Since neutral theory’s debut in the form of Hubbell’s book, various tests of this theory of biodiversity maintenance have been carried out. I review these and conclude that few are particularly conclusive. Instances of agreement with neutral theory involve parameter freedom that could result in agreement with a niche model also, and carrying out model selection instead is difficult. Rejections of neutral theory often involve simplifications that could be causing the model failure rather than the presence of niche differentiation or habitat filtering. Here I describe an approach to improving the use of neutral theory as a process-based null model, and present a key question of scale that arises under this approach. I propose a simulation methodology to answering this question of scale, and present preliminary results of a set of such simulations constructed with the data availability for the tree community on Barro Colorado Island in Panama in mind. I also highlight questions of scale and approaches to answering them.
Results/Conclusions
A key idea I highlight for improving neutral theory tests is to use data to inform the regional pool from which species immigrate in the neutral model, to reduce parameter freedom. The question of scale that then arises is what spatial and temporal scale of abundance data is required to inform the state of the regional pool. This question can be answered by comparing neutral model predictions generated from a large scale fully dynamic neutral model to those generated from a locally dynamic neutral model informed by regional pool data collected at various spatial and temporal scales from the full dynamic model. Preliminary results of such a comparison suggest that the data currently available to inform the regional pool for the Barro Colorado Island 50 ha plot is lacking in the spatial resolution adequate for this purpose.