Understanding how local climate patterns select for thermal and drought tolerance traits is needed to predict differential responses to climate change across complex ecosystems. Here, we used high throughput methods to measure traits that confer heat and drought tolerance across a tropical climatic variability gradient, and examined how forest type and leaf habit mediate these traits. Using standardized methods, we estimated thermotolerance thresholds (Tcrit, T50, T95), drought tolerance traits (water potential at turgor loss point [𝝭TLP] and osmotic potential [𝞹osm]), and morphological traits (wood density and leaf mass per area [LMA]) for 92 woody plant species across six sites along a tropical dry-to-wet gradient in Área de Conservación Guanacaste, Costa Rica. Among evergreen species, T95 showed a weak decline with elevation, LMA was positively associated with both T50 and T95, and drought tolerance increased at lower elevations and with greater wood density. T95 was higher on average in species from seasonal dry and ecotonal forests compared to wet forests. Evergreen species from dry and ecotonal forests also exhibited greater drought tolerance than evergreen species from wet forests, whereas drought tolerance did not differ among deciduous species. Across all species, thermal and drought tolerance traits were largely decoupled, indicating that coordination among stress‐tolerance traits is not generalizable across tropical forest communities.

Thermal indices

We used standardized methods to quantify thermal tolerances of plants. Individual sunlit branches were collected from the field in the morning on the day of measurement. Discs of healthy, mature leaves were collected from each individual. Following established protocols, leaf discs were exposed to eleven sous-vide water baths at different temperatures (room temperature (~23), 24, 28, 32, 36, 40, 44, 48, 52, 56, 60 °C) where they remained for 15 minutes under dim light. Following treatment, the discs were removed from the water baths and placed into humid petri dishes to recover for 24 hours, then dark-adapted for 20 minutes. Fv/Fm was measured using an LI-600 fluorometer (LI-COR Biosciences). Following Perez & Feeley (2020), we modeled the relationship between Fv/Fm and heat treatment temperature using the nls function from the stats package in R 4.4.2 (R Core Team 2024). From this relationship, we estimated three key thermal tolerance metrics: Tcrit, T50, and T95. Tcrit was defined as the temperature at which the slope of the Fv/Fm-versus-temperature curve reached 15% of its most negative value, indicating the onset of rapid PSII damage. T50 and T95 were calculated as the temperatures at which Fv/Fm had declined by 50% and 95%, respectively, relative to the control treatment.

Drought indices

We also used established, high-throughput methods to quantify osmotic potential and turgor loss point. Upon collection, branches of each individual were re-cut underwater in buckets and rehydrated overnight. The following morning, osmometer measurements were performed following Bartlett et al. (2012). Circular discs were collected from leaf samples and frozen in liquid nitrogen, and punctured with tweezers to facilitate equilibration before being placed into the vapor-pressure osmometer (VAPRO 5600). The output measurement of the vapor pressure osmometer represents the osmotic potential of the leaf symplast at full hydration. In addition to osmometer measurements, we also measured leaf thickness and density for each sample. Leaf thickness was calculated as the average of digital caliper measurements taken at the top, middle, and bottom of the leaf. Density was calculated as dry mass divided by area. Together, these values were used to estimate 𝝭TLP and 𝞹osm for each sample based on a global calibration dataset.

Wood density

Wood density for each individual was measured using the water displacement method (Perez-Harguindeguy et al., 2013). We cut wood cores from rehydrated branches with a minimum diameter of one centimeter on the day after collection, before submerging them in a graduated cylinder to measure volume based on water displacement. We then measured mass after drying the cores at 70 °C for 72 hours (Muller-Landau, 2004). Density was calculated as dry mass divided by volume.

Leaf mass per area (LMA)

To measure LMA, we cut 1.13 cm2 disks, avoiding midveins, out of the same rehydrated leaves that were used to measure drought tolerance via the vapor pressure osmometer. As such, we followed the same protocol: we used two leaves from each individual, except when only one individual was available, in which case we took four leaves to create an equal number of samples for each species, then used mean values for each species in the statistical analysis. We measured the mass of the discs after drying them at 70 °C for 72 hours. LMA was calculated as dry mass divided by area (cm2 g -1).