Population Genetic Structure of the Cayo Santiago Colony of Rhesus Macaques (Macaca mulatta)
Abstract
The rhesus macaque population at Cayo Santiago increases annually and is in urgent need of control. In-depth assessments of the colony's population genetic and pedigree structures provide a starting point for improving the colony's long-term management program. We evaluated the degree of genetic variation and coefficients of inbreeding and kinship of the Cayo Santiago colony by using pedigree and short tandem repeat (STR) data from 4738 rhesus macaques, which represent 7 extant social groups and a group of migrant males. Information on each animal's parentage, sex, birth date, and date of death or removal from the island were used to generate estimates of mean kinship, kinship value, gene value, genome uniqueness (GU), founder equivalents (fe), and founder genome equivalents (fg). Pedigree and STR analyses revealed that the social groups have not differentiated genetically from each other due to male-mediated gene flow (that is, FST estimates were in the negative range) and exhibit sufficient genetic variation, with mean estimates of allele numbers and observed and expected heterozygosity of 6.57, 0.72, and 0.70, respectively. Estimates of GU, fe, and fg show that a high effective number of founders has affected the colony's current genetic structure in a positive manner. As demographic changes occur, genetic and pedigree matrices need to be monitored consistently to ensure the health and wellbeing of the Cayo Santiago colony.
Cayo Santiago is a 38-acre island that is located approximately 1 km off the southeastern coast of Puerto Rico. The island is the habitat of the oldest continuously studied colony of free-ranging rhesus macaques (Macaca mulatta).8,24,34 This island colony is a provisioned population with no natural predators and is maintained with minimal human intervention. Although the animals are fed a commercial monkey diet (0.23 kg/animal daily), they forage extensively on natural vegetation for more than 50% of their feeding time.30 As a result, the colony experiences high rates of survival and fertility as well as marked population growth and density-dependent dynamics.20 Data on reproductive success reveal that adult males have a mean of 8.7 offspring (range, 0 to 47), and females on average have 7.7 offspring (range, 0 to 16).12
The Cayo Santiago colony currently comprises approximately 1700 rhesus macaques, which (at the time of writing) can be subdivided into 7 naturally formed multimale and multifemale social groups and 5 extragroup adult males. The extant population is approximately 4 times the optimal size proposed previously.8 This high density of animals threatens not only the colony's long-term viability but also the island's natural resources; as a consequence, the colony's growing population remains a major concern. Since the 1950s, trapping and culling have periodically been performed as a means of maintaining population size. Various methods of culling have been used including the removal of animals by sex, age and social group.19,33,44
The colony management of Cayo Santiago has been hampered by multiple factors including limited resources, logistic complexities, and the lack of specialized personnel. For example, culling intervention was suspended between 1973 and 1983 but was reestablished in 1984. In addition, past population-control approaches were applied without considering their effects on the genetic make-up of the colony. Therefore, a significant factor affecting the colony is the routine selective removal of animals to stabilize its population size.19
Among the factors that have influenced the population genetics of this rhesus colony is its geographic isolation since its establishment in 1938. In addition, the current population of Cayo Santiago was initiated from a total of only 409 Indian-origin progenitors that were released onto the island,8 and 90% of the rhesus macaques on Cayo Santiago are descendants of 15 unrelated female founders that were alive in the mid1950s.13 This number is low compared with the recommended founder group of at least 20 individuals with equal contributions.26 Since the colony's establishment, animal numbers also have fluctuated due to conflicts, infanticide, disease and inadequate food supplies.1,9,38 A significant genetic bottleneck resulted in the mid1950s due to a drop in population size to 70 animals.31 By the late 1950s, through regular provisioning, the animal numbers recovered.31 In addition, sires after the 1950s were probably unrelated to the 15 founder females, resulting in excess heterozygosity that countered the effects of earlier genetic bottlenecks.13,14,31,35
Although Cayo Santiago has remained a closed colony since the 1930s, the population perhaps is not genetically depauperate.34 However, an earlier study7 suggested that the island's diminished genetic heterogeneity relative to other macaque populations is indicative of founder effects and genetic bottlenecks over its 78-y history. This opinion is consistent with a recent study involving resident animals at the Sabana Seca Field Station (SSFS), which were derived from the Cayo Santiago colony. Short tandem repeat (STR)–based genetic diversity at SSFS was approximately 90% of the genetic variation at the California National Primate Research Center.23 In addition, single-nucleotide polymorphism data from animals at SSFS confirmed that the Cayo Santiago–derived animals exhibited the lowest level of variation compared with estimates from them California, Southwest, and Yerkes National Primate Research Centers.23 The effect of the colony's small founding population and subsequent genetic bottlenecks on its current genetic composition is underlined by other authors,43 who concluded that the lack of increases in colony inbreeding rates over a 2-decade period resulted from rare cases of breeding among kin in the colony.
Because demographic consequences and population management decisions can influence the genetic composition of populations, knowledge of the extant genetic structure of the Cayo Santiago rhesus population is critical for successful management of the colony. The integration of genetic management approaches into population management plans will help ensure the viability of the Cayo Santiago rhesus macaque colony. Despite its importance, an in-depth evaluation of the island colony's population genetic and pedigree structures has not been performed recently.
Traditional pedigrees are critical for accurately characterizing relationships between animals, assigning the expected proportion of genome-sharing (kinship) for the given relationship, and preventing close inbreeding. To compare population genetic structure and retain valuable genetic diversity through the prevention of allele loss across generations, a panel of highly informative genetic markers such as STR is necessary.23 Therefore, we here used STR and parentage data to evaluate the genetic variation and coefficients of inbreeding and kinship of rhesus macaques living on Cayo Santiago. This study is aimed at establishing baseline population genetic estimates to facilitate the Cayo Santiago colony's long term management.
Materials and Methods
The Caribbean Primate Research Center collects blood samples from yearlings during annual trappings for STR-based paternity analysis. STR genotyping and parentage determination were performed at the Veterinary Genetics Laboratory (University of California, Davis, CA). DNA extraction, PCR amplification, and fragment sizing protocols have been described previously.21 The Center currently has more 4700 STR profiles, which have been used for parentage determination for newborns since 1992.
The demographic and parentage records from Cayo Santiago provide an opportunity to collect longitudinal population genetic and mutigenerational pedigree data to facilitate future population management decisions of the island colony. For the genetic management of a population, a complete pedigree at the individual level should be available. In this study, we used the pedigree data management system PEDSYS15 to construct large multigenerational pedigrees of the 4738 animals for which STR data were available. Information on each animal's parentage, sex, birth date, and date of death or removal from the island were used to compute estimates of the inbreeding (F) and kinship (k) coefficients for each macaque. The F coefficient is the probability that an animal with 2 identical alleles at a particular locus inherited both of those alleles from a single ancestor.46 An animal's k coefficient is the average kinship between that animal and all other living members of its social group, including itself; k is the expected proportion of genetic similarity or genome sharing due to identity by descent among all individuals within a social group.2,4,27 We used the k coefficient data of each animal in the colony to manually calculate estimates of mean k, kinship value (kv), gene value (GV), genome uniqueness (GU), founder equivalents (fe), and founder genome equivalents (fg) for each of the 7 extant social groups (F, HH, KK, MM, R, S and V) and the extragroup of 5 males (group X) according to previously described methods.25
As such, the mean k of a social group is therefore the average of all members’ k coefficients. The kv of an individual animal is a variant of k that is weighted by the animal's reproductive value,17 that is, kv is age-specific and defined as the expected future lifetime reproduction. The mean kv is the average of all the individual kv in a group and takes into account each animal's age and its reproductive value or genomic contribution to future generations.
GV is defined as 1 –kv; therefore, maximizing a group's mean GV is an effective strategy for enhancing the genetic diversity of social groups with known ancestry. An individual macaque's GU is the probability that it exhibits founder alleles that are not present in any other member of the entire Cayo Santiago colony. GU is computed by using gene drop analysis29 and a previously described simulation technique.27 The group mean GU is the average probability across all members of the group that may have acquired unique founder alleles.
The group mean fe and fg values represent the mean numbers of equally contributing founders that would be expected to produce the same level of genetic diversity present in the current social groups. However, whereas mean fe is estimated under the assumption that no random loss of alleles due to genetic drift has occurred across generations, mean estimates of fg include effects due to genetic drift.25 Genetic drift refers to the loss, by chance, of allelic diversity present in the founders.
We analyzed 14 STR (Figure 1) that were described previously22 by using the software program Arlequin version 3.5.2.2.16 The program was used to compute allele numbers (Na), observed heterozygosity (OH), and estimated gene diversity (that is, expected heterozygosity [EH]) among the 7 extant social groups (F through V) and the 5 extragroup males (group X) in the island colony. The same software was used to compute genetic differentiation among the study groups due to genetic drift and isolation (pairwise FST40) and inbreeding coefficient (FIS), which reflects the combined loss of heterozygosity due to nonrandom mating within each study group.47
Results
The study involved a total of 1227 live rhesus macaques at Cayo Santiago, comprising 606 males and 621 females representing 4 to 9 generations of complete pedigrees (Table 1). The oldest animals among the 7 social groups were in their late 20s whereas the males in group X were 13 to 23 y old. The social group size ranged from 72 (MM) to 265 (F) animals. The ratio of males to females was approximately 1:1 in 5 of the 7 social groups studied; in groups R and V, the ratios were slightly skewed toward fewer females. Parentage data suggested that, on average, females are reproductively active from 3 to 22 y of age, whereas their male counterparts reproduced between 4 and 20 y of age. According to their pedigree records, the group X males sired 2 to 15 (average, 10.6) offspring with females from each of the 7 social groups (that is, F, HH, K, M, R, S and V).
Table 1.
Results of pedigree-based genetic analysis
| Estimated mean for group | |||||||||||
| Group | No. of males | No. of females | GN | F (range [mean]) | k | kv | GV | GU | fe | fg | |
| F | 128 | 137 | 7 | 0% to 12.5% (0.43%) | 1.40% | 1.30% | 98.70% | 9.50% | 50.59 | 60.84 | |
| HH | 50 | 68 | 9 | 0% to14.1% (0.58%) | 2.20% | 2.10% | 97.90% | 13.20% | 57.54 | 44.33 | |
| KK | 71 | 79 | 8 | 0% to 6.3% (0.30%) | 1.70% | 1.70% | 98.00% | 10.30% | 69.80 | 54.94 | |
| MM | 33 | 39 | 8 | 0% to 7.0% (0.80%) | 2.40% | 2.00% | 97.70% | 24.50% | 61.80 | 28.21 | |
| R | 156 | 145 | 8 | 0% to 25.0%(0.54% | 1.70% | 1.70% | 98.30% | 6.90% | 59.99 | 42.69 | |
| S | 60 | 74 | 7 | 0% to 25.0% (0.83%) | 2.50% | 2.50% | 97.50% | 10.60% | 42.62 | 27.02 | |
| V | 103 | 78 | 8 | 0% to 6.2% (0.31%) | 1.70% | 1.60% | 98.30% | 13.40% | 64.64 | 41.69 | |
| X | 5 | 0 | 4 | 0.00% (0.00%) | 0.50% | 0.00% | 100.00% | 94.90% | 15.69 | 4.85 | |
| Total | 606 | 621 | 8 | Average | 0% to 25.0% (0.47%) | 1.76% | 1.61% | 98.30% | 22.91% | 52.83 | 38.07 |
F, estimated inbreeding coefficient; fe, estimated founder equivalent; fg, estimated founder genome equivalent; GN, generations; GV, gene value; GU, genome uniqueness; k, kinship; kv, kinship value.
Most (that is, approximately 74%) of the 1227 study animals have a coefficient of inbreeding (F) of 0%, whereas only 2% (that is, 6 animals) of the remaining population exhibit F values of 12.5% to 25%.46 Groups S (0.83%) and HH (0.58%) exhibited the highest and second highest estimate mean F values (Table 1). Estimates of mean k and kv ranged from 1.4% and 1.3%, respectively, for group F to 2.5% for group S (Table 1); group X males exhibited a very low mean k of 0.5% and no measurable mean kv. Among all social groups, mean GV ranged from 97.5% (group S) to 98.7% (F); group X exhibited an estimated mean GV of 100% (Table 1). Estimates of mean GU ranged from 6.9% in group R to almost 25% in group MM; mean GU for group X was almost 95%. Estimated mean GU for the entire population of rhesus macaques at Cayo Santiago was only 1% (Table 1). Group S (42.62) showed the lowest mean fe among social groups, whereas group KK had the highest value for this parameter (69.8l Table 1). Among the groups, mean fg estimates were highest for group F (60.84) and lowest for group S (27.02; Table 1). The group X males demonstrated values of 15.69 for mean fe and 4.85 for mean fg.
Figure 1 summarizes the estimated allele numbers, observed heterozygosity, and expected heterozygosity for all 14 STRs used in this study. For each group, the average number of STR alleles ranged from 4 to 7.9, OH was 0.68 to 0.8, and EH was 0.68 to 0.71 (Table 2); the mean estimates for each of these indices were 6.57, 0.72, and 0.70, respectively. The FIS values suggest that the inbreeding levels in Cayo Santiago have remained at negligible levels, ranging from –0.14 (group X) to 0.02 (group HH; Table 2). Differentiation among the groups as reflected in Table 2 is also negligible, with pairwise FST values ranging from –0.008 (between KK and X and between MM and X) to –0.012 (between MM and R and between MM and S).
Table 2.
Estimated mean allele number (Na), observed heterozygosity (OH), expected heterozygosity (EH), and inbreeding coefficient (FIS) for each group
| Group | Na | OH | EH | FIS | F | HH | KK | MM | R | S | V |
| F | 7.29 | 0.72 | 0.71 | −0.02 | — | ||||||
| HH | 6.93 | 0.70 | 0.71 | 0.02 | 0.010 | — | |||||
| KK | 7.21 | 0.72 | 0.71 | −0.01 | 0.007 | 0.009 | — | ||||
| MM | 6.43 | 0.68 | 0.68 | −0.01 | 0.009 | 0.011 | 0.009 | — | |||
| R | 7.29 | 0.72 | 0.71 | −0.01 | 0.006 | 0.007 | 0.006 | 0.012 | — | ||
| S | 6.43 | 0.71 | 0.70 | −0.02 | 0.009 | 0.006 | 0.005 | 0.012 | 0.004 | — | |
| V | 7.00 | 0.71 | 0.70 | −0.01 | 0.006 | 0.008 | 0.007 | 0.008 | 0.008 | 0.010 | — |
| X | 4.00 | 0.80 | 0.71 | −0.14 | 0.000 | 0.000 | −0.008 | −0.008 | 0.001 | 0.004 | 0.002 |
| Average | 6.57 | 0.72 | 0.70 | −0.02 |
Pairwise estimates of the fixation index (FST) are below the diagonal.
Discussion
The number of rhesus macaques in each social group at Cayo Santiago is consistent with habitats where food subsidization by humans is high.28,39 Consistent with previous observations,18 female rhesus macaques at Cayo Santiago tend not to leave their social group, and there were no unaffiliated females in this study. Most male rhesus macaques on Cayo Santiago have been reported to leave their natal groups and disperse among and reproduce in different social groups during their lifetime.5 Similarly, the males in group X originated from different natal groups (that is, HH, FF, and R) and had reproductive access to females in all 7 social groups as well as reduced incidences of mating with kin.43 This pattern is characteristic of male migrants in wild rhesus populations that are looking for new mating opportunities.32
Group means of GU, kv, and GV are important relationship indices for ranking decisions for breeding or culling animals. Avoiding mating between animals from social groups with highly dissimilar kv is recommended because the genomes of resulting offspring might be half valuable and half nonvaluable, making it difficult to optimally manage the colony in the future.4 Kinship matrices obtained from the pedigree analysis of the Cayo Santiago colony (Tables 1 and and2)2) reveal that the social groups we examined have comparably low mean kv. Therefore, breeding between animals from different social groups on Cayo Santiago will not contribute toward any accumulation of inbreeding. In addition, breeding pairs reflecting such low mean kv contributes to equalizing the representation of founder genomes in the colony.3 Consequently, regarding a strategy for decreasing the size of the colony, either social group S (2.50%) or HH (2.10%) would be the best candidate for removal, given that these groups exhibit not only high but also very similar mean kv. Because the computation of both mean kv and GV per social group is relative to the current colony, the values for a social group or a specific animal might change as the colony's demography changes. For example, each time an animal reproduces or is removed from the colony, its social group's mean kv and GV estimates also change. As such, these estimates have to be recalculated to identify the genetically important individuals under the new situation.
The lack of a measurable mean kv among the group X males owing to their reproductive values indicates that they may have a positive effect on the colony's genetic diversity in the long run.17 Given their mean GU of more than 98%, these males carry founder alleles that are not present in other animals at Cayo Santiago; therefore, these unaffiliated males should remain in the colony.
The expected effective numbers of founders (mean fe and fg values) calculated from the pedigree data suggests that approximately 40 to 50 founders would be needed to produce the same amount of genetic diversity (mean GV exceeding 98%) among the present-day social groups. These estimates are more in line with previous recommendations26 of 20 or more founders than previously thought.13
Our current results demonstrate that this high effective number of founders has had positive effects on the Cayo Santiago colony's current genetic structure. Except for group MM, all of the social groups on Cayo Santiago showed higher Na, OH, and EH estimates than those of the conventional (nonSPF) colony at the Sabana Seca Field Station (Na, 6; OH, 0.69; and EH, 0.68).22 The SSFS colony was established over time through animal transfers from Cayo Santiago, including the relocation of an entire social group (group M) of about 40 animals in the mid1980s. The reduced gene flow due to the relative isolation of the SSFS colony from the other groups on the island is probably also reflected by the slightly lower genetic diversity at SSFS.
The mean Na and heterozygosity values of the Cayo Santiago colony are approximately 60% and 94% to 98% of the estimates generated from other captive Indian-origin animals.22 However, the occurrence of inbreeding due to consanguinity is extremely low, as indicated by the negative marker-based FIS values across all social groups except HH, whose FIS value of 0.02 likely reflects the fact that 1/3 of its members are products of consanguineous matings. Genetic, demographic, and behavioral data from the Cayo Santiago colony have shown a tendency toward inbreeding avoidance.36,37,42 Group fission has been linked to increased genetic differentiation among social groups.31 As large groups expand and then split into smaller groups, kinship among group members and differentiation among social groups increase in the Cayo Santiago colony.10,11 However, extragroup paternities were reported for almost 25% of births and might facilitate not only breeding between unrelated animals and decreasing inbreeding measurements41 but also curbing differentiation (pairwise FST) among the social groups. As such, in sharp contrast to those in previous studies,6,14 the Cayo Santiago social groups do not show significant intergroup differences. In addition, intergroup migration, which is predominantly male-biased on Cayo Santiago as in wild rhesus populations,32 might have minimized the pairwise FST values on Cayo Santiago.
In the current study, we characterized the genetic diversity of the rhesus macaque colony on Cayo Santiago by using pedigree and molecular genetic management approaches. Results show that, due to intergroup gene flow that is predominantly male-mediated, the colony comprises social groups that have sufficient genetic variation and have not differentiated significantly. Despite the use of different methods, our findings are in agreement with previous studies43,45 and indicate that the degree of inbreeding due to consanguineous mating is negligible.
However, there is a strong possibility that the inbreeding and kinship coefficients generated in the current study may have been underestimated. Genetic testing for parentage assessments, from which the pedigrees spanning 4 to 9 generations were derived, started in 1990; as such, the inbreeding and kinship estimates do not reflect the colony's entire genealogical history. Therefore, many animals that appear to be unrelated may, in fact, be related due to shared remote ancestry. In addition, because estimates of mean GV, GU, fe, and fg are based on kinship among animals, these matrices might have been overestimated as well. Regardless, the estimates of inbreeding and kinship generated in this study provide a starting point to gauge current levels of consanguinity to develop strategies to avoid risks of further inbreeding, minimize losses of heterozygosity, and maximize the colony animals’ genetic value and diversity over the long term.
Given its size, the population of rhesus macaques at Cayo Santiago is not sustainable, and colony managers urgently need to dramatically decrease animal numbers. The social group kinship estimates we generated in the current study, in conjunction with demographic data, provide a good starting point for culling strategies.19,20 Because animals in the Cayo Santiago colony are maintained for supporting numerous research projects onsite, plans to identify animals for removal and population control need to accommodate these priorities as well. For instance, colony managers plan to soon remove group HH because it is the group used least for research purposes and exhibits relatively high k, kv, and FIS values as well as intermediate levels of GU and genetic diversity. Because culling and new births lead to demographic changes, population genetic matrices including diversity estimates and kinship and inbreeding levels need to be monitored continuously.
Acknowledgments
We thank the dedicated staff at Cayo Santiago and Sabana Seca Field Station for their diligence and care in support of the CPRC's animal population. All research procedures of this study were approved by the CPRC and the IACUC of the University of Puerto Rico (protocol number 338300) and followed all international, national, and institutional guidelines for the proper care and use of animals.