Heat stress in temperate and tropical maize hybrids: A novel approach for assessing sources of kernel loss in field conditions

► We examined heat stress effects on final kernel numbers by means of a novel approach. ► Hybrids of contrasting genetic background were tested, temperate and tropical. ► Plant growth rate reduction was the main source of kernel loss due to heating. ► We identified traits associated with heat tolerance among tested genotypes. ► The new approach could be used in breeding programs as an aid to genotype selection. Temperate and tropical maize differ in their tolerance to heat stress but the ecophysiological bases for genotypic differences are poorly understood. Our objectives were (i) to assess the sources of kernel loss, and (ii) to identify the main differences in these traits among genotypes of contrasting genetic background. We used the classic relationships that associate final kernel number per plant (KNP) with plant (PGRCP) and ear (EGRCP) growth rates during the critical period for kernel set and developed an alternative approach based on the combined analysis of these relationships for assessing sources of kernel loss in field conditions. We identified three sources of loss associated with (i) PGRCP reductions (ΔKNP1), (ii) changes in biomass partitioning to the ear (ΔKNP2), and (iii) constraints not directly related to assimilate allocation to the ear (ΔKNP3). A partitioning index was also established (PI=EGRCP PGRCP−1). Field experiments included three contrasting maize hybrids (Te: temperate; Tr: tropical; TeTr: Te×Tr) grown under two temperature regimes (control and heated) during daytime hours. We tested heating (ca. 33–40°C at ear level) along two 15-d periods (GS1: pre-anthesis; GS2: from silking onwards). Final KNP was severely reduced by heating, and this negative effect was larger (i) when it occurred during silking (−75% for GS2) than before anthesis (−52% for GS1), and (ii) for the Te hybrid (−77%) than the TeTr (−69%) and the Tr (−44%) hybrids. The contribution of each source of loss to the decrease in KNP was 47% for ΔKNP1, 27% for ΔKNP2, and 32% for ΔKNP3. Variations in ΔKNP2 were explained by changes in PI (r2=0.85, P