Yield gap analysis of rainfed alfalfa in the United States

Abstract

The United States (US) is the largest alfalfa (Medicago sativa L.) producer in the world. More than 44% of the US alfalfa is produced under rainfed conditions, although it requires a relatively high amount of water compared to major field crops. Considering that yield and production of rainfed alfalfa have been relatively stagnant in the country for decades, there is a need to better understand the magnitude of yield loss due to water limitation and how far from yield potential current yields are. In this context, the main objective of this study was to estimate the current yield gap of rainfed alfalfa in the US. We collected 10 year (2009-2018) county-level government-reported yield and weather data from 393 counties within 12 major US rainfed alfalfa producing states and delineated alfalfa growing season using probabilistic approaches based on temperature thresholds for crop development. We then calculated county-level growing season rainfall (GSR), which was plotted against county-level yield to determine attainable yield (Ya) using frontier function analysis, and water-limited potential yield (Yw) using boundary function analysis. Average and potential water use efficiencies (WUE) were estimated, and associated yield gap referring to attainable (YGa) or water-limited yields (YGw) were calculated. Finally, we used conditional inference trees (CIT) to identify major weather-related yield-limiting factors to alfalfa forage yield. The frontier model predicted a mean Ya of 9.6 ± 1.5 Mg ha^-1 and an associated optimum GSR of 670 mm, resulting in a mean YGa of 34%. The boundary function suggested a mean Yw of 15.3 ± 3 Mg ha^-1 at the mean GSR of 672 ± 153 mm, resulting in a mean yield gap of 58%. The potential alfalfa WUE was 30 kg ha^-1 mm^-1 with associated minimum water losses of 24% of mean GSR, which was three times greater than the mean WUE of 10 kg ha^-1 mm^-1. The CIT suggested that GSR and minimum temperature in the season were the main yield-limiting weather variables in rainfed alfalfa production in the US. Our study also revealed that alfalfa was only limited by water availability in 21% of the environments. Thus, future research on management practices to narrow yield gaps at current levels of water supply is necessary.

Keywords: Medicago sativa L.; attainable yield; conditional inference tree; frontier yield function; growing season rainfall; linear boundary function; water-limited potential yield; yield gap.