Rice yields and yield gaps in Southeast Asia: Past trends and future outlook

Europ. J. Agronomy 36 (2012) 9–20

Contents lists available at SciVerse ScienceDirect

European Journal of Agronomy journal homepage: www.elsevier.com/locate/eja

Rice yields and yield gaps in Southeast Asia: Past trends and future outlook Alice G. Laborte a,b,∗ , Kees (C.A.J.M.) de Bie a , Eric M.A. Smaling a , Piedad F. Moya b , Anita A. Boling b , Martin K. Van Ittersum c a b c

Faculty of Geo-Information Science and Earth Observation (ITC), University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands International Rice Research Institute, Los Ba˜ nos, Laguna, Philippines Plant Production Systems, Wageningen University, P.O. Box 430, 6700 AK Wageningen, The Netherlands

a r t i c l e

i n f o

Article history: Received 17 March 2010 Received in revised form 6 August 2011 Accepted 24 August 2011 Keywords: Food production Learning curve Rice Southeast Asia Yield gap

a b s t r a c t Rice production must increase to meet future food requirements amid strong competition for limited resources. Yield gap analysis is a useful method to examine how large the ranges are between potential, desirable rice yields and those actually realized in farmers’ fields. We analyzed farmers’ yields in wet and dry seasons in four intensively cropped rice areas in Southeast Asia and explored opportunities for reducing the yield gap to meet future food requirements. We found yield gaps of 2.0–5.0 t ha−1 between average and climatic yield potential and 1.2–2.6 t ha−1 between average and best farmers’ yields. In relative terms, average yields varied between 43% and 75% of the climatic yield potential and 61% and 83% of the best farmers’ yields. Farmers with best yields were generally more educated, and used fertilizers and labor more efficiently than average farmers. The yield gaps between average and best farmers’ yields are higher in rice-importing countries (Indonesia and Philippines) compared with rice-exporting countries (Thailand and Vietnam). Assuming no change in diet, closing the existing yield gap between average and best-yielding farmers can sufficiently cover the yield increase needed for 2050 in the three countries, except for the Philippines, where yield increase must be even higher. Trend analysis of yield increases of a population of farmers in Central Luzon (Philippines), which included a learning curve analysis, well described the process of technology adoption from 1966 to 2008, leading to higher yields. Using this analysis, for the Philippines, we predicted yields to increase (from 2007/2008 to 2050) by only 18% with current cultivars, production technologies, and prevailing conditions. Therefore, structural changes are needed to boost farmers’ yields to close the yield gap faster. Investments in technology transfer and institutional arrangements are suggested. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Global agricultural production must increase by 70% to meet demand by 2050 (Bruinsma, 2009). Approaches to bridge the gap between projected demand and current level of supply include (1) expansion of land under cultivation, (2) intensification on existing farmland by growing two or three crops a year, (3) narrowing the yield gap in farmers’ fields, (4) raising the yield ceiling by introducing higher yielding varieties, and (5) reducing postharvest losses and food waste (Koning and Van Ittersum, 2009). The first two approaches result in higher output with higher input (e.g., land, capital), whereas, the next two result in higher output due to pro-

∗ Corresponding author at: International Rice Research Institute, Los Banos, ˜ Laguna, Philippines. Tel.: +63 2 580 5600; fax: +63 2 580 5699. E-mail addresses: [email protected], [email protected] (A.G. Laborte). 1161-0301/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.eja.2011.08.005

duction efficiencies. The third approach could also involve higher inputs, depending on the current level of input use. Making farming more efficient as well as reducing waste (fifth approach) were also cited by Smil (2000) as ways to increase future food supply. In the developing world, 80% of the growth in crop production is expected to come from higher yields and increased cropping intensity and the remaining 20% is expected from land expansion (Bruinsma, 2009). This estimate does not yet consider the impending competition between food and energy for limited land and water resources. Although growth in global production has kept pace with population growth, a continuous linear increase in yields as in the past five decades will not be sufficient to meet food, feed, and fuel needs in the next 40 years (Fischer et al., 2009). About half of the world’s cereal production and 89% of the world’s harvested rice are from Asia. Various studies have indicated that arable land in the region may still be expanded by 20–30% over 1990 levels, but, in reality, less than half of this land may be suitable for cultivation (Young, 1999). Moreover, expansion of cultivated land has associated environmental costs such as deforestation,

10

A.G. Laborte et al. / Europ. J. Agronomy 36 (2012) 9–20

biodiversity loss, and soil degradation. Agricultural intensification, on the other hand, would require ample freshwater supply to allow multiple cropping in a year. In 2000, total crop water requirement was 7130 km3 and water withdrawn from surface water and groundwater resources was 2630 km3 (Comprehensive Assessment of Water Management in Agriculture, 2007). Considering current water productivity and production patterns, total crop water requirement is projected to increase to 12,050–13,500 km3 by 2050 (Comprehensive Assessment of Water Management in Agriculture, 2007). In some areas, however, the water tables have been declining and this situation is expected to get worse in future. Groundwater tables have been falling by 0.5–1 m year−1 in some states in India and by 1–3 m year−1 in some parts of northern China (Tuong and Bouman, 2003). Conventional cultivation of rice, the staple crop in Asia, requires at least double that of water needed for irrigating other food crops such as maize and wheat. Approximately 150 million ha of crop land in Asia is irrigated (Bouman et al., 2007) and rice accounts for 40–46% of the net irrigated area of all crops (Dawe, 2005). By 2025, about 15 million ha of irrigated rice in China and some parts of South Asia are estimated to experience “physical water scarcity” and 22 million ha in Asia may face “economic water scarcity” as irrigation water becomes too expensive for rice farmers in the future (Tuong and Bouman, 2003). Water-saving technologies for rice have been developed (e.g., alternate wetting and drying, saturated soil culture, aerobic rice), but these result in yield decline and cause weed problems (Tuong and Bouman, 2003; Bouman et al., 2007). The genetic yield potential of inbred rice has stagnated since the introduction of the IR8 variety in the late 1960s (Peng et al., 1999) in spite of efforts to raise its yield ceiling. Among these are the development of a new plant type (rice with low tillering capacity and large panicles) and C4 rice (incorporation of the C4 photosynthetic pathway into rice), which promise higher yields and more efficient input use (Peng et al., 1999; Hibberd et al., 2008). While these innovations are not yet available and, in the case of C4 rice, will probably take more than a decade to be realized, substantial increases in production must come from reductions in the yield gap in farmers’ fields. It has been estimated that reducing the yield gap alone could provide an additional 60% rice by 2025 (Chaudhary, 2000). The term “yield gap” has been commonly used to refer to the difference between the average farmers’ yields and an estimate of a reference yield (potential or water-limited) at a specific area in a given time. Potential yield (Van Ittersum and Rabbinge, 1997) can be defined and measured in a variety of ways such as using crop growth models, maximum yield trials, and other research experiments, or best yields from farmers’ fields as in Lobell et al. (2009). Yield gaps exist because the best available production technologies are not adopted in farmers’ fields. This could be, among others, due to farmers’ characteristics (e.g., lack of knowledge and skills, risk aversion), farm characteristics (e.g., poor soil, difficult terrain, inaccessibility), and inappropriateness of the technology to farmers’ circumstances (e.g., labor-intensive, high investment costs, poor access to inputs). Because average crop yields are critical drivers of food prices, cropland expansion, and food security, yield gaps should be better quantified and understood (De Bie, 2000; Nidumolu, 2004; Lobell et al., 2009). This paper aims to analyze farmers’ yields across time (year and season) in four intensively cropped rice areas in Southeast Asia. Specifically, yield gaps were quantified and trends in increases in farmers’ yields were analyzed to assess opportunities for increasing rice production using existing technologies to meet future food requirements. Prospects for reducing yield gaps in farmers’ fields and the implications on national food self-sufficiency and food availability in 2050 are discussed.

2. Data and methods 2.1. Data sets Time series of farm survey data available for different years and for the wet and dry seasons were used (Table 1). Farm survey data for Suphan Buri (Thailand), West Java (Indonesia), and Can Tho (Vietnam) were from the “Reversing Trends of Declining Productivity in Intensively Irrigated Rice Systems (RTDP)” project, which was conducted at nine sites in Asia. Data refer to the wet and dry seasons from 1994 to 1999. All of the farms surveyed were irrigated. Details of the RTDP farm surveys are described in Dobermann et al. (2004) and Moya et al. (2004). The data for Central Luzon, Philippines, came from another project and had a longer time series compared with the other three sites. The loop survey, which includes Central Luzon as one of the sites, started in 1966 and is conducted every 4 years. The most recent survey refers to the wet season of 2008. Some of the farms surveyed in the early years in Central Luzon were not irrigated, but starting in the 1980s, all of the farms surveyed were irrigated. Daily solar radiation and maximum and minimum temperature from weather stations in Nueva Ecija (Central Luzon, 1987–2000), Suphan Buri (1985–1999), Sukamandi (1985–2000) and Can Tho (1985–1999) were used as inputs to the ORYZA2000 crop growth simulation model. Current demand and supply for rice as well as estimates of future food requirements were based on rice production and consumption and trade data from FAOSTAT (FAO, 2011a) and population projections from the UN (2011).

2.2. Study area The study area consists of four sites where rice is intensively cultivated. There is one site each in the major rice-importing (Indonesia and Philippines) and -exporting countries (Thailand and Vietnam) in Southeast Asia (Fig. 1 and Table 1).

2.2.1. Central Luzon, Philippines This site consists of six provinces: Bulacan, La Union, Nueva Ecija, Pampanga, Pangasinan, and Tarlac, covering a total land area of 20,000 km2 . The provinces are actually part of two administrative regions: Region I (Ilocos) and Region III (Central Luzon), but here, the six provinces will be collectively referred to as Central Luzon. In 2010, total harvested rice area in Central Luzon was 0.9 million ha (double-cropped areas are counted twice) and average rice yield was 4.2 t ha−1 per crop, 17% higher than the national average. Eighty-one percent of the rice areas are irrigated (BAS, 2011). Two crops of rice are grown each year: wet-season rice from June/July to September/October and dry-season rice from December/January to March/April (farm survey). Average farm size in 2008 was 1.2 ha, down by about one-third of a hectare from the average farm size in 1999 (farm survey).

2.2.2. West Java, Indonesia The province of West Java is one of the most important rice-growing areas of Indonesia. In 2010, harvested rice area in the province was 2.0 million ha (double-cropped areas are counted twice) and average rice yield was 5.8 t ha−1 , 15% higher than national average (Ministry of Agriculture, 2011). Two crops of rice are grown in a year: wet-season rice is from November/December to February/March and dry-season rice from April/May to July/August (farm survey). Surveyed farms are located in three villages in Sukamandi in the district of Subang: Karang

A.G. Laborte et al. / Europ. J. Agronomy 36 (2012) 9–20

11

Table 1 Farm surveys. Site

Average number of farmers surveyed per season

Years covered by season

Source

Central Luzon, Philippines (6 provinces)

Wet season: 100

IRRI Loop survey

West Java, Indonesia

Wet season: 25 Dry season: 24

Wet season (11 years): 1966, 1970, 1974, 1979, 1982, 1986, 1990, 1994, 1999, 2003, 2008 Dry season (9 years): 1967, 1971, 1980, 1987, 1991, 1995, 1998, 2004, 2007 Wet season (4 years): 1995–1996, 1998–1999 Dry season (4 years): 1995, 1997–1999

Suphan Buri, Thailand

Wet season: 22 Dry season: 22

Wet season (5 years): 1994–1995, 1997–1999 Dry season (5 years): 1995–1999

IRRI RTDP Project

Can Tho, Vietnam

Wet season: 25 Dry season: 28

Wet season (5 years): 1994–1995, 1997–1999 Dry season (5): 1995–1996, 1998–1999

IRRI RTDP Project

Dry season: 51

IRRI RTDP Project

RTDP: Reversing Trends of Declining Productivity in Intensively Irrigated Rice Systems. See Dobermann et al. (2004) and Moya et al. (2004) for a description of the survey data sets.

Hegar, Sukareja, and Bojongjaya. Average farm size in this area in 1999 was 0.7 ha (farm survey). 2.2.3. Suphan Buri, Thailand The province of Suphan Buri, located in Central Thailand, accounts for only 5% of total rice production of Thailand, but, the average rice yield in the province was 61% higher than that of the country. In 2006, total planted rice area was 0.3 million ha (doublecropped areas are counted twice) and average rice yield was 4.4 t ha−1 (FAO, 2011b). Two to three crops of rice are grown each year—i.e., during the dry season from December to March/April, early wet season from March to June, and wet season from July/August to November/December (farm survey). Surveyed farms are located in Muang, Sriprachan, and Donchedi districts. Average farm size in 1999 was 4 ha (farm survey). 2.2.4. Can Tho, Vietnam The province of Can Tho is located in the Mekong River Delta, which accounts for more than half of the total rice production of Vietnam. In 2009, 0.2 million ha was planted to rice (doublecropped areas are counted twice) with an average yield of 5.2 t ha−1 , 4% higher than the national average (GSO, 2011). The farm survey was conducted in six villages in Omon district where two to three crops of rice are grown each year—i.e., during the dry season from November/December to February/March, early wet season from March/April to June/July, and late wet season from May/June to August/September (farm survey). Average farm size in 1999 was 1 ha (farm survey). 2.3. Trend analysis of farmers’ yields Time series of rice yield data were analyzed and rates of yield increases were calculated for Central Luzon, where longer timeseries farm survey data are available. To get a clear picture of the changes over time, we used quantile regression to compute for the different regression lines that correspond to various percentage points of the distributions of farmers’ yields (Koenker and Hallock, 2001). In addition, we used nonlinear regression to model farmers’ yields in relation to time and the diffusion of successful innovations over time according to the S curve (learning curve). Initially, it follows a slow cumulative adoption, then a fast accumulation, and then flattens out at some point as the maximum level of adoption is reached. Portions of this diffusion curve can be attributed to different types of adopters—i.e., innovators, early adopters, early majority adopters, late majority adopters, and laggards (Rogers, 2003). We assume that the diffusion of modern rice varieties intro-

duced in the 1960s followed this pattern. We hypothesize that, initially, there are more farmers getting low yields (subsistence level) while there are only a few farmers achieving high yields (positively skewed distribution) and that this distribution will change over time (Fig. 2). With improved conditions such as availability of higher yielding varieties and accessibility to markets and credits, the entire density curve will shift to the right, representing higher yields. Also, given the time and positive perception on the benefits of new technologies, more farmers will adopt the new technologies, and there will be a change in the shape of the distribution of farmers’ yields from positively skewed to negatively skewed (i.e., more farmers will have relatively high yield levels but there will still be a few achieving lower yields). Skewness was used as an input to model yields and not made dependent on time in the nonlinear regression analysis. Therefore, we model farmers’ yields as follows: Yti = a ln(t + b) +



1 1 + ecSt



+ε

where Yti is the yield at the tth year of the ith farmer; t is the number of years from the base year (1966); St is the skewness coefficient of farmers’ yields in year t; the coefficients a, b, and c are parameters of the nonlinear model; and ε is the error term. The first term accounts for the logarithmic increase in yield with time, whereas the second term accounts for the S curve in the diffusion of new technologies. The model was tested using farm survey data for Central Luzon, Philippines. We performed the regression analyses in the R statistical environment using the following R packages: quantreg for quantile regression (Koenker, 2009) and nls2 to estimate the parameters of the nonlinear model (Grothendieck, 2007). 2.4. Yield gap in farmers’ fields Three yield gaps were estimated based on climatic yield potential, economic yield, and best farmers’ yields (Fig. 3). Climatic yield potential refers to the maximum (theoretical) yield that can be attained, given the genotype and prevailing climatic conditions with no other production constraints (Van Ittersum and Rabbinge, 1997). This can be estimated using crop growth models or experimentally through maximum yield trials. In this paper, we calculated climatic yield potential from 1985 to 2000 using the ORYZA2000 crop growth model (Bouman et al., 2001). The model components for potential production are based on the SUCROS concept—i.e., the daily rate of canopy CO2 assimilation is calculated from daily incoming radiation, temperature, and leaf area index. The model contains a set of subroutines that

12

A.G. Laborte et al. / Europ. J. Agronomy 36 (2012) 9–20

Fig. 1. Farm survey sites.

Fig. 2. Conceptual framework for the change in distribution of farmers’ yields in time. Each plot shows the frequency distribution of farmers’ yields at a given time t.

A.G. Laborte et al. / Europ. J. Agronomy 36 (2012) 9–20

13

Table 3 Non-linear regression results on farmers’ yields in Central Luzon, Philippines from 1966 to 2008. Parameter a b c Residual standard error Degrees of freedom * **

Fig. 3. Conceptual comparison between climatic yield potential, economic yield goal, and farmers’ yields. Three definitions of yield gaps are indicated based on the differences between average farmers’ yields and climatic yield potential (YieldGapp ), economic yield goal (YieldGape ), and best farmers’ yields (YieldGapf ). Adapted from Lobell et al. (2009).

calculate the daily totals by integrating instantaneous rates of leaf CO2 assimilation. The model has been used extensively to simulate rice production in regions in Southeast Asia (Bouman and Van Laar, 2006; Boling et al., 2010; Wikarmpapraharn and Kositsakulchai, 2010). For each site and season, we used standard crop parameters for rice variety IR72 (Bouman et al., 2001; Bouman and Van Laar, 2006), crop development rates calculated from observed phenological stages, actual crop establishment method and actual average planting dates of farmers, solar radiation, and maximum and minimum temperature. Based on the farm surveys from 1994 to 1999, average planting dates for the wet and dry seasons were July 18 and January 6 for Central Luzon, December 5 and May 12 for Sukamandi, July 30 and December 18 for Suphan Buri, and April 30 and November 22 for Can Tho. Typically, potential rice yields in the subhumid to humid subtropical and tropical regions in Asia range from 9 to 11 t ha−1 during the dry season and from 6 to 8 t ha−1 during the wet season (Dobermann and Witt, 2004). In general, the higher simulated potential yields in the dry season are caused by higher solar radiation levels. In our simulations for West Java, yields in the wet season were higher than in the dry season, yet, simulated yields were within the range of values in Dobermann and Witt (2004): i.e., 8–9 t ha−1 and 6.5–7.5 t ha−1 for the wet and dry season, respectively. The simulations for this site show that the crop growth duration was longer and accumulated radiation was higher in the wet season compared with the dry season. For the other three sites, simulations for the wet season were within or close to the range of values for subhumid to humid subtropical and tropical regions in Asia. However, for the dry season, our simulations are lower. This could be due to the differences in planting dates. Previous studies have shown large differences in potential yield estimates due to differences in crop establishment dates (Kropff et al., 1993; Boling et al., 2007). Economic yield goal is the target yield that is still considered economically viable for farmers to attain (cf. De Koeijer et al., 1999). In this study, we estimate economic yield goal as 80% of the climatic yield potential, though the precise level depends on price ratios. Yields beyond 80% of the climatic yield potential have lower nutrient use efficiencies (De Wit, 1992; Witt et al., 1999), and those within 70–80% are, in most cases, associated with the highest profits (Dobermann and Witt, 2004; Lobell et al., 2009). The best farmers’ yields refer to the average of the upper 10 percentile in each year and each season. This could mean that farmers

Wet season 0.98** 8.44** 2.05** 1.29 1,100

Dry season 1.18** 4.17* 1.02 1.53 459

Significant at 5% level. Significant at 1%.

with best yields can differ each year and each season, depending on their performance. From hereon, best-yielding farmers in a particular year and season refer to those farmers with rice yields belonging to the upper 10 percentile. 2.5. Future rice requirements Assessments of future rice requirements were done at the country level for the Philippines, Indonesia, Thailand, and Vietnam. Estimates of future domestic demand were based on population projections for 2050 and the latest available data on per capita total rice consumption (2007); we assumed no change in diet. The additional domestic demand for rice in 2050 to attain self-sufficiency was calculated by multiplying the difference in population between 2050 and 2009 with per capita total rice consumption (2007), and adding the current imports (2009). 3. Results 3.1. Growth in rice yields Rice yields in farmers’ fields in the case study sites were generally higher during the dry season than during the wet season, except for West Java (Table 2). For Central Luzon, average wetseason rice yields increased from 2.3 t ha−1 in the late 1960s to 4.5 t ha−1 in 2008, whereas the dry-season rice yields increased from 1.9 to 5.1 t ha−1 (Table 2). The distribution of yields was positively skewed in the late 1960s and became less skewed and wider in range of values after the 1960s (Fig. 4). The quantile regression for Central Luzon shows that the median yield during the wet season had been increasing annually by 0.1 t ha−1 from 1966 to 1979 and slowed down to only 0.02 t ha−1 from 1982 to 2008 (Fig. 5). The ordinary least squares (OLS) estimate does not show the annual change in yields for different yield quantiles. Yields increased at a much faster rate for the upper quantiles compared with the lower quantiles from 1966 to 1979 (Fig. 5a). For the upper 10 percentile, yields increased annually by 0.18 t ha−1 , nearly twice that of the median. In contrast, there was no significant change in yields for the lowest 10 percentile, which implies that, for more than a decade since the introduction of modern varieties, there was no improvement in yields for some farmers. From 1982 onwards, the lower 30 percentile had slightly higher coefficients and further, the annual rates of change were not so different across quantiles (Fig. 5b). Results of the nonlinear regression for Central Luzon show significant positive coefficients for the first term of the equation (a * ln(t + b)) for both seasons (Table 3). This confirms that farmers’ yields had been increasing logarithmically with time but the increase was faster in the dry season compared with the wet season. The coefficient for the second term (1/(1 + ebS )) was significant for the wet season, confirming that farmers’ learning of improved production technologies follows an S curve. However, for the dry season, the coefficient for this term was lower and not significant. This could be due to a shorter time series and fewer records per

14

A.G. Laborte et al. / Europ. J. Agronomy 36 (2012) 9–20

Table 2 Comparison of rice yields (t ha−1 ) from farmers’ fields and research stations or on-farm trials. Survey site

Year

Wet season

Dry season

Farmers’ yields Average Central Luzon, Philippines

West Java, Indonesia

Suphan Buri, Thailand

Can Tho, Vietnam

a

Farmers’ yields a

Best yields

1966 1967 1970 1971 1974 1979 1980 1982 1986 1987 1990 1991 1994 1995 1998 1999 2003 2004 2007 2008

2.3

3.8

2.6

4.2

2.2 3.8

4.6 6.6

4.1 3.6

6.2 5.5

3.7

5.8

4.0

6.4

4.5

6.4

1995 1996 1997 1998 1999

5.9 5.5

7.7 7.9

5.8 5.3

7.9 9.1

1994 1995 1996 1997 1998 1999

5.4 4.6

6.4 6.2

5.0 4.8 5.3

6.2 5.8 6.3

1994 1995 1996 1997 1998 1999

4.2 4.4

5.4 7.3

3.8 3.8 3.6

4.8 4.8 5.3

3.4 4.3

Average

Best yieldsa

1.9

3.4

2.9

5.6

4.3

6.9

4.3

6.7

4.6

6.8

4.7 4.6

7.7 7.2

4.5 5.1

7.1 7.4

4.1

5.7

5.4 2.8 4.4

7.8 4.9 6.4

5.3 5.2 4.3 5.2 5.5

6.4 6.3 5.6 6.6 6.9

6.2 6.2

7.6 7.2

6.1 5.6

7.5 6.7

6.1 6.4

Average for high-yielding farmers (upper 10 percentile).

year in the dry season. Fig. 6 shows the predicted yields based on partial (including the first term only) and full model specifications. The result of the full model correlates well with the average farmers’ yields in each year for both seasons. The Pearson’s correlation coefficients between predicted yields from the full model and average farmers’ yields were 0.92 and 0.98 for the wet and dry season, respectively. 3.2. Yield gap in farmers’ fields The gap between average and climatic yield potential (YieldGapp ) in the 1990s ranged from 2.8 to 5.0 t ha−1 (average yields were 43–67% of potential) in the wet season and 2.0 to 5.0 t ha−1 (48–75% of potential) in the dry season (Table 4). The gap with the economic yield goal (YieldGape ) ranged from 1.1 to 3.3 t ha−1 in the wet season and from 0.4 to 3.1 t ha−1 in the dry season. Among all sites, Central Luzon had the highest yield gap between average farmers’ yields and climatic yield potential as well as between average yield and economic yield goal for both seasons. Compared with the best-yielding farmers, the gap (YieldGapf ) ranged from 2 to 2.6 t ha−1 (average yields were 61–69% of best farmers’ yields) for the sites in the importing countries (Central Luzon and West Java) and only 1.2–1.6 t ha−1 (71–83% of best

farmers’ yields) for the sites in the exporting countries (Suphan Buri and Can Tho) (Table 4). Best-yielding farmers were already obtaining yields near or, in some cases, over the level of the economic yield goals in the four sites. Based on available data for 1995, farmers with the best yields had, on average, more years of schooling than average farmers (Table 5). The only exceptions were Central Luzon (both seasons) and Suphan Buri (wet season), where average farmers spent a fraction of a year more in school than the best-yielding farmers. Except for Suphan Buri and Can Tho (dry season), the best yields were from smaller farms. In the dry season, farmers who obtained the best yields were using more fertilizers per hectare in all sites. In the wet season, the difference was not as consistent but farmers with best yields were producing more rice per unit of nitrogen and labor input used. The adoption of modern varieties led to higher yield gaps between average and best-yielding farmers. Compared with the mid 1960s, YieldGapf increased in 1970s by 0.7 t ha−1 in the wet season and 1.2 t ha−1 in the dry season as more farmers started adopting modern rice varieties (Table 6). The gap was lower than in the 2000s but still higher than in the late 1960s as the late adopters of the modern varieties started catching up. Best farmers’ yields increased by 25–30% from 1970s to 2000s, and average farmers’ yields grew much faster (52–66%) in both seasons.

A.G. Laborte et al. / Europ. J. Agronomy 36 (2012) 9–20

15

Fig. 4. Distribution of farmers’ rice yields in Central Luzon, Philippines, selected years from 1966 to 2008. Vertical lines refer to average yields for the year and season.

4. Discussion and conclusion In recent decades, much of the growth in agricultural production has been due to the growth in yields (per hectare per crop) and will increasingly need to do so in the future (Bruinsma, 2009). Many farmers in developing countries are still cultivating rice at subsistence level. Thus, the yield gap needs to be narrowed to increase production and ensure availability of food at affordable prices. We found wide yield gaps between average farmers’ yield and climatic yield potential in the four study sites. The yield gap ranged from 2.0 to 5.0 t ha−1 or average yields were 43–75% of potential yields. This is within the range of yield gap values of Asian rice farmers reported by Lobell et al. (2009). Climatic potential yields (and hence yield gaps between average farmers’ yield and climatic yield potential) could be higher than presented here by using optimal crop establishment dates rather than actual planting dates. We did not optimize planting dates to calculate maximum potential yields in our simulations

because farmers do not normally follow optimal planting dates for various reasons. In Central Luzon, for instance, farmers rely on the release of water from large reservoirs. In some areas, however, this is determined based on the requirements for electrical power generation and domestic water use rather than to maximize potential rice yields. For policy analysis and projections of future rice production potential, yield gaps based on actual crop establishment dates seem more relevant. The gaps between average farmers’ yield and climatic yield potential cannot be closed entirely because achieving the potential yields is not economically attractive to farmers. In addition to climatic yield potential, we also calculated yield gaps based on economic yield goal and best farmers’ yields. These two yield levels are both attainable by farmers, hence, are better reference yields for calculating bridgeable yield gaps. Best farmers’ yields were either less than 1 t lower or, in some cases, higher than our estimate of the economic yield goal in the different sites and seasons. On average, best-yielding farmers were more educated than average farmers

16

A.G. Laborte et al. / Europ. J. Agronomy 36 (2012) 9–20

Fig. 5. Quantile regression results: coefficient of time (t year−1 ) vs. quantile of rice yields in farmers’ fields in Central Luzon, wet season, 1966–1979 (a) and 1982–2008 (b). The horizontal dashed line refers to the ordinary least squares (OLS) estimate and the other two horizontal dotted lines represent the 90% confidence interval of the estimate. Each dot represents every 2 percentile points. The shaded gray area refers to the 90% confidence band for the quantile regression estimates.

Fig. 6. Actual and predicted rice yields based on full and partial non-linear regression models, Central Luzon, Philippines, 1966–2008, by season. Error bars refer to the standard deviation of actual farmers’ rice yields.

Table 4 Rice yield gaps in farmers’ fields in intensively cropped systems in the 1990s.a Central Luzon, Philippines

West Java, Indonesia

Suphan Buri, Thailand

Can Tho, Vietnam

Wet season

Dry season

Wet season

Wet season

Wet season

Yield (t ha ) Average farmers Best farmers Economic goal Climatic potential

3.7 6.1 7.0 8.7

4.6 7.2 7.7 9.6

5.6 8.1 6.7 8.4

4.2 6.2 6.2 7.7

5.0 6.2 6.4 8.0

5.1 6.4 6.6 8.2

3.9 5.5 6.2 7.8

6.0 7.2 6.4 8.0

Yield gap (t ha−1 ) YieldGapf YieldGape YieldGapp

2.4 3.3 5.0

2.6 3.1 5.0

2.5 1.1 2.8

2.0 2.0 3.5

1.2 1.4 3.0

1.3 1.5 3.1

1.6 2.3 3.9

1.2 0.4 2.0

Dry season

Dry season

Dry season

−1

Average yield as % of Best farmers’ yields Economic yield goal Potential yield a

61 53 43

64 60 48

69 84 67

Data refer to averages for the 1990s where available (see Table 2).

68 68 55

81 78 63

80 77 62

71 63 50

83 94 75

A.G. Laborte et al. / Europ. J. Agronomy 36 (2012) 9–20

17

Table 5 Education, farm size, and input use associated with average and best farmers’ yields during wet and dry seasons in 1995. Season/indicator

Wet season Yield (t ha−1 ) Years in school Farm size (ha) Input use N fertilizer (kg ha−1 ) Labor (days ha−1 ) Input use efficiency (kg rice per unit input) N fertilizer Labor Dry season Yield (t ha−1 ) Years in school Farm size (ha) Input use N fertilizer (kg ha−1 ) Labor (days ha−1 ) Input use efficiency (kg rice per unit input) N fertilizer Labor a

Central Luzon, Philippinesa

West Java, Indonesia

Suphan Buri, Thailand

Can Tho, Vietnam

Average

Average

Average

Average

Best yields

4.0 7.1 1.8

6.4 6.2 1.1

Best yields

5.9 7.3 1.3

7.7 8.0 1.0

4.6 4.7 2.1

Best yields 6.2 4.0 2.1

4.4 6.8 0.9

Best yields 7.3 7.3 0.7

93 70

111 114

104 103

84 119

91 12

101 14

114 80

111 86

43 57

58 56

57 57

92 65

51 383

61 443

38 55

66 85

4.7 6.7 1.6

7.7 5.7 0.9

4.1 7.3 1.2

5.7 11.0 0.9

5.3 4.7 2.0

6.4 5.0 2.1

6.2 6.8 0.9

7.6 10.5 1.8

122 55

162 84

124 107

142 89

112 17

120 17

89 61

97 54

39 85

48 91

33 38

40 64

47 312

53 376

70 102

78 141

Wet season of 1994.

(though not consistently for all regions). They apply fertilizer and use labor more efficiently. Input use inefficiencies have been found to be related to farmers’ information and skills and input supply problems (Ali and Byerlee, 1991). With limited data, we were not able to do a more comprehensive analysis of the differences in farmers’ characteristics and farm management between average and best-yielding farmers. Several studies have attempted to explain the rice yield gaps in farmers’ fields in different sites. A study in the mid-1970s in six Asian countries found yield gaps varying from 0.4 to 2.2 t ha−1 between farmers’ practice and on-farm trials with high inputs of fertilizer, weed, and insect control (Herdt, 1979). The average yield gap was 0.9 t ha−1 (25% yield difference) in the wet season and 1.3 t ha−1 (30% yield difference) in the dry season. The study concluded that it was not economically attractive for farmers to apply high rates of fertilizer and insect control, which are needed to close the gap in the wet season. But with a larger average yield gap, it was, in most cases, economically attractive to use more chemical inputs in the dry season (Herdt, 1979). De Bie (2000) found a yield gap of 2.6 t ha−1 (90% yield difference) between the average and best yields of farmers who were growing sticky rice in northern Thailand in 1992. The yield gap was attributed to water shortage (41%), incidence of diseases (22%), untimely planting (18%), lodging (10%), and poor soil quality (8%). In a similar analysis for southeastern India, Nidumolu (2004) found

a yield gap of 2.1 t ha−1 (33% yield difference) between average and best rice yields and attributed the yield gap to water shortage (38%), number of fertilizer applications (22%), date of harvesting (21%), and second weeding (19%). In both studies, water shortage was the most important constraint to high rice yields. Another common attribute considered was the cropping calendar (date of planting/harvesting). Similar to the findings of Nidumolu (2004), studies conducted elsewhere have shown that nutrients are often a yield-limiting factor (Regmi et al., 2002; Van Ittersum et al., 2003). The cropping calendar cannot be the cause of the yield gap between average farmers’ yields and climatic yield potential in our study because we used the average crop establishment dates as in the farm survey. It is possible that water and nutrient limitations partly caused the yield gap. In simulations for Central Java, Indonesia, Boling et al. (2010) found that yield gaps caused by water limitations ranged from 0 to 28% and that caused by nitrogen limitations ranged from 35 to 63%. Other studies have listed more constraints that contribute to yield losses in farmers’ fields (Fischer et al., 2009; Lobell et al., 2009). Some of these constraints relate to farmers’ characteristics (e.g., risk aversion, skills) and environmental conditions (e.g., soil problem, weather-related conditions). There is also evidence that yield gaps do not arise from consistent factors each year (Lobell et al., 2009).

Table 6 Gap between average and best farmers’ rice yields in Central Luzon, Philippines from 1966 to 2008. Season

Decade Mid-1960s

1970s

1980s

1990s

2000s

Wet season Average farmers’ yields (t ha−1 ) Best farmers’ yields (t ha−1 ) Yield gap (YieldGapf , t ha−1 ) Average as % of best farmers’ yields (%)

2.3 3.8 1.5 61

2.9 5.1 2.2 57

3.8 5.9 2.1 64

3.7 6.1 2.4 61

4.4 6.4 2.0 69

Dry season Average farmers’ yields (t ha−1 ) Best farmers’ yields (t ha−1 ) Yield gap (YieldGapf , t ha−1 ) Average as % of best farmers’ yields (%)

1.9 3.4 1.5 56

2.9 5.6 2.7 52

4.3 6.8 2.5 63

4.6 7.2 2.6 64

4.8 7.3 2.5 66

18

A.G. Laborte et al. / Europ. J. Agronomy 36 (2012) 9–20

Table 7 Current and projected rice demand and supply in four countries in Southeast Asia. Indicator

Current situation Population, 2009 (million persons) Demand Rice consumptiona , 2007 (kg capita−1 ) Food All uses Supply Rice production per capita, 2007 (kg capita−1 ) Total rice production, 2009 (million t) Rice trade as % of productiona , 2007 Import Export Rice area, 2009 (million ha) Rice yield, 2009 (t ha−1 ) Projection for 2050 Average population growth rate 2009–2050, medium variant (% year−1 ) Population (million persons) Difference in population, 2009–2050 (million persons) Additional domestic demand (million t) Needed production to meet additional domestic demand and attain self-sufficiencyc (million t) % increase over 2009 levels Estimated yield gap as % of average yieldsb YieldGapf YieldGape YieldGapp

Rice importers

Rice exporters

Philippines

Indonesia

Thailand

92

237

69

87

194 207

182 237

153 241

251 319

183 16.3

246 64.4

474 31.5

423 38.9

18 0 4.5 3.6

4 0 12.9 5.0

0 43 11.0 2.9

0 19 7.4 5.2

1.3

0.5

0.1

0.4

155 63

293 56

13.0

13.3

32 98

80 25

32 2

44 14

43 60 99

46 34 67

24 28 60

26 21 52

71 2 0.5

Vietnam

104 17 5.4

Sources of basic data: rice production, consumption and trade: FAO (2011a); population: UN (2011). a In rough rice equivalent. b Estimated based on yield gaps and average yields from 1994/1995 to 1999 in the study sites in Indonesia, Thailand, and Vietnam, and from 2001 to 2008 in the study site in the Philippines. The values are averages in the wet and dry seasons weighted by production except for Indonesia wherein simple averages were used because production data by season is not available. c Assuming the same level of stocks for all countries, no imports for the Philippines and Indonesia, and the same level of exports for Thailand and Vietnam as in 2009.

Based on calculations of national rice requirements in 2050 and assuming that self-sufficiency is a goal and consumption patterns remain the same, production in the Philippines and Indonesia must increase by 98% and 25%, respectively, compared with 2009 levels (Table 7). If the national yield gaps are similar to those in the study sites, then, reducing the yield gap between average and bestyielding farmers would be more than adequate to make Indonesia become self-sufficient in rice. However, for the Philippines, average yield level must be raised even higher than that of the best-yielding farmers. On the other hand, for Thailand and Vietnam, current national production is enough to meet domestic demand by 2050. However, these two countries are major rice exporters, together accounting for 56% of global rice exports in 2008. Based on global demand modeling, there will be significant price increases by 2050 if current yield growth rates do not improve (Tweeten and Thompson, 2008). To cover the anticipated larger export demand, rice yields in Thailand and Vietnam must also increase. There seems to be ample scope to increase export by closing the yield gap between average and best farmers’ yields. It appears that meeting future demand can be achieved in three countries (Indonesia, Thailand, and Vietnam) by bridging the yield gap in farmers’ fields without the need to raise the current yield ceiling. On the other hand, in the Philippines, yields would have to increase to the present climatic potential. Whether such yield increases can be achieved needs to be examined. Using the nonlinear equation developed for Central Luzon, Philippines, and making an optimistic assumption that skewness of yields will be −1 for both the wet and dry seasons by 2050 (i.e., the distribution of farmers’ yields will be more negatively skewed as more farmers

achieve high yields in the future), average rice yields will increase by at most 18% over 2007/2008 levels. With the current trends in yield growth, existing production technologies, and prevailing conditions, self-sufficiency by 2050 does not appear to be likely for the Philippines. Although there could be a change in the diet of some consumers who will most likely replace their rice consumption with meat, there may still be many poor people for whom rice will continue to be the major source of energy and protein. In addition, calculations were based on current aggregate rice areas. With pressure from other land uses such as settlements, crops for feed and biofuels, and projected scarcity of water for irrigation, rice areas in 2050 could be smaller than current levels. Hence, yields should be much higher than the indicated yield here for both Philippines and Indonesia to meet self-sufficiency requirements by mid-century. It should be noted, however, that the four study sites represent irrigated rice areas where yields are much higher than the national average. Although the study areas show wide yield gaps, the unfavorable rice environments (rainfed lowland and upland areas) which comprise a non-trivial proportion of total rice areas are reported to have much higher yield gaps that persist due to economic reasons (Lobell et al., 2009). New varieties are generally more readily adopted by farmers than new management techniques. Hence, a promising solution for bridging the yield gap involves targeted breeding to make varieties more resilient (Fischer et al., 2009). Breeding for biotic and abiotic stresses does not involve an increase in potential yield, but it would result in rice varieties that are more resistant, for example, to weather risks (flooding, drought, cold temperatures) and other stresses. Mega varieties of rice with tolerance for submergence and

A.G. Laborte et al. / Europ. J. Agronomy 36 (2012) 9–20

salinity have been developed or are in the pipeline (Septiningsih et al., 2009; Thomson et al., 2010). In addition to closing the yield gap, the needed increase in rice supply in the future can be supplemented by further raising the rice yield ceiling. Some best-yielding farmers are already at or near the economic yield goal. Although there is still room for increase based on potential yields, this may not be economically attractive for farmers, unless there is a change in the prices of rice and/or production inputs that will make it more attractive for best-yielding farmers to invest more to achieve higher yields. Yield growth has been slowing in the past decade; hence, higher yielding rice varieties should be made available to boost yield growth. However, our analysis shows that many farmers adopt new technologies slowly. Moreover, farmers with very low yields are even slower to benefit from the higher yield ceiling. It took more than a decade for low-yielding farmers to achieve higher yields. In conclusion, wide yield gaps persist in farmers’ rice fields in Southeast Asia and closing this yield gap remains a challenge. Our study quantifies the yield gaps and gives indications of causal factors by comparing average and best-farmers’ yields and inputs. However, a more in-depth farm survey could shed more light on the explanations of the yield gaps and the differences in performance between average and best-yielding farmers. Production inputs need to be used more efficiently; our study demonstrated that best-yielding farmers do so for nitrogen and labor. Furthermore, the results of the study imply that years of schooling tend to be higher with best-yielding farmers. This reinforces the observation that intensive knowledge delivery and programs that enhance farmers’ skills as well as institutional arrangements are crucial. These would require substantial investments in research for crop improvement and extension.

Acknowledgements We thank the Social Sciences Division, International Rice Research Institute for providing the data used in this paper: farm survey data from the loop survey (Central Luzon) and the RTDP Project (Can Tho, Suphan Buri, and West Java). We also thank the Climate Unit of IRRI, the Thai Meteorological Department, the Indonesian Center for Rice Research (ICRR), Ministry of Agriculture – Sukamandi, Indonesia through the IRRI Outpost office in Indonesia, and Jarrod Welch for providing the weather data used in the potential yield simulations. Finally, we thank the anonymous reviewer for comments and suggestions that helped us improve this paper.

References Ali, M., Byerlee, D., 1991. Economic efficiency of small farmers in a changing world: a survey of recent evidence. J. Int. Dev. 3, 1–27. BAS (Bureau of Agricultural Statistics, Philippines), 2011. CountrySTAT Philippines, Available at http://bas.gov.ph/ (verified 21.07.11) [Online]. Boling, A.A., Bouman, B.A.M., Tuong, T.P., Murty, M.V.R., Jatmiko, S.Y., 2007. Modeling the effect of groundwater depth on yield-increasing interventions in rainfed lowland rice in Central Java, Indonesia. Agric. Syst. 92, 115–139. Boling, A.A., Tuong, T.P., Van Keulen, H., Bouman, B.A.M., Suganda, H., Spiertz, J.H.J., 2010. Yield gap of rainfed rice in farmers’ fields in Central Java, Indonesia. Agric. Syst. 103, 307–315. Bouman, B.A.M., Van Laar, H.H., 2006. Description and evaluation of the rice growth model ORYZA2000 under nitrogen-limited conditions. Agric. Syst. 87, 249–273. Bouman, B.A.M., Kropff, M.J., Tuong, T.P., Wopereis, M.C.S., Ten Berge, H.F.M., van ˜ Laguna. Laar, H.H., 2001. ORYZA2000: Modeling Lowland Rice. IRRI, Los Banos, Bouman, B.A.M., Lampayan, R.M., Tuong, T.P., 2007. Water Management in Irrigated ˜ Rice: Coping with Water Scarcity. IRRI, Los Banos, Laguna, 54 pp. Bruinsma, J., 2009. The resource outlook to 2050: by how much do land, water, and crop yields need to increase by 2050? In: Expert Meeting on How to Feed the World in 2050, FAO, Rome, Available at www.fao.org/fileadmin/ templates/wsfs/docs/expert paper/05-Bruinsma ResourceOutlookto2050.pdf (verified 19.10.09) [Online].

19

Chaudhary, R.C., 2000. Strategies for bridging the yield gap in rice: a regional perspective. In: Papademetriou, M.K., Dent, F.J., Herath, E.M. (Eds.), Bridging the Rice Yield Gap in the Asia-Pacific Region. UN FAO, Bangkok, pp. 201–214. Comprehensive Assessment of Water Management in Agriculture, 2007. Water for Food, Water for Life: A Comprehensive Assessment of Water Management in Agriculture. Earthscan/International Water Management Institute, London/Colombo. Dawe, D., 2005. Increasing water productivity in rice-based systems in Asia—past trends, current problems, and future prospects. Plant Prod. Sci. 8, 221–230. De Bie, C.A.J.M., 2000. Comparative performance analysis of agro-ecosystems. ITC dissertation no. 75. Enschede, 232 pp. De Koeijer, T.J., Wossink, G.A.A., Van Ittersum, M.K., Struik, P.C., Renkema, J.A., 1999. A conceptual model for analysing input–output coefficients in arable farming systems: from diagnosis towards design. Agric. Syst. 61, 33–44. De Wit, C.T., 1992. Resource use efficiency in agriculture. Agric. Syst. 40, 125–151. Dobermann, A., Witt, C., 2004. The evolution of site-specific nutrient management in irrigated rice systems of Asia. In: Dobermann, A., Witt, C., Dawe, D. (Eds.), Increasing Productivity of Intensive Rice Systems through Site-specific Nutrient ˜ Laguna, pp. 75–99. Management. Science Publishers/IRRI, Enfield, NH/Los Banos, Dobermann, A., Simbahan, G.C., Moya, P.F., Adviento, M.A.A., Tiongco, M., Witt, C., Dawe, D., 2004. Methodology for socioeconomic and agronomic on-farm research in the RTDP project. In: Dobermann, A., Witt, C., Dawe, D. (Eds.), Increasing Productivity of Intensive Rice Systems through Site-specific Nutri˜ ent Management. Science Publishers/IRRI, Enfield, NH/Los Banos, Laguna, pp. 11–27. FAO (Food and Agriculture Organization), 2011a. FAOSTAT. FAO, Rome, Available at http://faostat.fao.org/ (verified 16.07.11) [Online]. FAO (Food and Agriculture Organization), 2011b. Regional Data Exchange System (RDES)—Thailand, Available at www.faorap-apcas.org/thailand.html (verified 21.07.11) [Online]. Fischer, R.A., Byerlee, D., Edmeades, G.O., 2009. Can technology deliver on the yield challenge to 2050? In: Expert Meeting on How to Feed the World in 2050, FAO, Rome, Available at www.fao.org/fileadmin/templates/wsfs/docs/ expert paper/11- Fischer-etal Technology-Challenge.pdf (verified 13.10.09) [Online]. Grothendieck, G., 2007. nls2: Non-linear Regression with Brute Force. R Package Version 0.1-2, Available at http://cran.r-project.org/web/packages/nls2/index.html (verified 12.11.09) [Online]. GSO (General Statistics Office of Vietnam), 2011. Agriculture, Forestry and Fishery Statistical Data, Available at www.gso.gov.vn/default en.aspx?tabid= 469&idmid=3 (verified 21.07.11) [Online]. Herdt, R., 1979. An overview of the constraints project results. Farm-level constraints ˜ Laguna, pp. 395–411. to high rice yields in Asia: 1974–77: papers. IRRI, Los Banos, Hibberd, J.M., Sheehy, J.E., Langdale, J.A., 2008. Using C4 photosynthesis to increase the yield of rice—rationale and feasibility. Curr. Opin. Plant Biol. 11, 228–231. Koenker, R., 2009. quantreg: Quantile Regression. R Package Version 4.43, Available at http://cran.r-project.org/web/packages/quantreg/index.html (verified 02.12.09) [Online]. Koenker, R., Hallock, K.F., 2001. Quantile regression. J. Econ. Perspect. 15, 143–156. Koning, N., Van Ittersum, M.K., 2009. Will the world have enough to eat? Curr. Opin. Environ. Sustain. 1, 77–82. Kropff, M.J., Cassman, K.G., Van Laar, H.H., Peng, S., 1993. Nitrogen and yield potential of irrigated rice. Plant Soil 155/156, 391–394. Lobell, D.B., Cassman, K.G., Field, C.B., 2009. Crop yield gaps: their importance, magnitudes, and causes. Annu. Rev. Environ. Resour. 34, 179–204. Ministry of Agriculture, Republic of Indonesia, 2011. Agricultural Statistical Database, Available at http://aplikasi.deptan.go.id/bdsp/index-e.asp (verified 21.07.11) [Online]. Moya, P.F., Dawe, D., Pabale, D., Tiongco, M., Chien, N.V., Devarajan, S., Djatiharti, A., Lai, N.X., Niyomvit, L., Ping, H.X., Redondo, G., Wardana, P., 2004. The economics of intensively irrigated rice in Asia. In: Dobermann, A., Witt, C., Dawe, D. (Eds.), Increasing Productivity of Intensive Rice Systems through Site-specific Nutrient ˜ Management. Science Publishers, Enfield, N.H. and IRRI, Los Banos, Laguna, pp. 29–58. Nidumolu, U.B., 2004. Integrating geo-information models with participatory approaches: applications in land use analysis. ITC Dissertation No. 117. Enschede, 153 pp. Peng, S., Cassman, K.G., Virmani, S.S., Sheehy, J., Khush, G.S., 1999. Yield potential trends of tropical rice since the release of IR8 and the challenge of increasing rice yield potential. Crop Sci. 39, 1552–1559. Regmi, A.P., Ladha, J.K., Pathak, H., Pasuquin, E., Bueno, C., Dawe, D., Hobbs, P.R., Joshy, D., Maskey, S.L., Pandey, S.P., 2002. Yield and soil fertility trends in a 20-year rice–rice–wheat experiment in Nepal. Soil Sci. Soc. Am. J. 66, 857–867. Rogers, E.M., 2003. Diffusion of Innovations, 5th ed. Free Press, New York, 550 pp. Septiningsih, E.M., Pamplona, A.M., Sanchez, D.L., Neeraja, C.N., Vergara, G.V., Heuer, S., Ismail, A.M., Mackill, D.J., 2009. Development of submergence-tolerant rice cultivars: the Sub1 locus and beyond. Ann. Bot. 103, 151–160. Smil, V., 2000. Feeding the World: A Challenge for the Twenty-first Century. Massachusetts Institute of Technology, Massachusetts, 360 pp. Thomson, M.J., de Ocampo, M., Egdane, J., Rahman, M.A., Sajise, A.G., Adorada, D.L., Tumimbang-Raiz, E., Blumwald, E., Seraj, Z.I., Singh, R.K., Gregorio, G.B., Ismail, A.M., 2010. Characterizing the Saltol quantitative trait locus for salinity tolerance in rice. Rice 3, 148–160. Tweeten, L., Thompson, S.R., 2008. Long-term agricultural output supplydemand balance and real farm and food prices. Working Paper AEDE-WP

20

A.G. Laborte et al. / Europ. J. Agronomy 36 (2012) 9–20

0044-08. The Ohio State University, USA. Available at http://aede.osu.edu/ resources/docs/pdf/QY1R4X2Y-AG7A-VXX7-J003PSOREUML1A3F.pdf (verified 10.01.10) [Online]. Tuong, T.P., Bouman, B.A.M., 2003. Rice production in water scarce environments. In: Kijne, J.W., Barker, R., Molden, D. (Eds.), Water Productivity in Agriculture: Limits and Opportunities for Improvement. CABI Publishing, Wallingford, pp. 53–67. UN (United Nations), 2011. World Population Prospects: The 2010 Revision, Available at http://esa.un.org/unpd/wpp/unpp/panel population.htm (verified 16.07.11) [Online]. Young, A., 1999. Is there really spare land? A critique of estimates of available cultivable land in developing countries. Environ. Dev. Sustain. 1, 3–18.

Van Ittersum, M.K., Rabbinge, R., 1997. Concepts in production ecology for analysis and quantification of agricultural input-output combinations. Field Crops Res. 52, 197–208. Van Ittersum, M.K., Leffelaar, P.A., Van Keulen, H., Kropff, M.J., Bastiaans, L., Goudriaan, J., 2003. On approaches and applications of the Wageningen crop models. Eur. J. Agron. 18, 201–234. Wikarmpapraharn, C., Kositsakulchai, E., 2010. Evaluation of ORYZA2000 and CERESRice models under potential growth condition in the Central Plain of Thailand. Thai J. Agric. Sci. 43 (1), 17–29. Witt, C., Dobermann, A., Abdulrachman, S., Gines, H.C., Wang, G.H., Nagarajan, R., Satawathananont, S., Son, T.T., Tan, P.S., Tiem, L.V., Simbahan, G.C., Olk, D.C., 1999. Internal nutrient efficiencies of irrigated lowland rice in tropical and subtropical Asia. Field Crops Res. 63, 113–138.