ORIGINAL ARTICLE Testing Bergmann’s rule in the presence of potentially confounding factors: a case study with three species of Galerida larks in Morocco Alban Guillaumet 1,2 *, Jean-Baptiste Ferdy 3 , Eric Desmarais 1 , Bernard Godelle 3 and Pierre-Andre ´ Crochet 4 INTRODUCTION In general, species are not homogeneous throughout their range but show instead more or less well-marked geographical variation in, for example, morphology, coloration or life- history traits. Among these, body size is one of the most commonly measured attributes in animal species (Laiolo & Rolando, 2001), and has been the subject of many studies of intraspecific geographical variation (Ashton, 2001). While the main forces responsible for population divergence have been identified (see below), their respective magnitude in natural populations is still poorly understood (Merila ¨ & Crnokrak, 2001; Marroig & Cheverud, 2004). First, part of the geographical variation in any given trait might result from 1 Institut des Sciences de l’Evolution, C.C. 63, Universite´de Montpellier II, Place E. Bataillon, 34095 Montpellier Cedex 5, France, 2 EPHE, CEFE CNRS, UMR 5175, 1919 Route de Mende, 34293 Montpellier Cedex 5, France, 3 Institut des Sciences de l’Evolution, C.C. 65, Universite´de Montpellier II, Place E. Bataillon, 34095 Montpellier Cedex 5, France, 4 CEFE CNRS, UMR 5175, 1919 Route de Mende, 34293 Montpellier Cedex 5, France *Correspondence: A. Guillaumet, EPHE, CEFE CNRS, UMR 5175, 1919 Route de Mende, 34293 Montpellier Cedex 5, France. E-mail: alban.guillaumet@cefe.cnrs.fr ABSTRACT Aim To test Bergmann’s rule (which predicts a larger body size in colder areas within warm-blooded vertebrate species) in three partially sympatric species of larks (Galerida theklae, Galerida cristata and Galerida randonii) that occur in Morocco. Location Morocco. Methods Restriction fragment length polymorphism techniques applied on cytochrome b haplotypes were used to discriminate G. cristata and G. randonii, and to investigate the effects of interspecific hybridization in their contact zone. A comprehensive statistical framework was then designed to test Bergmann’s rule in our three Galerida species (using altitude as a proxy for cold temperatures), while controlling for the possible influence of interspecific hybridization and competition and accounting for spatial autocorrelation. The method we propose is conservative in the sense that potentially confounding factors are adjusted so as to maximize their influence on the variable of interest. Results Bergmann’s rule was strongly supported in G. theklae and G. randonii. However, body size did not respond to altitude in G. cristata, a result that was not simply explained by species-specific differences in geographical ranges and altitudinal span. In G. cristata, we detected a tendency for body size to increase with aridity, in agreement with an alternative definition of Bergmann’s rule. However, since G. cristata also hybridizes with G. randonii in a contact zone located in the most arid part of the range of G. cristata, we could not tease apart the relative contribution of selection and hybridization in driving this pattern. Main conclusions This study highlights the need for careful statistical designs that allow meaningful variables to be picked out from large sets of potential factors. When taking these factors into account, we found that Bergmann’s rule was still strongly supported in two out of the three species examined. Keywords Bergmann’s rule, body size, competition, confounding factors, Galerida, hybridization, Morocco, morphology, spatial autocorrelation. Journal of Biogeography (J. Biogeogr.) (2008) 35, 579–591 ª 2007 The Authors www.blackwellpublishing.com/jbi 579 Journal compilation ª 2007 Blackwell Publishing Ltd doi:10.1111/j.1365-2699.2007.01826.x