Fitness consequences of variation in natural antibodies and complement in the Barn Swallow Hirundo rustica



    Corresponding author
      †Author to whom correspondence should be addressed. E-mail:
    Search for more papers by this author

    1. 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
    Search for more papers by this author

†Author to whom correspondence should be addressed. E-mail:


  • 1The immune system and its component parts have evolved and are maintained due to the fitness benefits that hosts acquire in terms of parasite resistance. However, there are relatively few studies of the fitness consequences of natural variation in immunity in free-living populations, mainly due to the complexity of the immune system and logistic problems of assessing immunity and fitness components.
  • 2We quantified two aspects of humoral immunity in a population of Barn Swallows Hirundo rustica, natural antibodies and complement that constitute the first line of defence against parasites.
  • 3The magnitude of both immune components increased during the breeding season.
  • 4Very old adults had reduced levels of natural antibodies, providing evidence consistent with immuno-senescence, while that was not the case for lysis. However, there was no evidence that survivors from one year to the next had stronger agglutination or lysis than nonsurvivors.
  • 5Females with a second clutch had higher levels of natural antibodies than females that only had a single clutch, while males showed no differences. Annual fecundity of females decreased with natural antibody levels, that was not evident in males. Therefore, in females, natural antibody and complement levels are important predictors of vital rates.
  • 6Tail length of male partners, but not of females, explained the variation in female natural antibodies and complement, while there was no similar effect in males. This pattern may have arisen from differential parental investment by females mated to the most attractive males.
  • 7The abundance of four different ectoparasites was not predicted by either immune component.


Parasites constitute an important component of the environment of all living beings. Not surprisingly, parasites exert strong selection pressures on their hosts through the exploitation of resources that could otherwise have been used by the host, but also indirectly through pathology and activation of the immune system (e.g. Noble & Noble 1976; Cox 1982; Combes 2001). While the parasitic effect on host phenotype (including their secondary sexual characters and their life history) is increasingly appreciated (e.g. Møller, Christe & Lux 1999; Martin et al. 2001; Tella, Scheuerlein & Ricklefs 2002), we still have a poor understanding of this effect. Hosts are not passive bystanders to parasite challenge, and a range of behavioural and physiological responses have evolved to enable hosts to avoid, cope with and modify the effects of parasitism.

The most important defence system against parasites is the immune system and its component parts. The immune system has the ability to recognize foreign antigens and produce defences that are either innate or acquired to such antigenic challenge. The immune system of vertebrates constitutes a number of different components that partly interact and complement each other in providing a broad range of defence mechanisms against many different parasites. The complexity of the immune system complicates the assessment of the overall effectiveness of defence ability, particularly if different components seem to be independent, as shown by Matson et al. (2006). To add to this difficulty, few studies of different components of the immune system have been undertaken under field conditions that are most likely to realistically reflect the environment under which this system has evolved, and also the environment in which the costs of immunity (energetically, autoimmunologically, etc.) are likely to be significant. This is mainly caused by logistic difficulties coupled with difficulties of obtaining large sample sizes required for reliable tests. Hence, our current knowledge of the ecological and evolutionary consequences of intraspecific and interspecific differences in the magnitude of the immune responses, is relatively poor.

Immune responses can either be specific, nonspecific, acquired or innate (Roitt, Male & Brostoff 1996; Pastoret et al. 1998). Constitutive innate immunity represents the first line of defence against parasite attack, with the two components being natural antibodies (NAbs) and complement. The function of these two components is to recognize and initiate the complement enzyme cascade (NAbs) that eventually ends in cell lysis (Carroll & Prodeus 1998). NAbs and complement are connected to adaptive immunity, providing a link between the innate and the acquired arms of defence (Carroll & Prodeus 1998; Thornton, Vetvicka & Ross 1994: Ochsenbein & Zinkernagel 2000). The function of complement is best known through the effects of deficiency syndromes in human disease (e.g. Schur 1983) and animal models (Barta & Hubbert 1978; Ellis, Arp & Lamont 1989; Kapil, Das & Kumar 1989; Reis 1989; Vestey & Lochmiller 1994; Olaho-Mukani et al. 1995; Barman et al. 1997; Parmentier et al. 2002). NAbs occur in immunologically naïve animals and thus do not require prior antigen exposure (Pereira et al. 1986; Ochsenbein & Zinkernagel 2000; Belperron & Bockenstedt 2001), being directly encoded in nuclear DNA (Naparstek et al. 1986; Avrameas 1991; Belperron & Bockenstedt 2001). NAbs have been shown to provide resistance to malarial parasites (Congdon et al. 1969); correlate with the abundance of chewing lice (Whiteman et al. 2006); and naturally occurring concentrations can kill bacteria and spirochetes in vivo (Belperron & Bockenstedt 2001; Ochsenbein et al. 1999), as well as clear lipopolysaccharides in vitro (Reid et al. 1997). In chickens, NAbs respond to selection on other immune components (Parmentier et al. 2004), and are generally believed to be relatively insensitive to short-term changes in environmental conditions (Matson, Ricklefs & Klasing 2005). However, the fitness consequences of natural variation in NAbs and complement in free-living animals, are virtually unknown.

The objective of this study was to assess the fitness consequences of intraspecific variation in two components of humoral immunity, namely natural antibody production (as reflected by agglutination) and complement (as reflected by lysis). Matson et al. (2005) have shown that natural antibody production and complement are only weakly positively correlated, implying that they represent independent measures of immunity. The relationship between natural antibody production and complement, which are important components of immunity (Roitt et al. 1996; Pastoret et al. 1998), and fitness components such as reproductive success and survival prospects, has not yet been assessed under field conditions. While several components of immunity, reliably predict survival prospects (review in Møller & Saino 2004) and mating success (Møller et al. 1999), our current level of knowledge is poor. Consequently, we assessed the associations between natural antibody production and complement, prior to the start of the breeding season in a population of Barn Swallows Hirundo rustica and then related this to subsequent: (1) sexual selection; (2) fecundity; (3) survival; and (4) parasitism. More specifically, we tested: (1) whether mated males had higher levels of agglutination and lysis than unmated males; (2) whether male tail length (a secondary sexual character; Møller 1994a) was positively related to the level of agglutination and lysis; (3) whether male tail length, but not female tail length was positively correlated with agglutination and lysis in females (as would be expected if females mated to attractive males, invested differentially in reproduction); (4) whether the presence of a second clutch was more common among individual females with greater levels of agglutination and lysis (with annual fecundity mainly being determined by the number of clutches; Møller 1994a); (5) whether fecundity, independent of the number of clutches, was negatively predicted by agglutination and lysis (females were predicted to show a negative relationship because they invest disproportionately in reproduction; Møller 1994a); (6) whether the survival of adults from one year to the next had stronger agglutination and lysis than nonsurvivors; and (7) whether individuals with strong agglutination and lysis, had fewer parasites.

Materials and methods

field procedures

The field site consists of ≈ 45 km2 at Kraghede (57°12′N, 10°00′E), Denmark (see Møller (1994a) for details), with an altitude range of 9–28 m a.s.l. Although neighbouring farms are separated by distances ranging from 75 to 3000 m, breeding Barn Swallows rarely moved between farms, during the breeding season. This study population of Barn Swallows, has been the subject of a long-term investigation since 1971 (Møller 1994a). The study site is an open farmland habitat with Barn Swallows breeding in the barns and stables, and only rarely elsewhere, such as under bridges and in culverts. The areas around the farms consist of fields and scattered trees, with the exception of a few plantations, ponds, ditches and streams. The main crops are grass, barley, wheat and potatoes.

Adults were captured regularly in mist nets before the start of the breeding season and each adult in a given site given colour identification bands. The total number of individuals was 156, consisting of 74 males and 82 females of known age. Upon capture, a number of morphological variables were recorded, including the length of the two outermost tail feathers, the central tail feather, flattened wing length, keel length, tarsus length (to the nearest 0·01 mm) and body mass (to the nearest 0·1 g).

Blood was collected from adult Barn Swallows by puncturing the brachial vein and collecting two 75 µL aliquots in heparinized capillaries and stored in a cooling box at a temperature just above freezing. Within a period of 2 h, the capillaries were centrifuged for 10 min at 4000 r.p.m. The plasma and cells were separated and stored at −20 °C until analysis.

Adult plumage was carefully screened for louse flies and mites, when measured and handled. In brief, the number of louse flies was recorded while recording the 12 different morphological traits, and a search for the presence of mites of the species Ornithonyssus bursa (see Summary of details in Møller 1991, 1994a) was undertaken in the head feathers. The abundance of chewing lice was recorded indirectly from the number of holes in the wing and tail feathers (see Summary of details in Møller 1991, 1994a) and the number of feather mites was recorded in all wing and tail feathers. All these measures of parasite abundance are highly repeatable, as shown by repeat measurements of the same individuals on different days, and they have also been shown to be reliable as demonstrated by cross-validation using counts of mites in nests and the abundance of chewing lice recorded when handling birds (Møller 1994a; and unpublished data).

The mating success of males was assessed from regular observations, with males either being unmated (singing vigorously in their territory, chasing females, and being unaccompanied by a female) or mated monogamously (occupying a territory and breeding with a female; Møller 1994a). Specific adults were assigned to nests using colour bands and colour codes painted on breast feathers. We recorded laying date as the date of laying of the first egg, clutch size as the maximum number of eggs in the nest, and brood size as the number of fledglings. This was recorded for both first and second clutches (when a second clutch was found). Annual fecundity, was determined by combining the number of fledglings produced in the first and second clutches. Blood samples were analysed from seven first clutch broods, when the nestlings were 12 days old, as described above.

Adults were captured annually between 1984 and 2005, and capture probability exceeded 98% as revealed by capture–mark–recapture analyses (Møller & Szép 2002). We assumed that adults captured the first time were yearlings, given that all but one of 175 local recruits were captured the first time when 1 year old, and that adults after initially breeding in one location, generally continue to breed in the same farm. In fact, only three of 4187 adults ever moved, and even then it was only to the nearest neighbouring farm (a maximum distance 400 m). We recorded survival of adults tested for NAb and lysis in 2005 during 2006 by assuming that birds recaptured in 2006 were the only survivors. This assumption is supported by the very high recapture probability of adults in the population (Møller & Szép 2002).

haemolysis–haemagglutination assay

To estimate the levels of circulating natural antibodies and complement we used the procedure developed by Matson et al. (2005). The agglutination assay estimates the interaction between natural antibodies and antigens in rabbit blood. The lysis assay estimates the action of complement from the amount of haemoglobin released from the lysis of rabbit erythrocytes. The quantification of agglutination and lysis is determined by serial dilution (in polystyrene 96-well assay plates), and determining the point when the agglutination or lysis reaction has stopped. This was undertaken by using fresh rabbit blood (stored at 4 °C) with Alsever's anticoagulant, 96 round well assay plates and an Epson 4490 photo scanner. After determination of the level of haematocrit, we diluted the sample to obtain a solution of 1% of erythrocytes.

The protocol for haemolysis and haemagglutination was as follows. Plasma samples were thawed and homogenized using a vortex. Subsequently, 25 µL of plasma was pipetted into each well. Subsequent wells were serially diluted from a solution of 1 in 2, down to 1 in 2048. The final well contained a sample of erythrocytes, thus serving as a negative control. Subsequently 25 µL of a 1% solution of rabbit blood was added to all wells. The assay plate was then covered and shaken for 10 s, followed by incubation for 90 min at 37 °C. The assay plate was then left at an inclination of 45° at ambient temperature for 20 min. Finally, the assay plate was read by CH and scanned. Scoring was based on ‘negative’ samples having a small, round agglutinate forming a well-defined red mass, at the bottom of the wells, while ‘positive’ samples having a diffuse film at the bottom of the wells.

This was followed by incubating the assay plate at ambient temperature for 70 min, followed by assessment of lysis and scanning of the plate. Wells are assumed to be positive when there is no sign of whole blood (the action of complement had destroyed the red blood cells and only left haemoglobin).

All tests were made carried out blindly by CH with respect to the phenotype of the individual birds. Repeat tests on 17 individuals on two different days showed that individuals accounted for 73% of the variance for haemagglutination reaction (F = 3·31, d.f. = 16, 17, r2 = 0·73, P = 0·0088). Similarly, repeat tests on 17 individuals on two different days showed that individuals accounted for 82% of the variance for lysis (F = 5·36, d.f. = 16, 17, r2 = 0·82, P = 0·0005).

We also tested for individual differences in the scoring of scans of assay plates by APM assigning scores to 40 individuals. This was undertaken blindly with respect to the first series of scores by CH. Again, there was highly significant consistency in scores between the two scorers (haemagglutination: F = 406·97, d.f. = 39, 40, r2 = 0·99, P < 0·0001; lysis: F = 16·18, d.f. = 39, 40, r2 = 0·94, P < 0·0001).

statistical analyses

All analyses were made using the software JMP (2001). Repeatability analyses of agglutination and lysis scores were made using one-way analyses of variance with individual as a factor.

The possibility for immuno-senescence was investigated by determining the relationship between agglutination and lysis, respectively, and the linear and quadratic terms of age estimated in years. We used a backward stepwise procedure, with the probability of a variable entering set to 0·25, and the probability of a variable leaving set to 0·10. Forward and backward elimination procedures produced identical models in all cases.

We developed the best-fit models by including predictor variables (including date that accounted for seasonal variation in NAbs and complement) and other variables such as the linear and quadratic term of age, sex and the phenotypic variables of interest. In the analyses of effects of sexual selection we included data, age, age squared, tail length and tail length of the mate as predictor variables.

We investigated the relationship between survival and NAb level using logistic regression.

However, not all information was available for all individuals and sample sizes vary slightly among analyses.


Agglutination in adults ranged from 0 to 11 with a mean value of 3·35 (SE = 0·22), n = 156, CV = 84%, while lysis in adults only ranged from 0 to 2 with a mean value of 0·082 (SE = 0·025), n = 156, CV = 398%. Agglutination and lysis were only weakly positively related, accounting for 16% of the variance [F = 28·86, d.f. = 1, 156, r2 = 0·16, P < 0·0001, slope (SE) = 3·58 (0·67)]. The low degree of determination implies that subsequent analyses did not have any serious problems of colinearity.

Adults and nestlings differed significantly in agglutination [F = 7·29, d.f. = 1, 165, r2 = 0·04, P = 0·008, mean (SE) for nestlings = 0·63 (0·17)], but not in lysis [F = 0·14, d.f. = 1, 164, r2 = 0·00, P = 0·71, mean (SE) for nestlings = 0·13 (0·11)].

Agglutination increased significantly with sampling date [Fig. 1; F = 6·13, d.f. = 1, 154, r2 = 0·04, P = 0·014, slope (SE) = 0·043 (0·017)], and that was also the case for lysis [F = 7·15, d.f. = 1, 154, r2 = 0·04, P = 0·008, slope (SE) = 0·0046 (0·0017)]. In contrast, there was no significant effect of laying date in these analyses (agglutination: F = 0·34, d.f. = 1, 124, r2 = 0·003, P = 0·56; lysis: F = 0·21, d.f. = 1, 124, r2 = 0·002, P = 0·65), nor was there an effect of time before start of laying (agglutination: F = 1·27, d.f. = 1, 123, r2 = 0·010, P = 0·26; lysis: F = 0·12, d.f. = 1, 123, r2 = 0·001, P = 0·73). Analysis of repeat samples of 10 individuals with an average interval of 16 days showed a significant increase (paired t-test: agglutination: t = 2·81, d.f. = 9, P = 0·021; lysis: t = 4·58, d.f. = 9, P = 0·0013).

Figure 1.

The levels of agglutination in adult Barn Swallows in relation to sampling date (1 = 1 May). Line = linear regression line.

In addition to the date effect, agglutination showed a bell-shaped relationship with age [Fig. 2; quadratic term: F = 4·16, d.f. = 1, 152, r2 = 0·03, P = 0·043, slope (SE) = −0·38 (0·19)] but not for lysis (linear term: F = 0·01, d.f. = 1, 152, P = 0·92; quadratic term: F = 0·32, d.f. = 1, 152, P = 0·57).

Figure 2.

The levels of agglutination in relation to age (years) in adult Barn Swallows. Values are means (SE).

There was no significant sex effect for agglutination (F = 2·21, d.f. = 1, 151, r2 = 0·01, P = 0·14) or lysis (F = 2·67, d.f. = 1, 151, r2 = 0·01, P = 0·10). Male mating success was not predicted by agglutination (F = 0·67, d.f. = 1, 151, r2 = 0·01, P = 0·41) or lysis (F = 0·06, d.f. = 1, 151, r2 = 0·04, P = 0·98).

The prevalence and intensity of parasitism was as follows: chewing lice: 0·849, 14·27 (SE = 0·96), n = 159, feather mites: 0·711, 33·69 (3·06), n = 159, louse flies: 0·151, 0·12 (0·03), n = 159, mites: 0·0126, 0·016 (0·011), n = 159. There were no significant relationships between agglutination or lysis and the abundance of the four different parasites (F < 1·03, d.f. = 1, 151, r2 < 0·01, P > 0·31).

There was no significant relationship between agglutination and tail length or any other morphological character (F < 1·71, d.f. = 1, 151, r2 < 0·01, P > 0·19), and none of the interactions between morphology and sex reached statistical significance (F < 0·20, d.f. = 1, 149, r2 < 0·01, P > 0·65). There was no significant assortative mating for agglutination (Kendall rank order correlation: τ = −0·10, n = 75, P = 0·26), while there was assortative mating for lysis (Kendall rank order correlation: τ = 0·39, n = 75, P = 0·0008). Male tail length predicted agglutination of his mate, while there was no effect of the female's own tail length (Fig. 3a; Table 1a). In males, neither male tail length nor that of the mate, predicted agglutination (Fig. 3b; Table 1). Similar conclusions applied to lysis in males and females (Table 1b).

Figure 3.

(a) Residual agglutination in female Barn Swallows in relation to tail length (mm) of their mates. (b) Agglutination in male Barn Swallows in relation to tail length (mm) of their mates.

Table 1.  Stepwise regression models of (a) agglutination and (b) lysis in adult Barn Swallows in relation to age, age squared, date, tail length and tail length of mates. The models had the statistics F = 6·21, d.f. = 3, 70, r2 = 0·21, P = 0·0008, F = 6·38, d.f. = 2, 54, r2 = 0·19, P = 0·0032, F = 6·21, d.f. = 3, 70, r2 = 0·21, P = 0·0008, no model predicted male lysis, and F = 12·43, d.f. = 2, 54, r2 = 0·32, P < 0·0001
VariableSSFd.f.PSlope (SE)
(a) Agglutination
 Age0·10 0·02 10·90 
 Age squared19·58 3·43 10·068 
 Date70·5212·37 10·00080·08 (0·02)
 Error398·98 70  
 Age34·59 4·18 10·0461·28 (0·63)
 Tail length of mate73·01 8·82 10·0044−0·13 (0·04)
 Error447·17 54  
(b) Lysis
 No model found
 Age1·9117·56 10·00010·30 (0·07)
 Tail length of mate0·84 7·73 10·0075−0·014 (0·005)
 Error5·86 54  

We developed a best-fit model that accounted for 19% of the variance in agglutination (Table 2). In addition to the already reported effects of sampling date and age, there were significant effects of presence of a second clutch, a significant interaction between annual fecundity and sex, and a significant interaction between presence of a second clutch and sex (Table 2). In general, individuals with a second clutch had a stronger agglutination than individuals without a second clutch (Table 2). The interaction between sex and second clutch was due to females without a second clutch having much weaker agglutination than females with a second clutch, while there was no difference in males (Fig. 4). The interaction between sex and annual fecundity was due to a strong negative relationship between fecundity and agglutination in females, while no significant relationship was present in males (Fig. 5). The fecundity effect in females was entirely due to the second clutch, because fecundity in the second clutch decreased with agglutination, while this was not the case in the first clutch [partial regressions after controlling for the effects of date, age, age-squared: first clutch: F = 0·32, d.f. = 1, 57, r2 = 0·006, P = 0·58; second clutch: F = 23·15, d.f. = 1, 41, r2 = 0·36, P < 0·0001, slope (SE) = −1·85 (0·38)].

Table 2.  Best-fit model of agglutination (dependent variable) in relation to age, age squared, date, sex, annual fecundity and presence of a second clutch. The model had the statistics F = 3·40, d.f. = 8, 118, r2 = 0·19, P = 0·0015
VariableSSFd.f.PSlope (SE)
Age  1·600·24  10·62 
Age squared 26·244·01  10·048−0·44 (0·22)
Date 43·436·63  10·0110·05 (0·02)
Sex (S) 25·503·89  10·051 
Annual fecundity (A) 17·912·73  10·10 
Second clutch (C) 27·584·21  10·0421·76 (0·86)
S × A 35·775·46  10·021−0·44 (0·19)
S × C 46·907·16  10·00852·29 (0·86)
Residual753·20 115  
Figure 4.

Residual agglutination in relation to presence or absence of a second clutch and sex in the Barn Swallow. Values are means (SE).

Figure 5.

Residual agglutination in relation to annual fecundity (the number of fledglings) in male and female Barn Swallows.

A total of 25·6% of the 156 adults survived from 2005 to 2006. Adult survival from 2005 to 2006 was not predicted by agglutination or lysis (logistic regression based on 156 individuals: agglutination: Wald χ2 = 0·17, P = 0·68; lysis: Wald χ2 = 0·55, P = 0·46). Likewise, a logistic regression that included sex, capture date, tail length and their interactions did not reveal significant effects of agglutination or lysis (logistic regression based on 154 individuals: agglutination: Wald χ2 = 0·07, P = 0·79; lysis: Wald χ2 = 1·10, P = 0·29).


The main findings of this study were that: (1) very old individuals had suppressed NAb levels, but not complement levels; (2) females that did not produce a second clutch had lower NAb levels than females that produced a second clutch; (3) NAb levels were negatively related to annual fecundity in females, but not in males; (4) female NAb levels decreased with the expression of a secondary sexual character in their mates; and (5) NAb levels did not predict adult survival. We will briefly discuss these findings.

The mean levels of agglutination in the Barn Swallows investigated here were at the low end of recently reported values from a range of bird species (Matson et al. 2005; Mendes et al. 2006). The level of lysis relative to agglutination was likewise at the low end, and this category included several species such as Waved Albatross Phoebastria irrorata, Zebra Finch Taeniopygia guttata and Mourning Dove Zenaida macroura (Matson et al. 2005; Mendes et al. 2006). The underlying ecological factors that characterize this group of species remain far from clear and await the accumulation of NAb and complement data for many species under natural conditions.

Both NAbs and complement increased with sampling date, and an analysis of same individuals on two different occasions also showed a general increase. The latter finding precludes the possibility that the ‘date effect’ was simply due to individuals of different quality being sampled on different dates. In contrast, there was no evidence that breeding date predicted NAbs or complement levels.

NAbs and complement were related to two aspects of fecundity. The annual reproductive success of Barn Swallows mainly depends on the number of clutches produced rather than the number of eggs per clutch (Møller 1994a). Therefore, factors that determine the frequency of second clutches will have a strong effect on annual fecundity. Here, we have shown that females, but not males, that produce two clutches had higher NAb levels than females that did not. In addition, there was a negative relationship between annual fecundity and NAbs in females, but not in males, independent of the number of clutches. This latter effect was due to a strong negative relationship between fecundity in the second clutch and NAb, while there was no indication of a relationship in the first clutch. The simplest interpretation of these two results is that only females in prime condition, as reflected by high levels of Nabs, are able to produce a second clutch. This is indicative of only high-quality females producing two clutches per season. The negative relationship between fecundity in the second clutch, but not the first clutch, and Nab, suggests that females with low residual reproductive value, as reflected by low levels of NAb, produce large second clutches as a terminal investment in reproduction. Previous experiments in Barn Swallows have shown that enlarged first clutches and parasite load of first clutches both reduce the frequency of second clutches (Møller 1993). These findings can be reconciled with the present study if female parental investment in a larger first clutch with more parasites, reduces the ability of females to produce NAbs, or if such investment reduces the ability of females to reproduce successfully with a given level of NAb.

Parasites may mediate sexual selection because continuous coevolution between parasites and their hosts can maintain genetic variation in viability that may constitute a major advantage for ‘choosy’ individuals during mate selection (Hamilton & Zuk 1982). A quantitative review of the literature on parasite-mediated sexual selection provided evidence consistent with this hypothesis, especially when considering measures based on immune function rather than parasite abundance (Møller et al. 1999). The analyses of agglutination and lysis in this investigation failed to find evidence of differences in measures between mated and unmated males. Likewise, there was no evidence that NAbs and complement were related to tail length of males, a secondary sexual character in the Barn Swallow (Møller 1994a). However, we did find evidence of positive assortative mating for lysis, but not for agglutination, suggesting that individuals with similar lysis (and hence similar parasite loads; Møller 1991) tend to mate with each other. In general, these results suggest that sexual selection in terms of mating success was not based on these components of immunity. However, females mated to long-tailed males had both weaker NAb and complement levels than females mated to short-tailed males, while there was no effect of the tail length of the female itself. Neither male nor female tail length predicted male NAb or complement. This surprising result suggests that male tail length can be considered the extended phenotype of a female, as shown in several studies of the Barn Swallow (de Lope & Møller 1993; Møller 1994a,b; Kose & Møller 1999; Kose, Mänd & Møller 1999). Previous analyses have shown that females mated to long-tailed males provide differential parental investment (Burley 1986; Møller & Thornhill 1998). They do so by contributing more to nest building and feeding of offspring, but also by laying eggs earlier, laying more eggs and more often raising a second clutch than females mated to short-tailed males (de Lope & Møller 1993; Møller 1994a,b; Kose & Møller 1999; Kose et al. 1999). Such differential investment is bound to be costly to females, although the currency of such cost has not yet been identified. The results presented here on the relationship between male tail length and female levels of NAb and complement, are consistent with this hypothesis, suggesting that females that mate with long-tailed males and invest differentially in reproduction, in fact have reduced levels of NAb and complement. If low levels of NAb and complement are indicative of low residual reproductive value, for example due to immuno-senescence, the differential parental investment by such females could be considered the result of terminal investment.

Immuno-senescence occurs as a consequence of a decline in the efficiency of the immune system with age as observed in humans (Miller 1996). Recently, two studies of humoral immunity in free-living birds have provided evidence of immuno-senescence, with immune response decreasing linearly with age (Cichon, Sendecka & Gustafsson 2003; Saino et al. 2003). Here we found that agglutination, but not lysis declined with age in a nonlinear fashion (Fig. 2). The reason why senescence was demonstrated in one measure, but not the other cannot readily be explained, although it is noteworthy that nestlings and adults differed significantly in agglutination that showed a strong decline with age, but not in lysis. One possibility is that immune components that increase rapidly at early age (as agglutination) show evidence of immuno-senescence, while that is not the case for complement that does not show a rapid increase between nestlings and adults. The dramatic decrease in agglutination among individuals aged 4–5 years is consistent with expectations based on immuno-senescence. The alternative hypothesis that the age effect was due to selection would predict that survivors would have stronger measures than nonsurvivors leading to an increase in immune measure with age. The fact that we observed the opposite pattern suggests that this explanation is unlikely to apply. While immune responses often have been found to predict survival of both nestlings and adults in birds and other organisms (Møller & Saino 2004), we found no evidence of that being the case for agglutination or lysis, despite the fact that mortality removed most birds. Capture probability of adults in the study population exceeds 98% (Møller & Szép 2002), making it highly unlikely that the data are biased by differential capture probability.

We quantified the abundance of four kinds of ectoparasites in Barn Swallows, but found no evidence that abundance was related to levels of NAbs or complement. Previous studies have linked NAbs levels to resistance to malarial parasites (Congdon et al. 1969), and naturally occurring concentrations kill bacteria and spirochetes in vivo (Ochsenbein et al. 1999; Belperron & Bockenstedt 2001). Recently, Whiteman et al. (2006) have shown a negative correlation between abundance of chewing lice in Galápagos Hawks Buteo galapagoensis and NAb levels. However, we were unable to replicate the latter effect despite a very high prevalence and intensity of chewing lice in our study population of Barn Swallows. Nor did we find an association between NAb and abundance of three other kinds of ectoparasites. Given the high prevalence and intensity of three of these ectoparasites, and given our large sample sizes, it seems unlikely that NAb and lysis will predict abundance of any of these common ectoparasites. We note that our previous studies of some of these ectoparasites have shown evidence of correlation with specific types of leucocytes (e.g. Saino, Møller & Bolzern 1995), as expected from the role of leucocytes in regulating ectoparasites (Roitt et al. 1996).

Innate immunity as reflected by NAbs and complement is assumed to be less sensitive to short-term variation in environmental conditions than acquired immunity (Matson et al. 2005). While the lack of relationship between body condition, as reflected by body mass, or body mass adjusted for a skeletal measure of body size such as keel length, and NAbs and complement may be consistent with this suggestion, other results were not. For example, the strong negative relationship between annual fecundity and female NAbs, the strong positive association between sampling date and NAbs and complement, respectively, even when sampling the same individuals repeatedly, suggests that both NAbs and complement respond to changes in environmental conditions.

In conclusion, we have found evidence consistent with the hypothesis that intraspecific variation in levels of natural antibodies and complement, predict variation in a range of fitness components related to life history and sexual selection.


K. D. Matson kindly provided invaluable help with the assays.