J. Kilpimaa, Section of Ecology, Department of Biology, University of Turku, FI-20014 Turku, Finland. Tel.: +358-2-333-5556; fax: +358-2-333-6550; e-mail: firstname.lastname@example.org
Maternal investment in offspring immunity via egg quality may be an adaptive evolutionary strategy shaped by natural selection. We investigated how maternal investment in eggs can influence offspring immunity by conducting two experiments. First, we manipulated foraging performance of the mothers before egg laying by attaching a small weight to their back feathers. During the nestling period, we investigated offspring total antibody production at the age of 7 days and after antibody challenge, and conducted a partial cross-fostering design to separate the effects of the experiment and rearing-related variation on offspring immunity. In a separate experiment, partial cross-fostering with antibody challenging without female pied flycatcher manipulation was conducted for another set of nests. Total antibody levels at the age of 7 days were reduced in nestlings of the experimental female pied flycatchers when compared with the set of unmanipulated nests. Maternal investment in the eggs may affect some aspects of offspring immunity during the early nestling period and this investment is costly. However, antibody response to a set of novel antigens (sheep red blood cells) at the end of the nestling period was not affected by the female pied flycatchers treatment. Instead our results suggest that general antibody responsiveness is mainly determined by the rearing environment and total antibody levels before the injection.
Variation in immune function has been suggested to influence fitness, because immune function reflects an important aspect of parasite resistance. Many aspects of vertebrate adaptive immunity are inducible and hence strongly environmentally determined (Frost, 1999). Some immune traits have also been demonstrated to have a genetic basis (Wakelin & Apanius, 1997; Kilpimaa et al., 2005). Because immune function can be energetically or nutritionally costly, environmental conditions may also influence individual immune efficiency (Klasing et al., 1987; Sheldon & Verhulst, 1996; Zuk & Stoehr, 2002). Phenotypical variation in offspring immune function has also been suggested to be influenced strongly by the phenotype of the mother (Grindstaff et al., 2003). This is because mothers can directly influence offspring immune function by the transmission of antibodies to offspring via eggs. Because the offspring are expected to live in the same environment as their parents, mothers can pass on specific antibodies that can protect the offspring against the particular pathogens at the given environment. Mothers can also enhance the general ability to resist pathogens via investment of nutrients and antioxidants in the eggs.
The physiological mechanisms of how mothers transfer immunity to offspring are well understood from numerous poultry studies (Patterson et al., 1962; Apanius, 1998). However, there is little evidence of the evolutionary consequences of maternal effects on offspring immunity in wild populations. In a study of kittiwakes (Rissa tridactyla), Gasparini et al. (2002) demonstrated maternal transfer of antibodies to offspring against locally virulent pathogens. Saino et al. (2003) manipulated the egg carotenoid levels which affected nestling T-cell-mediated immune response in barn swallows (Hirundo rustica). Experimental exposure of mothers to parasites before egg laying has been suggested to enhance offspring survival via maternal transfer of antibodies (Heeb et al., 1998). These results suggest that maternal transfer of antibodies may be adaptive, because it affects offspring fitness or traits closely associated with fitness. However, maternal allocation can be regarded as an evolutionary strategy only if there is a cost associated with it. There are few studies investigating the costliness of maternal investment in offspring immunity in vertebrates. In captive Japanese quails (Coturnix japonica) diet quality did not affect maternal antibody transmission to egg yolk (Grindstaff et al., 2005). Similar results suggesting that some aspects of offspring immunity may not be limited by the protein or energy intake of the mother have been found in studies in ungulates (Sams et al., 1995; Landete-Castillejos et al., 2002).
In this study, we investigated how maternal investment in the eggs can influence offspring immunity and whether there is a cost of doing so. To achieve this, we manipulated maternal investment by handicapping female pied flycatchers by attaching a small weight to their back feathers before egg laying. To distinguish between the effects of the experiment and rearing-related variation a partial cross-fostering design was conducted. We also used a different subset of broods to investigate genetic and environmental variation in nestling immune function.
Materials and methods
Experiment 1: handicapping of female pied flycatchers
This study was conducted in Central Finland, at the Konnevesi research station in May–June 2002. To investigate prehatching maternal effects on nestling immune function, we manipulated maternal condition by handicapping flight performance of female pied flycatchers. To achieve this, we glued a small weight (1.8 g) on the back feathers of the experimental female pied flycatchers before the onset of egg laying (mean 9.02 ± 4.32 SD days before egg laying). We assumed that this would increase energy expenditure and handicap female pied flycatcher foraging performance, which could possibly affect female pied flycatcher resources to invest in her eggs. During capture, female pied flycatchers were also blood sampled (a sample of maximum 75 μL from brachial vein) and their wing length (accuracy 0.1 mm) and weight (accuracy 0.1 g) measured. Control female pied flycatchers were only blood sampled and measured as above. In total, we captured 21 experimental and 22 control female pied flycatchers before their egg laying. Pied flycatchers usually lay one egg per day, and clutch size varies from five to eight eggs (Lundberg & Alatalo, 1992). When a female pied flycatcher started to incubate her eggs, she was recaptured in her nest box. The weight was removed (unless it had already fallen off before this capture), and the female pied flycatcher was blood sampled and weighed again. Eggs were measured using a calliper (accuracy 0.01 mm). Most of the female pied flycatchers managed to get rid of the weight before incubation, but observations of female pied flycatchers suggest that these female pied flycatchers did carry the weight at least for a few days.
The experiment could affect nestling immune function both via egg quality (prehatching maternal effects) and through impaired nestling feeding performance. To distinguish between these sources of variation, partial cross-fostering was conducted. Half of the two-day-old nestlings were swapped between pairs of nests with the same hatching date and approximately the same number of nestlings. Hatching date was defined as the day on which at least half of the nestlings had hatched. For each pair (dyad) one brood originated from an experimental female pied flycatcher and the second from a control female pied flycatcher. As a result within each pair, there was an experimental female pied flycatcher raising half of her own young and half of the brood of the control female pied flycatcher and vice versa. Because hatching dates and clutch sizes of the original control female pied flycatchers and the experimental female pied flycatchers did not match perfectly, we also had to use female pied flycatchers that were captured for the first time during their incubation period as controls. As a result 12 of the 21 control female pied flycatchers used in the experiment were captured for the first time during incubating. Claws of fostered nestlings were painted to distinguish original from fostered nestlings, before they were banded with individually numbered aluminium rings at the age of seven days.
Experiment 2: genetic and environmental effects of total antibody production
We also conducted a separate cross-fostering experiment without experimental manipulation of female pied flycatcher condition, to study origin-related versus environmental variation on nestling immune function (n = 16 broods). The hatchlings were cross-fostered when day two, blood sampled on day seven (for measuring pre-injection total antibody levels), challenged with sheep red blood cells (SRBC) on day seven and again blood sampled (for total antibody levels) on day 12 (see below).
Measuring total immunoglobulins and antibody challenge
In the middle of their nestling period (when the nestlings were seven days old) all of them (nestlings from both experiments 1 and 2) were first blood sampled (maximum 75 μL) by puncturing the brachial vein to measure the initial total antibody levels. After this, they were all injected intraperitoneally with 100 μL of a phosphate buffered saline (PBS) containing a concentration of 5 × 105 SRBC to measure their general antibody responsiveness. Five days later (at the age of 12 days; pied flycatchers fledge when c. 14 days old) all the nestlings were blood sampled again to measure their total antibody levels.
Immunoglobulin concentrations were measured from plasma samples with an ELISA method for total immunoglobulin levels for seven-day-old nestlings, and for increase in plasma total antibody levels between nestling dates seven and 12. In brief, 96-well ELISA plates (NuncTM Immunoplate; Nunc Co., Nunc A/S, Roskilde, Denmark) were first coated with antichicken IgG (Sigma C-6409; Sigma Chemical Co., St. Louis, MO, USA). Using commercial antichicken immunoglobulins in an ELISA is a reliable technique to detect total serum immunoglobulins from a variety of bird species including the pied flycatcher (Martinez et al., 2003). The plates were incubated in +4 °C overnight. After emptying the wells, they were masked with 1 % BSA–PBS (Roche Diagnostics GmbH, Mannheim, Germany) for 1 h and washed 3 × 200 μL with PBS–Tween. Samples and their replicates (50 μL well−1) were diluted with 1 % BSA–PBS and added to the wells. For each plate, a standard with different dilutions (50 μL well−1; diluted also with 1 % BSA–PBS) was added. As a standard we used a mixture of plasma of all individuals measured, that was given an arbitrary concentration of 106. All values were subsequently expressed relative to this standard (antibody index = relative antibody titer/10 000). Samples and standards with different dilutions were incubated for 3 h at room temperature. After washing the plates (3 × 200 μL with PBS–Tween), an alkaline phosphatase conjugated antibody (Sigma A-971 antichicken IgG; diluted with 1 % BSA–PBS) was added to the wells and incubated overnight at +4 °C. Finally, after washing the wells (3 × 400 μl with PBS–Tween) an alkaline phosphatase substrate PNPP (p-nitrophenyl phosphate, Sigma 104® phosphatase substrate; Sigma Chemical Co.) in 1 m diethanol amine buffer (1 mg mL−1) was applied (50 μL/well). The absorbance of the wells was read several times in an ELISA reader at 405 nm for up to 1 h (until the highest standard point reached the absorbance 2.0).
In the avian egg, maternal immunoglobulin (type G) is located in the yolk (Klasing & Leshchinsky, 1998). Antibodies in the blood of 1-day-old nestlings are considered to be of maternal origin (e.g. Apanius, 1998; Klasing & Leshchinsky, 1998). However, in the middle of the nestling period the chick's own immune system has most probably started producing its own antibodies. This has been shown, e.g. in magpies (Pica pica; Pihlaja et al., 2006), and furthermore, total immunoglobulin levels in the middle of the nestling period are strongly correlated with levels just after hatching (which are of maternal origin, Pihlaja et al., 2006). When the nestlings first start producing immunoglobulins, they produce immunoglobulin type M (Frank, 2002). IgM is a naïve antibody, which is then transformed to a more specific antibody IgG (Frank, 2002). Western blotting and immunostaining with the antibody showed that commercial antichicken IgG antibody recognized both the IgG and IgM isotypes of plasma of several bird species: great tit (Kilpimaa et al., 2005) magpie, black grouse Tetrao tetrix, and black-headed gull Larus ridibundus; I. Jokinen, personal communication). Therefore, our results are presented as total immunoglobulin concentration.
Post-injection total antibody levels were significantly higher than pre-injection total antibody levels (paired t-test: t = 22.55, P < 0.001, n = 264), and only 6.8 % of nestlings had lower post-injection total antibody levels than pre-injection levels. In the analysis, log transformed post-injection total antibody levels were used as a measure of general antibody responsiveness and log transformed pre-injection levels were used as a covariate.
The effects of the female pied flycatcher handicapping experiment were estimated using mixed anova with proc mixed in sas (version 9.1; SAS Institute, Cary, NC, USA). We used proc mixed, because proc mixed allows fitting both fixed and random effects. As a method of estimation restricted maximum likelihood (REML) was used. Because of the unbalanced data set (unequal family sizes) we specified the ‘ddfm = kenward’ option on the ‘MODEL’ statement to obtain Kenward–Roger estimates of the denominator degrees of freedom (d.f.) for the tests of fixed effects. In our model the term ‘experiment’ was regarded as a fixed effect and it relates to the effects of the female pied flycatchers manipulation. Random effects in the model were ‘rearing’ (nest of rearing) and ‘dyad’ (pair of nests). Interactions included in the model were experiment × dyad and experiment × rearing nested in dyad and their effects were regarded as random. The term ‘rearing’ estimates the effects of common rearing environment during nestling period and its effects were nested within the term ‘dyad’. Each dyad (pair of nests) consisted of two partially cross-fostered broods. The term ‘dyad’ reflects any temporal variation faced by offspring among pairs of nests during breeding season. Experiment × dyad interaction if significant would suggest origin-related variation apart from the female pied flycatchers manipulation. Significant experiment × rearing interaction would suggest that the effects of the experiment would vary depending on the foster environment.
In the broods of unmanipulated female pied flycatchers (experiment 2), origin- and rearing-related effects on offspring immune function were estimated using nested anova with proc glm in sas. In each model, the term ‘origin’ (nest of origin) and the term ‘rearing’ (nest of rearing) were nested as a factor within the ‘dyad’. The term ‘rearing’ estimates the effects of common rearing environment during nestling period. The term ‘origin’ accounts for any precross-fostering effects including additive genetic variation and dominance variation and maternal effects if present. Because of the unequal family sizes we performed random effect analysis of variance tests for each effect in the model. Variance components were estimated with the sas varcomp Procedure using REML estimates.
For both experiments pre-injection immunoglobulin levels (term ‘pre-immunoglobulins’ were included in the model when analysing general antibody responsiveness.
The effects of the handicapping experiment on female pied flycatchers
The handicapping experiment delayed the onset of egg laying. There was no difference in the capturing date of experimental and control female pied flycatchers caught before egg laying (anova, F1,39 = 0.07, P = 0.829), but the handicapped experimental female pied flycatchers laid the first egg on average 2 days later than the control female pied flycatchers (mean laying date of handicapped female pied flycatchers: 34.86 ± 0.47; mean laying date of control female pied flycatchers: 33.05 ± 0.0.63; anova, F1,43 = 5.285, P < 0.05). There was a tendency for experimental female pied flycatchers to have lower total immunoglobulin concentration in their plasma after egg laying (during early incubation) (mean log transformed Ig concentration of handicapped female pied flycatchers: 9.01 ± 0.39; mean log transformed Ig concentration of control female pied flycatchers: 9.23 ± 0.29; anova, F1,40 = 3.801, P = 0.059). Female pied flycatchers body weight, measured before egg laying and during early incubation, was not affected by the treatment (repeated measures analysis: F1,37 = 0.202, P = 0.655). Mean egg volume (mean egg volume of handicapped female pied flycatchers: 1602.10 ± 23.78; mean egg volume of control female pied flycatchers: 1613.69 ± 82.69; anova, F1,39 = 0.136, P = 0.714) or clutch size (mean clutch size of handicapped female pied flycatchers: 6.38 ± 0.16; mean clutch size of control female pied flycatchers: 6.24 ± 0.54; anova, F1,42 = 0.511, P = 0.479) were not affected by the treatment.
The effects of the handicapping experiment on nestling total immunoglobulins at the age of seven days
The experiment affected the nestling immunoglobulin levels at the age of seven days (Table 1). The nestlings originating from the nests of the handicapped female pied flycatchers had lower total antibody levels than nestlings originating from the nests of control female pied flycatchers (Fig. 1). Variance estimates for random effects were small (Table 1). The nest of rearing had no effect on nestling total immunoglobulin levels at the age of seven days. There was no significant effect of the term ‘dyad’, suggesting that the timing of breeding had no effect on total immunoglobulin levels. There was no significant experiment × foster interaction, suggesting that the handicapping experiment had similar effects regardless of the foster environment. There was a tendency for experiment × dyad interaction which may also suggest origin-related variation other than caused by female pied flycatcher manipulation.
Table 1. Mixed model of the effects of female pied flycatcher handicapping experiment on nestling total immunoglobulins at the age of seven days.
Tests of fixed effects
Covariance parameter estimates
Exp × dyad
Exp × rearing (dyad)
The effects of the female pied flycatcher handicapping experiment on nestling general antibody responsiveness after immune challenge at the age of 12 days
The experiment did not affect the nestling general antibody responsiveness (measured as a change in total antibody levels) at the age of 12 days (Table 2). Instead, the post-injection total antibody levels were strongly affected by the pre-injection immunoglobulin levels. None of the random effects included in the model were significant.
Table 2. Mixed model of the effects of female pied flycatcher handicapping experiment on nestling antibody responsiveness at the age of 12 days.
Tests of fixed effects
Covariance parameter estimates
Exp × dyad
Exp × rearing (dyad)
Cross-fostering experiment without female pied flycatcher manipulation: the effects of the nest of origin and the rearing environment on nestling total immunoglobulin levels at the age of seven days
The nestling total immunoglobulin levels at the age of seven days were determined by the nest of origin, suggesting genetic variation or prehatching maternal effects of nestling total immunoglobulin levels (Table 3). Common rearing environment had no effect on nestling immunoglobulins. There was no significant effect of the term ‘dyad’ on nestling immunoglobulin levels. There was no significant origin × rearing interaction, and the interaction was left out of the final model. Variance components for each effect are given in Table 5.
Table 3. The effects of the nest of origin and the common rearing environment on nestling total immunoglobulin levels at the age of seven days.
Table 5. Causal components of variance (REML estimates) for the total immunoglobulins at the age of seven days and at the age of 12 days after immune challenge in the cross-fostered unmanipulated broods.
Total IG at the age of 7 days, Var (%)
Total Ig after immune challenge at the age of 12 days, Var (%)
Cross-fostering experiment without female pied flycatcher manipulation: the effects of the nest of origin and the rearing environment on nestling general antibody responsiveness after immune challenge at the age of 12 days
Post-injection total antibody levels were mainly determined by the common rearing environment, and the nest of origin also had an almost significant effect (Table 4). Pre-injection total antibody levels had an almost significant effect on general antibody responsiveness. There was no significant origin × rearing interaction and the interaction was excluded from the final model. Variance components for each effect are given in Table 5.
Table 4. The effects of the nest of origin and the common rearing environment on nestling general antibody responsiveness at the age of 12 days.
Experimental manipulation of maternal condition before egg laying resulted in reduced nestling total immunoglobulin levels at the age of seven days, which represents the middle of their nestling period. However, general antibody responsiveness (measured as change in total antibody titre after challenging with SRBCs) at the end of the nestling period was not affected by the treatment. Instead the general antibody responsiveness was correlated with initial antibody levels before immune challenge. In a cross-fostering experiment where maternal investment was not manipulated, immunoglobulin levels at the age of seven days were mainly determined by the nest of origin, suggesting genetic and/or prehatching maternal effects. The effects of manipulation of female pied flycatcher condition suggest that this variation may be at least partly accounted for by prehatching maternal effects. Variation in antibody responsiveness at the age of 12 days was mainly because of the common rearing environment.
Several mechanisms may explain why the offspring of experimentally handicapped female pied flycatchers had less immunoglobulins in the middle of their nestling period. For example, carrying extra weight during egg laying may have caused impaired foraging performance or increased energy expenditure in experimental female pied flycatchers which may have resulted in a trade-off between investment in self-maintenance and investment in eggs. The additional weight was used to mimic natural variation in female pied flycatcher body condition over the nesting period, as female pied flycatchers normally acquire energy reserves for the incubation period amounting to the same level of body mass increase (Lundberg & Alatalo, 1992). However, we did not glue the extra weight too tightly to allow most of the female pied flycatchers to get rid of the weight already before incubation. As we did not sample the eggs, we do not know which component of the egg quality that influences offspring immunity was affected. It is possible that lower total immunoglobulin levels in the experimental chicks are a result of lower total immunoglobulin concentration in the egg yolk. In our study, experimental female pied flycatchers tended to have lower plasma total immunoglobulin concentrations than the control female pied flycatchers after egg laying (at the beginning of the incubation period). In captive Japanese quails, dietary status of mothers did not affect maternal transmission of antibodies to egg yolk (Grindstaff et al., 2005). However, captive female pied flycatchers fed ad libitum may be able to compensate the poor food quality with increased food intake. In our study we manipulated feeding performance, probably reducing both amount and quality of the food plus increasing the total energy demand. Also, in the wild immunoglobulin production may be traded off against many competing traits, increasing the cost of antibody transmission.
Total immunoglobulin levels of nestlings at the age of 7 days probably represent at least partly immunoglobulins produced by the offspring themselves and less maternally derived antibodies. Maternal antibodies start declining with the onset of active antibody production by offspring, but can be crucial for chick survival until their own adaptive immunity has fully developed (Grindstaff et al., 2003). For example, in chicken, offspring begin to synthesize antibodies independently at the age of 5 days (Patterson et al., 1962; Brambell, 1970; Apanius, 1998). The case might be different between precocial and altricial species, and is poorly understood, especially in wild species. Evidence from magpies shows that the mother's condition may limit antibody transmission into the eggs, and at hatching, antibody levels are positively related to chick's own antibody production in the middle of their nestling period (Pihlaja et al., 2006). In any case, maternal immunoglobulins may continue to affect the strength and the diversity of offspring immune responses after it has been catabolized (Lemke & Lange, 1999; Lundin et al., 1999).
Another possibility is that eggs of experimental female pied flycatchers had less carotenoids or nutrients that may affect offspring immunity. Carotenoids may be especially important for offspring immunity, because the functioning of the immune system produces oxidative stress and carotenoids act as antioxidants (Chew, 1996; Møller et al., 2000). In the barn swallow, experimental manipulation of egg carotenoids affected the nestling T-cell-mediated immunity (Saino et al., 2003). However, the antibody production to Newcastle disease virus vaccine was not affected. In lesser black-backed gulls (Larus fuscus) carotenoid-fed female gulls produced eggs with increased concentration of carotenoids but decreased concentration of immunoglobulins compared with the control female gulls, suggesting that high antioxidant production may compensate for passive immunity provided by immunoglobulins (Blount et al., 2002). Regardless of the mechanism of affecting offspring immunity, our results suggest that maternal investment in offspring immunity is costly and it is likely that this investment is adaptive.
General immune responsiveness was not affected by our handicapping experiment, but was mainly determined by growth conditions. In the female pied flycatcher handicapping experiment, the change in antibody titre was, however, affected by pre-injection total immunoglobulin levels (term ‘pre-immunoglobulins’ in the model, Table 4), which may in turn reflect prehatching maternal and/or genetic effects. However, maternal effects may be more important in more specific immune responses. For example, mothers can pass on specific antibodies against locally virulent pathogens to protect their offspring during the early nestling period. Specific antibodies may also have long-lasting effects on both strength and diversity of antibody responsiveness (Grindstaff et al., 2003). Our results are in concordance with a study of serins (Serinus serinus) where food availability during nestling period affected antibody response to SRBCs (Hoi-Leitner et al., 2001). There are also several studies suggesting strong environmental variation in T-cell-mediated immune response to PHA (phytohaemagglutinin) but also origin-related variation in some studies (Saino et al., 1997; Brinkhof et al., 1999; Christe et al., 2000; Tella et al., 2000; Kilpimaa et al., 2005). This response has been considered as a measure of general immune efficiency, because PHA is a mitogen that stimulates many specific T-cells to proliferate (Goto et al., 1978; McCorkle et al., 1980). SBRCs are broadly antigenic immunogen which may also reflect general immune capacity or general antibody responsiveness. General antibody responsiveness is probably a trait closely linked to fitness and it may be under strong directional selection, which may explain the low genetic variation found in our study and in previous studies. Origin-related variation in nestling total immunoglobulin levels at the age of 7 days may suggest heritable variation in this trait. However, because our female pied flycatcher handicapping experiment suggests that this trait is also affected by the phenotype of the mother it is likely that at least a part of the origin-related variation accounts for nongenetic maternal effects.
To conclude, according to our experiment, the mother's phenotype may affect offspring immune function at least during the early phase of the nestling period. Moreover, our results indicate that maternal investment in offspring immunity is costly for the mothers. This suggests that maternal investment may be an evolutionary strategy shaped by natural selection. General antibody responsiveness to antigenic immunogen in a broad manner was mainly environmentally determined with little origin-related variation. However, maternal effects may be more important for specific resistance to virulent pathogens at the local level.
We thank Konnevesi research station for facilities, Matti Halonen for the help in the field, Elina Virtanen for the assistance in the laboratory and the Academy of Finland for funding the study. The experiment was carried out under the permission of the Animal Care Committee of the University of Jyväskylä.