26 Demographic and Genetic Management of Captive Populations JONATHAN D. BALLOU AND THOMAS J. FOOSE The last fifteen years have been a time of revolutionary change in the focus and goals of zoos. This change has been motivated by a fundamental philosophical shift in the zoo community toward establishing conservation, through both captive propagation and public education, as a primary pur- pose. A large part of the material included in this volume has resulted as much from this shift as from the develop- ment and application of new technology. The material discussed in this chapter has also resulted from this new conservation ethic. Two decades ago, the pri- mary focus of zoo animal rtianagement was the exhibition, maintenance, and occasional breeding of individual speci- mens. Collection from the wild, rather than reproduction in captivity, was the usual means of acquiring animals for zoo exhibits. As animals became increasingly difficult to obtain from the wild, however, captive propagation became a more important means for replenishing zoo collections. At the same time, zoos began to recognize their potential contri- bution to wildlife conservation: the preservation of wild ani- mal species in captivity over long periods of time and the potential to reintroduce captive-bred animals into the wild. Significant biological problems confront this effort. Small and fragmented populations, such as those that exist in zoos and to an increasing extent in the wild, have a low probability of long-term survival. Random catastrophic demographic events, reduced vigor due to inbreeding de- pression, and loss of genetic diversity due to genetic drift and inbreeding all have an effect on the survival of small populations (Gilpin and Soul? 1986). Success in managing these populations ultimately rests on managing them as populations, not just as individuals. There are also organizational problems involved in managing groups of individual zoo collections as cross- institutional biological populations. The international zoo At the time of writing, Thomas J. Foose was with the Captive Breeding SpeciaUst Group, Apple Valley, Minnesota. community has responded to this additional responsibility by forming associations and programs to organize and coor- dinate cooperative population management efforts (Hutch- ins and Wiese 1991; Shoemaker and Flesness, appendix 4, this volume). The primary purpose of such programs is to contribute to the conservation of species by providing res- ervoirs of genetic and demographic material that can be used periodically to reinforce, revitalize, or reestablish pop- ulations in the wild. This goal requires the development of propagation programs oriented to the maintenance of ge- netic diversity and demographic security. It is envisioned that conservation programs for many endangered species will interactively manage both wild and captive populations for mutual support and survival. This chapter delineates the principles, concepts, and techniques necessary to manage captive populations, con- centrating on those aspects critical to the long-term main- tenance of genetic diversity and demographic security. However, the scope of "population" management is broad and includes many subjects considered elsewhere in this volume: reproductive and behavioral research, data man- agement, genetic research, program administration, and of course, basic husbandry. The chapter begins with a discussion on establishing man- agement goals to guide the formulation of a propagation and management plan. Genetic and demographic analyses are a critical part of this process, but can only be conducted after basic pedigree and demographic data on the current and historical population have been compiled and organized. Methods of data collection and organization are discussed in the second section, as well as by Shoemaker and Flesness (appendix 4, this volume). This is followed by a discussion of methods and techniques for both genetic and demo- graphic analyses, using examples to illustrate various ana- lytic procedures. The chapter concludes with a section on population management, which covers the basic strategies and concepts for combining propagation goals with the re- 263 264 DEMOGRAPHIC AND GENETIC MANAGEMENT suits of the demographic and genetic analyses to form the basis for the program's specific institution-by-institution and animal-by-animal breeding recommendations. CAPTIVE MANAGEMENT GOALS The goals of most captive management programs approach one of two extremes: individuals are either intentionally se- lected to be well adapted to captive environments, or they are managed to preserve genetic diversity (Foose et al. 1986; Frankham et al. 1986). Which goal is chosen depends on the managers' objective for the population. Frankham et al. (1986) distinguish among four types of captive populations of primary interest to zoos: 1. Common display species Population goal: Establish a tractable, easily managed population well adapted to the captive environment Management strategy: Select for traits adapted to captivity Example: Fallow deer, Cervus dama (Hemmer 1986) 2. Endangered species in captivity for long-term conservation Population goal: Long-term maintenance of a viable population and preservation of genetic diversity Management strategy: Maximize retention of the foun- ders' genetic diversity and maintain a demographically stable population compatible with the limits of the cap- tive environment's carrying capacity Example: Golden lion tamarin, Leontopithecus rosalia (Kleiman et al. 1986); Siberian tiger, Panthera tigris al- taica (Foose and Seal 1986) 3. Rare species being propagated for immediate release into natural habitats Population goal: Rapid population growth and large- scale reproduction for immediate release Management strategy: Maximize reproduction in a cap- tive environment as similar as possible to the natural environment Example: Guam rail, Rallus ownstoni (Derrickson 1987) 4. Rare species not yet capable of self-sustaining reproduc- tion in captivity Population goal: Develop husbandry techniques for achieving self-sustaining capabilities Management strategy: Preferentially propagate individ- uals capable of reproducing in the captive environment. Once the population has grown to self-sustaining status, manage the population as type 2 above Example: Micronesian kingfisher, Halcyon cinnamom- ina (Bahner 1991) Adapting a species to captivity is appropriate for popu- lations of type 1 and, initially, type 4. Frankham et al. (1986) and Foose et al. (1986) discuss basic population management strategies for these types of populations. Main- tenance of genetic diversity is paramount for populations of type 2 and, eventually, type 4. Although the intent of this chapter is to describe popula- tion management techniques specifically for populations designated for long-term conservation, many of the con- cepts and techniques can be appropriately applied to popu- lations under other objectives. For example, data should be collected and demographic and genetic calculations rou- tinely performed for all types of populations. MANAGEMENT OF GENETIC DIVERSITY Maintaining genetic diversity and demographic security are the primary population management goals for long-term conservation. Management for genetic diversity minimizes change in the genetic constitution of the population while in captivity so that if and when the opportunity arises for ani- mals to be reintroduced into the wild, they will represent, as closely as possible, the genetic characteristics of the original founders used to establish the captive population (Hedrick et al. 1986; Lacy et al. 1995). Genetic variation is also the basis for adaptive evolution and must be retained to main- tain the population's potential to adapt to changing envi- ronments. Furthermore, there are a growing number of studies that indicate a general, although not universal, posi- tive relationship between genetic variation and both indi- vidual and population fitness (Hedrick et al. 1986; Allen- dorf and Leary 1986; Mitton and Grant 1984). These include a number of studies that have documented the dele- terious effects of inbreeding in captive populations (Rails, Ballon, and Templeton 1988; Lacy, Petric, and Warneke 1993) as well as the effects of low levels of variation on reproductive condition (Wildt et al. 1987; O'Brien et al. 1985). Last, maintaining genetic diversity preserves fu- ture management options, a strategy that will become in- creasingly important as knowledge of the genetic and demographic requirements of wild and captive populations expands. There are several kinds of genetic variation. Animals carry many thousands of genes located on chromosomes. The exact location of a gene on the chromosome is referred to as its locus. The terms gene and locus are often used in- terchangeably. Each gene may occur in alternative forms, called al?eles, that may produce slightly different genetic ef- fects, such as different pelage colors, eye colors, and so forth. The number of al?eles per gene or locus can vary from one (no diversity, or monomorphic locus) to many (poly- morphic locus). Most vertebrate species carry two copies of each gene (they are diploid), one inherited from each par- ent. A few species carry only one copy of each gene (they are haploid). Both copies may represent the same al?ele, in which case the animal is described as homozygous for that locus, or different al?eles, in which case the animal is heterozygous. Genetic diversity comprises both allelic diversity and het- erozygosity. Allelic diversity refers to the number of differ- ent al?eles at any given locus in the population. Heterozy- gosity is the percentage of heterozygous loci in a population or individual. Thus, genetic diversity can be measured in both individuals and populations. Both allelic diversity and heterozygosity are desirable, in captive populations; allelic diversity is important for a population's long-term ability to adapt, while heterozygosity is important for immediate ad- aptation ( AUendorf 19 8 6 ). Both allelic diversity and heterozygosity are lost in small populations (numbering a few tens to a few hundreds) BALLOU AND FOOSE 265 through the process of genetic drift. The al?eles passed from parents to offspring represent only a sample of the allelic variation of the parental generation. When only a few off- spring are produced, the genetic diversity of the offspring may be unrepresentative of the genetic diversity present in the parents. By chance alone, some al?eles may not be passed to the offspring; others may increase or decrease in fre- quency. These changes in the number and frequency of al- ?eles, as well as changes in heterozygosity due to this biased sampling process, are termed genetic drift. The population's average heterozygosity is often used as an overall indicator of genetic diversity since it lends itself well to theoretical considerations and usually provides a simple, accurate indicator of the loss of allelic diversity (Allendorf 1986). This is not always the case, however. Dur- ing bottleneck events, allelic diversity is lost much more readily than heterozygosity, and using heterozygosity alone overestimates the amount of genetic diversity retained after a bottleneck. Moreover, rare al?eles (those occurring at low frequencies in the population) may also be lost faster than average heterozygosity. The genetic goals of most captive propagation programs are currently based on maintaining heterozygosity but also consider bottleneck effects in the pedigrees. In managing genetic variation, it is important to distin- guish between single-locus and multilocus, or quantita- tive, variation. Single-locus variation is variation in traits regulated by single genes, while quantitative variation is variation in traits regulated by many genes. Although quan- titative variation is probably more important than single- locus variation for long-term evolutionary adaptation, het- erozygosity and additive quantitative variation are lost at approximately the same rate. Consequently, management strategies based on maintenance of heterozygosity generally apply to maintenance of additive genetic diversity as well (Lande and Barrowclough 1987). Selection can potentially retard or accelerate loss of ge- netic diversity. However, little is known about the role selec- tion plays in captive populations (Frankham et al. 1986; Ar- nold 1995). Variation can be selective (influenced by selection pressures) or selectively neutral (influenced not by selection pressures but by the random process of genetic drift). The conservative approach is to assume selective neu- trality, particularly in small populations in which genetic drift isiikely to be a stronger force than selection. Discus- sions of the management and maintenance of genetic varia- tion in this chapter refer primarily to single-locus neutral variation. (See Lacy et al. 1995; Lande and Barrowclough 1987 for further discussions of this issue.) Loss of genetic diversity is a function of population size and time. In general, the smaller the population, the faster the loss; the longer the period of time, the greater the total loss (fig. 26.1). Therefore, those developing management plans to conserve genetic diversity must consider the ques- tions, "How much genetic diversity is required?" and "How long should it be maintained?" The question of how much genetic diversity is required to retain long-term fitness and evolutionary potential in cap- tive populations is difficult to answer, since little is known < ^ 0.6 O ^ 0.4 O s 0.2 b X 0.0 ^ ^^ \ \ ^^ --^ Ne=100 \\ ^ Ne=50 ^ .JJe=10 ^ Ne=20 0 20 40 60 80 IOC GENERATION FIG. 26.1. Percentage of original heterozygosity retained over 100 generations for effective population sizes (N.) varying from 10 to 100. about minimum genetic requirements of populations. One approach might be to maintain as much genetic variation as possible. This does not provide useful guidelines for the development of management programs, however, because populations would have to be "as large as possible" to achieve this result. The finite availability of captive re- sources strictly limits the size and number of captive popu- lations that can be managed. Because loss of genetic diversity is a function of time, it is also important to consider how long genetic diversity must be maintained. The time scale for management programs will vary. Some species may need the support of a captive population for a relatively short time before they can be re- turned to the wild. However, for many, if not most, species, captive populations will have to be maintained for the long term, often over hundreds of years. The safest approach is to initiate all programs as if they will be for the long term. They can always be concluded earlier if conditions permit. A crude but general strategy that has been suggested in response to the questions "how much" and "for how long" is to preserve 90% of the founders' heterozygosity over a period of 200 years (Soul? et al. 1986). This "90%/200 year rule" originated from considerations of how long human population growth and development will continue to re- duce wildlife habitat. Its authors estimated that this "demo- graphic winter" will last between 500 and 1,000 years. However, they observed that some stabilization of human population growth is expected in the next 150 to 200 years. More importantly, they hypothesized that the current rapid development of biological technology, especially long-term storage of germ plasm (cryopreservation), will decrease de- pendence on populations of living animals for the preserva- tion of gene pools by the end of the twenty-first century. The authors despaired of the feasibility of developing human- managed programs that would continue for hundreds of years and concluded that 200 years would be a reasonable time frame for management of captive populations. The rec- ommendation to retain 90% of the original heterozygosity was based on the authors' consensus that the 10% loss "represents, intuitively, the zone between a potentially dam- aging and a tolerable loss of heterozygosity" (Soul? et al. 1986,107). Although this 90%/200 year rule of thumb is somewhat arbitrary, it does provide a starting point for establishing 266 DEMOGRAPHIC AND GENETIC MANAGEMENT population size goals. We have chosen to use this approach to determine genetic goals for captive propagation pro- grams and will focus our discussions on populations being managed under this goal. However, the questions "how much" and "for how long" will normally require species- specific answers (see "Demographic Analyses" below). In- dividual programs can modify both the time period of con- cern and the level of genetic diversity to be retained in response to the circumstances of the species. For example, populations that are to be reintroduced soon after a captive colony is established will require less concern about long- term maintenance of genetic diversity than populations destined to remain in captivity for many generations. In fact, more recently, population size objectives have been for- mulated in terms of 100 rather than 200 years, since this results in smaller, more realistic population sizes (Foose et al. 1995). Nevertheless, the general techniques for devel- oping a plan remain the same. Small populations are subject to demographic as well as genetic problems, and similar questions about demographic security should be considered in establishing goals for cap- tive propagation programs. Many demographic threats are stochastic (random) in nature. They include such easily ap- preciated chance events as environmental variation, natural disasters, and disease epidemics (Dobson and May 1986; Goodman 1987) as well as more subtle fluctuations in birth and death rates, including distortion of sex ratios, due to simple variation among individuals. Risks of demographic problems, like genetic risks, are functions of population size and time. The smaller the popu- lation and the longer the time period of management, the greater the risks. The relevant question is, what is the proba- bility of a population surviving (i.e., not going extinct) for a specified period of time? Or, in other words, what popula- tion size is necessary to achieve a high probability (e.g., 95%) of survival over a long time period (e.g., 200 years) (Shaffer 1987)? In most cases, captive populations large enough to achieve the usual genetic objectives will also be large enough to insure high survival probability over the time period of concern because management in captivity will usually, though not always, be able to moderate demo- graphic stochasticity caused by environmental variation. This is not likely to be the case for small wild populations, in which environmental variation has a tremendous effect on population survival (Goodman 1987). DATA COMPILATION The most important task in the development of a captive propagation plan is compiling the basic data required for population analysis and management. Data may already have been compiled in a variety of different forms if a cap- tive population exists or has existed in the past. The best source of compiled data is a studbook, which is a chro- nology of a captive population listing vital information on animal identities, sexes, parentage, and birth and death dates, as well as information on animal movements between institutions (Shoemaker and Flesness, appendix 4, this vol- ume; Glatston 1986). Currently there are approximately 250 international and regional studbooks (T. J. Foose, pers. comm.), many of which are available as computerized data bases, and the number is growing annually. If a studbook does not exist or is out of date, one must be compiled from original sources. Historical and current data should be collected from all institutions that have had or currently have individuals of interest. Historical data are critical for determining the ancestry of living animals and estimating certain genetic and demographic parameters (e.g., population growth rates, generation lengths, effective population sizes). Data compilation should begin with reference to a num- ber of sources that summarize data on captive populations. Potential sources of data are: ? International Species Information System (ISIS). ISIS is a computerized data base containing information on animal identities, birth and death dates, genealogies, and move- ments (Seal, Makey, and Murtfeldt 1976; Shoemaker and Flesness, appendix 4, this volume). ISIS collects data from institutions worldwide and is the best starting point for compiling population data if no studbook is available. ? International Zoo Yearbook (IZY). IZY provides an an- nual list of birds, mammals, reptiles, amphibians, and fishes bred in captivity (Olney 1986). Although only numbers and locations are presented, these annual listings are useful for identifying institutions that once had or currently have spec- imens of interest. ? Species Registries. Registries are single-species listings of numbers and locations of animals. They do not contain the detailed vital information necessary for population analysis, but do provide a starting point for locating institutions that once held specimens (Shoemaker and Flesness, appendix 4, this volume). ? In-House Institutional Records. In-house inventory rec- ords are the primary source of data. Once institutions that have had or currently have specimens of interest are identi- fied, they can be contacted for information on the history, status, and details of their collection. Again, it is important to stress the importance of collecting historical data. The basic data required on each animal for population analysis and management are: ? Individual identification: a simple numeric lifetime iden- tity (i.e., studbook number). To achieve this identification, it may be necessary to link a series of different ID numbers (e.g., the local ISIS specimen numbers) the animal has had as it has moved among institutions ? Sex ? Birth date ? Death date (it is vital to record stillbirths and aborted fetuses) ? Parentage ? Whether the individual is wild-caught Date and site of capture Possible relationship to other wild-caught animals (i.e., several animals captured in a nest or herd) Date animal entered captivity ? Date animal left captivity or was lost to follow-up (rein- troduced into the wild, escaped, lost track of) ? Institutions/facilities where it has been, with dates or shipments BALLOU AND FOOSE 267 2 3 4 5 t If i_f fW ?\_ 16 ID Sex Sire Status 1 F Wild Wild Dead Founder 2 M Wild Wild Dead Founder 3 F Wild Wild Dead Founder 4 M Wild Wild Dead Founder 5 F Wild Wild Dead Founder 6 M 3 2 Dead 7 F 3 2 Dead 8 M 1 6 Living 9 M 7 4 Living 10 F 5 4 Dead 11 M 5 4 Dead 12 M 7 8 Dead 13 F 7 8 Dead 14 F 10 11 Living 15 M 10 11 Living 16 M 13 9 Living FIG. 26.2. Pedigree of a population founded with 2.3 individuals. Squares, males; circles, females; open squares and circles, living animals. Numbers are unique identifiers for each individual. The pedigree listing is presented on the right. ? Information on circumstances and cause of death ? Reproductive condition (e.g., castrated male, postrepro- ductive female) ? Group compositions (who is housed with whom) ? Reproductive opportunities (whether animal was given opportunities to breed, and when) ? Information on past breeding experience (e.g., proven breeder.') ? Tattoo or other permanent identification marks (e.g., transponder number) ? Carcass disposition and tracer (e.g., "Sent to Uiliv. Kansas Museum, #12345") ? Miscellaneous comments (unusual behavior or pheno- type, etc.) When dealing with unknown or missing data, record as much information as possible. Dates or events that are par- tially or completely unknown should be noted as such. Un- known dates are a particular problem. Usually events are dated to the nearest day, month, or year. Uncertain parent- age is also a common problem, particularly in herd situa- tions. Record all potential parents and, if possible, indicate the likelihood (e.g., based on behavioral data) of each being the actual parent (e.g., '^Comment: Potential Sires/Likeli- hood: Stbk 123/50%; Stbk 1221/25%; Stbk 1212/25%). Most analyses require that the data be computerized for easy access and manipulation. Standard formats for pedi- gree data have been developed (Shoemaker and Flesness, appendix 4, this volume), and a number of computerized studbook management and analysis software packages are available, including the Single Population Animal Record Keeping System (SPARKS: ISIS 1991), and the Zoo Re- search Studbook Management System (Princ?e 1989). GENETIC ANALYSES The purpose of genetic analyses is to describe the genetic characteristics of a population that are important for its management. These include information on the number of founders, the distribution of their genes among living ani- mals, the relationships among individuals in the living population, and the capacity of the population to retain ge- netic variation. Results of genetic analyses are used in con- junction with results of demographic analyses to arrive at a carrying capacity for the captive population and formu- late recommendations for managing the population at this carrying capacity. A step-by-step procedure for calculating these genetic characteristics follows. 1. Construct the pedigree for each animal in the population. This "pedigree" can be in the form of a standard pedigree chart and/or simply a listing of each individual with its par- ents (fig. 26.2) that will be used with various pedigree analy- sis algorithms and computer programs. Pedigree charts are particularly useful for identifying pedigree bottlenecks as well as ancestors of special interest. 2. Identify the founders of the population. A founder is an animal that ? is from outside the population (usually the wild) ? has no known ancestors in the population at its time of entry ? has descendants in the living population or is currently liv- ing and capable of reproduction (a potential founder) Unless it is known otherwise, founders are assumed to be unrelated to each other. When the relationships of wild- caught animals are known or suspected (e.g., several chicks captured in the same nest), it is necessary to create "hypo- thetical" parents (or other ancestors) to define those rela- tionships. These hypothetical ancestors are then defined as founders. It is useful to name all hypothetical founders in an easily identifiable fashion (e.g., studbook numbers begin- ning with the letter H). Figure 26.3 illustrates the identification of founders in the captive population of black-footed ferrets, Mustela nigripes (Ballou and Oakleaf 1989). "Willa," "Emma," "Annie," 268 DEMOGRAPHIC AND GENETIC MANAGEMENT 653-54 ?40-41 WILL* EMMA ANNIE -?- J#f # TABLE 26-1. Founder Contributions for the Pedigree Illustrated in Figure 26.2 ? D DEXTER CODY 680-81 ROSE CREEK PAIR 668^?9 671-72 NEVER IN CAPTIVITY m (8IS-8T MOLLY 1148-49 MOM SCARFACE JENNY CUTLIP COLLENE ROCKY SUNDANCE FIG. 26.3. Pedigree of the founding population of black-footed ferrets. Squares, males; circles, females; solid squares and circles, founders; double-bordered squares and circles, living animals. A question mark indicates uncertain parentage. (Reprinted with permission from U. S. Seal et al., eds., Conservation Biology and the Black-footed Ferret [New Haven: Yale University Press, ?1989].) "Mom," "Jenny," "Dean," and "Scarface" are shown as founders since they are wild-caught, have no known ances- tors in the captive population, and are thought not to be closely related to each other. Although "Molly" has known relatives, they were either never in captivity or died without producing offspring; she is therefore considered a founder. Female "653-54" and male "640-41" are also founders be- cause "Dexter," who is living, is an offspring of both and "Cody" is an offspring of male "640-41." Molecular genetic analyses (e.g., DNA fingerprinting) can be useful in examining relationships of wild-caught ani- mals or even captive-born animals without pedigrees (Ryder and Fleischer, chap. 25, this volume; Morin and Ryder 1991; Avise et al. 1995; Ashworth and Parkin 1992; Geyer et al. 1993). However, these techniques may be useful for determining only first-order relatedness (e.g., full sibling or parent-offspring relationships). 3. Compute the genetic contribution of each founder to each living individual as well as to the living population as a whole. Founder contribution is the percentage of an individual's or a population's genes that have descended from each foun- der. Calculations are based on the Mendelian premise that each parent passes (on average) 50% of its genes to its off- spring, 25% to its grandoffspring, and so forth. Each foun- der's genetic contribution to living individuals can be calcu- lated by constructing each individual's pedigree back to the founders and applying these Mendelian rules of segregation. Living individuals Pop. Founder 8 9 14 15 16 avg. Retentiot 1 .50 0 0 0 .13 .126 .500 2 .25 .25 0 0 .31 .162 .484 3 .25 .25 0 0 .31 .162 .487 4 0 .50 .50 .50 .25 .350' .803 5 0 0 .50 .50 0 .200 .612 Mean kinship .150 .228 .238 .238 .244 Note: Average proportion of each individual's genes that has descended fron each founder. See text for explanation of retention and mean kinship. The founder's genetic contribution to the current popula- tion's gene pool is its contribution averaged across all living individuals (table 26.1). Algorithms and computer programs are available for calculating founder contributions from pedigree data (Ballou 1983; Lacy 1990a). Founder contributions in most captive populations are highly skewed, usually due to disproportionate breeding of a small proportion of the founders early in the population's history (fig. 26.4). Genetic diversity potentially contributed by the underrepresented founders is thus lost or at high risk of being lost due to genetic drift. 4. Calculate the loss of founder al?eles due to genetic drift and pedigree bottlenecks. Further loss of genetic diversity occurs when genetic drift causes founder al?eles to be lost from the population. Ex- treme cases of genetic drift are often referred to as pedigree bottlenecks, occurring when the genetic contribution of a founder passes through only one or a few individuals. For example, only 50% of a founder's genes survive to the next generation if it produces only one offspring, 75% survive if it produces two offspring, and so forth. Bottlenecks may oc- cur during the first generation of captive breeding if only one or two offspring of a founder live to reproduce. How- ever, the genetic drift caused by such bottlenecks can occur at any point in the pedigree, resulting in gradual erasion of the founders' al?eles. The rhore "pathways" a founder's genes have to the living population, the less loss will occur. Therefore, even though a large proportion of a population's gene pool may have descended from a particular founder, the population may represent only a fraction of that foun- der's genetic diversity. The proportion of a founder's genes that survive to the current population is referred to as gene retention or gene survival. Although exact methods for calculating retention have been developed (Cannings, Thompson, and Skolnick 1978), it is often estimated using Monte Carlo simulation procedures ("gene dropping": MacCluer et al. 1986). "Gene drop" procedures assign two uniquely identifiable al?eles to each founder. Al?eles are passed, randomly, from parents to offspring according to the rules of Mendelian segregation, and the distribution and pattern of al?eles among living ani- mals are examined after each simulation (fig. 26.5). The simulations are repeated several thousand times, and the re- ?ALLOU AND FOOSE 269 0.40 0.35 0.30 0.25 O 0.20 ? 0.15 0.10 0.05 0.00 FIG. 26.4. Founder contributions in the 1988 captive population of Przewalski's horses, Equus przewalskii. The distribution is heavily skewed due to disproportionate breeding among the founders early in the population's history. "DOM" is a Mongolian domestic mare that was bred to a Przewalski's horse in 1960. 212 39 211 40 1 231 52 5 18 17 11 DOM 12 FOUNDER GENE DROP ANALYSIS ^ ^ ???U ????o GENE DROP ANALYSIS ^ c (se) Us) 4 4 n2\ 15 ^3) 55 us) 55 (45) FIG. 26.5. Gene drop analysis, (a) Each founder is assigned two unique al?eles, (b) The al?eles are then "dropped" through the pedigree according to the rules of Mendelian segregation; each al?ele has a 50% chance of being passed on to an offspring. At the end of the simulation, the pattern and distribution of al?eles in the living population (bottom row) are examined. The simulation is repeated several thousand times and results are averaged across simulations to give gene retention. Note that al?ele 2 from founder A and al?ele 6 from founder C have been lost. tention for each founder is calculated as the average per- centage, across all simulations, of the founder's al?eles that have survived to the living population. The retention esti- mates for the sample pedigree shown in figure 26.2 are listed in table 26.1. The retention for founder 1 is only 50% be- cause she produced only one offspring, while the retention for founder 4 is higher because his genes have multiple path- ways to the living population. Gene drop analyses provide information about the distri- bution of founder genes in the living population that data on founder contribution do not. This is particularly true for deep, complex pedigrees, in which using founder contribu- tion alone to determine the founders' genetic contribution to the population can be very misleading. Figure 26.6 illus- trates the effects of pedigree structure on gene flow in two pedigrees that have equal levels of inbreeding and founder contribution but different levels of gene retention. Since both skewed founder contributions and loss of al- ?eles due to genetic drift result in the loss of founders' ge- netic diversity, the genetic contribution of the founders to the gene pool may be less than expected. Lacy (1989) intro- duced the concept of founder genome equivalent (fg) to il- lustrate the combined effect skewed founder contribution and genetic drift have on the genetic diversity of a popula- tion, fg is the number of founders that would be required to obtain the levels of genetic diversity that are observed in the current population if the founders were all equally repre- sented and had retained all their al?eles in the living popu- lation. It is calculated as /. 1 (26.1) where N^is the number of founders, /?, is the founder contri- bution of founder / to the population, and r, is founder ?'s retention. Our sample population in figure 26.2 has 5 foun- 270 DEMOGRAPHIC AND GENETIC MANAGEMENT 1 A j 1 s *, tJ tJ mk? N (26.2) (yn^ RETENTION = 0.47 I I ' RETENTION = 0.61 FIG. 26.6. Gene drop analysis applied to two similar pedigrees. Squares, males; circles, females; open squares and circles, living animals. Pedigrees A and B each have four founders and four living animals. Founder contribution to living animals is identical for both pedigrees (25% from each founder) and all animals are non-inbred. However, genes from each founder in pedigree A pass through only one individual, creating a severe bottleneck. Genes from each founder in pedigree B pass through two individuals, creating less severe bottlenecks. As a result, more of the founder al?eles in pedigree B are retained in the living population than in pedigree A. ders, but because of retention problems and skewed founder contribution, they have an /j of only 2.8. In essence, they behave genetically hke 2.8 idealized founders. The fg values are often calculated with living founders excluded from the analysis. Living founders have 100% re- tention, and including them assumes that their al?eles have been "captured" in the population, even though they may have no living descendants. Excluding living founders pro- vides a more realistic summary pf the genetic status of the population, particularly if there are many founders who are not likely to contribute offspring to the gene pool. Compar- ing the fg calculated with living founders excluded with the fg when they are included shows the contribution that ge- netic management can make if 100% of the living founders' genes can be retained in the population. 5. Calculate measures of genetic importance for each individual. When selecting animals for breeding, it is useful to rank in- dividuals according to their genetic "importance." Individ- uals carrying genes from overrepresented founders are not as genetically valuable as those carrying genes from under- represented founders. A number of methods have been used to rank animals according to their genetic importance. One is the Founder Importance Coefficient {fie: Ballou and Lacy 1995), which is the weighted average of the founder contri- butions within each individual, with the population founder contribution acting as the weights. However, the ftc does not take into consideration the effect of retention on an individual's genetic value and could potentially produce misleading results. For this reason, it has not been exten- sively used to measure genetic importance. A more appropriate measure of genetic importance is mean kinship {mk). Mean kinship is the average of the kin- ship coefficients between an individual and all living indi- viduals (including itself) in the population (Ballou and Lacy 1995; Lacy 1990a): where mk, is the mean kinship of individual i; k?? is the co- efficient of kinship between individuals / and /; and N is the number of living animals in the population (Ballou and Lacy 1995). The kinship coefficient is the probability that two ai- leles, taken at random from two individuals, are identical by descent (Crow and Kimura 1970). It is a measure of the genetic similarity of the individuals and is the same as the inbreeding coefficient of any offspring they would produce. Individuals who are carriers of rare genes will have low val- ues of mk because they have few relatives in the population, whereas individuals who carry genes shared with many in- dividuals will have a high mk. Ranking individuals accord- ing to their mk values provides a quick method for identi- fying genetically important animals. Both mk and fg relate directly to maintenance of genetic diversity in populations. Gene diversity, the expected level of heterozygosity of the population, can be calculated as either 1 ? mk or 1 ? 'i/lfg-, the two expressions are equiva- lent. Thus, breeding strategies to minimize mk or to maxi- mize fg will both maximize gene diversity (Ballou and Lacy 1995). Computer programs are available for calculating mk values (Lacy 1990a). Values of mk for the sample pedigree in figure 26.2 are shown in table 26.1. Another measure of genetic importance is genetic unique- ness, which is the probability that a gene carried by an in- dividual is unique (i.e., not carried by any other living ani- mal). Genetic uniqueness is calculated using the gene drop analysis described above and can also be used to rank indi- viduals by genetic importance (Ballou and Lacy 1995). 6. Calculate inbreeding coefficients of all individuals in the population as well as kinship coefficients between all living individuals. Inbreeding is the mating of related individuals. If two par- ents are related, their offspring will be inbred, and the more closely related the parents are, the more inbred their off- spring will be. The degree to which an individual is inbred is measured by its inbreeding coefficient (f), which is the probability of its receiving the same al?ele from each parent (i.e., that the al?eles are identical by descent). Offspring of father/daughter, mother/son, or full-sib matings are 25% inbred; offspring of first-cousin matings are 6.25% inbred. Inbreeding coefficients are used to examine the effects of in- breeding in the population (see below) and to determine the degree of relatedness between individuals. Algorithms and computer programs for calculating inbreeding coefficients from computerized pedigrees are available (Ballou 1983; Boyce 1983; Lacy 1990a; ISIS 1991). 7. Estimate the population's effective population size. The extent and rate of loss of genetic diversity depends on the size of the population. However, the size of relevance is not simply the number of individuals; rather, it is the ge- netically effective population size (N^). The effective size of a population is a measure of how well the population maintains genetic diversity from one generation to the next. BALLOU AND FOOSE 271 Genetic diversity is lost at the rate of I/IN^ per generation. Populations with small effective population sizes lose ge- netic diversity at a faster rate than those with large effective population sizes (see fig. 26.1). The concept of N^ is based on the genetic characteristics of a theoretical or ideal population that experiences no se- lection, mutations, or migration and in which all individuals are asexual and have an equal probability of contributing offspring to the next generation. Extensive population ge- netic models have been developed to examine the loss of ge- netic diversity over time in an ideal population (Kimura and Crow 1963). However, real populations differ greatly from the ideal. Estimating how rapidly a real population loses ge- netic diversity requires comparison of the genetic character- istics of the real population with those of the ideal popula- tion. A real population that loses genetic diversity at the same rate as an ideal population of size 50 (1% per genera- tion) has an effective population size of 50, regardless of its actual size. Strictly defined, the effective size of a population is the size of a theoretically ideal population that loses ge- netic diversity at the same rate as the population of interest (Hedrick 1983). Once an effective population size is calcu- lated, the rate at which the population loses genetic diversity can be estimated. In general, the effective size of a population is based pri- marily on three characteristics: the number of breeders, their sex ratio, and the relative numbers of offspring they produce during their lifetime (their "lifetime family size"). Each of these characteristics can strongly influence a population's ef- fective size. In general, a large number of breeders will pass on a larger proportion of the parental generation's genetic diversity than will only a few breeders. A heavily biased sex ratio in the breeders will likely result in loss of genetic diver- sity since the underrepresented sex will contribute an un- equally large proportion of the offspring's genetic diversity. An equal sex ratio is preferable since it assures that the gene pool will receive genes from a larger number of breeders than when the sex ratio is highly skewed. Differences in family size also result in loss of genetic diversity since some individuals contribute few or no offspring to the gene pool while others producing large numbers of offspring contrib- ute more to the gene pool. The amount of genetic diversity passed from one generation to another is, in general, maxi- mized when all breeders produce the same number of young (i.e., family sizes are equal and the variance in family size is zero). One method commonly used to calculate N^ is to assume that the population is not growing, has nonoverlapping gen- erations, and that family sizes have a Poisson (random) dis- tribution?this is the theoretically expected distribution if each individual in the population has an equal opportunity to breed. Under these assumptions the effective population size can be calculated from R (26.3) where N? and Nf are the total numbers of different adult males and females in the population over one generation. Unfortunately, in captive populations, family size distri- butions are rarely determined by random mating and are not Poisson in form. In unmanaged populations many more adults than expected may fail to produce offspring, while in intensively managed populations fewer adults than ex- pected may fail to produce. A more accurate method of es- timating Ne incorporates information on family size. The family size of an individual is the total number of offspring it produces during its lifetime and that survive to adulthood. Ideally, only individuals who have completed their repro- ductive lives (are postreproductive or have died) should be used for these calculations. However, Lande and Barrow- clough (1987) describe a method for estimating future re- productive performance for individuals still breeding. Both the mean family size (k) and the variance in family size ( Vk ) need to be calculated across all individuals; individuals who fail to breed must be included, contributing family sizes of zero. These parameters can be calculated directly from stud- book data. Since k and V^ are measured over individual lifetimes, they provide accurate estimates only if the population has been stable for a relatively long time (several generations). This is unlikely for most rapidly changing captive popula- tions, and calculations of k and Vk may not represent current population trends. Accurate estimates of current ef- fective sizes are therefore difficult to calculate. With estimates of k and V^ for each sex (if available), it is possible to calculate N^ separately for each sex. The effec- tive size of males is: Ne,?| = N? 1 + -^ (26.4) 1 where k^ is the average number of young surviving to adult- hood across all males; N? is the number of adult males in the population during a generation; and Vk(m)5 the variance in number of young surviving to adulthood, is defined as: V, l<(m) (26.5) where the sum is over the number of adult males in the population, and k? is the number of offspring surviving to adulthood for each male. The effective size for females {N^^i-,) is calculated from equation (26.4) using family size data for females. The effec- tive size for the overall population is then determined using equation (26.3), replacing N? with Ne(?| and Nf with N^if,. The effective population size can be compared with the true population size (N) by calculating the ratio of N^/N. It is theoretically possible for the effective size to be almost twice the true size if the variance in family size is zero. In reality, N^ is almost always less than the true size. Ratios of NJN in captive populations not genetically managed have been measured at between 0.3 and 0.5 (Flesness 1986; table 26.2). Effective size may change radically over time. Lack of ge- netic management in the past may have caused NJN ratios to be very low. Therefore, the data used to estimate the cur- rent effective size should be relatively recent (over the last 5 years). Likewise, estimates of future effective sizes should be based on future management goals (i.e., attempts to maxi- 272 DEMOGRAPHIC AND GENETIC MANAGEMENT TABLE 26.2. Effective Population Size for the Captive Population of Golden Lion Tamarins during a One-Generation (6-year) Period between 1981 and 1987 Variable Males Females Number of adults in the population* Mean number of offspring'' Variance in number of offspring' Effective size by sex (equation 26.4) N? = 269 N, = 275 k^ = 1.7 k, = 1.6 Vk. , = 12.1 Vu = 13.5 N? ? = 58.4 N,, = 48.6 Overall effective size (equation 26.3) N^ = 106 Actual population size'' N = 357 Ratio of effective size to real size NJN = 0.30 " Calculated from total number of males and females that lived in the population between 1981 and 1987. ''Mean number of offspring surviving to age of sexual maturity (18 months) per adult. 'Variance in number of offspring surviving to age of sexual maturity. ''Harmonic mean of the population size between 1981 and 1987. mize Ne while also aspiring to other objectives, such as zero population growth). Moreover, NJN ratios may be very different during the growth and carrying capacity phases of the population. Table 26.2 illustrates a calculation of the ef- fective population size for one generation in the captive his- tory of the golden lion tamarin. There is an appreciable literature on effective population sizes (Lande and Barrowclough 1987; Ballon 1987a; Ry- man et al. 1981; Hill 1972; Kimura and Crow 1963). Most computational methods (including those above) are derived for populations with nonoverlapping generations, rarely the case in vertebrate species. Lande and Barrowclough (1987) and Harris and AUendorf (1989) present methods for calculating effective population sizes in populations with overlapping generations. Those interested in a more detailed discussion should refer to their original articles. 8. Conduct various biochemical analyses that measure genetic variability and relationships. Estimates of genetic variation are helpful primarily for iden- tifying the extent of genetic differences between populations or taxa (Wayne et al. 1986; Ryder and Fleischer, chap. 25, this volume). If large differences (e.g., chromosomal differ- ences) are found within a managed population, it may be necessary to reevaluate the goal of the program and possibly manage the population as two separate units (Templeton et al. 1986). In the words of an emerging terminology, large genetic differences may be evidence that there is more than one "evolutionarily significant unit" (ESU) within a species (see Barrowclough and Flesness, chap. 24, this volume). Interbreeding individuals from different ESUs may result in reduced survival and reproduction (outbreeding depres- sion; see below). Unfortunately, criteria have not yet been developed to indicate what magnitude of genetic differences constitutes separate ESUs. Where different ESUs are sus- pected, additional analyses on morphological, behavioral, and biogeographical considerations should be conducted and considered. Levels of genetic variation may provide information on the demographic and genetic history of the population. However, the goal of maintaining genetic diversity should not be abandoned if Httle or no variation is measured. It is not yet clear how representative currently measurable varia- tion may be of the actual genetic diversity in an individual or population. There may be more diversity than can be de- tected by existing methods, and in the face of such uncer- tainty, the only prudent course of action is to manage as if diversity were present. In any case, it may be imperative to maintain what little genetic variation is present for the long- term fitness of the population. As mentioned earlier, biochemical analyses may also be useful in resolving questions regarding parentage and in identifying relationships among founders. Long-term bio- chemical studies can be used to monitor the change of ge- netic variation in a population over time (Wayne et al. 1986). A comparison of the empirically estimated loss of variation with the theoretical loss of variation estimated from "gene drop" or similar analyses might provide insight into the types and degree of selection acting on captive populations. It is not recommended that selection of breeding individ- uals be based on individual levels of heterozygosity esti- mated from biochemical methods. Heterozygosity at a few isozyme loci is often a poor indicator of overall individual heterozygosity (Hedrick et al. 1986; Lande and Barrow- clough 1987). In addition, specific selection for known het- erozygous loci (e.g., MHC loci: Hughes 1991) may select against heterozygous loci not sampled electrophoretically and decrease the overall level of genetic diversity in the population (Haig, Balloti, and Derrickson 1990; Miller and Hedrick 1991; Gilpin and Wills 1991; Vrijenhoek and Le- berg 1991). 9. Adjust for uncertain parentage. Lack of individual identification and uncertain parentage will complicate genetic analyses. This problem is common in species managed as herds, in which individual dams are often not identified, and in species in which more than one breeding male has access to females, resulting in uncertain paternity. Depending on the extent of unknown parentage, a number of different approaches can be taken. a. Exclude individuals with unknown parentage or ancestors from the managed population. This approach is practical only if few individuals are involved and they are not otherwise important to the population. In such cases, a determining factor in the decision will be the percentage of an individual's genes that have descended from unknown ancestors. Small percentages of unknown ancestry may be acceptable. Animals who have some degree of unknown an- cestry but also have ancestors whose genes are relatively rare could be kept in the population to perpetuate the con- tribution of underrepresented founders (see "Population Management" below). WiUis (1993) points out that exclud- ing animals of unknown parentage may result in maintain- ing lower levels of diversity than retaining them. b. If questionable parentage is limited to only a few individuals, run the genetic and demographic analyses under all possible combinations to give the complete range of outcomes. If the results are insensitive to parent- age possibilities, the questionable parentage should have little effect on management decisions. If the results are BALLOU AND FOOSE 273 highly dependent on parentage, other options for analyzing the pedigree should be explored. An alternafive strategy is to select the worst-case scenario as the basis for manage- ment decisions. c. Use the potential parent most likely to be the true parent for the pedigree analysis. d. Create hypothetical parents that represent an ag- glomeration of all potential parents. If the potential par- ents are all equally likely to be the true parent, then a new, average, "hypothetical" parent can be created. It is given a "dummy" ID number for the genetic analysis and consid- ered as the sire (or dam) of the offspring in question. The founder contribution of the "hypothetical" parerit is then calculated as the average of the founder contributions of the possible parents. Creating an "average" parent is most ap- propriate if the founder contributions of the potential par- ents are not too different. If the differences between the potential parents are very large (especially if the potential parents are founders), other options should be considered. Inbreeding coefficients are calculated by assuming that the "hypothetical" parent is unrelated to its mate and the rest of the population. In most cases, this v/ill underestimate in- breeding coefficients for the descendants of the unknown parent(s). It is better to assume worst-case scenarios: that is, the closest relationships among putative parents. e. When groups have been managed for several gen- erations without individual animal identification, create hypothetical pedigrees. "Black box" populations are com- mon in herding species kept in large groups. An example of how a worst-case strategy can be used to utilize at least some of the founder potenti?il in such groups is the AZA Species Survival Plan for Grevy's zebra, Equus grevyi. With this species, there were a number of very large herds in which individual parentage was not recorded. However, considerable useful information was known: each herd had been established by a number of founder animals (usually one stallion and several mares); there had been a limited number of further immigrants of known origin to the herds; only one stallion was in each herd in any breeding season; and the dates of birth of all individual foals born into the herds were documented. It was first assumed that a single founder female estab- lished the herd; that is, all actual founder females were amalgamated into a "hypothetical" founder female that was assigned a "dummy" ID number. All offspring born during the first few years (or a period of time equal to the age of sexual maturity for the species) were then considered to be offspring of the herd stallion and this hypothetical dam. Af- ter this first cohort, it was assumed that daughters of this pair would have matured and bred with their father. There- fore, an Fi hypothetical female was created. The parents of this female were the herd stallion and the hypothetical foun- der female. Thereafter, all offspring born in the herd traced 75% of their genes to the founder stallion and only 25% to the hypothetical founder female. Such a strategy is most useful if the herd was established by known founders. Obviously, this strategy will underesti- mate the actual number of founders for the herd as well as the genetic diversity involved. Inbreeding coefficients will be overestimated when a number of different breeding animals are combined under one "hypothetical" parent. However, within the herd, inbreeding coefficients will be relative, and closely related individuals will have higher coefficients than less closely related individuals. When "hypothetical" parents or founders are created to satisfy genetic analysis requirements, individuals with un- known ancestors in their pedigree should be clearly labeled to indicate that both their founder contributions and in- breeding coefficients are based on hypothetical data. DEMOGRAPHIC ANALYSES The purposes of demographic analyses are to calculate basic life tables and population dynamics for the captive popula- tion; determine a carrying capacity for the captive popula- tion compatible with genetic, demographic, and resource limitations; and determine lifetime and annual reproductive objectives for each individual. 1. Demographic Characteristics of the Population Basic demographic data of interest to the captive manager are: a. Size of the current population and number of in- stitutions and geographic regions over which it is distributed. This tabulation will usually be an immediate result of the data collection process (see above). It may also be useful to estimate the numbers and distribution of other taxa with similar "captive ecologies" (i.e., space and re- source requirements) in order to estimate the total captive carrying capacity for the species. b. Age and sex structure of the population. These dis- tributions show the proportion and sex ratio of the popula- tion in each age class (fig. 26.7). Males FIG. 26.7. Male and female age structures for the 1983 captive Sumatran tiger, Panthera tigris sumatrae, population. The solid area shows the number, sex ratio, and ages of proven breeders in the population. (From Ballon and Seidensticker 1987.) 274 DEMOGRAPHIC AND GENETIC MANAGEMENT c. Age-Specific survivorship and fertility rates. Survi- vorship and fertility rates are calculated separately for each sex from age-specific tallies of birth and death events in the population and are usually presented in the form of a life table (Caughley 1977). A variety of different procedures have been developed to calculate life tables (Caughley 1977; Foose and Foose 1983; Foose and Ballou 1988). Table 26.3 is a life table for the captive population of golden lion tamarins. TABLE 26.3. Life Table Calculations for Captive Female Golden Lion Tamarins for the Period 1981-1987 Age class (years) P.' tj- m/ 0 0.61 1.00 0.00 1 0.87 0.61 0.00 2 0.90 0.53 0.32 3 0.92 0.48 0.60 4 0.92 0.44 0.74 5 0.91 0.40 0.78 6 0.90 0.37 0.72 7 0.90 0.33 0.59 8 0.89 0.30 0.53 9 0.88 0.27 0.53 10 0.86 0.23 0.53 11 0.82 0.20 0.65 12 0.76 0.17 0.89 13 0.78 0.13 0.88 14 0.88 0.10 0.52 15 0.97 0.09 0.13 16 1.00 0.08 0.00 17 1.00 0.08 0.00 18 1.00 0.08 0.00 19 0.75 0.08 0.00 20 0.25 0.06 0.00 21 0.00 0.02 0.00 '?Proportion of females surviving from age class x to age x + 1. ''Proportion of offspring surviving to age class x. 'Average number of female offspring born to a female of age x. Source: Ballou 1987b. Note: Demographic calculations (see Caughley 1977): Net reproductive rate (Ro) 21 R? = 'ZLm, = 2.43 Exponential growth rate (r) 21 '2,e-"Lm^ = 1.00 Solution for r = 0.149 ?= e'= 1.16 Generation length (T) 21 T = ^xe-"Lm, = 5.28 x=0 Growth rate per generation = A^ = 2.19 No. female births required for zero population growth at carrying capacity of 250 females (see Keyfitz 1968) = 250* ^7- = 41 Because life tables derived from historical data may re- flect past rather than current population trends, life tables should be based on relatively recent data (e.g., the last 5 years). However, in many populations there may not be enough recent data to construct a reliable life table, and in those cases historical data can be used to increase sample sizes. In situations in which no data are available (e.g., because no captive population has previously existed), life table data can be estimated from basic life history data on the species (age of first reproduction, age of last reproduction, litter size) as well as from data on similar species in captivity. ISIS is a valuable source for rough survival and fecundity rates of captive populations (Shoemaker and Flesness, ap- pendix 4, this volume). d. Any factors that adversely affect survival and re- production rates and patterns. Evidence of reproductive failure and high mortality rates should be investigated im- mediately. In addition to medical, nutritional, physiological, and behavioral causes, potential genetic causes should be examined. The deleterious effects of inbreeding on survival and reproduction (inbreeding depression) have been ob- served in many captive populations (Rails and Ballou 1983; Templeton and Read 1984). Although genetic in cause, its effects are demographic and can include lower population growth rates, smaller populations, and, consequently, even higher rates of inbreeding (Gilpin and Soul? 1986). If such inbreeding depression is severe, inbreeding should be minimized. Outbreeding depression, a reduction in fitness caused by hybridization between individuals from differently adapted or coadapted populations, can also reduce breeding success (Templeton et al. 1986). Although rarely documented in mammals or other vertebrates, outbreeding depression is most likely when ESU boundaries are transgressed (see Barrowclough and Flesness, chap. 24, this volume), either knowingly or not (e.g., owl monkeys, Aotus trivirgatus: Cicmanec and Campbell 1977). Templeton and Read (1984), Templeton et al. (1986), and Lynch (1991) discuss methods for examining the effects of outbreeding in captive populations. Careful consideration of the ESU status of pop- ulations should mitigate potential outbreeding problems. 2. Demographic Parameters Estimated from Life Table Data A number of different demographic parameters are calcu- lated directly from life tables: a. Generation length (T). Generation length is the average age at which a parent produces young (Hedrick 1983). It can be calculated directly from estimates of sur- vival and fecundity rates (see table 26.3; Caughley 1977) and is used to estimate a minimum viable population (MVP) size for the captive population (see below). It should be calculated for each sex separately. b. Population growth rate (A). Life tables provide esti- mates of the expected growth rate of the population, assum- ing the estimated survival and fecundity rates remain stable over time (Caughley 1977). The growth rate is used to esti- mate the MVP size and the capacity of the population for BALLOU AND FOOSE 275 self-sustainment. If it is less than 1, the population is declin- ing; if greater than 1, the population is growing; and if equal to 1, the population size is stationary (constant). If the growth rate is inadequate for the population to be self-sustaining, the focus of the management program should shift to research on reproductive, behavioral, and other biological and husbandry aspects of management to resolve the problems. c. Stable age structure. The stable age structure is the eventual sex and age structure of the population if survival and fecundity rates remain stable over time (Caughley 1977). d. Net reproductive rate ( RQ ) ? The net reproductive rate is the number of same-sex offspring produced by an average individual during its lifetime (see table 26.3). Conceptually, it is the number of same-sex animals an individual "re- places" itself with in the population. For stationary popu- lations (zero population growth), Ro is 1: each animal ex- actly replaces itself in the population every generation. R^ is used to calculate animal lifetime reproductive objectives for population management. e. Annual reproductive rates. The number of animals born each year in a stable population with growth rate A is a function of the survival rate and the population growth rate. The annual number of female (or male) births required to achieve a population growth rate of A for a stable popu- lation of size N is: Number of births N 2A- L (26.6) where the summation in the denominator is over all age classes. The number of births are for the same sex from which the t values are calculated. For zero population growth (A = 1), the number of births is N/S/;(. For example, 41 female golden lion tamarins need to be born each year to achieve zero population growth at a stable population size of 500 animals (250 females: see table 26.3). If the survival rates for males and females are the same, and if the sex ratio at birth is 1:1, the total number of births required is double the number calculated above. f. Fertility, survival, and harvest rates necessary to maintain a stationary population. These parameters are used to develop management recommendations for desig- nation of surplus and breeding rates, as well as to predict what effects managerial modifications of survival and fer- tility rates will have on the population (Beddington and Tay- lor 1973; see below). Several computer programs are available to calculate these demographic parameters from life table data as well as to estimate life table data directly from computerized ped- igrees (Bingaman and Ballou 1993; ISIS 1991). Table 26.3 presents calculations of values for generation length, popu- lation growth rate, and fertility rates for the captive popu- lation of golden lion tamarins. 3. Carrying Capacity Determination For a managed population, the carrying capacity is an ana- lytically established target size to which the program as- pires. The process of establishing a carrying capacity in- volves reconciling the genetic and demographic goals of the population with the limited resources of zoological institutions. Losses of both genetic diversity and demographic sto- chasticity are functions of population size. The smaller the population, the faster the loss of diversity, and the more un- stable and susceptible to extinction it is. Thus, purely for genetic and demographic reasons, the captive population should be as large as possible. However, limited resources place severe restrictions on the sizes of captive populations. Similar species "compete" for captive habitat. Maintaining too large a population of any one taxon may deprive other needy taxa of captive resources; therefore, the carrying ca- pacity for each taxon needs to be a compromise between maintaining some minimum viable population (MVP) size large enough for genetic and demographic goals to be real- ized and still allowing enough resources for other similar species' programs. While the MVP size determines the lower limit of the carrying capacity, the upper limit is determined by resource limitations. a. Determine the lower Umit of carrying capacity: minimum viable population size. The lower limit of the carrying capacity, the MVP, will be determined primarily by the long-term genetic and demographic objectives of the program and the biological characteristics of the pop- ulation. As discussed above, the primary goal of most conservation-related captive propagation programs is to maintain genetic diversity and demographic security. More specifically, MVP size depends on: ? The kind and amount of genetic diversity to be preserved ? The length of time the population is to be managed ? The probability that the population will survive this time period ? The biological characteristics of the population (i.e., num- ber of founders, generation length, effective size, and popu- lation growth rate) The 90%/200 year rule of thumb (see "Captive Manage- ment Goals" above) is a common approach to determining how much diversity is required and for how long it should be maintained (Soul? et al. 1986). Calculating the MVP size necessary to maintain 90% of the original heterozygosity for 200 years requires modeling the population's growth from its founding number to its carrying capacity through the 200-year period. Loss of genetic diversity can be con- veniently visualized as occurring during three phases: the founding event, the growth to carrying capacity, and the management of a stable population at carrying capacity (fig. 26.8; Ballou 1987c). During each phase genetic diver- sity is lost at a different rate, and the overall loss of hetero- zygosity is the cumulative loss over all three phases. Loss of heterozygosity during the founding phase is a function of the effective size of the founding population. Equation (26.1) can be used to roughly estimate the effective size of the founding population from its sex ratio. Loss of hetero- zygosity during the growth phase is a function of the popu- lation's growth rate and how long (measured in animal gen- eration lengths) it remains in the growth phase. Loss of heterozygosity during the carrying capacity phase is deter- 276 DEMOGRAPHIC AND GENETIC MANAGEMENT CO z o p < -I Z) a. O ?. CAPACITY PHASE FIG. 26.8. Theoretical growth curve of a captive population showring the founding, growrth, and capacity phases used to model loss of genetic diversity. 180 200 YEAR mined by the effective size of the population at carrying ca- pacity and how long it stays at capacity, again measured in generations. If the founder effective size, generation length, N^/N ra- tio, and population growth rate are known, the carrying ca- pacity required to maintain X% (e.g., 90%) of the variation for Y (e.g., 200) years can be calculated (Ballou 1987c). Graphs, algorithms, and computer programs have been de- veloped for this purpose (Gilpin 1987; Ballou 1986; Soule et al. 1986). Table 26.4 illustrates the MVP sizes required for a range of N^/N ratios and effective founder sizes for a population growing at 8% per year. A secondary consideration for determining the MVP size is demographic stochasticity and the susceptibility of the population to extinction due to chance or catastrophic events. Populations smaller than 50 or even 100 may be par- ticularly vulnerable to "crashes" or extinctions due to ran- dom demographic events such as disease epidemics, natural disasters, or sex ratio distortions (Gilpin and Soule 1986). Therefore, for demographic reasons, MVPs should be no smaller than 50 individuals (Foose et al. 1986). The MVP TABLE 26.4. Carrying Capacities Necessary for Maintaining 90% of the Founders' Heterozygosity for 200 Years in a Population writh Generation Length of 5 Years and a A of 1.08 (8% Growrth Rate per Year) NJN Effective founder number ratio 10 15 20 25 30 0.1 3620 2452 2084 1917 1822 0.3 1207 817 695 639 607 0.5 724 490 417 383 364 0.7 517 350 298 274 260 0.9 402 272 232 213 202 1.1 329 223 189 174 166 Source: After Soule et al 1986. based on genetic considerations will probably be large enough to insure a high probability of survival for 200 years for most captive populations, since distribution of animals among many zoos decreases the likelihood of total popula- tion extinction due to environmental stochasticity or ca- tastrophe (Shaffer 1987). Demographic simulations can be conducted to estimate how susceptible populations with given demographic characteristics are to such chance events (Ewens et al. 1987; Goodman 1987; Lacy 1990b; Shaffer 1987). Demographic MVP models are discussed in detail in Soule (1987). Finally, it should be emphasized that there is no single MVP size that applies to all species or populations, nor is there a single magic-number MVP size that categorically ap- plies to any one species all the time. Determination of MVPs depends on a number of factors. The simplistic determina- tion of MVP described here is an example of the more gen- eral process known as population viability analysis (PVA: Soule 1987; Gilpin and Soule 1986). b. Determine the upper limit of carrying capacity: captive resources. The upper limit on carrying capacity should be derived from an analysis of the amount of "cap- tive habitat" (space and resources) currently being used by the target population and other taxa with similar "captive ecologies" (enclosure and resource requirements, exhibit value, etc.). Current population sizes and information on expansion plans can be used to estimate the captive habitat available. For example, Foose and Seal (1986) calculated the total number of enclosure spaces utilized by all large fe- lids to determine the captive carrying capacity for Siberian tigers. In addition, the number of different taxa in need of assis- tance from captive propagation programs and potentially competing for this captive space should be determined. This may require additional information on the status and trends of wild populations as well as consideration of taxonomic BALLOU AND FOOSE 277 uniqueness. Division of the available "captive" habitat by the number of "competing" taxa will suggest the upper limit on carrying capacity for each taxon. These estimates can be refined if additional information is known about the MVP size requirements of the other "competing" taxa (Conway 1987; Foose, Seal, and Flesness 1987). c. Establish the carrying capacity. With the lower limit of carrying capacity determined by MVP size requirements and the upper limit determined by captive resource alloca- tion, a carrying capacity for the captive population can be established. The carrying capacity should be as large as pos- sible within these limits. If the MVP size is larger than the population size allowed by resource allocation, then either the MVP requirements will have to be relaxed (reduce the time frame of concern and/or decrease the levels of diversity to be retained) or the biological characteristics of the popu- lation will have to be improved (acquire new founders, ex- tend generation length, increase NJN ratio, and/or increase growth rate). Otherwise, it will be necessary to prioritize and select among competing taxa. The development of cri- teria for prioritizing taxa in need of conservation measures is complex and continues to be discussed by a number of conservation organizations (Foose, Seal, and Flesness 1987; Oates 1986; Foose 1983). POPULATION MANAGEMENT Genetic and demographic analyses provide the basis for formulating a captive propagation and management pro- gram, such as an AZA Species Survival Plan Masterplan (Foose and Seal 1986). Ideally, the program should provide specific recommendations for each individual in the popu- lation. More specific guidelines for population management include: 1. Attempt to obtain a sufficient number of founders to sample adequately both the heterozygosity and the allelic diversity in the source population. AUelic diversity is lost much more rapidly than heterozy- gosity during bottleneck and founding events (Allendorf 1986; Fuerst and Maruyama 1986). Therefore, the primary concern is adequate sampling for allelic diversity, since this may require more founders than sampling for heterozygos- ity alone. Sampling for heterozygosity does, however, estab- lish a lower limit for the effective founder size required. A sample of N effective founders retains on average {1/2N) * 100% of the source population's heterozygosity. A general rule of thumb is to try to sample at least 95% of the source population's heterozygosity; this requires an effective foun- der size of at least ten (Denniston 1978). The number of founders required to sample allelic diver- sity adequately depends on the al?ele frequencies in the source population. Marshall and Brown (1975), Denniston (1978), and Gregorius (1980) discuss the effective founder sizes required given various al?ele frequency distributions (table 26.5). Unfortunately, information on the distribution of al?ele frequencies in the source population is often not available. Marshall and Brown (1975) suggest that founder numbers adequate for effectively sampling allelic diversity be based on the most likely al?ele distributions, and con- TABLE 26.5. Founder Sizes Necessary to Sample All Al?eles with Frequencies Equal to or Greater Than p with 95% and 99% Certainty AUele frequency iP)- Certainty 95% 99% .500 .300 .200 .100 .040 .010 .008 6 11 22 51 152 754 972 15 28 66 192 916 1,174 Source: After Gregorius 1980. elude that effective founder sizes between twenty-five and fifty are sufficient in most cases. They emphasize that poten- tial differences in genetic variation over the range of a popu- lation should be considered. Sampling strategies should at- tempt to compensate for and/or exploit known geographic patterns of genetic variation to optimize the levels of genetic diversity sampled, while at the same time striving to remain within the geographic boundary of the ESU. Additional perspectives on the number of founders re- quired will also derive from the MVP analysis described above. Specifying genetic and demographic objectives and other population characteristics will prescribe a minimum number of founders. The MVP analysis approach will also consider demographic as well as genetic factors in establish- ing minimum founder numbers. It should be appreciated that founders will not necessar- ily or optimally enter the population only at the inception of a captive propagation project. Immigrants from the wild should periodically be incorporated into the captive popu- lation if possible. It should also be noted that failure to ob- tain an optimal genetic number of founders is not justifica- tion for cancelling plans to establish a captive propagation program. Wild-caught specimens, however, should be ob- tained only after extremely careful consideration of the po- tential effects of such removals on the wild population. 2. Expand the population size as rapidly as possible to the carrying capacity. Genetic diversity is lost when growth rates are slow because small populations lose genetic diversity at a faster rate than large populations. 3. Stabilize the population at carrying capacity. The current population size and growth rate determine whether the population is at, or when it will be at, carrying capacity. If the population is at or is approaching carrying capacity, demographic analysis can be used to determine how fertility and survivorship rates can be managed by "removals" of animals (harvests, culls) and/or regulation of reproduction (birth control) to stabilize the population at the desired carrying capacity (Beddington and Taylor 1973). This process may entail much "what if" analysis to determine how such managerial modifications of survivor- ship and fertility patterns will affect population size, growth rate, age distribution, and so forth. Table 26.6 shows how such an analysis can be applied to Siberian tigers. The ef- 278 DEMOGRAPHIC AND GENETIC MANAGEMENT TABLE 26.6. Management Options for Stabilizing the Siberian Tiger Population at 250 Animals Option If reproduction is Percentage of 0-1-year-olds to be removed Percentage of each age class to be removed Equivalent to 1956-1980 and litter size of 2.43 cubs Adjusted to compensate for mortality and equally distributed over all ages One litter of 2.43 cubs at any age One litter of 3 cubs at age 4 Two Utters of 2.43 cubs at age 5 Two Utters of 2.43 cubs at ages 5 and 10 Two litters of 2.43 cubs at ages 4 and 7 TUree litters of 2.43 cubs at ages 4, 9, and 12 Litters of 2.43 cubs in alternate years 46 0 TUis level of reproduction appears insufficient to sustain the population with present mortality rates 0 (probably) 0 (probably) This level of reproduction appears insufficient to sustain the population with present mortality rates 30 5 35 8 53 10 72 15 Source: Foose and Seal 1986. fects of these modifications on the effective population size should also be examined, since a reduction in the number of breeders can reduce N^. Modifications that maintain a large effective size while still accomplishing the goal of stabilizing the population at carrying capacity should be explored (Dyke et al. 1986; Ryman et al. 1981). 4. Extend generation length as much as possible without jeopardizing demographic security. Because genetic diversity is lost each generation, extending generation length (T) will reduce the amount of diversity lost during a given number of years. Alternatively, the same level of diversity can be retained with a smaller N^ if T is extended. Generation time can be lengthened by shifting the mean age of reproduction to later in life. This strategy, how- ever, incurs greater risk of stochastic losses of animals be- fore they can breed and may also result in reduction of fer- tility due to age-dependent factors. 5. Adjust representation of founder lineages to be proportional to the probable distribution of founder al?eles surviving in the living population. The most substantial part of any captive propagation pro- gram is the process of identifying which animals are to breed, how often, and with whom. The basic objective of this process is to compensate for highly skewed founder contribution distributions and loss of founder al?eles by preferentially breeding descendants of underrepresented founders and restricting reproduction in those of overrepre- sented founders. However, this must be done within the framework of the demographic requirements of the popu- lation. The process involves four steps: (1) determining the target founder contributions that define the objectives for adjusting the population founder contributions; (2) deter- mining individual lifetime reproductive objectives accord- ing to the genetic characteristics of each individual (i.e., how often animals are to breed); (3) determining individual annual reproductive objectives so that yearly population demographic needs are met (i.e., which animals are to breed each year); and (4) recommending pairing to accomplish all of the above (i.e., who is to breed with whom). a. Determine the target founder contribution objec- tives for the population. Equalization of founder represen- tation is usually not the optimal objective. Founder repre- sentation should not be equalized if some proportion of a founder's al?eles have been lost as a result of bottlenecks in the pedigree (see "Genetic Analyses" above). With fewer genes to contribute, adjusting this founder's contribution to the same level as that of other founders not having experi- enced a bottleneck will overrepresent this founder's remain- ing genes in the population. For such founders, the founder contribution goals must be reduced according to their level of retention. For example, if a founder's retention is 50%, it could be considered only "one-half" of a founder and its contribution to the living population managed to a level of one-half that of the other founders. Results of the gene drop and founder contribution analy- ses (see "Genetic Analyses" above) are used to compute a target distribution of founder representation that more ac- curately reflects how much of the founders' genetic diversity has survived to the living population (retention): TF, (26.7) E where TF? is the target founder contribution for founder i; r? is the proportion of founder ?'s al?eles surviving to the liv- ing population (retention); and Np is the number of foun- ders in the living population. Genetic representation of founders with low retention should be managed at a lower level than that of well-represented founders. This strategy will increase the number of unique founder al?eles main- tained in the population. Table 26.7 shows the target foun- der representations for the pedigree in figure 26.2. Note that founders 2 and 3 are adequately represented according to the target founder contribution goal, whereas before they appeared to be underrepresented. The objective of genetic management then becomes attempting to shift the observed founder contributions toward the target founder contributions. b. Combine genetic and demographic objectives to ar- BALLOU AND FOOSE 279 TABLE 26.7. Target Founder Contributions for Genetic Management of the Pedigree Shown in Figure 26.2 Current Parity" % genes Target^ founder founder surviving '' founder Status of Founder contribution contribution (retention) contribution contribution ?' 1 .13 .20 .50 .17 Under 2 .16 .20 .48 .17 Adequate 3 .16 .20 .49 .17 Adequate 4 .35 .20 .80 .28 Over 5 .20 .20 .61 .21 Adequate Note: Target founder contributions are based on the proportion of a founder's genes surviving to the hving population (retention). ''Parity (equal representation) = l/(number of founders). ''Proportion of each founder's genome surviving to the living population is based on a gene drop analysis with 5,000 simulations. 'See equation 26.7. ?'Under, founder's genetic contribution is below that of its target?it is genetically underrepresented in the population; over, founder's genetic contribution is above that of its target?it is genetically overrepre- sented in the population; adequate, founder's genetic contribution is approximately that of its target. rive at animal-by-animal breeding recommendations. As discussed above, demographic analyses can be used to de- termine the average hfetime reproductive requirements for individuals in the population (Ro)- However, in order to modify current founder contributions to match target ob- jectives, it is desirable for individuals descended from un- derrepresented founders to produce more than the average number of progeny, while individuals descended from over- represented founders should produce fewer. These individuals can be identified by examining the founder contribution within each individual. Animals with a high founder contribution from underrepresented foun- ders (as defined by the target founder contribution) should be considered high-priority breeders (e.g., individual 8 in the sample pedigree, table 26.1). However, identifying pre- ferred breeders by scanning founder contribution charts is often difficult. An individual may be descended from both over- and underrepresented founders, and contribution from overrepresented founders may be highly correlated with contribution from underrepresented founders. Addi- tionally, if there are a large number of founders, the quantity of information that must be considered is formidable. Diagnostic methods like mean kinship and genetic uniqueness simplify the process of identifying priority indi- viduals (see "Genetic Analyses" above). Ranking of animals according to mean kinship provides a fast and simple diag- nostic tool for identifying animals that are genetically valu- able to the population. Furthermore, these ranking methods can be combined with the demographic requirements of the population to develop specific animal-by-animal breeding recommendations for each individual in the population. One strategy is to use the distribution of mean kinship to determine lifetime reproductive objectives for each indi- vidual. Different lifetime reproductive objectives are as- signed to different levels of mean kinship: the lower the in- dividual's mk value, the higher its lifetime reproductive objective. However, the scale must be established such that average lifetime reproductive objectives across all individ- uals achieve the overall demographic objectives. Scaling reproductive contribution according to mk should not change the mean reproductive rate, only the variance across individuals. The scaling of lifetime reproductive objectives with mk is dependent on how rapidly the current founder contribu- tions are to be adjusted to match the target founder contri- butions. Large differences in lifetime reproductive objec- tives between descendants of over- and underrepresented founders will result in rapid convergence between current and target founder contributions. The range of lifetime re- productive objectives will most likely be determined by the reproductive biology of the species and how intensely the population can be managed. An example of scaling lifetime reproductive objectives according to mean kinship is shown in table 26.8. If and when the population obtains its target distribu- tion of founder representation, lifetime founder objectives (family sizes) should be equalized to maximize the effec- tive size. c. Schedule lifetime reproductive objectives to meet annual population growth objectives. Animal-by-animal breeding recommendations combine the population's yearly demographic requirements with the lifetime reproductive TABLE 26.8. Lifetime Reproductive Recommendations for Individual Captive Golden Lion Tamarins Based on Distribution of Mean Kinship (mk) Values in the Population Mean kinship range /o of population Offspring objectives < 0.019 12% 4 0.020 to 0.034 17% 3 0.035 to 0.050 45% 2 0.051 to 0.058 14% 1 > 0.058 12% 0 Mean = 2.0 Note: The total number of offspring that should be produced and eventually bred per individual is a function of the individual's mk value. Individuals with low mk values carry genes that are uncommon in the population and should be bred more than individuals with com- mon genes. Note that the mean number of births across all individuals is 2.0: each breeding pair only replaces itself and zero population growth is achieved. 280 DEMOGRAPHIC AND GENETIC MANAGEMENT goals for each individual. Demographic analyses determine the yearly number of births required to meet population growth objectives (e.g., rapid population growth or zero population growth). Given the species' litter size and breed- ing rates, this number can be translated into the number of pairs that need to be bred per year to accomplish the desired birth rate. For example, in golden lion tamarins, 82 births (41 female births) are needed each year for zero population growth (see table 26.3). The average litter size is 2.0, 25% of the breeding pairs fail to breed, 65% produce 1 litter per year, and 35% produce 2 litters per year. Sex ratios at birth are equal. Therefore, we need to maintain 40 breeding pairs per year to produce the 82 required births. The animal-by-animal breeding recommendations must specify which individuals should be bred each year to meet these annual breeding requirements. The individuals se- lected for breeding are chosen from those who have not yet fulfilled their lifetime reproductive goals. Scheduling of which animals are to reproduce each year should take into consideration age and genetic importance so that aging, important animals are given reproductive priority. Schedul- ing should also take advantage of existing pairings or groups and institutions' capacities and interests. A schedule of which individuals should reproduce over the next 1-5 years is recommended. d. Select pairings of animals among those scheduled for breeding. At this point, the specific individuals to be bred over the next several years will have been identified. The next, and final, step is to recommend pairings among those individuals. There are two principal criteria for deter- mining who is to breed with whom. The first is to try to avoid pairing an animal from underrepresented founders with an animal from overrepresented founders. The off- spring of such a pair would have founder contributions that are the average of the parents', and the underrepresented founder contribution would be linked to the overrepre- sented founder contribution; they could then no longer be managed independently. Therefore, when making pairing decisions, pair animals that are descended from underrepre- sented founders (i.e., low mean kinship) with similar ani- mals. This strategy allows the underrepresented founder contributions to be increased in the population indepen- dently of the overrepresented founder contributions. At a later generation, when founder contributions are more evenly distributed, this is less of a concern. It is often the case that pairings of animals with similar, underrepresented founder contributions will result in mates that are closely related to each other (a valuable brother/ sister pair will have similar, if not the same, mean kinship). Therefore, it is also important to examine relationships among potential mates and exclude pairings of closely re- lated animals. This objective accomplishes two goals: (1) it reduces the degree of relationship within the population as a whole, therefore retaining higher levels of heterozygosity than would otherwise be retained; and (2) it reduces the potentially detrimental effects of inbreeding depression on survival and reproductive rates. Inbreeding effects are highly variable among species, and it is not possible to pre- dict at what level inbreeding will have significant deleterious effects on the population (Brewer et al. 1990). Inbreeding should be avoided or minimized if inbreeding depression is observed. Unfortunately, inbreeding levels will increase in popula- tions in which immigration is restricted. The minimum amount of inbreeding possible in a population depends on the levels of relatedness among the living animals. A rule of thumb often used in determining acceptable levels of in- breeding in a population uses the mean kinship value aver- aged across all living individuals. The average mk is equal to the expected inbreeding coefficient of the offspring that would be produced if all individuals in the population were bred randomly. To keep inbreeding levels low, pairings should be selected so that offspring have inbreeding coeffi- cients no higher than the average mk. It should be noted that minimizing inbreeding is not the primary criterion for genetic management. However, there should be an attempt to minimize inbreeding within the constraints of adjusting founder contributions toward tar- get objectives and mating individuals to avoid linking over- with underrepresented founder contributions. There may be cases in which inbreeding is deliberately employed to purge a population of its deleterious al?eles, as in the Speke's ga- zelle, Gazella spekei, program (Templeton and Read 1984). Intentional inbreeding of small populations is generally con- trary to the goal of maintaining genetic diversity since it can drastically change the genetic characteristics of the popula- tion, possibly making it unsuitable for later release into the wild (Templeton and Read 1983). 6. Select against individuals with extreme outlying morphological and reproductive characteristics. Such characteristics would include traits such as albinism and dwarfism. This stabilizing selection should help control levels of genetic load in the population (Frankham et al. 1986). Selection can and should be imposed within families by replacing individuals to be selected against with their siblings. This strategy will allow selection to be imple- mented while working within the constraints of equalizing family size. 7. Consider dividing the population into several subdivisions or demes among which gene flow (usually exchange of animals but also exchange of gametes or embryos) is regulated. Subdivision of a population is advantageous for epidemio- logical protection (Dobson and May 1986) as well as for other practical reasons, such as reduction of shipping costs and hazards and simpUfication of management logistics. In addition, genetic advantages may accrue based on the theo- retical argument that, without selection, random genetic drift will drive different al?eles to fixation in different demes and, overall, maintain a higher level of allelic diversity (Chesser, Smith, and Brisbin 1980). However, the role of selection in captive populations is uncertain, and it is pos- sible that similar types of selection, conscious or uncon- scious, will actually fix similar al?eles in each deme, thereby decreasing the overall levels of genetic diversity. Further- more, the smaller size of semi-isolated subdivisions may ren- der them more vulnerable to demographic stochasticity. Subdivided populations with large numbers of animals in BALLOU AND FOOSE 281 each division will benefit from the practical advantages without the consequences of some of the genetic and demo- graphic uncertainties. It is not possible in this chapter to de- scribe methods to determine the optimal numbers or sizes of subdivisions or the extent and rate of genetic exchange among them. More analyses are needed on the role of population subdivision in maintaining genetic diversity. The reader is referred to Lacy (1987) for discussion. 8. If possible, continually introduce new wild-caught founders into the population. Additional genetic material will help to minimize loss of variation due to genetic drift. In some cases, it may be ap- propriate to devise a program of continual exchange of in- dividuals (or genetic material) between wild and captive populations, taking care to minimize the associated epide- miological problems. 9. Utilize available reproductive technology to the fullest extent possible. Reproductive technology (semen/ovum collection and stor- age, embryo transfer and freezing, etc.) should be considered a primary tool for assisting captive propagation programs in the long-term maintenance of genetic diversity. Such tech- nology can facilitate exchange of germ plasm between wild and captive populations as well as effectively increasing the reproductive lifetimes of founders and their immediate de- scendants. By increasing generation length, adequate levels of genetic diversity can be maintained in smaller popula- tions, leaving more room for populations of other needy species (Ballou and Cooper 1992). Living founders who have not yet contributed to the population should be considered immediate candidates for germ plasm storage. Although reproductive technology is not yet available for most exotic species, it is a major focus of research by the reproductive community (Wildt 1989). CONCLUSIONS Zoological institutions are making a major contribution to the conservation of threatened and endangered species through captive propagation programs. Their commitment to the long-term preservation and management of captive populations is dependent upon developing cooperative cap- tive propagation programs based on sound genetic and demographic principles. The essence of any captive propa- gation program is the recommendations indicating which animals are to breed, how often, and with whom. These individual-by-individual and institution-by-institution rec- ommendations are based on the results of genetic and demo- graphic analyses. The final result is a conservation strategy that pools the genetic and demographic potential of all the individuals in the population to satisfy the genetic and demographic needs of the population. Whenever possible, conservation strategies for wild and captive populations should be integrated to provide a comprehensive conserva- tion plan for the species. The technology of genetic and demographic manage- ment is growing rapidly, and experts in many areas are di- recting attention to the problems unique to this field. Ana- lytic and biochemical techniques are becoming increasingly available for population managers to use in the numerous analyses required. In addition, our understanding of the genetic and demographic needs of captive populations con- tinues to expand as more is learned about natural popula- tions. Advances will continue to be incorporated into the procedures for developing captive propagation programs as they become available. ACKNOWLEDGMENTS We thank Katherine Rails, Laurie Bingaman, Susan Haig, and Kathy Cooper as well as an anonymous reviewer for making many helpful comments on this chapter. REFERENCES Allendorf, F. W. 1986. Genetic drift and the loss of al?eles versus heterozygosity. Zoo Biol. 5:181-90. Allendorf, F. W., and Leary, R. F. 1986. Heterozygosity and fitness in natural populations of animals. In Conservation biology: The science of scarcity and diversity, ed. M. E. Soul?, 57-76. Sun- derland, Mass.: Sinauer Associates. Arnold, S. 1995. Monitoring quantitative genetic variation and evolution in captive populations. In Population Management for Survival and Recovery, ed. J. D. Ballou, M. Gilpin, and T. J. Foose, 295-317. New York: Columbia University Press. Ashworth, D., and Parkin, D. T. 1992. Captive breeding: Can ge- netic fingerprinting help? Symp. Zool. Soc. Lond. 64:135-49. Avise, J. C, S. M. Haig, O. A. Ryder, M. Lynch, and C. G. Geyer. 1995. Descriptive genetic studies: Applications in population management and conservation biology. In Population Manage- ment for Survival and Recovery, ed. J. D. Ballou, M. Gilpin, and T. J. Foose, 183-244. New York: Columbia University Press. Bahner, B. 1991. Micronesian kingfisher SSP Program. In AAZPA Annual Report on Conservation and Science 1990-91, eds. M. Hutchins, R. J. Wiese, K. Willis, and S. Becker, 141-43. Wheeling, W.Va.: American Association of Zoological Parks and Aquariums. Ballou, J. D. 1983. Calculating inbreeding coefficients from pedi- grees. In Genetics and conservation, ed. C. M. Schonewald-Cox, S. M. Chambers, B. MacBryde, and L. Thomas, 509-20. Menlo Park, Calif.: Benjamin/Cummings. . 1986. CAPACITY Lotus 1-2-3 software to calculate car- rying capacities for captive populations. Washington, D.C.: Na- tional Zoological Park. -. 1987a. The concept of effective population size and its role in the genetic management of captive populations. AAZPA Regional Conference Proceedings, 43-49. Wheeling, W.Va.: American Association of Zoological Parks and Aquariums. -. 1987b. 1986 international golden lion tamarin studbook. Washington, D.C.: National Zoological Park. -. 1987c. Small populations, genetic diversity, and captive carrying capacities. AAZPA Annual Conference Proceedings, 33-47. Wheeling, W.Va.: American Association of Zoological Parks and Aquariums. Ballou, J. D., and Cooper, K. A. 1992. Application of biotechnol- ogy to captive breeding of endangered species. Symp. Zool. Soc. Lond. 64-.183-296. Ballou, J. D., and Lacy, R. C. 1995. Identifying genetically impor- tant individuals for management of genetic diversity in captive populations. In Population management for survival and recovery, ed. J. D. Ballou, M. Gilpin, and T. J. Foose, 76-111. New York: Columbia University Press. 282 DEMOGRAPHIC AND GENETIC MANAGEMENT Ballou, J. D., and Oakleaf, R. 1989. Demographic and genetic captive-breeding recommendations for black-footed ferrets. In Conservation biology and the black-footed ferret, ed. U. S. Seal, E. T. Thorne, M. A. Bogan, and S. H. Anderson, 247-67. New Haven: Yale University Press. Ballou, J. D., and Seidensticker, J. 1987. The genetic and demo- graphic characteristics of the 1983 captive population of Suma- tran tigers (Panthera tigris sumatrae). In World conservation strategies for tigers, ed. U. S. Seal and R. Tilson, 329-47. Park Ridge, N.J.: Noyes. Beddington, J. R., and Taylor, D. B. 1973. Optimum age specific harvesting of a population. Biometrics 29:801-9. Bingaman, L. E., and Ballou, J. D. 1993. DEMOG 4.1: Lotus 1-2-3 spreadsheet demographic model. Washington, D.C.: Na- tional Zoological Park. Boyce, A. J. 1983. Computation of inbreeding and kinship coeffi- cients on extended pedigrees. /. Hered. 74:400-404. Brewer, B. A., Lacy, R. C, Eoster, M. L., and Alaks, G. 1990. In- breeding depression in insular and central populations of Pero- myscus mice. /. Hered. 81:257-66. Cannings, C, Thompson, E. S., and Skolnick, M. H. 1978. Proba- bility functions on complex pedigrees. Adv. Appl. Prob. 10: 26-61. Caughley, G. 1977. Analysis of vertebrate populations. New York: John Wiley and Sons. Chesser, R. K., Smith, M. H., and Brisbin, I. L. 1980. Management and maintenance of genetic variability in endangered species. Int. Zoo Yrbk. 20:146-54. Cicmanec, J. C, and Campbell, A. K. 1977. Breeding the owl mon- key (Aotus trivirgatusj in a laboratory environment. Lab. Anim. Sei. 17-.512-17. Conway, W. 1987. Species carrying capacity in the zoo alone. In AAZPA Annual Conference Proceedings, 20-32. Wheel- ing, W.Va.: American Association of Zoological Parks and Aquariums. Crow, J. F., and Kimura, M. 1970. An introduction to population genetic theory. New York: Harper and Row. Denniston, C. 1978. Small population size and genetic diversity: Implications for endangered species. In Endangered birds: Man- agement techniques for preserving threatened species, ed. S. A. Temple, 281-89. Madison: University of Wisconsin Press. Derrickson, S. R. 1987. Current status and captive propagation of the endangered Guam rail. In Proceedings of the Jean Delacour/ IFCB Symposium on Breeding Birds in Captivity, ed. A. C. Ris- ser, 187-95. North Hollywood, Calif.: IFCB. Dobson, A. P., and May, R. M. 1986. Disease and conservation. In Conservation biology: The science of scarcity and diversity, ed. M. E. Soul?, 345-65. Sunderland, Mass.: Sinauer Associates. Dyke, B., Gage, T. B., Mamelka, R M., Goy, R. W., and Stone, W. H. 1986. A demographic analysis of the Wisconsin Regional Primate Center rhesus colony, 1962-1982. Am./. Primatol. 10: 257-69. Ewens, W. J., Brockwell, R J., Gani, J, M., and Resnick, S. I. 1987. Minimum viable population size in the presence of catastrophes. In Viable populations for conservation, ed. M. E. Soul?, 59-68. Cambridge: Cambridge University Press. Flesness, N. R. 1986. Captive status and genetic considerations. In Primates: The road to self-sustaining populations, ed. K. Be- nirschke, 845-56. New York: Springer-Verlag. Foose, T. J. 1983. The relevance of captive propagation to the- conservation of biotic diversity. In Genetics and conserva- tion, ed. C. M. Schonewald-Cox, S. M. Chambers, B. Mac- Bryde, and L. Thomas, 374-401. Menlo Park, Calif.: Benjamin/ Cummings. Foose, T. J., and Ballou, J. D. 1988. Population management: Theory and practice. Int. Zoo Yrbk. 27:26-41. Foose, T. J., de Boer, L., Seal, U. S., and Lande, R. 1995. Conser- vation management strategies based on viable populations. In Population management for survival and recovery, ed. J. D. Bal- lou, M. Gilpin, and T. J. Foose, 273-94. New York: Columbia University Press. Foose, T. J., and Foose, E. 1983. Demographic and genetic status and management. In The biology and management of an extinct species: Pere David's deer, ed. B. B. Beck and C. Wemmer, 133- 86. Park Ridge, N.J.: Noyes. Foose, T. J., Lande, R., Flesness, N. R., Rabb, G., and Read, B. 1986. Propagation plans. Zoo Biol. 5:139-46. Foose, T. J., and Seal, U. S. 1986. Species Survival Plans for large cats in North American zoos. In Cats of the world, ed. S. D. Miller and D. D. Everett, 173-98. Washington, D.C.: National Wildlife Federation. Foose, T. J., Seal, U. S., and Flesness, N. R. 1987. Captive propa- gation as a component of conservation strategies for endangered primates. In Primate conservation in the tropical rainforest, ed. C. W. Marsh and R. A. Mittermeier, 263-99. New York: Alan R. Liss. Frankham, R., Hemmer, H., Ryder, O. A., Cothran, E. G., Soul?, M. F., Murray, N. D., and Snyder, M. 1986. Selection in captive populations. Zoo Biol. 5:127-38. Fuerst, P. A., and Maruyama, T. 1986. Considerations on the con- servation of al?eles and of genie heterozygosity in small managed populations. Zoo Biol. 5:171-80. Geyer, C. J., Ryder, O. A., Chemnick L. G., and Thompson, E. A. 1993. Analysis of relatedness in the California condors from DNA fingerprints. Mol. Biol. Evol. 10:571-89. Gilpin, M. E. 1987. HETSIM: Basic software for simulating loss of heterozygosity in captive populations. San Diego: University of California, San Diego. Gilpin, M. E., and Soul?, M. E. 1986. Minimum viable popula- tions: Processes of species extinction. In Conservation biology: The science of scarcity and diversity, ed. M. E. Soul?, 19-34. Sunderland, Mass.: Sinauer Associates. Gilpin, M. E., and Wills, C. 1991. MHC and captive breeding: A rebuttal. Conserv. Biol. 5:554-55. Glatston, A. R. 1986. Studbooks: The basis of breeding programs. lnt.ZooYrbk.l5:\61-67. Goodman, D. 1987. The demography of chance extinction. In Vi- able populations for conservation, ed. M. E. Soul?, 11-34. Cambridge: Cambridge University Press. Gregorius, H. 1980. The probability of losing an al?ele when dip- loid genotypes are sampled. Biometrics 36:643-52. Haig, S. M., Ballou, J. D., and Derrickson, S. R. 1990. Manage- ment options for preserving genetic diversity: Reintroduction of Guam rails to the wild. Conserv. Biol. 4:290-300. Harris, R. B., and Allendorf, F. W. 1989. Genetically effective population size of large mammals: An assessment of estimators. Conserv. Biol. ?>:U\-9\. Hedrick, P. W. 1983. Genetics of populations. Boston: Science Books International. Hedrick, R W., Brussard, V. F., Allendorf, F. W., Beardmore, J. A., and Orzack, S. 1986. Protein variation, fitness and captive propagation. Zoo Biol. 5:91-99. Hemmer, H. 1986. Nutztierdamhirsch. In Domestikationundver- haltensgerechte Haltung des Damwildes als landwirtschaftliche Nutzart, 71-91. Neumuhle: Herausgeben. Hill, W. G. 1972. Effective size of populations with overlapping generations. Theor. Popul. Biol. 3:278-89. Hughes, A. L. 1991. MHC polymorphism and the design of captive breeding programs. Conserv. Biol. 5:249-51. Hutchins, M., and Wiese, R. J. 1991. Beyond genetic and demo- graphic management: The future of the Species Survival Plan and related AAZPA conservation efforts. Zoo Biol. 10:285-92. BALLOU AND FOOSE 283 International Species Information System (ISIS). 1991. SPARKS (Single Species Animal Record Keeping System). Apple Valley, Minn.: ISIS. Kimura, M., and Crow, J. F. 1963. The measurement of effective population number. Evolution 17:279-88. Kleiman, D. G., Beck, B. B., Dietz, J. M., Dietz, L. A., Ballou, J. D., and Coimbra-Filho. A. F. 1986. Conservation program for the golden lion tamarin: Captive research and management, eco- logical studies, education strategies, and reintroduction. In Primates: The road to self-sustaining populations, ed. K. Be- nirschke, 959-79. New York: Springer-Verlag. Lacy, R. C. 1987. Loss of genetic diversity from managed popula- tions: Interaction effects of drift, mutation, immigration, selec- tion, and population subdivision. Conserv. Biol. 1:143-58. . 1989. Analysis of founder representation in pedigrees: Founder equivalents and founder genome equivalence. Zoo Biol. 8:111-24. 1990a. GENES: Pedigree analysis software. Brookfield, 111.: Chicago Zoological Park. 1990b. VORTEX: Population viability analysis software. Brookfield, 111.: Chicago Zoological Park. Lacy, R., Ballou, J. D., Starfield, A., and Thompson, E. 1995. Pedi- gree analysis for population management. In Population man- agement for survival and recovery, ed. J. D. Ballou, M. Gilpin, and T. J. Foose, 57-75. New York: Columbia University Press. Lacy, R. C, Petric, A., and Warneke, M. 1993. Inbreeding and out- breeding in captive populations of wild animal species. In The natural history of inbreeding and outbreeding, ed. N. W. Thorn- hill. 352-74. Chicago: University of Chicago Press. Lande, R., and Barrowclough, G^ 1987. Effective population size, genetic variation, and their use in population management. In Viable populations for conservation, ed. M. E. Soul?, 87-124. Cambridge: Cambridge University Press. Lynch, M. 1991. The genetic interpretation of inbreeding depres- sion and outbreeding depression. Evolution 45:622-29. MacCluer, J. W., VandeBerg, J. L., Read, B., and Ryder, O. A. 1986. Pedigree analysis by computer simulation. Zoo Biol. 5: 147-60. Marshall, D. R., and Brown, A. H. D. 1975. Optimal sampling strategies in genetic conservation. In Crop genetic resources for today and tomorrow, ed. O. H. Frankel and J. G. Hawkes, 53- 80. Cambridge: Cambridge University Press. Miller, P S., and Hedrick, R W. 1991. MHC polymorphism and the design of captive breeding programs: Simple solutions are not the answer. Conserv. Biol. 5:556-58. Mitton, J. B., and Grant, M. C. 1984. Associations among protein heterozygosity, growth rate, and developmental homeostasis. Annu. Rev. Ecol. Syst. 15:479-500. Morin, P. A., and Ryder, O. A. 1991. Founder contribution and pedigree inference in a captive breeding colony of lion-tailed ma- caques, using mitochondrial DNA and DNA fingerprint analy- ses. Zoo B?o/. 10:341-52. Oates, J. F. 1986. SSC/IUCN action plan for African primate con- servation: 1986-1990. Merges, Switzerland: lUCN. O'Brien, S. J., Roelke, M. E., Marker, L., Newman, A., Winkler, C. A., Meltzer, D., Colly, L., Evermann, J. F., Bush, M., and Wildt, D. E. 1985. Genetic basis for species vulnerability in the cheetah. Science 227:142S-34. Olney, P. J. S. 1986. International zoo yearbook. London: Zoo- logical Society of London. Princ?e, F. 1989. Zoo research studbook management. Amster- dam: Dutch Zoo Federation. Rails, K., and Ballou, J. D. 1983. Extinction: Lessons from zoos. In Genetics and conservation, ed. C. M. Schonewald-Cox, S. M. Chambers, B. MacBryde, and L. Thomas, 164-84. Menlo Park, Calif.: Benjamin/Cummings. Rails, K., Ballou, J. D., and Templeton, A. R. 1988. Estimates of lethal equivalents and the cost of inbreeding in mammals. Con- serv. Biol. 2:1^5-93. Ryman, N., Baccus, R., Reuterwall, C, and Smith, M. H. 1981. Effective population size, generation interval, and potential loss of genetic variability in game species under different hunting re- gimes. Oikos 36:257-66. Seal, U. S., Makey, D. G., and Murtfeldt, L. E. 1976. ISIS: An ani- mal census system. Int. Zoo Yrbk. 16:180-84. Shaffer, M. 1987. Minimum viable populations: Coping with un- certainty. In Viable populations for conservation, ed. M. E. Soul?, 69-86. Cambridge: Cambridge University Press. Soul?, M. E. 1987. Viable populations for conservation. Cam- bridge: Cambridge University Press. Soul?, M., Gilpin, M., Conway, W., and Foose, T. J. 1986. The millennium ark: How long a voyage, how many staterooms, how many passengers? Zoo Biol. 5:101-13. Templeton, A. R., Hemmer, H., Mace, G., Seal, U. S., Shields, W. M., and Woodruff, D. S. 1986. Local adaptation, coadapta- tion and population boundaries. Zoo Biol. S : 115-25. Templeton, A. R., and Read, B. 1983. The ehmination of inbreed- ing depression in a captive herd of Speke's gazelle. In Genetics and conservation, ed. C. M. Schonewald-Cox, S. M. Chambers, B. MacBryde, and L. Thomas, 241-61. Menlo Park, Calif.: Ben- jamin/Cummings. . 1984. Factors eliminating inbreeding depression in a cap- tive herd of Speke's gazelle. Zoo Biol. 3:177-99. Vrijenhoek, R. C, and Leberg, P. L. 1991. Let's not throw the baby out with the bathwater: A comment on management for MHC diversity in captive populations. Conserv. Biol. 5:252-54. Wayne, R. K., Forman, L., Newman, A. K., Simonson, J. M., and O'Brien, S. J. 1986. Genetic monitors of zoo populations: Mor- phological and electrophoretic assays. Zoo Biol. 5:215-32. Wildt, D. E. 1989. Reproductive research in conservation biology: Priorities and avenues for support. /. Zoo Wildl. Med. 20: 391-95. Wildt, D. E., Bush, M., Goodrowe, K. L., Packer, C, Pusey, A. E., Brown, J. L., Joslin, P., and O'Brien, S. J. 1987. Reproductive and genetic consequences of founding isolated lion populations. Nature 329:32S-31. Willis, K. 1993. Use of animals with unknown ancestries in scien- tifically managed breeding programs. Zoo Biol. 12: 161-72.