Scientific evidence of yield and productivity declines in irrigated rice systems of tropical Asia

K.G. Cassman,a D.C. Olkb and A. Dobermannb

aAgronomy Department, University of Nebraska, Lincoln, NE 68583-0915 USA;
bInternational Rice Research Institute, PO Box 933, 1099 Manila, the Philippines

Rice is the primary source of calories for the human population of Asia. Sustaining the rate of growth in rice production is essential for food security. In recent years, however, there has been a significant reduction in this growth rate, which was less than the rate of population growth in the past decade (FAO, 1996). Reversing this trend is crucial to ensure adequate rice supplies at reasonable prices. Hence, identifying the causes of the deceleration in rice production growth rate is fundamental to the goal of food security. This report summarizes recent information about rice yield decline in long-term experiments and the results from on-farm studies of factor productivity (i.e. grain output per unit of applied input) in relation to broad measures of soil quality and management practices in irrigated rice systems. The discussion will focus on intensive systems in which farmers harvest two and sometimes three rice crops from the same field each year. Justification for this emphasis comes from increasing reports of declining rice yields, yield stagnation or declining factor productivity in some of the most favourable production environments in which these systems predominate (Cassman and Pingali, 1995b; FAO, 1996). In addition, irrigated double- and triple-crop rice systems account for more than 28 million ha of harvested area and 30 percent of annual rice production in the developing countries of tropical Asia (Cassman and Pingali, 1995a).

Present rice yields average about 5 tonnes/ha in these intensive rice systems, which is less than 60 percent of the climate-adjusted yield potential of existing high-yielding varieties (HYVs) (Cassman and Harwood, 1995; Mathews et al., 1995). Although an exploitable yield gap still exists, it must be rapidly closed to meet the projected increase in rice demand over the next 30 years. While it is recognized that market forces and government policies also have a major influence on production trends, this review is narrowly focused on some of the biophysical relationships that govern productivity. At issue is whether our knowledge of these relationships and available technology are sufficient to assure adequate growth rates in rice production from these important agro-ecosystems.

YIELD TRENDS, LONG-TERM EXPERIMENTS AND ON-FARM STUDIES

National yield trends are based on aggregate output data from different districts and agro-ecological zones within a country. They reflect the influence of technological changes such as expansion of irrigated area, increased use of fertilizers, improved varieties and pest management and mechanization. Such aggregation can conceal yield declines, stagnant yields or decreasing factor productivity in key production zones (Cassman and Pingali, 1995b). A focus on long-term experiments and on-farm factor productivity in specific production domains is needed to identify the reasons for the observed yield trends. Long-term field studies provide a "constant" management regime in defined treatments so that yield trends can be interpreted in relation to changes in soil properties and climate. Results from these studies can be used to identify the key factors that sustain productivity, and these hypotheses can be validated in on-farm studies.

FACTORS AFFECTING YIELD DECLINE IN LONG-TERM EXPERIMENTS

Although aggregate rice yields reported at the national level continue to rise in most Asian countries, albeit at a reduced rate, yield declines have been documented in a number of long-term experiments on double-crop and triple-crop irrigated rice systems in the Philippines and India (Cassman and Pingali, 1995b). These experiments utilize "best management practices" that include recommended fertilizer rates, the best varieties, proper floodwater control and pest management as required to avoid yield loss, and they are typically established in important rice-growing domains.

Nitrogen management and soil nitrogen supply

Nitrogen (N) was initially identified as the most limiting factor in these experiments (Nambiar and Ghosh, 1984; De Datta, Gomez and Descalsota, 1988). At two sites in the Philippines, rice yields declined by 30 percent over a 23 to 26 year period but yields could be restored to original levels by increasing the amount of applied N and improved timing of N applications (Cassman et al., 1995). Restoration of yields from increased N rates indicated that the indigenous N supply of the soil-floodwater system had decreased over time because crop uptake efficiency of the applied N and the physiological efficiency of the acquired N in producing grain had not changed.

Despite the apparent decrease in soil N supply, soil organic carbon and total soil N were conserved or increased compared with original levels at the beginning of the experiment, especially in treatments receiving balanced nutrient inputs (Table 1). Conservation of or increases in soil organic matter occurred even when all above-ground crop residues were removed so that only roots were returned to soil. This ability to conserve or sequester organic carbon appears to be a common characteristic of intensive irrigated rice systems. Another unique characteristic is the apparent lack of association between the native soil N supply and soil organic matter content or total soil N (Cassman et al., 1996b). This unique property results from the processes of humus formation and decomposition, mediated by soil micro-organisms under the anaerobic or nearly anaerobic conditions of flooded soils. These processes and the associated microbial communities differ in the aerated soils in upland cropping systems where it is more difficult to maintain soil organic carbon and N levels. This difficulty is particularly acute in the tropics where high temperatures promote microbial activity throughout the year.

But how can we explain the increase in soil organic matter with a decrease in soil N supply? With continuous rice cropping under submerged soil conditions, phenols were found to accumulate in the organic matter of soils from long-term experiments (Olk et al., 1996a). Phenol accumulation was observed in whole soil (Table 2), and also in labile humic acid fractions extracted from these soils. Increased phenol content is known to inhibit N mineralization of crop residues and green manures in flooded rice soils (Becker et al., 1994), and it may also reduce N mineralization from labile organic matter fractions. Thus, it would be possible to observe a reduction in soil N supply capacity despite the conservation of or increase in soil organic matter and total N in intensive rice systems, where cropping mostly occurs on flooded soils. Recent research has documented a threefold difference in N mineralization from humic acid fractions extracted from lowland rice soils in Viet Nam and the Philippines, and these differences were associated with chemical properties related to humification (Bao Ngyuen Ve, 1996). Moreover, the N contained in the extracted humic acids represented 20 to 30 percent of total soil N.

Table 1

Soil organic carbon and total N content in soils from the unfertilized control treatment (0-0-0) and treatment with recommended rates of N, P and K (N+P+K) in long-term experiments at the International Rice Research Institute (IRRI) in Laguna Province and the Philippine Rice Research Institute (PRRI) in Nueva Ecija Province

Site

Year

Organic carbon

Total

0-0-0
(g/kg-1)

N+P+K
(g/kg-1)

N 0-0-0

N+P+K

IRRI

1964

20.0

20.0

1.78

1.78

1982

18.9

20.41

1.99

2.091

1988

19.3

21.21

1.86

2.011

1991

18.2

19.91

1.87

2.011

PRRI

1968

8.7

 8.7

0.80

0.80

1977

10.6

12.01

0.97

1.201

1988

10.8

12.51

0.92

1.031

1991

10.4

12.61

0.96

1.171

1Indicates a significant difference (P <0.05) in organic carbon or total N between the unfertilized and fertilized treatments in a given year.
Source:
Modified from Cassman et al., 1995.

The poor association between the native soil N supply and organic matter in intensive rice systems was also indicated by the lack of increase in soil N supply after long-term use of legume green manure or Azolla microphylla (Cassman et al., 1996a). Application of these organic N sources to 21 consecutive rice crops resulted in a significant increase in total soil N compared with treatments that provided an equivalent rate of N as urea. Despite this increase, there was no long-term benefit on rice yields from the use of green manure (Table 3), and the apparent N uptake from native soil reserves was similar in both urea and organic N source treatments. Analysis of the labile humic acid fractions extracted from these soils revealed a similar accumulation of phenols in both green manure and urea treatments (Olk et al., 1996a).

Results from these long-term experiments provide strong evidence that soil N supply is governed more by the chemical qualities than the amount of soil organic matter, and that a decrease in soil N supply can occur although organic matter and total N increase in intensive irrigated rice systems. These characteristics reflect the microbial processes that control the formation and degradation of soil humus in flooded soils, and they may also affect the availability of applied N. Recent evidence indicates that the processes of carbon sequestration and the decline in soil N supply per unit of organic soil N can be reversed by incorporating rice straw or other crop residues during the fallow period between rice crops, when soil is aerated, rather than during the puddling operation before transplanting or direct wet seeding. While residue incorporation under aerated soil conditions requires mechanization, it can increase the overall N-use efficiency of the cropping system by increasing the availability of soil N and the N contributions from recycled rice straw and crop residues, thus reducing the requirements for applied N fertilizer.

Table 2

Soil organic carbon and phenol content under four cropping systems that differ in annual rice cropping intensity and water regime, IRRI research farm in Laguna Province, the Philippines

Cropping system
(water regime)

Total organic carbon
(g/kg-1 soil)

Phenol content

Whole soil
(g/kg-1 soil)

Soil organic matter
(g/kg-1 C)

Continuous dryland rice (rainfed)

13.0

0.18

14

Rice-soybean (irrigated)1

13.3

0.20

15

Rice-rice (irrigated)1

22.5

0.61

27

Rice-rice-rice (irrigated)1

28.8

1.02

35

1 Irrigated rice grown in flooded soil.
Source: Modified from Olk et al., 1996a.

Table 3

Long-term effects of organic and inorganic N sources on rice yields at two sites in Laguna Province, the Philippines1

N Source

Grain yield

IRRI

Victoria

Dry season
(n=10)2

Wet season
(n=11)

Dry season
(n=9)

Wet season
(n=9)

(tonnes/ha-1)

Control

3.79c 3

3.39c

3.83b

3.22b

Urea

5.99a

4.24a

5.89a

3.99a

Azolla

5.63b

4.02b

5.87a

3.96a

Sesbania

5.72b

3.90b

-

-

Urea + rice straw

5.73b

3.99b

5.68a

3.89a

1 Except for the control without applied N, equivalent amounts of N were applied in all treatments during a 10- to 11-year period with 18 to 22 consecutive rice crops.
2 Values in parentheses are the number of consecutive rice crops during the experiment.
3 Values shown are mean yields for each treatment in the 10- to 11-year period of the experiment. Means within columns followed by the same letter do not differ significantly (P <0.05).
Source: Modified from Cassman et al., 1996a.

Deficiencies of other nutrients

Although N management is typically the initial driving force of rice yield potential in long-term experiments, deficiencies of other macronutrients are also important constraints to sustaining high yield levels. This is illustrated by results from a site where phosphorus (P) and potassium (K) deficiencies developed despite the application of recommended rates of these nutrients from the start of this long-term experiment in 1968 (Table 4). With the original levels of applied N, there was little response to applied K although yields responded to P application. After the N rate was increased, a twofold increase in the amount of applied K was required to maintain yields above 8 tonnes/ha. Thus, the original recommendation for K application was sufficient to produce yields above 8 tonnes/ha in the initial years of this experiment before soil K reserves were depleted (De Datta, Gomez and Descalsota, 1988), and this same K rate could sustain yields of 6 to 6.5 tonnes/ha in later years when soil K was deficient. These original K rates were inadequate, however, to support higher yields in the K-deficient soil. In contrast, the original recommendations for P were sufficient to maintain yields and soil P reserves throughout the experiment, although P rates were increased slightly to ensure a positive P balance.

Despite the importance of proper nutrient balance, existing soil testing methods are generally quite poor at identifying soils on which a response to applied nutrients can be expected. For intensive rice systems, this problem was documented by evaluating soil test data and crop response in 11 long-term experiments located in five Asian countries (Dobermann, Cassman and Sta. Cruz, 1996; Dobermann et al., 1996a and 1996b). Although plant tissue testing is a more accurate tool, it is not a feasible option to improve nutrient management by rice farmers.

Table 4

Mean dry season rice yields in consecutive three-year periods before (1989-91) and after (1992-94) as a result of increasing the amounts of applied nutrients in a long-term experiment at PRRI in Nueva Ecija Province, the Philippines

Fertilizer rate

Mean grain yield 1

1989-91

1992-94

1989-91

1992-94

N

P

K

N

P

K

(kg ha-1)

(tonnes/ha-1)

0

0

0

0

0

0

2.98c2

3.22d

140

0

0

200

0

0

4.39b

4.48c

140

26

0

200

30

0

6.36a

5.57b

140

0

50

200

0

100

4.40b

4.88c

140

26

50

200

30

100

6.97a

8.14a

1 Values for each treatment are pooled means from three dry season crops in the three-year period indicated.
2 Within each site and period, means followed by the same letter are not significantly different (P <0.05).

Disease pressure and nutrient-disease interactions

Increased disease pressure is also a constraint to sustaining high yields in intensive rice systems. In long-term experiments, the need for increased N fertilizer rates to maintain yields in the face of decreased soil N supply or to close the gap between present yield levels and yield potential often results in greater disease pressure from sheath blight (Rhizoctonia solani) and blast (Pyricularia grisea) (Mew, 1991). These diseases are much less of a problem when the rice crop is N deficient (Fig. 1). The large leaf area required to achieve high yields results in a dense, N-rich canopy which provides a conducive environment for disease development when climatic conditions favour these pathogens (Cu et al., 1996). Although varietal resistance provides a significant measure of protection against blast, there has been little progress in identifying sources of resistance to sheath blight. Hence, achieving high and stable yields requires preventative measures to avoid yield loss when climatic conditions are conducive to sheath blight development. Decision-support tools for taking preventative action in response to conducive climatic conditions are needed to sustain high yield levels.

Fig.1

Another suite of diseases are associated with nutrient imbalances or deficiencies other than N. Deficient K supply is often associated with diseases such as brown spot (Helminthosporium oryzae) and narrow brown spot (Cercospora oryzae), particularly when coupled with excessive N supply (Huber and Arny, 1985). In recent years, a new disorder called "red stripe" or "ugly-ripening" disease has been reported in intensive rice domains of Viet Nam, Malaysia, Indonesia and the Philippines. Although the pathogens responsible for this disease have not been identified, disease incidence appears to be associated with nutritional imbalance.

A MODEL OF DECLINING PRODUCTIVITY IN IRRIGATED RICE SYSTEMS

Based on the recent results from long-term experiments and general knowledge of intensive rice systems, a conceptual model of factors contributing to declining productivity is presented in Figure 2. In this model, all intensively cropped rice soils are considered N-deficient in that they require applied N to achieve high yields. Soil N supply is therefore a critical component of soil quality, but the effective soil N supply is affected by anaerobic N mineralization-immobilization processes that are unique to flooded soils. These processes are responsible for the lack of association between soil organic matter and native soil N supply, but it appears that they can be managed to improve the overall N-use efficiency of the cropping system. Constraints other than N are more site-specific and depend on the long-term nutrient balance, climate, pest pressure and the accumulation of salts or micronutrients to toxic levels.

Fig.2

FACTOR PRODUCTIVITY AND NUTRIENT MANAGEMENT ON-FARM

Partial factor productivity for a given input is the ratio of grain yield to the amount (or cost) of the applied input. For a given yield level, optimal factor productivity from applied nutrients is achieved when the use of indigenous soil nutrients is maximal and the efficiency of applied nutrients in producing economic yield is high. In the long term, specification of optimal factor productivity must also consider nutrient balance so that the depletion of nutrient stocks below critical threshold levels does not lead to increased requirements for applied nutrients to maintain yield levels.

At issue is how well rice farmers account for soil nutrient supply and nutrient balance in their management of soil and fertilizer to optimize factor productivity from applied nutrients. Poor management of these resources can result in yield reduction, yield stagnation or declining factor productivity. During the past five years, scientists from the International Rice Research Institution (IRRI) and the National Agricultural Research institutions of the Philippines, India, Indonesia, Thailand and Viet Nam set out to study this issue in on-farm experiments located in intensive irrigated rice production domains within a 20 km radius of a major rice research experiment station. Standard experimental protocols and soil-plant analysis methods were used in all five countries.

Initial studies established N fertilizer omission plots (N0) in a representative sample of 24 to 36 individual farmers' fields (Cassman et al., 1993). Three N0 plots, each 36 m2, were placed in different locations within each field, and these plots did not receive applied N at any time before or during the rice cropping period. Outside these omission plots, farmers applied their own nutrient management regime and nutrient rates were recorded by the researchers. If farmers applied P, K or other nutrients to the field at large, the same rate of these nutrients was applied to the N0 plots. Farmers implemented all other crop management operations in these fields. A soil sample was taken from each N0 plot at the beginning of the cropping season for soil testing, and grain yield and nutrient uptake were measured in both the N0 plots and in an adjacent area that was under the farmers' fertilizer management. Diseases, insect damage and weed pressure were monitored but did not indicate significant yield losses from these pests.

The N uptake by the rice crop and the grain yield supported by native soil resources in the N0 plots provides a direct measure of the effective soil N supply. It also provides a broad measure of soil quality because N supply and uptake are influenced by physical, chemical and biological soil properties. Although soil types within the relatively small study area were similar, there was a tremendous range in yields and N uptake in the N0 plots, indicating a large variation in the effective soil N supply (Cassman et al., 1996b). Grain yields ranged from 2 to 6 tonnes/ha, N uptake from 35 to 120 kg/ha, and grain yields were closely correlated with N uptake (r2 = 0.89). Statistical analyses indicated that the differences among fields in apparent soil N supply were highly significant (P <0.001). Despite the wide range in the indigenous soil N supply, farmers did not adjust the rate of applied N to account for these differences (Fig. 3). In many cases, farmers with relatively high soil N supply applied high N fertilizer rates, while farmers with low soil N supply applied small fertilizer rates. Soil analyses for organic carbon, total N and extractable inorganic N revealed that there was no relationship between these parameters and the grain yield or N uptake from N0 plots (Cassman et al., 1995). The lack of a significant relationship is consistent with the results from the long-term experiments which also indicated that soil N supply was not closely associated with soil organic carbon content and soil N. It therefore appears that both farmers and researchers have difficulty predicting soil N supply.

Fig.3

Subsequent studies conducted in India, Indonesia, Thailand, and India in 1994-1995 gave similar results (Olk et al., 1996b). In each of the five study areas, irrigated rice systems with two and sometimes three rice crops each year had been practised for the past 10 to 20 years. Although soil properties differed between the five study areas, soil types within each of the relatively small study areas were similar. Despite the similarity of cropping systems and soil types within study areas, there were two- to threefold differences among farmers' fields in grain yield and N uptake in N0 plots at each location. Soil organic carbon or total N explained little of the variation in apparent soil N supply capacity. Grain yields were tightly correlated with N uptake, but there was no relationship between the N fertilizer rates that farmers applied and the effective soil N supply measured by crop N uptake in the N0 plots. This mismatch between the availability of indigenous soil resources and applied inputs contributed to the relatively low N fertilizer efficiencies of 5 to 15 kg increase in grain yield per kg of applied N, with a mean efficiency of 11 kg/kg. In contrast, N fertilizer efficiencies of 24 to 39 kg of grain per kg of applied N can be achieved at yield levels of 8 to 9.7 tonnes/ha when N fertilizer application is properly timed to complement the soil N supply (Peng et al., 1996).

Measurements of P and K uptake by the rice crop were also taken in unfertilized plots established in farmers' fields within the five study areas. Like the N0 plots in the earlier study, nutrient uptake and grain yield measured in unfertilized plots provide a broad measure of soil quality. Using a new soil test method that measures the dynamic soil supply of P and K (Dobermann, Cassman and Sta. Cruz, 1996; Dobermann et al., 1996a), the indigenous soil supply capacity of these nutrients also varied greatly among farmers' fields in the study areas where these measurements were taken (Dobermann et al., 1996b). Again, despite this variation, there was no correlation between the amount of P and K applied by farmers and the dynamic P and K supply measurements. In addition, the measurement of plant nutrient status in the field under the farmers' nutrient management regime indicated yield reductions from P and/or K deficiency on a number of farms. These results indicate a mismatch between soil P and K supply and the amount of fertilizer P and K that farmers apply, and this imbalance would reduce the factor productivity: in many fields nutrients were applied when they were not required while yields in other fields were limited by too little or a lack of P and/or K application. In addition, the data indicated that K applications were far below levels required to maintain soil K stocks while P inputs were somewhat less than the amounts removed in harvested grain and straw.

The initial focus of these studies was on N, P and K because these nutrients are required in the greatest amounts and the costs of purchasing these fertilizers are greater than for other essential nutrients. Although sulphur and zinc can also limit rice yields in some irrigated rice areas, deficiencies of these constraints are relatively inexpensive to rectify and are therefore mostly a problem of diagnosis and access to the appropriate source of fertilizer. However, a similar approach using nutrient omission plots in farmers' fields can also be used to evaluate the response to these nutrients.

SUMMARY AND CONCLUSIONS

While many factors affect rice production trends at the farm level, proper nutrient management has a major influence on both yields and productivity in the short and long term. Yield declines may occur when management practices are held constant in long-term experiments on intensive irrigated rice systems, owing to changes in soil properties as a result of intensive cropping and improper nutrient balance. It is not certain that yield declines occur or are widespread in farmers' fields because farmers continuously change management practices and adopt new technologies in response to environmental conditions and market forces. In the five irrigated rice domains studied, however, it is clear that nutrient management practices currently used by farmers do not provide a good balance between soil nutrient supply, crop requirements and the amount of applied nutrients. Such an imbalance reduces factor productivity and/or yields. It also leads to a depletion of soil fertility when inputs do not replenish extracted nutrients in soils that are already nutrient-deficient, or potential environmental problems when nutrients are applied in excess. Although the present studies were limited to five domains, we suspect that poor nutrient management is widespread in intensive rice systems elsewhere because present soil test methods do not accurately predict the need for applied nutrients, and nutrient availability is extremely dynamic in the soil-floodwater system, particularly for N.

If our findings are representative of intensive rice systems in general, a "field-specific" approach to nutrient management will be required to improve factor productivity and yields (Dobermann et al., 1996b). Sustaining the needed increase in rice production growth rates from these key rice production domains will not be possible without a field-specific nutrient management approach. At present, however, most national agricultural research systems provide nutrient management recommendations on a regional or district basis, and this level of aggregation is far too large.

The challenge of developing field-specific nutrient management techniques for rice farmers in the developing countries of Asia is enormous. It will require new approaches to participatory on-farm research on nutrient management, and it should be coupled with similar efforts in integrated pest management because of nutrient-pest interactions. Researchers will be challenged by the need for a greater understanding of nutrient cycling processes that govern short- and long-term availability of essential elements to develop improved management strategies. Diagnostic tools and practical decision-support systems will be needed for use by extension workers and farmers. We are convinced that present knowledge and soil test methods in widespread use are not adequate for the task faced.

Despite the magnitude of this challenge, there is a decreasing pool of scientists being trained in soil fertility and plant nutrition. Funding priorities have shifted to other areas such as molecular biology and biotechnology. This trend is occurring in both developed and developing countries. Unfortunately, improved soil and nutrient management for sustainable rice production cannot be achieved by improved germplasm alone. Fortunately, a yield gap between potential rice yields and actual on-farm yields still exists, and improved nutrient management will make an important contribution to closing this gap.

REFERENCES

Bao Ngyuen Ve. 1996. Characterization of two humic acid fractions and their contribution to nitrogen supplying capacity of tropical lowland rice soils. August, 1996. Department of Soil Science, University of the Philippines, Los Ba�os.

Becker, M., Ladha, J.K., Simpson, I.C. & Ottow, J.C.G. 1994. Parameters affecting residue nitrogen mineralization in flooded soils. Soil Sci. Soc. Am. J., 58: 1666-1671.

Cassman, K.G. & Harwood, R.R. 1995. The nature of agricultural systems: food security and environmental balance. Food Policy, 20: 439-454.

Cassman, K.G. & Pingali, P.L. 1995a. Intensification of irrigated rice systems: learning from the past to meet future challenges. GeoJournal, 35: 299-305.

Cassman, K.G. & Pingali, P.L. 1995b. Extrapolating trends from long-term experiments to farmers fields: the case of irrigated rice systems in Asia. In V. Barnett, R. Payne & R. Steiner, eds. Agricultural sustainability in economic, environmental and statistical considerations, p. 63-84. London, UK, Wiley.

Cassman, K.G., Kropff, M.J., Gaunt, J. & Peng, S. 1993. Nitrogen use efficiency of irrigated rice: What are the key constraints? Plant Soil, 155/156: 359-362.

Cassman, K.G., De Datta, S.K., Olk, D.C., Alcantara, J.M., Samson, M.I., Descalsota, J.P. & Dizon, M.A. 1995. Yield decline and the nitrogen economy of long-term experiments on continuous, irrigated rice systems in the tropics. In R. Lal & B.A. Stewart, eds. Soil management: experimental basis for sustainability and environmental quality, p. 181-222. Boca Raton, USA, Lewis/CRC Publishers.

Cassman, K.G., De Datta, S.K., Amarante, S., Liboon, S., Samson, M.I. & Dizon, M.A. 1996a. Long-term comparison of the agronomic efficiency and residual benefits of organic and inorganic nitrogen sources for tropical lowland rice. Experimental Agric., 32: 427-444.

Cassman, K.G., Dobermann, A., Sta. Cruz, P.C., Gines, G.C., Samson, M.I., Descalsota, J.P., Alcantara, J.M., Dizon, M.A. & Olk, D.C. 1996b. Soil organic matter and the indigenous nitrogen supply of intensive irrigated rice systems in the tropics. Plant Soil, 182: 267-278.

Cassman, K.G., Gines, G.C., Dizon, M.A., Samson, M.I. & Alcantara, J.M. 1996c. Nitrogen-use efficiency in tropical lowland rice systems: contributions from indigenous and applied nitrogen. Field Crops Res., 47: 1-12.

Cu, R., Mew, T.W., Cassman, K.G., & Teng, P.S. 1996. Effect of sheath blight on yield in tropical, intensive rice production systems. Plant Dis., 80: 1103-1108.

De Datta, S.K., Gomez, K.A. & Descalsota, J. 1988. Changes in yield response to major nutrients and in soil fertility under intensive rice cropping. Soil Sci., 146: 350-358.

Dobermann, A., Cassman, K.G. & Sta. Cruz, P.C. 1996. Fertilizer inputs, nutrient balance and soil nutrient supplying power in intensive, irrigated rice ecosystems. II. Effective soil K supplying capacity. Nutr. Cyc. Agroecosyst., 46: 11-21.

Dobermann, A., Cassman, K.G., Sta. Cruz, P.C., Adviento, M.A. & Pampolino, M.F. 1996a. Fertilizer inputs, nutrient balance and soil nutrient supplying power in intensive, irrigated rice systems. III. Phosphorus. Nutr. Cyc. Agroecosyst., 46: 111-125c.

Dobermann, A., Cassman, K.G., Peng, S., Pham Sy Tan, Cao Van Phung, Sta. Cruz, P.C., Bajita, J.B., Adviento, M.A.A. & Olk, D.C. 1996b. Precision nutrient management in intensive irrigated rice systems. In T. Attanandana et al., eds. Proceedings of the International Symposium on Maximizing Sustainable Rice Yields through Improved Soil and Environmental Management, Khon Kaen, Thailand, p. 38-52. Bangkok, International Society of Soil Science, Thailand Department of Agriculture, Soil and Fertilizer Society of Thailand, and Thailand Department of Land Development.

FAO. 1996. Report of the Expert Consultation on Technological Evolution and Impact for Sustainable Rice Production in Asia and the Pacific, 29-31 October 1996, Bangkok, Thailand. Rome.

Huber, D.M. & Arny, D.C. 1985. Interaction of potassium with plant disease. In R.D. Munson, ed. Potassium in agriculture, p. 467-488. Madison, Wisc., USA, American Society of Agronomy.

Mathews, R.B., Horie, T., Kropff, M.J., Bachelet, D., Centeno, H.G., Shin, J.C., Mohandass, S., Singh, S., Zhu Defeng & Moon Hee Lee. 1995. A regional evaluation of the effect of future climate change on rice production in Asia. In Mathews et al., eds. Modeling the impact of climate change on rice production in Asia, p. 95-139. Wallingford, UK, CAB International.

Mew, T. 1991. Disease management in rice. In D. Pimentel, ed. CRC handbook of pest management in agriculture, 2nd edition. Vol. 3, p. 279-299. Boston, Mass., USA, CRC Press.

Nambiar, K.K.M. & Ghosh, A.B. 1984. Highlights of research on long-term fertilizer experiments in India. LTFE Bulletin No. 1. New Delhi, Indian Council of Agricultural Research.

Olk, D.C., Cassman, K.G., Randall, E.W., Kinchesh, P., Sanger, L.J. & Anderson, J.M. 1996a. Changes in chemical properties of organic matter with intensified rice cropping in tropical lowland soil. Eur. J. Soil Sci., 47: 293-303.

Olk, D.C., Cassman, K.G., Simbahan, G., Sta. Cruz, P.C., S. Abdulrachman, S., Nagarajan, R., Pham Sy Than & Satawathananont, S. 1996b. Congruence of N fertilizer management by farmers and soil N supply in tropical lowland rice systems. In T. Attanandana et al., eds. Proceedings of the International Symposium on Maximizing Sustainable Rice Yields through Improved Soil and Environmental Management, Khon Kaen, Thailand, p. 29-38. Bangkok, International Society of Soil Science, Thailand Department of Agriculture, Soil and Fertilizer Society of Thailand, and Thailand Department of Land Development.

Peng, S., Garcia, F.V., Laza, R.C., Sanico, A.L., Visperas, R.M. & Cassman, K.G. 1996. Increased N-use efficiency using a chlorophyll meter on high-yielding irrigated rice. Field Crops Res. 47: 243-252.

Setter, T.L., Peng, S., Kirk,G.J.D., Virmani, S.S., Kropff, M.J. & Cassman, K.G. 1994. Physiological considerations and hybrid rice. In K.G. Cassman, ed. Breaking the yield barrier: Proceedings of a Workshop on Rice Yield Potential in Favorable Environments, p. 40-71. Los Ba�os, the Philippines, IRRI.

PREUVES SCIENTIFIQUES DE LA BAISSE DE RENDEMENT ET DE PRODUCTIVIT� DES MODES DE RIZICULTURE IRRIGU�E EN ASIE TROPICALE

Des exp�rimentations de longue dur�e ont document� la baisse de rendement des syst�mes de riziculture irrigu�e � double r�colte aux Philippines et en Inde. De r�cents rapports imputent cette tendance principalement � la baisse de l'approvisionnement en azote du sol, � l'�puisement d'autres macronutriments et � l'alourdissement du poids des maladies. Il faut signaler que cet approvisionnement en azote (N) par le sol baisse malgr� la conservation ou l'accroissement de la teneur du sol en mati�re organique. Cette rupture entre la quantit� de mati�re organique dans le sol et l'azote (N) disponible pour les plantes est particuli�re aux rizi�res de bas-fonds et est associ�e � des changements de la nature biochimique de la partie labile de la mati�re organique du sol suite � l'�tat ana�robie de ce m�me sol.

La pression accrue des maladies, allant de la pourriture � scl�rotes de la gaine � la pyriculariose, tient souvent au fait que, devant la baisse de l'approvisionnement en azote (N) du sol, il est n�cessaire d'accro�tre l'�pandage d'engrais azot�s afin de maintenir les rendements. Pour obtenir des rendements �lev�s et stables, des mesures de pr�vention s'imposent afin d'�viter qu'ils ne faiblissent lorsque les conditions climatiques sont propices au d�veloppement des maladies. Dans les champs des cultivateurs, la productivit� totale des facteurs a accus� une baisse spectaculaire dans quatre lieux de production intensive aux Philippines et en Inde. Cela est � pr�voir � mesure que les rendements moyens augmentent et que l'avantage marginal de facteurs de production additionnels diminue; toutefois, d'autres exp�rimentations au champ ont permis d'identifier des pratiques de gestion qui contribuent � une m�diocre productivit� des facteurs dans les lieux de production intensive aux Philippines, en Inde, en Indon�sie et au Viet Nam. Tout comme dans le cas des exp�rimentations de longue dur�e, on a constat� que l'approvisionnement en azote (N) par le sol des champs des riziculteurs n'avait pas de relation avec sa teneur en mati�re organique et que les rendements de riz sans adjonction d'engrais oscillaient entre 2 et 6 tonnes/ha dans des lieux de production relativement petits o� le type de sol est assez uniforme et o� le probl�me des ravageurs est n�gligeable. En outre, les planteurs n'adaptaient pas leurs �pandages d'engrais pour tenir compte des grandes diff�rences de teneur initiale du sol en �l�ments nutritifs et de l'effet sur les rendements d'applications d'engrais bas�es sur les r�sultats obtenus dans les parcelles d'observation sans application de nutriments.

Les parcelles exp�rimentales donnent une mesure g�n�rale de la qualit� du sol s'agissant de permettre un rendement de riz sans apport ext�rieur de nutriments. La diff�rence de rendement et d'absorption de substance nutritives entre ces parcelles non fertilis�es et l'ensemble du champ g�r� par le cultivateur sur le plan nutritif permet de quantifier l'efficacit� des engrais. Le plus grand �cart s�parait l'approvisionnement en azote (N) du sol et le taux d'application d'engrais azot�s par les fermiers, mais il y avait aussi plusieurs cas o� les d�s�quilibres de phosphore (P) et de potassium (K) r�duisaient les rendements et la productivit� des facteurs. Ces r�sultats ont �t� v�rifi�s dans 24 � 30 champs dans chacun des cinq lieux de production pendant trois p�riodes v�g�tatives cons�cutives. Ils indiquent qu'avec une gestion des nutriments adapt�e au champ vis�, le potentiel d'accroissement des rendements et de la productivit� des facteurs est consid�rable. Ils montrent aussi qu'avec l'application indiff�renci�e d'engrais recommand�e actuellement par la plupart des organismes et des services de vulgarisation publics pour un lieu de production donn�, les riziculteurs n'obtiennent pas un rendement optimal ni une fertilisation �conomique parce que la teneur initiale en nutriments varie consid�rablement d'un champ � l'autre.

En conclusion, les recommandations actuelles en mati�re de gestion des nutriments ne sont simplement pas adapt�es au caract�re dynamique de l'approvisionnement en nutriments du sol dans les bas-fonds de riziculture intensive en Asie tropicale et subtropicale. Pour soutenir l'accroissement des rendements, la viabilit� �conomique et la qualit� environnementale, il faudra adopter une gestion des nutriments adapt�e au champ donn�. Si cet examen a insist� sur les limites des nutriments et la gestion des �l�ments nutritifs, la d�marche de la recherche adopt�e dans les exp�rimentations � long terme et les �tudes de surveillance au niveau de l'exploitation peuvent aussi contribuer � identifier d'autres facteurs importants de nature � limiter les rendements, tels que les probl�mes de ravageurs, la m�diocre qualit� des semences et les propri�t�s physiques du sol. Cela est d� au fait que la mesure du gain de rendement ramen� au gain d'azote dans la plante (ou efficacit� physiologique de N) est, dans les vari�t�s de riz modernes, �troitement centr�e autour d'un gain d'environ 50 kg pour chaque kilogramme d'azote suppl�mentaire. Lorsque l'efficacit� physiologique est inf�rieure � ce chiffre, c'est parce que d'autres facteurs ont limit� les rendements (par exemple les maladies, le d�ficit hydrique), facteurs qui peuvent alors �tre identifi�s. De m�me, l'efficacit� de l'�pandage de N b�n�ficiant d'une bonne gestion agronomique devrait �tre sup�rieure � 20 kg de grain par kilo d'azote (N). Un sol m�diocre, les maladies des racines, les n�matodes ou d'autres facteurs qui limitent l'absorption d'azote (N) par le plant de riz r�duiraient donc cette valeur et des �tudes ult�rieures au niveau de la ferme pourraient les identifier.

DATOS CIENT�FICOS DEMUESTRAN UN DESCENSO DEL RENDIMIENTO Y DE LA PRODUCTIVIDAD EN SISTEMAS DE ARROZ DE REGAD�O DEL ASIA TROPICAL

Diversos experimentos de larga duraci�n han documentado el descenso de los rendimientos en sistemas de arroz de regad�o que producen dos cosechas en Filipinas y la India. La disminuci�n de la aportaci�n de nitr�geno al suelo, la reducci�n de otros macronutrientes y el aumento de la presi�n de las enfermedades son las causas principales de estas tendencias.
La aportaci�n de nitr�geno al suelo disminuye a pesar del mantenimiento o el incremento en �ste del contenido de materia org�nica. Esta desconexi�n entre el contenido de materia org�nica del suelo y la aportaci�n al suelo de nitr�geno disponible para las plantas es exclusiva de los suelos de tierras bajas dedicados al cultivo del arroz y est� asociada con cambios en la estructura qu�mica de fracciones l�biles de la materia org�nica del suelo como resultado de las condiciones anaerobias de �ste.

El aumento de la presi�n de enfermedades como el tiz�n de la vaina y el a�ublo suele ser el resultado de la necesidad de aumentar las dosis de fertilizantes nitrogenados para mantener los rendimientos ante la disminuci�n de la aportaci�n de N al suelo. Para conseguir rendimientos altos y estables es necesario adoptar medidas preventivas con el fin de evitar p�rdidas de rendimiento cuando las condiciones clim�ticas son propicias al desarrollo de la enfermedad. La productividad total de los factores en las explotaciones ha disminuido mucho con respecto a la situaci�n anterior en cuatro zonas dedicadas a la producci�n intensiva de arroz en Filipinas y la India. Aunque este es un resultado que cabe prever cuando aumentan los rendimientos medios y disminuyen los beneficios marginales de los insumos, sobre el terreno se han identificado pr�cticas de gesti�n que contribuyen a la escasa productividad de los factores en zonas dedicadas al cultivo intensivo de arroz en Filipinas, la India, Indonesia, Tailandia y Viet Nam.
Al igual que en los experimentos de larga duraci�n, se observ� que la aportaci�n de N al suelo en los campos de los agricultores no guardaba relaci�n con el contenido de materia org�nica del suelo y que los rendimientos del arroz sin la aplicaci�n de fertilizantes variaban entre 2 y 6 t/ha dentro de zonas de producci�n peque�as en las que el tipo de suelo era relativamente uniforme y los problemas de plagas eran insignificantes. Adem�s, los agricultores no modificaban las cantidades de fertilizantes aplicadas para tener en cuenta las grandes diferencias en la aportaci�n de nutrientes al suelo y la respuesta de los rendimientos a los fertilizantes aplicados tomando como base el rendimiento obtenido en parcelas de observaci�n establecidas en el campo de cada agricultor en las que no se hab�an aplicado nutrientes.

Las parcelas experimentales permiten medir aproximadamente la calidad del suelo en lo que respecta a su capacidad para mantener el rendimiento del arroz sin la aportaci�n externa de nutrientes. Las diferencias en el rendimiento y en la absorci�n de nutrientes entre estas parcelas a las que no se hab�an aplicado fertilizantes y los campos de los agricultores que hab�an recibido nutrientes permiten cuantificar la eficiencia de los fertilizantes. La discrepancia mayor se observ� entre la aportaci�n de N al suelo y la dosis de fertilizantes nitrogenados aplicada por los agricultores, pero tambi�n hubo varios casos en que los desequilibrios de P y K redujeron los rendimientos y la productividad de los factores. Estos resultados se verificaron en 24 a 30 campos en cada una de las cinco zonas de producci�n durante tres per�odos vegetativos consecutivos, e indican un potencial de aumento de los rendimientos y la productividad de los factores con un enfoque del manejo de nutrientes espec�fico para cada campo. Dadas las enormes variaciones en la aportaci�n natural de nutrientes al suelo de un campo a otro, las recomendaciones generales que hacen actualmente los organismos y servicios de extensi�n gubernamentales en cuanto a los fertilizantes que han de aplicarse a una determinada zona productora de arroz no se traducir�n en rendimientos �ptimos o en pr�cticas rentables en cuanto al uso de los fertilizantes.

Las recomendaciones que suelen hacerse sobre manejo de nutrientes no tienen en cuenta el car�cter din�mico de la aportaci�n de nutrientes a los suelos de las tierras bajas dedicadas a la producci�n intensiva de arroz en el Asia tropical y subtropical. Para mantener los incrementos de rendimiento, la viabilidad econ�mica y la calidad del medio ambiente, es necesario un enfoque del manejo de nutrientes espec�fico para cada campo. Aunque en este examen se ha hecho hincapi� en las limitaciones y el manejo de los nutrientes, el m�todo de investigaci�n utilizado en los experimentos de larga duraci�n y los estudios de vigilancia sobre el terreno permiten identificar tambi�n otros factores importantes, como los problemas de plagas, la calidad insuficiente de las semillas y las propiedades f�sicas del suelo que pueden limitar los rendimientos. Ello se debe a que en las variedades modernas de arroz el aumento del rendimiento por incremento unitario de la eficiencia fisiol�gica del nitr�geno en la planta se conserva estrictamente, con un aumento de unos 50 kg de arroz por cada kg de incremento de la absorci�n de N. Cuando la eficiencia fisiol�gica es inferior, ello significa que hay otros factores (como por ejemplo una enfermedad, un d�ficit de agua) que limitan el rendimiento, factores que es posible identificar entonces. Del mismo modo, la eficiencia de un fertilizante nitrogenado con una buena gesti�n agron�mica ser� superior a 20 kg de grano por cada kg de N aplicado. Ese valor se ve reducido por el mal estado del suelo, enfermedades de las ra�ces, nematodos u otros factores que limitan la absorci�n de N, los cuales podr�an identificarse mediante estudios posteriores sobre el terreno.