Molecular Ecology (2011) 20, 3237–3250
doi: 10.1111/j.1365-294X.2011.05158.x
Alien eggs in duck nests: brood parasitism or a help from Grandma? RALPH TIEDEMANN,* KIRSTEN B. PAULUS,* KATJA HAVENSTEIN,* SVERRIR T H O R S T E N S E N , † A E V A R P E T E R S E N , ‡ P E T E R L Y N G S § and M I C H E L C . M I L I N K O V I T C H – *Unit of Evolutionary Biology ⁄ Systematic Zoology, Institute of Biochemistry and Biology, University of Potsdam, KarlLiebknecht-Strasse 24-25, Haus 26, D-14476 Potsdam, Germany, †Langahlı´ð 9a, IS-603 Akureyri, Iceland, ‡Icelandic Institute of Natural History, IS-125 Reykjavı´k, Iceland, §Christiansø Fieldstation of Natural Science, Christiansø, DK-3760 Gudhjem, Denmark, –Laboratory of Artificial and Natural Evolution, Department of Genetics and Evolution, University of Geneva, Geneva, Switzerland
Abstract Intraspecific brood parasitism (IBP) is a remarkable phenomenon by which parasitic females can increase their reproductive output by laying eggs in conspecific females’ nests in addition to incubating eggs in their own nest. Kin selection could explain the tolerance, or even the selective advantage, of IBP, but different models of IBP based on game theory yield contradicting predictions. Our analyses of seven polymorphic autosomal microsatellites in two eider duck colonies indicate that relatedness between host and parasitizing females is significantly higher than the background relatedness within the colony. This result is unlikely to be a by-product of relatives nesting in close vicinity, as nest distance and genetic identity are not correlated. For eider females that had been ring-marked during the decades prior to our study, our analyses indicate that (i) the average age of parasitized females is higher than the age of nonparasitized females, (ii) the percentage of nests with alien eggs increases with the age of nesting females, (iii) the level of IBP increases with the host females’ age, and (iv) the number of own eggs in the nest of parasitized females significantly decreases with age. IBP may allow those older females unable to produce as many eggs as they can incubate to gain indirect fitness without impairing their direct fitness: genetically related females specialize in their energy allocation, with young females producing more eggs than they can incubate and entrusting these to their older relatives. Intraspecific brood parasitism in ducks may constitute cooperation among generations of closely related females. Keywords: eider duck, indirect relatedness, Somateria mollissima
fitness,
intraspecific
brood
parasitism,
microsatellites,
Received 1 March 2011; revision received 29 April 2011; accepted 5 May 2011
Introduction Intraspecific brood parasitism (IBP; also called conspecific brood parasitism or prehatch brood amalgamation), i.e. laying of eggs into the nest of another female, is a remarkable, yet widespread, phenomenon in Anatidae (ducks and allies). Indeed, among the 234 bird species known to exhibit IBP (Yom-Tov 2001), 32% are anatids, a number vastly disproportionate relative to Correspondence: Ralph Tiedemann, Fax: +49 331 977 5070; E-mail:
[email protected] 2011 Blackwell Publishing Ltd
the fraction (1.5%) of anatid species within birds (Monroe & Sibley 1993). As brood parasitism has been observed in most of the extant anatid species analysed to date (Yom-Tov 2001), it is probable that duck species generally exhibit this very peculiar trait. Several hypotheses have been forwarded to explain this phenomenon. Some of them assume that parasitizing females do not build a nest themselves because of either (i) limitation of nest sites, (ii) intrinsic individual limitations (lack of experience and ⁄ or poor physiological condition; Sorensen 1991), (iii) or as an alternative reproductive strategy maintained by balancing selection
3238 R . T I E D E M A N N E T A L . (recently reviewed in Reichart et al. 2010). Other hypotheses assume parasitizing females to have established a nest themselves, but to additionally lay eggs in other females’ nest, either (i) because they lost the eggs in their nest, or (ii) for decreasing the risk of predation by distributing eggs into several nests (‘bet-hedging’), or (iii) because they are in such excellent condition that they can increase their fitness by laying additional eggs into other females’ nest (‘fecundity enhancement’; Sorensen 1991; Lyon & Eadie 2008). While the evidence for the bet-hedging hypothesis is scarce (Lyon & Eadie 2008), there is support for the ‘fecundity enhancement’ ˚ hlund & Anhypothesis: it has been demonstrated (A dersson 2001) that parasitic duck females can double their reproductive output by laying eggs in other females’ nests in addition to incubating eggs in their own nest. The mentioned hypotheses have in common that they concentrate on the fitness benefit for the parasitizing female. From the host’s perspective, IBP might either decrease the fitness, under which scenario one should assume the evolution of an avoidance mechanism (Svennungsen & Holen 2010). Alternatively, host females might also gain (indirect) fitness, when experiencing IBP. Ducks are characterized by female natal philopatry; contrary to many bird species, it is the female, not the male, that returns to its birth site (Greenwood 1980; Greenwood & Harvey 1982; Anderson et al. 1992). Several authors (Andersson 1984, 2001; Eadie et al. 1988; ˚ hlund 2000; Loeb McRae & Burke 1996; Andersson & A et al. 2000; Lyon & Eadie 2000; Lo´pez-Sepulcre & Kokko 2002; Loeb 2003) have raised the possibility that kin selection could explain the occurrence of brood parasitism. Indeed, if parasitic and parasitized females are close relatives (a situation that is greatly facilitated in ducks by the combination of coloniality and female philopatry), the host female could gain an indirect fitness benefit from IBP. Although this hypothesis has gained recent experimental support (Andersson & Waldeck 2007), its evolutionary implications are controversial: While the potential gain in inclusive fitness is evident from the perspective of the parasitizing female, one of the models of IBP based on game theory predicts that — from the host’s perspective — the loss in direct fitness exceeds the gain in indirect fitness, hence making IBP evolutionary unstable (Zink 2000; but see Andersson 2001; Lo´pez-Sepulcre & Kokko 2002). Ducks, including eiders, can reach high ages of more than a decade. Although longitudinal studies on ducks are scarce, there is evidence that clutch size decreases with higher age (Dow & Fredga 1984). Our study is the first investigating IBP in relation to the age of the host female: If the ability to produce own eggs decreases with age, older females might lay fewer eggs than they
can incubate. Then, the incubation of additional eggs from related females might constitute a gain in indirect fitness without any associated loss of direct fitness. Here, we specifically test the hypotheses that (i) IBP preferentially occurs among relatives and (ii) IBP increases with the age of the host female. Although very large clutch sizes, high egg-laying rates, and ⁄ or differences in egg appearance can be indicative of IBP in ducks, these traditional methods often underestimate the level of IBP (Semel & Sherman 1992). The use of molecular data in duck species should allow a better investigation of both IBP and the kin˚ hlund 2000, selection hypothesis (e.g. Andersson & A 2001; Lyon & Eadie 2000). Hence, using DNA microsatellites, we analysed IBP and genetic relatedness between breeding females and their newly hatched ducklings in a common duck species, the Eider (Somateria mollissima), in two breeding colonies, one in the Subarctic (Northern Iceland) and the other in the Southern Baltic Sea. In species with natal philopatry and parental care in both sexes, both males and females may potentially increase their inclusive fitness by tolerating IBP from relatives (McRae & Burke 1996). However, as (i) male Eiders are not philopatric and disperse across colonies (Tiedemann et al. 1999, 2004), (ii) pairing and mating take place outside the breeding colony at wintering grounds where eiders of different origins mix (Spurr & Milne 1976), and (iii) theoretical models on IBP in ducks concentrate on the egg-layer’s (i.e. the female’s) perspective (Andersson 1984, 2001; Eadie & Fryxell 1992; Zink 2000; Lo´pez-Sepulcre & Kokko 2002), we focus here on the analysis of females.
Materials and methods Study areas and populations Christiansø is a rocky Danish island of about 600 m in diameter, east of Bornholm in the Southern Baltic Sea (5519¢N, 1511¢E), where approximately 2600 eider females nest annually (Lyngs 2000). Akureyri airport is situated at the bottom of the Eyjafjo¨rður fjord in Northern Iceland (6541¢N, 1804¢W). The runway is build into the fjord and provides a suitable habitat of about 2,000 by 300 metres. This colony consists of about 380 nesting females with an average clutch size of 4.1 eggs per nest (Petersen & Thorstensen 1990). According to the traditional taxonomy, eiders at Christiansø belong to the palaearctic subspecies S. m. mollissima, whereas the Icelandic eiders belong to the nearctic subspecies S. m. borealis (cf. Tiedemann et al. 2004). Eider ducks are, in general, restricted to areas mostly inaccessible to mammalian predators. More specifically for the two 2011 Blackwell Publishing Ltd
I N T R A S P E C I F I C B R O O D P A R A S I T I S M , A G E A N D K I N 3239 sites analysed here, the level of predation by mammals is further restricted at Akureyri airport because of human activities, and mammalian predators are absent in Christiansø archipelago. Large gulls are present at both sites but typically do not attack adult females at the nest (personal observations).
Field work Female eider ducks were caught at the nest in early May at Christiansø (springs of 1998–2001; n = 63) and in late May at Akureyri (years 2000 and 2001; n = 77). Specifically, females were captured at Christiansø by pulling over their head a dark cotton bag, kept open by a metal ring of about 20 cm in diameter and attached to a pole of about 1.5 m. At Akureyri, females were either driven into hand-held mist nets (see Otnes 1990 for details) or captured by pulling over their head a loop attached to a 2- to 3-m-long pole. Each female was blood-sampled at the leg vein and ring-marked (with one individual-specific steel ring and 1 year-specific coloured ring). In Akureyri, many females had already been ring-marked as a result of intensive annual campaigns during the last decades (Petersen & Thorstensen 1990; see Appendix I). Each nest of marked females was identified with a stick (to be found in subsequent study years) and located on a map on which pairwise Euclidian distances among nests were measured. Eggs were monitored at least twice a day to ensure that ducklings were sampled shortly after hatching. Altogether, 506 eider ducklings were blood-sampled in the nest within at most 12 h after hatching. Five females had their ducklings sampled once, whereas 45, 7 and 6 females were investigated 2, 3 and 4 consecutive years, respectively. Altogether, 140 nests of 63 females (41 from Akureyri, 22 from Christiansø) were analysed. Blood samples were directly transferred into a preservation buffer (Seutin et al. 1991) and kept at 4 C.
Production of molecular data DNA was isolated from blood samples using the SuperQuik-Gene DNA extraction kit (Analytical Genetic Testing Center, Denver, CO, USA) according to the manufacturer’s instructions. To determine whether offspring had been mothered by the incubating female, we genotyped seven polymorphic autosomal microsatellite loci for 140 clutches, both of the nesting females and corresponding 506 ducklings. The 7 loci (HrU2, Primmer et al. 1995; Sfil4 and Sfil7, Fields & Scribner 1997; Smo4, Smo6, Smo9, and Smo10, Paulus & Tiedemann 2003) were amplified via PCR according to Paulus & Tiedemann (2003). It has been demonstrated (Paulus & Tiedemann 2003; Tiedemann et al. 2004) that these loci 2011 Blackwell Publishing Ltd
are polymorphic and do not deviate significantly from Hardy–Weinberg equilibrium in the two populations studied here. Fragment size was determined on an ABI 373 automatic sequencer, using the GENESCAN 2.0 software (Applied Biosystems) and an internal size standard (MapMarker, Eurogentec). In case of allelic mismatch between a nesting female and a duckling (i.e. if the two individuals did not share at least one allele at any given locus), all specimens of that clutch (female and all ducklings) were reanalysed by a different investigator in a different laboratory, performing a new PCR and determining fragment size on an AB 3100 multicapillary sequencer. Only mismatches confirmed by both analyses were considered here. For each locus, we calculated the specific exclusionary power as in Reichart et al. (2010).
Inference of intraspecific brood parasitism Because neither parent was known with certainty beforehand, we followed the maternity assignment approach of Reichart et al. (2010), i.e. genotypes of nesting females and offspring in the nest were compared locus-by-locus. In principle, true biological offspring should share at least one allele at each locus with the nesting female. Lack of a shared allele at a locus (called ‘mismatch’ hereafter) between a nesting female and a duckling in its nest can occur by one of the following four phenomena: (i) genotyping error for nesting female and ⁄ or duckling; (ii) mutation in the germ line of the nesting female or at an early development stage of the duckling (mosaicism); (iii) homozygote mismatches because of a null allele in the adult female; or (iv) the nesting female not being the mother of that duckling, i.e. ‘intraspecific brood parasitism’ (IBP). Genotyping errors were minimized by confirming all mismatches in two separate genetic analyses. For point (ii), assuming mutation rates at polymorphic microsatellites between 10)4 and 10)3 (e.g. Paetkau et al. 2004), we can expect — on average — between 0.7 and 7 such events in our data set (506 ducklings * 7 loci * 2 alleles ⁄ locus * 10)4 or 10)3 mutations ⁄ locus), of which half (0.35–3.50) can be expected to affect an allele received from the true biological mother. Homozygous mismatch (i.e. when the nesting female and the duckling are — at any specific locus — both homozygous but for different alleles) should be considered with caution because it could be caused by the occurrence of null alleles or by allelic drop-out (Goosens et al. 1998), rather than by nonparentage. In our data set, the overall frequency of homozygous mismatches for ‘nesting female vs. offspring’ pairs is 0.8% across all loci (i.e. 28 mismatches of 3542 pairwise comparisons), corresponding to 11 ⁄ 243 ducklings (4.5%) in the Christiansø site and 15 ⁄ 263
3240 R . T I E D E M A N N E T A L .
Computation of pairwise genetic identity and relatedness For each of the 140 nests, mean allele identity (number of shared alleles divided by the number of scored alleles, across the 7 analysed microsatellite loci) was computed between the nesting female and (i) biological offspring within its nest (Inf ⁄ bo), (ii) parasitic ducklings within its nest (Inf ⁄ pd), (iii) all ducklings from other nests in the colony (Inf ⁄ rd; background identity with random ducklings), and (iv) all females from other nests in the colony (Ir; background identity among nesting females,
Match Mismatch Confirmed mismatch
(a) 600
Number of ducklings
500
400
300
200
100
0 HrU2
Sfiµ4
Sfiµ7
Smo4
Smo6
Smo9
Smo10
Microsatellite loci
(b) 600 Expected mismatches (3% typing error, 10e-3 mutation rate) 500
Number of ducklings
ducklings (5.7%) in the Akureyri sample. Homozygous mismatches occur at three loci with the following frequencies: 2.0% (10 ⁄ 506) for sfil4, 0.8% (4 ⁄ 506) for Smo4 and 2.8% (14 ⁄ 506) for Smo9. It is unlikely that a significant proportion of homozygous mismatches are, in our sample, because of null alleles because we would then expect some individuals to exhibit homozygosity for null alleles (i.e. individuals that would exhibit no amplification product at one locus, a case we did not encounter). Similarly, it is unlikely that allelic drop-out yielded a large number of homozygous mismatches because genomic DNA quality was high, genotyping was duplicated, and the loci used in the present study did not exhibit any significant excess of short alleles in homozygous eider individuals analysed elsewhere (Tiedemann et al. 2004). The expected frequency of mismatches because of the combined effect of genotyping error and mutation was calculated according to a Poisson distribution and compared with observed values (Fig. 1). Each of the 506 ducklings was assigned as a ‘biological offspring’ or ‘parasitic duckling’ of the corresponding nesting female on the basis of genotype comparisons. Specifically, an offspring was assigned ‘parasitic’ if full parentage was incompatible with the allele distribution of the offspring and the corresponding nesting female, as even single mismatched offspring are very likely to be parasitic ducklings (see Fig. 1 and the following Results section). Nevertheless, we additionally analysed our data set after excluding all ducklings that exhibited a mismatch with the nesting mother at a single locus (cf. Reichart et al. 2010). It should be noted that this approach is conservative (=reduces type I statistical error) because it excludes the rare potential ‘true biological, yet singlelocus mismatched, offspring’ (i.e. exhibiting either one mutated allele, or one null allele, or one allelic drop out), at the expense of an increased type II error, because many individuals that are truly parasitic may be excluded (i.e. the frequency of ‘parasitic single-locus mismatched offspring’ is expected to be high, given the high genetic relatedness among breeding females).
Expected mismatches (3% typing error, 10e-4 mutation rate) Observed mismatches
400
300
200
100
0 0
1
2
3
4
5
6
7
Number of mismatched loci
Fig. 1 (a) Comparison of genotypes between nesting female and corresponding ducklings: number of ducklings with no mismatch, with one or several mismatch(es) after one round of genotyping, and one or several mismatch(es) confirmed after re-typing at the corresponding loci. (b) Observed number of mismatches between duckling and nesting female (only those confirmed by re-genotyping) compared with expected numbers because of genotyping error and ⁄ or mutations. See text for details.
i.e. ‘genetic identity’ according to Nei 1987, page 221). We infer genetic relatedness between a nesting female (nf) and a parasitic female (pf) using the following equation (see Appendix II for derivation): rnf=pf ¼
Inf=pd Ir : Inf=bo Ir
We put our molecular data in perspective with nesting females’ age obtained from ring-marking in the 2011 Blackwell Publishing Ltd
I N T R A S P E C I F I C B R O O D P A R A S I T I S M , A G E A N D K I N 3241 20 years prior to our study at the Akureyri site. When a female is captured for ring marking, there is some uncertainty whether this individual is in its very first nesting season (i.e. at age 2) because we cannot guarantee that any given female was ringed in its first breeding year. Therefore, we assigned a minimum (rather than exact) age to each ring-marked female (Appendix I). Females caught without a ring were assigned to minimum age 2. Eider females with a minimum age of 3 and more were assigned to age classes (age 3–6, age 7–10 and age ‡11) for computing summary statistics (mean occurrence of IBP, average clutch size and number of own eggs).
Statistical analyses As some females contributed more than once to the analysis (because they were observed in different years, as confirmed by ring marking), we used Fisher’s exact test to identify a possible individual-specific effect. Likewise, as our study spans more than one breeding season, we tested for a possible year-specific effect for all parameters associated with IBP, applying v2 homogeneity tests for frequency data, and Analyses of Variance (data from more than 2 years) or t-tests (data from 2 years) for numerical data (Sokal & Rohlf 1995). The correlation between female minimum age and IBP occurrence was assessed by using the product– moment correlation coefficient (e.g. Sokal & Rohlf 1995), whereas the correlation between nest distance and genetic identity among nesting females was evaluated, separately for sites and study years, by a Mantel test using MANTEL ver. 2.0 (Liedloff 1999).
Results Inference of intraspecific brood parasitism For the typed microsatellites, numbers of alleles (and—in parentheses—locus-specific exclusionary power for Christiansø ⁄ Akureyri) were as follows: HrU2: 3 alleles (0.185 ⁄ 0.166); Sfil4: 9 alleles (0.451 ⁄ 0.334); Sfil7: 60 alleles (0.748 ⁄ 0.726); Smo4: 32 alleles (0.645 ⁄ 0.650); Smo6: 12 alleles (0.361 ⁄ 0.332); Smo9: 14 alleles (0.718 ⁄ 0.613); Smo10: 23 alleles (0.466 ⁄ 0.477). When one or several genotype mismatches were observed between a nesting female and any duckling in the nest, all animals (nesting female and all ducklings) were retyped in a separate laboratory, yielding an overall estimate of genotype error of about 3% (Fig. 1a). From here on, and in all subsequent figures and tables, only confirmed mismatches are considered. Of 506 ducklings, 423 proved to match the nesting female at all loci after retyping. The total exclusionary power of our analysis was 0.996 2011 Blackwell Publishing Ltd
and 0.993 at Christiansø and Akureyri, respectively, translating into an estimated probability of only 0.004 and 0.007, respectively that an unrelated duckling matches a females’ genotype at all loci by random. Hence, ducklings matching the nesting female at all loci are considered true biological offspring of that female. The remaining 83 ducklings differed from the nesting female at 1–6 loci. We assessed the possible impact of mutation and genotyping error on our mismatch data: Fig. 1b shows expected dyad frequencies of matches ⁄ mismatches under the assumption that all ducklings in a nest originate from the nesting female (‘null hypothesis’; Poisson distribution of 1 ) b where b is the sum of the genotype errors — i.e. 3% in each of two subsequent typings — and the mutation rate; data for two mutation rates are shown). According to this calculation, we can expect only 3 (low mutation rate) to 6 (high mutation rate) ducklings exhibiting a singlelocus genotype mismatch with their mother. We observe 83 mismatching ducklings, a number that vastly exceeds the expectation under the null hypothesis (Fig. 1b). Even for the 49 confirmed single mismatches, the vast majority (i.e. an estimated 88–94%) can be considered to be ‘parasitic offspring’. For mismatches at more than one locus, the probability that such a duckling is parasitic exceeds 99.9%, even when considering a high mutation rate.
Occurence of intraspecific brood parasitism Among 63 and 77 analysed nests at the Christiansø and Akureyri sites, respectively, 24 (38%) and 23 (30%) were inferred to contain ducklings not mothered by the nesting female (i.e. ‘parasitic’ ducklings; Table 1). The overall percentages of parasitic ducklings per parasitized nest (see Appendix III for data of individual females) is 41% and 42% for the Christiansø and Akureyri sites, respectively, whereas the proportions of parasitic ducklings across all nests are 17% and 16%, respectively (derived from Table 1). The percentage of parasitized nests has a tendency to be higher at Christiansø (Fisher’s exact test; P = 0.085), while the percentages of parasitic ducklings within a nest do not significantly differ between sites. As we re-sampled some females in consecutive years, we tested whether the probabilities of IBP in 2 consecutive years are independent. Fisher’s exact test for independence indicates that there is no sign of an individual-specific effect, i.e. the likelihood of IBP in the nest of a given female is independent among years of sampling, both for Christiansø and Akureyri [P = 0.190 and P = 0.177, respectively; the power of these tests is 1 ) b = 0.86 and 1 ) b = 0.72 for Christiansø and Akureyri, respectively, to detect with significance (at the 5%
3242 R . T I E D E M A N N E T A L .
Parasitized females Site
Parameter
Christiansø No of nests
Akureyri
Year
1998 1999 2000 2001 Total No of 1998 ducklings 1999 2000 2001 Total No of 2000 nests 2001 Total No of 2000 ducklings 2001 Total
Nonparasitized females 1 mismatch >1 mismatch Sum Total 10 12 10 7 39 59 68 45 29 201 27 27 54 116 106 222
4 4 3 2 13 10 6 5 5 26 7 3 10 13 10 23
3 5 2 1 11 5 7 3 1 16 3 10 13 5 13 18
level) a high correlation (r > 0.75) in IBP across subsequent years]. Furthermore, the age of host females and the various estimates of genetic identity related to IBP do not significantly differ across sampling years, except for relatedness between nesting females and both own and parasitic offspring at Akureyri where both these measures were significantly lower in 2001 than in 2000 [t-test: t-values 2.843 (P = 0.011) and 2.650 (P = 0.015), respectively]. As there is no identifiable individual-specific or year-specific effect on age and genetic-identity parameters in relation to the occurrence of IBP, we used mean values of these parameters among study years for further analyses such that more robust estimates based on larger sample sizes are obtained (for year-specific data refer to Appendix III).
Age dependence of intraspecific brood parasitism Because all studied eider females of the Akureyri colony have been individually ring-marked as nesting adults during the past decades (except for eiders of minimum age 2 which we caught without ring; see
7 9 5 3 24 15 13 8 6 42 10 13 23 18 23 41
17 21 15 10 63 74 81 53 35 243 37 40 77 134 129 263
Table 1 Occurrence of intraspecific brood parasitism (IBP) at Christiansø and Akureyri. Nonparasitized females are defined as exhibiting no mismatch with any duckling in their nest. Evidence of IBP for parasitized females is separated into single and multiple mismatch(es), at 7 microsatellite loci, between females and ducklings in their nest
Material and methods), we could unambiguously determine in this colony that probability and levels of IBP are correlated with mean minimum age of nesting females (Table 2). For the 3- to 6-year-age class, only 17% of the nests contained parasitic ducklings, while this proportion increases to 35% and 37% for the 7- to 10-year- and >11-year-age classes, respectively (Table 2). It should be noted that IBP was also common (42%) among the females of the 2-year-age class, i.e. among those eiders of unknown age which we caught without ring. Except for this first age class, the percentage of alien eggs increases significantly (r = 0.529, P = 0.052, n = 14) with the age of nesting females, such that, on average, over 20% of the clutch is not mothered by the oldest nesting females. This latter value increases to a striking 64% of alien eggs when only parasitized nests are considered. Finally, the three females that incubated alien eggs only (i.e. their nests did not contain any biological offspring) were all at least 15 years old. Note that the total number of eggs (own plus parasitic) remains relatively stable across age classes of
Table 2 Occurrence of intraspecific brood parasitism in relation to the minimum age of the nesting host females at the Akureyri site (mean values ± standard errors; in parentheses are values if only parasitic ducklings with more than one mismatch are considered)
Age class
n
Proportion of nests with IBP
Proportion of alien eggs
Average total clutch size
Average number of own eggs
2 3–6 7–10 ‡11 All
12 29 17 19 77
0.417 0.172 0.353 0.368 0.299
0.201 0.051 0.162 0.236 0.145
2.9 3.5 3.7 3.3 3.4
2.3 3.3 3.1 2.5 2.9
(0.000) (0.103) (0.235) (0.316) (0.169)
± ± ± ± ±
0.079 0.023 0.060 0.086 0.030
(0.000 (0.027 (0.088 (0.133 (0.058
± ± ± ± ±
0.000) 0.015) 0.041) 0.059) 0.016)
± ± ± ± ±
0.3 0.2 0.4 0.3 0.1
± ± ± ± ±
0.3 0.2 0.3 0.4 0.2
(2.9 (3.4 (3.3 (2.9 (3.2
± ± ± ± ±
0.3) 0.2) 0.3) 0.3) 0.1)
2011 Blackwell Publishing Ltd
Mean relative frequency of alien eggs
I N T R A S P E C I F I C B R O O D P A R A S I T I S M , A G E A N D K I N 3243 0.2 n = 36 P = 0.008
0.1
n = 41
0 2–6
7+
Minimum age (years)
Fig. 2 Percentage (means and standard errors) of alien eggs (two or more mismatches to the nesting female) in the nests of eider females of different ages. P-value is for pairwise t-test.
nesting females (average total clutch size in Table 2). This is explained by the observation (Fig. 3) that the number of own eggs in the nest of parasitized females significantly decreases with age (r = 0.669, P = 0.024, n = 14), whereas the number of own eggs in the nest of nonparasitized females remains stable across age classes (r = 0.077, P = 0.822, n = 14). When the conservative approach of ‘counting only ducklings with more than one mismatch as parasitic’ (cf. Table 1) is followed, the age dependence of IBP appears even more pronounced (Table 2, Fig. 2).
Relatedness, nest distance and intraspecific brood parasitism As expected, mean genetic identity between nesting females and their biological offspring is significantly higher (0.636 and 0.604 at Christiansø and Akureyri, respectively; Table 3) than that between nesting females and offspring randomly taken from the colony (0.360 and 0.298; P < 0.001; see Table 3 for details). The mean genetic identity between a parasitized nesting female and an inferred parasitic offspring in its nest reaches intermediate values (0.486 and 0.410 at Christiansø and Akureyri, respectively). These latter values remain significantly higher than those between a nesting female
Parameter
Tested against
Ir Inf ⁄ rd Inf ⁄ pd
— — Inf ⁄ rd
Inf ⁄ bo
Inf ⁄ rd
2011 Blackwell Publishing Ltd
and a random offspring (P < 0.001; see Table 3 for details); hence, parasitic offspring are genetically significantly more closely related to their host female than expected by chance. This result strongly suggests that the biological (parasitic) mother and the host female are genetically closely related. Using mean allele identities between nesting females and various cohorts in the population (see Material and methods), we estimated that the mean genetic relatedness between a nesting female and a parasitic female is r = 0.39 ± 0.07 (0.48 ± 0.09 and 0.28 ± 0.11 for Christiansø and Akureyri populations, respectively). This significantly higher-thanrandom genetic identity between host and parasitic females (for two randomly chosen females, an average of r = 0 is expected according to the Formula 3 in Appendix II) is unlikely to be a by-product of relatives nesting in close vicinity because we were unable to evidence any significant relationship between distance among nests and levels of genetic identity among nest holders (Mantel tests: Christiansø 1999: r = 0.12, P =0.110; Christiansø 2000: r = )0.09, P = 0.238; Akureyri 2000: r = )0.03, P = 0.223; Akureyri 2001: r = )0.11, P =0.014). Indeed, although the correlation between these two parameters (in different nesting seasons and at different sites) is significant in Akureyri for the year 2001, these statistics are characterized by r2 values between 0.009 and 0.014; in other words, only 0.9– 1.4% of the variation in genetic identity is explained by distances among nests. Hence, there is no indication that female relatives nest in close vicinity to each other.
Discussion Relatedness and intraspecific brood parasitism Our previous analyses of mtDNA restriction patterns (Tiedemann & Noer 1998) and mtDNA sequence data (Tiedemann et al. 1999, 2004) confirmed that eider females are strongly philopatric such that colonies are composed of closely related matrilines. To test the possibility of kin selection as a causal mechanism for IBP, we analysed here genetic relatedness (by typing 7
Christiansø
Akureyri
0.358 ± 0.045 0.360 ± 0.009 0.486 ± 0.026 t = 4.580; d.f. = 46; P < 0.001 0.636 ± 0.012 t = 18.504; d.f. = 45; P < 0.001
0.316 ± 0.037 0.298 ± 0.006 0.410 ± 0.031 t = 3.547; d.f. = 44; P < 0.001 0.604 ± 0.012 t = 23.729; d.f. = 41; P < 0.001
Table 3 Mean allele identities (I), computed from 7 microsatellite loci, at the Christiansø and Akureyri sites between nesting females (nf) and either ducklings randomly chosen from the colony (rd), parasitic ducklings (pd), or biological offspring (bo). Ir gives mean allele identity among all nesting females. P values are given for comparisons with respective random expectations (onetailed t-test)
3244 R . T I E D E M A N N E T A L . polymorphic autosomal microsatellites with high total exclusionary power of over 0.99) between breeding Eider duck females and their newly hatched ducklings at two different sites and demonstrated that over all, 34% of nests contained ducklings not mothered by the nesting female (i.e. ‘parasitic’ ducklings). This proportion is remarkably similar to the estimated 31% of IBP in a Canadian Eider colony (Waldeck & Andersson 2006), as detected by protein fingerprinting (Andersson ˚ hlund 2001). It has been previously suggested that & A IBP preferentially occurs among relatives such that host and parasitic females are closely related (Andersson & Eriksson 1982; Andersson 1984; Eadie et al. 1988; McRae & Burke 1996; Loeb et al. 2000; Lo´pez-Sepulcre & Kokko 2002; Loeb 2003; Andersson & Waldeck 2007). Based on genetic identity (measured by the use of molecular data) between host females and parasitic ducklings, we infer here that the overall mean genetic relatedness between a host female and the corresponding parasitic female is 0.39 ± 0.07, i.e. all our analyses suggest that the mean coefficient of relatedness between a parasitic and a host female is between the theoretical expectations among first-order (0.5 for mother ⁄ daughter or full sisters) and second-order (0.25 for grandmother ⁄ granddaughter or cousins) relatives. This r value might actually constitute a slight underestimate, as the background identity of females randomly chosen from the colony might be elevated, because of female philopatry. Given that female Eider ducks are philopatric, the close relatedness between parasitic and host females could (i) be because of individual recognition of relatedness among these females, or (ii) just reflect a side effect of close relatives breeding in close vicinity. The latter possibility has been postulated for the moorhen (Gallinula chloropus), where random host-nest selection (under strong natal philopatry) yields an almost one-in-five chance of laying parasitic eggs into a first-order relatives’ nest (McRae & Burke 1996). Eiders, however, are—albeit philopatric with regard to their colony—not as faithful as moorhen to a specific nest location: in a study of a Norwegian Eider colony (Bustnes & Erikstad 1993), it has been demonstrated that only 25% of the females were faithful to a specific local nest, whereas 84% nested
