Agronomy Journal 92:860-867 (2000)
© 2000 American Society of Agronomy
TROPICAL SOIL MANAGEMENT
Sequential Cropping as a Function of Water in a Seasonal Tropical Region
Ricardo Radulovich
Dep. of Agricultural Engineering, Univ. of Costa Rica, San José, Costa Rica
ricardo.radulovich{at}undp.org
 |
ABSTRACT
|
|---|
In the seasonal (wet–dry) tropics, yields of rainfed staple crops are usually low and variable. However, our simulations have indicated that rainfall could be used more efficiently, increasing the length of the rainfed cropping season, the number of crops grown, and their yields, while decreasing yield variability and risk. To evaluate these predictions, cropping schemes with two or three annual rainfed crops grown in sequence were field-tested. Irrigated plantings were subsequently added to evaluate year-round cropping. Work was conducted in Costa Rica during 4 yr at one site and 1 yr at a second site. Both sites have a half-year-long, bimodal rainy season, and have deep soils with high water-holding capacity. The first crop of each year was planted early using preseason rains, and the last was planted to mature after the rainy season to maximize depletion of available soil water. Rainfed cropping sequences averaged 255 d long, which is from 50 to 100 d longer than local practices. Adding an irrigated planting during the dry season brought the mean length of the cropping season up to 346 d. Of 115 plantings, only 7 had low yields; all others had medium or high yields. Adding an irrigated crop increased yield potential of a year-round cropping season composed of two or more plantings. All low yields were attributed here to extended water excess conditions. However, water excess effects do not remain after the rainy season, nor into the next rainy season.
Abbreviations: AQUA, Agricultural Query and Analysis, a water balance model • CATIE, Tropical Agronomy Center for Research and Teaching, Turrialba, Costa Rica • CIAT, International Center for Tropical Agriculture, Cali, Colombia • FAO, Food and Agriculture Organization of the United Nations
|
INTRODUCTION
|
|---|
THE MAJOR LIMITATION to all-year cropping in the tropics is imposed by water, where rainfall distribution dictates the seasons. Water shortages and excesses contribute significantly to low and highly variable yields, and often to complete crop failure. This is particularly important for the seasonal (wet–dry) tropics, due to large and often unpredictable variations in water conditions. For example, water deficit and excess explained from 70 to 90% of regional yield variability of bean (Phaseolus vulgaris L.), maize (Zea mays L.), and rice (Oryza sativa L.) in a seasonal tropical area (Radulovich, 1990).
Optimizing the use of available rainfall will contribute to increased productivity and to decreased yield variability. Thus, it will be possible to improve further annual cropping in seasonal tropical regions, which occupy 1.0 to 1.5 billion ha of potentially cultivable land area in the world (Sanchez, 1976; Lal, 1987). Because both economic and water resources available to expand irrigation are limited, selecting the right crops while increasing the use of rainfall water are, in many cases, the only options to increase productivity. Moreover, this must be done while dealing with water excess during the peaks of the rainy season.
Multiple cropping has clearly been recognized as an important method to increase resource use (e.g., Papendick et al., 1976). With sequential cropping, benefits are derived from obtaining the yield of more than one crop per year, grown one after another. In general, the longer actively growing crops are in the field, the more radiation will be intercepted and used for biomass and harvestable yields. However, largely due to highly variable rainfall characteristics, traditional rainfed farmers optimize using cautious, low-input cropping schemes—mainly a single maize crop, planted late to mature soon after the end of the rainy season, often intercropped with sorghum or a climbing bean. Thus, while securing a modest yield, they do not take full advantage of otherwise favorable conditions. Alternative cropping systems must be designed and validated if these farmers are to improve their socioeconomic condition within what their resources may allow.
Work reported here is an effort to evaluate the possibility and the effects of intensifying cropping through time in seasonal tropical regions by managing water as the main limiting factor. The aims are to increase the number of crops grown per year, yields, yield stability, and cropping season length based on a cropping season concept that optimizes yields not only per individual crop, but also through time. Such a cropping system approach entails three steps: (i) basic characterization of the region, emphasizing relations between farming systems, climate, soils, and crops; (ii) design and testing through simulation of cropping schemes that would best fit the region's or a farm's characteristics in relation to water; and (iii) field-testing of these designs, managing them with the aid of computerized models based on probabilistic and real-time rainfall data. This paper deals with the latter step, namely field validation of simulations. The other steps have been presented elsewhere (Radulovich, 1987a and b, 1989, 1990; Carmona and Radulovich, 1988), and they are related to other work in the literature (e.g., Muchow and Bellamy, 1991). Particularly, the approach used here is of the response farming type, which entails reacting to current rainfall characteristics within probabilistic possibilities (Stewart and Hash, 1982; Stewart and Kashasha, 1984; Sivakumar, 1988, 1990), applied to cropping sequences (Radulovich, 1989).
Accordingly, cropping schemes were managed using real-time rainfall data with the water balance model AQUA (Radulovich, 1987a, 1989, 1990; Carmona and Radulovich, 1988; Radulovich and Sanchez, 1993), specifically translating to field situation cropping schemes previously designed for this region (Radulovich, 1989; Radulovich et al., 1989). Thus, the objectives of this work were to field-test the simulated results using both probabilistic and real-time rainfall data in a dynamic, responsive manner (e.g., Stewart and Kashasha, 1984; Radulovich, 1989) to produce two or three sequential rainfed crops of medium to high yields per year. Irrigated plantings were added after rainfed sequences to evaluate the potential and limitations of these regions for year-round cropping.
|
Materials and methods
|
|---|
Work was conducted in Costa Rica at two stations, Alajuela and San Pedro, with seasonal and bimodal rainfall regimes (Fig. 1) . Alajuela, 34 km from and at a lower elevation than San Pedro, is warmer, and it also receives slightly more rain (Table 1) . Following a pattern common in the seasonal tropics, where half-year-long bimodal rainfall distributions are induced by the sun's apparent path, the rainy season normally begins from late April to mid-May and ends by early to late November, but it may be shorter, beginning in late May and/or ending by mid-October (Radulovich, 1989). There is a decrease in rainfall around July, which does not normally cause severe drought in these stations (Carmona and Radulovich, 1988). There are high rainfall peaks in May–June and September–October. The average length of the rainy season is approximately 180 d, plus up to a month or even longer of highly variable preseason rains and a similar period of equally variable postseason rains.
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 1 Mean monthly rainfall for the two stations of the study, in Alajuela (La Central, solid squares) and San Pedro (Sabanilla, solid circles), 20-yr averages
|
|
In Alajuela, two soils were used: an Ustic Dystropept (Soil Survey Staff, 1992), used only the first year, which is a clayey soil of poor structure with a compacted layer below 0.5 m that inhibits deep root penetration; and a Thapto-Vertic Dystric Haplustand, which exhibits good physical and chemical characteristics to a depth of about 1.0 m, where a buried Vertisol inhibits deeper root penetration. In spite of these physical limitations, neither soil exhibited drainage problems in the form of surface or subsurface free water accumulation. The soil at the San Pedro site is a Typic Haplustand, which is a deep, friable clayey soil with no known physical or chemical limitations to root penetration in the upper 1.5 m. Available water holding capacity of the Haplustands, determined by establishing field capacity in situ, is 228 mm m-1 for Alajuela and 237 mm m-1 for San Pedro. The soils at the Alajuela site had been used for intensive experimental cropping for 23 yr without interruption and were planted to sugar cane for decades before that. All plots were infested with nutsedge (Cyperus rotundus L.), which was controlled after 2 yr of field work. The soil at the San Pedro site, previously a coffee plantation, had not been cropped for at least 10 yr and supported an early successional stage of grasses and small bushes.
Crops grown were bean (cv. Huetar; secondarily, cv. Talamanca; and occasionally three cultivars from CIAT, Bat-477, Dor-363, and San Cristobal), cotton (Gossypium hirsutum L., cv. Pexa), maize (cv. Diamantes), peanut (Arachis hypogaea L., cv. Flowerson), rice (cv. CR5272), sorghum [Sorghum bicolor (L.) Moench, cv. Costasem II], soybean [Glycine max (L.) Merr., cv. IAC-8], and sunflower (Helianthus annuus L., cv. Mamut). Bean and maize were the primary crops, given their regional importance. Excepting the three bean cultivars from CIAT, which were used to expand work with bean, all cultivars are the ones locally recommended by the government, though not necessarily the ones normally used by all farmers. The bean cultivars used (in particular, cv. Huetar and Talamanca) are considered here as short-duration crops (75–85 d), while all other crops reported are considered long-duration crops (i.e., growing cycle > 100 d). These cultivars are in general of shorter cycle than traditional varieties used by many farmers. It should be noted that there are many available cultivars of these crops with even shorter growing cycles than those used here, allowing for a wide range of cropping possibilities with regard to time.
Studies were conducted at Alajuela during 4 yr from May 1988 through January 1992. During 1988, limited plantings were implemented on the Dystropept; for the three other years, plantings were on the Haplustand. Yearly, two cropping sequences were begun during or immediately after preseason rains with plantings of maize and of bean, followed by plantings of these and other crops (see Table 2
for examples of sequences). The bean–bean sequences also branched into third plantings in 1989 and 1990. For those same years, irrigated plantings were added after the rainfed sequences. In 1989, 1990, and 1991, extra maize and bean plantings were grown in these plots as part of more specific experiments on planting and harvesting dates. Yield data for these sole, nonsequential plantings (12 of maize and 24 of bean) are used here only in Fig. 2
. Plantings at San Pedro were conducted during the 1989 rainy season, ending in May 1990. Plantings were conducted for only 1 yr at this second site, as the purpose was to verify the transferability of the longer-term study in Alajuela. Three rainfed sequences were started using bean and soybean. Irrigated plantings were added after a sequence of three bean crops (see Table 2 for examples).
View this table:
[in this window]
[in a new window]
|
Table 2 Examples of rainfed and rainfed + irrigated cropping sequences implemented during 1989–1990. Yield data on each row are for the last crop of that sequence. The sequence's yield may be obtained by compounding individual yields
|
|
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 2 Bean and maize yields as affected by date of planting. Data used are from Alajuela (Haplustand soil) for all plantings of (a) a single bean cultivar (Huetar) and (b) a single maize cultivar (Diamantes). In the equations, x represents days of the year. ** Significant at the 0.01 probability level
|
|
The experimental design consisted of planting crops as indicated by a water balance model to validate the predictions that medium to high yields could be obtained from sequential cropping. All plantings followed a randomized complete block design with three replications per planting. Plot sizes within each block were 300 m2 for the first planting of each year and 20 to 50 m2 for the other plantings in Alajuela, and 40 to 100 m2 for the first planting and 20 to 50 m2 for the other plantings in San Pedro. The smaller plots were used with bean. Yields were obtained by harvesting from 3 to 20 m2 from at least two randomly chosen portions of each plot, excepting border areas. Yields of grains are reported on a 120 to 140 g kg-1 moisture basis, while those of peanut and cotton are reported after air-drying. For general yield analysis, a yield is considered low when its value is <50% of the mean yield obtained for that crop (see Table 3) , and high when yields are larger than that mean yield, with medium yields in between.
Agronomic practices followed are those recommended by the Costa Rican Ministry of Agriculture for intermediate commercial yields in the region. Thus, these practices do not represent current peasant practices, nor did the experiments attempt to maximize yields. Input and management levels (e.g., fertilizer levels) used were far below those normally used for yield maximization. Light cultivation was used for land preparation. Seeding was by hand or by using a manual planter. Plant population densities were intermediate within ranges normally recommended. Fertilization was 80 kg N ha-1 for bean, 120 kg N ha-1 for maize and rice, and 100 kg N ha-1 for the other crops, in two or three split applications (one of them incorporated at seeding time, the others sidedressed). Due to high P immobilization in volcanically derived soils, 30 kg P ha-1 (40 kg P ha-1 for maize and cotton) was incorporated at seeding time. For rainfed plantings close to the end of the rainy season, all the fertilizer was incorporated at seeding time. A mixture of granular soil insecticide-nematicide was applied with the fertilizer at seeding time. Two or three low-dose preventive applications of mixtures of insecticide and fungicide were applied during each crop's growing cycle, one of them with a commercial formula of foliar micronutrients. Five rainfed second-plantings required damage control with fungicide applications on a weekly basis for 3 to 4 wk, actually with little or no crop recovery from fungal attack. Preplant herbicide was used; postemergence weed control was mostly done by hoeing, plus some direct-contact herbicide applications using a wet wick. Irrigated plantings received amounts of water equivalent to full potential evapotranspiration weekly by furrows.
The water balance model AQUA (Agricultural Query and Analysis) was used to design and manage planting sequences. This model operates on daily water balance calculations and on probabilistic estimates of rainfall, producing cumulative days with water stress for the different periods and cropping options tested (Radulovich 1987a, 1989, 1990; Radulovich and Sanchez, 1993). Decisions based on these results are thus made choosing the cropping schemes with the least number of days with water stress. Plantings followed the cropping schemes designed earlier by simulation for Alajuela (Radulovich, 1989); however, each station was operated separately, according to its own rainfall and water balance characteristics. The basic cropping scheme followed consists of growing at least two sequential rainfed crops, attempting to maximize the length of the rainfed cropping season while decreasing water stress. Key aspects of this strategy, which combines avoidance and confrontation of water stress periods, are
- taking advantage of the preseason rains by planting early, thus lengthening the rainfed cropping season, entering rainfall peaks during June with well-established crops, and allowing reproductive growth and/or harvest during the midseason period of low rainfall if short-duration crops or cultivars are used;
- minimizing the time between harvest and the following sequential planting (minimizing turnaround time as feasible), thus having already well-established crops during the rainfall peaks of September and October; and
- maximizing use of late-season and postseason rains and of stored soil water after that, instead of using these only partially, as is currently done.
According to simulation, the rainfed cropping season begins during the preseason rains with the earliest safe planting date. This date is determined each year by real-time daily rainfall data within a probabilistically predetermined period based on 25-yr historic records. A minimum rainfall of 30 mm within that period is needed to trigger planting. By using this date, seeding can be done before the full start of the rainy season, while avoiding preseason water deficits from planting too early into a soil at or near permanent wilting point. In this manner, time is gained safely and the high rainfall events once the season starts are confronted with an established crop (Carmona and Radulovich, 1988; Radulovich, 1989). Given that in this region, the midseason period of low rainfall is not a major water deficit hazard (Carmona and Radulovich, 1988), it was considered here mostly as a low-rainfall window of opportunity for harvesting and planting; however, its relevance can be much larger in other regions.
A critical point of the rainfed cropping season is the last date during the rainy season in which the soil water is replenished to field capacity by rainfall before water deficits begin. This date is considered here the last day of the rainy season. The model requires that all rainfed seeding be done on or before that date as determined probabilistically. The rainfed cropping season is continued after that date by cropping on the basis of stored soil water plus any late rains, and it ends with the last harvest (Carmona and Radulovich, 1988; Radulovich, 1989; Radulovich et al., 1989). Analogous to work conducted in Africa (Stewart and Kashasha, 1984; Sivakumar, 1988), prediction of some late-season rainfall characteristics is also included in these schemes (Radulovich, 1987b, 1989). Briefly, it has been established that there is a negative relationship between May and November rainfall, which permits some prediction of rainfall amounts for November, and it also affects the last date when it is expected that the soil will be replenished to field capacity by rainfall.
Thus, taking advantage of rainfall characteristics and using conservative values of soil water storage capacity and of crop water extraction capabilities, simulations specified that the Alajuela site should have dependable rainfed cropping seasons of 209 to 236 d, with an average of 220 d (Radulovich, 1989). Given the similarities in soil and rainfall (Fig. 1), it was assumed that the San Pedro site would behave similarly to the main Alajuela site. Thus, similar cropping schemes were planned for both stations. However, it was considered as management criteria that a deeper soil and lower temperature and evapotranspiration rates in San Pedro (Table 1) would allow a longer cropping season than in Alajuela.
|
Results
|
|---|
Except for two periods of unusually high rainfall in 1988 due to hurricanes in the Caribbean, the four rainy seasons of this study did not depart from expected rainfall patterns. Following model recommendations, sowing was 1 to 4 d after all the criteria for earliest safe planting date were met each year at each site. In all cases this was during early May, and up to 40 d earlier than is recommended in the region. Every first planting of each year was successfully established before the rainy season reached its first peak (Fig. 1). After the first plantings of each season, one or two sequential rainfed plantings followed, sown from one to a few days after the harvest of the previous crop. Seeding for rainfed plantings continued until before the estimated last day of the rainy season, in November, when the soil is expected to be recharged fully to field capacity for the last time and before water deficits begin to set in. Such late plantings allowed for rainfed crop growth through March in some instances. In comparison with local practices, hardly any rainfed crop is seen growing through December. Irrigated crops were established as third or fourth plantings in 1989–1990 at both sites and in 1990–1991 in Alajuela.
A total of 115 plantings are reported here: 63 rainfed and 16 irrigated plantings were included in cropping sequences, while 24 bean and 12 maize nonsequential plantings are used only in Fig. 2. Selected examples of the cropping sequences for both sites are given in Table 2, showing that third or fourth irrigated plantings were implemented after two or three rainfed crops. The irrigated planting can even be a long crop, as shown in Table 2 for a 112-d, high-yielding sorghum crop planted after three bean crops. Harvest of this sorghum was on 25 April, totaling 357 d of cropping season length for that sequence.
The harvest of the last crop of each rainfed sequence begun in May was from December through March (see Table 2). The length of the rainfed cropping season, calculated as the number of days from the first planting to the last harvest of each rainfed sequence, varied from as low as 203 d to as high as 328 d in Alajuela, and from 273 to 319 d in San Pedro. Mean values for Alajuela in 1988–1989 were 232 d (SD = 17.6); in 1989–1990, 264 d (34.4); in 1990–1991, 236 d (31.1); in 1991–1992, 243 d (16.6); and for all four years, 250 d (31.0). This value was 289 d (16.0) for San Pedro in 1989–1990, which is not significantly larger by t-test than the value for Alajuela that year. The overall mean length of the rainfed cropping season for both stations was 255 d, which is from 50 to 100 d longer than traditional practices. Adding an irrigated crop after rainfed sequences extended cropping to 346 d (12.3) for the 14 cases in Alajuela and to 351 d (2.1) for the two cases in San Pedro, for a weighted mean length of cropping season of 346 d for both stations.
Mean rainfed yields obtained (Table 3) were much higher than values reported for Central America (Food and Agriculture Organization, 1992), particularly for bean, maize, and sorghum. For Costa Rica, national mean yields at that time were 1.5 t ha-1 for maize, 1.2 t ha-1 for sorghum, 0.7 t ha-1 for bean and 3.3 t ha-1 for rice (CNP, 1989). The difference between experimental and farmer's yields is much smaller for rice, because this crop is frequently grown mechanized and irrigated, in much better conditions than traditional farming. Mean rainfed yields obtained for the other crops were substantially higher than those reported for tropical Latin American countries, and similar to those reported for the USA (Food and Agriculture Organization, 1992). Only seven low individual-planting yields were obtained (three for bean, three for maize, and one for sunflower); all of them were second rainfed plantings in their respective sequence (i.e., grown during the second peak of the rainy season) (Fig. 1). No yields of first rainfed or of irrigated crops were low. Late rainfed plantings, sown even up to mid-November, produced medium to high yields, despite the ending of the rainy season soon after that. Thus, medium to high yields were generally obtained through a variety of rainfed cropping sequences, namely two long crops (cotton, maize, peanut, rice, sorghum, and soybean) or three crops (three short crops, i.e., bean crops, or two short crops plus a long crop), as well as combinations of one long and one short crop (see examples in Table 2, and number of plantings for each crop in Table 3). Such wide-range testing is important to count not only with a menu of crops and cultivars with different length of growing cycles, but also with a variety of cropping sequences to use for different conditions.
As indicated, all low yields were of second rainfed plantings. The general recurrence of lower yields in second plantings was analyzed for maize and bean against yields of first and of irrigated plantings from Alajuela with the Haplustand soil. For both crops, the mean of yields from first plantings was significantly larger than that of second plantings. The means of bean yields from first plantings and from fourth-irrigated plantings were not significantly different, while both means were significantly larger than that of second plantings (Table 4)
.
View this table:
[in this window]
[in a new window]
|
Table 4 Comparison between mean yields within sequence, for all plantings of maize and bean in Alajuela (Haplustand soil) and for all crops planted in both stations. A relative yield is obtained by dividing a crop's yield by that crop's mean yield from Table 3
|
|
It was necessary to draw upon relative yields for a more thorough analysis when comparing different crops. Yields of all sequential plantings in both stations were converted to relative yield values using each crop's mean yield from Table 3 as a reference. The mean of first rainfed plantings was found to be significantly larger than that of second rainfed plantings (Table 4). Also, the mean of first rainfed plantings was not significantly different from that of irrigated plantings; however, the mean of second rainfed plantings was significantly lower than that of the irrigated plantings. Thus, the pattern found for maize and bean in one station and soil was found for all crops in both stations when analyzed together (Table 4). No significant differences in absolute or relative yields were found between stations, nor between years in Alajuela.
In general, yields decreased as sowing date was delayed into the rainy season, recovering as sowing date approached and went beyond the end of the rainy season. This relationship was explained by polynomial equations of second degree, and it was found to be highly significant for all cases analyzed, including all crops from both stations together on a relative yield basis (R2 = 0.42**; n = 115). As expected, the strongest relationship was found when analyzing a single cultivar within a crop species for a single site, which is shown in Fig. 2 for bean and maize in Alajuela (Haplustand soil). As seen, yield variabilities of 66% for bean and 57% for maize were explained by planting date alone (although slightly larger R2 values were obtained with polynomial equations of third degree, these are not used here for the sake of parsimony). As mentioned earlier, traditional planting times for maize are usually well into the rainy season so that the ear of long cycle varieties will mature and be left drying in the field once the rainy season is over. However, delayed planting results in decreased yields (Fig. 2).
A clear and significant negative relationship between amounts of rainfall received and yields occurred for second-planted bean and maize in Alajuela (Haplustand soil) (Fig. 3)
. Rainfall amounts for Fig. 3b are for 60 and 80 d after planting for bean and maize, respectively, selected for giving the largest coefficients of determination. Interestingly, no relationship was found between rainfall amounts and yields of first-planted bean and maize, even though some first plantings were subjected to excessive amounts of rainfall, though later in their growth cycle than second plantings. Considering these results, and for these stations, much of the difference in yields throughout the cropping season can be attributed to water excess effects, which develop with the rainy season but do not remain after it is over (Fig. 2, Table 4). The main limitation for crop production was, therefore, when crop growth, particularly from germination and establishment, took place within excessive rainfall periods.
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 3 Relationship between cumulative rainfall and second-planting yields of (a) bean (cv. Huetar) and (b) relative yields of bean and maize (cv. Huetar and Diamantes, respectively) for Alajuela (Haplustand soil). Amounts of rainfall used are for 60 d after planting for bean and 80 d after planting for maize. *,** Significant at the 0.05 and 0.01 probability levels, respectively
|
|
For cumulative annual yield analysis, yields of 11 cropping sequences consisting of only one crop species each are shown in Table 5
, for which absolute yields were added up. High sum yields were obtained in all cases. However, sequences with different crop species should be expected as the norm, for which relative yield analysis is necessary. Using a method analogous to the area x time equivalency ratio of Hiebsch and McCollum (1987), which allows one to determine how efficiently a given field has been used through time, the yields obtained represent 93.0 and 94.8% overall cropping efficiency for rainfed and rainfed + irrigated crops, respectively. That is, on average, planting areas were successfully cropped, obtaining relatively high yields, over 90% of the time they were used for cropping. Such low rate of crop failure is encouraging given the intermediate management and input levels used.
|
Discussion
|
|---|
The mean length of the rainfed cropping season obtained in Alajuela, 249 d, is about 30 d longer than the values predicted earlier by simulation for this station, and from 50 to 100 d longer than local practices (Radulovich, 1989). A similarly long rainfed cropping season was successfully implemented in San Pedro. Similarities between stations, together with the similar responses obtained by using different crops, speak in favor of the transferability of the approach and of model recommendations. The difference between simulated and experimental lengths of rainfed cropping seasons were due in part to the large amounts of soil water available for extraction, and to a much larger than expected ability of crops to extract and grow on stored soil water (R. Radulovich and J.R. Chaves, unpublished data, 1990). Overall, long rainfed cropping seasons were obtained basically because (i) first plantings were seeded earlier than the normal practice, (ii) there was a short turnaround time of one to a few days between crops, and (iii) late plantings used postseason rains and available soil water, extending the rainfed season through March in some instances. Another important element is that the use of crops and cultivars of relatively short growing cycles allowed us to fit two or three crops into a cycle and thus to extend the rainfed season to its limit. Noticeably, and apparently due to low-risk strategies practiced by farmers, actively growing rainfed crops are rarely seen in these regions through December. (This fact actually ruined experiments in a third, drier site in Guanacaste, because animals, from rodents to horses, ate the very early and very late rainfed crops, fences not withstanding, as these were the only green vegetation around, moreover, adequately supplied with N).
The longest mean lengths of the rainfed cropping season were those of San Pedro and Alajuela in 1989–1990. This reflects more aggressive planning implemented in both stations that year, which implicated taking a larger risk using longer cropping sequences. Given the low rainfall in May that year, medium to high rainfall was expected in November, so longer sequences could be planted—actually, rains extended into early December that year. For the other 3 yr, high rainfall in May forecasted shorter rainy seasons—this assumption turned out to be true for 1988 and 1991. Overall, the limited predictive capability for the level of November rainfall contributed to better planning in 3 of the 4 yr. In the fourth year, 1990, the result was that planning was overcautious, since the rainy season did not end early as expected. However, and at least in this case, where soils with high water-holding capacity were used, the focus of any predictive capabilities may have to be shifted from water deficit to excess. Water deficit did not play a significant role here due to the soil and crop characteristics indicated above, and because simulations considered deficits adequately.
After the end of the rainy season, late-seeded rainfed plantings and irrigated plantings yielded as well as first plantings. First plantings in Alajuela were preceded by the plantings of the previous year, and by decades of intensive cropping. Yet first plantings in both stations, seeded during preseason rains, were not affected by excessive amounts of rainfall, while second plantings were (Table 4, Fig. 2 and 3). Water excess stress does develop during the rainy season, yet it does not continue in any form beyond the current rainy season, and thus it is not present once the dry season begins. No water excess effects were present for irrigated crops, nor did they affect first plantings at the beginning of the following rainy season. This relationship held true even when using the same crop and cultivar three or four times in a year and then again the following years. Far from endorsing a continuous monoculture, it is clear that the temporality of water excess effects has strong positive implications for intensifying cropping in seasonal tropical regions.
Rainfed strategies to increase the composite annual yields shown in Table 5 should be based mainly on increasing yields of second rainfed plantings; i.e., on decreasing the effects of water excess. Crops or cultivars more tolerant to water excess conditions should be grown, such as rice and perhaps soybean, while developing and applying technologies for these conditions. In general, and as noted by Lizaso and Ritchie (1997), there is little understanding of water excess conditions at the field level. Also, growing a third or fourth irrigated crop for year-round cropping largely increases the yield potential. Maximum composite yield potential is achieved by growing an irrigated crop after a sequence of three short, two long, or one long and one short rainfed plantings. However, the use of water for irrigation is limited, and the irrigated crop should be one of much greater value than a staple crop. The use of fast-maturing varieties and/or relay plantings should allow production of three long (e.g., maize or soybean) crops per year in this environment, aided by high temperatures year-round.
As proposed by Radulovich (1991), building on Soria (1976) and Sanchez (1976), optimization of cropping in terms of yield(s) per growing season should constitute the paradigm of tropical cropping, instead of the paradigm of temperate regions, which has been traditionally to optimize the yield of a single crop per year. A shift in paradigm allows a different view on tropical cropping. Given the longer cropping season, it follows that yields can be higher in the tropics, not on a per planting basis, but on a per length of cropping season basis. Three medium-level yields per year may be much higher than one high-level yield, and perhaps more sustainable, profitable, and equitable if medium-level inputs and local labor are used throughout the year instead of using high inputs in a highly mechanized operation during a single season. The fact that an irrigated crop is possible after an intensive rainfed season adds to the existing possibilities.
The generality of medium to high yields reported here, which included an instance of maize yielding over 7.0 t ha-1, were no surprise, even though these yields are from two to four times higher than those normally obtained by traditional farmers. Experimental rainfed maize yields over 6.0 t ha-1 have been common in the Alajuela experimental station since at least the early 1970s (e.g., Vives and Chacón, 1972). Furthermore, over 20 yr ago, Soria (1976) cited commercial yields with improved technology in Central America at 3.8 t ha-1 for maize, 4.5 t ha-1 for sorghum, 1.3 t ha-1 for bean, and 5.2 t ha-1 for rice. Besides planting in a more careful manner regarding water, yields reported here are attributed to using soils that are not marginal or highly degraded, and to using consistently recommended crops, practices and inputs for commercial production. Adding soil and water conservation measures, such as contour planting, in situ water harvesting, mulches and wind barriers, and measures to deal with water excess together with other sound agricultural practices, will contribute toward increasing the sustainability of higher productivity cropping in the seasonal tropics.
|
Conclusions
|
|---|
Implementation and transferability of model recommendations and of cropping based on water did not present any major limitation, at least at the research stage. In general, medium to high yields were obtained and the length of the cropping season was increased. The large potential for cropping in the sites tested was evidenced, as was the water excess limitation. The procedure followed during this work and the results obtained contribute to the conclusion that managing crops and cropping sequences in a dynamic manner regarding rainfall is an adequate avenue for the development of alternative rainfed annual cropping in the seasonal tropics. It can also be a cost-effective alternative or complement to the development of irrigation schemes, particularly given the ubiquitous shortages of water for irrigation.
Evidently, there is a need for farmers to have access to, understand, and use climatic data and other recommendations such as timely planting if they are to implement what are actually seen as high-risk strategies. To achieve this, the concept of an agrometeorological district is offered as a management possibility. An agrometeorological district would be an advisory unit that provides timely recommendations to farmers regarding planting, harvesting, and other opportunities and problems in relation to climate, mainly rainfall. To manage the set of criteria used here (i.e., the earliest safe planting date, the last day when the soil is taken to field capacity by rains, and the predictive capability of gross amounts of late rainfall, among others) would be the priority task of any agrometeorological district or similar operation.
While the dynamics under water deficits and excesses are beyond the scope of this paper, it is clear that problems associated with water excess need to be solved before these cropping schemes can be safely and fully implemented. These solutions will have to deal not only with direct crop well-being, but also with field (land preparation, planting, and harvesting) and postharvest operations at an appropriate scale. Work on water excess for the humid, subhumid, and semidry tropics represents a most promising line of research and extension that can generally benefit large areas of the world characterized by poor and unsustainable agriculture.Consejo Nacional de Producción 1989
|
ACKNOWLEDGMENTS
|
|---|
Thanks are given to Mr. Rodolfo Chaves, now with Chiquita Banana, Costa Rica, for his excellent assistance in conducting field work, and to Dr. Donald Kass, of CATIE, Turrialba, Costa Rica, for his kind revision of a final draft of this paper. Special thanks are given to the editor and the three reviewers, who kindly invested much time during the review process.
|
NOTES
|
|---|
Mailing address: SJO 492, P.O. Box 025216, Miami, FL 33102-5216. Research supported by grant No. 8.337 of the Program in Science and Technology Cooperation, USAID, Washington, DC, and by the University of Costa Rica.
Received for publication January 18, 1999.
|
REFERENCES
|
|---|
- Carmona G., Radulovich R. Metodología de evaluación de veranillos y de siembra temprana como estrategia para minimizar sus efectos. Turrialba 1988;38:215-222.
- Consejo Nacional de Producción (National Production Council). 1989. Boletín Agrotécnico, Nov. 1989. CNP, San José, Costa Rica.
- Food and Agriculture Organization Production yearbook. Rome: FAO, 1992 Food and Agriculture Organization Statistics Ser. no. 104..
- Hiebsch C.K., McCollum R.E. Area-x-time equivalency ratio: A method for evaluating the productivity of intercrops. Agron. J. 1987;79:15-22.[Abstract/Free Full Text]
- Lal R. Tropical ecology and physical edaphology. London: John Wiley & Sons, 1987.
- Lizaso J.I., Ritchie J.T. Maize shoot and root response to root zone saturation during vegetative growth. Agron. J. 1997;89:125-134.[Abstract/Free Full Text]
- Muchow, R.C., and J.C. Bellamy (ed.). 1991. Climatic risk in crop production: Models and management for the semiarid tropics and subtropics. CAB. International, London.
- Papendick, R.I., P.A. Sanchez, and G.B. Triplett (ed.). 1976. Multiple cropping.ASA Spec. Publ. 27. ASA, CSSA, and SSSA, Madison, WI.
- Radulovich R. AQUA, a model to evaluate water deficits and excesses in tropical cropping: I. Basic assumptions and yield. Agric. For. Meteorol. 1987;40:305-321 a.
- Radulovich R. Predicción de la precipitación de fin de estación lluviosa en el régimen del Pacífico. Agron. Costarric. 1987;11:55-64 b.
- Radulovich R. Optimization of rainfed tropical cropping in semi-dry areas: A case study. Agric. Water Manage. 1989;16:337-352.
- Radulovich R. AQUA, a model to evaluate water deficits and excesses in tropical cropping: II. Regional yield prediction. Agric. For. Meteorol. 1990;52:253-261.
- Radulovich R. Desarrollo agrícola en el trópico Latinoamericano: El caso del pequeño agricultor vs. la economía nacional. Interciencia 1991;16:125-130.
- Radulovich R., Sanchez F. AQUA: Modelo de balance hídrico para agricultura tropical. San José: Editorial Universidad de Costa Rica, 1993 Version 1.0..
- Radulovich R., Vargas G., Chaves J.R., Sánchez F. Agua y cultivos secuenciales en zonas semi-secas: Evaluación de campo preliminar. Agron. Costarric. 1989;13:219-225.
- Rojas, O.E. 1985. Estudio agroclimático de Costa Rica. Interamerican Institute for Cooperation on Agriculture (IICA), Serie Publicaciones Misceláneas no. 617. IICA, San José, Costa Rica.
- Sanchez P.A. Properties and management of soils in the tropics. New York: John Wiley & Sons, 1976.
- Sivakumar M.V.K. Predicting rainy season potential from the onset of rains in southern Sahelian and Sudanian climatic zones of West Africa. Agric. For. Meteorol. 1988;42:295-305.
- Sivakumar M.V.K. Exploiting rainy season potential from the onset of rains in the Sahelian zone of West Africa. Agric. For. Meteorol. 1990;51:321-332.
- Soil Survey Staff. Keys to soil taxonomy, 5th ed Blacksburg, VA: Pocahontas Press, 1992 Soil Management Support Services Tech. Monogr. 19..
- Soria J. Los sistemas de agricultura en el Istmo Centroamericano. Rev. Biol. Trop. 1976;4(Suppl. 1):57-68.
- Stewart J.I., Hash C.T. Impact of weather analysis on agricultural production and planning decisions for the semiarid areas of Kenya. J. Appl. Meteorol. 1982;21:477-494.
- Stewart, J.I., and D.A.R. Kashasha. 1984. Response farming of maize and beans at Katumani, Machakos District, Kenya: Recommendations, yield expectations and economic benefits. East Afr. Agric. For. J. 44(Special issue, Oct.):58–79.
- Vives L.A., Chacón A. Análisis de épocas de siembra del maíz. Vol. 5, no. 5. Boletín Técnico Alajuela, Costa Rica: Estación Fabio Baudrit, 1972.
TOP
ABSTRACT
INTRODUCTION
