A comparison of random draw and locally neutral models for the avifauna of an English woodland

BMC Ecology

Table 1 Species composition predictions.

	Random draw model	Neutral model
Mean (± SD) number of species correctly predicted as present	42.1 ± 1.12	33.9 ± 1.69
Mean (± SD) number of species incorrectly predicted as absent	2.87 ± 1.12	11.08 ± 1.69
Mean (± SD) number of species incorrectly predicted as present	17.5 ± 1.29	10.6 ± 1.41
Mean (± SD) number of correct predictions (absences and presences)	60.7 ± 1.68	59.3 ± 2.21

The performance of the random draw and neutral models with regards to predicting the species composition of Eastern Wood. Note that care is needed when interpreting these statistics, as they simulate assemblages of differing mean species richness. Comparisons based on the number of species correctly or incorrectly selected by a model are biased towards models that generate assemblages with higher and lower species richness, respectively. The last comparison is unbiased with respect to simulated species richness.

ISSN: 1472-6785