Determinants of within- and among-clutch variation in levels of maternal hormones in Black-Headed Gull eggs

Authors

  • Groothuis T. G.,

    Corresponding author
    1. Department of Animal Behaviour, Biological Centre, University of Groningen, PO Box 14, 9750 AA, Haren, the Netherlands, and
      †Author to whom correspondence should be addressed. E-mail: T.Groothuis@biol.rug.nl
    Search for more papers by this author
  • Schwabl H.

    1. Center for Reproductive Biology, School of Biological Sciences, Washington State University, Pullman, WA 99164–4236, USA
    Search for more papers by this author

†Author to whom correspondence should be addressed. E-mail: T.Groothuis@biol.rug.nl

Summary

1. Females of egg-laying vertebrates may adjust the development of their offspring to prevailing environmental conditions by regulating the deposition of hormones into their eggs. Within- and amng-clutch variation in levels of steroid hormones were studied in the egg yolks of the Black-Headed Gull (Larus ridibundus, Linnaeus) in relation to environmental conditions at the nest site. This species breeds in colonies of different densities and in different habitats, and the chicks hatch asynchronously.

2. Egg yolks contained very high levels of androstenedione, substantial levels of testosterone and moderate levels of 5α-dihydrotestosterone. Oestrogen (17β-oestradiol) was not detected.

3. Androgen levels increased strongly with laying order, irrespective of egg or yolk mass. This may compensate for the disadvantages of the later hatching chicks. These results have implications for adaptive hypotheses that were proposed for asynchronous incubation.

4. Eggs of lighter clutches contained more androgens, perhaps to compensate for a lower nutritional quality of these eggs.

5. Birds breeding in the periphery of a colony, being relatively more aggressive and having relatively large territories, laid eggs that contained more androgens than those of birds breeding in the centre. These high yolk androgen levels may facilitate growth and motor development of the chicks, which may be especially important for chicks developing at the periphery of a colony. Reduced levels may be adaptive for birds breeding in the centre, where risk of infectious diseases is high, since steroids may be immunosuppressive.

6. Corrected for nest distance, clutches of birds in high vegetation, where predation risk is less severe and therefore competition for nest sites perhaps high, contained relatively high levels of androgens. It is suggested that the level of yolk androgens reflects the hormonal condition of the female, that in turn is influenced by her characteristics such as her age and aggressiveness, and the level of social stimulation.

Introduction

Apart from transfer of genetic material parents may maximize gene propagation via non-genetic modifications of their offsprings’ phenotypes. For example, females of egg-laying vertebrates may transfer substances to their eggs that modify development and thereby the survival of their offspring. The advantage of such maternal influences could be that they are flexible, allowing the mother to vary them with the prevailing conditions that are likely to be met by her young early or later in life. Transfer of maternal steroid hormones to the developing embryo might play an important role here. It is well known that early exposure to these hormones profoundly influences development, not only leading to differentiation of the sexes (Balthazart 1997; Schlinger 1998), but also to phenotypic variation among individuals of the same sex (Clark & Galef 1995; Crews 1998; Moore, Hews & Knapp 1998).

In birds the concentrations of yolk steroids vary within a clutch as well as among clutches. The functions of within-clutch variation in yolk steroid levels have received most of the attention so far. In almost all bird species studied to date yolk hormone levels varied with laying order. In most species androgen levels increase over the laying sequence (Canary, Schwabl 1993; Red-Winged Blackbird, Lipar, Ketterson & Nolan 1999a; American Kestrel, Sockman & Schwabl 2000; Common Tern, French, Nisbet & Schwabl 2001); in some others it decreases or remains constant (Cattle Egret, Schwabl, Mock & Grieg 1997; Zebra Finch, Gil et al. 1999; Tree Swallow, Whittingham & Schwabl 2002). Since in canaries yolk levels correlated with the social rank of the offspring later in life and elevation of yolk hormone levels resulted in enhanced growth (Schwabl 1993, 1996a), increased levels of yolk androgens are thought to be beneficial for the chick. It is generally assumed that by allocating more hormones to the later laid eggs of a clutch mothers can counteract the disadvantages of the later hatching chicks in sibling competition (Schwabl 1993, 1996a). In contrast, by depositing more hormone in first laid eggs mothers of siblicidal species may enhance the effects of hatching asynchrony facilitating facultative brood reduction by the oldest chicks (Schwabl et al. 1997). However, this interpretation is not entirely in line with recent data. The generally low yolk levels of androgens decline slightly over the laying sequence in the Zebra Finch (Gil et al. 1999), a non-siblicidal species with a similar degree of hatching asynchrony as the Canary. In Tree Swallows no relationship between laying sequence and yolk androgen levels was evident (Wittingham & Schwabl 2001) and in American Kestrels yolk androgens handicapped chicks by delaying hatching and growth (Sockman & Schwabl 2000). It is possible that the levels of hormones that females deposit in their eggs depend on other variables that also influence offspring survival, such as egg quality, female age or experience, social competition and food availability. This called for a field study integrating the causes of both within- and among-clutch variation in yolk hormone levels, hatching asynchrony, and ecological conditions in one and the same species.

The causes and functions of among-brood variation in within-brood patterns as well as in absolute levels of yolk steroids have received far less attention than within-brood variation. There is some evidence for influences of environmental factors on yolk androgen levels. For example, in House Sparrows, yolk testosterone contents and breeding density were positively correlated (Schwabl 1997a); in Tree Swallows, yolk testosterone concentrations were correlated with the number of intrusions into the territory (Wittingham & Schwabl 2001); and in Canaries yolk testosterone levels increased with the progress of the breeding season (Schwabl 1996b). In many bird species social stimulation by the same or opposite sex increases androgen production (e.g. Wingfield et al. 1990; Beletsky, Ovian & Wingfield 1992) and female circulating levels of androgens during yolk formation have been shown to correlate with yolk hormone levels in the egg (Schwabl 1996a). Thus there is the intriguing possibility that the deposition of maternal steroid hormones into eggs is influenced by the mother’s social experience. In this way the female may be able to prepare the phenotype of her offspring for the level of social competition after hatching or fledging.

We studied ecological and social factors that may influence the within- and among-clutch variation of steroid hormone concentrations in eggs of the Black-Headed Gull (Larus ridibundus, Linnaeus). Gulls have a modal clutch size of three eggs that are usually laid and hatch at more than 1-day intervals and the effects of laying order and hatching asynchrony on chick growth and survival are well studied (e.g. Parson 1975; Royle & Hamer 1998; Royle 2000). Thus they are ideal models to test functional hypotheses of the variation of yolk steroid levels with laying order. Furthermore, gulls breed in colonies that vary substantially in nest density and vegetation. Both factors have been shown to influence the number and quality of social interactions (Bukacinska & Bukacinski 1993). Females have relatively high blood plasma levels of androgens during egg formation (Groothuis & Meeuwissen 1992; Malickiene 1999), which are influenced by social stimulation (Groothuis 1992). This makes them ideal for studies of the effect of habitat and behaviour on the deposition of maternal hormones into eggs. Finally, chicks are very aggressive and defend space around the nest with behaviour that depends on androgens (Groothuis 1989; Groothuis & Meeuwissen 1992, Ros 1997). Therefore maternally derived androgens may have important functions not only for sibling competition, but also for nest defence and competition among broods in this species. This provides an excellent model system to study the causes and function of among-clutch variation of maternal steroids in the yolk.

Methods

Egg collection and habitat measurements

Eggs were collected in 1996 and 1997, with permission from the landowners and under government and the university licences. The 1996 study was explorative. Eggs were collected from nests that just had one single egg when first encountered in order to select eggs of incomplete clutches that were not yet incubated. In some cases eggs were collected only when present in nests that had been empty the previous day. In the first subset some eggs contained recognizable embryos. These single eggs may not have been first laid eggs owing to predation and were excluded from the analysis and only the second subset was used for some of the further analyses. Several colonies along the north-east coast of the Netherlands and one colony on the island of Griend just in front of the north coast were used. From each colony eggs were taken from nests located in the centre, where nest densities where high, and from the periphery, where the nest densities were much lower.

In 1997 collections were made only in the colonies at the coast in which nests were monitored daily. We collected, at the day of clutch completion, 10 full clutches of which the laying sequence of each egg was known, and 6 clutches of which the position in the laying sequence was known of at least one egg (sometimes two eggs were laid within the 24-h interval). Only clutches in which all three eggs had a similar shape and colour were used, to avoid the possibility that a clutch contained eggs of different females (egg dumping has occasionally been observed in this species in captivity, Van Rhijn & Groothuis 1985). First, second and third laid eggs will be referred to as a-, b- and c-eggs, respectively.

In 1997, nest density and vegetation height around the nest were measured, environmental variables that probably influence the number of social interactions between breeding pairs. Breeding density was estimated as the sum of the distances of a nest to the three nearest neighbour nests. Vegetation height and composition around nests were estimated and then nests were categorized into three classes. Class 1 nests were located in very short vegetation, mainly grass species, a few cm long. Class 2 nests contained vegetation around the nest of about 30 cm on average in height, consisting of grasses and Artemisia maritima. Class 3 nests were in vegetation that was substantially higher, up to 100 cm, and often consisted of grass species and Halimione portuclacoides. To avoid confounding influences of laying date on yolk androgen levels all clutches were collected within 1 week over the 6-week laying period.

Egg analyses

After collection eggs were weighed to the nearest 0·01 g in the laboratory and stored at –20 °C until they were shipped to the laboratory of HS on dry ice for further analyses.

Yolks were separated from albumin and weighed to the nearest 0·01 g. In some cases only a fraction of the whole yolk could be obtained. Since yolk and total egg mass were positively correlated (Pearson correlation, R = 0·365, P = 0·026, n = 37) egg mass was used as covariate in the general linear models (GLMs).

The procedures for extraction and radioimmunoassays of testosterone (T), androstenedione (A4), 5α-dihydrotestosterone (DHT) and 17β-oestradiol (E2) from yolk homogenates have been described previously (Schwabl 1993). Intra-assay coefficients of variation for the 1996 samples were 4·5% for A4, 5·2% for DHT, 5·4% for T% and 4·5% for E. Mean recoveries were 64% for A4, 45% for DHT, 56% for T and 59% for E. Since DHT and E were present only in very low concentrations, only T and A4 were analysed in the 1997 samples. Intra-assay coefficients of variation were 5·5% for A4 and 4·7% for T. Mean recoveries were 67% and 63%, respectively.

Statistical analyses

Parametric tests were used for statistical analyses. The data did not deviate from a normal distribution except for T levels in 1996, which were therefore ln-transformed. For indication of the effect size of independent factors and covariates in the GLMs, we present b-values, estimates of the slope in the regression.

Results

Mean androgen levels

All three androgens were clearly present in the eggs, with especially very high concentrations of A4. E2 was not detectable with our assay (detection limit 1 pg mg−1 for 150 mg yolk samples) (Table 1). Androgen concentrations in 1997 were slightly higher than those in 1996. T and A4 levels were positively correlated, while DHT did not correlate with the other two androgens (Table 2).

Table 1.  Levels of gonadal steroids (mean and standard error in pg/mg yolk) detected in yolk of Black-Headed Gull eggs in two different years
YearT (N)DHT (N)A4 (N)E (N)
19967·10 ± 1·11 (46)5·03 ± 0·47 (29)194·42 ± 14·19 (46)Not detectable (29)
19979·99 ± 0·87 (48)Not measured270·02 ± 20·15 (48)Not measured
Table 2.  Correlations (Pearson R and significance) between androgen levels in yolk of Black-Headed Gull eggs in two different years
Year T-A4 (N)T-DHT (N)A4-DHT (N)
1996R 0·293 (46)−0·143 (29)0·07 (29)
P 0·048 0·460·73
1996R 0·598 (48)  
P<0·001  

Within-clutch variation

Concentrations of T (Fig. 1) and A4 (Fig. 2) increased with laying order (repeated measures anova for T: F2,18 = 21·119, P < 0·001; for A4: F2,18 = 51·063, P < 0·001). The increase from the a- to the b-egg was significant for both hormones (paired t-test for T: T9 = 9·0, P < 0·001, resp. A4: T9 = 7·7, P < 0·001). The increase from the b- to the c-egg was not significant for T (T9 = 1·3, P = 0·227), but significant for A4 (T9 = 3·30, P < 0·009). T concentration in one c-egg was extremely low, even lower than the lowest value measured in a-eggs. When this outlier was removed from the analysis, the increase in T from the b- to the c-egg was significant (T8 = 2·74, P < 0·024).

Figure 1.

Testosterone concentration in yolk of first, second and third laid eggs of Black-Headed Gulls. Lines connect values of the same clutch.

Figure 2.

Androstenedione concentration in yolk of first, second and third laid eggs of Black-Headed Gulls. Lines connect values of the same clutch.

The increase in T concentrations was higher from the a- to the b-egg (5·40 ± 0·60) than from the b- to the c-egg (1·76 ± 1·36). However, this difference was not significant (T9 = 2·22, P = 0·053; T8 = 1·840, P = 0·103 without the outlier). For A4 this difference in increase between eggs was significant (147·37 ± 19·19 resp. 57·87 ± 17·56, T9 = 3·38, P = 0·008).

Total egg mass varied with laying sequence in a quadratic fashion (repeated measures anova, F2,18 = 8·861, P = 0·003), with the b-egg being heavier than the a- or c-egg (a-egg: 38·3 ± 0·7; b-egg: 40·2 ± 0·9; c-egg: 39·1 ± 0·7: a- vs b-egg: T8 = 3·563, P = 0·006; a- vs c-egg: T8 = 1·762, P = 0·112, b- vs c-egg: T8 = 3·403, P = 0·008). Since this within-clutch pattern in egg mass differs from that in hormone concentrations, the latter cannot simply be explained by heavier eggs having a lower hormone concentration while the total hormone content of the eggs remains constant. Indeed, the effect of laying order on T and A4 levels was also significant when the concentrations in each egg were multiplied by the mass of that egg (T: F2,18 = 21·426, P < 0·001; A4: F2,18 = 50·053, P < 0·001).

Between-clutch variation

A-eggs: data of 1996

For the statistical analysis we used GLM with hormone concentration as dependent variable, location (central vs periphery) as independent factor and egg mass as covariate. Only data for eggs that were probably first eggs (found in previously empty nests, see method section) were used. For ln-transformed T concentrations this yielded a significant model (F1,9 = 4·759, P = 0·030, adjusted R2 = 0·484, n = 9 (periphery) and n = 4 (central)) with significant effects of both location and egg mass and their interaction (all P-values < 0·031). T concentrations were lower in the centre (median: 4·28 pg mg−1 yolk, range: 1·84–10·39) than in the periphery of the colony (6·63, 4·79–47·76). Smaller eggs contained more T (b = –0·395). A similar, but not significant effect was found for A4 in the same sample, while DHT showed a nonsignificant opposite trend in a very small sample (n = 5 and n = 3).

Complete clutches, data of 1997

The mean hormone concentration in each of 16 clutches was used as the dependent variable in a backward stepwise regression in a GLM, with vegetation height as independent factor (three levels) and nest distance and clutch mass as covariates. All possible interactions between the factor and the two covariates and between the covariates were included in the model. This revealed a significant model (F4,11 = 8·063, P = 0·003, adjusted R2 0·653) with a strong effect of vegetation (F2,11 = 5·874, P = 0·018, b = – 6·129 for low, b = 5·164 for intermediate and b = 0 for high vegetation). Nest distance (F1,11 = 7·451, P = 0·030, b = 0·004) and clutch mass (F1,11 = 21·853, P = 0·001, b = –0·689) also had significant effects. Figure 3 shows that the T concentrations, when corrected for the effect of clutch mass, increase with increasing nest distance, while T concentrations in low vegetation are relatively low. Figure 4 shows that the T concentrations, when corrected for the effect of nest distance, decrease with increasing clutch mass. This figure also depicts that total clutch mass was lower in lower vegetation. A GLM with clutch mass as dependent variable and vegetation and nest distance as predictors revealed a significant model (F3,12 = 5·648, P = 0·012, adjusted R2 = 0·482) with a strong effect of vegetation (F2,12 = 7·790, P = 0·007) and a trend for nest distance (F1,12 = 3·326, P = 0·093).

Figure 3.

Effect of nest distance and vegetation on concentrations of testosterone in eggs of the Black-Headed Gull. Hormone levels are means of the three eggs of complete clutches and are corrected for the effect of clutch mass on these levels as estimated in a general linear model. Low (▴), intermediate (▪), and high (•) vegetation.

Figure 4.

Effect of total clutch mass and vegetation on concentrations of testosterone in eggs of the Black-Headed Gull. Hormone levels are means of the three eggs of complete clutches and are corrected for the effect of nest distance on these levels, as estimated in a general linear model. Low (▴), intermediate (▪), and high (•) vegetation.

A similar model was run for the mean concentrations of A4. The backward procedure yielded a significant final model with an effect of vegetation only (F2,13 = 10·342, P = 0·002, adjusted R2 = 0·555; b for low vegetation: 450·720, intermediate: 136·888, high: 0). Thus, in contrast to T, A4 levels increased with lower vegetation. However, in this model the effect of vegetation was not corrected for nest distance, since nest distance was removed from the final model, being non-significant. Unfortunately, both factors are negatively correlated (Rs = –0·865, P < 0·001, n = 16), meaning that nests were further apart in lower vegetation. Since both factors did not affect A4 differently, as in the case of T, it is impossible to decide which factor is the biological relevant one. When vegetation height was substituted for nest distance, the latter was highly significant too (F1,14 = 9·890, P = 0·007, adjusted R2 = 0·372, b = 0·179). The data are plotted in Fig. 5.

Figure 5.

Effect of nest distance and vegetation on concentrations of androstenedione in eggs of the Black-Headed Gull. Hormone levels are means of the three eggs of complete clutches. Low (▴), intermediate (▪), and high (•) vegetation.

To analyse whether the increase in yolk androgens with laying order in a clutch was related to habitat, nest distance and vegetation were used as covariates in two separate ancovas with laying sequence as repeated factor. In none of these four tests was the interaction between the covariate and laying order significant.

Discussion

Types and levels of maternal steroids in black-headed gull egg yolk

All three main androgens, T, A4 and 5α-DHT, were detected in the yolk of eggs of the gulls. No oestrogen was detected, despite the presence of elevated levels of this hormone in egg-laying females (60 pg ml−1, Malickiene 1999), and the sensitivity of our assay (detection limit 1 pg mg−1 yolk) while the same assay detected (low levels of) oestrogen in eggs of other avian species (Schwabl 1993; French et al. 2001). Since the eggs were collected before the onset of incubation, these hormones must be of maternal origin. The somewhat lower levels in 1996 relative to 1997 might be explained by the fact that in 1996 only a-eggs were collected, which have lower levels of androgens than b- and c-eggs while in 1997 b- and c-eggs were also collected. All four steroids have been detected in the eggs of other species when assaying for them (T, A4, DHT and E2 in Canary, Schwabl 1993; T and E2 in Zebra Finch, Schwabl 1993; E2 in Quail, Adkins-Regan, Ottinger & Park 1995; T in House Sparrow, Schwabl 1997; T, DHT and A4 in Cattle Egret, Schwabl et al. 1997; T and DHT in Zebra Finch, Gil et al. 1999; T and E2 in Red-Winged Blackbird, and T in Dark Eyed Junco, Lipar et al. 1999b; T, DHT and A4 in American Kestrel, Sockman & Schwabl 2000; T and DHT in Lesser Black-Backed Gull, Royle, Surai & Hartley 2001; T in Tree Swallow, Whittingham & H. Schwabl 2002; A4, DHT, T and low levels of E2 in Common Tern, French et al. 2001). It seems that in small altricial birds T concentrations are much higher than those of A4, while the reverse seems to be the case for bigger semi-altricial or precocial species. It is possible that the larger eggs of precocial species, which have a longer incubation time than those of altricial birds, require relatively more testosterone, and that part of the amount is provided by the mother in the form of A4 as a T precursor and reservoir. In this way the embryo itself would be able to convert A4 into T, thereby avoiding possibly toxic high levels of T at the beginning of development. Indeed, detrimental effects of high T levels early in development have been demonstrated for the Black-Headed Gull (Groothuis & Meeuwissen 1992).

Not surprisingly, yolk levels of A4, the precursor for T, were correlated with levels of T. However, DHT did not correlate with the two other androgens. Furthermore, in contrast to A4, T concentrations were related not only to vegetation height but also to nest distance. This suggests that habitat characteristics influence activity levels of enzymes that control the pathways of androgen production in the female.

Within-clutch variation in androgens: Are high levels of androgens beneficial or detrimental to the chicks?

Concentrations of T and A4 clearly increased with laying order. Such a pattern has been found in several other, but not all, bird species. This increase is generally interpreted as a compensatory maternal mechanism for the disadvantages of hatching asynchrony (see Introduction). This interpretation contradicts the idea that hatching asynchrony, in creating disadvantages for the last hatched chicks, is advantageous by facilitating brood reduction in case food supply is not sufficient for the entire brood (for a review see Stoleson & Beissinger 1995). There are several possible explanations for this apparent contradiction. First, the assumption that yolk hormones are always beneficial for the chicks may not be valid. Indeed, in the American Kestrel, for example, experimentally increased yolk androgen levels (A4 and T) resulted in delayed hatching, retarded growth and decreased nestling survival (Sockman & Schwabl 2000); in the Canary, in contrast, elevated T was beneficial for the chick, as it increased begging behaviour and growth during early development (Schwabl 1996a). Moreover, yolk androgen levels were positively correlated with social rank later in life (Schwabl 1993). In addition, manipulation of the effective yolk testosterone levels by injection of T or a T antagonist, respectively, increased and decreased the size of the musculus complexus in Red-Winged Blackbirds (Lipar & Ketterson 2000). This muscle is important for hatching as well as begging (Schwabl & Lipar, 2002). Thus, the results with Red-Winged Blackbirds suggest that increased concentrations in yolk testosterone levels in later laid eggs likely leads to shorter hatching intervals, thereby diminishing the size and age handicaps of later hatching chicks. Furthermore, in the Black-Headed Gull and in other species, all eggs contain substantial levels of androgens, but it seems unlikely that the mother will counteract the survival of all her offspring. It is possible that beneficial vs detrimental effects depend on hormone dose within a narrow range, becoming detrimental above a certain level; or effects may depend on other characteristics of eggs that can vary with laying order including the sex of the embryo. Important for our considerations here, we have experimental evidence for beneficial effects in the Black-Headed Gull. Higher yolk androgen levels resulted in earlier hatching and increased growth rate over more than 3 weeks after hatching (Eising et al. 2001). This suggests that in the Black-Headed Gull, just like in Canaries and Red-Winged Blackbirds, yolk androgens function to at least in part mitigate the handicaps resulting from asynchronous incubation and hatching asynchrony.

If hatching asynchrony and androgen allocation were both adaptive mechanisms to optimize reproductive output then the increase of androgen levels in eggs with laying order should depend on the prevailing food conditions and the ability of the parents to find food, and occur only under favourable conditions. In the present data set, as well as in data collected in a year with relatively poor food conditions (C. M. Eising et al., unpublished results) virtually all clutches showed this increase in maternal hormones, suggesting that condition-dependent adjustment in clutch androgen allocation patterns is unlikely. It has been suggested (Royle et al. 2001) that antioxidant capacity may vary with laying order depending on food conditions, and would maximize reproductive success in conjunction with variation in androgen levels. However, experimental testing of this suggestion and its underlying assumptions has not yet been undertaken.

Finally, the function of hatching asynchrony in Black-Headed Gulls might not primarily be to facilitate brood reduction. Many parents may usually be capable of rearing only two chicks and the third egg may serve as an insurance against hatching failure of the a- and b-egg (Stoleson & Beissinger 1995; Forbes et al. 1997). Elevated levels of T in the yolk may then help to overcome the disadvantage of delayed hatching of this chick during the first days of life, but not thereafter. The ‘insurance chick’ will only survive when one other sibling is present, but will die soon if more than one chick has already hatched. This would explain why in gulls the last hatched chick in complete broods has a much higher mortality than its siblings (Parson 1975; Graves, Whitten & Henzi 1984; Sydeman & Emslie 1992; Royle 2000). This is indeed the case in the Black-Headed Gull, where c-chicks do survive almost only in broods of two, despite elevated levels of yolk T in their eggs (Eising et al. 2001).

Hatching asynchrony may be an inevitable consequence of the start of incubation before clutch completion, which may function to prevent predation or reduction in egg viability (Webb 1987; Veiga 1992; Vinuela 2000). This is especially likely for open field breeders such as gulls, in which the eggs are frequently predated and not protected against influences of sun radiation. In that case the increase of androgens over the laying sequence may be adaptive to counteract the inevitable consequence of hatching asynchrony resulting from egg protection.

Discussions of the function of within-clutch variation of yolk androgens have been focusing on the consequences for the chick that will hatch last. However, not only c-chicks but also b-chicks have much higher levels than a-chicks. In fact the increase in hormone levels from the a- to the b-egg is much stronger than from the b- to the c-egg. One would expect the opposite if these hormone levels function to reduce sibling competition, since the hatching interval between a- and b-egg is equal or even less than between the b- and c-egg (Glutz von Blotzheim & Bauer 1982). This suggests that there is a limitation in the amount of androgens the mother may be able to deposit in the c-egg. There might be physiological constraints in the mother, or it is possibly an adaptation to prevent potential detrimental effects to the embryo of yolk androgen levels above a certain threshold (see above).

The fact that even a-eggs contain quite substantial levels of androgens suggests that these hormones do not only fulfil a function in relation to hatching asynchrony. The hormones might adjust the development of all chicks to the prevailing conditions in which they will be raised. This hypothesis can be investigated on the basis of between-clutch variation in hormone levels.

Ecological and behavioural correlates of among-clutch variation

Birds breeding in low vegetation or low density (these two factors are strongly correlated) produced relatively light eggs. Previous studies indicate that younger Black-Headed Gulls are more likely to breed in the periphery of the colony (Patterson 1965) and egg size in gulls increases with age (Sydeman & Emslie 1992). Thus, it is possible that the difference in yolk androgen levels between clutches at the periphery or the centre reflects the age of females. Interestingly, not only T concentration but also total T content of the eggs correlated negatively with clutch mass. This suggests that young inexperienced females may compensate for the lower ‘quality’ of their eggs by means of greater androgen deposition.

High vegetation correlated with low nest distances, in line with earlier studies that indicate that low visibility reduces the level of competition and permits the birds to nest closer together (Bukacinska & Bukacinski 1993). Despite this correlation both factors influenced T yolk levels independently. Large nest distances correlated with high T, as well as with high levels of A4. This contrasts with results of earlier studies that have suggested that high nest density and high levels of social stimulation enhances T production in the female and thereby T levels in her eggs (Schwabl 1997; Whittingham & Schwabl 2001). However, it meets the observations in Black-Headed Gulls that birds that breed in the periphery of the colony are much more aggressive than those that breed at the centre and that aggressiveness may be the cause of high nest distances in the periphery (Patterson 1964). As in many species, high levels of aggressiveness in Black-Headed Gulls correlate with high levels of testosterone (Groothuis 1989; Groothuis & Meeuwissen 1992), which may then translate into high T levels in the eggs (Schwabl 1997). Predation rate in the periphery of colonies is high (Patterson 1964) and increased T levels in the yolk of the eggs of these birds may function to enhance the locomotor development of chicks (Schwabl 1993), allowing them to run faster for cover when a predator approaches. This is less important in high-density areas that are located in high vegetation. In addition, low T levels in the centre might be beneficial because high levels may suppress immune function (Peters 2000). Moreover, the risk of infectious diseases is high for gull chicks (Hario & Rudback 1999), and might be much higher in high-density central than low-density peripheral areas of a colony.

Clutches in high vegetation tend to have higher T levels than clutches in low vegetation, independent of nest distances. It might be that competition for nest sites in high vegetation is very intense. These sites may be desirable because of their much lower nest predation due to a lot of cover.

In conclusion, we propose that egg mass, as a measure of quality, and nest location in relation to habitat, perhaps through individual characteristics of the female such as her age, explain a substantial portion of among-clutch variation in yolk androgen levels. We would like to emphasize, however, that there is a need for more ecological studies before we will be able to fully understand functions and limitations of intra- and interclutch variation in yolk androgens.

Acknowledgements

We like to thank Dr H. Visser for help in transporting eggs and Dr M. Loonen for statistical advice.

Ancillary