• Adams, R. M., K. J. Bryant, B. A. McCarl, D. M. Legler, J. J. O’Brien, A. Solow, and R. Weiher, 1995: Value of improved long-range weather forecasts. Contemp. Econ. Policy,13, 10–19.

  • Barnston, A. G., and Coauthors, 1994: Long-lead seasonal forecasts—Where do we stand? Bull. Amer. Meteor. Soc.,75, 2097–2114.

  • Cane, M. A., G. Eshel, and R. W. Buckland, 1994: Forecasting Zimbabwean maize yield using eastern equatorial Pacific sea surface temperature. Nature,370, 204–205.

  • Cantagallo, J. E., C. A. Chimenti, and A. J. Hall, 1999: Number of seeds per unit area in sunflower correlates well with a photothermal quotient. Crop Sci., in press.

  • Carlson, R. E., D. P. Todey, and S. E. Taylor, 1996: Midwestern corn yield and weather in relation to extremes of the Southern Oscillation. J. Prod. Agric.,9, 347–352.

  • Chen, D., S. E. Zebiak, A. J. Busalacchi, and M. A. Cane, 1995: An improved procedure for El Niño forecasting: Implications for predictability. Science,269, 1699–1702.

  • Cleveland, R. B., W. S. Cleveland, J. E. McRae, and I. Terpenning, 1990: STL: A seasonal-trend decomposition procedure based on Loess. J. Off. Stat.,6, 3–73.

  • Cleveland, W. S., and S. J. Devlin, 1988: Locally-weighted regression:An approach to regression analysis by local fitting. J. Amer. Stat. Assoc.,83, 596–610.

  • Connor, D. J., and V. O. Sadras, 1992: Physiology of yield expression in sunflower. Field Crops Res.,30, 333–389.

  • Dai, A., I. Y. Fung, and A. D. Del Genio, 1997: Surface observed global land precipitation variations during 1900–88. J. Climate,10, 2943–2962.

  • Díaz, A. F., C. D. Studzinski, and C. R. Mechoso, 1998: Relationships between precipitation anomalies in Uruguay and southern Brazil and sea surface temperature in the Pacific and Atlantic Oceans. J. Climate,11, 251–271.

  • Díaz, R. A., 1995: Seeking practical applications for short climate predictions in the agriculture of Argentina. International Research Institute for Climate Prediction Rep. IRIP-TL-95/1, 40 pp. [Available from INTA, Instituto de Clima y Agua, 1712 Castelar, Prov. Buenos Aires, Argentina.].

  • Dilley, M., 1997: Climatic factors affecting annual maize yields in the Valley of Oaxaca, Mexico. Int. J. Climatol.,17, 1549–1557.

  • Garnett, E. R., and M. L. Khandekar, 1992: The impact of large-scale atmospheric circulations and anomalies on Indian monsoon droughts and floods and on world grain yields—A statistical analysis. Agric. For. Meteor.,61, 113–128.

  • Hall, A. J., J. H. Lemcoff, and M. Trápani, 1981: Water stress before and during flowering in maize and its effect on yields, its components and their determinants. Maydica,26, 19–38.

  • ——, C. M. Rebella, C. M. Ghersa, and J. P. Culot, 1992: Field-crop systems of the Pampas. Ecosystems of the World: Field Crop Ecosystems, C. J. Pearson, Ed., Elsevier, 413–449.

  • Hammer, G. L., D. P. Holzworth, and R. Stone, 1996: The value of skill in seasonal climate forecasting to wheat crop management in a region with high climatic variability. Aust. J. Agric. Res.,47, 717–737.

  • Handler, P., 1990: USA corn yields, the El Niño and agricultural drought: 1867–1988. Int. J. Climatol.,10, 819–828.

  • Hansen, J. W., and Coauthors, 1996: ENSO effects on yields and economic returns of wheat, corn, and soybean in Argentina. Actas del VII Congreso Argentino y VII Congreso Latinoamericano e Ibérico de Meteorología, Buenos Aires, Argentina, Centro Argentino de Meteorólogos and Federacíon Latinoamericana e Ibérica de Sociedades de Meteorología, 89–90.

  • Hoogenboom, G., J. W. Jones, and K. J. Boote, 1991: A decision support system for prediction of crop yield, evapotranspiration, and irrigation management. Proc. Irrigation and Drainage, Honolulu, HI, Irrigation Division, American Society of Civil Engineers, 198–204.

  • Lanzante, J. R., 1996: Resistant, robust and non-parametric techniques for the analysis of climate data: Theory and examples, including applications to historical radiosonde station data. Int. J. Climatol.,16, 1197–1226.

  • Legler, D. M., K. J. Bryant, and J. J. O’Brien, 1999: Impact of ENSO-related climate anomalies on crop yields in the U.S. Climatic Change, in press.

  • Manly, B. F. J., 1997: Randomization, Bootstrap and Monte Carlo Methods in Biology. Chapman and Hall, 399 pp.

  • Meinke, H., R. C. Stone, and G. L. Hammer, 1996: SOI phases and climatic risk to peanut production: A case study for northern Australia. Int. J. Climatol.,16, 783–789.

  • Messina, C., A. Beltrán, and A. Ravelo, 1996a: El fenómeno ENSO:Su relación con la productividad de maíz en la región pampeana argentina (The ENSO phenomenon: Its relationship with maize productivity in the Argentine Pampas). IV Congreso Colombiano de Meteorología, Santa Fe de Bogota, Colombia, Sociedad Colombiana de Meteorología, 236–241.

  • ——, ——, and ——, 1996b: La variabilidad interanual de los rendimientos de trigo en la región pampeana y su relación con el fenómeno ENSO (El Niño/Southern Oscillation) (Interannual variability of wheat yields in the Pampas, and their association with the ENSO phenomenon). Actas del VII Congreso Argentino y VII Congreso Latinoamericano e Ibérico de Meteorología, Buenos Aires, Argentina, Centro Argentino de Meteorólogos and Federacíon Latinoamericana e Ibérica de Sociedades de Meteorología, 55–56.

  • Meyers, S. D., J. J. O’Brien, and E. Thelin, 1999: Reconstruction of monthly SST in the tropical Pacific Ocean during 1868–1993 using adaptive climate basis functions. Mon. Wea. Rev.,127, 1599–1612.

  • Mjelde, J. W., T. N. Thompson, F. M. Hons, J. T. Cothren, and C. G. Coffman, 1997: Using Southern Oscillation information for determining corn and sorghum profit-maximizing input levels in east-central Texas. J. Prod. Agric.,10, 168–175.

  • Nicholls, N., 1985: Impact of the Southern Oscillation on Australian crops. J. Climatol.,5, 553–560.

  • Phillips, J. G., C. Rosenzweig, and M. Cane, 1996: Exploring the potential for using ENSO forecasts in the U.S. Corn Belt. Drought Network News,8, 6–10.

  • ——, M. A. Cane, and C. Rosenzweig, 1998: ENSO, seasonal rainfall patterns and simulated maize yield variability in Zimbabwe. Agric. For. Meteor.,90, 39–50.

  • Pisciottano, G., A. Díaz, G. Cazes, and C. R. Mechoso, 1994: El Niño–Southern Oscillation impact on rainfall in Uruguay. J. Climate,7, 1286–1302.

  • Rao, V. B., L. D. A. Sá, S. H. Franchito, and K. Hada, 1997: Interannual variations of rainfall and corn yields in Northeast Brazil. Agric. For. Meteor.,85, 63–74.

  • Rebella, C. M., R. Díaz, and L. A. Funston, 1984: Tendencias de lluvias y rendimientos de maíz en la región maicera típica argentina (Precipitation trends and maize yields in Argentina’s corn belt). Proc. III Congreso Nacional de Maíz, Pergamino, Argentina, Asociación Argentina de Ingenieros Agrónomos del Norte de Buenos Aires, 129–133.

  • Ropelewski, C. F., and M. S. Halpert, 1987: Global- and regional-scale patterns associated with the El Niño/Southern Oscillation. Mon. Wea. Rev.,115, 1606–1626.

  • ——, and ——, 1989: Precipitation patterns associated with the high index phase of the Southern Oscillation. J. Climate,2, 268–284.

  • ——, and ——, 1996: Quantifying Southern Oscillation–precipitation relationships. J. Climate,9, 1043–1059.

  • SAGPyA, 1994: Estadísticas agropecuarias y pesqueras 1994. Ministerio de Economía y Obras y Servicios Públicos, Secretaría de Agricultura, Ganadería, Pesca y Alimentación, Argentina, 459 pp.

  • Smit, B., R. Blain, and P. Keddie, 1997: Corn hybrid selection and climatic variability: Gambling with nature? Can. Geogr.,41, 429–438.

  • Sokal, R. R., and F. J. Rohlf, 1969: Biometry. W. H. Freeman, 776 pp.

  • Sonka, S. T., J. W. Mjelde, P. J. Lamb, S. E. Hollinger, and B. L. Dixon. 1987: Valuing climate forecast information. J. Climate Appl. Meteor.,26, 1080–1091.

  • Tanco, R., and G. J. Berri, 1996: Acerca del efecto del fenómeno El Niño sobre la precipitación en la Pampa húmeda argentina (On the effects of the ENSO phenomenon on precipitation in the Argentine Pampas). Actas del VII Congreso Argentino y VII Congreso Latinoamericano e Ibérico de Meteorología, Buenos Aires, Argentina, Centro Argentino de Meteorólogos and Federacíon Latinoamericana e Ibérica de Sociedades de Meteorología, 319–320.

  • Travasso, M. I., 1990: Evaluation des facteurs agroclimatiques limitant le rendement du ble dans le sud de Buenos Aires, Argentine. Essai de modelisation a l’echelle regionale (Assessment of agroclimatic factors limiting wheat yields in southeastern Buenos Aires, Argentina. A modelling test on regional scales). Ph.D. thesis, Universite de Paris, 96 pp. [Available from INTA, Instituto de Clima y Agua, 1712 Castelar, Prov. Buenos Aires, Argentina.].

  • Trenberth, K., 1996: El Niño–Southern Oscillation. Climate Change:Developing Southern Hemisphere Perspectives, T. W. Giambelluca and A. Henderson-Sellers, Eds., John Wiley and Sons, 145–173.

  • ——, 1997a: Short-term climate variations: Recent accomplishments and issues for future progress. Bull. Amer. Meteor. Soc.,78, 1081–1096.

  • ——, 1997b: The definition of El Niño. Bull. Amer. Meteor. Soc.,78, 2771–2777.

  • Whiteman, P. C., and G. L. Wilson, 1965: Effects of water stress on the reproduction development of Sorghum vulgare. Pers. University of Queensland, Paper 14. [Available from University of Queensland, Brisbane, Queensland 4072, Australia.].

  • Widmann, M., and C. Schär, 1997: A principal component and long-term trend analysis of daily precipitation in Switzerland. Int. J. Climatol.,17, 1333–1356.

  • Wilks, D. S., 1995: Statistical Methods in the Atmospheric Sciences. Academic Press, 467 pp.

  • View in gallery

    The study area. Black circles indicate location of rainfall stations.

  • View in gallery

    Time series of maize yields, 1900–95. Open squares indicate warm ENSO events; filled squares correspond to cold events. Neutral years are shown as open circles. The dashed line indicates the estimated low-frequency trend.

  • View in gallery

    Time series of relative yield residuals for summer crops, for the overlapping data period (1972–95). Warm ENSO events are indicated by gray bars; black bars indicate cold events; (a) maize, (b) sunflower, (c) sorghum, and (d) soybeans.

  • View in gallery

    Boxplot of PC1 amplitudes for Nov–Jan, by ENSO phase. Lower and upper boundaries for each box are the 25th and 75th percentiles. The line inside each box indicates the median. Whiskers mark the range of the values.

  • View in gallery

    Yield anomalies for summer crops as a function of PC1 amplitudes for the Nov–Jan series. Negative PC1 amplitudes indicate negative precipitation anomalies, and vice versa; (a) maize, (b) sunflower, (c) sorghum, and (d) soybeans.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 199 199 13
PDF Downloads 77 77 11

Associations between Grain Crop Yields in Central-Eastern Argentina and El Niño–Southern Oscillation

View More View Less
  • a Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, Florida
  • | b Facultad de Agronomía, Universidad de Buenos Aires, Buenos Aires, Argentina
  • | c Instituto de Clima y Agua, Instituto Nacional de Tecnología Agropecuaria, Castelar, Argentina
© Get Permissions
Full access

Abstract

Associations are investigated between yields of major crops in the Argentine Pampas (central-eastern Argentina) and El Niño–Southern Oscillation (ENSO) phase. For maize and sorghum, higher (lower) yield anomalies occur more frequently than expected by chance alone during warm (cold) ENSO events. For both crops, the depression of yields during cold events is, on average, larger and less variable than yield increases are during warm events. A yield decrease during cold events also is observed in soybean yields, although the effect of warm events is not statistically significant. There is a marginally significant tendency for low sunflower yields to occur less frequently than expected during cold events. Wheat, the only winter crop considered, did not show an association with ENSO. Precipitation anomalies during October–February (the period with strongest ENSO signal in the Pampas) are summarized through principal component analysis. Precipitation anomalies during November–January are significantly correlated with maize, sorghum, and soybean yield anomalies. In turn, those precipitation anomalies show a distinct ENSO signal. Late spring–early summer precipitation, then, appears to mediate associations between ENSO phase and yields of maize, sorghum, and soybean in the Pampas.

Corresponding author address: Dr. Guillermo P. Podestá, University of Miami, RSMAS/MPO, 4600 Rickenbacker Causeway, Miami, FL 33149-1098.

gpodesta@rsmas.miami.edu

Abstract

Associations are investigated between yields of major crops in the Argentine Pampas (central-eastern Argentina) and El Niño–Southern Oscillation (ENSO) phase. For maize and sorghum, higher (lower) yield anomalies occur more frequently than expected by chance alone during warm (cold) ENSO events. For both crops, the depression of yields during cold events is, on average, larger and less variable than yield increases are during warm events. A yield decrease during cold events also is observed in soybean yields, although the effect of warm events is not statistically significant. There is a marginally significant tendency for low sunflower yields to occur less frequently than expected during cold events. Wheat, the only winter crop considered, did not show an association with ENSO. Precipitation anomalies during October–February (the period with strongest ENSO signal in the Pampas) are summarized through principal component analysis. Precipitation anomalies during November–January are significantly correlated with maize, sorghum, and soybean yield anomalies. In turn, those precipitation anomalies show a distinct ENSO signal. Late spring–early summer precipitation, then, appears to mediate associations between ENSO phase and yields of maize, sorghum, and soybean in the Pampas.

Corresponding author address: Dr. Guillermo P. Podestá, University of Miami, RSMAS/MPO, 4600 Rickenbacker Causeway, Miami, FL 33149-1098.

gpodesta@rsmas.miami.edu

Introduction

Agriculture is one of the human activities most affected by climate and weather variability. The El Niño–Southern Oscillation (ENSO) phenomenon, the result of a two-way interaction between the ocean and the atmosphere in the tropical Pacific Ocean, is the major single source of climate variability on seasonal to interannual scales in many parts of the world (Trenberth 1996; Trenberth 1997a). The ENSO phenomenon involves two extreme phases: warm events, also known as El Niño years; and cold events, referred to as La Niña or El Viejo years. Years that do not fall in the extreme phases are labeled as “neutral.”

Recent advances in our understanding of air–sea interactions and in observational systems have made it possible to forecast the occurrence of ENSO events with usable skill and with lead times of several months (Barnston et al. 1994; Chen et al. 1995). It is assumed that the routine availability of ENSO-related climate forecasts will benefit the agricultural sector by allowing farmers to mitigate potential negative consequences of climate variability or, alternatively, to capitalize on potentially beneficial effects (Adams et al. 1995; Mjelde et al. 1997). However, a necessary initial step in assessing the usefulness of climate forecasts for agriculture in a region is to characterize the vulnerability of this sector to ENSO-related climate variability.

The goal of this work is to explore associations between grain crop yields and ENSO-related climate variability in the region of central-eastern Argentina known as the Pampas (Fig. 1). The Pampas is one of the major agricultural regions in the world; a large proportion of Argentina’s crop production originates in this region. Hall et al. (1992) provide a thorough description of the climate, soils, and crop production systems in the Pampas.

ENSO events influence precipitation and temperature regimes in southeastern South America, including the Pampas. Ropelewski and Halpert (1987) found a tendency for enhanced precipitation in warm ENSO events during November–February of the following year. During cold ENSO events, Ropelewski and Halpert (1989) reported below-normal precipitation: of the 19 cold events in the series they analyzed, 16 were associated with dry June–December seasons. Pisciottano et al. (1994) found higher than average rainfall in Uruguay, a region close to the Pampas, during warm ENSO events from November to the following January. In contrast, cold events tended to have lower than average rainfall from October to December. Ropelewski and Halpert (1996) expanded their earlier results but emphasized shifts in the probability distribution of precipitation totals. Díaz et al. (1998) confirmed reported associations between ENSO and precipitation, but also explored regional modulation of the ENSO signal by sea surface temperature (SST) anomalies in the Atlantic Ocean.

Links between ENSO and crop yields have been shown for the United States (Handler 1990; Carlson et al. 1996; Phillips et al. 1996; Mjelde et al. 1997; Legler et al. 1999), Australia (Nicholls 1985; Meinke et al. 1996), Mexico (Dilley 1997), northeastern Brazil (Rao et al. 1997), and southern Africa (Cane et al. 1994; Phillips et al. 1998). In Argentina, several studies have investigated the influence of ENSO-related climate variability on crop yields. Garnett and Khandekar (1992) found that warm ENSO events appeared to favor wheat yields and cold events tended to lower yields of this crop. Díaz (1995) explored associations between SSTs in the tropical Pacific Ocean (as indicators of ENSO conditions), rainfall anomalies, and yields of maize and soybean in key crop districts in the Pampas. He found extremely low correlation between rainfall and maize yield anomalies in four major districts. Díaz also found rainfall anomalies to be more closely related with national-level maize production than with yield anomalies. In contrast, Messina et al. (1996a) showed significant correlation between Pacific SST anomalies in December and maize yields in the central part of the Pampas. Messina et al. (1996b) detected positive correlation between ENSO-related SST anomalies and wheat yields, but only for the southwestern portion of the Pampas. Hansen et al. (1996) used biophysical crop models to study ENSO effects on the yields of wheat, maize, and soybean in three sites in the Pampas of Argentina. They showed improved net returns when crop management was modified in response to knowledge about ENSO phase.

In this work, long yield series (in some cases approaching a century) for the major crops in the Pampas are used to explore the existence and nature of associations with ENSO phase. Crop yield data are detrended to remove effects of technological improvement, and yield anomalies are then computed. Associations between yield anomalies and ENSO phase are explored through categorical analyses (contingency tables) and tests of differences in the central tendency and spread of yields among ENSO phases. Finally, a spatial summary of precipitation anomalies through the Pampas is derived through principal component analysis. These precipitation anomalies are then related both to ENSO phase and to crop yields.

Data

Crop data

Records for the five major crops in Argentina (maize, wheat, sunflower, grain sorghum, and soybean) were obtained from Argentina’s Secretaría de Agricultura, Ganadería, Pesca y Alimentación (SAGPyA 1994). Maize, wheat, and sorghum made up 91%–93% of total production of cereals; soybean and sunflower accounted for 94%–98% of oilseed production [in both cases figures are for 1987–93; SAGPyA (1994)]. Although data are aggregated at the national level, the majority of the area and total production of these crops is in the Pampas. Production from the four Argentine provinces encompassing most of the Pampas (Buenos Aires, Córdoba, Santa Fe, and La Pampa) accounts for over 80% of the nation’s production of sorghum and over 90% for the other crops [average 1988–92; SAGPyA (1994)]. Except for limited areas (and only in recent years), the crops considered are grown without irrigation.

This work focuses on yield (estimated as the ratio of total production to area harvested) as an indicator of a crop’s vulnerability to climate variability. The time unit considered is the cropping cycle between July and June of the following calendar year; under this definition both winter and summer crops fall within the same “agricultural year.” A cropping cycle is noted by the year in which a crop was sown, even though harvest may have taken place in the following calendar year (e.g., the 1982/83 cropping season is noted as 1982). Lengths of the yield series analyzed are shown in Table 1. Some earlier data were excluded because yields from very small sown areas were not considered to be representative. The number of cold and warm ENSO events (according to a definition presented below) and neutral years for each series also is shown in Table 1.

Precipitation data

Associations between crop yields and rainfall were explored using a database of monthly precipitation totals at 33 locations throughout the Pampas (Fig. 1), encompassing the period 1912–90. The data originally were compiled by the Argentine Meteorological Service.

ENSO phase

An ENSO phase (warm, cold, or neutral) was assigned to each cropping cycle. There are several alternative definitions of ENSO events (Trenberth 1997b). Here, ENSO events were categorized according to an index developed by the Japan Meteorological Agency (JMA). The JMA ENSO index is based on a 5-month running mean of spatially averaged SST anomalies in the region of the tropical Pacific Ocean between 4°N–4°S and 90°–150°W. If index values between July and June of the following year are 0.5°C or greater for at least six consecutive months (including the quarter of October–December, considered to be the typical peak of ENSO-related anomalies), then the cropping cycle is categorized as a warm event. Similarly, if the index is −0.5°C or lower for at least six consecutive months (including October–December), then the cropping cycle is categorized as a cold event.

The JMA index is based on observed data for the period from 1949 to the present. For years prior to 1949, the index was derived from reconstructed monthly mean SST fields, estimated using an orthogonal projection technique (Meyers et al. 1999). Identified ENSO events are listed in Table 2. The number of events for each crop changes with the length of its series (Table 1).

Crop yield trends and anomalies

Low-frequency trends

Agricultural yield data typically have an upward low-frequency trend (LFT) because of technological improvements in crop genetics and management techniques (Hall et al. 1992). An LFT was fitted to the yield series using LOESS, a smoother based on locally weighted regression (Cleveland and Devlin 1988). This flexible technique follows patterns suggested by the data, and its robust fitting procedure guards against the possibility of outliers distorting the trend. A problem with the removal of trends (although not unique to the use of LOESS) is that the estimated LFT also may reflect long-term climate effects. Conversely, the variability after removal of the trend may not entirely be climate related.

For the maize and wheat series, smoothing roughly equivalent to a low-pass filter with a cutoff frequency of about 40 yr was used (various bandwidths yielded similar results). For the shorter series, smoothing was adjusted to maintain a bandwidth similar to that of the longer series. The use of LOESS is illustrated in Fig. 2 for maize. LFTs are not shown for the other crops, but they all reveal sharp yield increases starting around the 1970s that were caused by introducing technological improvements such as hybrids, improved cultivars, and fully mechanized labor and harvest (Hall et al. 1992).

Absolute and relative yield residuals

After LFTs were estimated, attention was focused on interannual yield variability. Absolute yield residuals were computed by subtracting the LFT from the annual yields. Relative yield residuals, defined as the ratio (as percentage) of the absolute residuals to the expected yield (the LFT) for a given year also were computed. Exploratory analyses produced qualitatively similar results using both absolute and relative residuals. For ease of agronomic interpretation (as yields of all crops have changed considerably through the years), results here are based on relative residuals. Descriptive statistics for absolute and relative yield residuals are shown in Table 3. Years in which maximum and minimum yields occurred are noted together with their corresponding ENSO phase. Time series of relative yield residuals for the four summer crops are shown in Fig. 3 for the overlapping data period (1972–95).

Crop yields and ENSO phase

We explored associations between crop yield residuals and ENSO phase following two approaches. First, we used contingency tables to detect associations between crop yields and ENSO phase. Then, we quantified differences among ENSO phases in central tendency and spread of yield anomalies.

Contingency tables

We built a two-way contingency table for each crop (Tables 4a–e) by classifying each cropping cycle according to (i) ENSO phase (see Table 2) and (ii) relative yield residual tercile. Boundaries between terciles, the 33d and 66th percentiles, are shown for each crop in Table 3.

If there were no association between yield and ENSO, the number of years in each yield tercile for a given ENSO phase should be relatively similar. In contrast, deviations from expected frequencies would suggest that ENSO phase and crop yield categories are not independent. The statistical significance of such deviations was assessed via χ2 tests (Wilks 1995). An exact Fisher test was used when expected cell counts were low (<5), and thus χ2 results would be suspect (Sokal and Rohlf 1969).

As an example, let us examine the maize contingency table (Table 4a). There are 21 warm events in the maize series. Assuming no association with ENSO, there should be about seven years in each yield tercile. However, for the warm phase there are 13 years in the upper tercile. That is, high maize yields are almost twice as likely as by chance alone. Correspondingly, there are only three warm event years in the lower yield tercile, about half as many as would be expected. During cold ENSO events the opposite pattern is observed. There are 16 years in the lower tercile, more than twice the number expected by chance. In contrast, only two cold events show yields in the upper tercile. The χ2 probability value (P ≪ 0.001) indicates that counts in the table cells are significantly different from those expected if relative maize yields were independent of ENSO phase.

The wheat table (Table 4b) suggests a lack of association between ENSO phase and yield residuals. Because warm ENSO events may be associated with slightly higher than normal precipitation during March–April (Tanco and Berri 1996), we also examined a possible association between ENSO phase for a given year and wheat yield anomalies during the following cropping cycle. Such association might result, for example, from higher soil water content prior to wheat’s winter planting, caused by ENSO-enhanced March–April rains. A contingency table (not shown) linking ENSO phase and wheat yield anomalies in the following year did not show any association (χ2 = 0.355, P = 0.986). The sunflower and sorghum contingency tables (Tables 4c,d) show no significant overall association between ENSO phase and yields, whereas the association is significant for soybeans (Table 4e).

The χ2 test examines deviations from expected cell counts over an entire contingency table. However, we are interested particularly in departures from expected counts during extreme (warm and cold) ENSO phases. For example, how unusual is the large number of years with high maize yields during warm events? The hypergeometric distribution (Sokal and Rohlf 1969) provides a more specific test of such departures.

For maize and soybean, the hypergeometric probabilities confirm the overall significance of the χ2 tests. For example, the probability of observing 13 years or more in the upper maize yield tercile during 21 warm events (given a total of 32 years in the upper tercile and 64 years in middle or lower terciles) is 0.002. Conversely, the probability of observing 16 years or more in the lower maize yield tercile during 23 cold events is 0.0005. Therefore, maize yield departures from expected frequencies during both warm and cold events are highly significant. The soybean table (Table 4e) shows four upper-tercile years during warm ENSO events; this outcome has a probability of 0.069. At the same time, all four cold events fall in the lower yield tercile—a frequency with a probability of 0.007. That is, the effect of warm events on soybeans is significant only at probability levels of 0.07 or more, whereas cold events have a highly significant effect.

For sunflower and sorghum, some patterns emerge when attention is focused on extreme ENSO phases, even though the contingency tables show no significant overall association. For sunflower, the number of lower tercile residuals during cold events (three years) is about half what might be expected by chance; this outcome has a probability of 0.059. Low sunflower yields thus appear to be less likely during cold events than in other years. For sorghum, there is only one low-yield year (about three are expected) during warm events—an outcome with a probability of 0.007. At the same time, the number of lower-tercile years in cold events (six) is twice what would be expected by chance alone; the probability of observing at least this many years in this category is 0.029. As with maize, sorghum high yields (low yields) are more frequent during warm (cold) events.

Central tendency and spread of yields by ENSO phase

Differences in the central tendencies and spread of yield residuals among ENSO phases were quantified. Two estimators of the central tendency of yield residuals, the mean and median, are listed in Table 5 for each entire yield series and by ENSO phase. A Kruskal–Wallis test (Wilks 1995) was used to detect differences in the central tendency of residuals among all ENSO phases; probability values are listed in Table 6. One-tailed Wilcoxon tests (Wilks 1995) were then used to examine differences in median yield residuals between pairs of ENSO phases (warm vs neutral, neutral vs cold, warm vs cold); probability values for these tests also are listed in Table 6.

Although attention frequently is focused on the central tendency of yields, their spread or dispersion also is important for quantifying vulnerability to climate variability. Two estimators (Table 5) describe the spread of yield residuals: the standard deviation (SD) and a pseudostandard deviation (PSD). The PSD is a resistant estimator of spread (Lanzante 1996) and is computed by estimating the median absolute deviation of residuals and scaling it by a factor of 1.482 to make it a consistent estimator of the standard deviation for a Gaussian model (Wilks 1995). The significance of differences in the spread of yield residuals between pairs of ENSO phases was estimated via a test analogous to the parametric ratio of variances. The test statistic, the ratio of the squared PSDs for the two phases considered, has an unknown distribution; therefore, its significance was tested through a randomization procedure (Manly 1997).

The central tendency of maize residuals was higher for warm events than for cold events, with neutral years falling in between (Table 5). Differences were highly significant for the neutral–cold and warm–cold phase pairs (Table 6), but the warm–neutral difference was significant only at probability levels of 0.10 or more. Mean and median wheat yield anomalies were similar for all ENSO phases, and no significant differences were detected. Mean and median sunflower yield anomalies were higher during cold events than during other ENSO phases, but results showed no statistical significance. The central tendency of sorghum yield anomalies tended to be higher for warm events than for cold events. As for maize, differences in yields were highly significant for the neutral–cold and warm–cold phase pairs, but the warm–neutral difference was significant only at probability levels of 0.10 or more. Median yield residuals for soybeans were higher during warm events than for cold events, with neutral years in between. Both neutral–cold and cold–warm comparisons were highly significant; the warm–neutral test, in contrast, did not show statistical significance.

For most crops, no significant differences were detected in the spread of yield residuals between ENSO phases. The only significant result (at a level of 0.05) was the lower spread of maize residuals during warm events when compared to that of neutral years.

Yield and precipitation anomalies

Previous sections established significant statistical associations between yields of some crops in the Pampas and ENSO phase. However, the factors that might mediate these associations have not been discussed so far. There is considerable evidence that ENSO influences precipitation regimes in the Pampas (see introduction). Furthermore, rain-fed crop yields frequently are tied to available soil water and precipitation. For these reasons, we focus on associations between precipitation and crop yield residuals in the Pampas.

Precipitation anomalies throughout the Pampas were summarized using principal component analysis (PCA), a useful technique for extracting information from multidimensional datasets (Wilks 1995). Analyses were based on monthly precipitation totals at 33 stations for the period 1912–90 (Fig. 1).

PCA is typically conducted on centered data or anomalies (Wilks 1995). As a first step, we removed the seasonal cycle of precipitation to work with precipitation residuals. Significant low-frequency rainfall fluctuations have been reported in southeastern South America (Dai et al. 1997). These changes, however, appear to have occurred mostly during austral summer months. A flexible technique called STL (seasonal trend decomposition based on LOESS) (Cleveland et al. 1990) allowed the seasonal cycle of precipitation to change on timescales of several years while simultaneously estimating a low-frequency component. This approach effectively dealt with trends concentrated in specific times of the year. Detailed results from the seasonal decomposition are beyond the scope of this paper. Subsequent analyses focused on precipitation residuals, computed by subtracting from the original precipitation series the seasonal and low-frequency components (estimated separately for each station).

To reduce intra-annual variability, monthly precipitation anomalies at each station were aggregated over 3-month overlapping periods. For each 3-month series, principal component decomposition was carried out on the correlation matrix of precipitation anomalies. We focus on results for three series, centered on November, December, and January, respectively (the series will be noted as OND, NDJ, and DJF, indicating the months they encompass). The first principal component accounted for 46.5%, 44.5%, and 41.1% of total variability in the OND, NDJ, and DJF series, respectively.

The time series of the first principal component [PC1, also referred to as amplitude or score; Wilks (1995)] can be viewed as an optimally weighted average (where the weights are estimated through the PCA) of precipitation anomalies for all stations. Therefore, the PC1 scores summarize in a single series the temporal evolution of precipitation anomalies over the Pampas. Further, this approach drastically reduces effects of possible inhomogeneities or erroneous values in the original data series (Widmann and Schär 1997).

There is a distinct association between ENSO phase and precipitation anomalies in the Pampas during November–January, as illustrated by a box plot of PC1 amplitudes by ENSO phase (Fig. 4). Although PC amplitudes have no physical units and their signs are arbitrary, in this case positive values indicate positive precipitation anomalies, and vice versa. Warm ENSO events clearly are associated with higher median precipitation anomalies, then follow neutral and cold events, in that order. Furthermore, the range of anomalies is much smaller during cold events. Box plots for the OND and DJF series (not shown) reveal similar patterns. These results agree with previous reports of a strong ENSO signature in the study region during October–February (e.g., Ropelewski and Halpert 1996).

To explore associations between precipitation and yield anomalies, PC1 amplitudes for the OND, NDJ, and DJF series were correlated with each of the summer crop series. Figures 5a–d show scatterplots of yield residuals for maize, sunflower, sorghum, and soybeans, as a function of PC1 amplitudes for the NDJ series. A LOESS fit is shown (solid line) to facilitate visualization of trends. The strength of the associations was quantified via the rank-based Spearman correlation coefficient (ρ) (Wilks 1995). This statistic reflects monotonic associations between two variables, even if the associations are nonlinear (as is typically the case with crop and climate variables).

The association between national-level maize yield anomalies and precipitation anomalies throughout the Pampas is remarkably tight (Fig. 5a): low precipitation (negative PC1 values) is associated with low maize yields, and vice versa. The close association is confirmed by a highly significant correlation (ρ = 0.816; P ≪ 0.001). The other precipitation series show similar patterns of association with maize yields, but correlations are lower (albeit still highly significant): ρ values are 0.580 and 0.561 for OND and DJF, respectively. Sunflower yield shows a flat response to precipitation (Fig. 5b), and no significant correlation is detected. For sorghum, the yield–precipitation association (Fig. 5c) is similar to that of maize, although correlation values are generally lower. The highest correlation (ρ = 0.497; P = 0.009) occurs in DJF, a month later than for maize. Correlation values for OND and NDJ are 0.235 (P = 0.199) and 0.395 (P = 0.034). Soybean yields behave differently depending on whether NDJ precipitation anomalies are negative or positive (Fig. 5d). When anomalies are negative yields tend to increase with precipitation. In contrast, for positive precipitation anomalies soybean yields are mostly positive, but they do not increase with increasing precipitation (the fitted trend is flat for this portion of the graph). A similar pattern is present in the OND and DJF series (not shown). Correlations for the OND, NDJ, and DJF series are 0.726, 0.665, and 0.643, and they are all significant (corresponding probability values are 0.008, 0.022, and 0.034).

Discussion

We explored associations between yield anomalies of major grain crops in Argentina and ENSO phase using various complementary approaches. Although high and low yields have occurred under all ENSO phases, maize, sorghum, and soybeans showed some degree of association with ENSO. For maize and sorghum, there is a tendency toward higher (lower) yields during warm (cold) ENSO events. In both crops, the depression of yields during cold events is, on average, larger and more consistent than are yield increases during warm events. A decrease during cold events is also observed in soybean yields, although there does not seem to be a significant corresponding effect from warm events. For sunflower, the overall association with ENSO is not significant, although low yields are less frequent than expected during cold events. Finally, wheat, the only winter crop considered in this work, did not show any association with ENSO. These summarized conclusions are expanded below.

Maize yields show the clearest association with ENSO. This association is probably mediated by the enhanced likelihood of higher (lower) than normal rainfall during October–February, typical of warm (cold) ENSO events. December precipitation and soil moisture in December–January are important in determining maize yields in Argentina (Rebella et al. 1984 cited in Hall et al. 1992). Response to nitrogen fertilization is strongly related to rainfall during December and January (Hall et al. 1992). These months encompass the flowering period for maize throughout the Pampas. This period is critical in defining maize yield, and sensitivity to water availability is extremely high (Hall et al. 1981). In turn, this period coincides with a strong ENSO-related precipitation signal over central-eastern Argentina, as indicated by an index of November–January precipitation anomalies in the Pampas (Fig. 5a) and previous reports (Ropelewski and Halpert 1987, 1989, 1996).

Soybeans, currently the most important summer crop in Argentina, show a significant association with the cold ENSO phase (although the number of cold events is relatively small for this series). This crop has an interesting response to precipitation. There is a clear association between negative precipitation anomalies and soybean yields: lower precipitation results in lower yields. In contrast, yields, while mostly high, seem to become relatively insensitive to precipitation when positive anomalies occur. This behavior suggests that, once the minimum water needs of the crop are satisfied, yields reach a maximum, the level of which may be defined by other limiting factors. A similar pattern was observed in a simulation of soybean growth under various irrigation levels by Hoogenboom et al. (1991).

Associations between sorghum yields and both ENSO phase and precipitation anomalies are similar to those for maize. However, correlation between precipitation and yield is lower than that for maize, as sorghum is generally more resistant to water stress. For example, sorghum is able to modify its maturity in response to water availability: if there is water stress, the crop may delay its reproductive stage for short periods (Whiteman and Wilson 1965). The result may be a somewhat lower sensitivity of sorghum yields to water shortages.

Sunflower yields do not show an immediately apparent association with ENSO phase, as indicated by the lack of significance seen in most tests performed here. Nevertheless, tests based on the hypergeometric distribution reveal a marginally significant tendency for low yield residuals to be less frequent during cold events, a pattern opposite to that of the other summer crops considered. Because of its deep rooting system, water shortages, frequent during cold ENSO events, may not affect sunflower as much as they do other crops (Connor and Sadras 1992). In contrast, enhanced precipitation during warm events may favor the spread of diseases such as sclerotinia or the loss of nitrogen through leaching, with a consequent decrease in yields. Furthermore, because of sunflower’s lower sensitivity to water shortages, factors other than precipitation may be more relevant for it than for other crops: solar radiation, for example, may influence the number of grains, thus influencing yields (Cantagallo et al. 1999). The association between ENSO and this climatic variable, however, has not been documented in the region so far.

Wheat, the single winter crop considered here, does not show any apparent associations with ENSO. Water availability during the early part of this crop’s cycle (from planting to preflowering) is tied to yield variability. For example, about 42% of wheat yield variability has been linked to water availability in September–October (Hall et al. 1992). Nevertheless, the ENSO signature during these months is more variable than it is later in the spring. Furthermore, wheat has the widest geographic distribution of all crops considered, and this fact has implications for wheat’s link with ENSO. First, regionally inhomogeneous ENSO effects may cancel out when analyzing national-level yields. Second, because the timing of wheat’s critical periods varies with latitude, associations will be most apparent at times and locations for which critical periods and the ENSO signal coincide. For example, in the southern Pampas, wheat flowering (crucial in defining yield) occurs around late November (Travasso 1990). A clear ENSO signal during this month results in tighter associations with wheat yields in the southern Pampas than at the national level.

So far, yield responses have been discussed on a crop-by-crop basis. Figure 3 offers a chance to explore simultaneous responses of summer crops to ENSO events. The responses of maize, sorghum, and soybeans generally coincide for most events and follow the patterns described previously. One exception is the warm event of 1982, when residuals were slightly positive for sorghum but negative for maize and soybeans. Another inconsistency in the pattern occurred in 1973, a cold event. In this case, soybeans showed the expected negative residuals, whereas maize and sorghum had positive residuals. This illustrates our previous warning that both high and low yields have occurred under warm or cold ENSO phases, and therefore ENSO alone cannot completely explain yield fluctuations. A reviewer raised the issue of the consequences of strong warm events such as those in 1982 and 1997. In 1982, despite strong positive SST anomalies in the Pacific, the core-producing region of the Pampas had below-normal rainfall (J. Aiello, J. Forte Lay and A. Basualdo 1997, personal communication), with consequent low yields for maize and soybeans. In contrast, the strong 1997 event (not analyzed here) resulted in enhanced precipitation throughout the Pampas, and yields for maize and soybeans (after accounting for technology trends) were excellent. In both events there were extensive floods, but they did not significantly influence grain crop yields. The differential regional response to two very strong ENSO events illustrates the difficulty in predicting extratropical response to ocean–atmosphere interactions in the tropical Pacific Ocean.

For many of the crops that show associations with ENSO, cold events appear to have stronger, more consistent effects than do warm events. Yield decreases associated with cold events are usually larger (and often less variable) than corresponding yield increases during warm events. On one hand, this difference may reflect a stronger effect of cold ENSO events on the regional climate. Indeed, Ropelewski and Halpert (1996) found the association between rainfall anomalies and the cold phase to be stronger than for warm events in this region. Tanco and Berri (1996) also stressed the importance of cold events, as the below-normal rainfall associated with this ENSO phase showed a larger areal extent and persisted longer than did the impacts of warm events. Therefore, although much attention has been given in this region to the consequences of warm (El Niño) events, future studies should not neglect the effects of the potentially more damaging cold (La Niña) events.

On the other hand, the relative effects on crop yields of warm and cold ENSO events may have been distorted by the relatively low level of input usage that prevailed in Argentine agriculture until the 1990s. Low-input agricultural systems may realize only a small portion of the yield potentially achievable under the favorable conditions tied to warm events (W. Baethgen 1998, personal communication). The diminishing effects of high precipitation anomalies on maize yields (Fig. 5a) suggest that potential yields during warm events are not being exploited fully, and there is an opportunity for yield increases through increased inputs. In the last few years, agricultural production systems in Argentina have changed radically: fertilizer and agrochemical usage has increased significantly. It should not be surprising, therefore, if the effects of warm ENSO events on crop yields become more marked in the future. For instance, the increased use of fertilization and other technology advances allowed farmers to capitalize on favorable conditions (enhanced rainfall) associated with the strong 1997 warm ENSO event. Thus, all-time record maize yields were achieved in Argentina.

Results presented here may be relevant to a broad range of decision makers in the agricultural sector. Individual farmers may consider our findings to gauge risks associated with ENSO events. The existence of a demonstrable ENSO impact may provide an incentive to farmers to consider more flexible management to take full advantage of climate information. This goal, however, will require further research and an appropriate set of decision-support tools. Differences in variability, however, must be kept in mind when extrapolating national-level results (as shown here) to district or even individual enterprise scales. Given Argentina’s significant participation in the global maize and soybean markets, traders in Argentina and elsewhere might make better projections of aggregated supply conditions for a crop, which in turn may determine the market price (Mjelde et al. 1997). As in other parts of the world (Smit et al. 1997), there is currently a trend in Argentina toward the government shifting some of the climate-related risks to individual farmers. Agricultural emergency laws are currently under review, and crop insurance will probably become more common in the future. For these reasons, public sector policy makers may benefit from quantitative information on potential impacts of ENSO.

Turning skillful but uncertain ENSO-related climate forecasts into useful information and beneficial decisions is a major challenge for the immediate future (Trenberth 1997a). A first necessary step toward the adoption and effective use of climate forecasts in agriculture is a characterization of the impact of ENSO. If there is no ENSO signal on crop yields or economic returns in a region, it is unlikely that agricultural stakeholders will benefit from the forecasts (unless one considers impacts on other world producers of the commodity, which may influence output prices). However, we stress that simply documenting that ENSO affects yields does not imply that the agriculture sector will benefit from the adoption of climate forecasts. For forecasts to have beneficial effects on this sector, and on society in general, they must induce changes in the decision-making process and in the actions taken by sector agents (Sonka et al. 1987; Hammer et al. 1996). The exploration of alternative management options in response to expected climate scenarios is the topic of future research.

Acknowledgments

This work was supported by grants from NOAA (Office of Global Programs) and the National Science Foundation (Methods and Models for Integrated Assessment Initiative) to a consortium of Florida universities (University of Miami, The Florida State University, University of Florida). The Inter-American Institute for Global Change Research (IAI) provided additional funding as part of its Initial Science Program. C. Messina and M. Grondona conducted part of the research during IAI fellowships at the University of Miami. Drs. J. J. O’Brien and D. Legler kindly provided the JMA SST anomalies. Helpful comments by Drs. J. Hansen and J. Jones are gratefully acknowledged.

REFERENCES

  • Adams, R. M., K. J. Bryant, B. A. McCarl, D. M. Legler, J. J. O’Brien, A. Solow, and R. Weiher, 1995: Value of improved long-range weather forecasts. Contemp. Econ. Policy,13, 10–19.

  • Barnston, A. G., and Coauthors, 1994: Long-lead seasonal forecasts—Where do we stand? Bull. Amer. Meteor. Soc.,75, 2097–2114.

  • Cane, M. A., G. Eshel, and R. W. Buckland, 1994: Forecasting Zimbabwean maize yield using eastern equatorial Pacific sea surface temperature. Nature,370, 204–205.

  • Cantagallo, J. E., C. A. Chimenti, and A. J. Hall, 1999: Number of seeds per unit area in sunflower correlates well with a photothermal quotient. Crop Sci., in press.

  • Carlson, R. E., D. P. Todey, and S. E. Taylor, 1996: Midwestern corn yield and weather in relation to extremes of the Southern Oscillation. J. Prod. Agric.,9, 347–352.

  • Chen, D., S. E. Zebiak, A. J. Busalacchi, and M. A. Cane, 1995: An improved procedure for El Niño forecasting: Implications for predictability. Science,269, 1699–1702.

  • Cleveland, R. B., W. S. Cleveland, J. E. McRae, and I. Terpenning, 1990: STL: A seasonal-trend decomposition procedure based on Loess. J. Off. Stat.,6, 3–73.

  • Cleveland, W. S., and S. J. Devlin, 1988: Locally-weighted regression:An approach to regression analysis by local fitting. J. Amer. Stat. Assoc.,83, 596–610.

  • Connor, D. J., and V. O. Sadras, 1992: Physiology of yield expression in sunflower. Field Crops Res.,30, 333–389.

  • Dai, A., I. Y. Fung, and A. D. Del Genio, 1997: Surface observed global land precipitation variations during 1900–88. J. Climate,10, 2943–2962.

  • Díaz, A. F., C. D. Studzinski, and C. R. Mechoso, 1998: Relationships between precipitation anomalies in Uruguay and southern Brazil and sea surface temperature in the Pacific and Atlantic Oceans. J. Climate,11, 251–271.

  • Díaz, R. A., 1995: Seeking practical applications for short climate predictions in the agriculture of Argentina. International Research Institute for Climate Prediction Rep. IRIP-TL-95/1, 40 pp. [Available from INTA, Instituto de Clima y Agua, 1712 Castelar, Prov. Buenos Aires, Argentina.].

  • Dilley, M., 1997: Climatic factors affecting annual maize yields in the Valley of Oaxaca, Mexico. Int. J. Climatol.,17, 1549–1557.

  • Garnett, E. R., and M. L. Khandekar, 1992: The impact of large-scale atmospheric circulations and anomalies on Indian monsoon droughts and floods and on world grain yields—A statistical analysis. Agric. For. Meteor.,61, 113–128.

  • Hall, A. J., J. H. Lemcoff, and M. Trápani, 1981: Water stress before and during flowering in maize and its effect on yields, its components and their determinants. Maydica,26, 19–38.

  • ——, C. M. Rebella, C. M. Ghersa, and J. P. Culot, 1992: Field-crop systems of the Pampas. Ecosystems of the World: Field Crop Ecosystems, C. J. Pearson, Ed., Elsevier, 413–449.

  • Hammer, G. L., D. P. Holzworth, and R. Stone, 1996: The value of skill in seasonal climate forecasting to wheat crop management in a region with high climatic variability. Aust. J. Agric. Res.,47, 717–737.

  • Handler, P., 1990: USA corn yields, the El Niño and agricultural drought: 1867–1988. Int. J. Climatol.,10, 819–828.

  • Hansen, J. W., and Coauthors, 1996: ENSO effects on yields and economic returns of wheat, corn, and soybean in Argentina. Actas del VII Congreso Argentino y VII Congreso Latinoamericano e Ibérico de Meteorología, Buenos Aires, Argentina, Centro Argentino de Meteorólogos and Federacíon Latinoamericana e Ibérica de Sociedades de Meteorología, 89–90.

  • Hoogenboom, G., J. W. Jones, and K. J. Boote, 1991: A decision support system for prediction of crop yield, evapotranspiration, and irrigation management. Proc. Irrigation and Drainage, Honolulu, HI, Irrigation Division, American Society of Civil Engineers, 198–204.

  • Lanzante, J. R., 1996: Resistant, robust and non-parametric techniques for the analysis of climate data: Theory and examples, including applications to historical radiosonde station data. Int. J. Climatol.,16, 1197–1226.

  • Legler, D. M., K. J. Bryant, and J. J. O’Brien, 1999: Impact of ENSO-related climate anomalies on crop yields in the U.S. Climatic Change, in press.

  • Manly, B. F. J., 1997: Randomization, Bootstrap and Monte Carlo Methods in Biology. Chapman and Hall, 399 pp.

  • Meinke, H., R. C. Stone, and G. L. Hammer, 1996: SOI phases and climatic risk to peanut production: A case study for northern Australia. Int. J. Climatol.,16, 783–789.

  • Messina, C., A. Beltrán, and A. Ravelo, 1996a: El fenómeno ENSO:Su relación con la productividad de maíz en la región pampeana argentina (The ENSO phenomenon: Its relationship with maize productivity in the Argentine Pampas). IV Congreso Colombiano de Meteorología, Santa Fe de Bogota, Colombia, Sociedad Colombiana de Meteorología, 236–241.

  • ——, ——, and ——, 1996b: La variabilidad interanual de los rendimientos de trigo en la región pampeana y su relación con el fenómeno ENSO (El Niño/Southern Oscillation) (Interannual variability of wheat yields in the Pampas, and their association with the ENSO phenomenon). Actas del VII Congreso Argentino y VII Congreso Latinoamericano e Ibérico de Meteorología, Buenos Aires, Argentina, Centro Argentino de Meteorólogos and Federacíon Latinoamericana e Ibérica de Sociedades de Meteorología, 55–56.

  • Meyers, S. D., J. J. O’Brien, and E. Thelin, 1999: Reconstruction of monthly SST in the tropical Pacific Ocean during 1868–1993 using adaptive climate basis functions. Mon. Wea. Rev.,127, 1599–1612.

  • Mjelde, J. W., T. N. Thompson, F. M. Hons, J. T. Cothren, and C. G. Coffman, 1997: Using Southern Oscillation information for determining corn and sorghum profit-maximizing input levels in east-central Texas. J. Prod. Agric.,10, 168–175.

  • Nicholls, N., 1985: Impact of the Southern Oscillation on Australian crops. J. Climatol.,5, 553–560.

  • Phillips, J. G., C. Rosenzweig, and M. Cane, 1996: Exploring the potential for using ENSO forecasts in the U.S. Corn Belt. Drought Network News,8, 6–10.

  • ——, M. A. Cane, and C. Rosenzweig, 1998: ENSO, seasonal rainfall patterns and simulated maize yield variability in Zimbabwe. Agric. For. Meteor.,90, 39–50.

  • Pisciottano, G., A. Díaz, G. Cazes, and C. R. Mechoso, 1994: El Niño–Southern Oscillation impact on rainfall in Uruguay. J. Climate,7, 1286–1302.

  • Rao, V. B., L. D. A. Sá, S. H. Franchito, and K. Hada, 1997: Interannual variations of rainfall and corn yields in Northeast Brazil. Agric. For. Meteor.,85, 63–74.

  • Rebella, C. M., R. Díaz, and L. A. Funston, 1984: Tendencias de lluvias y rendimientos de maíz en la región maicera típica argentina (Precipitation trends and maize yields in Argentina’s corn belt). Proc. III Congreso Nacional de Maíz, Pergamino, Argentina, Asociación Argentina de Ingenieros Agrónomos del Norte de Buenos Aires, 129–133.

  • Ropelewski, C. F., and M. S. Halpert, 1987: Global- and regional-scale patterns associated with the El Niño/Southern Oscillation. Mon. Wea. Rev.,115, 1606–1626.

  • ——, and ——, 1989: Precipitation patterns associated with the high index phase of the Southern Oscillation. J. Climate,2, 268–284.

  • ——, and ——, 1996: Quantifying Southern Oscillation–precipitation relationships. J. Climate,9, 1043–1059.

  • SAGPyA, 1994: Estadísticas agropecuarias y pesqueras 1994. Ministerio de Economía y Obras y Servicios Públicos, Secretaría de Agricultura, Ganadería, Pesca y Alimentación, Argentina, 459 pp.

  • Smit, B., R. Blain, and P. Keddie, 1997: Corn hybrid selection and climatic variability: Gambling with nature? Can. Geogr.,41, 429–438.

  • Sokal, R. R., and F. J. Rohlf, 1969: Biometry. W. H. Freeman, 776 pp.

  • Sonka, S. T., J. W. Mjelde, P. J. Lamb, S. E. Hollinger, and B. L. Dixon. 1987: Valuing climate forecast information. J. Climate Appl. Meteor.,26, 1080–1091.

  • Tanco, R., and G. J. Berri, 1996: Acerca del efecto del fenómeno El Niño sobre la precipitación en la Pampa húmeda argentina (On the effects of the ENSO phenomenon on precipitation in the Argentine Pampas). Actas del VII Congreso Argentino y VII Congreso Latinoamericano e Ibérico de Meteorología, Buenos Aires, Argentina, Centro Argentino de Meteorólogos and Federacíon Latinoamericana e Ibérica de Sociedades de Meteorología, 319–320.

  • Travasso, M. I., 1990: Evaluation des facteurs agroclimatiques limitant le rendement du ble dans le sud de Buenos Aires, Argentine. Essai de modelisation a l’echelle regionale (Assessment of agroclimatic factors limiting wheat yields in southeastern Buenos Aires, Argentina. A modelling test on regional scales). Ph.D. thesis, Universite de Paris, 96 pp. [Available from INTA, Instituto de Clima y Agua, 1712 Castelar, Prov. Buenos Aires, Argentina.].

  • Trenberth, K., 1996: El Niño–Southern Oscillation. Climate Change:Developing Southern Hemisphere Perspectives, T. W. Giambelluca and A. Henderson-Sellers, Eds., John Wiley and Sons, 145–173.

  • ——, 1997a: Short-term climate variations: Recent accomplishments and issues for future progress. Bull. Amer. Meteor. Soc.,78, 1081–1096.

  • ——, 1997b: The definition of El Niño. Bull. Amer. Meteor. Soc.,78, 2771–2777.

  • Whiteman, P. C., and G. L. Wilson, 1965: Effects of water stress on the reproduction development of Sorghum vulgare. Pers. University of Queensland, Paper 14. [Available from University of Queensland, Brisbane, Queensland 4072, Australia.].

  • Widmann, M., and C. Schär, 1997: A principal component and long-term trend analysis of daily precipitation in Switzerland. Int. J. Climatol.,17, 1333–1356.

  • Wilks, D. S., 1995: Statistical Methods in the Atmospheric Sciences. Academic Press, 467 pp.

Fig. 1.
Fig. 1.

The study area. Black circles indicate location of rainfall stations.

Citation: Journal of Applied Meteorology 38, 10; 10.1175/1520-0450(1999)038<1488:ABGCYI>2.0.CO;2

Fig. 2.
Fig. 2.

Time series of maize yields, 1900–95. Open squares indicate warm ENSO events; filled squares correspond to cold events. Neutral years are shown as open circles. The dashed line indicates the estimated low-frequency trend.

Citation: Journal of Applied Meteorology 38, 10; 10.1175/1520-0450(1999)038<1488:ABGCYI>2.0.CO;2

Fig. 3.
Fig. 3.

Time series of relative yield residuals for summer crops, for the overlapping data period (1972–95). Warm ENSO events are indicated by gray bars; black bars indicate cold events; (a) maize, (b) sunflower, (c) sorghum, and (d) soybeans.

Citation: Journal of Applied Meteorology 38, 10; 10.1175/1520-0450(1999)038<1488:ABGCYI>2.0.CO;2

Fig. 4.
Fig. 4.

Boxplot of PC1 amplitudes for Nov–Jan, by ENSO phase. Lower and upper boundaries for each box are the 25th and 75th percentiles. The line inside each box indicates the median. Whiskers mark the range of the values.

Citation: Journal of Applied Meteorology 38, 10; 10.1175/1520-0450(1999)038<1488:ABGCYI>2.0.CO;2

Fig. 5.
Fig. 5.

Yield anomalies for summer crops as a function of PC1 amplitudes for the Nov–Jan series. Negative PC1 amplitudes indicate negative precipitation anomalies, and vice versa; (a) maize, (b) sunflower, (c) sorghum, and (d) soybeans.

Citation: Journal of Applied Meteorology 38, 10; 10.1175/1520-0450(1999)038<1488:ABGCYI>2.0.CO;2

Table 1.

Countrywide crop data series for Argentina. The table shows the first cropping cycle and the length of each analyzed series;all series end in 1995. The number of warm and cold ENSO events and neutral years for each crop series also is shown.

Table 1.
Table 2.

Warm and cold ENSO events between 1900 and 1995, as defined by the JMA index. Note that an “ENSO year” encompasses the period between July of the listed year and June of the following year. For example, the warm event of 1982 includes the period between Jul 1982 and Jun 1983. Years not listed are considered to be neutral.

Table 2.
Table 3.

Descriptive statistics for yield residuals of all crops analyzed. The first row for each statistic shows absolute residuals (kg ha−1).The second row of values shows relative residuals expressed as a percentage of the expected trend for any given year. The numbers in parentheses next to the maximum and minimum absolute and relative anomalies indicate the year in which these yield extremes occurred, as well as the corresponding ENSO phase (warm, W; neutral, N; or cold, C).

Table 3.
Table 4.

Contingency table of yield residuals by yield terciles and ENSO phase. Results from a χ2 test are listed at the bottom of the table (“df” indicates degrees of freedom and P is the probability of rejecting the null hypothesis when it is really true). Rowwise percentage of years in each yield tercile is shown in parentheses for each ENSO phase; values may not add to 100 because of rounding.

Table 4.
Table 5.

Estimates of central tendency and spread of yield residuals (in percentage of expected yield) for an entire crop series, and by ENSO phase. The central tendency is described by the mean and the median. Estimates of spread are the SD and a PSD computed from the median absolute deviation of residuals. Here n indicates the number of years in each ENSO phase.

Table 5.
Table 6.

Probability values of (a) Kruskal–Wallis tests of differences in central tendencies of yield residuals among all ENSO phases;(b) one-tailed Wilcoxon tests of differences in central tendencies of yield residuals between pairs of ENSO phases; and (c) tests of differences in spread of yield residuals between pairs of ENSO phases.For this last set of tests, the statistic was the ratio of squared pseudostandard deviations, and significance was estimated via a randomization procedure. Probability values in parentheses for the wheat and sunflower Wilcoxon tests indicate lack of significance in central tendency among all ENSO phases, as indicated by the Kruskal–Wallis tests.

Table 6.
Save