The impact of climate variability on soybean yields in Argentina. Multivariate regression
Climate variability is examined and discussed in this work, emphasizing its influence over the fluctuation of soybean yield in the Pampas (central-eastern Argentina). Monthly data of rainfall, maximum and minimum temperatures, thermal range and seasonal rainfall were analysed jointly with the soybea...
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | JOUR |
| Materias: | |
| Acceso en línea: | http://hdl.handle.net/20.500.12110/paper_13504827_v14_n1_p3_Penalba |
| Aporte de: |
| Sumario: | Climate variability is examined and discussed in this work, emphasizing its influence over the fluctuation of soybean yield in the Pampas (central-eastern Argentina). Monthly data of rainfall, maximum and minimum temperatures, thermal range and seasonal rainfall were analysed jointly with the soybean yield in the period 1973-2000. Low-frequency variability was significant only in the minimum temperature during November in almost all the stations. This situation is favourable to the crop since during this month, seed germination, a growth stage sensitive to low temperatures, takes place. In the crop's core production region, 72% of the series of soybean yield presented a positive trend. Except in years with extreme rainfall situations, interannual variability of the soybean yield is in phase with the seasonal rainfall interannual variability. During these years, losses in the soybean crop occurred, with yield negative anomalies greater than one standard deviation. Soybean yield showed spatial coherence at the local scale, except in the crop's core zone. The association between each climate variable and yield did not show a defined regional pattern. Summer high temperature and rainfall excesses during the period of maturity and harvest have the greatest negative impact on the crop, whilst higher minimum temperatures during the growing season favour high yields. The joint effect of climate variables over yield was studied with multivariate statistical models, assuming that the effect of other factors (such as soil, technology, pests) is contained in the residuals. The regression models represent the estimates of the yield satisfactorily (high percentage of explained variance) and can be used to assess expected anomalies of mean soybean yield for a particular year. However, the predictor variables of the yield depend on the region. Copyright © 2007 Royal Meteorological Society. |
|---|