Demographic response to transplantation explained by traits
Six species decreased, and five species increased in cover in the warmer sites compared to the origin. Five decreased in frequency but increased or showed no change in cover (Table S1, Figures S8-S9). Of the full trait space, the first four axes explained 63% of the variation (21%, 15%, 15% and 12%). The first axis separated species based on their water use efficiency measured with the gas-analyser (WUEinst, WUEint, E). The second axis separated species with high values of δ13C and δ18O enrichment (i.e. “conservative” water use strategy) and large size (width, leaf area) from small species with “prolific” water use strategy. The third axis separated species which increased their WUEinst after transplanting (high WUEinst.PI), had a high sensitivity of respiration to temperature (high Q10) and low leaf nitrogen content from species with opposite traits. The fourth axis separated species with prostate growth form (high CSI), high stomatal conductance, high nighttime transpiration and low temperature optima for photosynthesis from tall, erect and deep-rooted species with opposite physiological traits (Figure S10).
A species’ position on the second trait axis correlated with its demographic response to increasing temperature both when measured as cover (χ2(1) = 5.69, p = 0.017, Figure S11a) and frequency (χ2(1) = 14.5, p < 0.001; interaction axis2 × temperature, Figure 1a, Tables S4-S5). Small species in terms of width and leaf area with high life-time stomatal conductance (low δ18O) and low water use efficiency (low δ13C; “prolific” water use strategy) declined in cover and frequency at the warmer relative to cooler sites. Additionally, a species’ position on the first PCA axis correlated with its frequency change: species with high WUEinst and WUEint and low E reduced in frequency more than species with opposite traits at all sites (χ2(1) = 5.55, p = 0.019; Figure S11b). A species’ position on the two other trait axes did not correlate with their demographic response to warming (Table S5). The best LMMs derived from model selection (including initial cover or frequency, temperature and its interactions with traits, and species identity) explained 59% of the variation in cover and 69% in frequency, of which species identity explained 25% and 12%, and traits explained 10% and 12% (Table S3).
At the community-level, turfs transplanted to lower elevations were more likely to contain larger species (height and width; (F(1,48) = 4.88, p = 0.03; F(1,48) = 15.08, p < 0.001) with a conservative water use strategy (high δ13C and δ18O; F(1,48) = 68.57, p < 0.001; F(1,48) = 4.54, p = 0.04; Figure 3, Figure S12bc and Table S6).
Several traits (WUEins, gsnight, Enight, δ18O, δ13C) showed plasticity, and the direction of the change differed between species (Figures S4-S7, S13). Notably, photosynthesis and respiration were on average similar when measured at +0°C and +3°C (i.e. there was no clear evidence of acclimation, Figure S4a,f). None of the plasticity indices nor the parameters describing species’ ability to rapidly adjust to temperature (Topt, Ω, Q10) were important predictors of the demographic changes. Even though transplantation to lower elevations favored species with conservative water use, it resulted in higher life-time stomatal conductance and lower water use efficiency (low δ18O enrichment and δ13C) within species (Figures S5-S6).