Measures of breeding inequality: a case study in southern elephant seals

Inequality in distribution of resources is a key aspect of evolutionary biology particularly in relation to distribution of mates and copulations. Notwithstanding its important role, inequality is not easily defined, and its measurement is complicated by theoretical and methodological issues. Although the formal treatment of inequality has been mostly limited to the evolution of lek mating system, a methodologically correct approach to measurement of inequality is generally valid for the study of any kind of mating system. In this paper, we analyze inequality in a large set of southern elephant seal (Mirounga leonina) harems. The observed distribution of fertilizations was significantly different from both the expected distribution with equal shares of resources and the expected distribution with equal propensities to acquire resources. We calculate and compare various measures of inequality, observing a wide variation particularly among unbounded and bounded indices. We check the effect of choosing a specific measure of inequality by considering the effect of two aspects of harem socionomy, the number of females in the harem (i.e., the total amount of resources to be shared) and the number of males associated with the harem (i.e., the number of competitors). The choice of a specific measure of inequality had a strong impact on the results obtained and should be considered a critical step in every study of functional and evolutionary correlates of inequality. Unbounded indices showed a strong relationship with both harem size and number of males, while no effect was evident in the analysis of bounded indices. This demonstrates that, in this species, the despotism of the mating system remains high even in large harems and with many competitors, i.e., the worst conditions for monopolization. Résumé : L’inégalité dans la répartition des ressources, particulièrement la répartition des partenaires sexuels et des accouplements, est l’un des facteurs clés de la biologie évolutive. Malgré son rôle important, l’inégalité est difficile à définir et à mesurer à cause de problèmes théoriques et méthodologiques. À ce jour, l’inégalité a été étudiée surtout dans le contexte de l’évolution du système d’accouplement sur leks; une approche méthodologiquement correcte de la mesure de l’inégalité devrait être valide pour l’étude de n’importe quel système d’accouplement. Nous analysons ici l’inégalité au sein d’une grande série de harems d’éléphants de mer (Mirounga leonina). La répartition observée des fécondations diffère significativement de la répartition théorique prévue lorsque les ressources sont réparties également et diffère aussi de la répartition à laquelle on peut s’attendre lorsque les animaux ont tous la même propension à chercher à obtenir la ressource. Nous calculons et comparons diverses mesures de l’inégalité et obtenons une grande variation, particulièrement entre les indices non bornés et les indices bornés. Nous vérifions les effets du choix d’une mesure spécifique de l’inégalité en tenant compte de deux aspects de la socionomie des harems, le nombre de femelles dans le harem (i.e., la quantité totale de la ressource à partager) et le nombre de mâles dans le harem (i.e., le nombre de compétiteurs). Le choix d’une mesure spécifique de l’inégalité a un impact important sur les résultats obtenus et doit être considéré comme une étape critique dans toute étude des facteurs fonctionnels et évolutifs reliés à l’inégalité. Les indices non bornés mettent en lumière une forte relation avec la taille du harem et le nombre de mâles, alors que l’analyse des indices bornés ne révèle aucun effet particulier. Nos résultats démontrent que, chez cette espèce, le despotisme du système d’accouplement demeure élevé même dans les grands harems et même en présence de plusieurs compétiteurs, i.e., dans les pires conditions pour la monopolisation. [Traduit par la Rédaction] 1249 Galimberti et al. Introduction The variation of breeding success among individuals is a key aspect of evolutionary biology (Darwin 1871) and behavioral ecology (Dunbar 1984). The whole study of breeding strategies, mating tactics, and sexual selection has its first motivation in the presence of differences between individuals in their share of breeding resources acquired (Andersson 1994). Selection requires the presence of a nonrandom variation in the distribution of resources. Therefore, the study of Can. J. Zool. 80: 1240–1249 (2002) DOI: 10.1139/Z02-117 © 2002 NRC Canada 1240 Received 28 August 2001. Accepted 20 June 2002. Published on the NRC Research Press Web site at http://cjz.nrc.ca on 19 August 2002. F. Galimberti.1,2 Elephant Seal Research Group (ESRG), Via Buonarroti, 35, 20145 Milano, Italy. A. Fabiani. Department of Biological Sciences, University of Durham, South Road, Durham DH1 3LE, England. S. Sanvito. Department of Biology, Memorial University of Newfoundland, St. John’s, NF A1B 3X9, Canada. 1Corresponding author (e-mail: [email protected]). 2From 1 September to 30 November, address for correspondence: ESRG, Sea Lion Island, Falkland Islands (e-mail: [email protected]). selection and evolution of phenotypic traits does not make sense if the distribution of reproduction is even (Kokko et al. 1999). The presence of observed inequality in the distribution of breeding success is not, per se, a sufficient indication of the presence of a true, nonrandom, inequality (Sutherland 1987). Even in an egalitarian social system, where all individuals have the same propensities to acquire resources, purely random processes may produce inequality in the observed distribution of resources. In this case, the expected distribution of resources is a Poisson distribution, with an expected variance of resource shares equal to the mean (Sutherland 1985). Hence, the first step in the analysis of potential phenotypic targets of selection is to establish if the observed variation in resource distribution is greater than the one expected from a random process. Notwithstanding the crucial role of inequality, the definition of inequality is not very clear, and there is much debate about how to measure inequality and how to test its significance in relation to random, even, or despotic processes of resource acquisition (Kokko et al. 1999). Moreover, the temporal and spatial scales of calculation strongly affect the observed inequality and its relationships with both socionomy (social group structure, i.e., the distribution of individuals of a social group among social classes, where a social class is defined by a combination of factors like sex, age, dominance rank, breeding status, and kinship) and individual phenotype (Galimberti et al. 2002). The formal analysis of inequality has been mostly limited to the study of reproductive skew in eusocial systems (Keller and Reeve 1994) and to the study of evolution of leks in birds (Mackenzie et al. 1995; Widemo and Owens 1995a). Elephant seals are a classic textbook example of sexual selection (e.g., Andersson 1994). They are the most dimorphic land-breeding mammals, and they have a clearly polygynous mating system based on the formation of harems of up to hundreds of females. Moreover, the distribution of fertilizations of females among males associated with harems is apparently very uneven (Le Boeuf and Laws 1994). Notwithstanding this, rigorous quantification and tests of inequality are lacking, even in the northern species (Mirounga angustirostris), which is probably one of the most studied polygynous mammals. In this paper, we analyze a large sample of harems of southern elephant seals (Mirounga leonina) for which good estimates of the number of females fertilized are available. We calculate a wide range of inequality measures and test deviations from randomness. We compare the observed inequality with the expected values at the extremes of the evenness–monopolization spectrum. We evaluate the effect of total resources (harem size) and number of competitors on inequality, and the role of the choice of a specific inequality measure. Materials and methods Data collection was carried out during two breeding seasons (1993 and 1994) at Punta Delgada (Valdés Peninsula, Argentina, hereinafter DEL; for details on the population see Campagna et al. 1993) and during five breeding seasons (1995–1999) at Sea Lion Island (hereinafter SLI), the main breeding colony of southern elephant seals in the Falkland Islands (for details on the population see Galimberti and Boitani 1999). Here, “harem” is a group of two or more females, with or without a male (Baldi et al. 1996). The dataset comprises 74 harems (23 at DEL, 51 at SLI). “Harem size” is the total number of females that breed in the harem during the whole breeding season. Harem size was 55.6 ± 35.2 (mean ± SD) females. “Harem holder” is the male that resides within the female group. Very rarely a second male was observed within the female group. In these cases, the harem holder is defined as the one male that is more inside the females’ group. The “associated male” of a specific harem is each male that was observed in the harem or within 10 standard body lengths from the nearest female of the harem (Deutsch et al. 1990; Baldi et al. 1996), during daily counts, for the majority of days that the male was on land. The number of associated males was 6.5 ± 4.4. The whole sample comprises 485 males (55–104 per year) and 4104 females (512–753 per year). Some males are represented in the dataset for more than 1 year (22.1% for 2 years, 11.0% for 3 years, 3.9% for 4 years, 1.7% for 5 years), raising a concern about pseudoreplication. This should not be a problem for the present study because (i) it involves a harem-level analysis, which is not directly related to individual phenotype; (ii) the social habitat, and hence the competition and resource-acquisition processes, changes among consecutive seasons; and (iii) the phenotypic quality of males changes among seasons, owing to growth and increase in experience. In general, inequality will always be measured in the context of season-specific competition, hence the effect of pseudoreplication should be slight. Our measure of acquired breeding resources is the estimated number of females inseminated (ENFI, Le Boeuf 1974; see also Deutsch et al. 1990). ENFI is calculated from the proportion of copulations achieved by each male in or around the harem and harem’s size. The proportion of copulations achieved by each male in each harem was estimated from a total of 6634 h of systematic observations (the observation protocol is described in Galimberti et al. 2000). We observed a total of 3090 copulations. We trust the quality of our estimates of the proportion of copulations because of the continuous monitoring of the breeding beaches for the whole length of the breeding season and the conspicuousness of elephant seal copulations (e.g., Le Boeuf 1972). Frequently, copulations of males of low social rank are more difficult to observe than copulations of males of high social rank, resulting in biased estimates of mating success (Drickamer 1974). This effect is unlikely in elephant seals, where females are grouped, the breeding habitat is open, and copulations are easy to spot. We carried out observations almost exclusively during the day, but we also did a small sample of observations with night-vision equipment (32 h in all). Activity was equal between day and night (F. Galimberti, A. Fabiani, and S. Sanvito, unpublished data; see also Baldi et al. 1996). The number of females breeding in each harem was estimated from individual records of marked females. Each year, about 99% of the females were marked by cattle tags. Moreover, 70–85% of the females were also marked by hair dye. Details of the marking protocol can be found in Galimberti and Boitani (1999). The reliability of the estimated number of inseminated fe© 2002 NRC Canada Galimberti et al. 1241