Reprint of “Quantifying yield gaps in rainfed cropping systems: A case study of wheat in Australia”

Abstract

To feed a growing world population in the coming decades, agriculture must strive to reduce the gap between the yields that are currently achieved by farmers (Ya) and those potentially attainable in rainfed farming systems (Yw). The first step towards reducing yield gaps (Yg) is to obtain realistic estimates of their magnitude and their spatial and temporal variability. In this paper we describe a new yield gap assessment framework. The framework uses statistical yield and cropping area data, remotely sensed data, cropping system simulation and GIS mapping to calculate wheat yield gaps at scales from 1.1km cells to regional. The framework includes ad hoc on-ground testing of the calculated yield gaps. This framework was applied to wheat in the Wimmera region of Victoria, Australia. Estimated Yg over the whole Wimmera region varied annually from 0.63 to 4.12Mgha−1with an average of 2.00Mgha−1. Expressed as a relative yield (Y%) the range was 26.3–77.9% with an average of 52.7%. Similarly large spatial variability was described in a Wimmera yield gap map. Such maps can be used to show where efforts to bridge the yield gap are likely to have the biggest impacts. Bridging the exploitable yield gap in the Wimmera region by increasing average Y% to 80% would increase average annual wheat production from 1.09Mtonnes to 1.65Mtonnes. Model estimates of Yw and Yg were compared with data from crop yield contests, experimental variety trials, and on-farm water use and yields. These alternative approaches agreed well with the modelling results, indicating that the proposed framework provided a robust and widely applicable method of determining yield gaps. Its successful implementation requires that: (1) Ya as well as the area and geospatial distribution of wheat cropping are well defined; (2) there is a crop model with proven performance in the local agro-ecological zone; (3) daily weather and soil data (such as PAWC) required by crop models are available throughout the area; and (4) local agronomic best practice is well defined.