Andean and California condors possess dissimilar genetic composition but exhibit similar demographic histories

Abstract While genetic diversity of threatened species is a major concern of conservation biologists, historic patterns of genetic variation are often unknown. A powerful approach to assess patterns and processes of genetic erosion is via ancient DNA techniques. Herein, we analyzed mtDNA from historical samples (1800s to present) of Andean Condors (Vultur gryphus) to investigate whether contemporary low genetic variability is the result of recent human expansion and persecution, and compared this genetic history to that of California condors (Gymnogyps californianus).We then explored historic demographies for both species via coalescent simulations. We found that Andean condors have lost at least 17% of their genetic variation in the early 20th century. Unlike California condors, however, low mtDNA diversity in the Andean condor was mostly ancient, before European arrival. However, we found that both condor species shared similar demographies in that population bottlenecks were recent and co‐occurred with the introduction of livestock to the Americas and the global collapse of marine mammals. Given the combined information on genetic and demographic processes, we suggest that the protection of key habitats should be targeted for conserving extant genetic diversity and facilitate the natural recolonization of lost territories, while nuclear genomic data should be used to inform translocation plans.


| INTRODUC TI ON
Genetic diversity is widely recognized as centrally important for biodiversity assessment and conservation planning (Willoughby et al., 2015). Such focus is especially important when considering threatened and endangered species as they typically display low levels of genetic diversity, which can lead to reduced adaptive potential to environmental change (Bijlsma & Loeschcke, 2012). A common strategy to restore genetic diversity is the translocation of individuals to augment small and declining populations (Moritz, 1999;Whiteley et al., 2015). However, the motivation for restoring genetic diversity is often based on genetic surveys of modern populations, which assumes that past genetic diversity was always higher (Matocq & Villablanca, 2001;Napier et al., 2020). Nevertheless, long-term effects of translocations have been rarely assessed, and findings are inconsistent across systems: Some translocations have succeeded in increasing population fitness by incorporating new alleles (Johnson et al., 2010;Whiteley et al., 2015), while others have failed by underestimating the negative effects of genetic homogenization, outbreeding depression, disease transmissions or behavioral divergence (e.g., Deredec & Courchamp, 2007;Grauer et al., 2017;Kock et al., 2010;Manlick et al., 2017). This points to the importance of considering the natural history of populations in management plans, especially when species are geographically widespread as they may exhibit considerable differences in the degree of adaptation, isolation, and gene flow, resulting in possible incompatible genetic traits (DiBattista et al., 2020;Moritz, 1999).
One powerful approach is to harness DNA from museum collections to study the historical genetic composition of populations and understand both the spatial and temporal changes experienced (Wandeler et al., 2007). For example, recent studies have utilized museum specimens to reconstruct historic colonization routes (Pauli et al., 2015) and genetic structure (Black et al., 2018), quantify the degree of genetic erosion in persisting species (Hailer et al., 2006), and evaluate the effectiveness of reintroductions (Godoy et al., 2004).
A useful comparison within conservation biology is the management of the two extant species of condors, the Andean condor (Vultur gryphus) and the California condor (Gymnogyps californianus). The latter has been an object of intense genetic studies due to its near extinction in the 1980s and its subsequent breeding program in the 1990s . The first studies aimed at assisting in management decisions of the captive population revealed low levels of genetic variability (Ralls & Ballou, 2004). Additional analysis involving historic samples from the past two centuries confirmed that current diversity levels are a direct consequence of recent human persecution, resulting in >80% genetic decline (D'Elia et al., 2016), and currently, whole genome sequencing is being used for guiding smart breeding programs (Ryder et al., 2016). In contrast, the genetics of Andean condors has received little attention despite population declines. Though Andean condors remain widespread across South America, ranging throughout the Andean mountains from Colombia to the southernmost headlands of Argentina and Chile, they have undergone notable declines in distribution and abundance (BirdLife International, 2017;. Indeed, in the early 20th century, Andean condors once were distributed as far north as Venezuela (Calchi & Viloria, 1991), while historical records from expeditions in the early 19th century indicate that the Andean condor once inhabited the Atlantic coasts of Patagonia (Darwin, 1841;Hatcher, 1903). However, human prosecution during European colonization precipitated a continental-scale decline, resulting in local extinctions and retraction across its distributional range.
In 1970, the species was declared "Endangered" (USFWS, 1970), and three decades later was listed as globally "Near Threatened" by the International Union for Conservation of Nature (BirdLife International, 2017) and is currently recommended to be listed as "Vulnerable" . At the local level, the species is "Critically Endangered" in its northern distribution: Condors are functionally extirpated in Venezuela, <150 individuals inhabit Colombia, ~100 individuals inhabit at Ecuador, ~250 to 1,000 remain in Bolivia, and ~300 to 2,500 are estimated to occupy Peru (Méndez et al., 2019;Naveda-Rodríguez et al., 2016;Piana & Angulo, 2015). In the southern part of its distributional range, Andean condors were extirpated from the steppe and the Atlantic coasts 100 years ago (Conway, 2005), but holdout in the high Andes Perrig et al., 2017Perrig et al., , 2020. These extant populations continue to decline due to persistent poaching and habitat degradation, and face new challenges such as dietary toxins (especially lead and pesticides) and collisions with power lines Pavez & Estades, 2016;Wiemeyer et al., 2017).
Conservation of condors has primarily focused on programs aimed to reinforce extant populations and repopulate extinct ones via human-assisted translocations .
The first releases of Andean condors occurred in Colombia in 1989 using captively reared birds (Lieberman et al., 1993). This was followed by subsequent translocations in Colombia, Venezuela, Chile, Bolivia, and Argentina and, notably, recent efforts to repopulate the historic range along the Atlantic coast   (Hendrickson et al., 2003). More recently, however, we have shown that condors exhibit nuclear genetic diversity levels similar to those of other widely distributed vultures and present some degree of genetic differentiation across regions at the core distribution of the species (Padró et al., 2018(Padró et al., , 2019. Thus, it is unclear if the low genetic variability found at the continental scale is the result of low sample size, recent demographic declines, and range contraction or due to an ancestral state of mtDNA. Herein, we investigated the evolution of genetic diversity of Andean condors across their entire historic range. We employed historical mtDNA from museum specimens and genetic data from contemporary populations to evaluate whether the observed low genetic diversity is ancient and potentially intrinsic to the species or, as in the case of the California condor, is related to the recent population decline and contraction of its distributional range. We also employed coalescent simulations to evaluate the existence and intensity of major demographic bottlenecks in both condor species to assess whether similar processes have shaped the demographic dynamics of these vultures exhibiting similar ecological and life-history traits and that face similar threats. Finally, we obtained previously published mtDNA dataset from 32 individuals collected across the continent between 1961 and 1999 plus five individuals collected between 1932 and 1946 (Hendrickson et al., 2003). The temporal distribution of the combined dataset exhibited clear bimodality, with a breakpoint between historical (1884-1946) and contemporary samples . The collection dates of historical samples occurred either before or during the decline and range contraction of Andean condors. Specifically, by the first half of the 20th century the species was considered extinct in Venezuela (Calchi & Viloria, 1991) and was disappearing in Colombia (Lieberman et al., 1993;Márquez et al., 2005), and the last sightings of condors nesting on the Atlantic coast were made (Adams, 1907). Given that condors can live up to 70 years of age (Kasielke & Wallace, 1990;Meretsky et al., 2000), our historical time frame does not include individuals from the current generation.

| Molecular techniques
We extracted historic DNA from several types of tissues including feathers, toe pads, skin snips, and bone fragments using a modified QIAamp kit protocol (Qiagen) by incorporating 10 μl 1 M of dithiothreitol (DTT) and 3 μg of RNA carrier to the digestion buffer.
Contemporary DNA was extracted from molted feathers collected from roosting sites in central Argentina (see Padró et al., 2018). All extraction procedures were conducted in an isolated pre-PCR ultraclean room, equipped with air filtering units and UV lights dedicated to low-template DNA analysis. As a precaution, we handled no more than 10 samples per batch, and in addition to negative controls, we also performed a subset of replicate DNA extractions in each batch to detect possible contamination.
To perform the combined analyses with the previous dataset, we targeted mtDNA at the Control region and 12S rRNA. We initially amplified ~600 bp of the Control region and ~360 bp of 12S using the primers designed for Andean condors L16652-H621, L798-H1455, and L798-H1795 (Hendrickson et al., 2003). Degraded PCRs began with 5-min denaturation at 94°C, followed by 38 cycles of 30-s denaturation at 94°C, 30-s annealing at 60°C, and 30-s extension at 72°C, followed by 10 m of final extension at 72°C. PCR products were purified with ExoSAP-IT (USB) and sequenced in an ABI 3730xl DNA Analyzer (Applied Biosystems) in both 5′ and 3′ directions two times to test for possible genotyping errors. Resulting sequence chromatograms were visualized and aligned in MEGA-X using the MUSCLE algorithm (Kumar et al., 2018).

| Genetic diversity analysis
We combined our historical and contemporary data with previously published sequences (GenBank accession numbers in Table S1) and trimmed and aligned all sequences with our short reads. We then collapsed Control region (145 bp) and 12S (165 bp) sequences into a single 310 bp haplotype using FaBox (Villesen, 2007) to construct a temporal haplotype parsimony network and calculate the statistical probability of each link being evolutionarily correct using the TempNet script (Prost & Anderson, 2011) in R (R Core Team, 2017). We inspected the consistency of the haplotype network by considering mutation rate heterogeneity between coding and noncoding regions in NETWORK 4.6 (Bandelt et al., 1999), giving smaller weights to nucleotide positions in the Control region. We then calculated haplotype diversity (H), nucleotide diversity (π), and the average number of nucleotide differences (k) in DnaSP 6.0 (Rozas et al., 2017). We assessed the extent of genetic differentiation between regions and time periods with φST pairwise comparisons (10,000 permutations) and with exact tests of population differentiation using 1,000,000 Markov chain steps and 500,000 dememorization steps as implemented in ARLEQUIN 5.3.2 (Excoffier & Lischer, 2010). Our analyses were performed on three regional groupings of samples based on dispersal expectations of condors. We defined a priori populations by geographic proximity according to the species' gene-flow distance that has been recently estimated as a radius of ~1,200 km of each putative population (Padró et al., 2018), dividing northern, central, and southern Andes, providing us with a balanced sample size of 24, 28, and 21 individuals, respectively ( Figure 1a). We then employed hierarchical analyses of molecular variance to evaluate the amount of population genetic structure across time periods for the whole continent and among regions within time periods using 10,000 nonparametric permutations. Finally, we employed Mantel tests to assess the correlation between the evolutionary genetic distance, taking into account differences between nucleotides and inequality of frequencies (maximum composite likelihood method; Kumar et al., 2018), and geographic distances (10,000 permutations) using adegenet R package (Jombart & Ahmed, 2011).

| Demographic analysis
To explore the regional demographic history of condors, we per-  Table S2). Both species typically begin breeding at 6-8 years of age (Houston, 1994;; we set the generational time at 7 years. The mutational model of each locus was determined by the likelihood scores in JModelTest 2.1.10 (Darriba et al., 2012), resulting in HKY for both Andean and California condors (parameter details in Table S3). We modeled four different scenarios for the observed patterns of genetic variability in condors ( Figure 2). First, we introduced the no-change scenario, assuming no significant variation in population size across time (Figure 2; Scenario 1). Given that both species of condors experienced populations declines and range contractions during the last centuries, we also simulated a large ancient population size with an historical-recent bottleneck scenario, using a prior timing uniformly distributed between 7 and 350 years ago (t 1 ) to take into account the possible influence of the early European settlers and modern civilization (Figure 2; Scenario 2). Populations could also have declined as the consequence of historic climatic changes or the extinction of megafauna during the late Pleistocene (Emslie, 1987;Tambussi & Noriega, 1999). Thus, we included an ancient bottleneck uniformly distributed between 1,000 and 500,000 years ago (t 2 ) without the historical-recent bottleneck ( Figure 2; Scenario 3). Given that Andean and California condors are the only surviving members of a once diverse guild, it is possible to consider that the extinction of competitors during the late Pleistocene (>10,000 years ago) favored some demographic boost (Emslie, 1988;Perrig et al., 2019;Tonni & Noriega, 1998)

or condors
initially benefited from the introduction of livestock in the Americas before population declines (Adams, 1907;Emslie, 1987). Thus, we included a scenario with an expansion time period uniformly distributed up to 50,000 years (t e ) between the ancient and historicalrecent bottlenecks (Figure 2; Scenario 4).
We set the prior information of the historic-recent effective female population size (Nef1) of Andean condors between 50 and 1,000 for the northern region and between 500 and 5,000 for the south. These priors were based on the estimates of regional popula-  Table S4). After running 5 × 10 6 simulations for each analysis, we pre-evaluated the prior distributions by comparing the distribution of summary statistics and observed data using principal component analysis. We choose the most likely scenario based on their posterior probability applying polychotomous logistic regressions of each scenario probability on the deviations between simulated and observed summary statistics (based on 1% of simulated datasets) and comparing their 95% confidence intervals (Cornuet et al., 2014). We then calculated the posterior distribution of parameters in the most likely scenario by local linear regressions using 1% of simulated data closest to the observed data. Finally, we evaluated confidence in the scenario by calculating type I and II errors and performed model checking computations to assess the "goodness-of-fit" by comparing the observed data with the 1% of simulated data under each model-posterior combination (Cornuet et al., 2010).

| Genetic diversity
We sequenced Control and 12S regions from 33 historical and 8 contemporary samples (41 samples showed that most of the genetic structuring was due to differences of the north against the central (φST: 0.55, p < .001, exact-p = .01) and southern Andes (φST: 0.16, p < .01, exact-p < .01), but not between central and south (p > .05 in all cases). The spatial sorting of the haplogroups between central-southern (Bolivia, Chile, and Argentina) and northern Andes (Peru, Ecuador, Colombia, and Venezuela) showed that HIII was exclusively found in the north; HIV, HV, and HVI were only present in the south, while HII was predominant in the south (93.75%) and HI was evenly distributed across regions (Figure 1c). Mantel test between genetic and logarithmic geographic distance revealed significant isolation by distance (n = 73, r = .07, p = .02), indicating a history of evolutionary divergence between north and south (Figure 1d). Nucleotide and haplotype diversity levels were higher in the contemporary samples and in central-southern South America (Table 1). Our comparisons of φST and exact tests between time periods showed significant temporal structure (φST: 0.17, p < .01, exact-p = <.001). However, our hierarchical AMOVA showed no statistical differences across time, while regions exhibited marked differences within time periods, with most of the genetic variation explained within regions (Table 2).   Table S6).

| Demographic scenarios
Our analyses showed that type I error rate was 0.35, 0.61, and 0.39 for California and Andean condors from the north and centralsouth, respectively. Notwithstanding, most of the error rates were explained by miss-classification between scenarios 2 and 4, both including recent demographic bottlenecks ( Figure 2; Table S7), and a combined posterior probability > 99% for California and southern Andean condors and > 71% in northern Andean condors (Table S5).
Type II error was 0.16 for the California condor, whereas in the

| Genetic diversity patterns
Our study of contemporary and historical mitochondrial data in conjunction with a continental-level sampling of the Andean condor allowed us to characterize the spatiotemporal genetic composition for this species of special conservation concern. We showed that Andean condors have lost at least 17% of genetic variation by the early 20th century. As expected, genetic erosion was associated with the loss of a mitochondrial lineage in the peripheral distribution during the contraction of its southern geographic margin in the Note: Sample size (n); number of haplotypes (n H ); nucleotide diversity (π); haplotype diversity (H); average number of nucleotide differences (K). Atlantic coast. However, we found no evidence of lost haplotypes in the northern distribution where populations have been driven to near-extinction Naveda-Rodríguez et al., 2016). A possible explanation is that the genetic decline and range contraction in the north may have started long before the collection dates. Lower genetic variation and detection of significant mitochondrial substructure against the central-southern region reinforce the idea of genetic drift acting in northern South America.

Sum of squares
Overall, our study yielded similar results to those reported by Hendrickson et al. (2003), indicating that the extreme low mtDNA diversity in the Andean Condor is mostly ancient. In fact, we found more genetic variation in the contemporary samples than historical ones, resulting in mixed results between time periods. Low levels of genetic diversity seem to be intrinsic to the life-history traits of many raptors, as long generational times result in low mutation rates, large body size often reflects smaller net effective population sizes, and high dispersal power may result in homogenizing effects (Berlin et al., 2007;Martínez-Cruz, 2011 Dowling et al., 2008;Toews & Brelsford, 2012). Many authors have stressed the importance of natural selection in shaping divergent patterns of mtDNA across species, including the role of selection in mito-nuclear fitness interactions, where depending on the variation in the nuclear genome could constrain mitochondrial variation (e.g., Dowling et al., 2008). In addition, demographic and behavioral factors can also promote mtDNA homogenization in systems with skewed sex ratios or female-biased dispersal (Toews & Brelsford, 2012). Although Andean condors do not exhibit sex-biased dispersal (Padró et al., 2018), skewed sex ratios have been suggested to be driven by higher female mortality (Lambertucci et al., 2012;, possibly promoting mtDNA homogenization. Moreover, as observed among Cinereous vultures (Poulakakis et al., 2008) and Spanish Imperial eagles (Martínez-Cruz et al., 2007), Andean condors from central Argentina exhibited low mtDNA diversity, but show no evidence of nuclear genome-wide erosion (Padró et al., 2018), suggesting that evolutionary constraints on genetic diversity only affected the mitochondrial genome.

| Demographic processes
In concordance with previous work, we found that California condors northern Andean condors might be explained by low statistical power due to the reduced genetic variability or because of older coalescence times compatible with the northern origin of the group (Emslie, 1988;Tambussi & Noriega, 1999). If true, demographic expansion in the central-south could also be attributed to the recolonization dynamic from the glacial refugia in the tropic (Hewitt, 2000). Our estimated timing of demographic expansion (1,792-20,020 ya) is consistent with the postglacial recolonization of higher latitudes by highly vagile species of the southern hemisphere such as whales, seals, and penguins (Fraser et al., 2012;Thatje et al., 2008). Regardless of the specific demographic century (Adams, 1907;Hatcher, 1903). Our finding of a unique mitochondrial lineage in that region suggests that those condors may have represented an endemic coastal subpopulation as first suggested by Lydekker (1895), and not an extension of the distributional range from the mountains due to the abundance of food offered by sheep farms as hypothesized by Adams (1907). Nevertheless, it is possible that the coastal subpopulation was already in decline due to the loss of marine subsidies, forcing condors to feed more on terrestrial mammals , resulting in a deadly conflict with Patagonian ranchers (Ballejo et al., 2020;Pauli et al., 2018). Paintings and narratives from this period illustrate the persecution of condors, suggesting that hunting was common (e.g., Castellanos, 1923;Darwin, 1841;Lydekker, 1895) and a likely cause  . Given that the introduction of cattle in California started back in the 18th century (Chamberlain et al., 2005), it is possible that, as in the case of the Atlantic Andean condors, the collapse of large marine mammal populations during industrial whaling in the 20th century (Rocha et al., 2014) triggered condor's extinction by pushing them into a direct conflict with farmers (Chamberlain et al., 2005;Lotze & Worm, 2009). Although the current recovery of marine mammal populations (Lotze et al., 2011) is a major step forward to re-establish condor populations in the coasts, for populations to be self-sustaining, conservation strategies need not only to enhance ecosystem protection (Kurle et al., 2016;Perrig et al., 2020), but also to include appropriate genetic management when considering breeding and translocation programs.

| CON CLUS IONS
Low levels of genetic diversity found in the Andean condor represent a natural state of mtDNA, and thus are unlikely to be an immediate threat to long-term viability. However, we showed that the detection of further genetic loss is difficult when ancient diversity is already low, highlighting the importance of using museum samples that include the historical range of distribution. We found that similar to California condors, genetic decline of Andean condors cor-

CO N FLI C T O F I NTE R E S T
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.