Chapter 6. Social organization


Genetic management of studbook (zoo) populations is based on manipulating individual breeding combinations in order to avoid inbreeding and to minimize genetic loss [Ballou and Lacy, 1995]. This type of management is referred to as intensive genetic management [Princée, 1995b; see Chapter 1]. Pedigree data are required to implement intensive management. It may be expected that such detailed data will not be available for in situ populations or species that are maintained in social groups under ex situ conditions [see Chapter 1]. Biochemical and molecular techniques can reveal actual parentages within these populations. However, these techniques can not always be applied, especially, on in situ populations. Princée [1995b, Chapter 1] introduced the concept of low-intensity genetic management which refers to management of such populations. This type of management is based on species-specific biology and requires limited pedigree data.

Data on natural dispersal patterns (sex and age classes) are used in Maximal Avoidance of Inbreeding (MAI) schemes to minimize inbreeding in populations which are composed of social groups [see Chapter 5 ]. These breeding schemes represents 'worst-case' scenarios in cases that populations are composed of multi-male/multi-female groups and pedigree data are difficult to obtain. Since all offspring which are born within breeding groups are considered siblings, effective population sizes are likely to be underestimated. Assumptions such as random mating or equal reproductive success in breeding groups can result in overestimation of effective population sizes. This involves risks that the adaptive potential (and viability) of such populations are reduced due to genetic loss which is larger than assumed.

Assuming 'worst-case' scenarios can result in recommendations to maintain larger population sizes than are required to maintain minimal levels of genetic variation. Such an approach can be considered appropriate in terms of prevention but can generate conflicts. For example, limited space is available for ex situ populations in zoological gardens [see Chapter 1]. This implies that maintaining larger populations than would be required for preserving sufficient levels of genetic variation limits the number of species that can be maintained. These 'worst-case' scenarios also have impact on management plans for in situ populations. Recommendations based on such scenarios may involve expansion of wildlife reserves and/or the construction of corridors between these reserves. Since this can generate conflicts with the local population living near (potential) wildlife reserves, better estimates for effective population sizes are required.

Management strategies which can be applied to reduce errors in estimating effective populations sizes will differ between in situ and ex situ populations. The minimal data which are available on zoo populations already act on the level of individual (breeding) groups. This means that potential breeders and possible breeding combinations can be identified. Such data are often missing for in situ populations. This means that use of the methods, as described below, can be limited for such populations.

Reproductive parameters such as reproductive life-span, gestation length and inter-birth interval can be used to reconstruct pedigrees. This method involves exclusion of all individuals which do not belong to the reproductive age classes as possible parents. Refinements can be made on the basis of inter-birth interval and gestation length whenever the (partial) breeding histories of individual females, such as date of last litter, are known [see Chapter 4; Appendix B]. Methods to reconstruct pedigrees require that individuals in breeding groups can be individually identified. Therefore, the use of pedigrees will in general be limited to retrospective studies on studbook populations.

Information on species-specific mating systems can be used to estimate effective population size. Although multiple males and females in a breeding group can belong to the reproductive age classes, reproductive success may be limited to specific individuals. For example, only the alpha female in a pack of African wild dogs, Lycaon pictus, is expected to reproduce [Frame et al., 1979]. However, knowledge of the mating system stretches beyond observations of mating activities in populations or breeding groups. Such observations in, especially, males do not necessarily reflect reproductive success. For example, subordinate males brown-tufted capuchins, Cebus apella, copulate with females at the end of oestrus. Janson [1984] suggests that these are infertile matings since ovulation probably takes place in the middle of oestrus. Difficulties in determining reproductive success and therefore in estimating effective population sizes are certainly not restricted to species with complex hierarchic social systems. Studies on various bird species which are (were) considered monogamous breeders revealed that promiscuity or extra-pair fertilization may occur more frequently than generally was assumed [for example, Quinn et al., 1987; Morton et al., 1990; McKilligan, 1990].

Biochemical and molecular techniques can be used to study relationships between mating activities and reproductive success. Such techniques have been used to study social rank and reproductive success in various species of macaques, Macaca sp. [Smith, 1981; Curie-Cohen et al., 1983; Stern and Smith, 1984; de Ruiter et al., 1992]. For example, isozyme studies in a troop of rhesus macaques, Macaca mulatta, showed that second-rank males participated less in observed copulations than dominant males, but that they sired more offspring [Curie-Cohen et al., 1983]. Studies on reproductive success should last several years, depending on the species's life-span. Short-term studies not necessarily reflect an individual's life-time reproductive success. For example, Curie-Cohen et al. [1983] showed that dominant male rhesus macaques have the highest life-time reproductive success. However, these males sired most of their offspring while they belonged to the second rank class. This means that studies on reproductive success should, preferably, at least cover one generation to determine effective population sizes.

Results of studies on life-time reproductive success can be used to predict effective population size and be used in future projections regarding retention of genetic variation. However, one needs to be cautious in extrapolating from such results to a relationship between social rank and reproductive success as this can be influenced by environmental conditions. This especially refers to potential differences between in situ and ex situ conditions. For example, a study on wild troops of long-tailed macaques, Macaca fascicularis, by de Ruiter et al. [1992] showed a high correlation between male social rank and reproductive success. Results of studies on various macaque species under ex situ conditions do not show this same consistency. De Ruiter et al. [1992] discuss the impact of ex situ conditions on results of studies on social rank and reproductive success in macaques.

Unfortunately studies on social rank and reproductive success and even information on the natural mating system, either in situ or ex situ, are scarce for most (endangered) species. This means that the breeding combinations in populations (or breeding groups) of such species need to be considered as 'black-boxes' in genetic management. Effective population size and genetic drift will consequently also be unknown factors. Although this seriously limits genetic management, strategies could be based on extremes in genetic loss that can occur within such populations. The present study includes simulations with the model GeneFlow [Princée, 1988, 1989b, 1995b; Chapter 2] to compare the effects of different mating systems on genetic loss. The simulation experiments include variation in number and composition of breeding groups and reproductive success within the breeding groups.

Material and methods

The model GsPed [see Chapter 2] has been used to create pedigrees of populations, that are composed of several breeding groups (see Table 16). Each population starts with 64 wild-caught individuals (potential founders). Pedigree patterns have been simulated for 10 generations. No generation overlap occurs and new breeding groups are composed each generation following MAI schemes [see Chapter 5]. No population growth is assumed in these simulation experiments. Specific characteristics of these populations are presented in table 16. Effective sizes have been calculated each population according to the method as described by Lande and Barrowclough [1987], see also equation 9.

The model GeneFlow [Princée, 1988; Chapter 2] is used to estimate heterozygosity and gene diversity in generation groups of populations that are created by GsPed. Genetic simulations have been carried out with 40 autosomal independent loci, 5 allelic variants per locus with equal frequencies for each variant and 50 iterations (runs). Gene diversity in the source population is 0.800. Gene diversity in the (potential) founder population is 0.794 (or 99.19% of the gene diversity in the source population). This value is in accordance with the expected gene diversity in a random sample of 64 individuals from an ideal population (see Equation 7). Values of genetic variation in generation groups are presented as proportions (fractions) of gene diversity the source population.

Results

Sex-ratio in breeding groups

Simulation experiments using populations [C], [D] and [G] (see Table 16) are used to compare effects of sex-ratio on genetic loss. Results of these experiments are presented in figure 21. In particular populations that are composed of harem units are very vulnerable for genetic loss. This dramatic effect is illustrated by genetic loss in population [C] in figure 22. Even 64 potential founders and optimal breeding conditions (i.e. equal reproductive success for all individuals), does not prevent for a rapid decline in gene diversity. Gene diversity in this population drops below the '90 percent criterion' [Soulé et al., 1986] within 5 generations. Genetic loss per generation in population [D] can also be considered high. Sex-ratios in the multi- male/female breeding groups of this population are still not optimal. Gene diversity in population [D] drops below 90 percent after 7 generations. Population [H] represents a genetic optimal situation as all individuals equally reproduce. Equal sex-ratio and equal reproductive success limit genetic loss to 0.044 percent per generation.

Size of breeding groups

Effects of group size on genetic variation are demonstrated in figure 22. Populations [A] to [C], that are composed of harem groups, have been used in this simulation experiment. While population size is the same for these populations, the size of harem groups increases from population [A] to [C]. Genetic loss decreases as the population is subdivided in smaller harem groups. Note that effective size in 'large' harem groups is slightly higher than in 'small' harem groups (see Table 16).

Reproductive success in breeding groups

Effects of differences in reproductive success in populations which are composed of four breeding groups with equal sex-ratio are shown in figure 23. The populations which are used for these experiments are listed as [E] to [H] in table 16. Male reproductive success in breeding groups of population [E] are limited to dominant males. Unequal distribution of offspring in both males and females are assumed in breeding groups of population [F]. Breeding combinations in these breeding groups are presented in table 17. Equal reproductive success for half the number of breeding animals is assumed for breeding groups of population [G]. Population [H] has an optimal mating system: all animals produce equal numbers of offspring.

Gene diversity in populations in which reproductive success in breeding groups is restricted to one (dominant) male drops below the '90 percent criterion' within four generations. However, in the case that half the individuals in the breeding group reproduce (population [G]), gene diversity will drop below the '90 percent criterion' after five generations.

Discussion

Results of GeneFlow simulations in the present study show effects of breeding structures on genetic loss. Unequal sex ratio and unequal distribution of offspring will increase genetic loss in populations. Effects of unequal sex-ratio on genetic loss are illustrated in figure 21. Populations which are subdivided in large harem groups (e.g. population [C]) will loose genetic variation more rapidly than populations with equal sex-ratio (e.g. population [H]). Effects of unequal distribution of offspring are shown for multi-male/female groups in figure 23. Genetic loss in population [E], in which only a single male is reproducing per group, is higher than in population [H] which assumes equal distribution of offspring. The effects of unequal sex-ratio and unequal distribution of offspring on genetic loss could be expected from computed values for effective population size (see Table 16; see also Effective population size). It must be noted, however, that the actual Ne values for the populations in this study will differ from these computed values. Lande and Barrowclough [1987] assume random mating while this study assumes composition of new breeding (generation) groups according MAI schemes. This means that, except for the founder groups, breeding groups are composed of sibling groups of the same sex [see Chapter 5].

Results of the present study can be used as guidelines for strategies in genetic management of small animal populations. These guidelines will differ between management of in situ and ex situ populations. First, data on ex situ (especially zoo) populations are likely to be more accurate than data on in situ populations. Although data on actual parentages may be incomplete, data on composition of breeding groups will in general be available for zoo populations. Information on in situ populations will often be limited to raw census data. This means that estimates of genetic loss will be less reliable for in situ than for ex situ populations. Second, numbers and composition of breeding groups can be more easily controlled (and manipulated) in zoos than under in situ conditions. Measures to reduce genetic loss can, therefore, be more easily implemented in management of zoo populations than in management of in situ populations.

Ex situ populations

A measure to maintain sufficient levels of genetic variation in ex situ populations is to increase population size and number of breeding groups. However, implementing this "simple" measure is not always feasible in especially management of zoo populations. The total land area that is used by the world zoos limits the number of individual animals and groups that can be housed adequately. Given the total number of endangered species that may require captive propagation (see Chapter 1), the available space in zoos needs to be optimally used. This means that extending the number of, for example, large harem groups of a given species will 'consume' space that could be dedicated to another species.

Since carrying capacity in zoos will be a limiting factor, strategies to maintain genetic variation need to be based on reducing genetic loss by optimizing Ne/N ratios. This implies that sex-ratio and distribution of reproductive success need to be equalized while maintaining species-specific social organizations. Male effective size in harem species can be optimized by dividing populations in small harem groups. Figure 22 shows that genetic losses in populations with small harem groups are lower than in populations with large harem groups. Ne/N ratios in harem species can also be optimized by equalizing sex-ratio through regularly replacement of breeding males within a generation time. In this, the MAI scheme for harem species, which is based on replacing male breeders with sibling males, can be followed [see Chapter 5]. The strategy to maintain small breeding groups which still reflect natural compositions can also be applied to species which live in multi-male/female groups and in which reproductive success is related to social rank. For example, subdividing breeding groups of population [E] (see Table 16) in smaller groups (e.g. 4 males and 4 females) would increase the number of males that can reproduce and will therefore reduce genetic loss.

Socio-ecological data on in situ populations, preferably collected (when relevant) during the mating and breeding season, are required to decide the extent to which size and composition of ex situ breeding groups can be manipulated. Optimal Ne/N ratios can be achieved - while maintaining the mating structure - by establishing groups which are composed of those individuals that are likely to breed. Such groups may not necessarily reflect the natural social organization and may not even be desirable in the framework of reintroduction (and education). This especially refers to species where non-breeding group members play a vital role in rearing success by protection against predators and/or provision of food. For example, in the wild rearing success of the alpha female in Lycaon pictus is related to pack size [Frame et al., 1979; see also Wilson, 1980]. Since this species is not exposed to shortage of food supply and predators under zoo conditions, a single breeding pair can be successful in rearing offspring, as indicated by data from the Lycaon pictus African wild dog international studbook [Brewer et seq., 1992]. Genetic management which is based on breeding pairs would optimize Ne/N ratios in ex situ populations of this species. However, such a management would lead to the risk that social skills, which are essential for reproductive success in the wild are not developed adequately. This can reduce success of (future) re-introductions and hinder education in zoos. Therefore, European zoos adopted a management protocol which is based on natural composition of packs of Lycaon pictus [Verberkmoes, pers. comm].

Replacement of male breeders, as recommended to increase Ne/N ratios in harem groups, would reflect natural social organization of those species in which male tenure is relatively short compared to the average life-span. For example, lions, Panthera leo, generally are not able to take over a pride until they are five to six years old and their tenure seldom lasts longer than three to four years [Kingdon, 1997]. Although replacing male breeders would reduce genetic loss in the population, it increases the risk of infanticide by the new harem male in various species. This phenomonon has been documented for example in species of the subfamily Colubines [Struhsaker and Leland, 1987], gorillas, Gorilla gorilla, [Stewart and Harcourt, 1987], and lions [Packer and Pusey, 1983, 1984]. Males which take over the harem often kill suckling infants to induce estrus in the females. This type of infanticide behaviour has consequences for ex situ management of such species. Infanticide can eliminate the contribution of the previous harem male to the gene pool. Therefore, males should not be replaced until a number of their offspring are over the age when infanticide is a risk. It is expected that this kind of infanticide by harem males occurs in species which do not have a specific mating season, consequently suckling infants can be expected throughout the year. This implies that infanticide most likely cannot be totally avoided. A strategy to replace harem males in species which show infanticide behaviour may therefore not be considered acceptable in management of ex situ populations for ethical reasons.

Implementation of a strategy which involves bachelor groups under ex situ conditions depends on natural male-male social interactions (e.g. territorial behaviour). This strategy can in general be implemented for species in which (sub-adult) males form bachelor group as for example in zebra species, Equus spp. [see e.g. Estes, 1991]. Nevertheless, bachelor groups may need to be maintained in separate areas from harem groups as territorial behaviour can be induced by visual contact, smell and/or vocalization. Furthermore, bachelor groups may need to be housed in larger areas in order to avoid injuries due to competitive behaviour. In the past zoos did not establish bachelor groups for these reasons. A solution to these practical problems can be maintenance of bachelor groups in large semi-reserves as for example has been implemented in the management of Przewalski's horses [Zimmerman, 1997].

The population models in this study do not assume that generations overlap. New breeding groups are established with individuals of the same age class according MAI schemes and replace parental groups each generation. These models do not reflect the natural organization of species which live in continuous multi-male/female groups under an hierarchical system. The establishment of new breeding groups with individuals of the same age class so as to avoid generation overlap can lead to distorts of behaviour [e.g. Rijksen, 1981]. This implies that genetic management of such species requires manipulation of existing breeding groups. Reproduction in this group of species will often be restricted to males and females which have achieved top rank. Replacement of these breeders, especially when top rank positions are achieved during life-time, can result in excessive aggressive behaviour as the hierarchical system needs to be re-established almost immediately. Although detailed management measures will be species-specific, a general guideline would be to follow natural dispersal patterns. These patterns indicate sex and age of individuals which can be introduced and/or removed from breeding groups without (heavily) disturbing the hierarchical system. For example, male Macaca fascicularis migrate at puberty (four to six years) from their natal group into adjacent groups [de Ruiter et al., 1994].

Although 'continuous' breeding groups are composed of individuals which belong to different generations, hierarchical systems can reduce actual overlap between those generations. New immigrants often need to achieve an alpha or beta social status in order to participate in reproduction. Male Macaca fascicularis do not reproduce until they achieve high rank in their host group, usually at the age of nine years, and male tenure often does not last longer than three years [de Ruiter et al., 1994]. This means that male immigrants replace the previous male breeding generation with a 'delay' of three to five years after immigration. During this period of time the older female generation has most likely become less reproductive. This reduces generation overlap in these breeding groups, especially when migration is synchronized with male tenure. Generation overlap requires more complex population genetic models which include data on life history. This does not mean, however, that generation overlap in management of ex situ populations needs to be avoided or minimized. Older generation groups (i.e. the generations that are closer to the founders) in small populations will have more genetic variation than younger generations. This means that 'back-crossing' with older generations as long as possible can reduce genetic loss [see Princée, 1988]. Generation overlap needs to be minimized as soon as inbreeding can not be avoided.

In situ populations

The IUCN has adopted the Mace-Lande criteria to determine the status of in situ populations [Mace and Lande, 1991; IUCN, 1994a]. These criteria involve estimates of expected genetic loss on the basis of actual and effective population sizes. Population size is subject to errors and depends on the quality of survey methods (and or the use of combined methods). Under- or overestimating population size not only refers to small species. Even census data of the African elephant, Loxondota africana, are subject to errors [see for example Jachmann, 1991]. The errors in effective population size are due to assumptions regarding mating structure (and reproductive success). Mace and Lande [1991] propose Ne/N ratios which range between 0.2 and 0.5. Foose et al. [1995] consider a ratio of 0.1 as a lower limit. Using a 'worst-case' scenario, by assuming lower limits in census data and Ne/N ratios, may be considered acceptable to evaluate the status of species or local populations with respect to risks on extinction. However, more accurate estimates need to be available whenever management measures are required.

Table 16 shows Ne/N ratios for the model populations as studied in this study. Equal reproductive success among females is, except for populations [F] and [G], assumed in these models. It may be expected that reproductive success in females, especially when related to social rank, will not be distributed equally. Nevertheless, the range in Ne/N ratios (0.23 - 1.97) in populations [E] to [H] illustrate the order of magnitude in errors that can be introduced by assumptions regarding reproductive success.