ORIGINAL ARTICLE Analysing spatial and temporal variation in colony size: an approach using autoregressive mixed models and information theory Pedro P. Olea Received: 26 December 2007 / Accepted: 10 July 2008 / Published online: 29 August 2008 Ó The Society of Population Ecology and Springer 2008 Abstract The spatial and temporal variation in popula- tion sizes of animal colonies are rarely studied simultaneously. I examined factors determining colony size (number of nests) for 23 colonies from the only breeding population of rook Corvus frugilegus in Spain over 7 years. Population sizes within colonies were highly predictable over time, with autocorrelations up to a distance (lag) of 6 years. Autoregressive mixed models were used to explain colony size as a function of environmental factors, while controlling for temporal autocorrelation. These factors included refuse tips, widely used as food resource, and a derived variable that incorporated the two factors most often related to avian colony size (inter-colony competition and foraging habitat around colonies). Autoregressive models provided a better fit to the data than models which did not consider temporal autocorrelation. The informa- tion-theoretic (AIC c -based) approach revealed uncertainty in the selection of the best model explaining colony-size, but relatively strong support for certain variables. The highest weights of evidence were for year (x i = 0.90) and the number of competitors per unit of foraging habitat (i.e., derived variable; x i = 0.63), showing that the size of rook colonies in Spain was negatively affected by inter-colony competition relative to the foraging habitat surrounding the colonies. This variable measured within a 6-km radius from the colonies had *30 times higher weight of evidence (more plausible) than the same variable measured within 3 km, indicating that food limitation may occur outside the breeding period. Sizes of colonies tended to decrease when distance between the colony and the nearest refuse tip increased. There was some evidence supporting the idea that the effect of the number of competitors per unit of foraging habitat on colony size varied from year to year, but statistical power was weak. These findings suggest that variation in number of rook nests within colonies reflects spatial and temporal heterogeneity of net food via both inter-colony competition and foraging habitat around the colony. Keywords AIC Á Derived variable Á Corvus frugilegus Á Rook Á Glmm Á Temporal autocorrelation Introduction The greatest challenge today in science is accurate and complete description of complex systems (Wilson 1998 in Green et al. 2005). Ecological systems show complex patterns both in time and space; description of these pat- terns and identification of the mechanisms that shape them are the core of ecology (Bascompte and Sole ´ 1995). One of the most intriguing patterns observed in nature, but yet poorly understood, is that of breeding colonies of birds widely varying in population size within and among spe- cies (Brown and Brown 2001, 2002). Multiple mechanisms may explain the highly patchy distribution of colony sizes. Heterogeneity in resources (e.g., food, nest sites) can account for some of this variation (Brown and Brown 2001), while social factors (e.g., conspecfic attraction; Safran 2004; Serrano et al. 2004) could act as proxies used Electronic supplementary material The online version of this article (doi:10.1007/s10144-008-0106-4) contains supplementary material, which is available to authorized users. P. P. Olea (&) Department of Exact Sciences and Environmental Biology, Faculty of Experimental Sciences, IE University, Campus Santa Cruz la Real, 40003 Segovia, Spain e-mail: pedro.perez@ie.edu 123 Popul Ecol (2009) 51:161–174 DOI 10.1007/s10144-008-0106-4