Predicting Genotype × Environment × Management (G × E × M) Interactions for the Design of Crop Improvement Strategies

Breeding crops for drought-affected environments and improved climate resilience

“…To target breeding for water-limited environments, utilizing contributions from drought resistance traits (escape, avoidance, and tolerance) and trait networks, it is important to distinguish between water deficits that reduce crop yield while still enabling sufficient yield levels to remain above the threshold for viable agricultural productivity, which we will refer to as agricultural drought, and the catastrophic, severe water deficits that reduce yield below the viability threshold but still allow plant survival, which we will refer to as survival drought; the threshold that distinguishes between agricultural and survival droughts will vary for different agricultural systems, ranging from small-scale subsistence to large-scale industrial ( Ceccarelli, 1994 ; Cooper and Hammer, 1996 ; Blum, 2011a ; Chenu et al, 2011 ; Van Ittersum et al, 2013 , 2016 ; Gaffney et al, 2015 ; Kholová et al, 2021 ; Tardieu, 2022 ). Within this review, we focus on enabling and accelerating breeding for yield productivity for agricultural droughts, where crop yields are reduced due to water limitations, but remain above the threshold for viable agricultural productivity ( Blum, 2011a ; Lobell et al, 2014 ; Gaffney et al, 2015 ; Cooper et al, 2020 , 2021a , 2023 ; Kholová et al, 2021 ; Messina et al, 2022a ; Tardieu, 2022 ). The potential of such breeding strategies, designed to develop products that can help to reduce yield losses from agricultural droughts, to contribute to long-term climate resilience of agriculture will be considered.…”

“…evapotranspiration) ( Figure 3 ; Woltereck, 1909 ; Gregorius and Namkoong, 1986 ; Boer et al, 2007 ; Jarquín et al, 2014 ; Van Eeuwijk et al, 2016 ; Irmak et al, 2019 ; Cooper et al, 2021a ; Messina et al, 2022a ). Breeding for drought tolerance requires consideration of and selection for genotype yield reaction-norms across environments, and therefore the extent and factors contributing to genotype-by-environment ( GxE ) interactions within the TPE; herein, this includes consideration of G x (E x M) interactions, as discussed above ( Figure 3 ; Cooper and DeLacy, 1994 ; Cooper and Hammer, 1996 ; Basford and Cooper, 1998 ; Cooper et al, 2001 , 2014a , 2014b , 2020 , 2021a , 2023 ; Messina et al, 2009 , 2022a , 2022c ; Gaffney et al, 2015 ; Van Eeuwijk et al, 2016 ; Gage et al, 2021 ; Li et al, 2021 ; Diepenbrock et al, 2022 ). The incidence of GxE interactions for crop productivity traits, such as yield, has long been recognized as a vexing challenge with major implications for many aspects of plant breeding, including trait prediction, selection, and on-farm product performance ( Figures 1–3 ; Haldane, 1946 ; Falconer, 1952 ; Comstock and Moll, 1963 ; Finlay and Wilkinson, 1963 ; Allard and Bradshaw, 1964 ; Eberhart and Russell, 1966 ; Knight, 1970 ; Ceccarelli, 1989 , 1994 ; Cooper and DeLacy, 1994 ; Cooper and Hammer, 1996 ; Blum, 2011a ; Gaffney et al, 2015 ; Van Eeuwijk et al, 2016 ; Ceccarelli and Grando, 2020a , 2020b ; Gage et al, 2021 ;…”

“… Van Ittersum et al (2013) suggested using a reference of 80% of the yield potential that could be achieved in the absence of abiotic or biotic limitations to crop productivity as a practical target for designing improved crop management strategies to reduce on-farm yield-gaps; the yield-gap being the difference in yield between the potential crop yield within the environment with no resource limitations and the actual yield that is achieved by the farmer ( Fischer et al, 2014 ). We adopt the same convention with respect to crop breeding for improved levels of drought resistance ( Cooper et al, 2020 , 2023 ; Messina et al, 2022a ). That is, consideration should be given to targeted breeding to improve drought resistance when the total supply of water (from stored soil moisture, rainfall, and irrigation) available to a crop during its lifecycle is below the crop demand needed to achieve consistent harvestable yields of at least 80% of the potential yield in the target agricultural environment ( Ludlow and Muchow, 1990 ; Fukai and Cooper, 1995 ; Campos et al, 2004 ; Messina et al, 2011 , 2022a , 2022b ; Van Ittersum et al, 2013 ; Cooper et al, 2014a , 2020 , 2021a , 2021b , 2023 ; Fischer et al, 2014 ; Gleason et al, 2022 ).…”

“…We adopt the same convention with respect to crop breeding for improved levels of drought resistance ( Cooper et al, 2020 , 2023 ; Messina et al, 2022a ). That is, consideration should be given to targeted breeding to improve drought resistance when the total supply of water (from stored soil moisture, rainfall, and irrigation) available to a crop during its lifecycle is below the crop demand needed to achieve consistent harvestable yields of at least 80% of the potential yield in the target agricultural environment ( Ludlow and Muchow, 1990 ; Fukai and Cooper, 1995 ; Campos et al, 2004 ; Messina et al, 2011 , 2022a , 2022b ; Van Ittersum et al, 2013 ; Cooper et al, 2014a , 2020 , 2021a , 2021b , 2023 ; Fischer et al, 2014 ; Gleason et al, 2022 ). Here, we consider trait networks to be coordinated combinations of multiple traits, which together operate in ways that contribute to enhanced responses of crops to specific environments, or a range of environmental conditions that occur within a target population of environments (TPEs), beyond the responses that can be achieved by the individual traits operating in isolation ( Hammer et al, 2006 ; Cooper et al, 2009 , 2021a ; Tardieu et al, 2021 ; Gleason et al, 2022 ).…”

“…The combined influences of both the E and M dimensions of the agricultural environment on crop water availability must be considered in breeding for drought resistance. Genetic improvement ( ΔG ; as predicted by the “breeder’s equation” is targeted at the combined environment–management ( E x M ) conditions of the agricultural environment and thus must deal with the opportunities and complexities of genotype-by-environment-by-management ( G x E x M ) interactions ( Figure 1 ; Cooper and Hammer, 1996 ; Lynch and Walsh, 1998 ; Cooper et al, 2001 , 2020 , 2021a , 2023 ; Messina et al, 2009 ; Walsh and Lynch, 2018 ; Irmak et al, 2019 ; Peng et al, 2020 ; Hunt et al, 2021 ; Kholová et al, 2021 ; Zhao et al, 2022 ). Furthermore, whenever genotype-by-management interactions are important for effective crop water use for agricultural environments, this may require optimization and targeted improvement of genotype–management ( G–M ) technologies and focused attention on the influence of (G–M)xE interactions for current and future environments ( Duvick et al, 2004 ; Hammer et al, 2014 , 2020 ; Gaffney et al, 2015 ; Irmak et al, 2019 ; Cooper et al, 2020 , 2021a , 2023 ; Peng et al, 2020 ; Hunt et al, 2021 ; Messina et al, 2022a , 2022c ; Zhao et al, 2022 ).…”