Introduction

The conservation of existing forests and active reforestation are pivotal to mitigating the worst impacts of climate change (Girardin et al., 2021; Griscom et al., 2017). However, periodic heat stress induced by rapid climate change (IPCC, 2022) threatens forest function and thereby the future success of these conservation efforts (Jordan et al., 2023). Increased frequency of hot, dry conditions has driven declines in forest tree populations (Hammond et al., 2022) and associated carbon accumulation in biomass (Anderegg et al., 2016; Brienen et al., 2015). Tropical rainforests may be particularly susceptible, given their canopies already experience temperatures in excess of their thermal limits for maximum photosynthesis (Mau et al., 2018). With the frequency of lethal temperatures predicted to increase with future global warming (Doughty et al., 2023), it is essential to understand the capacity of species to tolerate or avoid heat stress. The leaf thermal safety margin describes the difference between observed maximum leaf temperatures and thermal tolerance of photosynthesis, and as such is a useful proxy to determine vulnerability of ecosystems, species, or populations to climate warming. Photosynthetic heat tolerance is often determined by assessing irreversible damage to leaf photosystem II using chlorophyll fluorescence (Maxwell & Johnson, 2000). Two commonly used metrics are Tcrit and T50, representing the temperatures at which there is a 5% and 50% irreversible decline in photosystem II functioning. Higher photosynthetic heat tolerance is a common adaptation to warmer leaf temperatures (Perez & Feeley, 2020), with higher values generally found in leaves exposed to higher ambient temperatures (Geange et al., 2021; O’Sullivan et al., 2017), lower soil moisture (Cook et al., 2021), and higher radiation (Slot et al., 2019). While species originating from warmer habitats exhibit higher Tcrit, the increase observed across an increase in growth temperature is modest, ranging from 0.24 to 0.60 °C increase per 1 °C increase in mean annual temperature (O’Sullivan et al., 2017; Slot et al., 2021; Zhu et al., 2018). Consequently, higher observed leaf temperatures in warmer climates such as lowland tropical forests result in them having a narrower thermal safety margin, which has been observed across species in contrasting biomes (Kitudom et al., 2022; Perez & Feeley, 2020) and within species (Kullberg et al., 2023). Forest canopy leaf temperatures can be substantially different from ambient air temperatures (Blonder & Michaletz, 2018; Crous et al., 2023; Doughty et al., 2023; Mau et al., 2018; Rey-Sanchez et al., 2017; Song et al., 2017; Still et al., 2022). Leaf temperature is a result of the balance of sensible and latent heat fluxes, along with net incoming radiation, and is impacted by a broad suite of morphological and physiological traits that interact with the microclimate of the leaf (Campbell & Norman, 1998; Jones, 2013). Fully illuminated sunlit leaves are typically warmer than ambient air temperatures (Zhou et al., 2023), with the magnitude of this offset (∆T) varying due to differences in leaf thermal traits (Blonder & Michaletz, 2018), such as effective leaf width, solar absorptance profile, inclination angle and orientation, and dynamic stomatal conductance (g s) (Fauset et al., 2018; Guo et al., 2022; Perez & Feeley, 2020). Trait variation resulting in enhanced leaf cooling is an important strategy to avoid heat stress (Deva et al., 2020; Drake et al., 2018). Indeed, there is growing evidence that communities of plant species grown under warmer conditions preferentially express leaf trait combinations that decrease leaf warming compared to those grown under cooler conditions (Kitudom et al., 2022; Kullberg et al., 2023; Leigh et al., 2012; Middleby et al., 2024a; Posch et al., 2022; Wright et al., 2017), providing some support to the idea of plant thermoregulation and limited homeothermy (Blonder & Michaletz, 2018; Michaletz et al., 2016). However, it is unclear to what extent intra-specific genetic variation may also show similar patterns of limited homeothermy. A recent study in sagebrush reported lower canopy temperatures of warm origin populations, attributing this to differences in plant height altering canopy microclimate (Olsoy et al., 2023). Similarly, warm origin genotypes ofPopulus fremontii exhibited higher transpiration rates and consequently lower leaf temperatures compared to cool origin genotypes (Hultine et al., 2020). In addition, enhanced leaf cooling has been associated with more heat tolerant genotypes in common bean (Deva et al., 2020). On the other hand, evidence for increased thermal tolerance of warm-origin genotypes is mixed (Chen et al., 2016; Coast et al., 2022; Gimeno et al., 2008; Marias et al., 2016). If tropical trees do exhibit a simultaneous increase in thermal tolerance and decrease in leaf temperatures of warm origin provenances, there is a potential for warm origin genotypes to have greater thermal safety margins when planted under common conditions, such as in restoration plantings. However, if Tcrit acclimates to Tleafacross populations as it does across species and biomes (Perez & Feeley, 2020), this may ultimately lead to a convergence of thermal safety margins. Therefore, understanding how these two variables covary across provenances is important for the conservation and management of tropical rainforests. Here we tested whether upland and lowland provenances of four tropical tree species differed in their thermal safety margins when grown in a lowland common garden. This was achieved by measuring photosynthetic heat tolerance, leaf, and canopy temperatures, as well as leaf thermal traits which were then used to paramaterise a leaf energy balance model. We hypothesised that provenances from the warmer lowlands would have a lower ΔT compared to provenances from the cooler uplands (H1), and that this provenance variation could be explained by variation in both leaf thermal traits, and differences in microclimate in plants of differing canopy heights (H2). We also expected thermal tolerance to acclimate to Tleaf, such that Tcrit and T50 would show similar patterns of provenance-differentiation as ΔT, leading to a convergence of thermal safety margins across provenances (H3).