Heat tolerance decreases with elevation
The decrease in heat tolerance with elevation and associated change in temperature is not a function of species turnover and was observed both across and within species. This trend is consistent with patterns along an elevation gradient in Colombia (Feeley et al . 2020a), and with trends in heat tolerance with mean annual or mean growing season temperature across latitudes (O’Sullivan et al. 2017; Zhuet al . 2018). The fact that elevation was a better predictor of heat tolerance than MAT or the mean maximum temperature of the warmest month may reflect greater accuracy of elevation than of temperature data, as the WorldClim temperature data is spatially gridded at ~1 km2 resolution. T50increased by 1.6°C from a mean of 48.5°C at Cerro Jefe to 50.1°C in Panama City, while MAT increased by 4.9°C from 22.1°C to 27.0°C, with an average increase across sites of 0.4°C per °C MAT. Thus, while there is a clear adjustment to temperature along this gradient, the thermal safety margin based on MAT—i.e., the difference between MAT and T50—decreases with increasing growth temperature and is smallest in the already relatively hot lowlands.
The 0.4°C increase in T50 per °C change in MAT was comparable to the 0.38°C increase in T50 of F0 rise per °C mean temperature of the warmest month in the global dataset by O’Sullivan et al . (2017). However, along the tropical elevation gradient studied by Feeley et al . (2020a), T50 increased by only 0.08°C °C–1MAT. One possible reason for the strong response of heat tolerance to local temperature compared to Feeley et al . (2020a) is the relatively narrow ranges in environmental conditions in the current study. Heat tolerance can increase with drought, cold, and other stressors (Sapper 1935; Kappen 1964; Ladjal et al . 2000; Sastryet al . 2018), as mild conditioning to other stressors may increase antioxidant scavenging (Gill & Tuteja 2010) that would also benefit heat tolerance. T50 of the warmest site in Feeley et al . (2020a) was comparable to that observed in the current study, so it is the cooler sites in Feeley et al . that have relatively high heat tolerance. The coolest site in Feeley et al . (2020a) is at almost 3000 m asl. At such elevation UV radiation can be strong, potentially causing cellular damage against which antioxidant activity can provide protection (Foyer et al . 1994). At another high-UV site in Colombia, T50 values averaged 49.2°C (Leon-Garcia & Lasso 2019), despite mean annual temperatures of <10°C, supporting the notion that high heat tolerance is not solely associated with high growing temperatures. The growing conditions at our pre-montane sites are favorable for most plants, and the elevation gradient represents a temperature gradient not confounded by gradients in other stressors such as UV radiation, cold, or drought.
Three species were collected both in lowland and in pre-montane forest. Of these, two showed the expected pattern in heat tolerance with elevation and temperature; Aspidosperma spruceanum —primarily a lowland species—and Posoqueria latifolia —a species widely distributed across lowland and pre-montane sites. The species that did not show any trends was Podocarpus guatemalensis , a species of lower montane regions. This species maintained comparatively low heat tolerance when it was growing under warm lowland conditions. This is consistent with observations that while heat tolerance can be downregulated in response to (seasonal) low temperature (e.g., Sastry & Barua 2017; Zhu et al . 2018), there is limited evidence for the upregulation of heat tolerance in response to warming (but see Drakeet al . 2018), especially of tropical species (Krause et al . 2010, 2013).