Covariation between brain size and immunity in birds: implications for brain size evolution

Authors


Anders Pape Møller, Laboratoire de Parasitologie Evolutive, CNRS UMR 7103, Université Pierre et Marie Curie, Bât. A, 7ème étage, 7 quai St. Bernard, Case 237, F-75252 Paris Cedex 05, France.
Tel.: (+33) 1 44 27 25 94; fax: (+33) 1 44 27 35 16; e-mail: amoller@snv.jussieu.fr

Abstract

Parasitism can negatively affect learning and cognition, setting the scene for coevolution between brain and immunity. Greater susceptibility to parasitism by males may impair their cognitive ability, and relatively greater male investment in immunity could compensate for greater susceptibility to parasites, in particular when males have a relatively large brain. We analysed covariation between relative size of immune defence organs and brain in juvenile and adult birds. The relative size of the bursa of Fabricius and the spleen in adults covaried positively with relative brain size across bird species. The relative size of these two immune defence organs covaried with sex differences in relative size of the brain, indicating that the relationship between immune defence and brain size was stronger for males. In contrast, liver and heart size or sexual size dimorphism in size did not covary with immune defence. Thus, species in which males have relatively large brains also have relatively large immune defence organs.

Introduction

We propose a novel hypothesis linking immunity, sexual selection and brain size evolution, suggesting that selection for larger brain size has favoured individuals (in particular males) with strong immune responses, because such individuals would suffer the least in terms of impaired learning and cognition because of the negative effects of parasites. This hypothesis is based on the assumptions that (i) learning is facilitated by an absence of parasitism; (ii) the two sexes differ in susceptibility to parasites; and (iii) both immunity and secondary sexual characters are condition dependent. Here, we first present these three assumptions; secondly, we provide four different scenarios linking coevolution between immunity and brain size; and, thirdly, we present some testable predictions that we subsequently evaluate in the Results section of the paper.

The first assumption is that learning is facilitated by an absence of parasites and disease. Discriminative learning and spatial and nonspatial cognitive performance are impaired in individuals suffering from a range of parasitaemias, and parasitism may thus lead to a reduction in cognitive performance (Kershaw et al., 1959; Stretch et al., 1960; Olson & Rose, 1966; Dolinsky et al., 1981; Kvalsvig, 1988; Kvalsvig et al., 1991; Nokes et al., 1992; Kavaliers et al., 1995; Sakti et al., 1999; Al Serouri et al., 2000; Stolzfus et al., 2001; Fiore et al., 2002; Jukes et al., 2002). Given that discriminatory ability may be affected by disease status (Kershaw et al., 1959; Stretch et al., 1960; Olson & Rose, 1966; Dolinsky et al., 1981; Kvalsvig, 1988; Kvalsvig et al., 1991; Nokes et al., 1992; Kavaliers et al., 1995; Sakti et al., 1999; Al Serouri et al., 2000; Stolzfus et al., 2001; Fiore et al., 2002; Jukes et al., 2002), mate choice by females and other aspects of sexual selection may be directly affected by parasites through their effects on brain function. Bird song may provide an appropriate example. Repertoires are costly at least in terms of brain space utilization. Bird song is associated with an increase in the number of neurones and greater synaptic and dendritic development in the higher vocal centre of the brain (Nottebohm et al., 1981, 1986; Canady et al., 1984; review in Garamszegi & Eens, 2004). Annual cycles in the size of the higher vocal centre and the robust nucleus of the archistriatum used for song production suggest that the maintenance of these structures is costly in terms of brain space and/or energy (Nottebohm et al., 1986). As the size of the higher vocal centre is positively related to repertoire size in interspecific studies of birds (DeVoogd et al., 1993; Székely et al., 1996), this implies that the magnitude of the cost increases with repertoire size. Repertoire size may be linked to immune defence through the effects of parasitism and disease on song learning (Catchpole, 1996). A recent experiment in the European starling Sturnus vulgaris demonstrated that unpredictable short-term food deprivations after fledging caused suppressed humoural response that strongly affected song production in the next breeding season (Buchanan et al., 2003).

Neuroendocrine and neuroimmune secretions provide links between brain, endocrine system and immune system. The immune system has been termed our sixth sense, given its crucial role in sensing the potentially dangerous entry into the body of alien invaders (Blalock, 1994). The efficiency of the immune system depends to a high degree upon the ability of the individual to learn from past experience. Thus, the ability to avoid particular sites at particular times of the day or the year may considerably reduce the risk of serious parasite infection (Hart, 1997).

The second assumption is that sex differences in susceptibility to parasitism are common (Alexander & Stimson, 1989; Zuk, 1990; Zuk & McKean, 1996; Moore & Wilson, 2002), with males generally being more susceptible than females (Poulin, 1996; Schalk & Forbes, 1997; McCurdy et al., 1998). The cause of sex differences in parasitism is unclear. First, Alexander & Stimson (1989), Zuk (1990) and Zuk & McKean (1996) suggested that sex differences in parasitism may arise from sex differences in susceptibility caused by males suffering more than females from intense competition for access to mates. Secondly, Folstad & Karter (1992) proposed that males suffer from immunosuppression due to the negative effects of androgens on immune function, either as a consequence of elevated androgen levels in reproductive males (Folstad & Karter, 1992), or as an adaptive response to such levels (Møller & Saino, 1994; Wedekind & Folstad, 1994). Thirdly, males and females may be differentially exposed to parasites, with individuals of the more exposed sex suffering more from parasitism. Such sex differences in parasitism may also have important implications for learning and the evolution of cognitive abilities in the two sexes. If males suffer more from parasitism than females, then we should expect males to suffer from impaired learning and cognitive abilities. As sexual selection generally is more intense in males than in females in most taxa (Andersson, 1994), such impaired mental abilities would select for increased size of the brain, but also for increased immune function to ameliorate or compensate for the negative effects of sex differences in susceptibility.

The third assumption is that immunity and brain function are condition dependent. The immune system is condition dependent with responsiveness being directly associated with body condition and ingestion of protein-rich food and essential nutrients (Chandra & Newberne, 1977; Gershwin et al., 1985; Lochmiller et al., 1993; Saino et al., 1997; Møller et al., 1998c, 2003; Alonso-Alvarez & Tella, 2001). Likewise, condition can affect the normal development and functioning of the brain (e.g. Kershaw et al., 1959; Stretch et al., 1960; Olson & Rose, 1966; Dolinsky et al., 1981; Kvalsvig, 1988; Kvalsvig et al., 1991; Nokes et al., 1992; Kavaliers et al., 1995; Sakti et al., 1999; Al Serouri et al., 2000; Stolzfus et al., 2001; Fiore et al., 2002; Jukes et al., 2002). Why should brain function and immune function be condition dependent? Only individuals in prime condition are able to allocate resources differentially to several different costly functions without sacrificing any of these (Zahavi, 1975; Zahavi & Zahavi, 1997). Hence, characters that differentially affect fitness should demonstrate particularly high levels of condition dependence because condition dependence of such characters would allow individuals in prime condition to fine-tune the expression of these different characters to facilitate their efficient co-functioning. Thus it is not surprising that secondary sexual characters, life history characters, immune function and cognitive function all show high degrees of condition dependence (e.g. Chandra & Newberne, 1977; Gershwin et al., 1985; Price & Schluter, 1991; Lochmiller et al., 1993; Andersson, 1994; Saino et al., 1997; Møller et al., 1998c, 2003; Alonso-Alvarez & Tella, 2001).

We can imagine four different scenarios for the coevolution of brain and immune defence. First, relative brain size initially became enlarged and immune function evolved later. Secondly, immune function improved initially followed by subsequent brain size evolution. Thirdly, both brain size and immunity coevolved simultaneously in response to a third factor. Fourthly, selection due to parasites may directly have caused a decrease in brain size and an increase in size of immune defence organs, giving rise to a negative relationship between relative brain size and immune function. In the first scenario, species in which learning ability is of particular importance have evolved larger brains, and immune function later evolved increased efficiency to overcome problems of parasite-impaired learning deficiency. This should particularly be the case in males as compared with females if sex differences in susceptibility to parasites occur. Alternatively, a given species may initially have evolved a relatively large brain as a means to achieve a behavioural adaptation to a given problem. The behavioural change may have exposed individuals to greater parasite-mediated natural selection, causing brain size and immune defence to covary without cognition and parasitism being causally linked. In the second scenario increased investment in immune function allowed some individuals to learn more quickly or more efficiently, and this later favoured the evolution of a larger brain. In the third scenario a third factor is responsible for both brain size and immune defence organ evolution. An example of such a third factor could be body condition. In the fourth scenario species subject to intense selection from parasites may be unable to develop a large brain, giving rise to a negative relationship between brain size and immune function.

The first aim of this study was to test for covariation between relative brain size and relative size of immune defence organs, after accounting for the effects of allometry, using birds as a model system. We did this for the bursa of Fabricius, which is an immune defence organ where B-cells differentiate in juvenile birds. In contrast, the spleen is an important storage organ for B-lymphocyte differentiation and proliferation of B- and T-cells in both juveniles and adults. Therefore, we conducted separate tests of covariation between relative size of the brain and immune defence organs in two different age classes to determine whether the selection pressures that have resulted in covariation between brain size and immune defence organs acted differently among juveniles and adults. The second aim of this study was to investigate sexual size dimorphism in relative brain mass and mass of immune defence organs. If selection pressures acting on the immune system and the brain differ between the sexes, but have acted in a consistent way on these two types of organs, then we should expect covariation between sexual size dimorphism in brain size and size of immune defence organs. Again, we investigated this both in juveniles and in adults to determine to what extent age-specific selection pressures have been involved. As a control, we investigated to what extent the relative size of the heart and the liver covaried with brain size. These two organs play crucial roles in circulation, digestion and assimilation, potentially giving them an important role in a sexual selection context if the more strenuous activity of reproductive males has selected for more efficient circulation and digestion.

The bursa of Fabricius synthesizes antibodies in juvenile birds (Glick, 1983, 1994; Toivanen & Toivanen, 1987), but regresses before sexual maturity (Rose, 1981; Glick, 1983; Toivanen & Toivanen, 1987). The bursa is responsible for differentiation of the repertoire of B-cells in birds. The spleen is an immune defence organ of the peripheral lymphoid tissue, acting as the main site of lymphocyte differentiation (B-cells) and proliferation (B- and T-cells), producing cells involved in the production of humoural and cell-mediated immune responses (reviews in Arvy, 1965; Rose, 1981; Keymer, 1982; Molyneux et al., 1983; John, 1994). We assume that a larger bursa of Fabricius or spleen can provide better immune defence than a smaller organ for a bird of a given body size. More than 75% of the volume of these two immune defence organs is composed of lymphocytes (Rose, 1981; Alberts et al., 1983; Toivanen & Toivanen, 1987; John, 1994). Spleen size is recommended as a standard measure of immunocompetence in ecotoxicology studies (National Research Council, 1992). An intraspecific study has suggested a link between spleen mass and helminth infection (Shutler et al., 1999), while an interspecific study has revealed a positive association between nematode species richness and relative spleen mass in birds (John, 1995; Morand & Poulin, 2000). Individual birds in better body condition generally have a larger spleen, even when controlling for disease status (Møller et al., 1998c). In addition, individual birds killed by mammalian predators consistently had smaller average spleens than individuals that died from other causes (Møller & Erritzøe, 2000). Finally, bird species with relatively large spleens suffer more from parasite-induced mortality (Møller & Erritzøe, 2002). Therefore, the size of the spleen has implications for survival prospects of individual birds.

Material and methods

Data sources and tests for bias in data

The mass of brain, bursa of Fabricius, spleen, liver and heart were recorded by J.E. to the nearest milligram on a precision balance from post-mortem examination of dead birds. This was carried out blindly with respect to the hypothesis under test. Total body mass was also recorded on a balance to the nearest gram. Danish taxidermists are required by law to record the cause of death of all specimens in a log-book, and more than 95% of all specimens in the present data set were found dead, with the remaining being shot by hunters. Birds were frozen when received by J.E., but any effects of storage on measurements should only cause noise in the data set. In addition, there is no reason to expect sex differences in such effects.

We tested for one potential kind of bias in the present data set. Sampling date might influence size estimates of immune defence organs, as the brain, bursa of Fabricius and spleen have sometimes been demonstrated to show annual fluctuations in size (Nottebohm et al., 1986; Toivanen & Toivanen, 1987; John, 1994; Møller et al., 2003). We tested whether date of sampling differed among species. We found a significant difference for median sampling date among species for each of the three organs (Kruskall–Wallis anovas, brain: inline image =370.23, P < 0.001; bursa of Fabricius: inline image = 98.31, P < 0.001; spleen: inline image = 273.95, P < 0.001). However, we did not find indications for consistent bias in relative organ size caused by species-specific seasonal effects. In species for which data were available for all seasons, two-way anovas revealed nonsignificant interaction terms between species and a four-scale seasonal variable (brain: F = 1.282, d.f. = 36, 26, n.s.; bursa of Fabricius: F =0.420, d.f. = 33, 30, n.s.; spleen: F = 0.971, d.f. = 36, 28, n.s.). When we dropped species-season interaction terms from the models, only species as main effects explained significant amounts of variance in relative organ size, while main effects for seasons were not significant (brain: species, F = 1476.06, d.f. = 12, 12, P < 0.001; seasons, F = 0.777, d.f. = 3, 3, n.s.; bursa of Fabricius: species, F = 1091.52, d.f. = 11, 11, P < 0.001; seasons, F = 0.978, d.f. = 3, 3, n.s.; spleen: species, F =56.31, d.f. = 12, 12, P < 0.001; seasons, F = 1.81, d.f. =3, 3, n.s.). Therefore, we assumed that the results would not be confounded by potential seasonal effects, and the main pattern we intended to explain here was the consistent interspecific variation in relative brain size independent of seasonal variation. All data are provided in Appendices 1–2.

Comparative analyses

We controlled for allometric effects by using residuals from the phylogenetically adjusted linear regression of log10-transformed mass of organ on log10-transformed body size for each sex. We expressed absolute brain size dimorphism as log10(absolute female brain size/absolute male brain size). We avoided calculating brain size dimorphism based on sex-dependent relative brain sizes, because these variables are residuals from the relevant regression lines causing them to scale with the independent variable used (sex-dependent body size). Therefore, combining residuals from two regression lines may introduce bias. As the numerator and the denominator of the absolute brain size dimorphism measured as the log10-transformed ratio of absolute female and male brain size scale similarly, absolute brain size dimorphism larger than zero indicates that females have relatively larger brains than males, whereas values smaller than zero reflect the opposite trend. However, absolute brain size dimorphism may result from absolute body size dimorphism due to allometric effects. Thus the log10-transformed ratio of absolute female and male brain size should be corrected for the similar ratio in body size. This correction was based on the phylogenetically independent regression of log10(absolute female brain size/absolute male brain size) on log10(absolute female body size/absolute male body size) (slope: 0.18, intercept = −0.02, the corresponding phylogenetic model: κ = 0.24, λ =0.63, likelihood ratio (LR) = 7.56, d.f. = 1, P < 0.001, n = 127). Note that absolute brain and body size dimorphism were not residuals by definition, allowing them to be combined in a single regression. Residuals from this regression were subsequently termed relative brain size dimorphism and used in the subsequent analyses. Positive values for relative brain size dimorphism thus indicate that females have relatively larger brains when allometric effects were held constant.

Here we used data for 59 species for bursa of Fabricius with information on liver and heart size for 55 and 50 species, respectively. Data on spleen was available for 127 species, and among these species we had data on liver and heart for 111 and 80 species, respectively. These comprised all the species, for which we had data for both males and females. We calculated spleen mass for adult birds and the mass of bursa of Fabricius for juvenile birds, and we tested our predictions with age-specific (adult or juvenile) brain sizes. In both sets of analyses we also estimated the phylogenetic correlations between brain size and liver size and heart size to estimate the importance of coevolution of organs due to metabolic constraints. The correlation between adult spleen mass and mass of bursa of Fabricius in juveniles was weakly positive, but only accounted for 8% of the variance (linear regression of log10-transformed data based on species: F = 24.09, d.f. = 1,47, P < 0.05, r2 =0.08). Hence, the two series of tests were largely independent.

We applied the general method of comparative analysis for continuous variables based on generalized least squares (GLS) models using the statistical software Continuous (Pagel, 1997, 1999). The GLS approach tests phylogenetic hypotheses based on likelihood ratio statistics. This compares the log-likelihood of the model corresponding to a null hypothesis (H0) over the model for an alternative hypothesis (H1), where the likelihood ratio = −2 loge[H0/H1]. The likelihood ratio statistic is asymptotically distributed as a chi-squared variate with degrees of freedom equal to the difference in the number of parameters between the two models. Models contain three scaling parameters that can be used to scale branch lengths in the tree (κ), scale total (root to tip) path in the tree (δ), and to assess the contribution of phylogeny (λ). We first assessed the contribution of scaling parameters, κ: branch length scaling factor, and λ phylogeny scaling factor [recent simulations showed that the estimation of δ: overall path length scaling factor is biased (Freckleton et al., 2002), and therefore we avoided estimating this parameter]. Scaling parameters were fitted sequentially. Once an appropriate model with adjusted scaling parameters had been selected, we studied correlated evolution of traits of interest by comparing the goodness of fit of model H0 fitted to the data by allowing only independent evolution with that alternative H1 model that permits correlated evolution of the characters. The appropriate scaling parameters and the log-likelihood ratio statistics testing for correlated trait evolution are presented. When we controlled for potentially confounding factors we entered these variables together with the variables of interest in the same model, and tested for correlated trait evolution. If the model offering the best fit with the data allowed correlation among traits, we calculated partial phylogenetic correlation for the relationship in question.

Allometric effects were controlled by calculating residuals from the regression of the log10-transformed dependent variable on log10-transformed body mass, using Continuous. Based on this equation residuals were calculated on the raw species data (see also Purvis & Rambaut, 1995).

Phylogenetic information for our comparative analyses originated from a number of sources using molecular techniques. We constructed a composite phylogenetic hypothesis mainly based on information in Sibley & Ahlquist (1990) derived from extensive studies of DNA–DNA hybridization. This phylogeny for higher taxa was supplemented with information from Sheldon et al. (1992), Leisler et al. (1997), Cibois & Pasquet (1999) and Grapputo et al. (2001) to resolve relationships in taxa with many species. We applied branch lengths from the tapestry tree of Sibley & Ahlquist (1990) for higher taxonomic levels. Within families the distance between genera was set to 3.4ΔT50H units, and between species within genera to 1.1ΔT50H units (see also Sibley & Ahlquist, 1990; Bennett & Owens, 2002). The two composite phylogenies used in the phylogenetic analyses of young and adult birds are shown without branch length transformation in Fig. 1.

Figure 1.

Figure 1.

Composite phylogenies of bird species for the comparative analysis of covariation between relative brain mass and (a) relative spleen mass and (b) relative mass of bursa of Fabricius. The sources are given in Material and methods. The scale is given at the bottom left.

Figure 1.

Figure 1.

Composite phylogenies of bird species for the comparative analysis of covariation between relative brain mass and (a) relative spleen mass and (b) relative mass of bursa of Fabricius. The sources are given in Material and methods. The scale is given at the bottom left.

Results

Mass of bursa of Fabricius and mass of brain in juvenile birds

First we investigated the relationship between the mass of bursa of Fabricius and brain mass of juvenile birds in their first year of life. We made this restriction to ensure that the data on mass of bursa and mass of brain originated from individuals of similar age. For females we did not find a significant relationship between the two variables (Fig. 2). The phylogenetic correlation was 0.15 (Table 1), and with a sample size of 59 species this test was not powerful if the true effect size is small or intermediate (Cohen, 1988). For males the phylogenetic correlation was 0.28 and significant (Fig. 2; Table 1).

Figure 2.

Covariation between relative brain mass of juvenile birds and relative mass of bursa of Fabricius in different bird species. Relative mass was calculated as residuals from a phylogenetically corrected regression of log10-transformed organ mass on log10-transformed body mass. The lines are the linear regression lines for males and females, respectively. The corresponding phylogenetic correlations are reported in Table 1, where brain size dimorphism represents the interaction between male and female brain size.

Table 1.  Phylogenetic correlations between sex-dependent relative mass of brain, bursa of Fabricius, liver and heart in juvenile birds.
 κλPhylogenetic correlationLRd.f.PN
  1. Relative values were controlled for effects of allometry, using residuals from phylogenetically controlled linear regressions. See Results for further details.

Bursa of Fabricius
 Female brain0.620.890.150.6510.2559
 Male brain0.540.850.282.4410.0359
 Brain dimorphism0.000.10−0.467.081<0.00159
 Liver0.220.530.140.5410.3055
 Heart0.000.25<0.0010.0010.9850
Female brain
 Liver1.001.00−0.211.2810.1155
 Heart1.000.83−0.211.1810.1350
Male brain
 Liver1.001.00−0.282.2810.0355
 Heart1.000.80−0.130.4310.3650
Brain dimorphism
 Liver0.540.470.020.0210.8655
 Heart0.000.18−0.030.0210.8650

An analysis of sex differences in mass of bursa of Fabricius and mass of the brain for 59 species revealed a phylogenetic correlation of −0.46, which was highly significant (Table 1). Relative brain size in juvenile birds was not related significantly to the mass of the heart and the liver in females (Table 1). However, in males there was a significant negative association between brain size and liver size (Table 1). We found no significant evidence for bursa size being related phylogenetically to liver size or to heart size (Table 1). Controlling for the confounding effects due to covariation with liver and heart size we found a stronger association between brain size and bursa size in both sexes (partial phylogenetic correlations: females, r = 0.28, P = 0.03, n = 50; male, r = 0.41, P < 0.001, n = 50).

Mass of spleen and mass of brain in adult birds

In this analysis of spleen mass and brain mass we restricted the data to adult individuals to allow a test that was independent of the test based on bursa of Fabricius. For females we found a nonsignificant positive phylogenetic correlation between spleen size and brain size of 0.16, based on a sample of 127 species (Fig. 3; Table 2). For males the phylogenetic correlation was 0.21 and significant (Fig. 3; Table 2). Relative brain size was not significantly related to the mass of the heart or the liver in either sex (Table 2). We found a tendency for spleen mass to be positively related to liver mass but not to heart mass (Table 2). These relationships weakened previous associations between sex-dependent brain size and spleen size in males (partial phylogenetic correlations: females, r = 0.19, P = 0.08, n = 80; male, r = 0.14, P =0.21, n = 80).

Figure 3.

Covariation between relative brain mass of adult birds and relative mass of the spleen in adults of different bird species. Relative mass was calculated as residuals from a phylogenetically corrected regression of log10-transformed organ mass on log10-transformed body mass. The lines are the linear regression lines for males and females, respectively. The corresponding phylogenetic correlations are reported in Table 2, where brain size dimorphism represents the interaction between male and female brain size.

Table 2.  Phylogenetic correlations between sex-dependent, relative mass of brain, spleen, liver and heart in adult birds.
 κλPhylogenetic correlationLRd.f.PN
  1. Relative values were controlled for effects of allometry, using residuals from phylogenetically controlled linear regressions. See Results for further details.

Spleen
 Female brain0.480.810.161.6010.07127
 Male brain0.600.870.212.7210.02127
 Brain dimorphism0.200.32−0.161.6310.07127
 Liver0.000.530.171.6810.07111
 Heart0.000.53−0.120.6210.2680
Female brain
 Liver0.530.870.040.0810.70111
 Heart0.650.92−0.130.6810.2480
Male brain
 Liver0.700.930.080.3610.40111
 Heart1.000.970.150.9310.1780
Brain dimorphism
 Liver0.210.220.070.2510.48111
 Heart0.000.000.120.5510.2980

We analysed the sex difference in relationship between spleen mass and brain mass in adult birds for 127 species. The phylogenetic correlation was −0.16 (Table 1).

A global test of the two data sets revealed a phylogenetic correlation r = 0.16, P = 0.04, n = 186 for females. A test of heterogeneity showed no evidence of statistical significance (P = 0.95). For males the phylogenetic correlation was r = 0.23, P = 0.002, n =186. Again, a test of heterogeneity was not statistically significant (P = 0.62). Based on fixed effects, the sex difference in the relationship had a phylogenetic correlation r = 0.26, which was highly significant (P < 0.001, n = 186). However, a test of heterogeneity was statistically significant (P < 0.05). If calculations were based on random effect sizes, the phylogenetic correlation for sex difference was r = 0.31, at P = 0.12. If we used partial phylogenetic correlation coefficients to control for covariation between sex-dependent brain size and relative mass of liver and heart, the overall effect sizes were positive, significant and homogenous (females: r = 0.22, n = 130, P < 0.51, P for heterogeneity = 0.62; males: r = 0.25, n = 130, P < 0.01, P for heterogeneity = 0.11).

Discussion

The main findings of this comparative study were that (1) the relative mass of two important immune defence organs covaried positively with the relative mass of the brain. That was the case for juvenile birds for which the relative mass of the bursa of Fabricius covaried with the relative mass of the brain. Likewise, in adults we found positive covariation between relative mass of the spleen and relative mass of the brain. (2) The relative mass of immune defence organs covaried with sex differences in relative mass of the brain. We found little evidence of significant covariation between brain size and size of liver and heart, both when analysed for the two sexes separately, and when investigating sexual size dimorphism. Thus the patterns reported for immune defence organs are not common to all organs of physiological importance, suggesting a specific role for immune function in brain size evolution. These patterns are all the more striking given the inherent heterogeneity in the data sets and resulting conservative nature of our statistical tests. This study provides the first empirical evidence suggesting that the impact of parasites through the evolution of the immune system may have a direct role in the evolution of the nervous system. The patterns reported here are particularly striking given that nothing similar has been derived from the enormous research effort in neurosciences during the last decades. We briefly discuss each of our main findings in the following paragraphs.

Positive covariation between the relative mass of immune defence organs and relative mass of the brain was present in two data sets on birds (Figs 2 and 3). As this effect was present already in juveniles, we can infer that the selection pressures that have resulted in such covariation occur already among young age classes before sexual maturation. The relative mass of both bursa of Fabricius and spleen is positively correlated with sexual dichromatism, repertoire size, extra-pair paternity, a tropical distribution, bird migration, coloniality, hole nesting, scavenging, and predation risk (Møller & Erritzøe, 1996, 1998, 2000; Møller, 1997, 1998; Møller et al., 1998a, 2000; Garamszegi et al., 2003). Thus, a whole suite of ecological factors has acted on the evolution of relative size of these two immune defence organs. In addition, the relative mass of the spleen is positively correlated with the magnitude of parasite-induced mortality among nestling altricial birds (Møller & Erritzøe, 2002). Thus species of birds with relatively large spleens are currently subject to greater mortality caused by parasitism. These findings suggest that the positive relationship between relative size of immune defence organs and the relative size of the brain has arisen from a range of different selection pressures associated with parasitism. This is consistent with our hypothesis that hosts from a very early age invest in immune defence to mitigate negative effects of parasitism on neural tissue.

There was significant positive covariation between the mass of two immune defence organs and sex difference in relative mass of the brain, implying that the relationship between immune defence mechanisms and brain size is stronger for males than for females (Figs 2 and 3). That was not the case for the liver or the heart, which are important organs involved in digestion, assimilation and circulation. Sexual differences in immune function is common in birds, as revealed by consistent sex difference in the size of immune defence organs, which is greatest in species with the highest frequency of extra-pair paternity, implying that males of such species have disproportionately small spleens for their body size (Møller et al., 1998b). Sex differences in impact of parasites on their hosts are also common in many different taxa (Alexander & Stimson, 1989; Zuk, 1990; Poulin, 1996; Schalk & Forbes, 1997). While the two sexes of juvenile birds differ in the size of bursa of Fabricius, the sex difference in size of the spleen first appears at sexual maturity (Møller et al., 1998b). Positive covariation between sex difference in brain size and the size of immune defence organs would be consistent with an hypothesis suggesting that parasite-mediated selection has played a role in the evolution of sex differences in brain size. Our empirical findings are consistent with the hypothesis that sex differences in brain function have evolved as a consequence of sex differences in susceptibility to parasitism. Selection acting on the immune system at large may through increased susceptibility of males to parasitism impact negatively on cognitive performance including learning. Therefore, different components of the immune system (bursa and spleen) may have evolved to mitigate this negative impact of parasites on brain function. Obviously, we cannot make claims about causation based on our correlations. However, the assumptions and predictions presented here are open to experimental tests in particular species or even to experimental tests in different species allowing for tests of assumptions and predictions in a comparative framework.

The results presented here also have implications for the evolution of brain size and sex differences in brain size. Sexual size dimorphism in brains of birds covaries with sex differences in complexity of vocalizations (L.Z. Garamszegi, M. Eens, J. Erritzøe & A.P. Møller, unpublished), suggesting that sexual selection has played a role in the evolution of sexual size dimorphism of brains. This hypothesis was corroborated by a second study showing that sexual size dimorphism in brain mass covaried negatively with the extent of extra-pair paternity (L.Z. Garamszegi, M. Eens, J. Erritzøe & A.P. Møller, unpublished). Thus, female birds have evolved relatively larger brains than males in species with frequent extra-pair paternity. Thus behaviour and cognitive tasks associated with extra-pair paternity and paternity assurance have been significant selective factors acting on the evolution of brain size in the two sexes. The present study suggests that sex differences in susceptibility to parasites may also have played a role in the evolution of sexually size dimorphic brains in birds.

Acknowledgments

We are grateful to M. Eens for constructive comments. LZG was supported by a postdoctoral grant from the FWO-Flanders (Belgium).

Appendix

Table Appendix1.   Sex dependent body mass (g), brain mass (g), sample sizes of adult birds and mass of spleen (g) liver (g) and heart (g) with the corresponding body mass averaged across sexes.
Species body mass (g)Female body mass (g)Female brain mass (g)NMale body mass (g)Male brain mass (g)NSpleen mass (g)Liver mass (g)Heart mass (g)Body mass (g)
Accipiter gentilis1059.07.8154678.37.61960.82713.628.061851.0
Accipiter nisus247.33.10324136.72.648140.1494.181.926200.0
Agapornis taranta46.21.810145.31.97510.0200.950.53143.1
Alauda arvensis38.00.910140.60.96650.0281.250.53436.1
Alcedo atthis38.90.811538.30.83220.0312.090.62736.6
Alisterus scapularis168.23.9002202.25.05450.0764.151.792194.0
Alle alle138.41.7906156.51.89730.1183.131.104147.0
Amazona aestiva203.06.9801327.48.29020.56014.023.403312.0
Amazona amazonica312.37.6362350.88.68021.5107.942.756286.0
Amazona ochrocephala204.56.4202341.58.80020.3257.852.007320.0
Anas platyrhynchos1200.05.44011228.05.92070.74029.12 1145.0
Andigena bailloni149.52.9101178.12.90020.19711.17 161.0
Apus apus42.80.673438.60.73010.0221.55 30.2
Ara manilata250.07.8802291.58.84011.2005.64 265.0
Asio otus298.05.59038254.95.622160.1636.531.928284.0
Barnardius barnardi118.63.5802119.54.13760.168  104.0
Barnardius zonarius129.44.1001110.64.10020.1277.281.362127.0
Bombycilla garrulous65.01.046461.81.15940.0553.100.83261.1
Bubo bubo2992.017.18712380.017.63210.797  1690.0
Buteo buteo892.27.97126687.47.968390.68913.914.875761.0
Calidris alpina37.60.880237.40.85560.0181.46 36.9
Carduelis cannabina19.10.580218.60.680100.0300.450.28919.1
Carduelis carduelis13.30.551916.80.642130.0060.350.24515.6
Carduelis chloris27.30.8571126.30.888340.0280.950.41326.7
Carduelis flammea12.70.527111.00.55010.0050.330.19918.2
Carduelis spinus11.40.523412.30.60760.0130.420.18412.7
Carpodacus roseus21.21.030121.00.95520.0251.150.30820.3
Certhia familiaris10.40.57018.80.45010.0160.48 8.5
Chloebia gouldiae12.20.400113.60.46240.0120.46 10.4
Chloropsis hardwickii21.20.720126.30.80010.0581.880.38830.6
Ciconia ciconia3334.015.78023350.016.24011.439  4000.0
Cinclus cinclus58.01.440272.00.92010.0302.59 58.0
Coccothraustes coccothraustes56.71.492355.51.585150.0761.660.81452.6
Columba livia247.02.2851282.02.69430.250  304.0
Corvus corax1350.014.52011313.315.81531.207  1328.0
Corvus corone512.98.19711553.78.44350.38811.583.998467.0
Corvus frugilegus472.97.6402574.38.24820.39013.395.293461.0
Corvus monedula165.34.4103232.54.88360.1396.292.431155.0
Cuculus canorus93.21.4192118.61.50030.0312.081.10082.6
Emberiza calandra53.31.149157.81.13820.0301.530.66553.9
Emberiza citrinella28.20.7811228.70.785260.0400.930.42728.5
Emberiza rutila18.60.491116.50.63520.0200.740.26514.5
Emberiza schoeniclus20.10.750220.80.64520.0320.87 20.8
Erithacus rubecula17.20.713318.30.62990.0280.930.20017.0
Falco tinnunculus227.63.8425175.23.823120.0856.102.080198.0
Ficedula hypoleuca14.50.510112.70.44330.0150.66 13.6
Fratercula arctica318.54.2001383.83.72520.46917.98 394.0
Fringilla coelebs23.60.7431724.30.761390.0320.770.27929.9
Fringilla montifringilla26.70.660123.40.74950.0360.570.25922.7
Fulica atra509.72.9902527.03.43520.81153.365.291834.0
Galerida cristata42.61.009241.71.14330.0240.87 41.3
Gallinula chloropus281.81.9655343.12.11020.3367.992.906246.0
Gallus sonneratii321.02.7701915.03.61520.62416.902.080737.0
Garrulus glandarius161.23.86417171.43.999170.1833.981.386162.0
Gavia stellata984.05.20011480.06.00010.70848.42 1663.0
Hippolais icterina14.60.483313.10.51830.0380.80 13.4
Hirundo rustica17.10.538616.10.58460.0290.44 17.8
Jynx torquilla35.50.825240.10.86540.0361.570.47534.3
Lagopus lagopus561.82.4003600.42.69022.190  598.0
Lanius excubitor64.11.580155.21.38010.040  64.2
Larus canus345.34.0603470.54.10010.41315.143.137336.0
Larus ridibundus267.82.7756200.22.89740.41310.142.003228.0
Leiothrix argentauris27.00.932325.41.00450.023  17.9
Lessonia rufa15.60.500114.00.40010.0040.58 14.3
Lonchura bicolor9.70.46019.70.41010.0090.480.1268.2
Lonchura molucca9.70.37929.90.42410.0100.400.13910.4
Lonchura pallida10.80.560110.50.45210.0430.460.15310.7
Lonchura spectabilis7.80.47018.00.50010.0071.100.1137.5
Loriculus vernalis22.20.950126.21.38020.0461.05 24.0
Mergus merganser1285.06.32521635.87.31350.55545.3618.1341253.0
Motacilla alba19.00.530221.00.56990.0320.780.48419.3
Motacilla flava16.90.471117.70.41750.0110.69 16.7
Muscicapa striata12.50.396116.80.53110.0230.730.17515.5
Nucifraga caryocatactes183.05.5801204.56.06010.236  180.9
Oenanthe oenanthe25.10.72 4124.00.60010.0300.580.31327.0
Panurus biarmicus13.50.567315.50.57730.040  14.2
Parus caeruleus10.80.640311.00.596100.0130.390.15210.6
Parus major17.30.793717.80.880150.0270.590.20216.9
Parus venustulus10.60.510110.20.55220.0140.350.12410.9
Passer domesticus28.30.8921028.50.886330.0541.030.39929.3
Passer montanus23.00.779622.20.77090.0310.840.29521.4
Perdix perdix400.81.8954402.01.87690.1379.082.560371.0
Phalacrocorax carbo1932.810.06552283.08.38011.38659.30 2074.0
Phasianus colchicus1047.03.29431411.44.075460.56425.766.3601353.0
Phoenicurus phoenicurus15.80.350113.30.45470.0360.710.17516.4
Phylloscopus collybita7.30.32039.00.29220.007  8.1
Pica pica200.75.03615229.25.640130.3326.262.294205.0
Picus viridis155.53.8001184.34.62520.167  176.0
Pluvialis apricaria199.32.2534199.22.36010.0857.70 180.0
Podiceps cristatus952.83.65731024.33.50520.57743.707.909859.0
Polytelis swainsonii141.93.1712130.23.21010.0756.331.66481.5
Prunella modularis21.50.623620.30.690100.0190.950.22220.3
Psephotus varius46.11.780245.22.10010.136  47.2
Psittacula krameri111.22.968195.03.20010.0375.55 82.2
Psittacus erithacus240.37.4501344.98.79021.24814.443.117378.0
Pteroglossus aracari272.83.3752256.03.60010.43218.68 275.0
Pucrasia macrolopha932.03.3381940.53.58010.31721.48 1009.3
Pyrrhula pyrrhula24.00.834826.50.869300.0350.730.32123.7
Rallus aquaticus121.91.5801117.31.57330.095  124.0
Regulus regulus5.50.34365.90.380110.0100.310.0795.6
Scolopax rusticola331.22.5428330.82.476160.1818.513.022441.0
Sitta europaea24.11.010224.31.11010.0290.60 22.6
Somateria mollissima1740.08.04012244.77.909180.70373.1312.0042197.0
Spreo superbus56.61.760151.71.80450.1112.610.53148.1
Streptopelia decaocto134.41.4501202.71.58630.0454.242.364134.0
Strix aluco515.59.24921426.58.903170.24410.512.517458.0
Sturnus vulgaris75.51.720380.91.83770.0993.161.05969.4
Sylvia atricapilla18.70.594318.60.628100.0360.880.21519.6
Sylvia borin19.20.520218.40.65760.0200.960.28718.4
Sylvia communis15.20.540413.60.51850.0240.54 14.4
Sylvia curruca13.60.470311.50.48250.0350.530.15012.8
Sylvia hortensis20.90.680123.60.90010.0350.78 22.3
Sylvia melanocephala10.70.488411.00.56730.0200.40 10.3
Tachybaptus ruficollis144.21.6941143.51.75110.0106.851.615135.5
Tadorna tadorna676.04.88011410.05.16010.23624.07 819.0
Tangara icterocephala17.30.544112.70.62910.0391.280.15715.8
Trichoglossus haematodus106.13.3333120.63.64540.117  106.0
Tringa totanus139.21.820198.51.40620.0513.32 102.0
Troglodytes troglodytes8.70.44529.80.52070.0190.490.12410.4
Turdus iliacus62.31.076168.11.35410.0772.890.78367.3
Turdus merula101.91.7652295.81.773340.2513.920.92398.6
Turdus philomelos70.41.4291469.41.542190.1222.460.79869.1
Turdus pilaris100.51.760292.11.76230.1275.19 104.0
Tyto alba263.85.3697260.25.758200.1065.551.887308.0
Upupa epops34.51.370138.81.18020.0691.37 34.6
Uria aalge1092.34.9652633.04.36010.560  735.0
Vanellus vanellus211.62.1368240.42.00020.1668.272.205162.0
Table Appendix2.   Sex-dependent mass of body (g) and brain (g), sample sizes of juvenile birds and mass of bursa of Fabricius (g), liver (g) and heart (g) with the corresponding body mass averaged across sexes.
Species body mass (g)Female body mass (g)Female brain mass (g)NMale body mass (g)Male brain mass (g)NBursa mass (g)Liver mass (g)Heart mass (g)Body mass (g)
Accipiter gentilis977.47.8114699.07.37080.33217.615.535851.0
Accipiter nisus240.42.95340135.02.617220.1654.941.631200.0
Alcedo atthis38.70.798636.00.78450.0462.030.59736.6
Alisterus scapularis190.24.9801204.63.75010.0936.431.225194.0
Alle alle109.11.9874175.31.86410.0796.051.063147.0
Anas crecca316.72.8061336.62.62530.3248.003.278364.0
Apus apus30.50.616223.10.42010.0120.690.47630.2
Aratinga pertinax63.63.420163.34.22010.0641.940.90066.8
Ardea cinerea1074.07.60431637.58.27041.66327.319.5651794.0
Asio otus292.06.02632275.46.04270.1436.511.995284.0
Bombycilla garrulus60.51.1691660.01.184140.0403.510.90961.1
Buteo buteo787.77.81442680.47.689200.66315.114.352761.0
Carduelis chloris28.90.818827.90.964100.0240.940.38626.7
Cinclodes fuscus27.80.910133.10.95010.050  58.0
Coccothraustes coccothraustes56.41.905152.91.88870.0591.650.72252.6
Corvus frugilegus403.07.5032476.08.67720.76414.253.225461.0
Erithacus rubecula17.20.613316.20.667220.0270.810.17317.0
Falco tinnunculus192.03.65426201.33.57430.2655.091.967198.0
Fringilla coelebs23.50.8151022.80.831100.0360.910.28829.9
Gallinago gallinago113.81.410496.01.30250.0602.961.168103.0
Gallinula chloropus308.12.1252337.32.15040.14312.742.744246.0
Garrulus glandarius165.04.78210174.34.29970.2804.481.536162.0
Hirundo rustica16.60.538819.40.612150.0320.860.24217.8
Larus argentatus561.55.38521252.05.98910.86230.147.270530.0
Larus ridibundus225.12.7585224.12.98430.5149.722.147228.0
Loriculus galgulus30.01.470125.31.40410.0321.890.33030.0
Motacilla alba15.50.528318.70.65330.0540.47 19.3
Numenius arquata833.53.9512570.83.63030.77234.456.533669.0
Parus caeruleus11.10.648211.10.89010.0270.420.15010.6
Parus major15.20.796217.71.02230.0300.580.19816.9
Passer domesticus26.70.9931327.30.940180.0431.300.31029.3
Passer montanus20.40.7671120.20.788190.0280.950.25821.4
Phoenicurus phoenicurus16.30.525216.10.63050.0190.740.21616.4
Phylloscopus collybita7.90.35738.40.37730.011  8.1
Pica pica170.84.5033207.26.35860.4706.151.599205.0
Picus viridis149.24.1172188.54.34140.1976.252.132176.0
Pluvialis apricaria187.22.0485183.21.88340.2158.712.152180.0
Pluvialis squatarola184.72.1573174.22.50010.217  196.0
Prunella modularis18.40.642219.90.83130.0451.120.22120.3
Pyrrhula pyrrhula26.30.880526.80.95490.0190.740.33623.7
Rallus aquaticus89.71.5801128.51.73320.1063.950.826124.0
Regulus regulus5.40.38596.00.34350.0020.280.0765.6
Scolopax rusticola324.12.4009323.82.42570.1839.052.946441.0
Sitta europaea21.71.015222.61.11850.0320.74 22.6
Spreo superbus44.51.795631.61.76520.0482.75 48.1
Strix aluco484.711.3055395.89.43050.44210.792.094458.0
Sturnus vulgaris70.41.656873.31.77860.1183.810.71069.4
Sylvia atricapilla21.40.692418.50.71550.0441.150.22819.6
Sylvia borin14.80.579219.30.68650.0231.060.22018.4
Sylvia curruca14.10.549113.30.57450.0290.790.14612.8
Tringa nebularia149.71.6077153.51.62840.207  151.0
Tringa totanus102.71.295696.71.33840.2432.25 102.0
Troglodytes troglodytes8.80.501810.10.52520.0480.500.11410.4
Turdus iliacus66.41.520466.91.48950.0573.420.71067.3
Turdus merula99.32.0433999.12.014580.2374.980.84698.6
Turdus philomelos72.21.6711667.51.645230.1533.460.67369.1
Turdus viscivorus108.12.1654103.62.31340.2964.26 101.0
Tyto alba289.85.89020236.46.169120.2816.241.755308.0
Uria aalge843.54.4202608.54.54830.34727.285.258735.0

Ancillary