Within-species trait variation and competition across environmental gradients in Puerto Rican forests

Background/Question/Methods: Within-species trait variation is a recently debated mechanism for promoting plant coexistence. The hypothesis is that such variation allows individuals to avoid direct competition by having slightly different trait values than their neighbors while better matching their environments. Indeed, there is significant evidence that there is a large degree of intraspecific variation in plant characteristics or traits. Nevertheless, trait variation is often considered negligible in predictive models of plant trait distribution across environment. We use data for intra-specific variation in wood density (WD), specific leaf area (SLA), and stomatal density (SD) in 304 species across 124 forest plots in Puerto Rico. Species abundance and basal area are measured in plots along precipitation (10-3170 mm/yr), soil (limestone and volcanic), and elevation gradients (5-985 meters). We construct three Bayesian joint models of species occurrence and basal area as predicted by different environmental factors. The different modeling frameworks answer how these traits are distributed across environmental gradients. In doing so, they address how incorporating intraspecific trait variation into predictive models of species occurrence changes our species covariance matrix (i.e., the likelihood that species will coexist), after accounting for environmental constraints.

Results/Conclusions: Preliminary results show that the mean traits are correlated with environmental variables. Wood density positively responds to precipitation (β=0.031 ± 0.014) and negatively to elevation (β=-0.022±0.008). Stomatal density responds negatively to elevation (β=-0.006 ±0.002) and increases on volcanic soils (β=15.610 ± 7.691). Otherwise, these traits did not respond significantly to environmental covariates. We also see significant within-species variation in these traits. Intraspecific variation in wood density is only an order of magnitude higher than interspecific variation (0.081 versus 0.153 g cm-3), though there is much more inter- than intraspecific variation for specific leaf area (27.4 m2.kg−1 versus 14559.79 m2.kg−1). Future research on trait response to the environment should incorporate this within-species variation into their predictive models. Our work provides modeling frameworks to incorporate this variation rigorously in predictive models.