Apidologie 39 (2008) 119–132 Available online at: c INRA/DIB-AGIB/ EDP Sciences, 2008 www.apidologie.org DOI: 10.1051/apido:2007044 Original article Correlated evolution of mating behaviour and morphology in large carpenter bees (Xylocopa)* Remko Leys1,2, Katja Hogendoorn3 1 Evolutionary Biology Unit South, Australian Museum, North Terrace SA 5000, Adelaide, Australia 2 School of Earth and Environmental Sciences, University of Adelaide, Australia 3 School of Agriculture, Food and Wine, The University of Adelaide, Adelaide SA 5005, Australia Received 12 June 2007 – Revised 25 September 2007 – Accepted 3 October 2007 Abstract – Carpenter bees (Xylocopa) display variation in mating strategies. In several subgenera males defend territories that contain resources for females. In other subgenera males defend a small non-resource territory. Here, we investigate the correlation between three morphological traits and mating strategy. We found associations between mating strategy and male eye size, size of the mesosomal gland and sexual colour dimorphism, as well as correlative evolution between the morphological characters. Analysis of the evolutionary pathways shows that resource defence, small glands and monomorphic sexes are ancestral states. Increases in gland size seem to precede or coincide with changes in mating behaviour, but changes towards sexual dimorphism follow changes in mating behaviour. Once a non-resource defence strategy with correlated morphology has evolved there are no reversals to the ancestral states. We discuss the types of selection that may have caused these correlative changes. correlated evolution / molecular phylogeny / mating / sexual selection / Xylocopa 1. INTRODUCTION comparative analysis of related species with different mating strategies are needed to elu- The elaboration of male traits, such as sex- cidate patterns of causal variables (e.g. Brown ual dimorphism in size or colour, secondary et al., 1997; Markow, 2002; Kruger et al., sexual characters and ornamentation is mostly 2007). considered to reflect the action of sexual se- In this study, we use phylogenetic meth- lection (Darwin, 1871). Sexual selection com- ods to evaluate the correlation between mat- prises intra-sexual selection through male- ing strategies and sexually selected traits in the male competition and inter-sexual selection group of large carpenter bees of the genus Xy- through female choice. In addition, but of- locopa (Hymenoptera, Apidae). The genus is ten forgotten, sexual dimorphism can evolve eminently suitable for analyses of such cor- through natural selection (e.g. Temeles et al., related evolution because it is very speciose 2000; Kruger, 2005; Stuart-Fox and Moussalli, (approximately 470 species divided into 31 2007). However, because there are strong re- subgenera) and because mating strategies vary ciprocal causal links between mating sys- considerably between species (e.g. Gerling tems and sexual selection (Andersson, 1994), et al., 1989; Michener, 1990; Minckley, 1994). Corresponding author: R. Leys, Mating behaviour has been described for [email protected] 38 species belonging to 16 different subgen- * Manuscript editor: Eduardo A.B. Almeida era (Tab. I supporting information). A well- Online material is available at: resolved molecular phylogeny that includes http://www.apidologie.org 27 subgenera has recently become available Article published by EDP Sciences and available at http://www.apidologie.org or http://dx.doi.org/10.1051/apido:2007044 120 R. Leys, K. Hogendoorn (Leys et al., 2002). This allows testing of males give chase to females and conspecific phylogenetic correlations between mating be- males (Frankie et al., 1979; Minckley, 1994). haviour and morphological traits. Therefore large eye size of males may be as- Using the terminology developed by sociated with resource defence behaviour, as Alcock et al. (1978) and Eickwort and was suggested by Osten (1989) and Minckley Ginsberg (1980), the different strategies that (1994). Minckley (1994) also suggested an as- males of carpenter bees use to encounter sociation between enlarged glands and rela- receptive females are: (1) non-territorial tively small eyes. patrolling of nests or flowers, (‘non-defence In this paper we apply phylogenetic meth- polygyny or patrolling’ - P); (2) territorial ods to investigate the associations between defence of resources (flowers, nests) or of mating behaviour and the morphological char- females inside nests (‘resource defence polyg- acters mentioned in the previous paragraph. yny - RD’); and (3) territorial behaviour at Specifically, we test whether there has been locations that do not contain any resources for correlated evolution in Xylocopa between mat- females (‘non-resource defence polygyny’- ing strategy and (a) size of mesosomal glands NRD), a strategy that has also been called in males, (b) sexual colour dimorphism and a ‘dispersed lek’ system (Bradbury, 1981) (c) relative male eye size. We hypothesize as- because females are able to assess the quality sociations between non-resource defence and of males by comparing displaying males in enlarged glands, presence of colour dimor- dispersed territories. phism and small eyes, and vice versa, be- Several authors have suggested associations tween resource defence, small glands, absence between mating strategies and morphological of colour dimorphism and enlarged eyes. Fur- characters. The main characters are: (i) Size thermore, we reconstruct the ancestral char- of the mesosomal glands. The exocrine prod- acter states for the genus and investigate the ucts from the mesosomal glands are used as a evolutionary pathways from the reconstructed long range sex-attractant pheromone (Gerling ancestral states to the derived character states, et al., 1989; Minckley et al., 1991; Alcock and including possible reversals, of all pairs of Johnson, 1990). Enlarged mesosomal glands characters. are often encountered in species displaying non-resource defence behaviour (Minckley, 2. METHODS 1994). (ii) Sexual colour dimorphism.Inmany species males and females have approximately 2.1. Taxa examined the same colouration, but in others the coloura- tion differs considerably. In dimorphic species We used the dataset supplied by Leys et al. males are usually entirely covered with light (2002) supplemented with representatives of several pubescent hairs (e.g. white, yellow or olive additional subgenera, as well as additional species green), in contrast with females that are mainly in the subgenus X. (Koptortosoma). Forty-five Xylo- black often with bands of light coloured copa species, belonging to 28 subgenera are now in- pubescence (e.g. in X. (Koptortosoma)andX. cluded in the molecular data set. Information about (Neoxylocopa)). It has been suggested that the mating strategies is available for 20 of these species light colouration in males could function as a (16 subgenera). These species represent all major visual recognition cue for females or to de- taxonomic divisions in the genus, and represent all crease thermal stress during the energetically recognized mating behavioral patterns and associ- expensive hovering flights (Minckley et al., ated morphological variation. References for mat- 1991). Sexual colour dimorphism therefore ing strategies are listed in Table I of the online sup- plementary information. seems to be associated with species in sub- genera that perform non-resource defence be- haviour. (iii) Male eye size. In several species 2.2. Molecular methods males have large eyes compared to the fe- males. It is likely that vision may be an im- DNA extraction, PCR amplification and se- portant part of mate encounters in RD, where quencing were performed as described in Leys et al. Ecology and evolution of mating systems 121 Table I. Correlation between behavioural and morphological characters using Bayes factors. Any Bayes Factor with a positive ratio greater than 2 is considered positive evidence, greater than 5 is strong evidence, and greater than 10 is very strong evidence for correlated evolution (Pagel and Meade, 2006). Harmonic mean (Log) Independent Dependent Bayes Correlated Characters analysis analysis Factor Mating – Mesosomal glands –34.6 –27.2 14.6 Mating – Sexual colour dimorphism –38.1 –30.7 14.8 Mating – Male eye size –38.7 –35.0 7.3 Mesosomal glands – Sexual colour dimorphism –43.3 –38.7 9.1 Mesosomal glands – Male eye size –44.5 –40.7 7.4 Male eye size – Sexual colour dimorphism –47.7 –44.1 7.1 (2002). The present data set is based on CO1, obtained. A burn-in of 30000 iterations was chosen Cyt-b, EF-1α and PEPCK (GenBank Acc. No.’s. for each independent run of MRBAYES. AY005222 – AY005329), and new sequences are available on Genbank (Acc. No.’s EU180073 – EU180132). 2.4. Analyses of correlated evolution To study correlated evolution we used 2.3. Phylogenetic analyses BayesTraits, a computer package for analyzing trait evolution among species on phylogenetic trees Phylogenetic analyses of aligned sequence (available from www.evolution.rdg.ac.uk). This data, of 2437 bp, were carried out using the package allows analysis of pairs of discrete binary program MRBAYES v.3.1 (Huelsenbeck and characters using reversible jump Markov Chain Ronquist, 2001). A General Time Reversible model Monte Carlo (rj-MCMC) methods. The theory and (Rodriguez et al., 1990), with a proportion of methods, including model testing, are described in invariant sites and unequal rates among sites depth in Pagel and Meade (2006) and the manual of (Yang, 1996), modeled with a gamma distribu-