4 | DISCUSSION
The tropical tree species Tabebuia rosea showed clear signs of photosynthetic acclimation to elevated temperature and CO2: treatment plants had moderately higher thermal optima for photosynthetic parameters VCMax and JMax, and downregulated respiration rates at high temperature, thereby improving photosynthetic performance at treatment conditions and supporting our hypothesis of acclimation. Higher thermal optima of net photosynthesis were observed regardless of measurement CO2 concentration, indicating that the shift in the optimum temperature was not solely due to suppression of photorespiration by elevated CO2 in the treatment dome. The observed reduction in VCMax in treatment plants is consistent with acclimation to elevated CO2. However, reduced VCMax has also been observed as a result of warming alone—e.g. in (sub-) tropical Eucalyptus provenances (Crous et al. 2018)—suggesting that the observed decrease in VCMax can be related to increased CO2, increased temperature, or a combination of the two. Indeed, in the tropical tree species A. glandulosa neither warming nor elevated CO2 significantly reduced VCMax, but the combination of the two did (Fauset et al. 2018)., In contrast to common observations in studies in which only CO2 is increased (e.g., Ainsworth & Rogers 2007; Klein & Ramon 2019), stomatal conductance and stomatal density were not reduced in our combined warming and CO2 enrichment experiment, suggesting that the CO2 effect may be temperature dependent.
As hypothesized, net photosynthesis of leaves developed under control conditions and transferred to treatment conditions for 1–2 weeks, did not acclimate to the same extent as leaves that had developed under treatment conditions, as indicated by smaller shifts in temperature optima, and greater overlap in the credible intervals of parameter estimates. We did not find support for our hypothesis that stomatal conductance would exert stronger control over net photosynthesis under treatment conditions; plants were not more inhibited by low stomatal conductance under treatment conditions than under control conditions. Accounting for stomatal conductance by examining temperature responses at a fixed Ci showed that stomatal conductance had only a small influence on TOpt, as the shifts in TOpt were comparable for P270 and P505 as they were for P400 and P800. The activation of Rubisco was not limiting net photosynthesis under growth conditions, neither in control nor treatment plants, but transferred plants exhibited signs of Rubsico activase limitation. Collectively, these results indicate that T. roseaplants exhibited strong acclimation to the combined effects of 4°C warming and doubling of CO2. Acclimation appeared to be underpinned by changes in the biochemical parameters, particularly in shifts in TOpt of VCMax associated with adjustment of ΔS.
Our Bayesian approach for fitting temperature response curves was motivated by complications associated with fitting these nonlinear models with ordinary least square regression. With five saplings per treatment the plant-level replication in our study was modest, and small sample size may constrain the ability to accurately identify treatment effects. The Bayesian approach enabled us to more reliably estimate how treatments affected the key parameters, and to more thoroughly probe the uncertainty around the estimated treatment effects. This is valuable for difficult-to-estimate model parameters such as TOpt: Uncertainty of TOpt can be asymmetric, especially when constrained by the lower or upper limits of temperature control of leaf cuvettes of commercially available portable photosynthesis systems. For example, the posterior distributions of TOpt of P400 were left skewed, because of the small number of observations below TOpt, and those of VCMax tended to be right skewed (Fig. 2). Even with few data points above TOpt of VCMax and JMax, its estimates were relatively well constrained with the help of informed priors, with the 10% and 90% percentiles separated by on average only 2.3°C.