Discussion
We investigated the assumption that heat tolerances promote carbon assimilation at higher temperatures. We did not find support for our first hypothesis (H1) that PSII heat tolerance is coordinated with Tmax. One reason that Tmax may not be directly correlated with PSII heat tolerance is because it and the Tmin parameter used to fit Eq. 2 are purported to have no physiological significance (Cunningham  S. C. & Read  J. 2003), and in many cases Tmin is unrealistically low (Slot & Winter 2017a). It is also likely that stomatal closure ceases carbon assimilation before the actual thermal limits of plant biochemistry (i.e. electron transport or NADPH and ATP generation) are reached (Slot & Winter 2017a b). While PSII heat tolerance and Tmaxwere not correlated we did find limited support for our hypothesis H1 that PSII heat tolerance provides a conservative high-temperature limit for Tmax. Our estimates of Tmaxcorresponded to the temperatures that caused between 0 and 45% damage to FV/FM, indicating that T50 may provide a reasonable upper bound for estimates of Tmax. On the other hand, our estimates of Tcrit, which corresponded to temperatures causing ~2% damage to FV/FM, were still higher than Tmax for one third of our study species.
The positive correlations we observed between T50, T95 and Ω support our second hypothesis (H2) that PSII heat tolerance is characteristic of thermal generalists. This is a notable result given that it is one of the only examples in plants providing an explicit physiological mechanism for the macroecological hypothesis that greater thermal variability should select for broader physiological tolerance (Janzen 1967; Perez et al. 2016). Specifically, our results showed that species with the greatest thermal ranges for photosynthesis also tend to have the highest PSII heat tolerances. These results are generally consistent with the predictions of leaf thermoregulatory theory. Deviations from this expectation may occur if there are acclimatory shifts of the PSII heat tolerance or photosynthetic traits away from their optimal values for which the leaf thermoregulatory theory was developed. However, these these deviations from theoretical trait relationships are unlikely given our resuls.
The negative correlations that we observed between T50, T95 and Popt does not support our hypothesis H3 as proposed in accordance with leaf thermoregulatory theory. Instead these results are consistent with the prediction that species with low carbon assimilation rates are likely to exhibit greater stress tolerance (Wright et al. 2004; Reich 2014). Indeed, maintenance of PSII heat tolerance imposes a large metabolic cost that ‘fast’ species may not be able to incur (see below).
Our final hypothesis (H4) posited that if PSII heat tolerance promoted greater carbon assimilation at higher temperatures, it should correspond to higher Topt. However, our results suggest that high PSII heat tolerance may actually reduce Topt. This counterintuitive relationship may be explained by the metabolic cost of maintaining high PSII heat tolerance. PSII heat tolerance is linked to increased production of heat shock proteins (Wahid, Gelani, Ashraf & Foolad 2007), isoprenoids (Logan & Monson 1999), photoprotective pigments (Krause et al. 2015), membrane-fortifying solutes (Hüve, Bichele, Tobias & Niinemets 2006), and the saturation of lipid bilayers (Zhu et al. 2018). The production of some of these metabolites may deplete the pools of NADPH and ATP that are available for carbon fixation as they are redirected to PSII thermoprotection (Süss & Yordanov 1986; Gershenzon, 1994; Wahid et al. 2007; Taylor, Smith, Slot & Feeley 2019; Voon & Lim 2019), explaining why both Topt and Popt decrease as PSII heat tolerance increases.
An important assumption we made was that our data were phylogenetically non-independent before we tested our hypotheses. Given that we measured species from a diverse set of families and clades (i.e., 21 species in 20 families), the topology and branch lengths of our phylogenetic tree are likely to provide a reasonable hypothesis of species relatedness. However, our assumption of phylogenetic non-independence could be violated if plasticity in PSII heat tolerance and carbon assimilation actually caused our trait estimates to be unrepresentative of each species (Way & Yamori 2014; Sastry, Guha & Barua 2018). That said, our results currently suggest that there is strong covariation between some PSII heat tolerances and carbon assimilation parameters within phylogenies. Regardless of any phylogenetic correction, we confirmed that at the species-level Tmax occurs at lower temperatures than T50 but not Tcrit, and that a community’s mean Tcrit may provide a reasonable approximation for Tmax; however we found little evidence to support the assumption that heat tolerance promotes carbon assimilation at high temperatures.
According to our phylogenetically corrected results, the only way that PSII heat tolerance may promote carbon assimilation at higher temperatures is by expanding Ω, but this benefit may be offset by concomitant decreases in Topt and Popt. This is potentially explained by high PSII heat tolerance promoting electron transport or the production of NADPH and ATP at high temperatures (Genty et al. 1989; Baker 2008). We noted that the heat tolerances that signify greater PSII impairment (i.e., greater Fv/Fm damage) tended to have stronger correlations with carbon assimilation parameters. This is consistent with the hypothesis that larger reductions in the quantum yield have a greater effect on plant carbon economics, and may explain why Tcrit heat tolerance was not correlated with any metric of carbon assimilation (Perez & Feeley 2020). Consequently, T95 may characterize plant thermal ecological strategies more effectively than T50, but provide overestimates of Tmax.
Our results suggest that the heat tolerances of PSII measured with dark-adapted quantum yield (FV/FM) are not ideal proxies for carbon assimilation. Heat tolerances estimated with light-adapted quantum yield (Fq’/Fm’) may be better proxies for assimilation (although these heat tolerances estimates are also subject to biases; Baker 2008). Importantly, we show that T50 provides an upper bound for Tmax. We also show that high PSII heat tolerance is characteristic of thermal generalist plant species with ‘slow’ carbon acquisition strategies. These results increase our understanding of the high temperature limits of photosynthesis and can potentially be used to explain macroecological patterns in plant responses to climate change. More specifically, since PSII heat tolerance can characterize thermal specialization, it may prove as a useful tool for predicting the thermal specialists and generalists that are hypothesized to be most and least vulnerable to climate change, respectively (Perez & Feeley 2020).
Acknowledgements: The authors would like to thank Fairchild Tropical Botanic Garden for providing access to their collections.