principles. To determine optimum management levels for practices not readily specified or to which profitability is likely to be sensitive, field experiments replicated within or across fields or both are conducted. Management factors that require special consideration are those with high costs or sensitivity to timeliness.
3. Testing on farmer-cooperators' field!
Experimentation on management variables. By testing a cropping pattern in farmers' fields over a period of 3 to 4 years, the data necessary to compare alternative patterns are generated. Cropping pattern test data are also used to examine sensitivities to terrain features and variations in seasonal rainfall or pest regimes. To determine the sensitivity of critical field operations, inputs, and planting dates, field experiments are conducted on selected management variables. At annual workshops, unadapted patterns are culled and input levels refined.
4. Multilocational testing/Pilot production testing. Multilocational testing is conducted to verify the performance of cropping patterns over the more extensive target area, and adjust input levels for environmental variation found in the target area but not adequately represented in the site. In a multilocational testing program, farms are selected on the basis of land classes established when the target area was delineated.
In pilot production testing, the organizational arrangements for full-scale production program are pre-tested on a subarea. Using this experience,
organizations and operations are adjusted before the program is expanded.
Researcher's roles/Farmer-cooperator'roles. In cooperative research with farmers, roles must be well-defined for two reasons. First, to assure smooth operations and timely management, confusion must be avoided. Second, the contribution of both groups must be understood for interpretive purposes. Where the farmer provides land, draft power, labor, and seeds, and the project provides fertilizers and pesticides, interpretation will differ from cases where the farmer provides only land. In the latter case, power, labor, and seed regeneration constraints may be completely removed by project resources.
When formulating a project, the roles of those
participating in research and the criteria used to judge technology acceptability must be established. The roles and the criteria are established during the second phase.
Agronomic feasibility criteria/Economic viability
criteria. Farmers must be able to plant within the period specified in a cropping calendar, and yields must be sufficiently high and reliable to demonstrate that the crops are agronomically adapted to farm conditions. Planting and harvest dates and crop yields are important elements in tests of agronomic feasibility.
Size and dependability of profits from an activity and profitability relative to alternative activities determine economic viability. Although the technical relationships between environment and crops may be satisfied and agronomic feasibility achieved, economic viability may not be achieved if product prices are not sufficiently high to generate
THREE RAINFED LOWLAND ENVIRONMENTS
In this section, the performance of technologies
identified as critical for the acceptance of new cropping sequences in three lowland environments is discussed. Readers not actively engaged in on-farm research in similar lowland environments may wish to skim the discussions or
proceed directly to the section on miscellaneous issues.
Figure 2 summnarizes rainfall patterns at three
Philippine sites. Table 1 gives information on soils, terrain, and farm and family size. At all sites, many cropping patterns were tested. In this paper, however, discussion is restricted to double-crop patterns [i.e., rice-rice (R-R), upland crop-rice (UC-R), and rice-upland crop (R-UC)]
CAGAYAN ILOILO PANGASINAN
50 '- ---__I
JFMAMJJASON DJF J FMAM JJASONOJ F J F MAMJ JASON OJ
Fig.' 2. Long-term mean monthly rainfall and alternative cropping patterns at three IRRI cropping systems sites.
Table I. Predominant soils, terrain features, and farm and family size at the
6 cropping systems sites.
Farm size Family size
Site Soils Terrain (ha) (no.)
Iloilo Pelluderts, Eutropepts Gently sloping (1-21) 1.5 5.7
Tropfluvents, silty broad marine terrace
clay surface soils with interhill miniplains at the boundary
with range of hills
Pangasinan Eutropepts, clay Broad. flat plains with 1.1 5.8
loam to loamy clay slight depressions and
surface soils eroded levees in an
Cagayan Tropaquepts, silty Alluvial plain, no 3.0 6.4
clay to clay surface local relief, about
The hydrologic growing year determines the potential
sequences of crops that can be grown. It is influenced by
rainfall (seasonal distributions and year-to-year
differences in distribution), soil properties (internal
drainage), and terrain features (surface drainage).
When synthesizing new cropping patterns for a rainfed rice environment, the common strategy has been to increase frequency of harvest by replacing late-maturing traditional rice cultivars with modern rice cultivars which mature early (creating slack field time) and partition photosynthate efficiently (increasing yield potential but generally requiring improved management to exploit profitably). In the syntheses, crops are forced both earlier and later in the hydrologic year, thereby increasing drought hazards.
In this discussion, economic analysis and
interpretation are excluded. The economic aspects of adoption and the impact of new cropping patterns in Iloilo and Pangasinan are discussed in Price (1982). Detailed linear programming analyses of Pangasinan and Iloilo are given by Barlow et al (1983). The economics of the cropping patterns tested in Cagayan are discussed in recent IRRI annual reports (IRRI 1983, 1984, and 1985).
Rice-Rice Patterns (R-R)
In the mid-1970s, several rice cultivars that mature in less than 120 days became available. An examination of rainfall patterns or a combination of rainfall and terrain features suggested it should be possible to grow two rice crops in the three environments represented by Iloilo, Pangasinan, and Cagayan, Philippines. Whereas shortduration varieties triggered research on double rice cropping in rainfed environments, the hydrologic growing year and farm resource constraints forced scientists to focus on methods to establish the first rice crop early and reduce the interval between first crop harvest and secondcrop planting.
Iloilo. The objective was to replace a latematuring, tall, low-yielding, traditional crop with two early-maturing crops. Wet seeding (pregerminated seed broadcast on puddled soil), a strategy to avoid the risk of not being able to transplant seedlings in the optimum age range because early rains did not persist after nursery establishment, was used to establish the first crop early. Farmers could manage this establishment technique reliably. For most farmers in most years, an accumulation of about 200 mm rainfall was sufficient for wet seeding. Early drought (vegetative phase) had little effect on first-crop yield. Weeds were controlled by puddling, high seed rates, spot hand weeding, and herbicides. Direct dry seeding (on unpuddled soil but in bunded paddies) found favor on some of the lighter-textured soils and in some
years. In dry-seeded rice, however, weeds were more difficult to control.
Despite scientists' recommendation that the second crop be transplanted to shorten its field duration and thereby avoid drought, farmers commonly wet seeded it. Data showed that farmers who wet-seeded (the large majority) established their second crops about 10 days before those who transplanted (Roxas et al 1978). Wet seeding was easier and demanded less labor because nursery construction, seedling pulling, and transplanting were eliminated.
When a late-maturing crop was harvested in the early dry season traditional foot threshing was appropriate for processing field-stacked sheaves over 1 or 2 months. Foot
threshing, however, was unsuitable when the field had to be prepared for a second crop. Power threshing quickly replaced the traditional method. Power thresher adoption is shown in Figure 3.
'76 '77 '78
Fig. 3. Thresher adoption by rainfed rice farmers in Iloilo,
Philippines, 1976-78. (IRRI 1979).
Because it was pushed into the wet-dry season
transition period, the second crop was exposed to drought
and yielded lower than the first crop. Double rice cropping
is now commonly practiced only on low-lying terrain and/or where internal drainage is slow (heavy soil textures, high
water tables). Field water observations from 1976
illustrate differences between terrain positions (Fig.4).
The significance of these water regimes to second-crop
yields is obvious from Table 2. On the lower fields, the
second crop is successful although not as high yielding or
as stable as the first crop.
4 slope Low side slope
o I if ' \
o i II" I '__ I
4 (c) Woterway
0o 1' I I 11111
19232731 353943475119 2327 31 3539434751
Fig. 4. Flooded status days (FSD) for 6 heavy-textured rainfed Iloilo land units, weeks 19 to 51, 1976. (Iorris and Zandstra 1979)
Table 2. Average yields of single- and double cropped rice for Iloilo, crop years 1976-77
and 1977-78, by landscape position.a/(Morris and Zandstra 1979).
__ Yields (t/ha)
Crop Sideslope Plateau Nain and drainageways
year Tinq e Double Tinq -ToblT inTe Double
1976-77 5.1(33) 5.0+1.7(67) b.2( 6) 5.3+2.3(94) 4.4(20) 5.5+3.4(80) 1977-78 4.6(93) 6.9+1.0( 7) 5.2(96) 5.8+0 ( 4) 4.5(31) 5.8+1.8(69)
-Figures in parentheses are percentages of observations. These were 15 observations on the side slope, 32 on the plateau, and 5 on the drainageway in 1976-77, and 30, 28, and 32 in
The main remaining constraints to double rice cropping are the time interval between the first and second crop and the field duration of wet-seeded second crops.
Pangasinan. Except in a small area located above a
shallow water table where a few farmers have acquired skills necessary to make the sequence work, double rice cropping has not been successful in Pangasinan for several reasons. Early direct dry seeding was necessary, but tillage capacity was limited and fields were prepared late, losing the advantage of early rainfall. Figure 5 shows that the research staff planted before the early rains, but farmers (except for one) waited for the rains to start before planting, and therefore harvested late.
Harvest date Rainfall (mm)
.0 Farmer- cooperator's field
20 v Researched- managed
Sep 20 *-vv -40
20 30 10 20 30 9 19 29 9 19 Apr May Jun Jul
Fig. 5. Dry seeded rice planting and harvest dates in farmer-cooperator fields and in an experiment, and daily rainfall during the planting period, 1977.( tiorris et al 1981).
Weeds Were another constraint to direct dry seeding. Chemicals used to control weeds were very sensitive to soil moisture and therefore unreliable. The labor input for hand weeding was too high (899 man-hours/ha,mean of 1977 and 1978) to be satisfied by family members (Fig.6). Moreover, the breakeven point was frequently exceeded.
Cumuafive frequency (%)
25 13 1978
400 800 1200 1600 2000 2400 Labor (mon-hrs/ha)
Fig. 6. Cumulative frequencies of hand weeding labor used by owner-operators. 1977 and 1978 (Morris et al 191).
Even when early rains were favorably distributed and direct dry seeding was successful, second-crop yields were generally low. The more delayed the first, and consequently the second plantings, the lower the second-crop yields (Table 3).
Table 3. Mean yields of direct dry-seeded rice (first crop)
followed by transplanted rice (second crop) by
period of first crop seeding, 1977 (Morris et al
First crop Rough rice yield (t/ha)
First crop Second crop
21 May and earlier 5.3 4.0
22-31 May 4.8 2.4
1 Jun and later 4.1 1.5
It is instructive to compare selected rainfall
patterns for Iloilo and Pangasinan. The parameters in Table
4 were determined from long-term daily rainfall records. Pangasinan should expect a rainy season of more than 184 days in half the years but less than 146 days in 20% of the years, suggesting little margin for lost time for the second crop. Iloilo is much more favored for a double rice crop sequence, with rainy seasons of more than 205 days in half the years and less than 171 days in 20% of the years.
Table 4. Total days available for first and second rainfaed rice crops (Morris and Zandstra 1979).
Expected Expected Days(no.)
date of date of available
Proba- 75 mm 100 mm for two
Site bililty accumulation accumulation crops
Pangasinan 0.8 11 May 4 Oct 146
Pangasinan 0.5 27 Apr 28 Oct 184
Iloilo 0.8 25 May 12 Nov 171
Iloilo 0.5 9 May 10 Dec 205
Cagayan. Figure 7 shows a cross-section of the terrain on which rice is grown in Cagayan.
Active Village -2 km
Flood plain I
Upper Middle Lower
Fig. 7. Schematic representation of landscape, cropping systems site,
To double-crop rice successfully, close adherence to a planting schedule was necessary to avoid having young (recently transplanted) or old (postheading) crops in the field during late October and early November when flood probability is highest. With present technology, double rice cropping in this environment is unreliable. When the first rice crop was wet seeded on puddled soil in the middle stratum, short-maturing rices generally could not compete with weeds. When exposed to lengthy drought, rice did not die but developed late, delaying the second crop. In the lower stratum, first crops were transplanted. Even the lower fields were often dry when seedlings were in the optimum age range. As in the middle stratum, drought often prolonged development thereby forcing the reproductive stage into a period of greatest flood probability and/or delaying the second rice crop.
Experience showed that without means for surface
drainage and where rainfall is as erratic as it is in this environment (Fig. 8), wet seeding was not a suitable technique for sowing the first crop because water could not
1980 horlal CW o es
160 140 120
20 f nflnfln LM
Ftg. 8. Uneven rainfall has contributed to the erratic performance of crops in environments similar to
Solana. About half the growing-season rainfall is derived from tropical cyclonles (Baquiran et
be controlled. When a dry period occurred after seeding, drought and weeds limited crop development; when rains were heavy, floods reduced plant density and tillering.
Second-crop yields were low because dry field
conditions frequently delayed transplanting, forcing farmers to plant old seedlings of short-maturing varieties. Farmers, however, commonly transplant traditional varieties that are more than 60 days old. Figure 9 illustrates the relative insensitivity of traditional varieties to seedling age. Nevertheless, modern varieties were often higher yielding than traditional varieties. For example, yields from farmers' fields, where traditional varieties were grown, were 0.6 and 1.6 t/ha (n=59 and 73) in 1980 and 1981. Yields from farmer cooperators' fields, where modern varieties were grown in the same years, were 1.1 and 2.7 t/ha (n=52 and 55). Modern varieties, however, have enjoyed only limited acceptance by farmers in this environment because of their
2 Wogwag fino
H. IR 46
0 IR 52
30 40 50 60 70 80 90 100 110 120 Seedling age
Fig. 9. Grain yields of modern and traditional rice varieties
expressed as ratios relative to yields from 40-day-olc seedlings, means from 3 experiments. Solana, Cagayan
(Gines et al, accepted for publication).
Rice-Upland Crop Patterns (R-UC)
Where rainfall-soil-terrain complexes do not combine to produce hydrologic years of sufficient duration and intensity to grow double rice sequences, early planted rice crops can be followed by upland crops. Where wet season onset is abrupt and fields are quickly saturated, rice is the logical crop with which to start a sequence. Wet season onsets in Iloilo and Pangasinan were rather abrupt. Objectives were to find combinations of rice cultivars and establishment techniques that produced high and stable yields, but which left adequate time for an upland crop. Furthermore a balance was sought between early and late upland crop establishment. If established too early, flooding was a hazard. If too late, moisture was often inadequate for emergence and/or the crop encountered severe late drought.
Iloilo. As in the R-R sequence, the early planted
rice crop grew well. Provided that fields were ready for planting in October should rains cease early, the UC was not strongly sensitive to the way the rice crop was established or to its duration.
The upland crop was sensitive to terrain and soil
factors. Figure 4 shows how much longer water remained on lower fields (plateaus and waterways) compared to upper fields (plateaus and upper and middle sideslopes). Lighttextured fields were less risky with respect to flooding than low and heavy-textured fields.
No tillage technique (minimum, zero, and conventional) reliably promoted seed emergence. Furthermore, conventional tillage demanded power and time, and often resulted in a very cloddy seedbed so that seeds were not in good contact with the soil. On heavy-textured soils, upland crops performed poorly but on intermediate-textured soils, growth and yields were satisfactory.
Grain legumes generally were more attractive to farmers than cereals because they can escape drought due to their early maturity. They do not require nitrogen fertilizer or suffer from nutritional drought and they command strong market prices. Despite the less-than-ideal conditions for upland crop establishment and growth, farmers now plant greater hectarages of mungbean and other crops.
Pangasinan. Transplanted early-maturing, modern varieties quickly replaced transplanted.traditional and intermediate-maturing modern varieties in Pangasinan (Fig. 10). Although direct seeding on unpuddled soils would have left the soil in an improved state for the following upland crop, farmers rejected direct seeding for the reasons presented in the R-R section.
Percent of area
[- Modern, short-maturity
varieties other than IR36
] Modern, medium-maturily
1975 1976 1977 1978
fig. 10. Changing percentage of I4anaoag site rice area planted to traditional, modern, early-maturing varieties and modern, intermediate-maturing varieties (Morris et al
Farmers preferred grain legumes over coarse grains as a crop after rice for the same reasons as in Iloilo. Before the project started, farmers in the area commonly planted upland crops after rice, but after adoption of earlymaturing varieties, the second crop was established earlier. The importance of early establishment is illustrated by Figure 11 which shows the influence of planting delays on mungbean yields and the relationship between mungbean yield and plant density.
Y=-136.5 4 177.0DI+73.5D2 + 5. 1 X
500  
0 Nov 29
0)//o 0 Dec 14
40 80 120 160 200 240
Plant density (no./lOm2)
Fig. 11. Effect of planting date on planting density and grait
yield of mungbean; X is plant density and DI and D2 are dummy variables for 16 and 29 Nov (Morris et al
As in Iloilo, no tillage technique tested (zero, minimum, or conventional) gave reliably adequate plant populations. Where rains after planting were heavy, the crop .benefited from conventional tillage. Conventional tillage, however, was seldom economically superior to zero tillage or minimum tillage.
Upland Crop-Rice Patterns (UC-R)
For environments where the wet season onset is gradual and/or soils are permeable, harvest frequency can be increased by planting upland crops before rice. For reasonably reliable yields, the expected period between the date of earliest effective rains and the date of damaging soil saturation should be 70 to 90 days. Pre-rice crops should be able to cope with occasional short periods of excess moisture.
In this type of environment, by the time the first crop is ready for harvest, fields will have been very wet and crop residues and late weed growth heavy. Therefore, rice will commonly be transplanted on puddled soils. Rice may encounter drought and/or floods depending on seasonal rainfall distributions and terrain features. Therefore, in selecting rice cultivars and management practices, the most common local hydrologic patterns should be considered.
Cagayan. Mungbean was tested as a crop to fit into a 70-day period (last half of April to end of June) in the upper stratum. Mungbean was not a new crop in the area but farmers' average yields were less than 100 kg/ha and failure was common. Table 5 shows the traditional and cooperators' yields. Variation between years was high. An improved variety was tested at a seed rate double the farmers'. Yields exceeded 1.5 t/ha on some fields in some years, but complete crop losses occurred on other fields when early rains were heavy. Although mungbean was planted on schedule, crop damage from heavy rainfall, which caused temporary flooding, was severe in all but a few cases.
Table 5. Mungbean yields from local (traditional and
cropping systems project (cooperators) technologies.Yield (kg/ha)
1980 1981 1982
Traditional 0 67 43
(42) (39) (20)
Cooperators 88 472 309
(13) (16) (10)
a/Figures in parentheses are number of observations.
Figure 12 shows weekly rainfall against the cumulative number of fields planted in 1980. In this environment, mungbean yields are commonly depressed by short periods of heavy rainfall such as experienced between 11 and 21 May 1980.
Cumulative no. of planted fields Rainfall (mm)
10 20 30 10 20 30 10
Apr May Jun
Fig. 12. Effect of rainfall on inungbean planting dates, Solana, Cagayan, 1980.
Most of the differences between yields from farmers' traditional mungbean practices and yields from practices recommended by the project were attributed to the higher seed rate and therefore to greater plant density. Farmers, however, are reluctant to plant at higher seed rates because seeds are expensive and flood risks are high.
Rice crops were scheduled for transplanting in August or early September, but the schedule could seldom be kept because soils were too dry for field operations. Direct seeding proved unreliable because water was often inadequate to sustain newly emerged seedlings and weed competition was strong when rice was suffering drought stress. Limitations
to field operations and yields of the second crop in the UC-R pattern were similar to those of the second crop in the R-R pattern.
The following issues, seldom discussed in papers on farming systems research methods, may be recognized as issues only after a few years of on-farm research.
Relationships with research station-oriented scientists
The purpose of on-farm cropping systems research should be carefully defined to dispel notions that on-farm cropping systems research competes with conventional agricultural research programs in agricultural research institutions and universities. Figure 13 suggests that on-farm cropping systems research is an interface between farmers' production environment and agricultural scientists' academic environment. In developing countries, where small-scale producers lack the means to communicate directly with the scientific community, on-farm cropping systems research should enable scientists to identify high-priority research needs more accurately and to focus resources on those needs
TECHNOLOGY DEVELOPMENT CONTINUUM
INFORMATION FEEDBACK (PROBLEMS. ENVIRONMENTAL PATTERNS)
IJ Clossicail Scientific Classicol Scientific A Choice of Technique
Method, Reductionist Method, Reductionist (e g., choice of insecticide)
or Aalytcal or AalytcalStandard Field Testing Methods
8, Synthesis of Alternative
Croptxng Systemss On- form
Systems Analysis, sytm
Convergence of Iresearch
ig. 13. The technology development continuum shows on-farm cropping systems research as an
adaptive research activity positioned between applied research and dissemination
Figure 14 illustrates the effects of a cropping systems research program on productivity increases from a conventionally oriented research institute operating in a modernizing agricultural environment. Cropping systems cientists are responsible for conducting the on-farm research necessary to incorporate innovations into improved cropping systems, and diagnosing, as precisely as possible, the causes of poor crop adaptation, managerial infeasibility, or economic incompetitiveness. Figure 14 implies that a cropping systems research program should help discipline-oriented scientists identify critical limiting factors or operations for which productivity-increasing innovations can be developed. These identifications should be more frequent than is now the case.
Introduction of innovations
riv. 14. Increased agricultural production in an agroecological zone derived from innovations developed by research institutes with and without
on-f~'r cropping systeris research programs.
Whereas research methods, materials, and standards within disciplines are firmly established and widely accepted, this is not the case with cropping systems research methods. Because the role of cropping systems research and the methods it uses are not well crystallized, research credibility continues to challenge cropping systems scientists. Conceptual diagrams, 35-imm slides of rural scenes, and testimonies go only so far. To incr -ease credibility and communication with agricultural Ficientists
who limit research activities to experiment stations, greenhouses, and laboratories, cropping systems scientists must demonstrate that their research projects have been well-formulated, data have been collected by reliable methods, results are reproducible, and logical conclusions have been drawn. A heavy burden falls on the presentation of evidence and application of logic on that evidence. Whereas statistical procedures must be applied to various components of an on-farm research project (e.g., experiments in farmers' fields and observational studies), statistical methods can seldom be used at the systems or subsystems level. At the systems or subsystems level, cropping systems scientists must rely on convergence of evidence to support conclusions. Furthermore, where negative evidence accumulates against an innovation, they must present it tactfully to the innovators.
Relationships with extension programs
Relationships with extension programs present
particular problems. First, extension agents vary in biases, technical qualifications, familiarity with research procedures, and knowledge of local agricultural practices and field conditions. They may be motivated toward contacts with large'and powerful farmers, even to the point of asking local elite to nominate potential farmer-cooperators and project employees. Therefore, it may be wiser to involve regional subject matter specialists, rather than local field extension agents, in a project. Field extension agents, however, must be consulted and informed of plans, operations, and results as they become available. Extension agents must understand that the cropping systems research project is a research project, not a competitive extension project. Advice solicited from local agents can be used if it is sound, and quietly ignored if unsound.
Another reason for de-emphasizing participation of
field extension agents in the early phases of site research is that there may be more weaknesses than strengths in the innovations being tested. The performance of an extension program, however, is evaluated on impact, often in the short run. While cropping systems research programs can maintain viability by showing research progress, including the identification of heretofore unrecognized constraints to intensification, and expanding the general knowledge base, extension agents may be tempted to extend innovations that are marginally acceptable and will have little if any impact.
By involving subject matter specialists in the cropping systems research project, linkages with extension are maintained and the not-invented-here and the impact-reputation syndromes can be avoided. If research shows that innovations merit further examination, local extension agents can be encouraged to test them across locations (multilocation testing). At this point, the agents should be in a position to look at the innovation more objectively than they might had they invested time in initial testing.
Relationships with local political leadership
Rural communities are often strongly factionalized, sometimes by political ideologies and affiliations, sometimes by personal power rivalries, and sometimes by both. The field staff must quickly recognize local factions and maintain good working relationships across factions. If field assistants, enumerators, and laborers are hired locally, care must be taken to hire compatible personnel. Hiring suspicious candidates nominated by local officials should be avoided lest someone be hired who expects to be paid for token performance.
Size of cooperating-farmer test fields
A 1000-m2 field (4 to 12% of a rice farm in most of Asia) serves as a scale model for a series of sequential cropping activities. It takes a far er with a single draft animal about 4 hours to plow 1000 m Transplanting and hand weeding have high labor demands, but these operations are often done in short periods by groups of laborers. Many other field operations (e.g., interrow cultivation, row sowing, threshing), although not as time-conjuming as plowing, require more than an hour for 1000 m Because fields are of significant size, farmers react verbally or fail to act if a suggested field operation demands too much of his resources. When fields are this large, marginal field-suitable conditions (usually sub or supraoptimal moisture for a field operation) cannot be lightly disregarded as they might when2only 50 or 100 m are cultivated. Therefore, 1000-m fields will at least yield hints about labor and power constraints and unsuitable field conditions. 2
The 1000-m field is the observational unit and what
the farmer does or does not do on that field, together with
output from the field, are observed. These observations are analyzed to determine agronomic adaptation and managerial feasibility and serve as part of the data to test economic viability.
Research site staff recruitment
Experience has shown that it is advantageous to hire young men and women from the village for positions that require practical skills and familiarity with local agriculture.
These persons, hired on a project basis for 3 to 5
years, are strongly motivated and facilitate the development of rapport between the technical staff who must be transferred to the site area and farmers. Furthermore, by hiring staff locally, absenteeism arising from extended weekend trips home or to headquarters is eliminated. The transfer of fewer people to the project site also reduces chances for violations of social norms. The cost of breaking in a new staff more than pays off in a short period.
Baquiran, V.A., H.C. Gines, D. Marra and R.A. Morris. 1983.
Performance of transplanted rice varieties under
different water regimes. Phil. Jour. of Crop
Sci. 8(3): 125-128.
Barlow, C., S. Jayasuriya and E.C. Price. 1983. Evaluating
technology for new farming systems: case studies from
Philippine rice farms. International Rice Research
Institute, Los Bafos, Philippines.
CSWG (Cropping Systems Working Group). 1975. Report of the
second cropping systems working group meeting, 3-8
November 1975. Indonesia. Multiple Cropping
Department, International Rice Research Institute, Los
Baios, Laguna, Philippines.
Dillon, J.L. 1976. The economics of systems research.
Agric. Syst. 1:5-22.
Dudal, R. 1982. Land degradation in a world perspective.
Journal of Soil and Water Conservation 37:245-249.
Ebersohn, J.P. 1976. A commentary on systems studies in
agriculture. Agric. Syst. 1:173-184.
Gines, H.C., R.G. Pernito and R.A. Morris. 1985. The
rationale of photoperiod sensitive rice cultivars in rainfed rice-based cropping systems. Phil. Jour. of
Crop Sci. (Accepted for publication).
iRRI (International Rice Research Institute). 1979. Annual
report for 1978. IRRI, Los Bafos, Philippines.
IRRI. 1983. Annual report for 1982. IRRI, Los Bahos,
IRRI. 1984. Annual report for 1983. IRRI, Los Baffos,
IRRI. 1985. Annual report for 1984. IRRI, Los Bafios,
Morris, R.A. and H.G.Zandstra. 1979. Land and climate in
relation to cropping patterns. Rainfed Lowland Rice:
Selected Papers from the 1978 International Rice
Research Conference. International Rice Research
Institute, Manila, Philippines.
Morris, R.A., H.C. Gines, and R.O. Torres. 1981. Cropping
Systems research in the Pangasinan Project. IRRI Research Paper Series No. 92. International Rice
Research Institute, Los Bafos, Philippines.
Price, E.C. 1982. Adoption and impact of new cropping
systems in Iloilo and Pangasinan, Phlippines. Pages
709-720 in Report of a workshop on cropping systems
research in Asia, 3-7 March 1980, International Rice
Research Institute, Los Baffos, Philippines.
Roxas, N.M., F.R. Bolton, R.D. Magbanua, E.C. Price. 1978.
Cropping strategies in rainfed lowland area in Iloilo.
.A paper presented for the 9th Annual Meeting of the
Crop Sci. Soc. of the Philippines, Iloilo City, May
Witz, J.A. 1973. Integration of systems science
methodology and scientific research. Agric. Sci.
Wright, A. 1971. Farming systems, models and simulations.
Pages 17-23 in J.B. Dent and J.R. Anderson (ed).
Systems analysis in agricultural management. John
Wiley, Sydney, Australia.
Zandstra, H.G., E.C. Price, J.A. Litsinger, and R.A. Morris.
1981. A methodology for on-farm cropping systems
research. International Rice Research Institute, Los