Male ornamentation, timing of breeding, and cost of polygyny in the collared flycatcher

Abstract

Highly ornamented males are often thought to be better able to provide females with resources, parental assistance, or good genes. Individual variation in such male abilities may override the costs of polygyny and therefore largely explain within-population variation in mating patterns. We investigated the influence of variation in male ornamentation and the environment on the costs of polygyny for female collared flycatchers (Ficedula albicollis), using data from a long-term study involving 2733 breeding attempts over 19 years. We show that females suffer reduced reproductive success when mated polygynously but that the costs of polygyny depend on an interaction between male ornamentation and timing of breeding. Among early breeders, polygynously mated females experience higher reproductive success when mated to less ornamented males, but among late breeders, females mated polygynously to highly ornamented males were more successful. We suggest that a high effort spent on obtaining extrapair matings early in the season renders highly ornamented males less able to assist two females in caring for the young. Thus, a male's ability to simultaneously gain from extrapair matings and polygyny may be limited through direct effects on female reproductive success. Given such limitation, extrapair matings may be expected to be less frequent in species with biparental care and a high level of social polygyny.

Mating patterns (or systems) and parental roles result from the coevolution of reproductive behaviors of males and females against an ecological setting (Davies, 1991; Emlen and Oring, 1977). Sexual conflict over mating patterns and parental roles can result in evolutionary changes where morphology, behaviors, and life-history characteristics of either or both sexes are driven away from their optima (e.g., Ahnesjö et al., 1993; Alonzo and Warner, 2000; Holland and Rice, 1999; Houston and Davies, 1985; Lessells, 1999; Svensson and Sheldon, 1998; Westneat and Sargent, 1996). The evolutionary outcome (i.e., observed mating pattern and parental roles) depends, at least in part, on the relative strength of selection operating on males and females. When internal (e.g., individual quality) or external factors (e.g., food availability or breeding synchrony) cause fluctuating selection pressures, the evolutionarily stable outcome may be plasticity in reproductive behavior.

Many bird species display great variation both within and between populations in mating patterns and parental roles (Davies, 1991; Lack, 1968). Fluctuating selection pressures causing plasticity in reproductive behaviors of males and females might explain a large proportion of this variation. In this study we focused on within-population variation in the outcome of sexual conflict over the number of mates in a resource-based breeding system. Although a male would benefit from attracting several females, any individual female would benefit from not having to share the resources a male provides with other females. Thus, traits and behaviors that increase mating success will be selected for in males whereas females are selected to avoid already mated males. However, according to the polygyny threshold model, a female is expected to mate with an already mated male when the cost of polygyny is overridden by the quality of the resources the male defends (Orians, 1969; Wittenberger, 1976). Although the original polygyny threshold model was cast in terms of a female's gain in terms of some resource located on a territory defended by a male, it can be expanded. Females may face a trade-off between mating status and the genetic quality of the males, if males of higher genetic quality are also more likely to be mated already (Kempenaers, 1994; Petrie and Lipsitch, 1994). In fact, female pied flycatchers (Ficedula hypoleuca) have been shown, in aviary choice experiments, to trade off mating status against male plumage brightness (Slagsvold and Drevon, 1999). Besides indicating male quality in terms of heritable fitness, the expression of secondary sexual traits can also signal male ability to provide parental care (Heywood, 1989; Hoelzer, 1989; Kokko, 1998; Price et al., 1993; Wolf et al., 1999). Thus, while females choosing to mate with highly ornamented males may suffer an increased risk of sharing their mate, the cost of polygyny could be overridden in several ways. In other words, the observed compromise over number of mates may vary depending on individual variation in male ability to provide females with resources, parental care, and good genes.

Collared flycatchers (Ficedula albicollis) exhibit a resource-based breeding system with biparental care and mixed mating tactics, that is strict monogamy, polygyny, and extrapair copulations occur within the same population. The males display a conspicuous white forehead patch, the size of which increases their probability of establishing a breeding territory (Pärt and Qvarnström, 1997; Qvarnström, 1997), of becomingpolygynous (Gustafsson et al., 1995), and of obtaining extrapair fertilizations (Sheldon and Ellegren, 1999; Sheldon et al., 1997). Thus, females choosing to mate with large-patched males suffer an increased risk of becoming polygynously mated. However, females are expected not only to respond to the expected risk of becoming polygynously mated in relation to male phenotype but also to the expected costs. On average, late-breeding female collared flycatchers produce relatively more fledged offspring (controlling for age and clutch size) when mated to large-patched males, while the opposite is true among early breeders (Qvarnström et al., 2000). A possible explanation for this pattern is that early, large-patched males invest in mating competition at the expense of their parental abilities (Qvarnström, 1999a). As an adaptive response, females display a mate preference for large-patched males only late in the season (Qvarnström et al., 2000). Here we investigated whether the effect of variation inmale ornamentation and the environment on female reproductive success depends on the female's mating status (i.e., whether the female is monogamously or polygynously mated). Because male parental care becomes a more limited resource for polygynously mated females, we expect the reproductive success of such females to be more sensitive to variation in male reproductive tactics. Therefore, polygyny may play an important role in the seasonal fluctuation in selection on the female mate preference.

METHODS

Data collection

The data were collected from 1981 through 1999 in a collared flycatcher population breeding on the Swedish island of Gotland (57°10′ N, 18°20′ E). Each spring we checked all nest-boxes regularly to determine date of egg laying, clutch size, hatching date of eggs, and number of fledged offspring. Birds were ringed as nestlings or when breeding and measured using standard procedures (see Gustafsson, 1989; Pärt and Gustafsson, 1989). The age of previously unringed females was determined from the shape of the primary coverts, whereas the age (yearling vs. older) of previously unringed males was determined on the basis of the color of their remiges (Svensson, 1992). The white forehead patch is approximately rectangular in shape, allowing its area to be estimated as the product of its height and breadth (measured to the nearest millimeter). We controlled for year-related differences in forehead patch size by taking residuals from an ANOVA with year as a factor. We defined males feeding young at two nests as polygynously mated and the relative status (i.e. primary, secondary or tertiary) of females mated to the same male was defined by their laying date, such that primary females lay their eggs first. Tertiary females are rare and were therefore not included in our analysis.

As the classification of females as primary or secondary relied on the capture of a polygynous male at both of the nests, and males sometimes entirely neglect the secondary female, we probably misclassified the mating status of some primary females as monogamous. Furthermore, because unassisted secondary females do not enter our analyses (because the identity of their males was not known), it is likely that we underestimate the true cost of being a secondary female. The analysis presented here is therefore probably conservative (i.e., we will underestimate variation in male ability to assist two females in raising their broods). This is because the less able a male is to assist two females with the caring of young at two nests, the more likely he should be to desert one of the broods. Thus, males that desert their secondary females should have characteristics more similar to males helping their secondary females a little than to males helping their secondary females a lot. For example, if females mated to small-patched males are doing worse, small-patched males should be most likely to desert their secondary females, and we would underestimate the costs of being polygynously mated to small-patched males.

Data analysis

Our analyses excluded all cases where brood size or hatching date had been experimentally manipulated and all cases where female collared flycatchers bred with pied flycatchers or F₁ hybrids between these two species. Because the measure of fitness we used for individual reproductive attempts, number of fledged young, was not normally distributed, all analyses of this variable were performed using a generalized linear model (GLM) with Poisson errors and ln-link. Residuals from these models were normally distributed. Because each male appears twice in the year in which he is polygynous (at a primary and secondary nest), we constructed separate models to compare monogamously breeding females with primary and secondary females, respectively. However, we treated repeated breeding attempts by the same female in different years as independent. There are two reasons to suggest that this will not result in inflation of degrees of freedom due to pseudoreplication. First, repeated pairings of males and females, even if both survive to a subsequent year, are rare, amounting to less than 1% of all pairings in this population (Kruuk et al., 2001). Second, there is little evidence for individual-specific effects on measures of reproductive success; both heritability and repeatability of reproductive success are close to zero in this population (Merilä and Sheldon, 2000; Przybylo et al., 2000). Analyses were performed using GLMStat, version 5.0.4 (Beath, 2000).

RESULTS

Costs of polygyny

Of all female collared flycatchers (N = 2733) used in our analyses, 4.1% were classified as primary and 4.1% as secondary. Primary females bred on average 2.2 days earlier than monogamous females and 6.2 days earlier than secondary females (Table 1). Clutch size did not differ with respect to mating status once the decline in clutch size due to laying date was accounted for (Table 1). Reproductive success (number of fledged young) varied with mating status, and when the effect of laying date was accounted for, primary and secondary females did not differ from each other, although each suffered a significant reduction of approximately 15% interms of reproductive success when compared to monogamously mated females.

Male ornamentation and the costs of polygyny

We investigated whether the size of a secondary sexual trait could be used to predict a male's ability to compensate females for the costs of polygyny. The number of young fledged from a breeding attempt, while controlling for the effects of male and female age and year, was dependent on the three-way interaction between female mating status, laying date, and male forehead patch size (Table 2). Separate analyses within each of the three mating classes showed that the strength of the interaction between laying date and male patch area on female reproductive success increased as the mating status of the female decreased (GLM, Poisson errors: monogamous: χ _i^2 = 3.33, p =.068; primary: χ_i^2= 3.54, p =.060; secondary χ_i^2= 10.64, p =.001; standardized effect sizes, following Rosenthal, 1991; monogamous; 0.036; primary: 0.179; secondary: 0.308). Primary and secondary females did not differ with respect to the interaction between patch size and laying date in reproductive success (χ_i^2= 0.52, p =.57). Thus, monogamous, primary, and secondary females all fledged relatively more young from late breeding attempts when the male had a large forehead patch size, while thereverse pattern was found among early breeders. The main difference is that the seasonal change in the effect ofmale patch size on number of young produced became more pronounced when females were mated polygynously (Figure 1).

DISCUSSION

According to indicator models of sexual selection, the costs of polygyny may be overridden by variation in male quality such that females benefit from choosing highly ornamented males even if they have to share mate (Andersson, 1994). We show here that, on average, female collared flycatchers (Ficedula albicollis) pay a high cost in terms of reproductive success when paired to an already mated male (see also Gustafsson, 1989). Furthermore, the reproductive success of polygynously mated females depend on an interaction between forehead patch size (a sexually selected trait) of their mate and their timing of breeding. Among late breeders, polygynously mated females experience higher reproductive success when mated to large-patched males, but the pattern is reversed for early breeders. Thus, which males (with respect to level of ornamentation) better succeed in raising two broods depends on the environment.

There are reasons to expect polygynously mated females to always do better when their mates have relatively large forehead patches. First, large-patched males have an advantage in competition over nest sites (Pärt and Qvarnström, 1997; Qvarnström, 1997) and should be better able to defendseveral territories of high quality. Second, the forehead patch is condition dependent in its expression (Griffith and Sheldon, 2001; Gustafsson et al., 1995; Qvarnström, 1999b), with large-patched males being in better condition. This suggests that large-patched males should be more able to assist two females in caring for offspring as compared to small-patched males. Nevertheless, polygynously mated females breeding early were found to experience better reproductive success when paired to small-patched males. There are, at least, two possible explanations for this result.

First, small-patched males are better adapted to the physical and/or social environment experienced early in the season, and they are therefore better able to assist two females with raising the broods while the opposite is true late in the season. Second, large-patched males are always best adapted to the environment and able to spend more energy on reproduction in absolute terms as compared to small-patched males, butthey allocate resources differently. Early in the season, when many females remain unfertilized, large-patched malesmay benefit from allocating relatively more effort to mating competition at the expense of their parental care(Qvarnström, 1999a; Qvarnström et al., 2000; see also Kokko, 1998).

Because breeding conditions are most favorable early in the season (and hence offspring production is greatest then; Wiggins et al., 1994) small-patched males should, according to the first explanation, be considered relatively high quality and according to indicator models should be favored by female choice. However, early females display no preference for small-patched males in their social mate choice (Qvarnström et al., 2000), and more extrapair offspring are found in the nests of small-patched males (Sheldon and Ellegren, 1999). In addition, the expression of the forehead patch has been found to be related to both environmental (Griffith and Sheldon, 2001; Gustafsson et al., 1995; Qvarnström, 1999b) and genetically determined condition (Sheldon et al., 1997). Therefore, we consider the second explanation for our results, that large-patched males adaptively change their reproductive tactic in response to a seasonal change in the social environment, most likely. That mating status affects thestrength of the interaction between timing of breeding and male patch size on female reproductive success (Table 2, Figure 1) could therefore be interpreted in terms of male parental care becoming a more limited resource for polygynously mated females. In other words, polygynously mated females are relatively more sensitive to variation in male reproductive tactics (i.e., allocation of resources between mating competition and parental care).

By entering mating status in the analysis, we are likely to be partially controlling for differences in mating effort among males. Therefore, another source of male mating effort, not associated with becoming mated polygynously, needs be postulated. There are several reasons for suggesting that theeffort spent on achieving extrapair matings is a possible candidate. In general, males probably have to spend effort to achieve extrapair copulations when they repeatedly visit and display to neighboring females while withstanding aggressive attacks from their mates. In the collared flycatcher, large-patched males enjoy an advantage in sperm competition, and they are most successful when cuckolding males breeding later than themselves (Sheldon and Ellegren, 1999). Thus, large-patched males will have a higher fitness payoff from allocating effort to seeking extrapair matings instead of defending high-quality breeding sites and/or feeding chicks as compared to small-patched males, and there are better options for this alternative mating tactic early in the season. If the increased effort spent on achieving extrapair matings translates into less time and energy to defend high-quality breeding sites and/or lower male parental care, it could explain why it is more costly for females to be polygynously mated to large-patched males than to small-patched males early in the season. By extension, this reasoning means those males face a trade-off between the degree to which they pursue extrapair fertilizations and the number of offspring produced by their social mates.

We have no behavioral data in support for our supposition that early-arriving, large-patched males are more active in seeking extrapair matings. However, an earlier study has demonstrated an interaction between the timing of breeding and male ornamentation on male mating effort (estimated as change in condition). Large-patched males arrive in better condition at the breeding grounds, but early in the season they lose condition faster than small-patched males. In fact, early- breeding, large-patched males are in poorer condition at the time of feeding offspring than small-patched males, but no such pattern is found among late breeders (Qvarnström, 1999a). These results suggest that large-patched males spend relatively more effort on mate attraction early in the season.

Previously, another reason for trade-offs between male benefits from being polygynous and obtaining extrapair matings has been suggested. Males that divide their time and effort between at least two social mates may suffer reduced options to engage in extrapair copulations and to mate guard (i.e., prevent their own mates from engaging in extrapair copulations; Arak, 1984; Birkhead and Møller, 1992; Westneat et al., 1990). Thus, there are several reasons for assuming that males are constrained in their abilities to both successfully obtain extrapair matings and being socially mated to several females.

Matings patterns are not solely the result of males optimizing their reproductive behavior but also depend on female behavior (Ahnesjö et al., 1993). It has been questioned whether female birds can directly assess male mating status or prevent secondary females from settling with their mate. In the collared flycatcher, as in the closely related pied flycatcher, males often defend spatially separated breeding territories. Explanations for polyterritoriality include that males thereby deceive prospecting females as to their mating status (Alatalo et al., 1981) and/or prevent female–female aggression and mate guarding (Slagsvold and Lifjeld, 1994; Slagsvold et al., 1992). Whichever explanation applies, mate-choosing females should be selected to use male phenotype to assess the risk and expected cost of becoming polygynously mated. An earlier study demonstrated that females indeed seem able to adjust their reproductive decisions to changes in the expected benefits from choice of males with different forehead patch sizes. Only late-breeding females display a social mate preference for large-patched males and lay larger clutches when mated to such males (Qvarnström et al., 2000). Therefore, it is possible that adaptive adjustment of female reproductive behaviors may further suppress male fitness benefits from both becoming polygynous and seeking extrapair matings.

It may be questioned why early-breeding females do not display a preference for small-patched males because mating with large-patched males not only increases the risk of becoming polygynously mated but also incurs a higher cost of polygyny. One possible explanation is that females would have to pay genetic costs by pairing with small-patched males; offspring sired by small-patched males fledge in relatively poorer condition (Sheldon et al., 1997) and inherit their farthers' small forehead patches (Qvarnström, 1999).

In conclusion, highly ornamented males are often assumed to be able to attract more females because the females benefit from breeding with those males (Andersson, 1994). We found that which males (with respect to level of ornamentation) are more capable of raising two broods depends on the environment. Among early-breeding collared flycatchers, polygynously mated females experienced higher reproductive success when mated to less ornamented males. We propose that the explanation of this result lies in a trade-off between the degree to which males pursue extrapair fertilizations and the number of offspring produced by their social mates. Following this reasoning, the co-occurrence of these two mating strategies (extrapair copulations and social polygyny) may be limited in species with biparental care.

Figure 1

Contour plots of the interaction between laying date (1 May = day 1) and male forehead patch size (mm²) on residual number of fledged young (controlling for year and parental age effects) for (a) monogamous, (b) primary, and (c) secondary females; darker contours represent lower values of reproductive success. The strength of the interaction between laying date and male patch size on female reproductive success increased as the mating status of the female decreased (GLM, Poisson errors; monogamous: χ²₁ = 3.33, p = 0.068; primary: χ²₁ = 3.54, p = 0.060; secondary: χ²₁ = 10.64, p = 0.001). Forehead patch size is controlled for year related variation

Table 1

Phenology and costs of polygyny for 2733 female collared flycatchers of which 2509 were classified as monogamous, 112 primary, and 112 secondary

	Female mating status
Variable	Monogamous	Primary	Secondary	F	p
Laying date	25.11* ± 0.09	22.91** ± 0.39	29.15† ± 0.39	33.91	<.001
Clutch sizea	6.01* ± 0.02	6.07* ± 0.07	5.77** ± 0.07	7.99	<.001
Clutch sizeb	6.05* ± 0.02	6.00* ± 0.06	6.01* ± 0.06	1.30	.27
Fledged younga	4.56* ± 0.05	4.15** ± 0.18	3.49† ± 0.18	10.74	<.001
Fledged youngb	4.64* ± 0.04	4.02** ± 0.18	3.92** ± 0.18	8.21	<.001

Values in the table are least squares means (± SE) correcting for variation associated with year and male and female age. F ratios from general linear models with normal (laying date, clutch size) or Poisson (fledged young) errors. Means with different superscript symbols differ significantly (P < 0.05) from each other. Laying date refers to days passing from 30 April (1 May = day 1); clutch size refers to the number of eggs laid and fledged young refers to the number of young that left the nest.

^a Uncorrected for decline with laying date.

^b Correcting for decline in clutch size or fledging success with respect to laying date.

Table 2

Analysis of effects of breeding time (date of clutch initiation), male forehead patch size, female mating status (monogamous, primary or secondary), and their interactions on numbers of fledged young

		Monogamous vs. primary			Monogamous vs. secondary
Variable	Δ df	Δ Deviance	p	Δ Deviance	p
Year	18	189.6	<.001	207.5	<.001
Male age	1	15.99	<.001	11.44	<.001
Female age	1	8.62	.003	13.27	<.001
Laying date	1	86.42	<.001	93.80	<.001
Forehead patch size	1	0.24	.62	0.58	.45
Mating status	1	3.91	.05	28.53	<.001
Date × patch size	1	4.79	.03	4.71	.03
Date × status	1	1.24	.27	1.25	.26
Patch size × status	1	2.86	.09	0.01	.92
Date × patch size × status	1	3.55	.06	4.88	.03
Residual	2705	2968		3041

Year and male and female age are included as potentially confounding variables. Test statistics are changes in deviance from a general linear model with Poisson errors, and ln-link, where the effect of each predictor variable is tested by sequentially adding and deleting it from the model. Separate models were constructed for monogamous vs. primary and monogamous vs. secondary females to avoid pseudoreplication.

We thank Göran Arnqvist and Trevor Price for discussions and comments on the manuscript. Financial support was obtained from the Swedish Foundation for International Cooperation in Research and Higher Education (STINT), the Fulbright Commission (to A.Q.), a Royal Society University Research Fellowship (to B.C.S), and the Swedish Natural Science Research Council (to T.P. and L.G.).