The hypervariable region I (HVRI) is persistently used to discern haplotypes, to distinguish geographic subpopulations, and to infer taxonomy in a range of organisms. Numerous studies have highlighted greater heterogeneity elsewhere in the mitochondrial DNA control region, however–particularly, in some species, in other understudied hypervariable regions. To assess the abundance and utility of such potential variations in orang-utans, we characterised 36 complete control-region haplotypes, of which 13 were of Sumatran and 23 of Bornean maternal ancestry, and compared polymorphisms within these and within shorter HVRI segments predominantly analysed in prior phylogenetic studies of Sumatran (~385 bp) and Bornean (~323 bp) orang-utans. We amplified the complete control region in a single PCR that proved successful even with highly degraded, non-invasive samples. By using species-specific primers to produce a single large amplicon (~1600 bp) comprising flanking coding regions, our method also serves to better avoid amplification of nuclear mitochondrial insertions (numts). We found the number, length and position of hypervariable regions is inconsistent between orang-utan species, and that prior definitions of the HVRI were haphazard. Polymorphisms occurring outside the predominantly analysed segments were phylogeographically informative in isolation, and could be used to assign haplotypes to comparable clades concordant with geographic subpopulations. The predominantly analysed segments could discern only up to 76% of all haplotypes, highlighting the forensic utility of complete control-region sequences. In the face of declining sequencing costs and our proven application to poor-quality DNA extracts, we see no reason to ever amplify only specific ‘hypervariable regions’ in any taxa, particularly as their lengths and positions are inconsistent and cannot be reliably defined–yet this strategy predominates widely. Given their greater utility and consistency, we instead advocate analysis of complete control-region sequences in future studies, where any shorter segment might otherwise have proven the region of choice.
Citation: Banes GL, Galdikas BMF (2016) Effective Characterisation of the Complete Orang-Utan Mitochondrial DNA Control Region, in the Face of Persistent Focus in Many Taxa on Shorter Hypervariable Regions. PLoS ONE11(12): e0168715. https://doi.org/10.1371/journal.pone.0168715
Editor: Tzen-Yuh Chiang, National Cheng Kung University, TAIWAN
Received: July 19, 2016; Accepted: December 4, 2016; Published: December 29, 2016
Copyright: © 2016 Banes, Galdikas. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Novel sequences generated in this study are available in GenBank, under the accession codes KU523975-KU523977 and KX427542-KX427574.
Funding: Collection of samples at Camp Leakey was generously supported by the University of Aberdeen (UoA; http://www.abdn.ac.uk/) Expedition Fund; UoA Small Grants Fund; UoA Alumni Annual Fund; the John Reid Trust; the Royal Geographical Society with Institute of British Geographers (http://www.rgs.org/); the Royal Scottish Geographical Society (http://www.rsgs.org/); the Gilchrist Educational Trust (http://www.gilchristgrants.org.uk/); the Orangutan Foundation UK (http://www.orangutan.org.uk); the North of England Zoological Society at Chester Zoo (http://www.chesterzoo.org/); Darwin College, Cambridge (http://www.darwin.cam.ac.uk/); the Primate Society of Great Britain (http://www.psgb.org/), through a Charles A Lockwood Memorial Grant; the ARCUS Foundation (http://www.arcusfoundation.org/); the Leverhulme Centre for Human Evolutionary Studies, University of Cambridge (http://www.human-evol.cam.ac.uk/); the Ridgeway-Venn Studentship; the Miss Millie Foundation and the American Association of Zoo Keepers Los Angeles Chapter. The long-term research conducted at Camp Leakey, which facilitated sample collection in Tanjung Puting National Park, was primarily funded by the Orangutan Foundation International (http://www.orangutan.org/). Collection of samples in the United States, and subsequent laboratory analyses, was funded by the Sacramento Zoo Conservation Fund (http://www.saczoo.org/); Como Park Zoo and Conservatory (http://www.comozooconservatory.org/); Orangutan Outreach (http://www.redapes.org/); Henry Vilas Zoo (http://www.vilaszoo.org/); The Orang-utan Conservation Genetics Trust; the Max Planck Society (http://www.mpg.de); the Chinese Academy of Sciences (http://www.cas.ac.cn) and the National Natural Science Foundation of China (http://www.nsfc.gov.cn/; grant numbers 31450110070 and 91331203). GLB was supported by the Orang Utan Republik Foundation (http://www.orangutanrepublik.org/) through the 2012 LP Jenkins Memorial Fellowship and by the German Academic Exchange Service (DAAD; https://www.daad.de/). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Predominantly maternal inheritance, large copy number, and high replication and mutation rates render mitochondrial DNA (mtDNA) especially advantageous in studies of population genetics . If a female does not inter-breed or has no female offspring, and if her mtDNA haplotype does not occur elsewhere in the breeding population, her mitochondrial lineage will usually disappear . Mitochondrial loci have a smaller effective population size as compared to autosomal markers, therefore, and–as mtDNA lineages are often specific to geographic areas–such loci are often used to define taxonomic units and to identify genetically distinct subpopulations . Replication independent of the cell cycle  ensures large numbers of copies, typically up to several thousand per cell , facilitating mtDNA analyses from ancient or non-invasive samples when nuclear DNA quantities are insufficient or limited. High mutation rates–linked to increased susceptibility to DNA damage, higher copy-rate margin for error (reviewed in ) and greater oxidative stress –prove especially useful in studies of small populations or those comprising few generations .
Actual mitochondrial DNA mutation rates are likely to vary between cell types, and the abundance and distribution of individual mutations is known to vary between species . Nonetheless, it is commonly accepted that the highest mutation rates occur in the non-coding control region (or displacement ‘d’ loop, given its function in replication), which–in humans–comprises 1,122 bp of the 16,569 bp genome (i.e. 6.77%) . Variable blocks may evolve four to five times faster in the control region than elsewhere in the mtDNA molecule [11,12], probably due to preferential mutational hotspots . Much of this variation is concentrated within a short segment in the left domain of the control region, now commonly termed the Hypervariable Region I (HVRI) [11,14]. Given its increased and concentrated abundance of mutations, the HVRI is arguably the most commonly used mitochondrial marker in phylogenetic studies, even in the face of DNA barcoding, which typically relies on coding markers (e.g. cytochrome oxidase I) that are largely conserved within species . Accordingly, the HVRI has been chosen in countless taxa to infer species, subspecies and geographic subpopulations (e.g. ); to determine the origins of trafficked wildlife (e.g. [17–19]); to trace the provenance of confiscated wildlife parts (e.g. [20–22]) and to identify endangered species unlawfully sold for consumption (e.g. ).
Despite the inherent popularity of the HVRI as a phylogenetic marker, numerous studies have drawn attention to variations elsewhere within the control region. In particular, second (i.e. HVRII) or even third (i.e. HVRIII) hypervariable regions have since been discerned in a broad range of taxa. The sizes and positions of all such hypervariable regions are often arbitrarily defined, however–even within species–with little consistency across the literature . In humans, two hypervariable regions are commonly described, spanning either positions 16024–16365 (HVRI, 342 bp) and 73–340 (HVRII, 268 bp) (e.g. [25,26]) or positions 16024–16383 (HVRI, 360 bp) and 57–372 (HVRII, 316 bp) (e.g. [13,27]) in the revised Cambridge Reference Sequence . Some studies consider both regions to be much longer, occasionally spanning up to 610 bp each (e.g. ), while others consider both to be shorter and append a compensating third hypervariable region (i.e. HVRIII) (e.g ). Lack of a consensus definition proves especially problematic when comparing data from multiple research efforts. Different studies that target the same regions, in the same species, will often use different primer sets to produce sequences of varying length that may not fully overlap .
Nonetheless, it is accepted that the HVRI is relatively short–comprising, as commonly defined in humans, up to only 32% of the complete control region. As a consequence, variation outside of the HVRI is usually overlooked, in spite of the value of such variations having been established in numerous studies. When defining only two hypervariable regions, Meyer et al. (1999) observed greater overall numbers of polymorphic sites in the human HVRII than in the HVRI . Imes et al. (2012) reported a similar pattern in domestic dogs (Canis lupis familiaris) , while greater overall nucleotide diversity has been reported in the HVRII than in the HVRI of Mauritian long-tailed macaques (Macaca fascicularis, ). In some cases, variations outside of the HVRI have proved of equal or greater value in phylogeographic studies: for example, Wilkinson-Herbots et al. (2007) remarked that a single polymorphism in the HVRII could be used to help distinguish five major mtDNA clusters among European Caucasians . Indeed, both HVRI and HVRII by any common definition exclude frequent nucleotide variants characteristic of many human populations, including transitions at 7 positions that can be used to help diagnose at least 12 established haplogroups . At the very least, polymorphisms outside of the HVRI have been shown to have forensic utility in discerning distinct matrilines that might otherwise appear identical (e.g.[31,34]). It is therefore pragmatic to postulate that the complete control region–encompassing however many hypervariable regions–should be analysed as a single unit, rather than inconsistently sequencing arbitrary segments.
Orang-utans (Pongo spp.) are among the many species that owe their present-day taxonomy, at least partially, to studies of mitochondrial DNA. In conjunction with prior knowledge of a pericentric chromosomal inversion , complete mtDNA genome sequences of orang-utans from the islands of Borneo and Sumatra [36,37] were instrumental in conjunction with other mtDNA markers (e.g. [38,39]) in the decision to divide the genus into two distinct Bornean (Pongo pygmaeus) and Sumatran (Pongo abelii) species–though not without considerable debate (e.g. ). On the basis of short, ~245 bp sequences of a primary stretch of variation within the left domain of the mtDNA control region (i.e. a purported HVRI), Warren et al. (2001) later identified four geographic subpopulations on Borneo, implying that these might constitute subspecies of the Bornean orang-utan . Groves (2001), on the other hand, concluded on the basis of skull morphology that only three subspecies could be defined on Borneo–and, contrary to the available genetic data–his taxonomy was apparently favoured and continues to persist [42–44]. Nonetheless, the segment amplified by Warren et al. (2001) continues to be the predominant mitochondrial marker used in orang-utan phylogenetic studies, though most have produced sequences of varying lengths that may not encompass the full stretch of hypervariability [45–49] (Table 1). While many studies report amplifying the ‘hypervariable region I’, the length and position of such a stretch is unknown in either species, and thus is presumably considered similar to those reported in closely related taxa. In practice, analyses of nucleotide diversity over a large number of aligned complete control-region sequences is necessary to identify and define hypervariable regions (e.g.[11,14]).
Many were cut by the original authors from longer sequences (i.e. larger PCR amplicons, containing flanking regions) prior to analysis and/or publication; only those bases derived from the control region are specified here. Nucleotide positions correspond to those of the complete Sumatran orang-utan mitochondrial DNA genome (GenBank accession code X97707; ) and are estimated for unpublished sequences based on the lengths noted in each manuscript. The control region spans positions 15484–16499 of the reference genome.
At the time of submitting this manuscript, only five complete control-region sequences could be found in published databanks: two from Sumatran  and three from Bornean [36,37]) orang-utans. The two Sumatran sequences are identical at control-region resolution. The remainder of all published control-region sequences were all considerably shorter in length, and all comprise only the left domain or parts thereof. Two segments within this domain are predominant: among Sumatran orang-utans, a ~385 bp segment ; among Bornean orang-utans, a ~323 bp segment [45,46] encompassed within the former.
Here, we characterise the complete mitochondrial DNA control region in Sumatran and Bornean orang-utans. We first determine the length and position of the purported hypervariable region I within the complete control region, and assess the extent to which this section was sequenced and analysed in six major published orang-utan phylogenetic studies [41,45–49] (Table 1). Rather than arbitrarily defining further hypervariable regions, we then compare the extent and utility of polymorphisms present in the remainder of the control region–all of which have been ignored to date in all prior studies of intra-island phylogeography–to those occurring in the aforementioned predominantly analysed segments (i.e. 385 bp, Sumatran; 323 bp, Bornean). We determine the minimum extent to which these variations might prove informative and useful in this and future phylogenetic studies.
Materials and Methods
Faecal samples were collected from 15 free-ranging Bornean orang-utans in Tanjung Puting National Park, Central Kalimantan, Republic of Indonesia, and genomic DNA extracted, as previously described . Samples were frozen at -20°C for up to five years prior to DNA extraction. All required permits and approvals were obtained from the Indonesian State Ministry of Research and Technology (RISTEK), the Indonesian State Ministry of Forestry (PHKA), the Indonesian Institute of Sciences (LIPI), the Indonesian Agency for Natural Resource Conservation (BKSDA) and the Tanjung Puting National Park Office (BTNTP). As per the requirements of Indonesia’s domestic legislation, samples were exported to the European Union, via the United Kingdom, under permits 14459/IV/SATS-LN/2008 and 00459/IV/SATS-LN/2011 from the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES).
Blood samples (including whole blood, serum and plasma) and tissue biopsies were collected from 18 Sumatran, 22 Bornean and 11 hybrid orang-utans, of which 6 and 5 were descended maternally from each species respectively, from zoos in the United States of America. These samples had either been banked routinely by each institution over time, or–in the case of blood products–were drawn opportunistically for use in this and other studies, either during routine medical procedures or voluntary blood-draw training. Permission was granted by each Institutional Animal Care and Use Committee or equivalent body, and approved by recommendation of the Orangutan Species Survival Plan (SSP) Steering Committee. Samples were exported from the USA under CITES permit 11US49805A/9 and imported to the European Union, via the United Kingdom, under CITES permit 477248/01-27. Genomic DNA was extracted from samples of all blood products using the standard protocol of the QIAgen DNA Blood Mini Kit, with one modification: DNA from whole blood was eluted into 400 μl, and DNA from serum and plasma into 50 μl, of Buffer AE.
Of all 66 orang-utans, at least 46 were presumed to derive from different matrilines, on the basis of institutional or reintroduction records. The remaining 20 were known to be full siblings, maternal half-siblings or matrilineal relatives of other sampled individuals, and thus were assumed to share the same mitochondrial haplotype.
DNA amplification and sequencing
Improving on prior studies, we amplified the complete orang-utan mitochondrial DNA control region in a single polymerase chain reaction (PCR), in a protocol that proved successful even for highly degraded and low-quantity faecal- and serum-derived DNA extracts. We designed the orang-utan-specific primers CYTBMIDF (5-CAA TCC TAC GAT CCG TCC CC-3) and 133R (5-CGG GGA TGC TTG CAT GTG TAA C-3) to anneal to the cytochrome b and 12S rRNA regions respectively, on opposite sides of the control region (Fig 1). By targeting the complete control region in a single reaction, and by using species-specific primers, we intended to mitigate the risk of amplifying nuclear mitochondrial insertions (‘numts’), which are known to be paralogous to orang-utan mitochondrial DNA  and which have proven problematic in phylogenetic studies of a range of other hominoid taxa . We chose to use the Expand Long-Range dNTPack (Roche) to amplify the target region, which typically proved successful with only 1 μl template DNA irrespective of the source material. Though we successfully amplified the region using lower-fidelity enzymes and master mixes, such reagents required considerable optimization between reactions and, in the case of faecal samples, unsustainably large quantities (up to 10 μl) of template DNA in short supply from wild individuals.
The sizes of each region in this schematic are proportional to their sizes in the orang-utan mitochondrial genome.
Each 25 μl reaction comprised 5 μl 5x Expand Long Range Buffer with MgCl2 (12.5 mM), 1.25 μl dNTPs (10 mM each), 1.5 μl 100% DMSO, 0.75 μl each primer (10 μM), 0.35 μl Expand Long Range Enzyme Mix (5 U/μl), 14.4 μl ddH2O and 1 μl DNA extract. Thermal cycling conditions were as follows: initial denaturation at 92°C for 2 minutes; 10 cycles of 92°C for 10 seconds, 60°C for 15 seconds and 68°C for 2 minutes 30 seconds; 15–20 cycles of 92°C for 10 seconds, 60°C for 15 seconds and 68°C for 2 minutes 30 seconds (plus 20 seconds per cycle); and a final extension at 68°C for 7 minutes, followed by a 4°C hold. To reduce the risk of PCR error, we amplified DNA from each orang-utan twice, in two independent PCR reactions, each with template DNA derived from different DNA extracts.
PCR products were electrophoresed on 1% TAE agarose gels, and only those yielding single bright bands of the expected size were subject to DNA sequencing, using an ABI 3730xl DNA Analyzer with integrated KB basecaller under BigDye Terminator v3 cycling conditions (Applied Biosystems). Due to the limitations of the Sanger method in producing sufficiently long sequences, we ‘primer walked’ each pair of reactions with two different sets of primers, producing overlapping sequence contigs for each DNA extract. The first product of each pair was sequenced twice in the forward direction, using the primers CYTBMIDF and CRMIDF (5-CCC CTC AGT TAG TGG TCC CT-3, this study). The second product was sequenced twice in the reverse direction, using the primers CRMIDR (5-GGA GCG AGG AGA GTA GCA CT-3, this study) and 133R (Fig 1). Each base was therefore confirmed at least twice independently, from overlapping sequences generated from two separate reactions each from different DNA extracts.
Sequence contigs were assembled by mapping reads to a reference sequence of the complete Sumatran orang-utan mitochondrial genome, in GENEIOUS 8.1.8  (GenBank accession code X97707; ). Chromatograms were manually checked in all cases where a base was called with lower than 99% probability of being correct. We used RDP4  to check for recombination among sequences, sensu , as a means of identifying numts. Consensus sequences were collapsed into haplotypes and aligned using MUSCLE , with 3 of the 4 unique complete control-region sequences available in GenBank at the time of analysis: accession codes X97707 (P. abelii), X97709 and X98472 (both P. pygmaeus) (all ). As the published sequence of Horai et al. (1992)  (accession code D38115) features an 81 bp deletion not observed in these other published sequences, and given that another of those authors’ sequences was found to be a composite of several from multiple DNA sources , we chose to exclude this sequence.
Hypervariable regions were visualised by plotting average nucleotide diversity among the alignment of all sequences, computed in DnaSP 5.10.01  via sliding window and considering all alignment gaps across all sites. We determined the nucleotide positions of each region by first discerning those of conserved regions, as identified by inferring a conservation threshold given the observed number of segregating sites (S).
To demonstrate the forensic utility of progressively longer sequences, we cut our alignment to the various lengths of those published and analysed in six prior studies: 244 positions ; 325 positions [45,46]; 387 positions ; 409 positions  and 422 positions  (for nucleotide positions, see Table 1). In the case of the latter two studies, for which the authors chose not to release their sequences, we cut the alignment based on the length of the analysed sequences specified in each manuscript. We computed the number of haplotypes (h), number of polymorphic sites and total number of mutations, and mean number of nucleotide/pairwise differences (k) among alignments of each length in DnaSP 5.10.01 .
To assess the phylogeographic utility of those nucleotide positions not typically sequenced in Bornean orang-utans, we created two alignments comprising all of our novel complete control-region sequences, plus four chimpanzee (Pan troglodytes) sequences from GenBank to serve as an outgroup. For each, we either: 1) re-aligned the sequences with those derived from Bornean orang-utans of known geographic origin , and cut the alignment to the length of the predominantly analysed 323 bp segment; or 2) excised the 323 bp segment predominantly analysed for Bornean orang-utans, concatenated the remaining bases, and re-aligned the nucleotides. We repeated this procedure for Sumatran orang-utans, with published sequences from individuals of known origin on Sumatra , instead cutting the alignment to the 385 bp segment predominantly analysed in this species. We inferred phylogenetic trees from all alignments, and from an alignment of only all complete control-region sequences, with which to compare the consistency of inferred evolutionary relationships. Accession codes for published sequences used in these analyses are indicated in the S1 Table.
We used jModelTest 2.1.7 , computing likelihood calculations using Phyml  to determine the best model of nucleotide substitutions for alignments under corrected Akaike Information Criteria (AICc). Trees were estimated using MrBAYES 3.2.6  implementing Metropolis-coupled Markov chain Monte Carlo (MCMCMC) to approximate posterior probabilities, with four chains swapped at a frequency of 0.2. Analyses were run for a maximum of 10,000,000 generations from a randomly generated tree, sampling every 100 generations. The first 25% of trees generated before convergence occurred were discarded. Resulting trees were drawn as cladograms; nodes were labelled with posterior probabilities.
Complete control-region sequences were successfully generated from all 66 samples. In some cases, it was necessary to perform further DNA extractions–particularly from faecal samples–in order to yield sufficient genomic DNA for amplification. In all such cases, however, we were unable to amplify even short segments of the purported HVRI  using the initial DNA extracts: thus, the DNA was either of sufficient quality for complete control-region amplification, or entirely insufficient for amplification at all. In no circumstances could we amplify only shorter segments of the purported HVRI, but not the complete control region, in a single PCR.
The resulting 36 complete control-region haplotypes ranged from 999–1016 bp in length (Sumatran: 1011–1016 bp; N = 13; Bornean: 999–1000 bp, N = 23), and have since been submitted to GenBank (accession codes KX427542—KX427574). Three update shorter sequences previously generated from the same individuals in our prior study , and retain their original accession codes (KU523975-KU523977). One sequence (haplotype S4) proved identical at this resolution to a published full-genome sequence in GenBank (accession code X97708; ), known to have derived from a different matriline. Heteroplasmy was not observed in any sequence, nor were any bases heterozygous in sequence contigs. We did not detect recombination in any sequence, and those from maternally related individuals proved identical as expected, irrespective of sample type. This considered–and given our diligence in exercising all feasible precautions–we have no reason to believe our sequences feature nuclear mitochondrial insertions.
Presence and position of hypervariable regions
An alignment of all 35 novel sequences with those three previously published  comprised 1022 nucleotide positions. On the basis of sliding window analysis, we identified a primary stretch of variation, i.e. HVRI, spanning at least the first 376 sites (≤374 bp) and consistent among sequences of Bornean and Sumatran origin (Fig 2). When using the complete Sumatran orang-utan mtDNA genome of Xu and Arnason (1996) as a reference for sequences from both species (accession code X97707), we consider the HVRI to span at least positions 15484–15855. When referenced against the two pre-existing complete control-region sequences, the HVRI spans at least positions 1–372 in Sumatran orang-utans (accession code X97708) and at least positions 1–374 in Bornean orang-utans (accession code X97709). As defined here, the orang-utan HVRI aligns well with the human revised Cambridge Reference Sequence, from positions 16024–16400 (accession code NC_012920) .
Diversity was computed for alignments comprising, a) all 38 sequences (1022 sites), b) 13 sequences of Sumatran maternal ancestry (1018 sites), and c) 25 sequences of Bornean maternal ancestry (1002 sites). Data were computed under the sliding window method, using a window length of 100 sites and a step size of 25 sites, excluding sites with gaps.
Of all prior studies considered, only three amplified the complete hypervariable region I as defined herein [47–49]. However, the authors of two of these studies [47,49] elected not to submit their sequences for publication (e.g. in GenBank), though the latter manuscript comprised a detailed description of SNP positions. As a consequence, the only sequences presently available in GenBank that span the entire HVRI are those of Xu and Arnason (1996) (complete control region, N = 3), that of Horai et al. (1992) and of disputed quality (complete control region, N = 1) , and those of Nater et al. (2013) (≤ 385 bp, N = 28). All other published sequences span incomplete segments.
Though we identified further hypervariable stretches, their positions were inconsistent between Bornean and Sumatran sequences. Among those of Bornean origin, three especially hypervariable stretches were apparent, versus only two among those of Sumatran ancestry (Fig 2). Such inconsistencies could be the product of sample size, or sampling unrepresentative of the entire geographic range of each species. Having questioned the value of describing arbitrary hypervariable regions, and to avoid contributing further confusion to the literature, we chose not to formally define second or third hypervariable regions. Instead, we consider the value and utility of complete control-region sequences.
Nucleotide diversity in progressively longer control-region segments
The number of haplotypes, the number of polymorphic sites, the total number of mutations and the average number of nucleotide differences correlated positively with sequence length (Table 2). Only complete control-region sequences could discern the full total of 38 haplotypes and complete range of nucleotide diversity. Complete control-region sequences contained more than twice the number of polymorphic sites, total mutations and nucleotide differences (k) than any of the shorter segments previously studied.
Phylogeographic utility of predominantly analysed segments, versus the remainder of the control region
By inferring evolutionary relationships from the predominantly analysed segment (i.e. 385 bp), we inferred our sequences of Sumatran maternal origin derived ancestrally from two of the four subpopulations previously discerned on Sumatra  (Fig 3). These sequences were correctly resolved into distinct clades on a tree inferred only from all remaining control-region nucleotides. Thus, the remainder of the control region is phylogeographically informative, in that polymorphisms can distinguish at least two subpopulations. Similarly, sequences of Bornean origin were concordantly resolved into the same five different clades, representing different geographic subpopulations, using either the predominantly analysed segment (i.e. 323 bp) or all remaining control-region nucleotides (Fig 4). Polymorphic sites in the remainder of the Bornean orang-utan control region are therefore at least as informative as those in the predominantly analysed segment.
Evolutionary relationships for alignments of DNA sequences spanning a) 385 bp of the left domain of the Sumatran orang-utan mitochondrial DNA control region, as predominantly analysed in prior phylogenetic studies; b) all remaining nucleotides in the control region; c) all nucleotides of the control region. Sequences from orang-utans of known geographic origin are prefixed WAC (West Alas Cluster), LK (Langkat), NA (North Aceh) or BT (Batang Toru); novel sequences are prefixed 'S'; other published sequences are named with their accession numbers. Sequences from Bornean orang-utans and chimpanzees, used as outgroups, are not shown. Published sequences used in this inference are detailed in S1 Table.
Evolutionary relationships for alignments of DNA sequences spanning a) 323 bp of the left domain of the Bornean orang-utan mitochondrial DNA control region, as predominantly analysed in prior phylogenetic studies; b) all remaining nucleotides in the control region; c) all nucleotides of the control region. Sequences from orang-utans of known geographic origin are prefixed SA (Sabangau), GP (Gunung Palung), DS (Danau Sentarum), SE (Semenggoh), SN (Sangatta), KU (Kutai), TU (Tuanan), SL (Sungai Lading), SD (Sandakan) and DV (Danum Valley); novel sequences are prefixed 'S'; other published sequences are named with their accession numbers. Sequences from Sumatran orang-utans and chimpanzees, used an outgroup, are not shown. Published sequences used in this inference are detailed in S1 Table.
It is evident that there is considerable variation in orang-utans, in one or more additional hypervariable regions, in those areas of the control region that are typically ignored. Despite this, the hypervariable region I is broadly used as a representative sample of the control region in numerous taxa, on the understanding that it comprises the most abundant and useful of all polymorphisms for most phylogenetic applications. On this basis, prior orang-utan phylogenetic studies have demonstrably underestimated true control-region diversity, in choosing to focus only on the hypervariable region I or shorter segments thereof. It is probable that fewer haplotypes were characterised in such studies, as matrilineally unrelated individuals–that could have been distinguished by complete control-region sequences–may have been clustered together. Furthermore, as polymorphic sites in the remainder of the control region at least mirror the phylogeographic signals of the shorter and predominantly sequenced segments, it is possible that—with broader sampling—such sites could prove diagnostic of further geographic substructuring.
These observations are not intended as criticism of any previous study: several in particular have comprised especially monumental efforts to collect and analyse samples from across the entire geographic range of both species (i.e. [41,45,48]). Their analyses of short segments of the HVRI have significantly improved our understanding of orang-utan phylogeography, and all incorporated other markers–including at autosomal microsatellite and sex-chromosomal loci–from which to draw their conclusions. Indeed, in the case of orang-utans, it is unlikely that polymorphisms elsewhere in the control region would prove to be of taxonomic importance: studies at much higher resolutions, incorporating full mtDNA genome and nuclear-genome-wide variations, are already underway or published by other research groups, and are needed to explain the abundant idiosyncrasies observed in orang-utan taxonomy. Bornean orang-utan subspecies are known not to be reciprocally monophyletic based on mtDNA [41,45], though this may–to some extent–be the result of male dispersal or the unintentional translocation and reintroduction of non-native individuals [18,45,46]. More notably, through phylogenetic analysis of mtDNA, one Sumatran subpopulation has been shown to be more closely related to orang-utans on Borneo than to others on Sumatra . Nonetheless, complete orang-utan control-region sequences still have superior value: not least in their ability to distinguish distinct individuals, from different matrilines, that might otherwise share haplotypes at lower resolutions. Insertion/deletion polymorphisms are comparatively sparse in the remainder of the control region, as compared to the HVRI: in such instances, generating the complete control region provides substantially more data for phylogenetic analysis and reduces the potential for severe alignment errors. A consistent 15 bp deletion towards the end of the Bornean orang-utan control region could also prove useful, in facilitating their rapid discrimination from Sumatran sequences simply through visualization of product size via gel electrophoresis. Current methods utilize restriction enzymes applied to PCR products, and are therefore less feasible in field laboratories or in those in developing countries .
Our findings are relevant to all future control-region studies, however, irrespective of taxa. As mitochondrial genome structure is largely conserved, mutational hotspots can be expected throughout the complete control region in general [1,13]. Focusing only on short, arbitrarily defined and purported hypervariable segments has therefore potentially underestimated genetic diversity and population substructuring in a range of prior studies. This might occur when such segments are incorrectly considered to be broadly representative of complete control- region variation: when in fact, as in the case of orang-utans, useful polymorphisms are located elsewhere in the region that could be used to help distinguish distinct subpopulations or matrilines. The extent to which this might have occurred is unknown, as–in the clear majority of species–only purported hypervariable regions have been sequenced, rather than the complete control region: the abundance, value and utility of polymorphisms occurring elsewhere in such cases is therefore yet to be characterised. The issue might also predominate when aligning sequences of the ‘same’ region, produced by different studies, when each defines the length and position of this region differently. For successful phylogenetic analysis, such alignments must be cut to the length of the shortest sequence, encompassing only the narrowest definition. Consequently, genetic diversity within the author’s definition of the HVRI could be underestimated, having partially discarded the nucleotides. This will continue to be problematic, even when using complete control-region sequences: though we generated such sequences from orang-utans in our prior study, it was necessary to cut these to the shorter lengths of published sequences to infer their subspecies affiliations . In that scenario–discounting the benefits of avoiding amplification of numts, or extraneous human DNA–there was no superior value to complete control-region sequences, as most additional nucleotides were simply discarded.
The need for consistency in future studies is therefore paramount, yet segments such as the ‘hypervariable region’ cannot be consistently defined. Their lengths and positions must be determined through analysis of numerous sequences, using the sliding window approach adopted herein, or more complex change-point detection methods. As sequences are added and removed from the sample, particularly from genetically distinct individuals or populations, the lengths and positions of hypervariable and conserved regions will change. By contrast, the position of the control region is well-defined and conserved across taxa, and its length–excluding insertions and deletions–is typically uniform within species. Complete control-region sequences from multiple studies can therefore be aligned and analysed with no such ambiguity. Given the affordability and ease at which one can now amplify complete control-region sequences, even from highly degraded DNA, we therefore strongly advocate their use in all future studies where a shorter segment might otherwise have proven the locus of choice.
We acknowledge that the control region itself is simply a small segment of a larger, variable genome. In their analysis of 241 complete human mitochondrial genomes, for example, Coble et al. (2004) found that 18 haplotypes at the resolution of the HVRI and HVRII increased to 105 over the entire genome–though, having chosen only samples from European Caucasians, it is unclear to what extent these variations might be phylogeographically informative . Nonetheless, complete mitochondrial genome sequencing continues to be unaffordable for the vast majority of laboratories: such technology is as far in the distant future for cost-stretched Western research groups as it is for those in developing countries; particularly for researchers who study less charismatic fauna for which endeavours are comparatively poorly funded. For those without next-generation sequencing (NGS) capability, primer-walking and Sanger sequencing of mitochondrial genomes is possible, but still expensive–particularly when studies necessitate a large number of individual samples. Though the price of NGS continues to decline, the costs of library preparation are still insurmountable for many research groups, and thus, in both cases, only shorter segments such as the hypervariable region I are feasible for analysis.
We do not underestimate the value of nuclear markers, and we appreciate that–given limited funds–some might opt to sequence such a locus in addition to a short control-region segment, rather than devoting their additional budget to encompassing the complete control region. We do not consider these options to be equal, however. The costs of adding an independent marker are considerably greater, given that one requires more reagents and primers to perform additional PCRs. In contrast, it is not particularly more expensive to amplify and sequence the complete control region in a single PCR than to do so only for any shorter segment: the only supplementary cost is a second pair of sequencing reads. Given this minimal increase, and the fact that the procedure is no more complex or labour intensive, we suggest that complete control-region sequencing becomes the new de facto standard–and the new ‘bare minimum’–in favour of targeting any shorter segment thereof. Indeed, based on our findings, we find it hard to reconcile that any researcher would intentionally generate anything shorter.
Our view is shared by the forensic science community, for whom analysis of complete control-region sequences has long been the norm: guidelines to do so were introduced in 2007 , and–since 2013 –a leading journal, Forensic Science International: Genetics, has declined to accept submissions reliant only on partial control-region sequences . The zoological community is yet to catch up, however. Having compared and demonstrated their superior utility and potential in a non-human animal species, we hope to encourage the use of at least complete control-region sequences in future studies, where a short and arbitrary segment thereof might otherwise have proven the exclusive locus of choice.
We thank the Indonesian State Ministry of Research and Technology (RISTEK), the Indonesian State Ministry of Forestry (PHKA), the Indonesian Institute of Sciences (LIPI), the Indonesian Agency for Natural Resource Conservation (BKSDA) and the Tanjung Puting National Park Office (BTNTP) for granting their permission to undertake this research, and to our formal counterpart, the late S. H. Limin of CIMTROP at Universitas Palangka Raya, for his generous support of this work. We are indebted to A. E. Leiman OBE, S. Brend and P. A. Racey, who were instrumental in facilitating the fieldwork; to the Indonesian assistants who helped with sample collection; and to those zoos in the United States that contributed samples for analysis. We thank V. Städele and J. Hans for their comments on earlier versions of the text, and L. Howe and P. Khaitovich, who facilitated the work’s completion. GLB thanks Fapiao, his cat, for throwing up on the first handwritten draft of this manuscript, forcing substantial revisions that have greatly improved the text.
- Conceptualization: GLB.
- Data curation: GLB.
- Formal analysis: GLB.
- Funding acquisition: GLB BMFG.
- Investigation: GLB.
- Methodology: GLB.
- Project administration: GLB.
- Resources: GLB BMFG.
- Supervision: GLB.
- Validation: GLB.
- Visualization: GLB.
- Writing – original draft: GLB.
- Writing – review & editing: GLB.
- 1. Pakendorf B, Stoneking M. Mitochondrial DNA and Human Evolution. Annu Rev Genom Human Genet. 2005;6: 165–183.
- 2. Giles RE, Blanc H, Cann HM, Wallace DC. Maternal inheritance of human mitochondrial DNA. Proc Natl Acad Sci USA. 1980;77: 6715–6719. pmid:6256757
- 3. Paijmans JLA, Gilbert MTP, Hofreiter M. Mitogenomic analyses from ancient DNA. Mol Phylogenet Evol. 2013;69: 404–416. pmid:22705825
- 4. Bogenhagen D, Clayton DA. Mouse L cell mitochondrial DNA molecules are selected randomly for replication throughout the cell cycle. Cell. 1977;11: 719–727. pmid:560914
- 5. Robin ED, Wong R. Mitochondrial DNA molecules and virtual number of mitochondria per cell in mammalian cells. J Cell Physiol. 1988;136: 507–513. pmid:3170646
- 6. Cline SD. Mitochondrial DNA damage and its consequences for mitochondrial gene expression. Biochimica Biophysica Acta. 2012;1819: 979–991.
- 7. Yakes FM, Van Houten B. Mitochondrial DNA damage is more extensive and persists longer than nuclear DNA damage in human cells following oxidative stress. Proc Natl Acad Sci USA. 1997;94: 514–519. pmid:9012815
- 8. Brown WM, George M, Wilson AC. Rapid evolution of animal mitochondrial DNA. Proc Natl Acad Sci USA. 1979;76: 1967–1971. pmid:109836
- 9. Bandelt H-J, Kong Q-P, Richards M, Macaulay V. Estimation of mutation rates and coalescence times: some caveats. In: Bandelt HJ, Macaulay V, Richards M, editors. Human mitochondrial DNA and the evolution of Homo sapiens. Heidelberg: Springer; 2006. pp. 47–90.
- 10. Andrews RM, Kubacka I, Chinnery PF, Lightowlers RN, Turnbull DM, Howell N. Reanalysis and revision of the Cambridge reference sequence for human mitochondrial DNA. Nat Genet. 1999;23: 147–147. pmid:10508508
- 11. Greenberg BD, Newbold JE, Sugino A. Intraspecific nucleotide sequence variability surrounding the origin of replication in human mitochondrial DNA. Gene. 1983;21: 33–49. pmid:6301949
- 12. Horai S, Hayasaka K. Intraspecific nucleotide sequence differences in the major noncoding region of human mitochondrial DNA. Am J Hum Genet. 1990;46: 828–842. pmid:2316527
- 13. Stoneking M. Hypervariable sites in the mtDNA control region are mutational hotspots. Am J Hum Genet. 2000;67: 1029–1032. pmid:10968778
- 14. Vigilant L, Pennington R, Harpending H, Kocher TD, Wilson AC. Mitochondrial DNA sequences in single hairs from a southern African population. Proc Natl Acad Sci USA. 1989;86: 9350–9354. pmid:2594772
- 15. Hebert PDN, Cywinska A, Ball SL, deWaard JR. Biological identifications through DNA barcodes. Proc R Soc B. 2003;270: 313–321. pmid:12614582
- 16. Gonder MK, Oates JF, Disotell TR, Forstner MR, Morales JC, Melnick DJ. A new west African chimpanzee subspecies? Nature. 1997;388: 337–337. pmid:9237749
- 17. Changtragoon S. Wildlife Forensics in Thailand. In: Huffman JE, Wallace JR, editors. Wildlife forensics: methods and applications. Chichester: John Wiley & Sons Ltd; 2012. pp. 327–342.
- 18. Banes GL, Galdikas BMF, Vigilant L. Reintroduction of confiscated and displaced mammals risks outbreeding and introgression in natural populations, as evidenced by orang-utans of divergent subspecies. Sci Rep. 2016;6: 22026. pmid:26911345
- 19. Fernandes GA, Caparroz R. DNA sequence analysis to guide the release of blue-and-yellow macaws (Ara ararauna, Psittaciformes, Aves) from the illegal trade back into the wild. Mol Biol Rep. 2013;40: 2757–2762. pmid:23184047
- 20. Ishida Y, Georgiadis NJ, Hondo T, Roca AL. Triangulating the provenance of African elephants using mitochondrial DNA. Evol Appl. 2013;6: 253–265. pmid:23798975
- 21. Hsieh H-M, Lee JC-I, Wu J-H, Chen C-A, Chen Y-J, Wang G-B, et al. Establishing the pangolin mitochondrial D-loop sequences from the confiscated scales. Forensic Sci Int Genet. 2011;5: 303–307. pmid:20638354
- 22. Wu H, Wan Q-H, Fang S-G, Zhang S-Y. Application of mitochondrial DNA sequence analysis in the forensic identification of Chinese sika deer subspecies. Forensic Sci Int. 2005;148: 101–105. pmid:15639603
- 23. Sorenson L, McDowell JR, Knott T, Graves JE. Assignment test method using hypervariable markers for blue marlin (Makaira nigricans) stock identification. Conserv Genet Resour. 2012;5: 293–297.
- 24. Tzen JM, Hsu H-J, Wang M-N. Redefinition of hypervariable region I in mitochondrial DNA control region and comparing its diversity among various ethnic groups. Mitochondrion. 2008;8: 146–154. pmid:18248776
- 25. Miller KW, Budowle B. A compendium of human mitochondrial DNA control region: development of an international standard forensic database. Croat Med J. 2001;42: 315–327. pmid:11387646
- 26. Krings M, Geisert H, Schmitz RW, Krainitzki H, Paabo S. DNA sequence of the mitochondrial hypervariable region II from the neandertal type specimen. Proc Natl Acad Sci USA. 1999;96: 5581–5585. pmid:10318927
- 27. Meyer S, Weiss G, Haeseler von A. Pattern of nucleotide substitution and rate heterogeneity in the hypervariable regions I and II of human mtDNA. Genetics. 1999;152: 1103–1110. pmid:10388828
- 28. Parson W, Dür A. EMPOP—A forensic mtDNA database. Forensic Sci Int Genet. 2007;1: 88–92. pmid:19083735
- 29. Lutz S, Weisser HJ, Heizmann J, Pollak S. Location and frequency of polymorphic positions in the mtDNA control region of individuals from Germany. Int J Legal Med. 1998;111: 67–77. pmid:9541853
- 30. Imes DL, Wictum EJ, Allard MW, Sacks BN. Identification of single nucleotide polymorphisms within the mtDNA genome of the domestic dog to discriminate individuals with common HVI haplotypes. Forensic Sci Int Genet. 2012;6: 630–639. pmid:22436122
- 31. Badhan A, Eichstaedt CA, Almond NM, Knapp LA, Rose NJ. Analysis of full-length mitochondrial DNA D-loop sequences from Macaca fascicularis of different geographical origin reveals novel haplotypes. J Med Primatol. 2015;44: 125–136. pmid:25707924
- 32. Wilkinson-Herbots HM, Richards MB, Forster P, Sykes BC. Site 73 in hypervariable region II of the human mitochondrial genome and the origin of European populations. Ann Human Genet. 1996;60: 499–508.
- 33. Parson W, Bandelt H-J. Extended guidelines for mtDNA typing of population data in forensic science. Forensic Sci Int Genet. 2007;1: 13–19. pmid:19083723
- 34. Dunn JC, Shedden-González A, Cristóbal-Azkarate J, Cortés-Ortiz L, Rodríguez-Luna E, Knapp LA. Limited genetic diversity in the critically endangered Mexican howler monkey (Alouatta palliata mexicana) in the Selva Zoque, Mexico. Primates. 2013;55: 155–160. pmid:24276508
- 35. Seuanez HN, Evans HJ, Martin DE, Fletcher J. An inversion of chromosome 2 that distinguishes between Bornean and Sumatran orangutans. Cytogenet Cell Genet. 1979;23: 137–140. pmid:761478
- 36. Horai S, Satta Y, Hayasaka K, Kondo R, Inoue T, Ishida T, et al. Man's place in Hominoidea revealed by mitochondrial DNA genealogy. J Mol Evol. 1992;35: 32–43. pmid:1518083
- 37. Xu X, Arnason U. The mitochondrial DNA molecule of Sumatran orangutan and a molecular proposal for two (Bornean and Sumatran) species of orangutan. J Mol Evol. 1996;43: 431–437. pmid:8875856
- 38. Ryder OA, Chemnick LG. Chromosomal and mitochondrial DNA variation in orang utans. Journal of Heredity. 1993;84: 405–409. pmid:8409361
- 39. Zhi L, Karesh WB, Janczewski DN, Frazier-Taylor H, Sajuthi D, Gombek F, et al. Genomic differentiation among natural populations of orang-utan (Pongo pygmaeus). Curr Biol. 1996;6: 1326–1336. pmid:8939569
- 40. Muir CC, Galdikas BMF, Beckenbach AT. Is there sufficient evidence to elevate the orangutan of Borneo and Sumatra to separate species? J Mol Evol. 1998;46: 378–379. pmid:9541530
- 41. Warren KS, Verschoor EJ, Langenhuijzen S, Heriyanto , Swan RA, Vigilant L, et al. Speciation and Intrasubspecific Variation of Bornean Orangutans, Pongo pygmaeus pygmaeus. Mol Biol Evol. 2001;18: 472–480. pmid:11264398
- 42. Groves C. P. Primate Taxonomy. Washington, D.C.: Smithsonian Institution Press; 2001.
- 43. Brandon-Jones D, Eudey AA, Geissmann T, Groves CP, Melnick DJ, Morales JC, et al. Asian Primate Classification. Int J Primatol. 25: 97–164.
- 44. Singleton I, Wich S, Husson S, Stephens S, Utami Atmoko S, Leighton M et al. Orangutan Population and Habitat Viability Assessment: Final Report. Apple Valley, MN: IUCN/SSC Conservation Breeding Specialist Group; 2004
- 45. Arora N, Nater A, van Schaik CP, Willems EP, van Noordwijk MA, Goossens B, et al. Effects of Pleistocene glaciations and rivers on the population structure of Bornean orangutans (Pongo pygmaeus). Proc Natl Acad Sci USA. 2010;107: 21376–21381. pmid:21098261
- 46. Jalil MF, Cable J, Sinyor J, Lackman-Ancrenaz I, Ancrenaz M, Bruford MW, et al. Riverine effects on mitochondrial structure of Bornean orang-utans (Pongo pygmaeus) at two spatial scales. Mol Ecol. 2008;17: 2898–2909. pmid:18494768
- 47. Morrogh-Bernard HC, Morf NV, Chivers DJ, Krützen M. Dispersal Patterns of Orang-utans (Pongo spp.) in a Bornean Peat-swamp Forest. Int J Primatol. 2010;32: 362–376.
- 48. Nater A, Arora N, Greminger MP, van Schaik CP, Singleton I, Wich SA, et al. Marked population structure and recent migration in the critically endangered Sumatran orangutan (Pongo abelii). J Hered. 2013;104: 2–13. pmid:23077232
- 49. Rianti P, Perwitasari-Farajallah D, Sajuthi D, Pamungkas J, Nater A, Krützen M. Identification of Diagnostic Mitochondrial DNA Single Nucleotide Polymorphisms Specific to Sumatran Orangutan (Pongo abelii) Populations. Hayati. 2015.
- 50. Xu X, Arnason U. The mitochondrial DNA molecule of Sumatran orangutan and a molecular proposal for two (Bornean and Sumatran) species of orangutan. J Mol Evol. 1996;43: 431–437. pmid:8875856
- 51. Banes GL, Galdikas BMF, Vigilant L. Male orang-utan bimaturism and reproductive success at Camp Leakey in Tanjung Puting National Park, Indonesia. Behav Ecol Sociobiol. 2015;69: 1785–1794.
- 52. Soto-Calderón ID, Lee EJ, Jensen-Seaman MI, Anthony NM. Factors Affecting the Relative Abundance of Nuclear Copies of Mitochondrial DNA (Numts) in Hominoids. J Mol Evol. 2012;75: 102–111. pmid:23053193
- 53. Kearse M, Moir R, Wilson A, Stones-Havas S, Cheung M, Sturrock S, et al. Geneious Basic: an integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics. 2012;28: 1647–1649. pmid:22543367
- 54. Martin DP, Murrell B, Golden M, Khoosal A, Muhire B. RDP4: Detection and analysis of recombination patterns in virus genomes. Virus Evol. 2015;1: vev003–vev003. pmid:27774277
- 55. Anthony NM, Clifford SL, Bawe-Johnson M, Abernethy KA, Bruford MW, Jean Wickings E. Distinguishing gorilla mitochondrial sequences from nuclear integrations and PCR recombinants: Guidelines for their diagnosis in complex sequence databases. Mol Phylogenet Evol. 2007;43: 553–566. pmid:17084645
- 56. Edgar RC. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 2004;32: 1792–1797. pmid:15034147
- 57. Das R, Hergenrother SD, Soto-Calderón ID, Dew JL, Anthony NM, Jensen-Seaman MI. Complete Mitochondrial Genome Sequence of the Eastern Gorilla (Gorilla beringei) and Implications for African Ape Biogeography. J Hered. 2015;105: 846–855.
- 58. Librado P, Rozas J. DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics. 2009;25: 1451–1452. pmid:19346325
- 59. Darriba D, Taboada GL, Doallo R, Posada D. jModelTest 2: more models, new heuristics and parallel computing. Nat Meth. 2012;9: 772–772.
- 60. Guindon SXP, Gascuel O. A Simple, Fast, and Accurate Algorithm to Estimate Large Phylogenies by Maximum Likelihood. Syst Biol. 2003;52: 696–704. pmid:14530136
- 61. Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A, Hohna S, et al. MrBayes 3.2: Efficient Bayesian Phylogenetic Inference and Model Choice Across a Large Model Space. Syst Biol. 2012;61: 539–542. pmid:22357727
- 62. Perwitasari-Farajallah D. Discrimination of two species of Orangutans (Pongo sp.): A rapid protocol for rehabilitation centres and zoos. Biotropia. 2009;16: 65–70.
- 63. Coble MD, Just RS, O'Callaghan JE, Letmanyi IH, Peterson CT, Irwin JA, et al. Single nucleotide polymorphisms over the entire mtDNA genome that increase the power of forensic testing in Caucasians. Int J Legal Med. 2004;118: 137–146. pmid:14760490
- 64. Carracedo Á, Butler JM, Gusmão L, Linacre A, Parson W, Roewer L, et al. New guidelines for the publication of genetic population data. Forensic Sci Int Genet. 2013;7: 217–220. pmid:23375886