Evolution of sexual dichromatism in relation to nesting habits in European passerines: a test of Wallace’s hypothesis

Authors

  • J. J. SOLER,

    1. Estación Experimental de Zonas Áridas (EEZA-CSIC), Cañada de San Urbano, Almería, Spain
    Search for more papers by this author
    • 1Both authors have contributed equally to this article.

  • J. MORENO

    1. Departamento de Ecología Evolutiva, Museo Nacional de Ciencias Naturales (CSIC), Madrid, Spain
    Search for more papers by this author
    • 1Both authors have contributed equally to this article.


Juan J. Soler, Estación Experimental de Zonas Áridas (EEZA-CSIC), Carretera de Sacramento s/n, Cañada de San Urbano, E-04120 Almería, Spain.
Tel.: +34 950281045; fax: +34 950277100; e-mail: jsoler@eeza.csic.es

Abstract

Wallace proposed in 1868 that natural rather than sexual selection could explain the striking differences in avian plumage dichromatism. Thus, he predicted that nesting habits, through their association with nest predation, could drive changes in sexual dichromatism by enabling females in cavity nesters to become as conspicuous as males, whereas Darwin (1871, The Descent of Man and Selection in Relation to Sex, John Murray, London) argued that sexual selection was the sole explanation for dichromatism. Sexual dichromatism is currently used as indicating the strength of sexual selection, and therefore testing Wallace’s claim with modern phylogentically controlled methodologies is of prime interest for comparing the roles of natural and sexual selection in affecting the evolution of avian coloration. Here, we have related information on nest attendance, sexual dichromatism and nesting habits (open and cavity nesting) to male and female plumage conspicuousness in European passerines. Nest incubation attendance does not explain male or female plumage conspicuousness but nest type does. Moreover, although females of monochromatic and cavity nesting species are more conspicuous than females of other species, males of monochromatic and open nesting species are those with more cryptic plumage. Finally, analyses of character evolution suggest that changes in nesting habits influence the probability of changes in both dichromatism and plumage conspicuousness of males but do not significantly affect those in females. These results strongly suggest a role of nesting habits in the evolution of plumage conspicuousness of males, and a role for sexual selection also in females, both factors affecting the evolution of sexual dichromatism. We discuss our findings in relation to the debate that Darwin and Wallace maintained more than one century ago on the importance of natural and sexual selection in driving the evolution of plumage conspicuousness and sexual dichromatism in birds, and conclude that our results partly support the evolutionary scenarios envisaged by both extraordinary scientists.

Introduction

The association of extreme divergence in appearance of males and females (henceforth sexual dichromatism) with processes of sexual selection has been debated even before Darwin (1871) proposed his ground-breaking theory to explain natural ornamentation. Alfred R. Wallace questioned already in 1868 the capacity of Darwin’s theory to explain single-handedly the striking degree of variation in sexual dichromatism found in nature, especially when applied to avian plumage coloration. Thus, he claimed that ‘the sexual differences of colour and plumage in birds are very remarkable and have attracted much attention; and in the case of polygamous birds have been well explained by Mr. Darwin’s principle of sexual selection……; but his theory does not throw any light on the causes which have made the female toucan, bee-eater, parroquet, macaw and tit, in almost every case as gay and brilliant as the male, while the gorgeous chatteres, manakins, tanagers, and birds of Paradise, as well as our own blackbird, have mates so dull and inconspicuous that they can hardly be recognised as the same species’. After planting doubts about sexual selection as the unique explanation, Wallace (1868) associated sexual dichromatism with the nesting habits of birds in relation to the risk of nest predation. He considered that, assuming that (i) incubation attendance by either sex promotes cryptic plumage in open nesters, but (ii) not in cavity or domed nesters, (iii) conspicuous sexual monochromatism should be associated with cavity or domed nesting, and (iv) sexual dichromatism with conspicuous males and cryptic females should be related to open nesting (Table 1). Wallace (1868) offered support for the two last predictions by listing 23 phylogenetically related group of birds (i.e. families or genera) with conspicuous monochromatism nesting in cavities or domed nests and seven families with bright males and dull females with open nesting habits. Wallace (1868, 1889) also predicted that because of the higher phylogenetic lability of plumage colour, changes in nesting habits would come first and be followed by changes in coloration. Darwin (1871) disagreed with this view and forcefully argued that plumage coloration could select for changes in nesting habits while the opposite was less plausible. In nearly a century and a half elapsed since Wallace first presented his theory on avian sexual dichromatism in relation to nesting habits, few attempts have been made to empirically check its validity despite the attention that sexual dichromatism as variable reflecting the strength of sexual selection in different bird species has received during the last decades (see for instance, Amundsen & Pärn, 2006) and the huge increase in information on avian natural history and phylogeny. The recent literature on avian sexual dichromatism still does not recognize nest concealment as an important explanation for this phenomenon (Badyaev & Hill, 2003). According to recent views (Amundsen & Pärn, 2006), dichromatism is a product of differences in the strength of sexual and/or social selection (i.e. due to differential success in social competition, whatever the resource at stake (i.e. due to differential success in social competition, whatever the resource at stake; West Eberhard, 1983) acting on the two sexes. Natural selection for crypticity in open nesters would override sexual selection in both sexes (cryptic monochromatism, not covered by Wallace), or only in females (dichromatism), depending on the strength of sexual/social selection in males to reduce their nest attendance (Table 1). Thus, Wallace was correct in arguing that selection due to nest predation is crucial for understanding the evolution of sexual dichromatism in birds. In comparison with open nesters, natural selection for crypticity in cavity nesters would be weaker (Martin & Li, 1992), and sexual/social selection in females would therefore dominate in many cases, rendering cases of conspicuous monochromatism (Table 1). Wallace was therefore right against Darwin (1871) that natural selection could exert independent selective influences on the two sexes (Caro et al., 2008), which could explain flamboyant monochromatism. However, sexual/social selection in females may be weaker than in males in other situations resulting in cases of dichromatism not predicted to exist by Wallace (Table 1).

Table 1.   Explanations of the relationships between sexual dichromatism, nesting habits and plumage conspicuousness presented by Wallace (1889), Darwin (1871) and by recent literature.
 MalesFemales
Monochromatic cryptic
 Open nestingWallace: not coveredWallace: not covered
Darwin: weak sexual selectionDarwin: genetic correlation
Present view: natural selectionPresent view: natural selection
Dichromatic conspicuous males and cryptic females
 Open nestingWallace: no selectionWallace: natural selection
Darwin: sexual selectionDarwin: weak genetic correlation
Present view: sexual selectionPresent view: natural selection
Monochromatic conspicuous
 Cavity nestingWallace: no selectionWallace: no selection
Darwin: sexual selectionDarwin: strong genetic correlation
Present view: sexual selectionPresent view: sexual and/or social selection
Dichromatic conspicuous males and cryptic females
 Cavity nestingWallace: no selectionWallace: not predicted to exist
Darwin: sexual selectionDarwin: weak genetic correlation
Present view: sexual selectionPresent view: weak sexual and/or social selection

In this article, we tested some assumptions and predictions of Wallace’s theory by analysing plumage conspicuousness and dichromatism, nesting habits and incubation attendance of European passerines as described in Handbook of Birds of The Western Palearctic (HBWP; Cramp & Perrins, 1988, 1992, 1993, 1994a,b). We have also corrected for phylogenetic relationships in all analyses as nesting habits, and to a lesser degree sexual dichromatism, may show a marked phylogenetic component as already argued by Wallace (1889). According to the fundamental assumption of Wallace that incubation attendance by either sex promotes cryptic plumage in open nesters, but not in cavity nesters, conspicuousness in either sex should be related to incubation attendance, nest type and their interaction (Prediction 1). Moreover, the predictions by Wallace that conspicuous sexual monochromatism should be associated with cavity or domed nesting, and sexual dichromatism with conspicuous males and cryptic females should be related to open nesting, were tested by relating degree of male and female conspicuousness to nest type and sexual dichromatism. For predictions to be fulfilled, male conspicuousness should not depend on nest type or dichromatism, whereas female conspicuousness should be associated with nest type and dichromatism (Prediction 2). This is because dichromatism according to Wallace is the product of selection on female conspicuousness in open nesters, with selection on males neither explaining dichromatism nor being related to nesting habits. Furthermore, we have also tested the predictions derived from recent views on the importance of sexual/social selection that sexual dichromatism in cavity nesters should be negatively related to conspicuousness of the attendant sex, in most cases females (Prediction 3), whereas sexual dichromatism in open nesters should be positively related to conspicuousness of the nonattendant sex, in most cases males (Prediction 4). Finally, we also reconstruct character evolution in the phylogenetic tree and, by means of Pagel’s (1994) discrete method to test models of independent and dependent evolution, test Wallace’s (1889) prediction that plumage colour should be influenced by nesting habits and not the other way around as envisaged by Darwin (1871) (Prediction 5).

Materials and Methods

Characterization of plumage coloration and nesting habits

We characterized male and female plumage conspicuousness from the coloured plates of volumes V–IX of HBWP (Cramp & Perrins, 1988, 1992, 1993, 1994a,b). These plates represent birds in larger size than in field guides and constitute probably the most scientifically accurate representation of Western Palearctic birds. Only 163 species with European distribution (some species in HBWP are present only outside Europe) were included given the phylogenetic tree used. Morphology like tail length or bill length or coloration of structures other than body feathers was not considered. To reduce the possibility of involuntary bias, a noninformed layperson and two students without knowledge about the hypotheses being tested characterized images of females and males as cryptic or conspicuous during brief (a few seconds) observations of the images in the plates, the least hesitation leading to consideration of cases as intermediate. Only one image could be observed at a time as the rest of the plate was covered while scoring. From the first printed edition of HBWP, we extracted the information concerning sexual dichromatism (yes/no), male and female incubation attendance (females, mainly females, males and females equally) and nest type, classified as cavity, domed or open nests. We have used a dichotomous score of sexual dichromatism because this score predicts susceptibility to predation (Huhta et al., 2003; Møller & Nielsen, 2006; Møller et al., 2006).

Conspicuousness of attendants at open nests obviously depends on the surrounding environment and on light conditions. Background-dependent conspicuousness is surely very difficult to estimate for open nesters as there is a large inter and intraspecific variation in nesting environments. To attempt such an analysis requires sophisticated and detailed studies of light conditions in nesting environments, which are not available for most species in Europe or elsewhere. Our rationale rather than to explore the visual determination of crypticity/conspicuousness is to check whether modern phylogenetic methods sustain Wallace’s conclusions based on basic natural history knowledge and common human perception of avian conspicuousness, presumably similar to the one used by Wallace himself.

Sexual dichromatism as presented in reference works like Cramp is based on the assumption that human vision allows a reasonable approximation of dichromatism or of bird coloration generally, as perceived by the birds themselves, many of which can perceive UV light in contrast to humans (Bennett et al., 1994). This basic assumption has been recently validated for a majority of avian families that may include many avian nest predators (Seddon et al., 2010). However, conspicuousness to avian predators like raptors and corvids should not be equated to attractiveness to conspecifics (Håstad et al., 2005). Moreover, it is unknown to what degree human visual perception of avian plumage conspicuousness agrees with perception by potential mammalian or reptilian nest predators. In what follows, we assume that human perception of plumage conspicuousness is positively correlated with nest predator perception.

Phylogenetic analyses

We considered conspicuousness scores for plumage with two levels (1 = conspicuous; 2 = cryptic), with dubious cases scoring as 1.5. Because scores of the three estimations for male (r = 0.796, F269,540 = 12.72, P < 0.00001) and female conspicuousness (r = 0.663, F249,500 = 6.90, P < 0.00001) were moderately repeatable, average scores for the three observers were used in analyses as dummy variables for female and male conspicuousness, respectively. These variables have been considered as dependent variables in analyses. Moreover, under Wallace’s assumption species with domed nest should suffer from intermediate rates of predation and, thus, we used nesting habits as a dummy variable (cavity = 3, domed = 2 and open = 1). Degree of male incubation attendance was also considered as continuous dummy variable (only females = 1, mainly females = 2, and males and females equally = 3). The use of dummy variables in regression analyses is well established in the statistical literature (e.g. Zar, 1999) as it allows performing multivariate regression analyses including discrete variables that can be controlled for phylogenetic influences. Moreover, the effect of domes in affecting nest predation may be debatable because, for instance, it may increase probability of nest detection (Darwin, 1871). Thus, we also performed the analyses excluding species with domed nests. As results did not vary with respect to statistical significance depending on whether species with domed nests were excluded from the analyses or not, we will here present results from comparisons of open nesters and cavity nesters excluding domed nests as an intermediate category. We also checked whether body mass was implicated in the associations between nesting habits and conspicuousness by the inclusion of body mass in analyses. However, as body mass was unrelated to conspicuousness and because Wallace’s hypothesis does not deal with body mass, we do not present these results that do not change the conclusions here presented. See Appendix S1 for used values of considered species traits.

Taxonomic groups such as species cannot be considered statistically independent observations due to the confounding effects of common ancestry (Harvey & Pagel, 1991). Therefore, we tested how plumage conspicuousness of males and females is related to sexual dichromatism and nesting habits using the lambda statistic of Pagel (Pagel 1999; Freckleton et al., 2002) and a molecular phylogeny of European birds recently published (Thuiller et al., 2011) (Appendix S2). To control for the phylogenetic relationship among the sampled species, we used phylogenetic generalized least square regression (PGLS) models (Pagel, 1997, 1999) as implemented in the r statistical environment with the appropriate libraries (‘ape’, ‘MASS’ and ‘mvtnorm’) and additional functions by R. Freckleton (University of Sheffield) as implemented in the CAIC package for r (http://r-forge.r-project.org/projects/caic). The PGLS approach characterizes evolutionary changes along each branch of a phylogeny through the variance components of traits and controls for the nonindependence among species by incorporating a matrix of the covariances among species based on their phylogenetic relationships (Martins & Hansen, 1997; Pagel, 1997, 1999). Thus, phylogenetic information is incorporated to the error term, thereby controlling for the shared evolutionary history among species (Harvey & Pagel, 1991; Martins & Hansen, 1997). The method applies likelihood ratio statistics to test hypotheses of correlated trait evolution and also to estimate the phylogenetic signal (λ). The phylogenetic signal represents the importance of phylogenetic corrections in the models (Freckleton et al., 2002), which varies between 0 (phylogenetic independence) and 1 (species’ traits covary in direct proportion to their shared evolutionary history) (Pagel, 1997, 1999). Lambda was incorporated to the error term to control for the effect of phylogenetic relationship on the degree of phylogenetic dependence of the PGLS models.

For studying character evolution and the direction of changes along the phylogenetic tree, we used nesting habits (cavity vs. open) as the first binary variable and plumage conspicuousness of males and females [conspicuous (values ≥ 1.5) vs. cryptic (values < 1.5)], and sexual dimorphism (monochromatic vs. dichromatic) as the second variables in three different models. Briefly, we estimated ancestral states by using parsimony reconstruction methods as implemented in Mesquite (Maddison & Maddison, 2009). To analyse correlated evolution between nest type and sexual dichromatism or plumage conspicuousness of males and females, we used Pagel’s (1994) discrete method to test models of independence and dependent evolution. This method compares the ratio of likelihood of two models: one of the models where the rates of change in each character are independent of the state, and a second model where rates of change depend of the state of the other trait. As likelihoods associated with each of the eight possibilities of transition are estimated, this approach provides a good method to study evolutionary pathways through estimations of transition rates between pairs of binary character states. We performed these analyses as implemented in Mesquite (Maddison & Maddison, 2009) with 50 ML replicates over 500 repeated simulations.

Results

Prediction 1: Nest attendance and plumage conspicuousness

Neither nest attendance (PGLS; males: t133 = 0.33, P = 0.74, females: t133 = 0.99, P = 0.33) nor nest type (PGLS; males: t133 = 1.88, P = 0.062, females: t133 = 0.88, P = 0.38) explained degree of plumage conspicuousness of males (whole PGLS model, λ = 0.730, R2 = 0.027, F2,134 = 1.84, P = 0.16) or females (whole PGLS model, λ = 0.970, R2 = 0.013, F2,134 = 0.90, P = 0.41). Moreover, degree of male attendance did not associate with plumage conspicuousness of males when considering either cavity (PGLS; t34 = 0.14, P = 0.96) or open nesters (PGLS; t98 = 0.27, P = 0.78). Similar results came out when analysing conspicuousness of female plumage (PGLS; cavity nesters: t34 = 0.28, P = 0.78; open nesters: t98 = 0.71, P = 0.48) (Fig. 1). Thus, incubation attendance does not predict conspicuousness of female or male plumage. Phylogenetic signals were moderately strong.

Figure 1.

 Average values (±95% confidence intervals) of conspicuousness index of male and female plumages for species where only females (F) or males and females (M and F) contribute to incubation duties and that breed in open or cavity nests.

Prediction 2: Sexual dichromatism and nesting habits

In contrast to predictions by Wallace, conspicuousness of male plumage was explained by both nest type and sexual dichromatism (whole PGLS model, λ = 0.755, R2 = 0.296, F2,141 = 29.69, P < 0.0001). Males of cavity nesters (PGLS, t142 = 2.07, P = 0.040) and of dichromatic species (PGLS, t142 = 7.40, P < 0.0001) are more conspicuous than those of open nesters and monochromatic species, respectively (Fig. 2). However, the effect of sexual dichromatism was only detected for males of open nesters (PGLS, t142 = 7.16, P < 0.0001), but not for those of cavity nesters (PGLS, t142 = 1.54, P = 0.132) (Fig. 2). Partly in support of Wallace’s prediction for females, sexual dichromatism significantly explained conspicuousness of female plumage (whole PGLS model, λ = 0.881, R2 = 0.103, F2,141 = 8.08, P = 0.0004). Females of dichromatic species are less conspicuous than those of monochromatic species (PGLS, t142 = 3.83, P = 0.0002), and this effect appeared both for open (PGLS, t107 = 3.02, P = 0.003) and for cavity nesters (PGLS, t35 = 3.15, P = 0.0034) (Fig. 2). However, giving only weak support for Wallace’s hypothesis, average female conspicuousness of cavity and open nesting species does not differ (PGLS, t142 = 1.29, P = 0.20), although females of monochromatic cavity nesters tended to be more conspicuous than those of monochromatic open nesting species (PGLS, t129 = 1.90, P = 0.060) (Fig. 1). Phylogenetic signals in analyses were fairly strong.

Figure 2.

 Average values (±95% confidence intervals) of conspicuousness index of male and female plumages for monochromatic (Monochrom.) and dichromatic (Dichrom.) species that breed in open or cavity nests.

Predictions 3 and 4: Dichromatism and attending sex

Confirming predictions derived from recent views on the importance of sexual/social selection, we found that females of monochromatic species breeding in cavities are more conspicuous than females of other species (PGLS, t142 = 2.21, P = 0.029), whereas males of monochromatic species that breed in open nests are those with more cryptic plumage (PGLS, t142 = 2.36, P = 0.020). Moreover, conspicuousness of males of dichromatic species does not depend on nesting habits (PGLS, t86 = 0.24, P = 0.81), but males of monochromatic species are duller in open nesters than in cavity nesters (PGLS, t99 = 3.66, P = 0.0004) (Fig. 1).

Prediction 5: Character reconstruction

Together with previous results, the study of character reconstruction along the phylogenetic tree provides a more complete picture of the evolutionary association between nesting habits and plumage conspicuousness. First, the ancestral state of European passerines was equivocal for dichromatism, but was clearly open nesting for nesting habits, with females and males showing conspicuous plumage (see Appendix S3). The most parsimonious reconstruction of sexual dichromatism as estimated from Cramp & Perrins (1988, 1992, 1993, 1994a,b) did not however provide an unequivocal value. Moreover, and in accordance with Wallace’s theory, we find support for an evolutionary correlation between nest type and dichromatism [Pagel’s, 1994 test of correlated (discrete) character evolution, P = 0.046]. The best model generated includes nest type as the independent factor, thereby suggesting that changes in nest type influence changes in sexual dichromatism (Fig. 3). However, when analysing the correlated evolution between nest type and plumage conspicuousness of males and females separately, we find evidence of such a correlation for males [Pagel’s, 1994 test of correlated (discrete) character evolution, P = 0.024], but not for females [Pagel’s, 1994 test of correlated (discrete) character evolution, P = 0.23]. Again, resulting likelihood values suggest that changes in nest type influence changes in male plumage coloration, but not significantly in females (Fig. 3).

Figure 3.

 Flow diagrams showing correlated evolution between nesting habits (open vs. cavity nesting) and plumage conspicuousness [conspicuous (conspic.) vs. cryptic (crypt.)] of males and females, and plumage dichromatism. Numbers close to arrows represent maximum likelihoods and, together with arrow thickness, indicate which transition is expected to be the most common. Probability (P) in the centre of diagrams indicates probability of correlated evolution between the binary traits after 50 ML replicates and 500 repeated simulations.

Discussion

We have been unable to confirm several key assumptions and predictions of Wallace’s (1868) theory on the link between plumage conspicuousness and nesting habits. First, incubation attendance was not related to plumage conspicuousness for either cavity or open nesters. Males were no more cryptic when partaking in incubation duties in open nesters, and females were no more conspicuous when incubating in cavities. Second, several cavity nesters were dichromatic, a possibility not contemplated by Wallace. Third, some open nesters were cryptically monochromatic, a result which should not be due to male incubation attendance as we considered earlier. These cases were not discussed by Wallace (1868) although they cast some doubt on the link between nesting habits and sexual dichromatism. Darwin (1871) already commented on evidence contradicting Wallace’s theory. Thus, when discussing apparent exceptions to the theory he argued that ‘if we look to the birds of England we shall see that there is no close and general relation between the colours of the female and the nature of the nest which is constructed’. He then mentioned as examples several European passerines included in our analyses. He also cited cases of bright males attending open nests and stated that ‘if brilliant colours had been extremely dangerous to birds while sitting on their open nests, the males in these cases would have suffered greatly. It might, however, be of such paramount importance to the male to be brilliantly coloured, to beat his rivals, that this may have more than compensated some additional danger’. Similar cases probably explain in our data set the absence of significant effects of male participation in incubation duties on plumage conspicuousness in open nesters. Thus, Darwin introduced sexual selection into the debate with Wallace about plumage coloration and nesting habits.

One way of considering the strength of sexual selection as proposed by Darwin is to include sexual dichromatism in the analyses. Indices of sexual dimorphism, and especially of sexual dichromatism, are frequently used as surrogates for the strength of sexual selection in comparative analyses including predation risk (Huhta et al., 2003; Møller & Nielsen, 2006; Møller et al., 2006). If we interpret sexual dichromatism in this light, our analyses suggest that sexual selection explains the variation in conspicuousness of males of open nesters as suggested by Darwin (male plumage of dichromatic species is more conspicuous than that of monochromatic species). However, our results also suggest that probability of predation at the nest may constrain the evolution and/or maintenance of conspicuous coloration of males, because males of monochromatic species are more cryptic when nesting in open than when nesting in cavity nests. Moreover, changes in nesting habits and sexual dichromatism are significantly related throughout the phylogeny of European passerines. The evolution of conspicuous plumage of males in fact occurs more frequently in cavity nesting ancestors (Fig. 3). Thus, our results suggest that not only sexual selection, but also natural selection due to predation plays an important role in determining both sexual dichromatism and plumage conspicuousness of males. Nesting habits not only affect the probability of predation at the nest, but also other life history traits including those related to reproductive effort such as clutch size (Martin & Li, 1992) that, at the same time, may be affected by sexually selected traits (Andersson, 1994). Thus, the detected association between nesting habits and sexual dichromatism and conspicuousness of males could be due to reasons other than the risk of nest predation (Martin & Badyaev, 1996). In any case, we have, to our knowledge for the first time based on a phylogenetically controlled analysis, pointed out a potential role of nesting habits in affecting the evolution of avian plumage conspicuousness, an effect that was predicted by Wallace more than a century ago especially for females.

Darwin (1871) was reluctant to consider the possibility that female ornamentation could evolve driven by sexual selection on females. In his view, the slightly dimmer female plumage in monochromatic cavity nesters was due to the laws of inheritance, or what we today would interpret as genetic correlations between the sexes (Lande, 1980) (Table 1). However, given the evidence for striking dichromatism in many species, this interpretation was open to challenge. One could ask the reasons for the marked variation in the strength of inheritance among species. Wallace offered nest predation as a selective factor reducing the strength of genetic correlation among the sexes, whereas Darwin argued weakly ‘that the degree of limitation should differ in different species of the same group will not surprise anyone who has studied the laws of inheritance, for they are so complex that they appear to us in our ignorance to be capricious in their action’. Thus, Wallace was correct in stressing that natural selection could easily break genetic correlations for ornamentation, whereas Darwin was right in arguing that sexual selection was necessarily involved in plumage evolution (Caro et al., 2008). We have found that females of monochromatic species wear more conspicuous plumages than those of dichromatic species and, as Wallace predicted, it occurs mainly in cavity nesting species (Fig. 2), suggesting that selection on female plumage conspicuousness probably involves sexual selection. However, the effect of nesting habits was not as strong as in males because we failed to find a significant relationship after controlling for phylogenetic effects.

There are important differences between our analysis and that presented by Wallace (1868). The taxonomic scope was wider in Wallace’s analysis, which allowed including a larger range of variation in plumage conspicuousness. He also considered coloration of structures other than plumage such as toucan bills. By restricting ourselves to body plumage, we may have excluded some aspects of coloration that may affect predation risk. However, passerines are relatively uniform in morphology and constitute by far the largest segment of avian taxonomic diversity in almost any environment. Wallace (1868) explained away exceptions to his rule as Darwin (1871) correctly pointed out and mixed some rather cryptic monomorphic cavity nesters with conspicuous ones. However, the main reason for differences in results between Wallace’s analysis and the present ones probably resides in the absence of techniques for phylogenetic correction until fairly recently. Wallace could thus not separate the effects of ecology from those derived from phylogenetic ancestry. The moderately large phylogenetic signals detected in our analyses indicate that phylogenetic ancestry explains a large part of the variation used by Wallace (1868) to make his case.

Understanding the causes of sexual dimorphism and their relevance for the Darwinian concept of sexual selection has been a major objective for evolutionary biologists. Most studies have adopted the microevolutionary perspective and have tried to explain adaptive function (the fitness consequences of sexual ornaments; Andersson, 1994; Møller, 1994), by implicitly assuming that dimorphic species have evolved from cryptic monomorphic species (Friedman et al., 2009). However, the ancestors of European passerines with respect to dichromatism could be monochromatic or dichromatic, but were probably open nesters and with conspicuous plumage in both males and females. Thus, it is likely that selection pressures due to predation have selected for cryptic plumage and/or cavity nesting as they are the apomorphic traits in descendant species. Mating system, migratory behaviour, insularity, sympatry with related species and probability of predation are factors that have been suggested to influence sexual dichromatism in different taxa (Martin & Badyaev, 1996; Figuerola & Green, 2000; Friedman et al., 2009). The influence of nesting habits (as a proxy for probability of predation at the nest) on the evolution of sexual dimorphism has, to our knowledge, only been tested in Anseriformes without finding support for it (Figuerola & Green, 2000). Our results for European passerines are therefore the first evidence that cavity nesting has affected the evolution of sexual dichromatism and plumage conspicuousness of birds.

To conclude, we have found comparative evidence of nesting habits being an important trait driving the evolution of sexual dichromatism in birds, by favouring conspicuous plumages of males when nesting in protected holes and cryptic plumage in females when breeding in unprotected open nests. Moreover, we have also found evidence of a role of sexual selection in explaining conspicuous plumages of females of monochromatic species (i.e. mutual sexual selection). These results do not confirm Wallace’s predictions in a large sample of European passerines but, in accordance with a previous article (Martin & Badyaev, 1996), nest predation appears to be only one force among others in driving plumage colour evolution. Nest attendance however is not as crucial as envisaged by Wallace, and sexual selection on males drives dichromatism in both open and cavity nesters. Moreover, contrary to Wallace’s view, natural selection on females probably explains cryptic monochromatism in open nesters rather than conspicuous monochromatism of cavity nesters that would more appropriately be explained by mutual sexual selection. In any case, Wallace (1868, 1889) was correct in emphasizing other factors than genetic correlations in explaining female plumage conspicuousness and correctly conjectured a role of nesting habits in explaining the evolution of avian plumage conspicuousness, although he mistakenly expected this role to apply to females and not to males.

Acknowledgments

JJS and JM were supported by grants from Ministerio de Ciencia e Innovación/FEDER (grants CGL2010-19233-C03-01 and CGL2010-19233-C03-02, respectively) while writing this article. Consuelo Corral, Sonia González and Rafael Ruiz-de-Castañeda scored conspicuousness in plates of HBWP. T. Caro and an anonymous reviewer offered much appreciated comments on a previous version.

Ancillary