IRRI Saturday Seminar
March 18, 1978
The Dynamics of Water Regimes in Iloilo and Pangasinan Land Systems
R.A. Morris, R.D. Magbanua, H.C. Gines, and R.L. Tinsley*
Cropping systems research methods prescribe a describe-designtest sequence to be used in site-related studies to develop improved rice-based cropping patterns for defined environments (Zandstra, 1976). The method assumes a set of agronomic and economic performance criteria established prior to the testing phase. At different research steps, agronomic and economic criteria can be used separately, but in the final evaluation of a pattern, both economic and agronomic performance should be considered. The cropping pattern's domain of adaptation should be specified simultaneously for obvious reasons.
In application, however, it is futile to carry pattern evaluation beyond gross agronomic considerations if it is apparent that an attempt has been made to grow a pattern outside its domain of adaptation. However, the domain of adaptation is generally not fully recognized and must be investigated. Thus, studies to determine pattern performance and to specify pattern adaptation domains will usually proceed concurrently, as they have in the IRRI-BPI Outreach sites in Pangasinan and Iloilo. In the first year of research at these sites, researchers began to hypothesize differences in landrelated factors which affected pattern performance (IRRI, 1975; IRRI, 1976). This seminar paper examines the hypotheses implicit in the early division of the Pangasinan and Iloilo sites into complexes based on land-related factors. To examine the hypotheses, the behavior of flooding regimes and the performance of cropping patterns are examined as logical deductions from the implied hypotheses.
Taxonomically, most soils at the Pangasinan site fall within the Eutropept Great Group. The soil pH ranges between 6.7 and 7.9, with a raw mean of 7.4. Clay loam is the modal surface soil texture. Using a landform classification proposed by Desaunettes (1977), the
* Agronomist; Site Coordinator, Iloilo Outreach Site; Site Coordinator, Pangasinan Outreach Site; former Visiting Scientist, CrQpping Sygtems-rogram-Thellnternational Rice Research Institute, P.O. Box 933, Manila, Philippines.
Pangsinan land systemI/ is composed of flat plains of a river terrace sub-system (P.3.1) and levees (A.2.6) of an alluvial sub-system. The rainfall distribution at the site normally contains 3-4 months with total rainfall exceeding 200 mm (including one month during which rainfall exceeds 500 mm) and 5-6 months with less than 100 mm total rainfall.
Cropping systems researchers have divided the Pangasinan site area into several complexes (IRRI, 1976; Herrera et al., 1975; Gines et al., 1977). A major division between complexes has been based on deep versus shallow water table depth. Further divisions have been based on differences in soil texture, water source and relative elevation. Relative elevation simply refers to difference between fields located in waterways (i.e., broad natural occurring drainage ways) and in non-water ways. To a casual observer, the landscape appears rather level, grading away slightly from river levees to occasional deeply dissected creeks into which waterways drain.
Although a hypothesis that the complexes affect cropping pattern performance has not been formally stated previously, it is apparent that the complexes have been perceived to constitute different domains of pattern adaptation, as indicated by the.distribution of designed patterns. As logical deductions from this implied hypothesis, differences in complex characteristics such as flooding regimes, changes from designed to implemented patterns and differences in pattern crop performance would be expected to vary according to complex. If the characteristics are found to behave in accordance with the deductions, the implied hypothesis would be supported, For the evaluation, daily flooded status (FS), crop yields, and changes from designed to implemented patterns, were examined, using data collected in CY 76/77. Alos, the effect of water table depth (deep and shallow), relative elevation (waterway and non-waterway) and water source class (rainfed, partially irrigated, free-flowing well) have been examined over the season to better understand the role of each factor.
Flooding regimes. As an indication of water availability for lowland rice, a stress day (SD) concept has been effectively used by IRRI scientists in the past (IRRI, 1976; IRRI, 1972), In the analysis presented here, a flooded status (FS) concept has been used which is basically the reverse of theiSD-eoucept, but with two important differences. An FS is taken as any day on which there is standing
-!/Although used rather loosely here, the term land system is
meant to denote a naturally recurring pattern of land units, A land unit is an area which is homogenous with respect to land form, soil characteristics and hydrological features, Land use, agronomic practices and crop performance would be relatively uniform with aland unit.
water on the field, whereas SD is taken as any day beyond three successive days during which there is no standing water on the field. The other difference lies in the application of the two concepts. For this analysis, FS have been counted over approximately a sevenmonth period, whether or not a crop was in the field, while SD have been counted over a period when a crop was actually grown. Thus in this analysis, the FS concept has been used to indicate the duration over which standing water has been kept on the field. Such characteristics as the rapidity of FS increase at the beginning of the season, the reliability of FS over the season and the rate of FS decline at the end of the season are important factors which affect pattern performance and thereby influence the domain of pattern adaptation.
In the analysis, 13 separate models were computed, using the
number of FS in a 3-week interval as dependent variables and the same set of land-related classification facrors ad independent variables. The dependent variables were obtained sumrming the number of observed FS in weeks 1 to 3, in weeks 3 to 5, in weeks 5 to 7, and so forth starting with May 26, 1976. The one-week overlap was incorporated to reduce the effect of farmer's field operations, such as drainage prior to harvesting, on the number of FS observed.
The land-related factors which have been found to influence FS regimes are water table depth class, water source class within water table depth class, and relative elevation within the water table depth and water source classes. Figure 1 presents estimated FS frequencies for 9 land units for a 27-week period. Smooth curves have been drawn through the 13 estimates. An inspection of Figure 1 indicates water table depth was the major factor differentiating land ufiits; the factor divides the area into two sectors: deep water table (Lipit Pao) and-shallow water table (Caaringayan),
The major differvnces.between shallow and deep water table land units, as characterized by the FS regimes, were (1) an early increase in FS, and (2) a broader interval of peak FS frequencies in the shallow compared to deep units. Much less rainfall is required to recharge the subsoil and the saturated soll sub-strata retards deep percolation losses in the shallow units. By comparison, it is apparent that substantial early rainfall was used to recharge the deep water table units. The soils in the deep water table sector are very permeable, and rainfall rapidly infiltrates and percolates to deep soil layers. It is also apparent that the local irrigation system had no significant impact on FS regimes, either early or late in the wet season. Mo-reover,.the main irrigation canal runs along a river levee before its water enters the fields. Therefore, fields close the irrigation canal which are generally partially irrigated, tend to have deeper water tables and to be faster percolating than the lowerlying fields further from -the canal. Furthermore, the lower fields also receive run-on from higher fields, These two factors help to explain the slightly higher FS frequencies on rainfed fields in comparison to partially irrigated fields,
Fields located within relative depressions or waterways did not have greatly modified FS regimes as had been postulated, although lower units under rainfed conditions had slightly prolonged FS regimes as rainfall declined.
By contrast, within the shallow water table sector, water source differences were clearly expressed. The major irrigation difference arose between land units near free-flowing well and rainfed or partially irrigated units. However, the partially irrigated units, in comparison to rainfed units, had higher FS frequencies toward the end of the wet season. A comparison between the non-waterway land unit FS regimes indicates that the lower positions had higher FS frequencies, especially under rainfed conditions situations as was noted in deep water table fields. In partially irrigated fields the FS regimes were more favorable as the dry season was entered, although the frequencies were far from ideal.
Soil texture has also been postulated as a factor modifying FS regimes. However, easily models indicated that texture was less important overall than relative elevation differences. Therefore, soil texture was abandoned as an explanatory variable so that model degrees of freedom would be reduced. However, composite FS estimates from an early model which did include the textural class of the second soil horizon as an independent variable, are presented in Figures 2 and 3 for the two land units having the greatest number of observations. The estimates in Figure 2 are from rainfed, non-waterway deep water table land units, and in Figure.3, from partially irrigated, non-waterway shallow water table land units. In both figures, lighter textured classes were combined to form Set II, Whereas the heavier textured classes made up Set I. A comparison of the estimates indicates that soil texture did alter the regimes as expected, i.e., greater average FS frequencies over the season for the heavier soils. However, the major difference between heavy and light texture in the deep water table units occurred during the rise to maximum FS frequency, whereas the effect in the shallow land units was more general over the periodanalyzed.
Several model statistics are presented in Table 1 for the 13 Pangasinan models. R2 started at a high level and decreased with minor variation through Models 4 to 8, which correspond to the period of maximum FS frequencies. R2 again increased over the last
6 models. Except in Model 4, intercept values steadily increased to a maximum in Model 8. Total SS decreased to Model 8, than increased. Model 4, which has a lower than expected R2, corresponds to the reduction of FS frequencies at week 8 which was preceded by an interval of low rainfall. The general behavior of Total SS, Model SS and intercept values reflect the iiifluence of rainfall distribution on FS, especially after land units have been recharged, The extra SS suggest that water table depth was the dominant factor explaining the differences between land units, although Its importance
was greatest during the first three-quarters of the period examined. Water source class played a moderate role in the early weeks, minor in the middle weeks and major in the last weeks, as indicated by Models 1-4, Models 5-8 and Models 9-13, respctively. However, the impact of water source was important only in the shallow water table sector. Relative elevation played a moderate role as rainfall declined, as indicated by Models 7 to 10. Because of the model structure and the distribution of the data points, conclusions about the dynamic aspects over the season as suggested by the extra SS must be considered tentative although they behave as expected.
Changes in patterns. Table 2 shows pattern group distribution according to water table depth and water source (rainfed vs. partial irrigation). The distributions are for cropping pattern groups as designed and as implemented. As the CY 76/77 season commenced and the factors which control pattern adaptation became more obvious, there was a shift to more R-UC and less GC-R-UC and R-R-UC patterns in the deep water table sector; in the shallow water table sector, the shift was towards more intensive patterns. These pattern.shifts reflect realizations that water would be a less limiting factor in the shallow water table sector. Under rainfed conditions, across both water table classes, fields were shifted out of R-R-UC and into R-UC patterns. Initially, 46% of the R-R-UC patterns were designated for rainfed fields, but upon implementation only 24% were grown under rainfed conditions. For the R-UC group the shift was reversed; 41% were designed for rainfed conditions whereas 52% were implemented. Although the changes in pattern distribution are not overwhelming, they are in agreement with expectations. There was a shift toward more intensive patterns in the shallow water table sector and in partially irrigated units. An evaluation of the performance of the implemented pattern groups is presented in the next session,
Pattern performance. Because soil texture and relative elevation effects were of minor importance, pattern performance has been summarized only by water table depth and water source factors (Table 3). A comparison of average rice yields in the different rainfed and partially irrigated land units showed that yields of patterns containing single rice crops (R-UC and CC-R-UC) were not greatly different between land units. The rice component of both these patterns coincides with maximum FS intervals as depicted in FS frequency diagrams for Pangasinan. The rice yields for single rice pattern groups grown in the deep water table units averaged 3.6 and 3.2 t/ha for rainfed and partially irrigated units, respectively, In the shallow water table sector, yield averages for the same crops were 3.1 and 3,4 t/ha for the rainfed and partially irrigated units, respectively,
Between the deep and shallow sectors, there were no distinct differences in the yields of first rice crops in R-R-UC patterns, Within the deep water table sector, however, there was an advantage for partially irrigated over rainfed units, but there were limited observations in both cases. The first rice crop yield from R-R-UC patterns averaged 1.4 t/ha higher than rice yields of the later
planted single rice patterns, reflecting in part, a greater build-up of rice pests and late drought stress. The second rice crop in R-R-UC patterns was affected by late stress, yielding an average of 3.0 t/ha in rainfed and partially irrigated units of the deep water table sector and 3.6 t/ha in the partially irrigated shallow water table units. No R-R-UC were attempted on rainfed shallow water table land units.
In fields near free-flowing wells, the average yields for the first and second crops were 4.7 and 4.0 t/ha, respectively. One cooperator failed to complete the two crops in.time to start the third crop so it would be completed prior to the following wet season, and three farmers planted the thitd crop but experienced failures because wells ran dry towards the end of the dry season. The ;.average third crop yield of the three farmers who grew the full R-R-R pattern was
Green corn failures were common because of the early typhoon. All but one of the green corn crops in the shallow water table area were total failures. Only on the partially irrigated units near the canal, where internal and surface drainage is adequate, did the GC component approach satisfactory performance on more than half of the units tested. On this land unit, four of seven cooperators grew crops averaging 19,600 marketable ears and the remaining three cooperators experienced total failures. Five of six cooperators on the rainfed units experienced failures.
There appeared to be a slight UC yield advantage for deep water table land units over pattern group as indicated by average relative yields in Table 3. However, 9 of the 20 farmer-cooperators growing two rice crops, did not attempt the UC crop because of dry conditions prevailing at the harvest of the second rice crop, whereas all farmercooperators did plant UC after single rice crops,
In support of the division of the site into different units, pattern performance data from implemented patterns indicate that early season drainage characteristics influence the suitability of land units for green corn cultivation and the presence of partial irrigation leads to higher second rice crop yields. However, there appears to be no advantage of a shallow table over a deep water table relative to the performance of rice as a first crop in a double rice pattern or in a single rice pattern, disregarding performance in free-flowing well areas, and there appears to be lower UC yield performance in the shallow water table sector.
Taxonomically, the predominant soils in the Iloilo site fall
within the Pelludert, Eutropept, and Tropfluvent Great Groups. Soil surface materials are generally moderately to slightly acid clay or silty clay materials. Although the site area is primarily rainfed, in CY 76/77 an increased number of land units received partial irrigation of varying durations beyond the normal end of the wet season. In the upper section of the site, the land system is composed primarily of ramified inter-hill miniplains and lower colluiial slopes of a alluvio-colluvial sub-system (A.3.3 and A.3.7) using Desaunettes' classification (1977). The land system in the lower section of the site consists primarily of flat plains in a river terrace subsystem (P.3.1). The rainfall distribution at the site normally consists of 5-6 months with total rainfall exceeding 200 mm and 3-4 months with less than 100 mm total rainfall.
A statement of the expected effect of landscape and soil factors on pattern adaptation have been clearly expressed for the Iloilo site (IRRI, 1976; Magbanua et al., 1977). The effects of landscape, soil texture, and water sources on flooding regimes, and the logical extensions of differences in water regimes on pattern performance are examined in the following sections, The procedures used are the same as those applied to the Pangasinan data, but the period of Iloilo data collection was slightly longer and the factors examined are, of course, appropriate to the Iloilo land system.
Although rainfall must have a major influence on FS regimes, Figures 4 and 5 show that regimes were modified by landscape, soil texture2/ and water source class within landscape. Bund position or relative elevation within a landscape class has been found to be important only for sideslopes, Estimated FS frequencies per week have been plotted in Figure 4 for six landscape positions, which have heavy-textured soils and are rainfed and for plateau units which have heavy-textured soils but are under different irrigation durations. Smooth curves have been drawn through the 15 estimates. For the rainfed land units, a trend towards decreased FS frequency, when progressing from waterways to high sideslopes is apparent. The major differences between units occurs after the thirty-seventh week, reflecting the greater decline in frequency for the higher landscape positions. There is little difference between the medium
2/ In this analysis, soil texture refers to the texture of the second horizon, usually between 12 to 30 cm deep, This horizon was used rather than the surface because it was thought to have more of an effect on the stablishment of a perched watertable, Texture classes were heavy (c, sic, sc), medium (cl, sicl), and light (scl, 1, sil, Is, sl).
and high sideslope FS regimes, and between the plateau and low sideslopes FS regimes.
For the Iloilo rainfed units, all FS estimates began at high frequencies because of the heavy rainfall which preceded the start of FS data recording. In the lower landscape positions, the extension of FS beyond the end of the peak rainy periods results from late surface and base flow from higher elevations.
The influence of irrigation on FS frequencies for heavy-textured plateau units is also apparent in Figure 4. Irrigation greatly increased the frequency of FS after the thirty-eight week as is evident by a comparison between rainfed and irrigated heavy-textured plateau units. In Figure 4, Irrigation I refers to units receiving partial irrigation two months or less beyond the end of the wet season, whereas Irrigation II refers to units receiving partial irrigation more than two months beyond the end of the wet season. Differences between irrigation classes would become more distinct if data from a longer period had been analyzed.
Composite FS frequency estimates from four rainfed fields
located on plateaus having medium or light-textured soils and four fields having light-textured soils plus coarse (sl or ls) sub-soil strata of at least 10 cm thickness are presented in Figure 5. Figure 5 is interpreted as follows: (1) A medium- or light-textured second soil horizon hampers the formation of a perched water table, which by contrast causes a rapid and early increase in FS on land units with heavy-textured second horizons; (2) For a land unit with a coarse substrata, rapid and deep percolation is a major factor causing lower-than-average FS frequencies, even when sufficient rainfall to recharge coarse substrata has accumulated and has continued to fall frequently enough to maintain near maximum FS frequencies on land units with medium- and light-textured soils over heavy material;
(3) Although irri.ation water increases the FS frequency on a coarse substrata land unit, irrigation water is also subject to high seepage and percolation losses.
Several model statistics are presented for the 15 Iloilo FS models in Table 4. Inspection of the table indicates the dynamic role of factors over the period analyzed. A decreasing trend in total SS occurred up to the sixth model; following the7.eight model, it increased rapidly. As indicated by the R2, less variation has generally been explained by the first ten models, than by the remaining four models. The model intercept values started at moderate values, increased to high values towards the middle of the series, and then rapidly dropped to negative values. The changes in Total SS, R2, and intercept values are a reflection of the dominance of rainfall at the peak of the wet season.
The extra SS for model variables suggest that the role of landscape position was moderate over all intervals, The relative elevation of sideslope positions were moderately important during
a few dry intervals toward the middle of the wet season and again at the end of the wet season. Water source played a moderate r3le up to the eighth model and from then onward played the dominant role in explaining the differences in FS observations among the land units.
*The influence of a lanscape X texture interaction was moderate in a few models but is notthought to be important. The same cautions applied to the interpretation of extra SS in the Pangasinan models apply to the Iloilo models also.
Changes in patterns. Changes between implemented patterns
versus designed patterns occurred in Iloilo as in Pangasinan. Early and heavy rainfall from a typhoon triggered the bulk of the changes. Many green corn crops proposed to precede lowland rice were eliminated because of the early wet conditions, the method of establishment shifted heavily to WSR because seedlings were not available for transplanting or soils were too wet for DSR, and there was a shift towards double rice crops, mainly in rainfed and partially irrigated plateau and plain units. These observed changes although strongly induced by very early rainfall, are in agreement with results obtained from flooding regime analysis.
Pattern performance. Because of the similarities found in FS regimes and in yields, cropping pattern performance data were averaged by pattern group within the following heavy-textured land units:
1. High and medium sideslopes, rainfed
2. Low sideslopes and plateaus, rainfed
3. Plains and waterways, rainfed
4. Plateau and plains, irrigation I
5. Plateau, irrigation II and III.
Mean crop yields by pattern groups grown only on these heavy-textured units are presented in Table 5. Yield data from a group of mediumand light-textured soils, some with coarse-textured substrata and partialirrigation, have been interpreted separately.
Between land units there was very little difference in the yield of the first rice crop, regardless of landscape position, water source, or pattern group. However, yields of the second rice crop were much more influenced by water availability, as governed either by landscape position or water source. Upland crops generally suffered drought stress following two rice crops, although in areas covered by irrigation II and III systems, lateral seepage caused poor performance of many upland crops planted at a time when adjacent fields were maintained in a flooded state for the rice crop.
Within heavy-textured, rainfed low sideslopes and plateau units, most farmers were not able to successfully grow the full R-R-UC pattern, i.e., 39% did not grow the UC portion and of the 14 that did attempt to grow an UC crop, six experienced complete crop failures, resulting in ,a relative yield average of only 13% for the pattern group. Furthermore, the second rice crop in these units was very
sensitive to moisture regimes as affected by planting date and slight positional differences within land units. Of the 23 second rice crops attempted, five were total failures, whereas the top five yielders averaged 4.0 t/ha. The nine cooperators who grew only R-R patterns were eighter delayed in their crop plantings or located on less favorable fields within the land units, and therefore the second rice crop yields were only 0.8 t/ha) compared to the 2.8 t/ha of the 14 farmers who attempted the UC crop.
In the plain and waterway land units, only R-R-UC patterns were attempted. However, one farmer-cooperator did not grow the UC portion while of the five who did, four experienced total UC failure because of drought and the remaining individual harvested a mungbean crop of only 79 kg/ha.
In heavy-textured plateau and plain units receiving irrigation for a duration of less than two months beyond the end of the normal end of the wet season (Irrigation I), farmers easily grew the R-R portion of a R-R-UC pattern. However, four of the eleven farmercooperators did not plant a UC crop and furthermore, performance of the UC crop was generally poor for the seven cooperators who did. For heavy-textured plateau positions receiving irrigation for more than two months beyond the normal end of the wet season (Irrigation II and III), farmers easily grew at least two rice crops, Where attempted, yields were reduced on the third rice crop of R-R-R patterns, but they were still substantial at 3.3 t/ha. The R-R-R patterns were located primarily in Irrigation III units. Upland crops performed poorly in Irrigation II units, partly because of seepage from adjacent fields which farmers were trying to keep flooded to finish late rice crops.
Land units with medium- and light-textured soils, as expected were generally poorer performers than the heavy textured units, especially with regards to either the second rice crop or upland crops. The average yield of the first rice crop over all mediumand light-textured units was 4.9 t/ha, which is 0.5 t/ha less than the weighted average over all heavy-textured rainfed units. Further-more, UC yields were very much lower, after either one or two rice crops. Where partial irrigation existed, farmers were capable of producing an adequate second rice crop with yields sufficient to make a R-R pattern more profitable than a R-UC pattern.
:Based on Pangasinan CY 76/77 cropping pattern monitoring data, water table depth and water source factors were found to be important factors altering FS regimes whereas relative elevation and soil texture differences induced only minor modifications on FS regimes. Observed changes in the distribution of patterns over the land units
and the performance of the crops in the pattern groups were in general agreement with a division of the site into complexes based on water table depth and water source class, especially considering the start and end of the wet season. Using Iloilo CY 76/77 pattern monitoring data, landscape, water source and soil texture were found to be important factors altering FS regimes. Differences found in the performance crops within pattern groups over the land units were in general agreement with the division of the site into land units based on landscape position and water source class;. However, differences between the high and medium sideslopes and between low sideslopes and plateaus were not major, as indicated by FS regimes analysis and crop performance.
Similar analyses of cropping pattern monitoring data which is
being collected for GY 77/78 and is to be collected in CY 78/79, should lead to a clearer understanding of the dynamics of water regimes, especially under different rainfall patterns, and thereby aid in designing and testing improved patterns for other locations.
It is apparent that there are identifiable land unit differences within both the Pangasinan and Iloilo sites, and these differences influence at least the agronomicS/ domain of pattern adaptation. The analyses show that the landscape, soil texture, water source and water table depth factors play roles in modifying FS regimes, and that the roles change as the season progresses. A classification of land units, employing a land system concept, should aid in the extrapolation of information from the immediate site area to a more general region. A better understanding of the factors involved in FS regime dynamics within a land system is important because it provides an improved basis for defining the domain of rice-based cropping patterns. Furthermore, documentation of the regimes on different land units, and the roles played by factors at different stages of the season, form an important basis for convincintly communicating site-related research results to others.
Undoubtedly however, there are land-related and other factors that modify FS regimes and which are important but which have been either ignored or incorrectly characterized in the FS regime analysis presented inthis seminar paper. The impact of the watershed above the land units and the effect offfield level water management on FS regimes are undoubtedly important consideration.
-/By involving the farmer in cropping pattern research, a farmerresource factor has a bearing on pattern performance. This factor is operating to an unknown degree in these evaluations, and therefore the agronomic potentials as discussed here are not the same agronomic potentials which would arise if all farmer-resource constraints were removed,
Table 1. Pangasinan FS model statistics, Models 1 to 13, CY 76/77.
PANGASINAN M 0 D E L N U M B E R
DF 1 2 3 4 5 6
Total SS 70 2503 3624 1961 1685 2442 2297
Model SS 8 1917 2466 934 563 1371 834
WT 1 1207 (63.0) 2121 (86.0) 546 (58.5) 227 (40.3) 1133 (82.6) 563 (67.5)
WS/WT 3 579 (30.2) 337 (13.7) 366 (39.2) 308 (54.7) 104 (7.6) 130 (15.6)
RELEV/WS/WT 4 131 (6.8) 8 (0.3) 21 (2.2) 28 (5.0) 134 (9.8) 142 (17.0)
Error SS 62 587 158 1027 1122 1071 1462
R2 0.77 0.68 0.48 0.33 0.56 0.36
Probability level* 0.0001 0.0001 0.0001 0.0009 0.0001 0.0003
Intercept 0.9 0.9 3.6 2.3 4.6 9.8
7 8 9 10 11 12 13
Total SS 1503 913 2609 4343 3455 2995 2310
Model SS 630 316 1165 1781 1589 1656 1256
WT 425 (67.5) 136 (43.0) 451 (38.7) 470 (26.4) 58 (3.7) 482 (29.1) 357 (28.4)
WS/WT 60 (9.5) 54 (17.1) 378 (32.4) 838 (47.0) 1339 (84.3) 1084 (65.5) 763 (60.7)
RELEV/WS/WT 145 (23.0) 126 (40.0) 336 (28.8) 474 (26.6) 192 (12.1) 89 (5.4) 135 (10.7)
Error SS 873 598 1444 2562 1866 1339 1054
R2 0.42 0.35 0.45 0.41 0.46 0.55 0.54
Probability level 0.0001 0.0006 0.0001 0.0001 0.0061 0.0001 0.0001
Intercept 12.4 20.6 17.8 11.9 2.4 0.0 0.0
( ) % of Model SS explained by the extra SS due to inclusion of the variable in the model.
* Probability for the F value from the test of Model SS
Table 2. Distribution of designed and implemented patterns, by pattern group, water
source class and water table depth. Pangasinan CY 76-77.
PATTERN GROUP l/ Designed Implemented
Water source class Water source class
Total RF PI Total RF PI
Deep water table 15 8 7 11 6 5
Shallow water table 8 5 3 9 7 2
Total 23 13 10 20 13 7
Deep water table 7 4 3 17 10 7
Shallow water table 15 5 10 12 5 7
Total 22 9 13 29 15 14
Deep water table 12 9 3 6 5 1
Shallow water table 12 2 10 15 0 15
Total 24 11 13 21 5 16
-/A fourth group, R-R-R, was designed also. Seven R-R-R plots were designed for Caaringayan (all near free-flowing wells) and none for Lipit-Pao. Six of the seven were implemented.
Table 3. Mean yields of crops in the three pattern groups according to land units, Pangasinan, CY 76/77.
No.a/ Yield/ No. not-/ No. Yield No. not No. Yield No. not
Land unit Obsn. GC R UC planting UC Obsn. R R UC planting UC Obsn. R UC planting UC
Rainfed 6 2.1 3.4 47% 0 5 4.7 2.8 54% 2 11 3.7 39% 0
irrigated 7 11.2 3.6 41% 0 2 5.2 3.5 38% 0 8 2.8 56% 0
Rainfed 9 0.3 3.1 34% 0 N o n e 4 3.0 26% 0
irrigated 2 0 4.6 0 13 4.7 3.6 40% 7 7 3.0 46% 0
I I I
SNumber of observations in pattern group.
./As t/ha for rice; thousand green ears for corn; and relative yield for UC, where yields of 1.53, 1.29 and 4.50 t/ha mungbeans, cowpeas, sorghum were used as the basis for the relative yield computation. These were the maximum yields obtained by farmer-cooperators.
-Number of cooperators not planting UC GC = green corn
R = rice
UC = upland crops
*SS TOPOK JO 4X93 aqi uioag anTUA-a gql 103 AITTTquqoaj
'TaPOul 9lq3 ul OTqvTaVA aqj jo UoTsnTouT ol anp SS vj2xa aqq Kq PaUTvTdxa SS TaPOm jo % aLu
Z*C- C "7- 9 17- 6*E- S*T- Z *1 '7 T*LT VOZ :Idaz)aa:luj
TOOO*O TOOO'O Tooolo Tooolo TOOO*O TOOO'O 17 *0 Z00*0 T9AaT A3TTTqvqoaj
09*0 OL'O 8L*O LL*O 0910 zs; 0 EVO E *o
086T ZOTZ CL6T 99LT flz9z 06f7Z 9f79T 9TC SS Joaas
WO) 9Z (8*0 06T (9*0) 9C (TIO) 17 WO) 9 WO T f7 T WIM L9 WT) C ZIMUDS(imli
WT) LE (9*0) 6Z WT) 98 (9*T) Z6 (Z*Z) 98 (Z*T) CC WO 9z (9*0) Z. zjlx2ljl
(T'Z6)C89Z (8*98) SLZV (Z*V8)EgLg (C gg)V'76V (9*SL)VTTE (T*TL)968T Wo) 86Z Wsz) 09 doscINII/sod(imq/sm
(0*0) C (Z*Z) 80T Wl)-88 07*0 96T WO) 6ZC U-0 9OZ (T-ZT) LS WT) C aDS(INI/Soj(lMq
WT) OV (6*9) s6z (T*OT)Z69 (8 17) TH (9'9) T9Z WS) SZZ (T*Z) OT (0*0 L ciascimli
VT6Z C86'7, CC89 Mg C96E L99Z OL'7 M SS T9POW
176817 980L 9088 89S;L 9999 95;Tg 91oz Z95; SS TVIOI
st i7T CT ZT TT OT 9
L*6T VST T*ST VOT UTT L'CT T*CT :Idaojaiuj
TOOO*O TOO*O zo*o co*o 900*0 '7000*0 s;oo 0 *TaA9T A2TTTquqoja
OVO i7V 0 LUO gc*o OV 0 917*0 V7 0 z Id
6LZ zoz LLL 65;OT OLOT. L9ZT LSCZ ZL SS aoaaa
WO OZ WgT) '7Z (O*OT) 9 V (9'i7Z) OVT WZZ) 9T (C 17) LV (9*T) 9Z C NMI x CIDS(IMII
(9*0 CZ WO L (8*TT) g WTO T8T (C*9T) STT (8*9Z) V6Z (9 6Z) 9817 z Mail
(S*LT)VTT WLT) LZ (9*9T) LL WTT) 99 WST) TET (T*LZ) L6Z (Z'LT)OOZ OT dos(imlilsoacNqlsm
(L*T) TT WO V WO Be (CqT) C6 WOO T9T WO 9C WO) ZT z aos(INIlsod(INg
(9*6) -19 WgT) 17 Z WZZ) '701 (9 '7T) CS, 07 8 T) CET ('7*TT) M (O'L) 60T C CIDS(INII
Of79 9 ;T 89V 699 CZL L60T MT TZ SS TOPON
6T6 9 s;c S;EZT 8Z9T C 6L I '79EZ ZTOV C6 SS Tv301
L 9 9 17 C z T a([
U a q K fl N I a (1 0 K olliolli
LL19L AD 'ST 03 T sTOPOR 'S3T3ST3Vjs TaPOm Sa OTTOTI 'V 9Tqul
Table 5. Mean yields of crops in three pattern groups according to land units, Iloilo CY 76/77.
Yieldb/ No. not S/ No. Yield No. not No. Yield
Land unit Obsn. R UC planting UC Obsn. R R -UC planting UC Obsn. R R R
High & medium
rainfed 6 5.4 64% 05 5.2 1.3 21% 1 N o n e
rainfed 2 5.2 57% 0 23 5.3 2.0 13% 9N o n e
Plain & waterways,
rainfed N o n e 6 6.1 4.7 1% 1N o n e
irrigation I 2 4.5 64% 0 i 5.7 4.6 13% 4N o n e
Plateau, irrigation II &
N o n e 9 5.1 4.6 2% 1 7 5.4 4.7 3.3
a/Number of observations in pattern group.
k/Yield of upland crops is on a relative yield basis, with CP = 1.72 t/ha = 100; S = 8.33 t/ha
= 100; M = 1.20 t/ha 100. These were the highest yield obtained in any of the cropping pattern trials, and are believed to reflect the maximum performance obtainable by farmer-cooperators. Data from patterns
which included SB were not included in the relative yield computation because SB is not locally adapted due to the high and uncontrollable levels of insects which causes low yields.
C/Number of cooperators not planting UC.
FS per week
&-Ma, /707 'I'V0607.
(Oeepwfer A7ble) w0ferhay
0 L-i;n L4
Rainfall (MM/Week) 6 Rai7fed, mn-wa*rm-), Rainfall 500
(ShO11OW M7fff 10t*) P&*a 4'7 '
2 L 200
0 tr 31 0
wof&woy (9wlow *VAVM7Y m7fertable)
0 0 L
23 27 31 35 39' 43 47
6- 0" Free-flowrq
23 27 31 35 39 43 47 Week
Figure 1. Weekly total rainfall and estimated.FS regimes for 9 Pangasinan land units, weeks 19 to 47, 1976.
heavy texture g
o - -23 27 31 35 39
Figure 2. Estimated FS regimes for heavy-textured and
li ht-textured rainfed non-waterway deep water
table land units, Pangasinan, weeks 23 to 39,
,, /heavy texture
* light texture i
0 I i
21 25 29 33 37' 41 45
Figure 3. Estimated FS regimes for heavy-textured and
light-textured partially irrigated non-waterway
shallow water table land units, Pangasinan,
weeks 23-45, 1976.
FS per week
2 'I amfoI i mmweek
8 23 27 31 35 39 43 47 51 19 23 27 31 35 9 43 47 5
4 .Figure 4. Weekly total rainfall and estimated FS regimes for 8 heavy-textured, rainfed and 2 t irrigated Iloilo land units, weeks 19 to
0 K)110 11 200
0 0- ,
19 23 27 31 35 39 43 47 51 923273135394347
4- Figure 4. Weekly total rainfall and estimated FS
regimes for 8 heavy-textured rainfed and 2- irrigated Ililo land units. weeks 19 to
0 3V3 5 94 75