Ambient-condition performance provides trait-dependent prediction of wheat genotype responses under elevated CO2 and chamber-associated warming

Abstract

BACKGROUND: Future wheat-growing environments are exepected to be warmer, drier, and characterised by elevated atmospheric CO2 concentration (eCO2). While drought and elevated temperature (eT) generally reduce yield, eCO2 can partially offset these effects, although responses vary among genotypes. Identifying genotypes suited to future climates is therefore critical, yet large-scale screening under eCO2 remains constrained by the high cost and availability of specialised facilities. An alternative approach is to determine whether genotype performance under ambient conditions can predict responses under combined future climate conditions. However, the robustness of such predictive relationships under interacting environmental drivers remains unclear. METHODS: Nineteen wheat genotypes were grown outdoors under ambient CO2 and temperature (aCO2 + aT) and in open-top chambers under eCO2, where chamber conditions also produced passive daytime warming and reduced relative humidity. Plants were either well-watered or droughted, and biomass, grain yield, and days to maturity (DTM) were measured. Because open-top chambers impose a combined microclimatic shift, the chamber treatment represents an integrated eCO2 and warming environment rather than an independent manipulation of these drivers. Scaling relationships were used to test whether responses under ambient conditions predicted performance under these combined conditions. RESULTS: Genotypes differed significantly for all traits. Drought reduced biomass and yield, and increased DTM across environments. The combined chamber environment altered biomass, yield and development, with strong predictive relationships observed for biomass and DTM but weaker for yield. Predictive strength improved when comparisons were made under matched water treatments and was strongest when temperature conditions were aligned between environments. CONCLUSIONS: While genotype rankings were partially conserved across environments, these patterns were strongly trait-dependent, indicating that predictive relationships are context-specific. Low-cost experimental approaches that approximate combined future conditions may therefore provide an effective first step for genotype pre-screening prior to targeted validation under fully controlled climate manipulations.