Using yield prediction models to assess yield gains: a case study for wheat uri icon

abstract

  • A methodology for estimating yield gain using predictive models was tested in the Yaqui Valley of northwestern Mexico. The irrigated valley is representative of regions producing 40% of the developing world's wheat (Triticum aestivum L.) and has been the subject of intensive wheat research investment during the last 30 years. Over the 1968-1990 (year of planting) study period, the linear increase in yield (+/- standard error) was 57.2 (+/- 12.8) kg ha-1 y-1 (r2=0.49) or 1.27 (+/- 0.26) % y-1 (r2=0.51). However, considerable variation in both yield and average growing season temperature were observed over the period. In an attempt to allow for yearly weather variation, two yield models (one simulation-based and one regression-based) assuming no change in cultivar or management were used to predict the weather-based potential yield of each year. These predicted yields showed a significant linear decline over the study period apparently due to a small increase in temperature. When average Valley yields were adjusted against simulated yields to take into account the variation in weather, the overall linear rate of yield gain increased to 103 (+/- 14.8) kg ha-1 y-1 (r2=0.69) or 1.91 (+/- 0.22) % y-1 (r2=0.78). Consequently, the allowance for the temporal temperature change showed that the returns to improved crop management and breeding have been superior to those suggested by the increase in actual Valley yields.

publication date

  • 1994
  • 1994