Ecological Biogeography of Southern Ocean Islands: The Importance of Considering Spatial Issues

Abstract

Understanding patterns of among-island variation in species richness has long been an important question in ecology and biogeography. However, despite the clear spatial nature of the data used for such investigations, the spatial distribution of the different sampled locations is rarely explicitly considered, which may be critical for statistical and biological reasons. In a recent study, Chown et al. (1998) investigated the relationships between species richness of different indigenous and introduced taxonomic groups and a variety of variables characterizing Southern Ocean islands, and here, we use these data to address spatial issues. As predicted, we found spatial autocorrelation in species richness for terrestrial taxa with high dispersal ability or for terrestrial taxa that had time to disperse locally (introduced land birds and indigenous taxa) but not for taxa that had low opportunity to disperse to nearby islands (introduced plants, insects, and mammals), which suggests that colonization from nearby islands has played an important role in shaping present-day patterns of among-island variation in species richness. Interestingly, in several cases, the estimated effect of variables changed when spatial covariance was incorporated. Moreover, the absence of autocorrelation of some variables allowed us to confirm some important results of Chown et al. (1998), notably those involving the potential impact of human presence on the biodiversity of these islands. Overall, our results illustrate the importance of considering spatial structures in ecological studies. This is notably the case when dispersal processes can be expected to explain some of the observed patterns.

Red herrings revisited: spatial autocorrelation and parameter estimation in geographical ecology

“…First and foremost, it remains the case that OLS and spatially explicit regression models of the same data sets sometimes differ, as we observed in our data. Ecologists often interpret this as evidence of the bias generated by spatial autocorrelation and then claim that the OLS results are not dependable (Lennon 2000, Selmi and Boulinier 2001, Tognelli and Kelt 2004, Gimona and Brewer 2006, Dormann 2007, Kühn 2007). However, spatial autocorrelation is not the source of the problem.…”

“…However, in addition to this almost universally recognized issue of false precision, it is also widely believed that spatial autocorrelation creates a shift in the relative importance of coefficients in simple and multiple regression models (Lennon 2000). This sometimes causes workers to abandon the results of standard ordinary least squares (OLS) regression and to interpret instead coefficients generated using one of several spatially explicit modeling procedures (Lennon 2000, Selmi and Boulinier 2001, Tognelli and Kelt 2004, Bahn et al 2006, Dormann 2007, Kühn 2007. Because regression models using OLS, generalized least squares (GLS), or spatial autoregression (simultaneous or conditional autoregressive models; SAR or CAR) may sometimes differ from each other, it thus becomes important to know to what extent these differences are the consequence of RSA or may be due to other structural differences among the modeling approaches that arise independently of spatial autocorrelation.…”

Evaluating the Relationship between White-Tailed Deer and Environmental Conditions using Spatially Autocorrelated Data in Tropical Dry Forests of Central Mexico

“…On the other hand, the results of the SARerr and GLS methods do not differ greatly since both control spatial autocorrelation of the model residuals and are mathematically similar [40,41]. Nevertheless, consistently lower AICc values were found in GLS, which is possibly due to the fact that GLS is more flexible in the form in which SAC is incorporated into the models [8].…”

Diversity of microfungi in ornithogenic soils from Beaufort Island, continental Antarctic

“…Antarctica's terrestrial ecosystems are fragile and easily affected by human or animal activities (Chown et Gaston 2000, Selmi et Boulinier 2001, Frenot et al 2005. Seabirds and marine mammals, such as penguins and seals, feed in the surrounding ocean and reproduce and rest on land.…”