Figure legends
Fig. 1. Thermal thresholds T50 (a,b) and TCrit (c,d) as a function of elevation and mean annual temperature for 150 tropical species. Values for CAM plants and C4 plants are indicated in red and blue, respectively. Black and red lines and r2 values indicate linear regressions for the full dataset and for C3 plants only, respectively. For C3: a) T50=50.4–0.0026 × Elevation (m asl). b) T50=39.3+0.41×MAT. c) TCrit=47.4–0.0039 × Elevation. d) TCrit=31.2+0.60×MAT.
Fig 2. Thermal thresholds T50 and TCrit in relation to elevation for species that were measured at two contrasting elevations. Error bars indicate 95% confidence intervals approximated as T50 ±1.96 × standard error of T50. 1. Posoqueria latifolia , 2. Aspidosperma spruceanum , 3. Protium panamense , 4.Ceiba pentandra , 5. Coccoloba uvifera , 6. Podocarpus guatemalensis , 7. Guatteria lucens .
Fig 3. Relationship between thermal thresholds T50 and TCrit and species ‘ages’. Gray diamonds show data for Chrysophyllum cainito , a sister species ofC. argentum dated in Dick et al . (2013). Species names in blue represent Pleistocene; green, Pliocene; red, Miocene.
Fig. 4 . Phylogenetic tree of study species with branches color coded by the value of T50 standardized to sea level. Red and blue arrows indicate CAM-exhibiting species and C4species, respectively. Black nodes represent polytomies.
Fig. 5 . Leaf critical temperature (TCrit) as a function of the estimated thermal time constant (log τ) and elevation. Solid and dashed lines indicate significant and non-significant correlations for lowland and non-lowland data, respectively. Shaded areas indicate 95% confidence intervals of the predictions. Boxplots illustrate lowland (open bars) vs non-lowland (mid-elevation and pre-montane) sites (closed bars). ** P <0.001).
Fig. 6 . T50 as a function of leaf mass per area (LMA) and elevation. Solid and dashed lines indicate significant and non-significant correlations for lowland and non-lowland data, respectively. Shaded areas indicate 95% confidence intervals of the predictions. Boxplots illustrate lowland (open bars) vs non-lowland (mid-elevation and pre-montane) sites (closed bars). * P<0.05, ** P <0.01).
Table 1 . Geographic and temperature data for the sample sites. TMin is the mean minimum temperature of the coolest month, MAT is mean annual temperature and TMax is the mean maximum temperature of the warmest month.