Capturing heat stress induced variability in spikelet sterility using panicle, leaf and air temperature under field conditions

- Tropical summer seasons can be effectively used to determine heat stress induced sterility across diverse germplasm.
- Tissue temperature explains heat induced spikelet sterility better than air temperature.
- Spikelet sterility measured on whole panicle basis is a robust measure of heat damage.

Heat stress induced increase in spikelet sterility is often correlated with prevailing air temperature during anthesis. However, plant tissue temperature can be different from air temperature, depending on transpiration cooling which is a function of relative humidity (RH). Controlled environment studies involving heat stress are generally conducted at targeted RH, hence excluding the dynamics associated with this factor. Using 81 diverse indica cultivars over three years (wet season-2012, dry season-2013 and 2014), flag leaf temperature (FLT), panicle temperature (PT) and air temperature (AT) were measured during flowering under field conditions. On average the FLT and PT were 10.4 and 8.7 °C lower than AT during the wet season and were 9.9 and 8.0 °C lower when averaged across all three seasons. Spikelet sterility in dry seasons ranged between 1.6 and 67% compared to <1 and 35.5% during the wet season. Spikelet sterility was significantly correlated with PT (r = 0.50; p < 0.01) and FLT (r = 0.55; p < 0.01), compared with a weaker relationship with AT (r = 0.23; p < 0.05) among the tested 81 cultivars, pooled across both 2013 and 2014 dry seasons. By employing an unbiased Z score approach, cultivars with (i) consistent and contrasting for sterility percentage and (ii) phenology (duration from seeding till 50% flowering) based on flowering date confirmed the strong relationship between FLT or PT with spikelet sterility compared to AT. In conclusion, we recommend using either PT or FLT (tissue temperatures) instead of AT for future field based heat stress phenotyping. Where direct PT measurements are not available, mechanistic and empirical models can be used to derive PT or FLT from weather station data, or remotely sensed canopy temperature might be used as a proxy in large scale phenotyping experiments.