Effects of egg yolk testosterone on growth and immunity in a precocial bird


Tobias Uller, Department of Zoology, Göteborg University, Medicinaregatan 18, 405 30 Göteborg, Sweden.
Tel.: +46 (0)31 773 36 96; fax: +46 (0)31 41 67 29;
e-mail: tobias.uller@zool.gu.se


In oviparous vertebrates, maternal steroid allocation to eggs can have important fitness consequences for the offspring. However, elevated testosterone levels are not only associated with beneficial postnatal effects, such as enhanced growth and high social status, but may also entail costs by suppressing the immune system. In this study, testosterone levels in eggs of Chinese painted quail (Coturnix chinensis) were experimentally manipulated to evaluate its effects on growth and immunocompetence. Testosterone did not affect embryonic development, body size or growth during the first 20 days. However, elevated testosterone levels during embryonic development were immunosuppressive for chicks with inherently higher growth rate. Adaptive scenarios where only beneficial effects of increased testosterone levels are considered may therefore need to be re-evaluated.


Maternal effects occur when the phenotype of the offspring is influenced by that of the mother. Hormones are one of the proximate mechanisms whereby this can be achieved. Thus, information about the environment perceived by the mother is transferred, via steroid deposition, into a nongenetic influence on offspring performance. Accordingly, eggs from a variety of oviparous vertebrates contain measurable amounts of maternally derived steroids (e.g. Dickhoff et al., 1990; Schwabl, 1993; Bowden et al., 2001), and in mammals, maternal steroids can cross the placenta to the offspring (Winter et al., 1981). Prenatal steroid exposure does not only affect juvenile traits, but can also have consequences for individual reproductive success in adulthood (Clark & Galef, 1995), and thereby act as a hormonal link between generations.

Steroid hormone levels in bird eggs show a wide range of variation within a clutch (e.g. Schwabl, 1996a; Lipar, 2001), between successive clutches (Schwabl, 1996a), between females of the same species (Reed & Vleck, 2001) and between different avian species (Schwabl, 1997). The physiological mechanisms responsible for regulating steroid allocation have not yet been established, but there is a positive correlation between female hormone levels and yolk hormone levels (Schwabl, 1996a; but see Verboven et al., 2003), suggesting that passive leakage can account for much of the variation in testosterone content in eggs. Both photoperiod and social interactions have been demonstrated to influence the amount of steroids allocated (Schwabl, 1997), possibly by altering female hormone levels.

Positive post-hatching effects of androgens have been found in canaries (Serinus canaria) by enhancing begging behaviour, hatchling growth (Schwabl, 1996b) and social dominance among siblings (Schwabl, 1993). Contrary to the pattern found in canaries, experimentally increased levels of androgens in first-laid eggs, comparable with natural levels in later-laid eggs, delayed hatching and reduced nestling growth and survival rates in American kestrels (Falco sparverius; Sockman & Schwabl, 2000). Thus, the adaptive value of differential allocation remains elusive, and experimental manipulation of yolk hormone levels is required to resolve this issue.

Recently, many studies have focused on androgen-mediated trade-offs between immunocompetence and reproductive success (reviewed in Hillgarth & Wingfield, 1997). Androgen exposure can lead to increased susceptibility to infections by directly lowering the immunocompetence, possibly by suppressing the expression of disease resistant genes, decreasing energy allocation to the immune system, or by influencing behaviour (Hillgarth & Wingfield, 1997; Klein, 2000). Consequently, in the context of steroid allocation to eggs, testosterone allocation may not only be a costly maternal investment (Gil et al., 1999; Pilz et al., 2003), but may also result in suppression of the immune system in the offspring (Uller & Olsson, 2003).

To date, studies of prenatal testosterone exposure in birds have been directed towards altricial and semi-precocial species, in which the young hatch undeveloped and are completely dependent on parental care (e.g. Schwabl, 1993; Groothuis & Schwabl, 2002). Enhanced growth compared with siblings, possibly triggered by elevated androgen levels, is of vital importance for offspring survival. In contrast, precocial birds hatch well developed and are able to forage independently from their parents. Furthermore, almost all studies of yolk testosterone have been descriptive (but see Schwabl, 1996b; Sockman & Schwabl, 2000; Uller & Olsson, 2003), and experimental manipulations are necessary for understanding the potential adaptive value of differential maternal allocation from a cost-benefit perspective. In this study, testosterone levels in eggs of Chinese painted quail (Coturnix chinensis) were experimentally manipulated. Based on the above reasoning, we hypothesized that young from testosterone-manipulated eggs would have increased growth rate but suppressed immune response compared with control offspring.


The Chinese painted quail (C. chinensis) is the smallest species in the family Phasianidae (an adult male and female weight 45–50 and 50–60 g, respectively). They are easy to keep and breed under captive conditions and are therefore of good use in laboratory experiments (Tsudzuki, 1994).

Ninety quails were obtained from commercial breeders (N.V. Birds Trading Company, Lommel, the Netherlands). The quails were kept in pairs in cages (800 × 600 × 500 mm) covered with fishing net (mesh diameter 25 mm) to avoid escape. Food [Finch seed, Swedish mix (Natural Garden, Schoten, Belgium), eggfood (CèDé, Evergem, Holland), insectfood (Bogena, Ralte, Holland), vitamins and minerals (Natural Garden)] and water were available ad libitum. The quails were exposed to a 14-h light 10-h dark photoperiod with ambient temperature being 28 °C during the day and 22 °C during the night. Eggs were collected daily over a 2-week period (corresponding to a normal clutch of approximately 12 eggs), marked with cage and order in the laying sequence, and stored at room temperature. The eight last laid eggs from each pair were selected for the experiment. All eggs were weighed (to the nearest 0.01 g) and measured with a pair of callipers (total length, to the nearest 0.1 mm).

A testosterone suspension was prepared using 60 ng testosterone (4-androsten-17β-ol-3-one; Sigma Chemical Co., St Louis, MO, USA) per μL sterile peanut oil (see Schwabl, 1996a). Eggs were wiped with 70% ethanol and 5 μL of the solution was injected into four of the eggs (chosen randomly) of each clutch, using a 25 μL hamilton microsyringe. This dose corresponds to doses found in eggs of other bird species (Schwabl, 1996a). The remaining four eggs were injected with 5 μL peanut oil and functioned as controls. All injections were performed horizontally, through the top of the egg, into the yolk with the syringe being inserted equally deep into each egg. The hole in the shell was patched with glue (Super Attak; Loctite Sweden AB, Göteborg, Sweden). Eggs were incubated at 38.5 °C under humid conditions. The eggs were turned twice each day to prevent deleterious adhesion between embryo and shell membranes (Deeming, 1999). The relative humidity was increased 2 days before hatching. After hatching, chicks were kept in the incubator for 12 h and then transferred to cages (620 × 450 × 290 mm) equipped with a 40 W spotlight (painted red to reduce the illumination), which created a thermal gradient within the cage from 33 to 42 °C. Pots made of fired clay were cut into halves, and one half was placed in each cage to provide a shelter. All cages contained wooden chips as bottom substrate and a bowl of sand. Hatchlings were kept in pairs comprised of one testosterone-treated and one control chick, descending from the same parents, and raised according to breeding standards with continuous light and heat (Tsudzuki, 1994). Hatchlings with no sibling with the corresponding treatment available were excluded from the present experiment. On day 0, chicks were provided with water and high potency mash (BlåStjärnan, Göteborg, Sweden) by hand, to secure nutritional requirement until they found nourishment on their own. Food (Fågelfoder; Fågel start, Svenska Foder AB, Lidköping, Sweden) and water were available ad libitum.

Body mass was measured on day 0, 5, 10 and 20 (to the nearest 0.01 g) and tarsus length on days 0 and 20 (to the nearest 0.1 mm), using a pair of callipers. Body condition was obtained using the residuals of body mass regressed on tarsus length. Increase in mass was calculated as (mass at day x) − (mass at hatching). T-lymphocyte cell-mediated immune response was measured on day 10 using standard methods. The thickness of the left wing web was measured with a micrometer (Mitutoyo 7300) and 0.1 mg PHA (Sigma L8754) dissolved in 0.02 mL phosphate-buffered saline (PBS) was injected. The wing web was measured again 24 h later. The average of five measurements for each wing web was used to reduce measurement errors. The change in thickness provides a measure of the intensity of the cell-mediated immune response (Smits et al., 1999). To reduce handling-related stress, no PBS injection was performed into the right (control) wing as the response has been shown to be insignificant (Smits et al., 1999; Sofia Andersson, personal observations for C. chinensis). On day 20, chicks were transferred to the adult room. Sex was determined at 4–5 weeks of age by the presence or absence of male secondary sexual characteristics.


In total, 115 eggs from 35 families hatched, yielding an overall fertilisation and hatching success of 41%, with no difference between treatments (Fischer's exact test, n.s.). Testosterone treatment did not enhance embryonic development (mean time to hatching was 18.5 days for both testosterone-treated and control offspring). Furthermore, there was no difference in body size (mass, tarsus length) or body condition between treatments at the time of hatching (ancova with family and treatment, and the interaction between them as factors and egg size as a covariate; treatment: n.s. for all variables, Table 1). Using mean scores per female, an anova with treatment and sex and the interaction between them as factors, revealed no sex effects on size at hatching (mass: sex: F1,65 = 2.49, n.s., sex × treatment: F1,65 = 0.13, n.s.; tars: sex: F1,65 = 0.09, n.s., sex × treatment: F1,65 = 0.97, n.s.) or on mass increase (sex: F1,55 < 0.25, n.s., sex × treatment: F1,55 < 0.80, n.s., for all growth periods).

Table 1.  Mass and tarsus length at hatching, mass increase and T-cell-mediated immune response (PHA response), mean ± SE.
 Mass (g)Tarsus length (mm)Mass increase (g)PHA response
Day 0Day 0Day 5Day 10Day 20Day 10
Testosterone3.54 ± 0.0810.32 ± 0.093.25 ± 0.179.38 ± 0.2422.43 ± 0.3917.70 ± 1.21
Control3.59 ± 0.1110.30 ± 0.113.39 ± 0.159.30 ± 0.2722.46 ± 0.4418.41 ± 1.22

To look for effects on growth, we first ran partly nested ancova (controlling for size at hatching), incorporating only families containing replicate cages (i.e. more than one pair of siblings, Quinn & Keough, 2002). Testosterone treatment had no significant impact on mass increase for 5, 10 or 20 days (5 days: F1,9 = 0.41, n.s.; 10 days: F1,9 = 0.00, n.s.; 20 days: F1,9 = 0.00, n.s.; Table 1). To increase the number of families and thereby the power of the tests, we also reran the analysis incorporating all families represented by at least one pair of siblings (using mean values per family). However, this did not change the outcome of the result (paired t-tests, t23 < 0.75, n.s. for all growth periods).

An anova with family, treatment and the interaction between the two as factors showed no significant overall effect of testosterone treatment on cell-mediated immune response (family: F24,26 = 2.92, P < 0.05; treatment: F1,23 = 0.41, n.s.; family × treatment: F23,26 = 0.98, n.s.; Table 1). Furthermore, there was no significant difference between the sexes in immune response (anova on mean values per family, with sex, treatment and the interaction as factors, F1,56 = 1.78, n.s.) However, control offspring showed a significant positive correlation between growth and immune response, whereas testosterone-manipulated offspring showed no correlation (control: r = 0.60, P < 0.05; testosterone treated: r = 0.13, n.s.; Fig. 1). Separate analyses for chicks with high vs. low mass increase (i.e. above or below the mean of 9.4 g), showed that there was a significant effect of testosterone treatment for relatively fast-growing, but not slow-growing, chicks (anova, family: F19,18 = 1.62, n.s.; treatment: F1,18 =5.62, P < 0.05, and family: F20,14 = 2.63, P < 0.05; treatment: F1,14 = 0.07, n.s., respectively; Fig. 1).

Figure 1.

Correlation between mass increase and immunity in (a) control and (b) testosterone-treated offspring (illustrated with trendlines for ease of interpretation). See text for statistics.


In the present study, prenatal exposure to testosterone showed no positive post-hatching effects on growth in Chinese painted quail juveniles. This contradicts results found in canaries, where testosterone enhanced growth (Schwabl, 1996b). The difference between studies may be a consequence of different feeding strategies between the two examined species. In contrast to precocial species, altricial hatchlings, such as canaries, are depending on food begging to secure nutritional requirements. Two studies of altricial species have found a positive correlation between testosterone concentration and the development of the hatching muscle (musculus complexus, Lipar & Ketterson, 2000; Lipar, 2001). The hatching muscle has a positive effect on dorsal flexion and extension of the neck during begging. Thus, an increase in the mass of the hatching muscle could be favourable in the competition for parentally delivered food in altricial nestlings. In contrast, juvenile quail can feed independently from their parents. Furthermore, in our study, food was provided ad libitum, making competition between siblings less severe. However, because there was no difference in time to hatching, body size or body condition between treatments, and no evidence of agonistic behaviour between siblings, it is uncertain whether limited resources would have favoured testosterone-manipulated chicks. In broiler chickens, yolk testosterone had a negative impact on prehatching weight in female, but not male, embryos (Henry & Burke, 1999). In contrast, we found no sex effects or interaction between sex and treatment on size at hatching or growth rate.

Elevated testosterone levels are associated with an increase in metabolic rate, which causes an increase in free radicals and consequently oxidative stress (Råberg et al., 1998). Many studies have found negative effects of testosterone on immunocompetence in adult animals (e.g. Olsson et al., 2000; Peters, 2000; Casto et al., 2001; but see Hasselquist et al., 1999; Evans et al., 2000; Lindström et al., 2001 for studies with no effects). Furthermore, differences in androgen exposure during development are likely to be partly responsible for differences in immune function between males and females (Martin, 2000). Thus, we would expect a decrease in immunocompetence in testosterone-exposed chickens. There was no overall effect of prenatal testosterone on the immune response in our study. When body mass increase was taken into account, however, the results suggest that, for fast-growing offspring, yolk testosterone had a negative effect on cell-mediated immunity (Fig. 1). Because testosterone did not increase growth, there was no evidence of a testosterone-mediated trade-off between growth and immunity, but rather that chicks with an inherently higher growth rate suffered from relatively depressed immune function under high steroid exposure. This could result if testosterone has detrimental effects on the immune system which only becomes evident if most energy is allocated to growth and relatively little to immune function. However, other traits correlated with high increase in body mass could also be responsible for the lack of positive relationship between growth and immunity in testosterone treated chicks. In domestic chicken (Gallus gallus), experimentally exposed to elevated prenatal testosterone levels, the bursa of Fabricius failed to mature, resulting in lower numbers of B-lymphocytes, i.e. suppression of the humoral immune response (Glick, 1986). Thus, there are indications of suppressive effects of testosterone on both cell-mediated and humoral immune response in precocial chickens.

Testosterone does not act on receptive tissue itself, but is first enzymatically converted to oestradiol or dihydrotestosterone (Hillgarth & Wingfield, 1997). Thus, the response to testosterone manipulation could be dependent on the derivative used in the experiment. In the present study, we used the testosterone derivative testosterone (found in relatively high concentrations in egg yolk; Schwabl, 1997; Schwabl et al., 1997; Reed & Vleck, 2000). Thus, excess testosterone may be metabolized or converted to other steroids, which may therefore lead to a weaker response than, for example, by using dihydrotestosterone (see Uller & Olsson, 2003). To date, there are not enough studies aimed at experimentally investigating effects of prenatal steroid exposure to understand the potential differences in response based on differences in testosterone derivative.

In conclusion, injection of testosterone into Chinese painted quail eggs had no effect on hatchling body size, growth or body condition, or overall immune response. However, fast-growing, testosterone-manipulated hatchlings showed impaired cell-mediated immune response compared with fast-growing controls. These results indicate that there is an immunological cost associated with steroid allocation, which may override any positive post-hatching effects.


Two anonymous reviewers provided valuable comments on this manuscript.