An integrated, multivariate characterisation of water and photothermal regimes for faba bean in Australia

Abstract

Environmental characterisation provides a useful summary of a major source of variation in grain yield. Environments consist of water and photothermal regimes that covary in time and space, but previous characterisations have focussed on single-variable regimes such as drought, or downplayed the temporal pattern of multivariate regimes. Season-long, multivariate characterisations are needed to more realistically represent the complex growing conditions that crops encounter. We conducted two studies on faba bean, an important source of plant protein, in Australia, its largest exporter. From a database of yield with 299 variety trials, Study 1 tested the timing and strength of the association of seed yield with maximum and minimum temperature, heat stress, frost, solar radiation, vapour pressure deficit and simulated water supply:demand. Study 2 used cluster analysis of 30,096 simulated crops (1957–2022, three sowing dates, two varieties, 76 locations) to determine environment types for these variables, individually and combined. We tested the real-world relevance of the environment types with the seed yield data of Study 1. Water supply:demand, maximum temperature and vapour pressure deficit had the strongest links to grain yield in both studies. We identified four multivariate environment types that ranged from syndromes of ‘wet, cool and low evaporative demand’ to ‘dry, hot and high evaporative demand’. From least to most stressful environment type, median seed yield reduced by 62 %. Frequency of environment types varied with location, sowing date and variety, highlighting the potential value of earlier sowing and phenology for a large part of the country. From 1963–1992 compared with 1993–2022, the frequency of stressful environment types increased by 4 to 9 %, highlighting the need to adapt to a challenging future climate. Our findings can inform breeding, management and research of faba bean in Australia and beyond, and our multivariate method can be applied to other crops and environments.