Does environmental variability drive plant functional trait variability?

Background/Question/Methods

During community assembly, plant traits are hypothesized to be filtered by prevailing environmental conditions.  A number of studies have demonstrated that trait variation within species can be substantial, and that this variation can be driven by environmental gradients.  If trait variation is primarily a response to prevailing environmental conditions, species which express large variation in plant functional traits should be present in diverse environments. Using environmental and soil nutrient data collected on 138 plots spanning a riparian wetland-to-dry upland ecotone and plant traits (SLA, height, LDMC, leaf C:N and leaf area) for 49 species, we sought to link plant trait variability and environmental variability.  We hypothesized that the most variable species would be present in plots with a high range of environmental variablity, while species with low trait variability would be present in environmentally similar plots.  Average environmental dissimilarity was calculated for each species.  First, multivariate habitat similarity was quantified by extracting the fusion distances between pairs of plots generated by hierarchical cluster analysis, then averaging by the number of plot pairs where each species was present.  For each species, univariate trait variability was quantified with coefficients of variation, while multivariate trait-variability was quantified both with convex-hull volumes and average distance-to-centroid.  Correlations coefficients between environmental dissimilarity, species presence, and trait variability were then tested.

Results/Conclusions

There was considerable environmental variability between measured plots, with fusion distances ranging from 0.0004 to 4.2.  Environmental dissimilarity for the 49 species ranged from 0.68 to 0.85.  Traits also varied considerably within species.  CVs ranged from 0.01 to 0.96, while convex hull volumes ranged from 15.4-189.9, and distance-to-centroid ranged from 1.27 to 1.58.  Contrary to expectations, for each species, trait variability for both univariate and multivariate measures were poorly correlated with the average environmental dissimilarity.  Only variation in plant height and LDMC showed any correlation with environmental distance (correlation coefficients = 0.11 and 0.12, respectively) but this correlation was not statistically significant.  However, variance in SLA and C:N ratio increased significantly with increasing species presence (correlation coefficients 0.30 and 0.29, p=0.05).  These results indicate that observed variation in plant functional traits is not primarily driven by habitat heterogeneity and prevailing abiotic conditions.  Rather, plant trait variation may be driven by genetic variation or plasticity in response to microsite conditions and biotic interactions.