Projected Extinction Year and Lifetime Maintenance Costs of Captive Former Biomedical Research Chimpanzees

Simple Summary

Animal populations kept in captivity require special care. Populations of captive animals may require human care for many generations. Knowing how long a population will survive in captivity allows managers to plan effectively for budgeting necessary resources into the future. This knowledge requires the ability to predict future population trends. The problem addressed here was to project future trends in a population of captive former biomedical research chimpanzees. Demographic information was used to estimate the date of population extinction. Then we used current estimates of annual costs to care for the chimpanzees to project the total maintenance costs until extinction occurs. Results indicated that the chimpanzee population will continue to exist for another 38 years, until going extinct in 2062. Representative estimates of the cost of care, with and without inflation, were used to estimate lifetime maintenance costs. Without inflation, lifetime maintenance expenses were USD 148.0 M. With inflation, lifetime maintenance expenses were USD 229.3 M. Extinction projections and estimated maintenance costs are essential parts of any sound and reliable program for captive population management. These projections of extinction and cost can be used for the long-term planning and management of captive former research chimpanzees.

Abstract

A vital component of small population management in captivity is forecasting future population trends. Until recently, little was known about the demographic structure of the recently retired captive former biomedical research chimpanzee population (Pan troglodytes). We used age- and sex-specific survivorship probabilities from current life tables to model the likely population extinction date and estimate total lifetime maintenance costs until extinction. Leslie matrix projections indicated that population extinction will occur in 2062 or 38 years from now. That date was 26 years longer than a recent government study of 2036, 8 years longer than the original research chimpanzee projection of 2054, and 5 years longer than our pilot projection of 2057. Median, lower, and upper quartiles of colony per diems plus median biomedical inflation were used to estimate lifetime maintenance costs. Without inflation, median lifetime expenses were USD 148.0 M, with a lower bound of USD 72.7 M and an upper bound of USD 209.9 M. With inflation, median lifetime expenses were USD 229.3 M (lower bound USD 112.7 M, upper bound USD 326.1 M). Extinction projections and associated cost estimates are essential components of any science-based small captive population management plan. These projections can be used for the long-term planning and rational management of the now-retired captive former research chimpanzee population.

1. Introduction

Due to their close phylogenetic relation to humans, chimpanzees (Pan troglodytes) were, for many decades, an important animal model for studying human morbidity, psychology, and biology [1]. In 1986, the U.S. National Institutes of Health (NIH) initiated a Chimpanzee Breeding and Research Program (CBRP) to establish a self-sustaining population of chimpanzees for biomedical research, primarily involving HIV/AIDS [2]. The breeding program became highly successful [3], with population growth peaking in 1997 [4]. In the early 1990s, chimpanzees were the favored animal model for AIDS because they were the only research primate that could be infected with HIV-1 [5]. But the need for chimpanzees in HIV/AIDS research declined over the next decade, as they were replaced by immunogenetically defined macaques that could be infected with the chimeric SHIV virus and actually contract AIDS [6,7]. Model life tables were estimated from a small sample collected from a young population of research chimpanzees [8]. Age- and sex-specific survivorships were used in population projections to estimate the extinction of the CBRP population in 2054, with associated total lifetime maintenance costs of USD 299.9 to USD 611.6 M [9]. Those two studies highlighted the unexpectedly long lifetimes of chimpanzees and the enormous costs associated with breeding them for research. In 1995, a temporary breeding moratorium was instituted to control the population size and costs, while a special NIH council was convened to study the situation more closely [10]. In 2007, the breeding moratorium was made permanent by an NIH moratorium on funding for breeding [11]. Recognition of a population of “surplus chimpanzees” in excess of declining biomedical research needs led to the call for a national chimpanzee sanctuary to house the chimpanzees after their research careers were over [10,12]. The Chimpanzee Health Improvement, Maintenance, and Protection (CHIMP) Act of 2000 created a federal sanctuary system to provide lifetime care for the retirement of chimpanzees that had been bred for, used in, purchased, or supported by the NIH for biomedical research [13]. In 2011, a second special NIH council concluded that most then-current use of chimpanzees in biomedical research was unnecessary [14]. In 2015, the U.S. Fish and Wildlife Service changed the split-listing conservation status of chimpanzees, whereby wild chimpanzees were considered “endangered” under the Endangered Species Act but captive chimpanzees were considered “threatened,” a status which permitted biomedical research [15]. Soon after, the NIH announced that they would no longer support biomedical research on chimpanzees, thereby making all NIH-owned or supported chimpanzees eligible for retirement to the federal sanctuary system [16,17,18,19,20]. But not all chimpanzees owned, supported, or bred for research by the NIH were finally deemed eligible for federal support because the definition of “owned” and “supported” was restricted to only those chimpanzees housed at research facilities that were funded by the NIH in fiscal year 2020 [21,22,23,24]. The model for this truncation dated back to 2007, when the Chimpanzee Management Committee Working Group (CMCWG) discussed the high lifetime maintenance costs of chimpanzees (grossly underestimated at USD 500,000 per lifetime per animal, equivalent to a USD 27.40 per diem) and concluded that ownership was an important issue [3,25].

These programmatic developments raised important issues about the demographic and epidemiologic profile of the CFBR chimpanzee population. As the foundational population science, demography forms the basis for the study of any population [26]. Demography is particularly important for the management of small animal populations in captivity [27]. For many endangered species managed in captivity, changes in population structure expected over time are routinely projected from their current population size and composition and accurate mortality probabilities estimated from life tables. For example, there are 530 species with studbooks and lifetables that are used in population viability analyses for long-term demographic and genetic management in captivity [28].

Until recently, little was known about the demography of the contemporary captive former biomedical research (CFBR) chimpanzee population. The original research chimpanzee life tables [8] were based on a small sample N = 1227 of known sex, composed of 577 males (280 dead, 297 alive) and 650 females (258 dead, 392 alive). Those data were collected during a period of intense captive breeding, characterized by rapid population growth and a relatively young population [8]. The maximum ages, alive or dead, were estimated at 44.5 years for males and 57.5 years for females [8]. Neither the mean age nor the number of geriatric or extreme old age individuals were presented in that paper. But, clearly, the 1995 population was very young. The authors of those seminal papers were well aware that the sparse data did not sample the entire chimpanzee lifespan, especially in the older age classes [8]. Consequently, they recommended caution in interpreting the life tables and explicitly called for periodic demographic updates and extinction date projections, including associated maintenance costs [8,9], a plea repeated later by the governmental commission on costs and cost containment [17].

The 1995 research life tables were partially updated in our pilot studies using data from two to five CFBR colonies, excluding the National Chimpanzee Sanctuary [29,30]. But fundamental questions about basic demography, including age- and sex-specific life tables, were not resolved until our recent paper on CFBR chimpanzee demography [4]. Relative to its ancestral research population [8,9,17], the CFBR chimpanzee population had nearly doubled in size (from N = 1227 to N = 2344, for a 91% increase). It had also aged. The average age of the living CFBR population was 27.6 years (27.7 years for females, 27.4 years for males). There were 302 living geriatric animals aged 30–49 years (168 females, 134 males), and another 19 (11 females, 8 males) in extreme old age (50+ years old). Finally, the finding that both the median age and age at death of the retired CFBR population were continuing to rise indicated that the CFBR population had not yet fully realized its aging potential [4].

After decades of programmatic inertia, those basic demographic results represented a milestone in our understanding of the demography and epidemiology of the CFBR chimpanzee population. More importantly, they also formed the foundation for addressing important practical issues, including projecting changes in the age/sex structure of the CFBR population over time up to the year of extinction and estimating how much it will cost to maintain this population until then. These related issues, extinction, and costs, are the topics of this paper.

Background

Several earlier studies have addressed the question of CFBR chimpanzee population extinction and maintenance costs. Pioneering studies conducted 3 decades ago used Leslie matrix projection and model life tables drawn from then-current population structure to project that the research CBRP chimpanzee population would survive another 60 years, until going extinct in 2054 [9].

Twelve years later, a primate colony director opined that the CFBR population would survive another thirty years, thus projecting extinction in 2037 [25]. That was nearly two decades earlier than the demography-based projection of a 2054 extinction [9]. That guesstimate was not based on any demographic model but merely posited an extinction date based on generalizations about the current average age and captive chimpanzee longevity [25]. The informality of that estimate notwithstanding, the 2037 extinction was uncritically repeated by government commissions and science journalists alike [11,17].

Our pilot study of two CFBR chimpanzee colonies estimated an intrinsic rate of increase of λ = −3.6%, with extinction projected by Leslie matrix analysis in 2057 [29]. A follow-up pilot study of four CFBR colonies estimated an intrinsic rate of increase of λ = −6.68% and also projected population extinction using Leslie matrices by 2057 [30]. One short-coming of both pilot studies is that they did not include demographic data from the National Chimpanzee Sanctuary, where many chimpanzees had already been relocated from CFBR colonies and were, therefore, lost to follow-up. The inclusion of sanctuary chimpanzees in our analyses [4] made the CFBR population even older than previously indicated [29,30]. In turn, this implied an even more distant projected extinction date.

Once the extinction year has been projected, it becomes possible to address the closely related question of lifetime maintenance costs [9]. The same three studies on CFBR extinction also addressed this question of lifetime maintenance costs. Assuming USD 25 per diems for maintenance (in 1995), 4% inflation projected for 1995 [9], and the then-hypothetical no-births scenario (implemented shortly thereafter with the 1995 breeding moratorium), they estimated a 60-year cumulative total costs of USD 299.9 million in 1995 dollars, or USD 611.6 million including 4% inflation [9]. Two decades later, the same governmental report that promoted a 2037 extinction conducted an in-depth study on the per diem costs at the various chimpanzee-holding colonies [17]. This GAO report was not directly concerned with population management, demographic projection, or lifetime maintenance costs. Their focus was to assess the reliability of the limited financial data available to them at four chimpanzee-holding colonies, estimate colony per diems, and discuss factors associated with the wide (49%) variation in colony per diems (USD 41/day to USD 61/day) [17]. While they did not directly estimate lifetime maintenance costs, they did conclude that the NIH needed to develop clear long-term implementation plans in order to care for retired CFBR chimpanzees in a cost-effective manner [17]. Finally, our own pilot studies that projected extinction in 2057 used the observed median per diem of USD 52.66 to predict total expected lifetime maintenance costs of USD 337.5 million in 2015 dollars or USD 358.8 million in inflation-adjusted dollars [29,30].

Our demography paper collected records from all current and former CFBR chimpanzee colonies, updated the 29-year-old research chimpanzee life tables, and estimated age- and sex-specific survivorship rates [4]. We extend that foundation in basic demography here by applying population matrix methods to project the current CFBR chimpanzee population composition over time, up to the year of population extinction. Those projections are then used to estimate total care and maintenance costs until expected population extinction. The results of these analyses are expected to provide rational science-based guidelines for effective population management and program planning, offer guidance for the allocation of scarce resources into the future, lay the foundation for periodic updates of CFBR chimpanzee demography, and facilitate studies in the epidemiology of chimpanzee mortality during aging.

2. Methods

2.1. Ethical Considerations

The analyses presented here were based entirely on animal records and did not require any contact with live animals. Therefore, Chimpanzee Research Use (CRU) committee approval was deemed unnecessary by the National Institutes of Health. Similarly, IACUC approval was deemed unnecessary at all participating CFBR institutions. The Chimp Haven SCCC board approved this study, which adhered to the American Society of Primatologists (ASP) Principles for the Ethical Treatment of Non-Human Primates.

2.2. Demographic Data

The data analyzed here were essentially the same as those used in our paper on CFBR chimpanzee demography [4], with several emendations. Specifically, 7 records on duplicate animals were removed; 3 records were corrected for date of birth, date of entry, or exit date; 27 records with previously missing entry dates were rectified; 1 record on a previously unobserved animal was added, and another 1705.6 life-years of observation extending through 2019 were included. This resulted in 53,234 life-years of observation on 2344 chimpanzees, consisting of 841 captive-born females (368 deceased, 473 alive and censored at the time of observation), 315 African-born females (209 deceased, 106 live), 8 females of unknown provenance (2 dead, 6 alive), 847 captive-born males (449 deceased, 398 live), 295 African-born males (222 deceased, 73 live), and 38 males of unknown provenance (26 dead, 12 alive). Finally, the African-born founders of the captive population were predominantly from the P.t. verus subspecies [31].

2.3. Population Projection Methods

The population projections used the Leslie matrix approach to model expected changes in population size and structure over time [32]. Leslie matrices involve the multiplication of a matrix of initial age- and sex-specific population sizes by a matrix of age- and sex-specific survival rates. Each iteration of the Leslie matrix cycle projected the population sizes that survived a specific year of life, which were then graduated into the next year of life. The projections utilized age- and sex-specific survivorships estimated in the CFBR chimpanzee lifetables [4]. Those life tables included still-living chimpanzees at the oldest ages (62 years for females, 55 years for males) [4]. To reach extinction, it was necessary to set a maximum age limit to the chimpanzee lifespan in the projections, upon reaching which all chimpanzees in the simulation would experience mortality without advancing to the next higher age and bring the simulated population to extinction. We used a maximum age limit of 69 years, which was older than all of the still-living chimpanzees in the life tables and older than that used in previous studies, while accommodating reports of zoo chimpanzees reaching 69 years of age [9,29,30,33]. Survivorships were extrapolated from the oldest chimpanzees in the life tables to age 69, when all animals in the simulation experienced mortality and the iteration was terminated, and smoothed to produce mortality rates that increased steadily with increasing age, as recommended [34,35]. A customized Excel spreadsheet was then built to run the Leslie matrix projections [36,37,38], following the classic deterministic approach as a single-trial projection using a fixed set of survivorships iterated year over year until extinction occurred [32]. Extinction was defined as the year when the maximum number of chimpanzees at any age declined below 1.0, a step necessary to handle the fractional numbers of survivors that would otherwise continue aging in the projection for decades without ever reaching extinction.

2.4. Population Composition

Because the overall CFBR population has effectively been nonbreeding since 1995, fertility was fixed at zero and ignored. This simplified the analysis because, absent of natality and immigration, the only demographic force affecting population size is mortality [39]. The starting year for the population projections, time t = 0, was set at 2023, the year of the most recent NIH report on population size (N = 319) [40]. A correction for population size was necessary because the CFBR population is the entire set of all former research chimpanzees in captivity (N = 918). Of those, the NIH only owns or supports a small fraction (N = 319), yielding a correction factor of 34.7% (319/918). Because the NIH only reports raw colony headcounts, the initial age/sex composition reflected the composition in our CFBR dataset (54% female, 46% male). Those starting numbers were multiplied by the age- and sex-specific survivorships to estimate the number of survivors from each year’s mortality. The survivors of mortality for one year of life were then advanced into the next year of life as that year’s starting population. The projections were iterated until extinction occurred. Annual population sizes were recorded for both sexes as they declined in size over time until extinction. Due to their different age-specific survivorships and population sizes, males and females could and did go extinct in different years.

2.5. Maintenance Costs

The annual maintenance costs for the NIH-owned or supported CFBR chimpanzees were estimated beginning in 2024 and for each consecutive year until extinction. Chimpanzee per diems are notorious for their extreme variation [17,29,30,41]. Between 2011 and 2023, the per diems varied by an average of 239% (minimum to maximum) within the colony, ranging from USD 0 to USD 107.04 per day across years [40] (see Figure 1). Within years, the per diems varied by 296%, ranging from USD 0 to USD 130.32 per day across colonies [40] (see Figure 1). We therefore used the median 2023 per diem of USD 79.60 as a reliable estimator of central tendency [40]. The per diems were multiplied by 365.25 average days per year, then multiplied by the projected population sizes per year, to estimate the total annual costs. The estimated annual costs were summed over the entire population duration, from 2024 to extinction, to estimate the total lifetime maintenance costs. The median maintenance costs were bounded by the first (USD 39.11) and third quartiles (USD 112.57) of the 2023 per diems, to define a 50% confidence band around the median.

2.6. Inflation

Maintenance costs can be expected to increase due to annual inflation in biomedical research prices [9,41]. To account for expected inflation, we utilized historical data from the Biomedical Research and Development Price Index or BRDPI [42]. Over the past 73 years (1951 through 2023), the median BRDPI inflation rate was 3.648% [42]. This approach conformed to earlier studies by including two sets of cost projections, one set in current dollars with no inflation and a second set that included a typical level of inflation, measured here by the long-term median [9,29,30].

3. Results

3.1. Rate of Decline and Population Milestones Until Extinction

The projections indicated that about 50% of the chimpanzees alive in 2023 will have died in the following 13 years, that is, by 2036, leaving N = 181 still alive (Figure 2). By 2044, the population will have declined to about 25% (N = 91) of the initial 2023 population (Figure 2). By 2052, the population will have declined to 10% (N = 37) of the initial population size (Figure 2). In 2057, less than 5% (N = 13) of the 2023 population will still be alive. Males were projected to go extinct in 2056 and females in 2062.

A close inspection of the projected population sizes revealed three general phases based on the mortality rates (Figure 2). From 2024 to 2041, the rate of decrease will be approximately linear, with an average of 6% mortality per year. From 2042 to 2051, the mortality rates will have a higher average of 10% per year, but mortality will increase more slowly year after year. This will be the era by which many more vulnerable chimpanzees have already died and only those who survived the earlier mortality bottlenecks will attain the advanced old age of 50+ years. And somewhat counterintuitively, they will have slightly longer life expectancies than those seen at earlier chronological ages. From 2052 to 2056, the rate of increase in mortality will again increase, with an average mortality rate of 16% each year (Figure 2). The male population was projected to go extinct in 2052, followed by a steady decrease in the remaining female population, with female extinction projected in 2062 (Figure 2).

3.2. Lifetime Maintenance Costs

Maintenance costs were estimated each year until projected extinction using the median 2023 colony per diem (USD 79.60), bounded by the first quartile, Q1 (USD 39.11), and the third quartile, Q3 (USD 112.57), per diem. Two separate sets of estimates were made, one in today’s dollars (no inflation) and the second with median long-term biomedical inflation. The non-inflationary scenario is not realistic because long-run biomedical inflation has averaged 3.648% per year and should be accounted for in cost estimates. However, the non-inflationary estimates were of interest because the total costs over the entire 5-decade-long period can all be stated in today’s dollars.

3.3. Non-Inflationary Scenario

For the Q1 per diem, the total costs ran to USD 73,050,552 from 2024 to the 2062 extinction, for an average of USD 1,873,091 per year (

Table 1. Maintenance costs until 2062 extinction by 3 per diems.

Inflation Scenario	Cost Estimate	Q1 per Diem (USD 39.11)	Median per Diem (USD 79.602)	Q3 per Diem (USD 112.57)
No Inflation
	Total Cost	USD 73,050,552	USD 148,669,197	USD 210,247,381
	Average cost per year (39 yrs)	USD 1,873,091	USD 3,812,031	USD 5,390,958
Median Inflation (3.648%/yr)
	Total Cost	USD 113,457,318	USD 230,903,228	USD 326,542,417
	Average cost per year (39 yrs)	USD 2,909,162	USD 5,920,596	USD 8,372,882

Legend: Total costs and average annual costs per year are tabulated for the 2 inflation scenarios (no inflation, or median 3.648% inflation) across the 3 per diems (first quartile, median, and third quartile). See text for Section 4.

). For the median per diem, the total costs ran USD 148,669,197 until extinction, for an average of USD 3,812,031 per year. For the Q3 per diem, the total costs ran USD 210,247,381 until extinction, for an average of USD 5,390,958 per year (Table 1). It was also of interest to compare differences in the total and average costs per year across the three per diems (

Table 2. Differences in total and average annual maintenance costs among 3 per diems.

Inflation Scenario	Per Diem Comparison	Difference in Total Costs	Difference in Average Costs per Year
No Inflation	Q1 to Median	USD 75,618,645	USD 1,938,940
	Median to Q3	USD 61,578,184	USD 1,578,928
	Q1 to Q3	USD 137,196,828	USD 3,517,867
Median Inflation (3.648%/yr)	Q1 to Median	USD 117,445,910	USD 3,011,434
	Median to Q3	USD 95,639,189	USD 2,452,287
	Q1 to Q3	USD 213,085,099	USD 5,463,720

Legend: Differences in total and average annual costs for the three per diems, tabulated for each of the 2 inflation scenarios. Differences represent changes in cost (cost savings or extra expenditures) to be realized by shifting per diems.

). The difference in the total costs between the Q1 and median per diems amounted to USD 75,618,645 over 39 years, for an average difference of USD 1,938,940 per year (Table 2). The difference between the median and Q3 per diems was USD 61,578,184 over 39 years, for an average difference of USD 1,578,928 per year. The difference between the Q1 and Q3 per diems was USD 137,196,828 over 39 years, for an average difference of USD 3,517,867 per year. The Q1 per diem represents a 50.9% cost savings relative to the median per diem and a savings of 65.3% relative to the Q3 per diem, while the median per diem represents a savings of 29.3% relative to the Q3 per diem. These figures represent the amount of savings, in both the total costs until extinction and in the average annual costs, which could be realized in the future depending on which per diems are actually to be paid.

3.4. Inflationary Scenario

In the second scenario, the per diems were increased annually by the median BRDPI inflation of 3.648% per year. The total costs for the Q1 per diem amounted to USD 113,457,318, for an average of USD 2,909,162 per year, to be spent from 2024 to the 2062 extinction (Table 1). The median per diem totaled USD 230,903,228, to be spent until extinction, for an average of USD 5,920,596 per year. The Q3 per diem summed to USD 326,542,417 until extinction, for an average of USD 8,372,882 per year. Differences in the total and average annual costs among the three per diems were again of interest (Table 2). The total difference between the Q1 and the median per diems was USD 117,445,910 until extinction, with an average difference of USD 3,011,434 per year. The difference between the median and Q3 per diems amounted to USD 95,639,189 until extinction, for an average difference of USD 2,452,287 per year. The difference between the Q1 and Q3 per diems amounted to USD 213,085,099 over 39 years, for an average difference of USD 5,463,720 per year. As in the no-inflation scenario, these figures represent a savings of 50.9% (Q1 to median per diems), 65.3% (Q1 to Q3 per diems), and 29.3% (median to Q3 per diems), that could be saved both in total and annually over 39 years of the future program until extinction, simply by moving the animals to a colony with a lower per diem.

Regardless of the inflation scenario or the per diems used, the projected maintenance costs will decline over time with the same general shape as the population size. An examination of projected population sizes year over year until extinction revealed that the population will decline more or less steadily from 2024 to around 2039 to 2047 when it will slowly hit an inflection point and then begin to slow down and flatten out (Figure 1). The flattening out between 2039 and 2047 probably represents the counter-intuitive slowdown in mortality rates among the very aged survivors after surviving earlier mortality bottlenecks. This population trend is reflected in the projected annual maintenance costs (Figure 3). As with the projected population size, after 2047, the costs will again decline steadily and slowly approach zero at the 2062 extinction (Figure 3).

The colony per diems play an outsized role in generating the total and annual maintenance costs (Figure 4A,B). For example, the lowest per diem (Q1, at USD 39.11) was projected to cost USD 5.1 M in 2024, the first year of our projections (Figure 4A). By comparison, the highest per diem (Q3, at USD 112.57 per day) was projected to cost USD 14.6 M in 2024, or USD 9.5 M more in one single year (Figure 4A). Furthermore, the Q3 annual costs will not decline to a comparable level as Q1 in 2024, until 2041 or 16 years later, when it was projected to run USD 5.0 M in today’s dollars (Figure 4A). But by 2041, the Q1 per diem will only amount to USD 1.7 M (Figure 4A). After adjustment for expected inflation, Q1 will cost USD 3.3 M in 2041, while Q3 will cost USD 9.5 M in 2041 (Figure 4B). That is, in 2041, the Q3 per diem will cost USD 6.2 M more in inflated dollars than the Q1 per diem (Figure 4B). This comparison illustrates the important point that per diems have more influence on maintenance costs than the actual population size.

4. Discussion

Reliable demographic data are vital to the long-term management of small populations in captivity [8,9,27,43]. Probably the single most important application of demography to small population management involves the use of age- and sex-specific survival rates to project changes in population size and structure over time. In the context of the captive conservation of an endangered species, the purpose of such projections is to avert extinction while preserving genetic variation for the future [27,43]. In contrast, this is a population of chimpanzees that were bred, but are no longer needed, for research and whose extinction was implicitly planned decades ago [8,9,10]. In such a context, the only purpose of population projection is to predict the year of extinction and estimate maintenance costs until then.

Other aspects of colony management will also be impacted by projected demographic trends. Many chronic diseases, including cardiovascular, inflammatory, and neurodegenerative diseases, are associated with aging [44,45,46]. The association between many risk factors and mortality also increases with age, resulting in excess deaths relative to deaths expected at younger ages [47]. Consequently, population projections are also necessary for epidemiological monitoring, to allow advance planning for age-related trends in morbidity and mortality and associated healthcare needs [8,9,10,29,30,48].

4.1. Comparison with Previous Extinction Projections

The previous estimates of the extinction date varied between 2036 and 2076 [9,17,25,29,30,49]. Specifically, our projected extinction year of 2062 was 6 years longer than the 2057 extinction projected in our pilot studies [29,30]. It was 9 years later than the original research chimpanzee projection of a 2054 extinction [9]. It was 20 years later than the 2036 extinction opined to the GAO [11,17]. And it was 13 years earlier than a flawed pilot study [49]. Two key factors for this wide variation in projected extinction dates include biased survivorship rates and aging in the contemporary CFBR population.

The first serious study of extinction and costs was conducted 30 years ago [9], based on sparse lifetables estimated from a young population during a growth phase [8]. Therefore, survivorships in 1995 were systematically underestimated at almost every age [4]. In the intervening thirty years, the CFBR population aged considerably, as measured by the oldest individuals, median life expectancy at birth (MLE), and the number of geriatric and extreme-old-aged individuals. In 1995, the oldest male was 44.5 years old, and the oldest female was 57.5 years old [8]. Thirty years later, the oldest CFBR male was 56.0 years, and the oldest female was 61.9 years [4]. In 1995, the mean MLE was 25.1 years (20.8 years for males, 29.4 years for females) [8]. By comparison, CFBR MLEs increased by 15 years to 35.8 years for males and increased by 11 years to 40.4 years for females [4]. Those ages are older than the research males and older than the research females in 1995 [8]. The prevalence of geriatric (30+ years) and extreme-old-age (50+) chimpanzees also increased. In 1995, there were only 116 geriatric chimpanzees (43 males and 73 females) and 4 chimpanzees (0 males and 4 females) in extreme old age [8]. Thirty years later, the CFBR chimpanzee population had 655 geriatric animals (315 males and 340 females), while another 59 chimpanzees (17 males and 42 females) had lived to extreme old age [4]. Clearly, more chimpanzees could live into the geriatric and extreme-old-age categories than previously thought [8,9,50,51]. Survivorships estimated from a young population underestimated their true values, as seen, after 3 decades of population aging [4]. This helps explain why the extinction dates of 2054 [9] or 2057 [29,30] were too early. Finally, because the CFBR population had not yet realized its full aging potential [4], the results reported here may still have underestimated CFBR population longevity. In short, reanalysis using more representative data that sampled virtually the entire chimpanzee lifespan extended extinction projections by 6–9 years over previous projections.

Several casual publications reported extinction dates that were decades earlier or later. The earliest extinction date came from a governmental study of costs and cost savings, which announced that the CFBR population would survive “for twenty years or more” [8,17]. That proposed 2036 extinction was not based on a sound scientific analysis rooted in demography but merely combined a guesstimated average age of 21 years with a simplification about a 50-year-long maximum chimpanzee lifespan, thus perpetuating an earlier misunderstanding that the population would “largely cease to exist” by 2036 [4,25] (see

Table 3. Comparison of extinction dates, costs, and model assumptions in 5 prior projections.

Reference	[8]	[25]	[29,30]	[49]	This Paper
Projection Method	Deterministic Leslie matrix	None (simple addition)	Leslie matrix	Stage-based	Deterministic Leslie matrix
Mortality data source	1995 Research chimpanzee model life tables	None (assumed mean age of 21 yrs. and 50 yrs. maximum lifespan)	Chimpanzee life tables based on two CFBR colonies	Erroneous CFBR life tables collapsed into 5 or 10 yr. long stages	Full CFBR lifetables
Population size	781	1000	676	627	319
Per diem *1	USD 54.78	USD 26.00	USD 52.08	USD 33.38	USD 52.66
Inflation	4.0%	(none)	3.352%	2.5%	3.648%
Inflation source	1995 BRDPI level	(none)	100-year BLS average	2020 BRDPI level	70 yr BRDPI median
Extinction year	2054	2036	2057	2076 (2074)	2062
Years until extinction (from 2024)	30	12	33	52 [50]	39
Total costs until extinction	USD 320,745,490	N/A	USD 322,938,038	USD 77,095,468	USD 148,669,197
Cost/yr	USD 8,018,637	USD 9,500,000	USD 7,510,187	USD 1,427,694	USD 3,812,031

Legend: The population projections used different assumptions, data sources, and projection methods, as indicated. Cost per year was estimated as the total cumulative cost until extinction divided by the number of years from 2024 until extinction. Note that stage-based population modeling has become more popular in recent years, even though it is biased relative to Leslie matrix modeling based on a full life table [52]. *1 Per diems stated in the original articles were standardized to their projected value after inflation in 2024, the starting year for our projections.

). The latest extinction date of 2076 was reported in a flawed pilot study [49] that began with the wrong population size, collapsed known survivorships into irregular 5- or 10-year-long stages, and then used a stage-based projection model known to be biased relative to full Leslie matrix modeling [52,53]. None of these three casual reports [17,25,49] can be considered reliable.

4.2. Comparison with Previous Cost Projections

Our earlier demographic study portrayed a still-living CFBR population that was smaller, older, and longer-lived than in prior studies [4]. Consequently, an updated cost analysis was also required. Most previous studies overestimated lifetime maintenance costs, although one study grossly underestimated costs (Table 2). The estimates ranged from a low of USD 77.1 M [49] to a high of USD 322.9 M [30] (see Table 2). The original 1995 projection came in at USD 320.7 M [9].

It is important to realize that maintenance costs will have the same general shape as projected population sizes over time because costs are primarily a function of population size, as inflected by the other factors, including biased survivorship schedules, inaccurate starting per diems, and/or inflation rates. The original research chimpanzee extinction projection [9] assumed a USD 25/day per diem and 4.0% inflation. But the realized inflation from 2011 to 2022 was only 3.025%, which helps explain why the maintenance costs were over-estimated [9]. Alternately, a starting per diem of only USD 22.98 best fit the actual cost data, indicating that the starting per diem of USD 25 was also too high [9]. Finally, unpredictable changes in population sizes can result from one-time “black swan” events. For example, between 2014 and 2015, the NIH chimpanzee population declined when one CFBR colony retired 59 chimpanzees to a private sanctuary in Georgia, and then again when the NIH redefined “NIH-owed or funded” to include only those colonies that received federal funding in FY2020 [21]. Such sudden declines in population size were due to migration, not mortality, and could not be predicted by any matrix population model rooted in demography.

Any combination of errors in these factors would yield inaccurate estimates of annual costs [9,30]. Understanding how such assumptions have influenced prior projections underscores the need to verify the empirical adequacy of model assumptions [29,30] and demonstrates the need for periodic demographic updates as the population continues to age and decline in size over time [4,9,30].

4.3. Ethical and Policy Significance of Costs

The decision to retire all CFBR chimpanzees [53] implied that reliable estimates of lifetime care costs should be of considerable interest to federal program managers in the planning process [30]. Indeed, that was the rationale for the original extinction projections and cost estimation [9,10]. Our projections indicate that the CFBR population will not go extinct until 2062. This is 27 years longer than the de facto NIH assumption of a 2036 extinction. This extended population longevity implies a greater ethical obligation by the NIH, who bred these endangered animals for research [54]. Specifically, the longer population survival until extinction carries a greater financial obligation on the NIH to fully plan for and fully fund their lifetime maintenance [10,21,55].

In this new light, the enormous variation in per diems among colonies and across years gains new significance. Many factors influence per diems, including husbandry and housing standards, behavioral management and enrichment practices, cost-sharing mechanisms, creative accounting practices, institutional values, political favoritism, regional variation in labor costs, scrutiny by governmental watchdog agencies like the GAO, availability of large outdoor play areas, or the need to construct new facilities [17,41,56,57,58,59]. Therefore, it is important to recognize the influence that funding decisions have on the management of the CFBR program. Regardless of the exact extinction year, shifting to lower per diems would save taxpayers millions of dollars each year until extinction. For example, USD 1.94 M in today’s dollars could be saved per year simply by switching from the median to the lowest per diem. This would pay for the estimated USD 3.41 M cost (USD 2.55 M in 2012 dollars) [59] of new facility construction at the National Chimpanzee Sanctuary in only 1 year and 9 months. This would allow the NIH to realize the promised and legally mandated retirement of all NIH chimpanzees to the federal chimpanzee sanctuary operated by Chimp Haven [17,41,58,60,61]. However, other factors are often given priority over chimpanzee well-being. For example, the NIH recently announced that chimpanzees at Colony A would finally be transferred to the National Chimpanzee Sanctuary, not because it was best for the chimpanzees, nor because a federal judge had ruled back in 2022 that the NIH had violated Congressional law by not moving them, but because the human caretakers were retiring [62]. That rationale implicitly acknowledged an NIH commitment, not to chimpanzee well-being, but to preserving jobs [62]. Specific ethical considerations include housing standards and available outdoor space. Housing varies, but all former research facilities house their chimpanzees indoors in small concrete and steel dens. Some colonies continue to house chimpanzees separately by sex, making highly unnatural social groups for a species that naturally lives in age-graded, mixed-sex groups. Limited outdoor access is allowed at former research colonies by the Primadomes found at most colonies. However, the amount of outdoor space afforded by Primadomes is quite restricted compared to Chimp Haven, which has 200 acres of outdoor wooded space. These far-reaching policy implications of maintenance costs on psychological well-being raise hard questions about why former research colonies with per diems higher than the National Chimpanzee Sanctuary should continue to house chimpanzees. These and related topics concerning human ethical responsibility for chimpanzee psychological well-being deserve deeper consideration by federal program officers.

4.4. Short-Comings and Future Goals

This study is not without its shortcomings. First, there may be some age estimation errors. But we used a statistical imputation method to avoid systematic bias in the ages of animals, with multiple DOBs recorded in the chimpanzee studbook and colony records [4,63,64,65,66]. Secondly, it might be useful to apply stochastic Leslie matrix modeling to reveal a range of possible outcomes due to variations in vital rates [67,68]. But the deterministic approach is useful in homogenous environments, such as those found in captivity, and has been advocated for endangered species [67]. Third, a detailed study of factors that influence maintenance costs, especially per diems, could effectively address the persistent problem of cost control [17]. Fourth, the observation that the CFBR population had not yet realized their full aging potential [4] could have biased the estimated survivorships downwards. Therefore, demographic updates could yield longer estimated survivorships and push projected extinction dates further into the future than modeled here. On the other hand, survivorships may be biased upwards, because all recent chimpanzee demographic studies reported a significant association between provenance and longevity, with African-born chimpanzees outliving captive-borns by up to 13.9 years [4,69,70,71,72]. There were more African-born chimpanzees in the captive population prior to 1977 when the CITES convention banned the export of African chimpanzees to the US [4]. Yet all recent captive chimpanzee life tables were estimated from a mixture of shorter-lived captive-born chimpanzees and longer-lived African-born chimpanzees. Further demographic work is needed to construct new life tables and estimate survivorship separately for captive-born versus African-born chimpanzees [4]. We plan to address these issues in future publications.

Other future goals include continued demographic updates for population modeling and characterization of the epidemiology of aging in former research chimpanzees. First, population projections need to be updated periodically. Continued assessment, monitoring, and population projection were recommended in the original extinction and costs study 30 years ago: “For planning purposes, it is also important to re-run projections regularly (perhaps annually), updating parameter estimates when appropriate and re-evaluating the results” [9] (p. 197). The first ILAR report on chimpanzees in research and the underlying demographic and cost analyses that led to the permanent breeding ban also recommended continued periodic reassessment [8,9,10]. But federal program managers failed to heed their own advice, and another two decades passed before the population’s status was studied again [29,30]. Our reanalysis confirmed that periodic updates will be necessary to refine extinction projections and update cost estimates in order to guide lifetime planning and budgeting [9,10,17,29,30]. This is especially important given population aging, expensive husbandry, complex housing requirements, and competing funding demands [8,14,17,29,30]. The data analyzed here were already 10 years old, so demographic updates should begin soon. Finally, we reiterate earlier recommendations for routine epidemiological monitoring [9,29,30]. Despite 100 years of captive studies, knowledge of chimpanzee health and disease patterns remains rudimentary. Chimpanzee veterinary medicine remains dependent on imperfect analogy to human medicine [52,71,72,73,74,75,76]. Cardiovascular diseases (CVDs) are widely recognized as the largest burden on chimpanzee mortality [77,78,79]. Yet chimpanzee CVD is unrelated to human risk factors, including obesity, serum cholesterol, and systolic hypertension [75,80]. And myocardial fibrosis is a prevalent form of CVD in chimpanzees, yet it presents differently in chimpanzees than in humans [81].

Our work demonstrates that chimpanzee morbidity and mortality can be productively studied through collaborations among veterinarians, demographers, and epidemiologists at the National Chimpanzee Sanctuary [4,29,30,77,82]. With another four decades until population extinction, there will be ample opportunities to collect and analyze biomarker results and clinical signs from routine health screenings to characterize the epidemiology of CFBR chimpanzee aging. Such studies can help improve veterinary care and facilitate effective planning to ensure that the needs of aged chimpanzees will be met over their entire lifespan. This much was originally intended by the CHIMP Act [13]. In this sense, the unexpectedly long chimpanzee lifespan will be a happy accident for comparative medicine [75].

4.5. Research Highlights

• Leslie matrix modeling was used to project the extinction of the retired CFBR chimpanzee population in 2062 or 38 years from now.
• Depending on the per diem, maintenance costs until extinction were estimated between USD 73.0 M and USD 210.2 M, in today’s dollars, or USD 113.5 M and USD 326.5 M, in inflation-adjusted dollars.
• Due to their wide variation across years and among colonies, per diems had a much greater effect on maintenance costs than the projected population size.
• Periodic demographic updates and population projections will be necessary as the population continues to age and suffers as-yet-unknown patterns of morbidity and mortality.
• The projected extinction in 2062 offers the opportunity to conduct observational epidemiological studies of chimpanzee morbidity and mortality. Knowledge gained will empower program managers and care-givers to adapt behavioral and healthcare interventions to changing population structure and better care for chimpanzees across their entire lifespan.