A spatial birth–death process to model sessile organisms

Spatial patterns of sessile organisms, such as plants, reflect past processes and affect future changes. We introduce a spatial birth–death process with birth/death rates that explicitly depend on distances to other points. These demographic rates are defined to represent the combined effect of the ecological mechanisms that affect individual organisms. We show that the constructed spatio-temporal process converges exponentially fast to a saturated pairwise interaction Gibbs point process. This allows for statistical inference through an analysis of either the locations where births/deaths occur, or (assuming stationarity) of the sequence of recorded spatial patterns. A simulation study demonstrates that either of these inference techniques recovers the parameters of the spatial birth–death process with relatively few temporal observations. We then applied our model to 20 forest plots in northern Queensland, Australia, which were censused a dozen times. We find that births are more likely around conspecifics, and that this may explain the species’ clustered spatial patterns. This work unifies two distinct streams of research and provides a route to inference and prediction of evolving spatial structures in communities from ecological mechanisms that influence demographic rates. • We present a spatial birth–death process model, and derive it’s properties. • Inferences can be made about the spatiotemporal distributions of births and deaths. • Inferences can use (a) snapshot pattern, and/or (b) censuses of births and deaths. • We examine the behaviour of the model with simulated data. • In a case study of rainforest trees, we found births clustered around other trees.