Spatial pattern reveals process in a species‐rich seasonal tropical forest: Insights from a multi‐type point process model

The spatial arrangement of individuals of multiple species of sessile organisms is thought to be driven by a variety of entangled biotic and abiotic mechanisms. In particular, segregation along environmental gradients causes long‐distance clustering within species and separation between species. Dispersal limitations result in intermediate‐scale clustering, solely within species. Physical packing, resource competition and natural enemies as agents of negative density dependence cause regularity at finer scales, with the latter only within species. Functional groups representing structure and dynamics have distinct associations, with gap and understorey species more clustered than canopy. We hypothesise the combined effect of these mechanisms on a forest spatial pattern. We test our hypotheses using the saturated pairwise interaction Gibbs point process model, fitted to individually mapped stems in a 50‐hectare plot located in species‐rich, seasonal evergreen forest in the Huai Kha Khaeng Wildlife Sanctuary in western Thailand. This approach allows for modelling species' distance‐dependent, signed pairwise interactions in addition to environmental responses. We found that many of the interactions were species‐agnostic, yet functional group‐dependent. In general, and for canopy trees particularly, interactions over short distances were negative, within and between species, while those over longer distances were positive. Understorey trees were overwhelmingly clustered, especially at fine scales, and showed positive short‐distance associations with other species. Gap species were distinct in displaying positive intraspecific interactions and generally negative interspecific interactions, irrespective of distance. Synthesis . The pairwise interaction Gibbs process provides a way to quantify the relative importance of the underlying ecological drivers via a static snapshot of plants' spatial pattern. Our approach may enhance our understanding of mixed‐species forests around the world and may also be applied in other mapped plots with over about 100 plants.