Introduction
The effect of global environmental change on life on earth often exhibits threshold dynamics in which performance steeply and irreversibly drops off above or below a critical threshold (Scheffer 2009). One example is the heat tolerance of ectotherms: beyond a critical temperature threshold (TCrit) performance rapidly declines, eventually resulting in irreversible damage typically only a few degrees above TCrit, as shown in e.g. arthropods (García-Robledo et al . 2016; Franken et al . 2018), amphibians and reptiles (e.g., Duarte et al . 2012), as well as in primary producers such as phytoplankton (Padfield et al . 2016) and terrestrial plants (e.g., Sachs 1864; Sapper 1935; Knight & Ackerly 2003; Krause et al . 2010). To better anticipate the consequences of rising temperatures for ectotherms it will be important to understand variation in thermal thresholds among co-occurring species within a community and among different communities. Because of the dependence of virtually all organisms on primary producers, understanding thermal thresholds of photosynthesizing organisms is particularly important.
More than 150 years ago, Sachs (1864) reported that plants in his university’s botanical garden could withstand exposure to air temperatures up to 50°C without leaf damage, but that 51°C or higher temperatures killed the leaves. Later studies reported more variation in heat tolerance, with both lower and higher values being observed in the early 20th century (Sapper 1935), but it remained challenging to identify general patterns underlying this variation. In recent years leaf heat tolerance has received renewed interest as ongoing global warming and more frequent and intense heatwaves may push ecosystems past their critical thermal thresholds (Feeley et al. 2020a; Lancaster & Humphreys 2020; Perez & Feeley 2020a,b; Geange et al. 2021). Recent studies have shown that heat tolerance increases from the poles towards the tropics (O’Sullivan et al . 2017) and from high elevation to low elevation (Feeley et al . 2020a), in parallel with increasing temperatures under which plants develop. While consistent, the increase in thermal thresholds along these gradients is moderate: 0.38°C per °C increase in mean maximum temperature of the warmest month from poles to the tropics, and < 0.1°C °C–1 mean annual temperature from high to low elevation. Furthermore, at each latitude and elevation heat tolerance varies considerably among species, and the mechanisms underlying this variation have not been explored in detail.
Variation in heat tolerance among co-occurring species may represent functional or ecological differences in micro-climate adaptation among species. For example, Sapper (1935) already showed that sun species have higher heat tolerance than shade species, and Slot et al . (2018) showed that even within species, heat tolerance tends to be moderately higher in sun leaves than in shade leaves. Within site variation may also reflect different evolutionary histories of plants. Dick et al . (2013) reported that many common Neotropical tree species have emerged long enough ago to have previously experienced climatological conditions not unlike those predicted for the end of the current century. As such, Dick et al . (2013) proposed, these survivors of past extreme climates may be more likely to tolerate high temperatures than species in contemporary Neotropical forests that have emerged more recently under relatively cooler conditions. Neotropical forests have existed for at least ~58 million years (Wing et al . 2009) and are characterized not only by high species diversity, but also by great diversity at higher taxonomic levels. For example, on 50 hectares of forest in central Panama > 300 species of woody plants with >1 cm stem diameter have been identified, belonging to >60 families, including ancient families such as Fabaceae and Lecythidaceae and families of more recent origins such as Chrysobalanaceae and Apocynaceae. Within site variation of temperature sensitivity may thus also be related to the evolutionary history of species.
Tropical forest trees routinely experience high temperatures when exposed to direct irradiance, especially during moments of low wind speed, with leaf temperatures exceeding air temperature by as much as 10–18°C (Doughty & Goulden 2008; Rey-Sanchez et al . 2016; Fauset et al . 2018). Leaf traits can influence both the magnitude and rate of leaf heating (Jones 2013). For example, greater leaf width and leaf size are generally associated with a larger leaf boundary layer and greater decoupling of leaf and air temperatures (Jones 2013). Indeed, Fauset et al . (2018) parameterized a leaf energy balance model for tropical montane species and found that leaf width was the most important leaf morphological driver of species differences in the leaf-to-air temperature differential. Multiple leaf traits can also be combined into composite traits that characterize the dynamics of leaf temperatures. For example, the thermal time constant (τ; s) quantifies the thermal stability of a leaf, i.e. how rapidly leaf temperature responds to temporal variation in microclimate (Michaletz et al . 2015, 2016). The low τ of relatively small and thin leaves indicates that they heat up and cool down quickly, so that leaf temperatures essentially track changes in microclimate variables, including very high temperatures when in full sun. By contrast, leaves with a comparatively large thermal mass and a high τ are buffered against fluctuations in microclimate, so that leaf temperatures lag behind changes in microclimate variables (Michaletz et al . 2015, 2016). Leaf temperature can also be affected by transpirational cooling, especially in hot and dry environments (Lin et al. 2017), and thus stomatal conductance is another potentially relevant leaf trait. However, because of the prevalence of mid-day stomatal depression (Zotz et al. 1995; Goulden et al. 2004; Kosugi et al. 2008), the highest leaf temperatures are experienced when stomata are closed, and maximum stomatal conductance is therefore unlikely to distinguish maximum temperatures across species. Furthermore, a sensitivity analysis revealed that variation in stomatal conductance had virtually no effect on the thermal time constant τ (Michaletz et al. 2016)
One would expect a strong selective advantage of high heat tolerance in species that have traits that cause them to experience high maximum temperatures. Indeed, heat tolerance scaled with maximum recorded leaf temperatures in a botanical garden in Florida (Perez & Feeley 2020a). High heat tolerance would likewise seem advantageous for species with traits that result in high temperatures being maintained for long periods of time. The rapid response to warming of low τ species should predispose them to experiencing higher maximum temperatures, whereas the slow cooling high τ species maintain high temperatures longer once they are reached. The photosynthetic capacity of species with low τ peaks at higher ambient temperatures than for species with high τ, suggesting that species with low τ are better acclimated to higher temperatures than high τ species (Michaletz et al . 2016). Whether low τ species also have higher heat tolerances under field conditions has not yet been tested.
Leaves that are structurally relatively costly to produce need to last long enough for the plant to offset the investment, and therefore tend to be better protected against biotic (Coley 1987) and abiotic stressors (Nardini et al . 2012). Correspondingly, high heat tolerance of such leaves would be expected. This has indeed been observed for plants experiencing distinct temperature seasonality (e.g., Knight & Ackerly 2003; Sastry & Barua 2017), but whether this pattern is maintained in more thermally stable ecosystems such as tropical forests remains to be tested.
We measured leaf heat tolerance for 147 tropical species from lowland and pre-montane forest sites in Panama and tested the following hypotheses:
1) Heat tolerance will be greater in species growing in lowland than in pre-montane sites, consistent with previously observed relationships between heat tolerance and growth temperature. Based on previous observations we expect this difference to be smaller than the difference in growth temperature. Over a 20°C growing season temperature range from the arctic to the tropics, mean heat tolerance increased by only 9°C (O’Sullivan et al . 2017), and over a 17°C mean annual temperature range across a tropical elevation gradient, Feeley et al . (2020a) reported an increase in heat tolerance of less than 2°C. These observations suggest that upper temperature thresholds are relatively high in cool conditions and increase only moderately with growth temperature.
2) Heat tolerance will be phylogenetically structured and lineages that can be considered survivors of past hot epochs will have higher heat tolerance than species or lineages that emerged later and are naïve to such conditions.
3) Heat tolerance will increase with traits that enable leaves to reach high temperature extremes, such that species that quickly heat up will have higher heat tolerance than species that are more buffered against high temperatures. This would be consistent with Perez & Feeley (2020a) and with the higher temperature optimum for photosynthesis of low τ species (Michaletz et al . 2016). Alternatively, heat tolerance may be greater in thermally buffered species because they cool down more slowly and thus maintain stable temperatures for longer. Heat tolerance may differ with duration of exposure; when individual leaves are warmed gradually over a wide temperature range (commonly by 1°C per minute) heat tolerance is often relatively low, suggesting a negative effect of the cumulative heat exposure (Krause et al. 2010). If the negative impact of sustained high temperatures is greater than that of higher, but shorter peak temperatures, heat tolerance might correlate positively with τ across species.
4) Species that produce structurally expensive leaves will have higher heat tolerance than species with ‘cheap’ leaves that can be readily replaced. High investment in leaves that are relatively thick and dense (high LMA), or with high leaf dry matter content (LDMC), is associated with higher leaf longevity (Wright et al . 2004), and thus with greater likelihood of the leaves experiencing temperature extremes during their lifetime.