Our simulation study explored various methodologies for studying species–habitat associations through spatial point pattern analyses. The results indicate that all methods performed equally well, and their performance was mostly influenced by the initial point pattern characteristics. Therefore, we suggest applying the method that is most suitable for the available data.

Abstract

Question

Species-specific habitat associations are one of several processes that lead to a clustered spatial pattern of plant populations. This pattern occurs in tropical and temperate forests. To analyze species–habitat associations, four methods are commonly used when determining species–habitat associations from spatial point pattern and environmental raster data. Two of the methods randomize the spatial point pattern of plants, and two randomize the raster data of habitat patches. However, the strengths and weaknesses of the four methods have never been analyzed in detail.

Methods

We conducted a simulation study to analyze the strengths and weaknesses of the four most used methods. The methods are the gamma test, pattern reconstruction, the torus-translation test and the randomized-habitats procedure. We simulated neutral landscapes representing habitat patches and point patterns representing fine-scale plant distributions. We built into our simulations known positive and negative species–habitat associations.

Results

All four methods were equally good at detecting species–habitat associations. Detected positive associations better than negative ones. Furthermore, correct detections were mostly influenced by the initial spatial distribution of the point patterns, landscape fragmentation and the number of simulated null model randomizations.

Conclusions

The four methods have advantages and disadvantages, and which is the most suitable method largely depends on the characteristics of the available data. However, our simulation study shows that the results are consistent between methods.