Early maternal effects and antibacterial immune factors in the eggs, nestlings and adults of the barn swallow


Nicola Saino, Dipartimento di Biologia, via Celoria 26, I-20133 Milano, Italy. Fax: +39 0250314802; e-mail: n.saino@mailserver.unimi.it


Transfer of immune factors via the egg may represent a maternal adaptation enhancing offspring survival. Lysozyme is a major component of maternal antibacterial immunity which is transferred to the eggs in birds. In a population of barn swallows (Hirundo rustica), lysozyme activity declined during the prelaying and laying periods in females but not in males. Egg hatching failure decreased with maternal lysozyme activity. The first eggs in a clutch contained more lysozyme and produced nestlings with larger lysozyme activity when 5 days old than last-laid ones. In a cross-fostering experiment where brood size was manipulated, nestling origin but not post-manipulation brood size affected lysozyme activity. Hence, maternal lysozyme varies during the breeding season and may differentially enhance antibacterial immune defence of the eggs and nestlings in relation to laying order. These findings suggest that offspring innate immunity is influenced by early maternal effects.


Parasites can exert intense negative impact on their hosts and their powerful selection pressure is among the main determinants of evolutionary change in host populations (Price, 1980; Wakelin, 1996; Wikel, 1996; Clayton & Moore, 1997; Møller, 1997). The effect of parasites on host evolution is indirectly evidenced by the complex immune system of vertebrates, which consists of two branches. Innate immunity provides nonantigen specific defences which act rapidly upon infection and are not modified significantly by repeated exposure to the same pathogen. Conversely, acquired immunity consists of parasite-specific mechanisms of response (Stites et al., 1997; Pastoret et al., 1998). However, many of the effector cells and molecules are shared between the two branches of immunity.

Females have the opportunity to influence offspring phenotype via early maternal effects mediated by the quality of the eggs, which may have profound consequences for offspring ontogeny and survival (see Kaplan, 1985; Mousseau & Fox, 1998a, b; Price, 1998; Wade, 1998). Potentially major, but poorly investigated components of egg quality are immune factors, which may protect the egg itself and the offspring during their early life-stages from infection by parasites (Rose & Orlans, 1981; Naqi et al., 1983; Graczyk et al., 1994; Smith et al., 1994a, b; Kaspers et al., 1996; Lung et al., 1996; Ahmad & Siddique, 1998; Gasperini et al., 2000; Saino et al., 2002). In vertebrates, mothers transfer protective immune factors to their progeny via routes that vary idiosyncratically among taxa and the particular component of immunity under scrutiny. Birds and other oviparous vertebrates transfer immunoglobulins to the embryo via the egg yolk, whereas lysozyme, which acts as a major component of innate antibacterial immunity by digesting peptoglycanes of bacterial cell walls, is transferred to the albumen (Sato & Watanabe, 1976; Clark & Bueschkens, 1986; Grinde, 1989; Tizard, 1991; Trziszka, 1994; Braun & Fehlhaber, 1996; Kudo, 2000). Egg lysozyme may be carried over into the embryo and nestling body (Board & Fuller, 1974). Hence, transmission of lysozyme may represent an important form of early maternal effect on egg and offspring viability. However, no data on lysozyme are available, to the best of our knowledge, for bird species in the wild.

In this study, we first analysed variation of lysozyme activity in adult barn swallows (Hirundo rustica). Consistent with variation in circulating immunoglobulins we documented in a previous study (Saino et al., 2001a), we predicted that lysozyme levels of mothers, but not fathers, increased before egg laying in order to avoid reduction of innate immune defence, and then decreased during laying because of transfer of lysozyme to the albumen. Second, we investigated variation of lysozyme activity in the eggs in relation to laying order. Thirdly, we analysed the relationship between circulating lysozyme in mothers and hatching success of the eggs, while predicting a positive relationship as observed in poultry (Melek, 1977; Krykanov, 1982;Prusinowska&Jankowski, 1996; Prusinowska et al., 2000). Fourthly, lysozyme activity in the plasma of the nestlings at different ages was analysed in relation to lysozyme activity in their mother and laying order of their original eggs. If the concentration of lysozyme in the egg is proportional to that in the mother, a positive correlation between the mother and offspring is predicted. In addition, variation in egg quality according to laying order may result in variation of circulating lysozyme in the offspring according to laying order. Finally, we conducted a brood size manipulation experiment to estimate the effect of crowdedness in a nest on lysozyme concentration. The timing of onset of autonomous lysozyme production during ontogeny is unknown for altricial birds. Under the assumption of negligible autonomous production during the nestling stage and strong maternal and/or genetic effects, we predicted that bactericidal activity of siblings raised in different broods would be affected by nest of origin. If nestlings can produce lysozyme, however, brood enlargement could either result in depressed lysozyme levels because of compromised immunedefence by nestlings in poor condition (Saino et al., 1997a), or, alternatively, in enhanced levels if brood enlargement does not exceedingly depress nestling body condition and bacteria are more abundant in crowded nests.


General field procedures

The barn swallow is an insectivorous, semicolonial, biparental passerine. Females incubate one to three clutches of two to seven eggs per breeding season. Hatching asynchrony is small, with all eggs in a clutch usually hatching within 24 h. Nestling fledge when they are 18–20 days old. We studied barn swallows breeding in six colonies (farms) in an agricultural area near Milano (northern Italy) during spring–summer 1999–2001. We captured adults in the very early morning, starting at their arrival from migration. Each adult was individually marked with plastic bands and colour markings on breast and belly feathers. A blood sample was collected in heparinized capillary tubes by puncturing the brachial vein. Blood samples were kept in a cooling bag in the field, then taken to the lab and centrifuged within 12 h of collection. Plasma was stored at −20 °C.

Temporal variation in plasma lysozyme activity of adults

We determined the composition of breeding pairs by direct observation. Inspection of nests at least every second day allowed us a posteriori to determine the stage in the breeding cycle at which each adult had been captured. Breeding stage at capture was expressed as the difference, in days, between the date of capture and date of laying of the first egg by the focal female or by her social mate. In this study, we considered samples collected between 23 days before laying of the first egg and the day of laying of the last egg.

We collected a sample of nine clutches of eggs of known laying order. Eggs were collected the second day after clutch completion. The eggs were kept in a cooling bag in the field and taken to the lab where the albumen was separated from the yolk and kept at −20 °C.

Lysozyme activity in offspring in relation to egg laying order

Eggs in a large number of nests were individually marked according to laying order. Around the estimated day of hatching, we started visiting nests at short-time intervals (on average every 4 h during daylight) to associate the nestlings to the egg from which they hatched. In a subset of nests this procedure allowed us to identify two groups of siblings which could be assigned to either the first or last eggs laid. Nestlings were usually subjected to a first blood sampling on day 5 and a second sampling on day 12 after hatching. Blood sampling on day 5 was performed in order to obtain an estimate of lysozyme activity in the offspring as soon as possible after hatching. Based on our previous experience with barn swallows, we knew that it is difficult to extract blood from very young nestlings. This explains why we were unsuccessful in bleeding 20% or less of the nestlings on day 5 (see Results for a detailed account of the number of nestlings that were analysed). As nestlings had already been bled on day 5, we decided to collect a relatively small quantity of blood also on day 12. We could thus assay 85% or more of the nestlings present on day 12 (see Results). Exclusion of some nestlings might have produced bias if, for example, we systematically excluded relatively small nestlings. However, this turned out not to be the case, as nestlings that could not be bled adequately (or whose blood sample was not available for lysozyme concentration analyses for accidental reasons) did not significantly differ from those that were bled successfully in body mass measured on days 1, 4, 7 and 12 or in tarsus length on days 7 and 12 (mixed model analyses of variance withbrood and a binary factor indicating inclusion vs. exclusion from the analyses; effect of inclusion/exclusion on morphological variables on days 5 or 12: P > 0.35 in all cases).

Brood size manipulation and bactericidal activity in mothers and offspring

We manipulated brood size by cross-fostering an unbalanced number of nestlings between pairs of broods (hereafter ‘dyads’) from the same colony where hatching was completed on the same day, increasing brood size of one randomly chosen brood by one nestling and reducing brood size in the other by one nestling. Nestlings were individually marked with small colour patches on the legs, before receiving a colour band on the tarsus on days 4 or 5. All nestlings were removed from the nest and some were transferred to the other brood in the dyad, whereas the others were put back in their original nest (see Saino et al., 1997a, 1999 for details of the brood size manipulation protocol). Nestlings due to be cross-fostered were chosen randomly. When 12 days old, nestlings were subjected to a blood sampling.

Assay of plasma and albumen bactericidal activity

Lysozyme activity was determined using the lysoplate method as described by Osserman & Lawlor (1966), modified in order to carry out the assay using a micromethod. The test was performed on an agar gel with a dried strain of Micrococcus lysodeikticus (M-3770; Sigma Chemical Co., St. Louis, MO, USA), which is particularly sensitive to lysozyme activity, by inoculating each test hole with 12.5 μL of plasma. Crystalline hen egg white lysozyme (L-6876, Sigma Chemical Co.) was used to prepare a standard curve in each plate. Agar gels were incubated at 27 °C for 20 h and the area of the gel surrounding theplasma inoculation site where bacterial growth was inhibited was measured using an ad hoc ruler, and converted into hen egg lysozyme equivalents (HEL equivalents, expressed in μg mL−1) according to the standard curve. Hence, lysozyme activity will be expressed as HEL mL−1 throughout this study (μg mL−1) (see also Brightman et al., 1991). Mean intra-assay coefficient of variation of lysozyme activity estimates in the plasma was 9.85% (n=38). Intra-assay coefficient of variation for the albumen was 9.56%. Mean inter-assay coefficient of variation was 10.3%. Lysozyme activity data which were log10-transformed were necessary to achieve normality.


Lysozyme activity in relation to breeding stage, clutch size and egg hatchability

Mean lysozyme activity recorded in females was 3.97 HEL equivalents (0.17 SE, n=172) whereas in males it was 3.48 (0.17 SE, n=101). Lysozyme activity of females (log-transformed; n=172) showed an inverse U-shaped pattern of variation in relation to the date at blood sampling relative to date of laying of their first egg, as indicated by the significant effect of the quadratic term of breeding stage in a polynomial regression analysis (Table 1, Fig. 1). The maximum of the function fitted was achieved approximately 11.4 days before laying of the first egg. Lysozyme activity was negatively correlated with the linear term of breeding stage after the estimated peak [F1,134=16.38, P < 0.001; log-lysozyme activity= −0.018 (0.004 SE) breeding stage + 0.449]. However, no significant positive association with breeding stage was observed before the estimated peak (F1,34=0.00, n.s). Hence, the analysis of temporal variation of lysozyme activity did not provide conclusive evidence that the decline in lysozyme activity before and during egg laying was preceded by an increase (see Discussion).

Table 1.  Multiple regression analyses of lysozyme concentration (after log 10 -transformation) in relation to linear and quadratic terms of stage in the breeding cycle at blood sampling and laying date. Breeding stage was expressed as the difference, in days, between the day of blood sampling and the day of laying of the first egg (females) and laying ofthe first egg by the mate (males).
 CoefficientSEtP -value
 Breeding stage 5.314 × 10−4 0.0086 0.060.95
 Breeding stage2−3.745 × 10−4−5.266 × 10−4−0.710.48
 Laying date−0.0021 0.0015−1.460.16
 Intercept 0.613 0.070  
F3,97 =2.02, R2 =0.06, P =0.12
 Breeding stage−0.0223 0.0067−3.360.001
 Breeding stage2−9.722 × 10−4 3.807 × 10−4−2.550.011
 Laying date 0.0012 0.0011 1.030.31
 Intercept 0.410 0.056  
F3,168 =4.91, R2 =0.08, P =0.003
Figure 1.

Lysozyme activity (egg lysozyme equivalents, HEL, in μ g mL −1 ) in the plasma of 172 adult female and 101 adult male barn swallows during the prelaying and egg laying periods. Stage in the breeding cycle is expressed as the difference, in days, between calendar date of blood sampling and date of laying of the first egg (females) or date of laying of the first egg by the female mate (males). The curve represents the second order polynomial function fitted to the female data. No significant relationship between male lysozyme activity and breeding stage was observed.

Lysozyme activity of males (log-transformed; n=101) did not change in relation to the stage in the breeding cycle when sampled (Table 1, Fig. 1), and the same result was found when either the linear or the quadratic terms of breeding stage were included in the model. In both sexes, laying date did not predict lysozyme activity after controlling for the effect of breeding stage at blood sampling (Table 1). Exclusion of the laying date term from the regression model of females only slightly affected the coefficient of determination (model with laying date: R2=8.07%, see Table 1; without laying date: R2=7.50%).

An analysis of covariance with sex as a factor and polynomial terms of breeding stage and laying date as covariates showed no significant difference in lysozyme activity between the sexes (F1,272=2.53, n.s).

However, a significant sex × breeding stage interaction (F1,265=4.30, P=0.039) indicated that different patterns of variation existed in the two sexes. The effect of year, entered as a random factor, was never found to significantly contribute to these models (P > 0.1 in all cases).

Residual lysozyme activity after controlling for polynomial terms of breeding stage of females correlated negatively with clutch size (log-transformed lysozyme activity: r=−0.21, d.f.=155, P=0.009), and this significant correlation persisted after the effect of laying date was controlled for in a partial correlation analysis (rpar=−0.19, d.f.=154, P=0.017). No significant association between lysozyme activity and clutch size was observed for males (log-transformed lysozyme activity: r=0.08, d.f.=92, P=0.43).

Hatching failure was expressed as the ratio between the number of eggs that did not hatch and clutch size. A nonparametric correlation analysis revealed a significant negative association between hatching failure and residual lysozyme activity of females (Kendall's τ=−0.176, n=147, P < 0.01). However, lysozyme activity of males, considered as a control group, was not significantly related to hatchability of the eggs (τ=0.148, n=88, P=0.069). In a logistic regression analysis where hatching failure was expressed as a binary response variable (presence/absence of hatching failures), we found a significant interaction between residuals of lysozyme concentration in the two sexes and sex (P=0.019), indicating that the effect of maternal lysozyme concentration actually differed from that of males.

Lysozyme activity in eggs in relation to laying order

Lysozyme activity was analysed in eggs of known laying order from nine clutches. An analysis of covariance with mother as a factor and egg laying order as a covariate indicated that lysozyme activity declined in the last eggs (effect of laying order: F1,26=14.12, P < 0.01, coefficient: −5.69; Fig. 2) and a highly significant variation existed also among clutches (effect of mother: F8,26=6.71, P < 0.001). On average, very small differences existed between the first and the second egg whereas bactericidal activity declined more steeply in thethird and fourth eggs (Fig. 3). In fact, the quadratic term of egg laying order provided a better fit of lysozyme activity compared with the linear term (effect size for thelinear term: 0.352, quadratic term: 0.382) although the difference between the two models was very small. Egg mass showed no significant variation in relation to laying order (F1,22=0.04, n.s.; one clutch was excluded because egg mass value was not available for two eggs).

Figure 2.

Mean (+SE) lysozyme activity (hen egg lysozyme equivalents, HEL, in μ g mL −1 ) in the albumen of eggs from nine complete clutches of barn swallows in relation to the egg laying sequence. There was a significant decline in lysozyme activity during the laying sequence.

Figure 3.

Mean within-brood lysozyme activity (hen egg lysozyme equivalents, HEL, in μ g mL −1 ) of nestling barn swallows sampled at day 5 (a) and 12 (b) after hatching in relation to residuals of lysozyme activity in mothers after controlling for the significant effects of first and second order polynomial terms of breeding stage at blood sampling. There was a significant positive relationship. Equation of the fitted line at day 5: y =0.325 (0.086 SE) x  + 3.766; F1,36 =14.45, R2 =0.286, P =0.0005; day 12: y =0.236 (0.095 SE) x  + 4.267; F1,40 =6.17, R2 =0.134, P =0.017).

Lysozyme activity in parents and offspring

We analysed the correlation between lysozyme activity in mothers and offspring 5 and 12 days respectively, after hatching using residuals of maternal plasma bactericidal activity obtained from a regression on polynomial terms of breeding stage at blood sampling. These regressions were always made on the subset of females whose offspring were assayed for lysozyme activity, and they always gave results qualitatively similar to those obtained on the total sample of females, with both linear and quadratic terms of breeding stage significantly contributing to the regression models (t-values associated with P < 0.01 in all cases). In the analysis of lysozyme activity on day 5 we used 38 broods with a total of 144 nestlings, 115 of which (80%) were assayed for lysozyme. We found a positive relationship between residual lysozyme activity in mothers and mean activity in the plasma oftheir offspring (F1,36=14.45, P < 0.001; Fig. 3a). Regression analysis on blood samples collected at day 12 was based on 42 broods containing 161 nestlings, 152 of which (94%) could be assayed. Also in this case, the relationship was positive and statistically significant (F1,40=6.17, P < 0.05; Fig. 3b). However, neither the correlation coefficient nor the slope of the fitted equations differed between ages 5 and 12.

Paternal lysozyme activity did not significantly predict mean within-brood lysozyme in the offspring sampled on day 5 (F1,29=0.24, n.s) or on day 12 after hatching (F1,32=1.73, n.s). The association with offspring lysozyme activity was weaker when we used the residuals of paternal lysozyme activity from a regression on polynomial terms of breeding stage (details not shown).

The slope of the line fitted to offspring lysozyme on standardized residuals of maternal lysozyme after controlling for the effect of breeding stage was larger than that fitted to the corresponding values of paternal lysozyme (F1,35=4.57, P=0.039; the degrees of freedom are corrected for the number of broods for which we had data on both mother and father). However, no significant difference existed in the slope of the lines fitted to offspring lysozyme concentration measured on day 12.

Nestlings hatched from the first laid eggs had significantly higher lysozyme activity on day 5 compared with their nest mates hatched from last laid eggs. A mixed-model analysis of variance with brood as a random factor and laying order as a fixed effect based on 95 nestlings (86% of the those present in the 29 broods considered) disclosed a significant effect of laying order (F1,63=9.97, P < 0.01; Fig. 4).

Figure 4.

Lysozyme activity (mean + SE of hen egg lysozyme equivalents, HEL, in μ g mL −1 ) of nestling barn swallows hatched from the group of first or last laid eggs in each clutch, measured on day 5 ( n =29 broods) and day 12 ( n =25 broods). There was a significant effect of laying order on lysozyme activity at 5 but not at 12 days after hatching.

The difference in lysozyme activity in relation to laying order was absent when nestlings were 12 days old. A mixed-model analysis of variance on 81 nestlings of a total of 96 (84%) present in 25 broods indicated no significant effect of laying order (F1,65=0.02, n.s.; Fig. 4).

Lysozyme activity on day 5 was negatively correlated with brood size (r=−0.40, d.f.=36, P < 0.05), but not with clutch size (r=−0.17, d.f.=36, n.s). The relationships with brood size were significantly negative for both first and last laid nestlings, respectively [equation of the regression line for first laid nestlings: lysozyme activity=−1.127 (0.413 SE) brood size + 8.502; F1,27=7.45, P < 0.05; last laid nestlings: lysozyme concentration= −0.831 (0.320 SE) brood size + 6.322; F1,27=6.73, P < 0.05], with no significant differences in the slopes of the fitted equations. No significant correlation existed between mean within-brood lysozyme activity on day 12 and clutch or brood size, and the same results were obtained when first and last laid nestlings were considered separately (P > 0.30 in all cases).

Lysozyme activity did not change from age 5 to age 12 as indicated by paired comparisons of mean within-brood concentration at the two ages [mean at age 5: 3.84 (0.28 SE); age 12: 4.19 (0.28); t32=1.03, n.s.], and this lack of change was present both among first laid and last laid nestlings. Moreover, lysozyme activity at day 5 did not predict that at day 12. In fact, in a mixed model analysis of covariance on 25 broods with brood as a random factor, lysozyme activity at day 12 did not significantly covary with that at day 5 (F1,51=0.41, n.s). In the same analysis, laying order of the nestlings did not predict lysozyme activity on day 12 (F1,51=0.90, n.s), and no significant interaction existed between lay order and lysozyme on day 5 (F1,51=0.82, n.s).

Lysozyme activity of nestlings in size manipulated broods

We obtained data for 28 complete dyads (56 broods with 233 nestlings, 206 of which were sampled for lysozyme concentration). A mixed model analysis of variance with dyad as a random factor showed that nest of origin had a highly significant effect on nestling lysozyme activity (F55,88=2.51, P < 0.001), whereas no significant effect of brood size manipulation or its interaction with nest of origin existed (brood size manipulation: F1,88=0.00, n.s.; interaction: F55,88=0.68, n.s). Indeed, mean lysozyme activity of nestlings in reduced broods (= 3.75, 0.25 SE, n=28 broods) was strictly similar to that observed in enlarged broods (= 3.77, 0.22 SE, n=28 broods). However, brood size manipulation had a strong effect on offspring body mass (mixed model analysis of variance as for lysozyme data: effect of brood size manipulation: F1,106=39.65, P < 0.0001), with nestlings from reduced broods being heavier than those from enlarged broods. Hence, brood size manipulation affected nestling growth, consistently with previous experiments (see also Saino et al., 1997a, 1999), but not lysozyme concentration.


In this study, we showed that, in the barn swallow (i) male and female breeding adults have different temporal patterns of variation in lysozyme activity; (ii) lysozyme activity in the albumen and nestlings declines with egg laying order; (iii) clutch size and hatching failure correlate negatively with maternal, but not paternal, lysozyme activity; (iv) maternal lysozyme activity predicts lysozyme activity in the offspring; and (v) shared origin, but not altered brood size, affects nestling lysozyme activity. As the literature on lysozyme in birds concerns domestic species, and the information on environmental, early maternal and genetic components of variation in lysozyme activity is scant, present results are open to different interpretations. We will explore these interpretations both under the assumption that no adaptive mechanisms of regulation of lysozyme levels in mothers and offspring exist or, alternatively, that lysozyme levels are subjected to regulation mechanisms functioning to enhance egg and offspring viability.

Variation of lysozyme activity in parents and in eggs in relation to laying order

The decline of female lysozyme activity before egg laying could reflect variation in exposure to bacteria or represent a side-effect of physiological changes occurring before egg laying. Egg production is a costly activity that may require depression of immune function allowing for more resources to be allocated to egg formation. Reduced levels of lysozyme may therefore be one manifestation of the cost of reproduction. The lack of significant variation of paternal lysozyme supports the idea that change in female lysozyme activity is linked to egg laying.

One alternative interpretation is that mothers undergo a decline in lysozyme concentration before layingbecause lysozyme is accumulated in the reproductive organs and then passed into the eggs. Decline of lysozyme may thus be part of an adaptation functioning to ensure high immunological protection to the eggs.

In fact, eggs possess both physical mechanisms of defence against infection provided by the shell and shell membranes and the albuminous sac, and chemical defence by substances contained in the albumen. Lysozyme contributes to physical defence forming a fibrous network with other proteins in the albuminous sac, but also to chemical defence (Trziszka, 1994). Hence, large levels of egg lysozyme may allow for antibacterial defence during embryo development.

The positive correlation between proportion of eggs that hatched in a clutch and maternal lysozyme is consistent with the results of poultry studies and may suggest that egg lysozyme is part of a maternally derived mechanism of egg immune defence. Alternatively, this correlation may simply reflect an association between good female body conditions, resulting in highly viable eggs independently of their lysozyme concentration, and high levels of maternal lysozyme which are passively transferred to the albumen.

Mean lysozyme activity in the albumen was approximately 15 times larger than the absolute maximum (c.4.5HEL equivalents) of the polynomial function fittedto the female lysozyme data (Fig. 1), confirming observations on poultry, domestic ostriches(S.camelus,e.g. Bizzarri et al., 1999) andpheasants(Phasianuscolchicus; N.Saino, P. Dall'Ara, R. Martinelli and A. P. Møller, unpublished data) that lysozyme in the albumen is more concentrated than in maternal plasma. This fact argues in favour of the existence of an active mechanism of accumulation and transmission of lysozyme to the eggs.

The early start of the decline of lysozyme with respect to egg laying is puzzling, because a closer matching with laying would be expected. Unfortunately, the timing of biosynthesis of lysozyme which is later found in the eggs is unknown in birds. However, the overall lysozyme activity in each albumen is more than 20 times larger than that instantaneously present in the mother's plasma, assuming an average albumen mass of 1 g (unpublished data), a relative blood/total body mass ratio of 0.1 and an haematocrit of 0.5. Lysozyme may start to be accumulated in oviduct magnum well before egg laying because during laying mothers may not be able to sustain the production of the large amounts of the enzyme that are transferred to up to seven eggs laid at 1-day intervals.

Whether the decline in lysozyme is detrimental to maternal immunity is likely to depend on variation before the decline starts. In this respect, our analyses are not conclusive, as the polynomial function of breeding stage fitted to lysozyme data had a maximum at day −11.4 but a linear regression of lysozyme data between day −23 and day −12 was statistically nonsignificant.

The negative relationship between clutch size and lysozyme concentration may suggest that females are limited in lysozyme production. The decline of lysozyme concentration with laying order may thus simply reflect exhaustion of maternal lysozyme. However, a functional interpretation of this finding is also plausible. First laid eggs may be more vulnerable to infection because they remain in the nest for longer than last laid ones, because laying spread is larger than hatch spread (Saino et al., 2001b). In addition, first laid eggs may undergo more cycles of warming and cooling when the mother is in the nest to lay the other eggs, because females spend the night in the nest, or simply because of circadian variation in temperature. This may add to the risk of infection because bacteria might be ‘sucked’ into the egg when it cools (J. Blount, pers. comm.).

Lysozyme activity in offspring in relation to parental lysozyme

The positive correlation between lysozyme activity in the mother and offspring, together with the parallel pattern of variation of lysozyme activity in the eggs and in the offspring in relation to laying order (see also below), suggests that lysozyme in mothers and their eggs are positively correlated. This correlation would be predicted on the basis of poultry studies (e.g. Melek, 1977; Krykanov, 1982; Prusinowska et al., 2000). Hence, mothers that produce eggs with large amounts of lysozyme might accrue a benefit in terms of egg hatchability as well as offspring antibacterial defence after hatching (Bessarabov & Krykanov, 1986).

A direct test of the benefits of large lysozyme levels during the nestling stage is not possible for the barn swallow at present because of very low nestling mortality invariably recorded in our study population, lack of information on infestation by virulent bacteria (although data are now being collected), and very low natal philopatry preventing accurate estimates of offspring viability during their first year of life. However, bacteria can be extremely virulent to their hosts and high levels of antibacterial defence during the nestling stage may markedly enhance survival of the offspring during the critical post-fledging period.

An alternative interpretation of the positive association between lysozyme activity of mother and offspring is that members of a family have similar exposure to bacteria and thus activation of immune defence. The results do not corroborate the idea that environmental factors play a major role in determining lysozyme concentration in nestling barn swallows. If such effects were actually operating, then a correlation of lysozyme activity of the offspring with their fathers should also have emerged. However, this might not be a conclusive evidence of weak envrionmental effects because mothers have closer contact with the eggs and the nestlings soon after hatching. In addition, the high frequency of extra-pair paternity (Møller & Tegelström, 1997; Saino et al., 1997b) could have partly confounded the correlation between fathers and their biological offspring.

Weak environmental effects on lysozyme concentration are also suggested by the brood size manipulation experiment. In fact, nest of origin affected plasma lysozyme activity, whereas brood size manipulation, which could affect competition for food (Saino et al., 1997a), pathogen transmission, thermal conditions or stress, did not affect lysozyme activity. Hence, we found no evidence for environmental determinism of lysozyme activity in offspring. On the other hand, previous studies have shown that these factors affect lysozyme activity in poultry (Popova-Ralcheva et al., 1998; Swierczewska et al., 1998).

Lysozyme activity in offspring in relation to laying order

Variation of albumen lysozyme activity in relation to laying order was mirrored by a consistent difference in plasma lysozyme activity in the offspring 5 days after hatching. Mean within-brood lysozyme activity in the plasma of the nestlings correlated positively with that in their mother. These pieces of evidence are compatible with the idea that lysozyme circulating in the offspring isat least partly of maternal origin (Bessarabov & Krykanov, 1986), and larger concentration of lysozyme in first compared with last laid offspring results from larger amounts of maternal lysozyme in the first laid eggs. Larger concentration of lysoyzme in first laid offspring may contribute to establish a hierarchy of reproductive value among the progeny. First eggs laid in a clutch tend to hatch first, and this slight hatching asynchrony results in a stable size and body mass hierarchy within the brood (Saino et al., 2001b) which might reflect a hierarchy in reproductive value, because of a positive relationship between body mass and survival (Møller, 1994). High lysozyme concentration in hens is known to result in enhanced natural resistance of the progeny to infection (Bessarabov & Krykanov, 1986). Hence, larger lysozyme levels in particular nestlings could contribute to enhance survival of offspring with large reproductive value.

However, although the positive correlation between maternal and offspring lysozyme persisted also on day 12 after hatching, the difference between first and last laid nestlings was no longer present on day 12. In addition, no obvious change in lysozyme activity occurredbetween days 5 and 12. Post-natal variation in plasma lysozyme concentration has been shown to be complex and nonmonotonic in poultry chicks (Rosolowska-Huszcz, 1978). Lysozyme on day 12 may largely reflect autonomous production by nestlings depending on current infection. This speculation cannot be confronted with real data on bacterial infection, which are not available for barn swallows. In contrast, lysozyme activity on day 5may largely reflect maternally derived lysozyme, as bacterial infection is unlikely to elicit an immune response in a few days following hatching.

In conclusion, our findings do not allow to draw inferences about the existence of adaptive variation of circulating levels of lysozyme in the barn swallow, nor the existence of mechanisms of allocation of innate immune factors to the progeny. However, the observations we presented on sexual differences in lysozyme profile among the adults and the positive association between lysozyme activity in the mother, egg hatchability and lysozyme activity in the nestlings suggest that maternally derived lysozyme may affect offspring viability. As heritable variation exists in lysozyme profiles even in domestic birds where artificial selection is likely to have eroded genetic variance, natural selection may have promoted the evolution of adaptive allocation strategies of this immune factor. This speculation, however, will require experimental investigation of the existence of allocation strategies and the fitness benefits and costs of high lysozyme activity in both parents and offspring.