Plant ecological strategies are shaped by numerous functional traits and their tradeoffs. Trait network analysis enables testing hypotheses for the shifting of trait correlation architecture across communities differing in climate and productivity. We built plant trait networks (PTNs) for 118 species within six communities across an aridity gradient, from forest to semi-desert across the California Floristic Province based on 34 leaf and wood functional traits, representing hydraulic and photosynthetic function, structure, economics, and size. We developed hypotheses for the association of PTN parameters with climate and ecosystem properties, based on theory for the adaptation of species to low resource/stressful environments, versus higher resource availability with greater potential niche differentiation. Thus, we hypothesized that across community PTNs, trait network connectivity (i.e., the degree traits are intercorrelated) and network complexity (i.e., the number of trait modules, and the degree of trait integration among modules) would be lower for communities adapted to arid climates, and higher for communities adapted to greater water availability, similarly to trends expected for phylogenetic diversity, functional richness and productivity. Further, within given PTNs, we hypothesized that traits would vary strongly in their network connectivity and that the traits most centrally connected within PTNs would be those with the least across-species variation. Across communities from more arid to wetter climates, PTN architecture varied from less to more interconnected and complex, in association with functional richness, but PTN architecture was independent of phylogenetic diversity and ecosystem productivity. Within community PTNs, traits with lower species variation were more interconnected.  

Synthesis.  The responsiveness of PTN architecture to climate highlights how a wide range of traits contribute to physiological and ecological strategies with an architecture that varies among plant communities. Communities in more arid environments show a lower degree of phenotypic integration, consistent with lesser niche differentiation. Our study extends the usefulness of PTNs as an approach to quantify tradeoffs among multiple traits, providing connectivity and complexity parameters as tools that clarify plant environmental adaptation and trait associations that would influence species distributions, community assembly and ecosystem resilience in response to climate change.

We sampled three to five individuals per species, resulting in a total of 683 individuals from 118 unique species in six plant communities distributed across a gradient of climatic aridity, differing by 10 ^oC in mean annual temperature (MAT) and 6-fold in mean annual precipitation (MAP). The sites also varied strongly in soil characteristics and plant community composition. We targeted the most abundant tree and shrub species at each site, based on information from reserve managers and forest inventories. The sampling spanned 37 families, with greatest species representation in Asteraceae (17), Rosaceae (12), Rhamnaceae (9), Ericaceae (8) and Pinaceae (8). At each site we sampled from 19 to 28 species, collecting a mature, sun-exposed and non-epicormic branch, with no signs of damage and herbivory using pole pruners or a slingshot. Branches were transported to the lab in dark plastic bags with moist paper and rehydrated overnight in a dark saturated atmosphere before harvesting current-year grown, fully expanded leaves for subsequent analyses. For compound-leafed species, whole leaves, not leaflets, were used.