Print PageA common consequence of recent climate changes are range shifts of animals and plants. Many studies show poleward shifts of ranges which are attributed to global warming. However, many of these studies fail to test these hypotheses in a statistically rigorous fashion, and they treat edges of distributions arbitrarily. This fails to account for the fact that our observations are merely a sample of species’ true distributions. In this work we aimed at formally modeling the edges of a spatial distribution along a north-south axis. We further developed a statistical test to examine if indeed a shift in the spatial distribution has occurred between two time frames.
The model estimates a spatial distribution based on presence/absence grid-cell data. a flexible parametric model for abundance over a bounded range is used to estimate cell occupancy. The distribution’s edge parameters were estimated using an optimization procedure. Likelihood ratios were used to conduct a statistical test between the null (no shift in range) and alternative (possible shift in range) hypotheses. We first tested the model on simulated data and then applied it to data on British bird distributions in 10*10 km grid cells between two time frames (1968-72 versus 1988-91).
Results/Conclusions
The likelihood ratio of the simulated data was found to have a chi-square distribution (d.f. =1), which enabled us to obtain significance values for our empirical tests. We analyzed distributions of 64 bird species that have either a northern or a southern distribution boundary within Britain. Of these 52 did not show a significant change in their edges; 5 species significantly shifted their northern boundary northwards and 5 southwards; one species significantly shifted its southern boundary northwards and one southwards. We therefore conclude that our method does not support the notion of climate induced directional range shifts of British birds between these two time frames.
The methods presented here enable both good estimation of true distribution boundaries and reliably testing shifts in edges of a distribution. This work emphasizes the importance of carefully formulated statistical models when estimating distributions of organisms in space, as for example, is needed in testing the effects of climate change. Effects of global climate change on species ranges can be more complex than previously recognized. Extreme temperatures as well as synergistic biotic and abiotic effects can affect species ranges differently from what would be expected by mean annual temperature.
