SEARCH

SEARCH BY CITATION

Keywords:

  • birds;
  • haematozoa;
  • Haemoproteus;
  • Leucocytozoon;
  • life history;
  • parasite prevalence;
  • parasite richness;
  • Plasmodium;
  • Trypanosoma

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Identifying host traits associated with the number of different parasite species or strains harboured by a particular host species can have important implications for understanding the impact of parasitism on hosts. We investigated associations between host ecology and life history, and parasite richness and prevalence of the four major avian blood parasite genera. We used an extensive data on blood parasite infections and host ecology in 263 bird species from the Western Palearctic, combining species-specific data with a comparative approach to control for similarity in phenotype among host species due to the effects of common phylogenetic descent. Adult survival rate negatively correlated with the number of parasite species infecting a host species when controlling for similarity due to common descent and body mass. In addition, the prevalence of Haemoproteus, Plasmodium and Leucocytozoon was higher in species harbouring a richer parasite assemblage. These results suggest that the impact on host fitness caused by avian haematozoa may be underestimated in natural populations if the exacerbated virulence associated with exposure to multiple parasites is not taken into account.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Parasitic infections exert a strong selection pressure on their hosts, and are responsible for the coevolutionary dynamics of host defensive mechanisms and parasite counter-adaptations (Dieckmann, 2002). To fully understand the impact of parasitism, we cannot overlook the fact that many (if not all) host species are simultaneously infected by several species or strains of parasites (e.g. Richie, 1988; Bruce et al., 2000; Poulin & Morand, 2000). Assemblages of parasites on hosts can be studied at different levels (Poulin, 1995; Poulin & Morand, 2000), and we need to consider this multidimensional component of host–parasite interactions. At the species or population level, host ecological traits associated with exposure to multiple vectors or to multiply infected vectors can result in increased parasite species richness harboured by the host (Poulin, 1995; Morand et al., 2000; Paul et al., 2002). However, at the individual level, the composition of the parasite community may be determined by variation in the selection pressure imposed by host immunity on different parasite taxa, or by within-host competition among parasites for limiting host resources (Bruce et al., 2000; Mayxay et al., 2004). Coinfections (either by a mix of related parasite species or by conspecifics) are a major cause of evolution of virulence, and the selection pressure imposed by parasites on hosts is expected to be stronger when several parasites exploit the same host (Herre, 1993; Frank, 1996; Mosquera & Adler, 1998). Thus, understanding the factors that determine the diversity of parasites harboured by a particular host species may help elucidate the ecological and evolutionary implications of host–parasite interactions.

Avian blood parasites are a model system in empirical and theoretical studies of host–parasite interactions (Hamilton & Zuk, 1982; Loye & Zuk, 1991; Clayton & Moore, 1997). These parasites are widely distributed among most host species (Peirce, 1981; Bennet et al., 1982). The dynamics of infection by avian blood parasites and their impact on host fitness have been described in both domestic and wild birds (Atkinson & Van Ripper, 1991; Valkiūnas, 2005). The prevalence of avian haematozoa (i.e. the proportion of individuals from a host species or population infected by a particular parasite) has recently been shown to be species specific (Deviche et al., 2001; Ricklefs et al., 2005) and previous studies have shown that interspecific variation in parasite prevalence is associated with host ecology (e.g. Ricklefs, 1992; Møller, 1997; Tella et al., 1999; Scheuerlein & Ricklefs, 2004). In addition, microscopy and molecular techniques have revealed that coinfections by different blood parasite lineages or genera occur commonly in the same host individual (e.g. Pérez-Tris & Bensch, 2005; Marzal et al., 2008). Therefore, avian haematozoa represent a suitable study system for an analysis of the association between parasite species richness and prevalence and host life history and ecology.

Avian haematozoa are transmitted by arthropod vectors (Atkinson & Van Ripper, 1991), and ecological factors associated with vector abundance were assumed to explain differences in the prevalence of parasite species irrespective of host (e.g. Greiner et al., 1975; Garvin & Remsen, 1997; Piersma, 1997; Tella et al., 1999). However, interspecific variation in parasite prevalence and species richness also depends on host defensive mechanisms and coevolution between hosts and parasites (Frank, 2002; Gandon et al., 2002; Zuk & Stoehr, 2002).

If exposure was an important determinant of parasite prevalence and species richness in different host species, we could make the following predictions: (1) cavities provide protection against predation and inclement weather, resulting in favourable sites for breeding and roosting birds. Competition for this limiting resource results in the re-use of cavities by conspecifics and heterospecifics, thereby facilitating exposure to vectors and parasites (Møller & Erritzøe, 1996). Thus, we predicted that parasite prevalence and species richness should be associated with nest site. (2) Social aggregations attract vectors or facilitate parasite transmission (Møller et al., 1993, 2001; Møller & Erritzøe, 1996). Thus, we predicted that coloniality would be associated with higher parasite prevalence and species richness. (3) Migratory birds live in two different areas during their annual cycle, and they are therefore exposed to a more diverse range of parasite species than residents that remain in a single area throughout the year (Møller & Erritzøe, 1998). Consistently, migratory birds from the temperate zone have been shown to harbour blood parasites that they must have picked up in their tropical winter quarters (Waldenström et al., 2002; Ricklefs et al., 2005). Furthermore, some long-distant migrants winter in tropical regions, where pathogen species richness is considerably higher than in temperate zones (e.g. Guernier et al., 2004). Thus, we predicted that parasite prevalence and parasite richness would be higher in migratory than in resident birds.

However, the degree of exposure to vectors is not the only determinant of blood parasite prevalence (Bradley, 1972; Clayton & Moore, 1997). Parasite virulence and host resistance are theoretically expected to coevolve so a certain level of host defence is matched by an equivalent level of parasite offence (Van Baalen, 1998). The existence of such pairs of coevolutionary stable strategies are supported by the observation that host species invest more in immunity when parasites cause greater fitness loss in their hosts (Martin et al., 2001; Møller et al., 2001; Møller & Erritzøe, 2002). Investment in anti-parasite mechanisms (e.g. immune defence) should therefore be adjusted to the level of parasite pressure and host residual reproductive value (i.e. current and future reproductive success; Perrin & Christe, 1996; Møller, 1997). Species with high residual reproductive value should invest relatively more in their own immune defence, because an effective immune defence will promote survival and, therefore, future reproduction. Thus, we predicted that host traits with strong influence on factors that contribute to residual reproductive value, such as adult survival rate and fecundity, would predict variation in parasite prevalence and the number of parasite species harboured. In particular, (4) host species with high survival prospects (i.e. high adult survival rate) should have lower parasite prevalence and reduced species richness of blood parasites despite their longer duration of exposure to parasites and their vectors, if parasitism was affected by host immunity. (5) High fecundity will compromise future survival and reproduction because immune defence is traded against current parental effort (e.g. Moreno et al., 1999; Råberg et al., 2000; Bonneaud et al., 2003). Due to the trade-offs between immune defence and reproductive effort (Lochmiller & Deerenberg, 2000), we predicted that host species with high reproductive investment would have a higher prevalence of parasites. Thus, species with a large maximum number of broods, large clutches and early age at first reproduction for their body size should have a higher prevalence of parasites and thus suffer disproportionately from multiple infections.

Finally, host intrinsic features such as developmental rate, may explain variation in parasite prevalence through their effects on development of immune function or time-dependent exposure to vectors at the nest. In particular, (6) haematozoan prevalence appears to be inversely related to relative duration of embryonic development among species of birds (Ricklefs, 1992; Tella et al., 1999). Thus, we predicted that species with relatively long incubation periods would have lower parasite prevalence and reduced number of parasite species. However, (7) the duration of post-hatching development may also be associated with increased exposure to vectors at the nest and result in a positive association between duration of the nestling period and parasite prevalence and species richness.

The general aim of this study was to determine to what extent ecological, life-history and intrinsic host factors predicted interspecific variation in parasite species richness and parasite prevalence. In addition, we examined associations between infections by the main taxa of avian blood parasites in relation to the number of parasite species recorded for the host species. We tested these predictions using an extensive data set on blood parasite infections and host phenotype in bird species from the Western Palearctic. We combined species-specific analyses with a comparative approach to account for similarity in parasitism and ecological or life-history traits due to common descent. Finally, we considered the effect of heterogeneity in host sampling effort, because the number of host individuals screened for blood parasite infections varied enormously among species, potentially affecting estimates of parasite prevalence and species richness.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Parasite data

We used published information on the prevalence of the four most common genera of avian blood parasites (Haemoproteus, Plasmodium, Leucocytozoon and Trypanosoma) in bird species from the Western Palearctic based on microscopic examination of blood smears. We also extracted from the literature information on the number of blood parasite species reported for each host species to calculate parasite species richness. Therefore, our analyses of parasite prevalence referred to blood parasite genera, whereas the analyses of parasite species richness referred to the number of parasite species. This information was extracted from Peirce (1981) and Scheuerlein & Ricklefs (2004) combined with information from several sources listed in Møller & Nielsen (2007). Although molecular techniques are better at detecting weak infections, several studies have shown a positive association between estimates of parasite prevalence using both microscopic and molecular techniques (e.g. Waldenström et al., 2004; Ricklefs et al., 2005). Our own analyses of the data provided in the supplementary material in Ricklefs et al. (2005) showed positive consistency among the two estimates of parasite prevalence (Kendall tau = 0.307, z = 6.252, P < 0.001), although analyses of blood smears only revealed 28% of what was found with PCR. Thus, we believe that our analyses based only on microscopy are conservative. In total, these analyses were based on infection levels of 29 799 adult hosts belonging to 263 species. Finally, we extracted information on the number of individuals examined for each host species to control for the potentially confounding effect of sampling effort in the analyses.

Ecological variables

Hole nesting and coloniality. We obtained information on open vs. cavity nesting from Cramp and Perrins (1977–1994), and scored a species as hole nesting if it used holes as nest sites or open nester otherwise. Likewise, we obtained information on breeding sociality from Cramp and Perrins (1977–1994) and scored a species as colonial (1) if nesting in clusters or solitary (0) otherwise.

Migration. We scored all host species as either residents (0, all populations having the same range during breeding and nonbreeding) or otherwise as migrants (1), based on the information in Cramp and Perrins (1977–1994).

Life-history variables

We recorded information on clutch size, maximum number of broods per year, minimum age at first breeding (years) and adult survival rate (%) from Cramp and Perrins (1977–1994). The duration of incubation and nestling periods (days) were also obtained from the same source. We included body mass extracted from Cramp and Perrins (1977–1994) and Dunning (1993) as an additional predictor variable in the analyses because some life-history traits scale to body size, but still show considerable residual variation. If only a range of values was reported, we used the mean of the reported minimum and maximum values. If multiple estimates were reported, we used the one with the largest sample size. The entire data set is reported in Appendix S1. Host life-history traits for one domesticated species (Phasianus colchinus) were not included in the analyses.

Statistical analyses

We used Akaike’s information criterion (AIC) and general linear models to find the best-fit model explaining variation in our dependent variables (Burnham & Anderson, 2002). We computed differences in AIC (ΔAIC values) between the best-fit model and alternative models that excluded a particular variable, to allow comparison between candidate models and interpret the relative importance of variables that entered the final models. Models with ΔAIC > 10 are considered to fail in explaining substantial explainable variation in the data (Burnham & Anderson, 2002).

To obtain variables that would allow the use of parametric statistics, parasite prevalence and host adult survival rate were square-root arcsine-transformed before analyses. The number of blood parasite species, body mass, age at first reproduction, clutch size and incubation and nestling periods were log10 transformed. We used Pearson’s product moment correlations to examine associations among prevalence of different parasite genera.

In our data set, the number of individuals of each host species examined for blood parasites ranged from 1 to 1977. Because sampling effort for avian blood parasites differed enormously among host species, estimates of parasite prevalence varied in reliability (Jovani & Tella, 2006). In addition, parasite species richness positively correlated with the number of individuals sampled for blood parasites (Pearson’s r = 0.610, P < 0.001, n = 263). Thus, we weighted all analyses by sample size, to account for the possible effect of heterogeneity in sampling effort, but simultaneously maximize statistical power and the probability of finding a pattern in phylogenetic analyses without discarding data (see Møller & Nielsen, 2007). The weighted analyses allowed us to adjust the contribution of each host species to the outcome of the statistical analysis by weighting the parasite prevalence and richness in proportion to the number of individuals sampled. To avoid over-representation of taxa with large sample size, we used the log10-transformed number of individuals examined for blood parasites as the weighting factor. We used the software statistica (StatSoft©, Tulsa, OK, USA) for all analyses. Weighted mean and standard error are reported throughout.

Comparative analyses

We constructed a composite phylogeny of the bird species used in this study. The topology of the phylogenetic tree and literature used for construction of the phylogeny are shown in Appendix S2. As we relied on many studies that differed in methods used to construct a phylogeny, we had no consistent estimates of branch lengths, and, therefore, we adopted uniform branch lengths for the comparative analyses.

We used the phylogenetic hypothesis of the relationship among species to calculate standardized independent linear contrasts, thereby controlling for the effect of common ancestry on similarity in phenotype in our analyses (Felsenstein, 1985). We used the software CAIC for these analyses (see Purvis & Rambaut, 1995 for a detailed description of the method). Analyses of standardized contrasts can only handle one dichotomous variable at a time (Purvis & Rambaut, 1995). Thus, despite the fact that migration, coloniality and hole nestling were originally defined as categorical variables, we used them in the analyses as continuous variables. Intermediate states of these variables are biologically meaningful, and using them as continuous variables allowed simultaneously testing for the effects of all possible predictor variables and confounding factors. All regressions of standardized linear contrasts were forced through the origin because the comparative analyses assume that there has been no evolutionary change in the dependent variable when the predictor variable has not changed (Purvis & Rambaut, 1995). Analyses of independent contrasts had body mass as the main predictor because we were interested in testing hypotheses concerning residual variation in life history independent of body mass. Thus, we forced body mass into the models of parasite prevalence based on independent contrasts by choosing the model including body mass with the lowest AIC value. Furthermore, all analyses were weighted by the log10-transformed number of individuals examined for blood parasites as explained by Møller & Nielsen (2007). In brief, we calculated weights for each contrast by calculating the mean sample size for the taxa immediately subtended by a given node.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

The number of haematozoan taxa infecting different avian host species ranged from 0 to 12, with arithmetic mean 1.969, SE 0.128, n = 263 host species (see Peirce, 1981; Scheuerlein & Ricklefs, 2004 for information on parasite taxa). The most abundant blood parasite genus was Haemoproteus, detected in 50.57% of the host species sampled, with mean prevalence 0.114, SE 0.011 (n = 263). The genus Leucocytozoon was the second most abundant parasite, present in 46.39% of the host species, with mean prevalence 0.102, SE 0.011 (n = 263). Blood parasites from the genus Trypanosoma were present in 36.15% of the host species, with mean prevalence 0.044, SE 0.007 (n = 260) and Plasmodium only in 23.19% of the host species with mean prevalence 0.030, SE 0.006 (n = 263). The prevalence of Trypanosoma and Leucocytozoon significantly positively correlated (Pearson r = 0.53). All other parasite prevalence had weak Pearson correlation coefficients ranging from 0.10 to 0.27.

The best-fit model of the species-specific relationship between the number of blood parasite species and host features included body mass, adult survival rate, incubation period, coloniality, migratory habit and age at first reproduction as explanatory variables (Table 1a). However, only the effects of survival rate and development rate reached statistical significance when controlling for similarity among taxa due to common descent (Table 1b). Host species harbouring a larger number of blood parasite species had greatly reduced adult survival rate independent of their body size (Fig. 1). Parasite species richness was negatively associated with the duration of the incubation period, but weakly positively associated with the duration of the nestling period (Table 1b).

Table 1.   Best-fit models of the relationship between the number of blood parasite species and host life-history and ecological variables in Palearctic birds.
VariableSlope (SE)SSd.f.FPΔAIC
  1. The models for species weighted by sample size had the statistics F6,134 = 16.80, < 0.001, inline image = 0.41, AIC = −0.72, whereas the model for independent contrasts had the statistics F4,130 = 3.69, = 0.007, inline image = 0.09, AIC = −197.732. ΔAIC is the difference in AIC between the model presented and a model excluding a given factor.

(a) Species
 Body mass0.380 (0.105)1.225112.99< 0.00111.64
 Survival rate−0.327 (0.115)0.75918.050.0054.99
 Incubation period−0.355 (0.120)0.82218.720.0046.69
 Colonial−0.120 (0.074)0.25112.660.1050.45
 Migrant0.128 (0.067)0.33513.550.0611.80
 Age at first   reproduction−0.305 (0.120)0.60516.420.0125.32
 Error 12.639134   
(b) Contrasts
 Body mass−0.022 (0.096)0.00110.050.816−1.58
 Survival rate−0.234 (0.091)0.15516.620.0114.95
 Incubation period−0.218 (0.102)0.10814.600.0342.99
 Nestling period0.171 (0.105)0.06212.650.1061.05
 Error 3.054130   
image

Figure 1.  Relationship between the number of blood parasite species infecting different species of birds of the Western Palearctic and adult survival rate. The line is the regression line based on weighted analyses (see Methods) using (a) species-specific data and (b) independent contrasts.

Download figure to PowerPoint

To study possible determinants of the prevalence of different parasite species, we analysed interspecific variation in the prevalence of Haemoproteus, Leucocytozoon, Plasmodium and Trypanosoma, while considering all other blood parasite infections reported for the host species as potential predictors. The number of different haematozoan parasite species harboured by a host species was a significant predictor of the prevalence of Haemoproteus, Leucocytozoon and Plasmodium (Tables 2 and 3). The prevalence of these three parasite genera was higher in species attacked by a more diverse parasite fauna (Fig. 2). The prevalence of Haemoproteus was predicted by coloniality, migratory habits, maximum number of broods per breeding season and prevalence of Leucocytozoon, when controlling for similarity among species due to common descent (Tables 2 and 3).

Table 2.   Best-fit models of the relationship between parasite prevalence and host life-history and ecological variables in Palearctic birds.
VariableSlope (SE)SSd.f.FPΔAIC
  1. Models weighted by sample size for different parasite genera: Haemoproteus: F6,134 = 10.66, inline image = 0.31, < 0.001, AIC = −12.707; Leucocytozoon: F5,252 = 50.37, inline image = 0.42, < 0.001, AIC = −73.48; Plasmodium: F2,257 = 22.01, inline image = 0.14, P < 0.001, AIC = −130.583; Trypanosoma: F2,256 = 55.57, inline image = 0.30, P < 0.001, AIC = −179.349. ΔAIC is the difference in AIC between the best-fit model and the model excluding a given factor.

(a) Haemoproteus
 Body mass0.174 (0.101)0.24312.940.0880.52
 Survival rate0.252 (0.127)0.32513.950.0492.18
 No. of parasite   species0.538 (0.094)2.694132.70< 0.00124.63
 Leucocytozoon−0.211 (0.092)0.42915.200.0242.94
 Migrant0.176 (0.079)0.40714.940.0283.25
 Age at first   reproduction−0.333 (0.127)0.56216.820.0104.53
 Error 11.042134   
(b) Leucocytozoon
 Body mass0.339 (0.047)3.099151.78< 0.00145.16
 Colonial0.195 (0.046)1.052117.57< 0.00113.93
 Migrant0.132 (0.046)0.48718.150.0055.52
 No. of parasite   species0.364 (0.049)3.282154.84< 0.00138.97
 Trypanosoma0.350 (0.049)3.021150.48< 0.00156.34
 Error 15.083252   
(c) Plasmodium
 No. of parasite   species0.322 (0.062)1.145127.24< 0.00118.79
 Trypanosoma0.121 (0.062)0.16213.870.0502.84
 Error 10.80257   
(d) Trypanosoma
 Body mass−0.170 (0.053)0.390110.190.00110.30
 Leucocytozoon0.558 (0.053)4.2091109.79< 0.00196.22
 Error 9.813256   
Table 3.   Best-fit models of the relationship between parasite prevalence and host life-history and ecological variables in Palearctic birds based on independent contrasts.
VariableSlope (SE)SSd.f.FPΔAIC
  1. Models weighted by sample size for Haemoproteus: F6,128 = 5.009, inline image = 0.17, < 0.001, AIC = −175.914; Leucocytozoon: F8,126 = 19.86, inline image = 0.54, < 0.001, AIC = −311.991; Plasmodium: F4,130 = 4.77, inline image = 0.11, P = 0.0013, AIC = −346.416; Trypanosoma: F5,129 = 16.97, inline image = 0.38, P < 0.001, AIC = −395.129.

(a) Haemoproteus
 Body mass0.088 (0.081)0.03211.200.275−0.75
 No. of parasite   species0.482 (0.092)0.735127.22< 0.00123.64
 Leucocytozoon−0.314 (0.101)0.25819.570.0027.59
  Max. number of  broods−0.216 (0.093)0.14415.340.0223.44
 Colonial−0.127 (0.084)0.06312.320.1300.39
 Migrant−0.158 (0.090)0.08313.060.0831.14
 Error 3.46128   
(b) Leucocytozoon
 Body mass0.149 (0.067)0.04714.930.0283.11
 No. of parasite   species0.410 (0.067)0.361137.72< 0.001> 10
 Haemoproteus−0.172 (0.064)0.06917.170.0085.37
 Trypanosoma0.379 (0.067)0.305131.89< 0.001> 10
 Max. of number   of broods−0.314 (0.069)0.195120.36< 0.001> 10
 Nestling period−0.177 (0.073)0.05615.820.0174.01
 Colonial−0.155 (0.064)0.05515.780.0183.96
 Migrant −0.163 (0.069)0.05315.530.0203.72
 Error 1.206126   
(c)Plasmodium
 Body mass−0.154 (0.092)0.02112.770.100.81
 No. of parasite   species0.292 (0.082)0.095112.57< 0.0110.27
 Nestling period0.207 (0.097)0.03414.530.0352.56
 Colonial−0.130 (0.087)0.01712.230.1370.27
 Error 0.990130   
(d) Trypanosoma
 Body mass−0.036 (0.084)0.00010.1890.665−1.81
 Leucocytozoon0.555 (0.069)0.333163.60< 0.00151.32
 Nestling period0.179 (0.077)0.02815.370.0223.48
 Migrant−0.147 (0.070)0.02314.430.0372.48
 Age at first   reproduction−0.163 (0.079)0.02214.230.0422.38
 Error 0.676129   
image

Figure 2.  Relationship between parasite prevalence and number of blood parasite species infecting different species of birds of the Western Palearctic. The lines are regression lines based on weighted analyses (see Methods) for (a–c) species-specific data and (d–f) independent contrasts.

Download figure to PowerPoint

The best-fit model for the prevalence of Leucocytozoon accounted for 54% of the variance, and included body mass, number of blood parasite species, prevalence of Haemoproteus and Trypanosoma, maximum number of broods, nestling period, coloniality and migration as explanatory variables. All these variables remained significant in an analysis based on independent contrasts (Tables 2 and 3).

The model for Plasmodium prevalence based on independent contrasts included the number of parasite species, duration of the nestling period and coloniality as explanatory variables (Table 3). The prevalence of Trypanosoma was mainly predicted by the prevalence of Leucocytozoon, although nestling period, migration and minimum age at first reproduction also entered a model based on independent contrasts (Tables 2 and 3).

The results of our study do not seem to be biased by heterogeneity in sampling effort among host species. Additional analyses including sampling effort as an explanatory factor, using a threshold criterion of a minimum sample size of 20 individuals per species, or not considering the issue of sample size confirmed the main results of the study (see Appendix S3 for a summary of these results).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We analysed an extensive data set on parasite prevalence of the four most common avian blood parasite genera and overall parasite species richness, together with ecological and life-history variables of 263 species of Western Palearctic birds. Parasite species richness was an important predictor of interspecific variation in the prevalence of Haemoproteus, Leucocytozoon and Plasmodium, which may have major implications for understanding the impact caused by these blood parasites on their avian hosts.

We found a negative association between adult survival rate and blood parasite species richness when controlling for host body mass and similarity due to phylogenetic relationships among hosts. Previous studies have shown that probability of survival in birds is linked to immunity, the expression of secondary sexual characters and morphological symmetry (Møller & Saino, 2004 and references therein). Our study shows that interspecific variation in survival is significantly associated with species richness of blood parasites, which suggests that there may be a fitness advantage associated with the ability to limit the diversity of parasites that attack a particular host species.

Several previous studies have aimed at understanding the link between host traits and parasite diversity in a comparative approach (e.g. Morand & Harvey, 2000; Nunn et al., 2003; Ezenwa et al., 2006). Morand & Harvey (2000) found a negative relationship between host longevity and richness of parasitic helminths in a study on 23 species of mammals. Likewise, in a comparative study of parasitic infections of artio- and perissodactyl mammals, Ezenwa et al. (2006) found a negative correlation between host longevity and parasite richness of helminths and microparasites. By contrast, longevity has been shown to correlate positively with the diversity of helminths in freshwater fish (Morand, 2000) and with a broad estimate of parasite richness in primates (Nunn et al., 2003). These conflicting results and the negative correlation found in our study seem to challenge prevailing hypotheses based on simple host exposure, because long-lived host species should have been exposed to parasites and vectors for a longer time than short-lived species and therefore accumulate more parasites (reviewed in Poulin & Morand, 2000). By contrast, life-history theory predicts that host species with higher survival rate are expected to invest heavily in immunity (Lochmiller & Deerenberg, 2000). Thus, greater investment in immunity in species with high adult survival prospects may explain negative associations between survival rate and parasite richness.

Multiple infections (i.e. the ability of a parasite to infect an already infected host) can arise from hosts being exposed to a diverse vector fauna, or by several parasite strains or species being transmitted by the same vector species, as in avian haematozoa (e.g. Greiner et al., 1975; Atkinson & Van Ripper, 1991). In host species exposed to a large number of blood parasite species, there is higher probability that several parasite species or strains co-infect the same individual host. Thus, a potential proximate mechanism underlying the observed association between adult survival rate and number of blood parasite species is that the increased virulence associated with multiple infections (Herre, 1993; Gandon et al., 2002) results in higher morbidity in hosts harbouring several parasite strains. The exacerbated impact of multiple infections by avian blood parasites has recently been shown to reduce the probability of recapture of birds suffering from double infections compared with singly infected or uninfected individuals (Davidar & Morton, 2006; Marzal et al., 2008).

Host species exposed to several parasite strains or species are expected to have greater difficulty in controlling infections and therefore have higher parasite prevalence. Thus, it is likely that parasite taxa such as Haemoproteus and Plasmodium, generally considered low pathogenic based on their physiological effects on organs and tissues of the host (Atkinson & Van Ripper, 1991; Valkiūnas, 2005), occur in multispecies infections and have a stronger impact on host fitness than traditionally assumed. Møller & Nielsen (2007) have recently shown a positive association between prevalence of Haemoproteus, Plasmodium and Leucocytozoon and vulnerability to predation by two species of avian predators, suggesting a major impact of these parasites on the fitness of avian hosts through a mechanism different from direct pathogenicity of blood parasites.

Parasites of the genus Trypanosoma are transmitted by a variety of vectors, including simuliids, ceratopogonids and culicids, also known to transmit other avian blood parasites (Greiner et al., 1975). Therefore, parasite–vector specificity or vector–host specificity does not seem to restrict the access of this group of parasites to potential hosts and explains the lack of association between parasite species richness and prevalence of Trypanosoma. An alternative explanation could be that infection by Trypanosoma may elicit host immune responses that prevent infection by other blood parasites (i.e. concomitant immunity), as shown in certain infections by this group of parasites in domestic mammals (Read & Taylor, 2001). However, more studies are needed to elucidate the role of vectors in filtering or facilitating the diversity of parasite assemblages in birds (see also Gager et al., 2008; Hellgren et al., 2008).

Migration by avian hosts entered as a significant predictor of prevalence of Haemoproteus, Leucocytozoon and Trypanosoma, with migratory species having lower prevalence than resident species. Long-distance migration incurs considerable costs for hosts and is associated with increased parasitism (Møller & Erritzøe, 1998; Alerstam et al., 2003). However, our results seem to contradict predictions based on host exposure, but support the hypothesis that the selection experienced by migratory birds in their breeding and wintering areas has resulted in greater investment in immune defence (Møller & Erritzøe, 1998). Thus, the overall impact of parasitism on hosts, estimated through the prevalence of infection, could be reduced in migratory compared with resident host species despite greater diversity of parasites encountered by long-distance migratory birds. Alternatively, migrants may die during migration due to the deleterious effects of infections. Thus, mortality during migration may leave the impression that migrants have lower parasite prevalence than residents, when in fact they may have higher prevalence if sampled before migration.

Social living is associated with increased risk of parasitism through an increased rate of horizontal parasite transmission and a higher frequency of multiple infections in hosts (reviewed by Møller et al., 1993). Thus, coloniality has been associated with higher prevalence of blood parasites across bird species (Tella, 2002; but see Tella et al., 1999). However, the strong selection pressure imposed by parasitism on social host species is predicted to have selected for greater investment in immune defence (Møller et al., 2001). In accordance with this idea, colonial bird species have relatively large immune defence organs (Møller & Erritzøe, 1996), and stronger T- and B-cell immune responses than solitary species (Møller et al., 2001). Our results showed that among 263 bird species, colonial species had reduced prevalence of Haemoproteus, Leucocytozoon and Plasmodium, consistent with the hypothesis of higher investment in immunity associated with sociality (Møller et al., 2001).

Parasite richness weakly negatively correlated with the duration of the incubation period after controlling for body size. This result is consistent with the hypothesis proposed by Ricklefs (1992) of a trade-off between embryonic development and ontogeny of an efficient immune system. However, the relative duration of post-hatching development was positively associated with the prevalence of Trypanosoma and Plasmodium, suggesting that time-dependent exposure to vectors at the nest is an additional determinant of parasite prevalence.

In conclusion, we found a link between the number of blood parasite species of avian hosts and adult survival rate across a sample of 263 bird species from the Western Palearctic, when controlling for a number of potentially confounding variables and similarity in phenotype among taxa due to common phylogenetic descent. Due to the correlational nature of our analyses, we do not invoke causality in the relationships discussed here. However, the role of parasite species richness in predicting variation in the prevalence of blood parasites among host taxa suggests that the number of species or strains of blood parasites that comprise the parasite burden of a host species may play a crucial role in parasite-mediated natural selection. Although some blood parasites may have minor impact on parasite-mediated mortality, the selection pressure imposed by parasitism may be underestimated in natural host populations if the increased virulence associated with exposure to multiple parasites is not considered.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

EA was supported by Centre National de la Recherche Scientifique, France. We thank Javier Pérez-Tris and Simon Fellous for constructive comments on an earlier draft. The comments by anonymous reviewers helped to improve the manuscript.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Appendix S1 Data set used in the study.

Appendix S2 Phylogenetic relationships among species of Palearctic birds in the study of blood parasite infections.

Appendix S3 A figure showing the correlation between parasite richness and the number ofindividuals sampled in each host species, and two tables showing a summary ofthe additional statistical analyses performed to examine whether heterogeneityin sampling effort could influence the main results of our study.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
JEB_1613_sm_suppl.doc1074KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.