INTRODUCTION

The study of genetic variation in animal populations across space and through time is fundamental to our understanding of population and evolutionary biology (Mayr 1963, Harrison 1989) and for establishing effective conservation strategies (Frankham et al. 2006, Haig et al. 2011). For critically endangered species, with populations long confined to a small portion of their historical range, investigations of the contemporary population often provide little insight into the species' population history and structure prior to population decline (Matocq and Villablanca 2001, Johnson et al. 2007; but see Haig and Ballou 1995). Advances in DNA extraction from museum specimens now permit direct investigation of spatial and temporal changes in genetic variation that can provide crucial information for the development of effective conservation and reintroduction strategies (Murata et al. 2004, Draheim et al. 2012).

The observational record of California Condors (Gymnogyps californianus) suggests that they were widespread and locally abundant from southern British Columbia, Canada, to Baja California, Mexico, during Euro-American colonization (Snyder and Snyder 2000, D'Elia and Haig 2013). Population declines and range contractions were documented shortly thereafter, with condors disappearing from the Pacific Northwest, USA (north of San Francisco), in the early 1900s (D'Elia and Haig 2013), and from Baja California by the end of the 1930s (Wilbur and Kiff 1980). By the middle of the 20^th century, the species was reduced to ∼150 individuals limited to the mountains of southern California, USA (Koford 1953, Snyder and Snyder 2000). California Condor numbers continued to decline into the latter half of the century. The species was rescued from the brink of extinction in the late 1980s, when all of the remaining wild birds were caught for captive breeding (Snyder and Snyder 2000). Only 22 California Condors remained in 1982, and only 14 genetic founders (8 males, 6 females; Table 1) from 3 genetic clans were used to initiate the captive breeding program (Geyer et al. 1993, USFWS 1996, Ralls and Ballou 2004).

TABLE 1.

Mitochondrial DNA control region haplotypes of the genetic founders of the California Condor captive breeding program.

Through aggressive captive breeding efforts, followed by reintroductions beginning in 1992, population numbers have increased substantially. There are now >250 individuals in the wild, spread among 6 release sites in the southern half of their historical range (Hopper Mountain National Wildlife Refuge, southern California; Pinnacles National Park, central California; Big Sur, central California; San Simeon hills, central California; Vermillion Cliffs, Arizona, USA; and San Pedro Martir, Baja California, Mexico), and >150 individuals in captivity (Walters et al. 2010, Mace 2016, USFWS 2016). However, recovering population numbers belie significant remaining threats to the viability of the California Condor—primarily from continued contamination of their food supply by spent lead ammunition (Finkelstein et al. 2012). Furthermore, condors are still absent from the entire northern portion of their historical range, although reintroductions to this area are now being considered (D'Elia and Haig 2013, USFWS et al. 2014, D'Elia et al. 2015).

Studies of genetic variation in the California Condor have almost exclusively focused on the genetic founders of the captive population and their offspring. As expected, given their recent severe population bottleneck, these studies have revealed that the remaining population possesses low genetic diversity (Corbin and Nice 1988, Geyer et al. 1993, Chemnick et al. 2000, Adams and Villablanca 2007, Romanov et al. 2009). However, virtually nothing is known about the condor's genetic diversity or the spatial distribution of its genetic lineages prior to the population bottleneck. Technological advances in our ability to extract, amplify, and analyze ancient DNA from museum specimens now enable direct evaluation of genetic diversity through time, which can provide insights into a species' historical population structure and demographic history (Johnson et al. 2007, Wandeler et al. 2007, Draheim et al. 2012).

Studying variation in the historical California Condor population via genetic sampling of museum specimens was suggested as early as 1993 (Geyer et al. 1993), but was not feasible until recently (Clipperton 2005). Clipperton (2005) evaluated 324 base pairs (bp) from the mitochondrial DNA (mtDNA) control region of 41 individuals from the historical population (collected between 1886 and 1905 from Baja California, Mexico, to Marin County, California, USA) and 12 of the 14 genetic founders. Results from Clipperton (2005) indicated that only a single mtDNA haplotype in the historical population was not represented in the contemporary population, suggesting that the species possessed low levels of mtDNA before the demographic bottleneck in the 20^th century.

In this study, we further investigated historical California Condor mtDNA diversity using a larger sample of condor mtDNA from museum specimens. We tested the hypothesis that mtDNA diversity in the historical population of California Condors was not different from mtDNA diversity observed in the genetic founders of the captive population. We also investigated whether museum specimens from the Pacific Northwest, where reintroduction efforts are currently being considered, revealed any novel haplotypes which might suggest that the population there was once isolated from populations to the south.

METHODS

Samples

We obtained 93 California Condor tissue samples from museum specimens collected between 1825 and 1984 (Figure 1, Appendix Tables 5 and 6). These individuals represented a sample of the historical population (i.e. the population prior to the founding of the captive flock). We also obtained genomic DNA samples from the genetic founders of the captive population and their progeny from the San Diego Zoo's Institute for Conservation Research (Escondido, California, USA; Table 1). Samples of the genetic founders included those taken directly from 10 of the 14 founders, and 2 taken from progeny of female genetic founders that could not be sampled directly. Thus, we could directly or indirectly obtain the mtDNA haplotypes of 12 of the 14 genetic founders. The remaining 2 genetic founders were males (studbook #2 and #3; Table 1) that died in the wild in 1984, and were not available for sampling.

FIGURE 1.

Spatial distribution of mtDNA control region haplotypes of California Condors collected from 1825 to 1984. Singleton haplotypes are haplotypes that were observed only once in our sample (n = 67) of the historical population.

There are few specimens that were collected from the geographic periphery of the historical range (Wilbur 2012, D'Elia and Haig 2013). However, we sampled all available specimens with known provenance from the Pacific Northwest (n = 5; Figure 1, Appendix Table 5), and 2 specimens that were suspected to have been collected from this region (Appendix Table 6). Only 2 additional condor specimens that were definitively collected in the Pacific Northwest exist worldwide, both held at the St. Petersburg Zoological Museum in Russia (Specimens NHM1583 and NHM1584). However, our requests for samples from these individuals were denied due to museum policies. Another sample at the National Museum of the Czech Republic (Specimen P6V-41249) is suspected to have been collected in northern California (Wilbur 2012), but we were unable to secure a sample of this individual as it was housed in a sealed display case.

Laboratory Analyses

Ancient DNA samples can be extremely susceptible to external contamination (Wandeler et al. 2007); therefore, we used negative controls for the extraction and amplification process throughout the procedure and followed appropriate ancient DNA techniques in a dedicated clean laboratory using a UV irradiated flow hood (see Draheim et al. 2012). Before extraction, we surface-sterilized museum tissues with sterile dH₂0 and 80% ethanol washes, incubated them in a −80°C acetone bath, and then crushed them using a sterile mortar and pestle. We used a modified Qiagen DNeasy Kit (Qiagen, Valencia, California, USA) protocol for extractions, incorporating an additional 24-hr 55°C Protenase K digestion step.

Our analysis of genetic variation focused on the mtDNA control region, the major noncoding region of animal mtDNA that plays a role in replication and transcription of mtDNA molecules (Clayton 1992). We amplified the mtDNA of the genetic founders using the primer pair alt1 (Clipperton 2005) and CACO-R2 (Table 2), producing a 930 bp fragment containing domains I and II of the control region. Primer CACO-R2 was designed from an Andean Condor (Vultur gryphus) reference sequence (Genbank accession number AY129646.1; Hendrickson et al. 2003).

TABLE 2.

Mitochondrial DNA control region primer sequences developed for California Condors.

The degraded condition of some museum tissue samples required the use of multiple shorter PCR amplifications to generate a control region sequence. A combination of control region primers from Clipperton (2005) and California Condor primers designed from reference sequences generated from the genetic founders (CACO primers; see above) was used to produce 4 overlapping control region fragments ranging in size from 200 bp to 319 bp (primer combinations: alt1 and piel, Indel+ and CACO-R3, CACO-F5 and TDKD, and CACO-F3 and CACO-R4; Table 2). These fragments were aligned and assembled into a single 526 bp fragment that formed the basis of our analyses.

We performed PCR amplifications in 25 μL reactions containing 2.0 mM MgCl₂, 1 μM of primer, 100 μM of each dNTP, 1× PCR buffer, and 1 U AmpliTaq Gold DNA polymerase (Perkin Elmer, Waltham, Massachusetts, USA). PCR reaction conditions were: 5 min denaturation at 94°C, followed by 35 cycles of denaturation at 94°C for 30 s, annealing at 54°C for 30 s, and extension at 72°C for 30 s, followed by a 10-min 72°C extension. All PCR products were bidirectionally sequenced with BigDye 3.1 dye terminator sequencing chemistry (Applied Biosystems, Foster City, California, USA), and resolved on an ABI 3730 automated DNA sequencer (Applied Biosystems) at the University of Kentucky, Advanced Genetic Technologies Center (Lexington, Kentucky, USA). Resulting sequence chromatograms were aligned, edited, and trimmed using program BioEdit Sequence Alignment Editor (Hall 1999). Although studies of ancient DNA can be prone to miscoding substitutions, or deamination (Gilbert et al. 2003, Sefc et al. 2007), absence of double peaks, use of short fragment amplifications (rather than larger fragments that are more prone to false variation), and the fact that we generated sequences multiple times to verify variable positions meant that the variation in our dataset was likely real and not due to degradation of ancient DNA. All unique sequences were archived in GenBank (accession numbers KX379719–KX379736).

RESULTS

We successfully amplified 526 bp of mtDNA control region sequences in portions of domains I and II from 67 museum samples (72% of museum specimens sampled; Appendix Tables 5 and 6) and all 12 genetic founders (or their progeny) for which samples were available (Table 1). The historical population contained a minimum of 18 haplotypes with 19 variable sites (Table 3), while our census of genetic founders of the captive population revealed only 3 haplotypes with 4 variable sites (Table 4). This finding reflects a >80% decline in unique haplotypes from the historical population through the genetic bottleneck. The loss of most of these haplotypes occurred after 1900 (Appendix Table 5), although the specific timing of the decline cannot be determined with our dataset due to the small number of museum samples from the 20^th century (Appendix Table 5).

TABLE 3.

Variable sites observed in a 526 base pair (bp) fragment of the mtDNA control region assayed from California Condors collected from 1825 to 1984 (n = 67).

TABLE 4.

Comparison of California Condor mtDNA diversity in a sample of the historical population (1825–1984; n = 67) and in 12 of the 14 genetic founders of the captive breeding program. Numbers in parentheses represent ± 1 SD.

The haplotype network was starlike, with low levels of sequence divergence and a high frequency of unique mutations (Figure 2). Haplotype diversity (h) declined 20%, from a mean value of 0.851 (± 0.027 SD) in the historical population to a value of 0.682 in the genetic founders. Nucleotide diversity (π) declined by 19%, from a mean value of 0.0036 (± 0.0022 SD) to 0.0029 (Table 4).

FIGURE 2.

Median-joining network of a 526 bp (base pairs) fragment of California Condor control region mtDNA sequences. The proportions of individuals from the historical population (1825–1984) are shown in black and the proportions of genetic founders of the captive breeding program are shown in gray. Pie-chart diameter is proportional to the number of individuals with each haplotype.

Twenty-four of our samples were from individuals also sequenced by Clipperton (2005). Our sequences matched those of Clipperton (2005), except for samples 41 and 70 (Appendix Table 5), where we saw variation at position 195 in sample 41 and at position 217 in sample 70 that Clipperton (2005) did not observe. We also found 2 additional haplotypes (samples 46 and 68) in the portion of domain II outside the region investigated by Clipperton (2005). Restricting all of our historical sequences to the fragment of mtDNA examined by Clipperton (2005) resulted in the identification of 13 haplotypes, as compared with only 4 haplotypes found by Clipperton (2005). Five additional haplotypes were detected in our study simply as a matter of sampling a larger fragment of mtDNA.

The haplotypes detected in the Pacific Northwest samples (H1 and H7) were common throughout the historical range, occurring from the Pacific Northwest to the southern terminus of the historical range in Baja California (Figure 1). These haplotypes survived the genetic bottleneck and are still represented in the contemporary population (Mace 2016; Table 1).

DISCUSSION

Our analysis revealed that California Condors have lost more than 80% of their mtDNA haplotypes over the past 2 centuries. As expected by theory (Allendorf 1986), haplotypes of low frequency were lost when California Condors experienced a major range contraction and reduction in population size. The small number of haplotypes remaining in the California Condor is similar to values reported for other bird species that have gone through a severe population bottleneck, including the Whooping Crane (Grus americana; Glenn et al. 1999), Crested Ibis (Nipponia nippon; Zhang et al. 2004), and Pink Pigeon (Streptopelia mayeri; Swinnerton et al. 2004).

Our results contrast with those of Clipperton (2005), who found little genetic diversity in the historical population of California Condors. This appears to be primarily a function of the larger sample size of our analysis and the larger fragment of mtDNA that we investigated. These contrasting results suggest that studies with small samples from the historical population, and those sampling small fragments of mtDNA, should be cautious about inferences that suggest inherently low genetic diversity in a species.

Numerous studies have reported low mtDNA haplotype diversity in large raptors, including the Andean Condor (Vultur gryphus; Hendrickson et al. 2003), Spanish Eagle (Aquila adalberti; Martínez-Cruz et al. 2004), White-bellied Sea-Eagle (Haliaeetus leucogaster; Shephard et al. 2005), Bonelli's Eagle (Aquila fasciata; Cadahía et al. 2007), and Cinereous Vulture (Aegypius monachus; Poulakakis et al. 2008). Reasons for this low diversity have been attributed to: past population bottlenecks followed by long recovery times after a bottleneck due to K-type life histories (Hendrickson et al. 2003, Romiguier et al. 2014); the slow spread of new mutations through a sparsely or linearly distributed population (Hendrickson et al. 2003); island colonization and subsequent isolation and genetic drift (Bollmer et al. 2006); and recent demographic crashes caused by humans (Krüger et al. 2015). However, low mtDNA diversity is not a universal trait of large raptors, especially those with extensive geographic ranges and large population sizes. For example, the Harpy Eagle (Harpia harpyja), the largest Neotropical bird of prey, with a geographic distribution from central Mexico to east-central Brazil, was found to possess 23 mtDNA haplotypes from a sample of 66 individuals, and had relatively high haplotype diversity (h = 0.906; Lerner et al. 2009). White-tailed Eagles (Haliaeetus albicilla), which have a disjunct distribution across the Palearctic and Greenland, also have retained a substantial amount of genetic diversity (38 mtDNA haplotypes, n = 420, h = 0.797; Langguth et al. 2013). The Golden Eagle (Aquila chrysaetos), one of the most widespread birds of prey, covering essentially the whole Palearctic, was found to have 30 haplotypes (n = 252) and total haplotype diversity consistent with many other large raptors that have sizeable populations and extensive geographic ranges (h = 0.800; Nebel et al. 2015). The Lammergeier (Gypaetus barbatus), whose range once extended across large regions of Asia, Europe, and Africa, had more than 50 mtDNA haplotypes represented in its historical population and had high haplotype diversity in central Asia (h = 0.940), where its range was most extensive (Godoy et al. 2004, Krüger et al. 2015). In addition, the White-rumped Vulture (Gyps bengalensis), which was formerly considered one of the most common raptors in the world, numbering in the millions, was shown to have a large number of mtDNA haplotypes (13 haplotypes detected in only 23 individuals) and considerable haplotype diversity (h = 0.846) prior to a recent bottleneck (Johnson et al. 2008). High levels of mtDNA diversity in these species has been attributed to: the longevity of large raptors, which may buffer against immediate losses of habitat or human persecution (Hailer et al. 2006, Lerner et al. 2009); conservation of multiple local populations distributed throughout the species' range (Hailer et al. 2006); and their large geographic ranges and, consequently, potentially high effective population sizes (Nebel et al. 2015). Comparably high mtDNA haplotype diversity in the historical population of California Condors (Table 4) is consistent with the 19^th century observational record in the western United States, which suggests that condors were relatively widespread and numerous at the time of Euro-American contact (D'Elia and Haig 2013).

The appreciable genetic diversity revealed in the historical California Condor population, and reported for other wide-ranging raptors, contrasts with the results of studies of Andean Condors, which showed low amounts of mtDNA diversity (Hendrickson et al. 2003). This is unexpected, as Andean Condors are relatively widespread across the Andes Mountains and have an estimated population of at least 10,000 individuals (Birdlife International 2015). Reasons for this difference are not clear. It has been suggested that body size, metabolic rate, and generation time can influence rates of mtDNA evolution (Martin and Palumbi 1993), but these factors are likely similar for the 2 condors. There are several possible explanations for the observed differences in mtDNA diversity between the species. First, most of the control region sequences evaluated for Andean Condors were from domains II and III (n = 38), with only 5 sequences from domain I (Hendrickson et al. 2003). Given that considerable differences in diversity may exist between control region domains in avian species (Roques et al. 2004), and the more conserved nature of domain II (Brown et al. 1986, Marshall and Baker 1997), it is possible that the observed difference was the result of evaluating different portions of the mtDNA genome. Another possibility is that the small number of samples evaluated by Hendrickson et al. (2003; n = 38) was not sufficient to accurately estimate the amount of mtDNA diversity in the population. Finally, it is possible that Andean Condors truly have low levels of genetic diversity in their mitochondrial genome as a result of an evolutionary and demographic history that differs substantially from that of the California Condor. Future investigations using a larger number of Andean Condor sequences from domain I of the control region could provide additional insights into whether Andean Condor mtDNA diversity is truly much lower than in the historical population of California Condors.

We did not detect any unique haplotypes in our sample of California Condors from the Pacific Northwest, although we caution that sample sizes in this region were limited and mtDNA represents only a single locus. Despite our lack of statistical power to make definitive conclusions regarding the condor's historical population structure, the apparent lack of spatial structure is consistent with our understanding of condor movement ecology and the geography of its historical range. Condors are long-lived and known to move long distances while expending minimal energy due to their large wingspan and soaring mode of flight (Meretsky and Snyder 1992, Rivers et al. 2014). Mountainous areas are preferred by condors for soaring and looking for food because these areas provide upward-moving air currents that help large soaring birds to stay aloft (Rivers et al. 2014). Thus, the large north–south mountain chains of the Coast, Sierra, and Cascade ranges likely provided effective movement corridors for condors, facilitating genetic connectivity among local groups.

Various molecular studies of California Condors using an array of genetic markers have arrived at disparate groupings of the apparent genetic founders (e.g., Adams and Villablanca 2007; however, note that Adams and Villablanca 2007 incorrectly identified studbook #18 and #19 as genetic founders when neither individual produced any offspring; see Ralls and Ballou 2004). This is largely the result of using different genetic markers (i.e. nuclear vs. mitochondrial) or looking at different portions of the mitochondrial or nuclear genome. Nonetheless, Adams and Villablanca (2007) examined a portion of the mtDNA control region that largely overlapped with the region used in our study. Our results were identical to those of Adams and Villablanca (2007) with respect to the 12 genetic founders common to both studies, except that they identified an additional haplotype in studbook #1 (“Topa” in figure 4 of Adams and Villablanca 2007). This individual is known to be heteroplasmic (Clipperton 2005; Table 1), which explains the apparently different outcomes between our studies.

Examination of the current studbook for wild and captive California Condors (Mace 2016) suggests that all of the founder haplotypes are represented in the contemporary population, and thus that no net loss of haplotypes has occurred since the 1980s when the birds were brought into captivity. However, as of 2002, these mtDNA haplotypes were unequally distributed, with ∼62% of the population assigned to haplotype H1, ∼15% to H4, and ∼23% to H7 (Adams and Villablanca 2007).

Current genetic management of the California Condor population is based on a kinship matrix of the complete nuclear genomes of all of the founders, followed by pedigree information on their descendants, rather than being based on mtDNA haplotypes (K. Ralls personal communication). Future investigations that apply these newly developed genomic tools to the historical population have the potential to provide further insights into the historical genetic diversity and demography of the California Condor.