Hierarchical constraints in water availability shapes the functional structure of lowland tropical forest tree communities at regional and landscape scales in New Caledonia

Understanding the factors driving species distribution across spatial scales in highly diverse forests is challenging and critical for biodiversity conservation planning. Functional ecology provides powerful tools to assess the effect of environment on species assemblages and to better understand the role of niche-based processes. Here, we explored how a key environmental factor, water availability, drives species distribution at both regional and landscape scales in lowland species-rich forests in the biodiversity hotspot of New Caledonia.

We studied two landscapes located on the dry west-coast (~1000 mm.yr-1) and the wet east-coast (~3000 mm.yr-1) of the New Caledonian main island. Within each landscape, 20 tree communities were sampled (in 0.04-ha plots) along a topographical gradient that correlates with water availability (Topographic Wetness Index, TWI). We sampled and measured five functional traits (wood density, leaf area, leaf specific area, leaf dry matter content, and leaf thickness) involved in drought resistance and resource acquisition.

First, we examined trait covariation and identified functional trade-offs across species. Second, we assessed the functional composition of local communities using four trait-based statistics (community weighted-mean, community unweighted mean, trait range, and community weighted variance). We tested the deviation of these statistics from null models of random species distribution designed either at regional or landscape scale.

We identified two main ecological trade-offs driving trait variation across tree species: (i) one opposing high hydraulic efficiency to drought resistance and related to a wood economic spectrum and (ii) the other opposing resource acquisition to resource conservation and related to a leaf economic spectrum. At regional scale, species with drought-resistance strategies (i.e. high wood density and small leaves) were favored under low precipitation while strategies were more divergent under higher precipitation. At landscape scale, we found contrasted responses of tree communities. In the dry landscape, drought-resistant species were selected under low TWI, while in the wet landscape, low TWI increased the abundance of species with conservative strategies (i.e. low leaf specific area and high leaf thickness).

Our study showed that water availability drives the variation of tree community composition at both landscape and regional scales. In the low precipitation landscape, environmental filtering on the wood economic spectrum due to hydric constraints was the main driver of community assembly. However, in the high precipitation landscape, water stress was weaker and the selection of species across topographic heterogeneity rather rely on their resource-use strategies along the leaf economic spectrum.