A mobile laboratory for ancient DNA analysis

Mobile devices for on-field DNA analysis have been used for medical diagnostics at the point-of-care, forensic investigations and environmental surveys, but still have to be validated for ancient DNA studies. We report here on a mobile laboratory that we setup using commercially available devices, including a compact real-time PCR machine, and describe procedures to perform DNA extraction and analysis from a variety of archeological samples within 4 hours. The process is carried out on 50 mg samples that are identified at the species level using custom TaqMan real-time PCR assays for mitochondrial DNA fragments. We evaluated the potential of this approach in museums lacking facilities for DNA studies by analyzing samples from the Enlène (MIS 2 layer) and the Portel-Ouest cave (MIS 3 deposits), and also performed experiments during an excavation campaign at the Roc-en-Pail (MIS 5) open-air site. Enlène Bovinae bone samples only yielded DNA for the extinct steppe bison (Bison priscus), whereas Portel-Ouest cave coprolites contained cave hyena (Crocuta crocuta spelaea) DNA together, for some of them, with DNA for the European bison sister species/subspecies (Bison schoetensacki/Bb1-X), thus highlighting the cave hyena diet. Roc-en-Pail Bovinae bone and tooth samples also contained DNA for the Bison schoetensacki/Bb1-X clade, and Cervidae bone samples only yielded reindeer (Rangifer tarandus) DNA. Subsequent DNA sequencing analyses confirmed that correct species identification had been achieved using our TaqMan assays, hence validating these assays for future studies. We conclude that our approach enables the rapid genetic characterization of tens of millennia-old archeological samples and is expected to be useful for the on-site screening of museums and freshly excavated samples for DNA content. Because our mobile laboratory is made up of commercially available instruments, this approach is easily accessible to other investigators.


Introduction
Over the last twelve years, the field of ancient DNA has experienced deep changes that come both from technological advances and the attention paid to a variety of DNA sources. On the one hand, next generation DNA sequencing methods enabled deciphering complete animal and human genomes [1][2][3], including the genome of a human lineage not known from the hominin fossil record [4]. On the other hand, the survey of archeological specimens revealed additional and sometimes unexpected sources of ancient DNA such as the avian distal feather components [5] and eggshell [6], dental calculus [7], mollusk shell [8], sediments [9], with especially high yield from hair shafts [10], coprolites [11], and the petrous bone [12]. As a complementary approach, new experimental procedures for DNA extraction [13][14][15] or sampling [16,17] have been described.
Despite these advances, days to weeks are required from the time a sample has been excavated to the delivery of DNA data. This contrasts with the increasing use of physics methods to perform on-field analysis of archeological material, which enable for example to characterize the pigments of rock art paintings [18][19][20].
For studies on extant DNA, great attention is paid to on-field analysis to perform medical diagnostics at the point-of-care, for forensic sciences, food testing, environmental monitoring, and detection of biothreat agents (see [21,22] for review). Progress accomplished thanks to such studies for rapid DNA analysis, together with the increasing need to avoid the dissemination of museum collections and the observation that freshly excavated fossils are best for ancient DNA analysis [23,24] call for the evaluation of the on-field approach for ancient DNA studies.
Because ancient DNA is most often present in trace amounts in archeological samples, contamination by modern or previously amplified DNA is a matter of concern. With this limitation in mind, we attempted to obtain the proof-of-concept for an ancient DNA mobile laboratory by focusing on DNA from extinct species or species no longer present in the locations where our studies were performed. Also, we used the method of real-time PCR because it avoids the post-PCR processing of amplified DNA. We report here on the use of a mobile laboratory consisting of commercially available devices for DNA extraction and amplification, and of TaqMan probes [25] that we designed for real-time PCR analysis of the mitochondrial DNA of six animal species. We conducted studies on samples from three Paleolithic sites: the Enlène cave, the Portel-Ouest cave, and the Roc-en-Pail open-air site (Fig 1).
Enlène (450 m above sea level (asl)), together with the Tuc d'Audoubert and the Trois-Frères caves, is part of the Volp cavern system [26], a property of the Bégouën family located near Montesquieu-Avantès (Ariège, France). By contrast to the Tuc d'Audoubert and the Trois-Frères caves, which are famous for Magdalenian rock art, Enlène contains almost no such art [27,28]. Enlène is nevertheless a major site which provided clues to Magdalenian daily life. Interest for archeological remains in Enlène began in 1860 with excavations performed by canon Jean-Jasques Pouech, was strongly reinforced in 1911 when Jacques Bégouën found a carved propulsion device, and furthermore in 1914 when Jacques and brothers Louis and Max discovered the Trois-Frères cave and its subterranean connection with Enlène [28]. Excavations carried out in the 1920's and 1930's by Louis Bégouën [29], then from 1970 to 1990 by Robert Bégouën, Jean Clottes and coworkers revealed 30,000 years of human occupation beginning with the Gravettian period [30]. Notably, these studies yielded abundant Magdalenian tools and portable art pieces. They also provided evidence for the sedentary settlement of magdalenian people in the deep cave sector (the Salle-du-Fond) which displays paving stones, bowl-shaped fire structures, and remains of hunted animal species. A palynological study carried out on this level revealed the presence of on open landscape, dominated by Poaceae and Asteraceae, with steppe species such as Artemisia, Helianthemum, Armeria, and Rubiaceae, growing in a cold and dry climate [31]. In the present study, we performed the DNA analysis of Bovinae bone remains excavated from the 18,000-year-old (Marine Isotope Stage 2 (MIS2)) Magdalenian layer of the Enlène Salle-du-Fond. This was done in the laboratory of the Bégouën Museum in Montesquieu-Avantès.
The cave of Le Portel (490 m asl) is also located in Ariège, near the village of Loubens, about 30 km away from Enlène. The site, a property of the Vézian family, consists of two caves which were connected to each other until a rockfall sealed the communication during MIS 5. The largest cave (eastern entrance) displays Magdalenian and Gravettian rock art pictures that were discovered in 1908 [32]. The Portel-Ouest cave (western entrance), discovered some ten years later, contains a rich, more than 5-m thick archeological sequence. Excavations carried out from 1949 to 1987 in the Portel-Ouest cave by Joseph and Jean Vézian disclosed 21 layers that yielded Neandertal bones and teeth [33], as well as lithic tools and animal remains spread over a time period of about 100,000 years, from MIS 5 to MIS 2 [34,35]. In this study, we analyzed cave hyena coprolites from layers D and B1 of the Portel-Ouest cave, dating back to MIS 3. Occupation by the cave hyena in these levels is well documented by the presence of numerous skeletal remains, coprolites and toothmarks on large mammal remains [36,37]. Layer D yielded Mousterian tools [38] and micro and macrofaunal remains that are consistent with the warming and dampening phase [39,40] of a cold climate, associated with the presence of a semi-open landscape [41]. Layer B1 yielded Upper Paleolithic tools. Faunal remains in this layer indicate a shift from a cold steppe to a partially forested environment [39,40], and the pollen record from north Pyrenean sites for this time period points to the dominance of steppe taxa with a progressive increase of forested taxa with the altitude [41]. Layers D and B1 correspond to the Middle to Upper Paleolithic transition. The archeological remains of Le Portel are stored in the Centre Européen de Recherches Préhistoriques (CERP, Tautavel, France) were we performed the DNA studies.
Roc-en-Pail (Chalonnes-sur-Loire, Maine-et-Loire; 30 m asl) is an open-air site from Western-Central France that is shared between several owners. It is located on the left bank of the Layon River. The site was discovered during the winter of 1870-1871 on the fringes of a quarry supplying lime kilns. Fossil fauna, mainly reindeer, and flint tools were unearthed at the foot of the small Devonian limestone hill exploited by the quarry [42]. The site was excavated by Dr. Gruet from 1943 to 1956 and shortly reopened in 1969. Gruet's fieldwork was very briefly described in several preliminary publications, and a synthetic description of the stratigraphy was later published [43]. The site was additionally described in studies devoted to sedimentology [44] and the relative chronology is only known from pollen studies [45]. In 2014, new excavations were launched with the aim to provide a comprehensive stratigraphy and establish the chronology. With almost 5-m thick deposits and numerous archeological layers, Roc-en-Pail is the longest sequence described so far for the Middle Paleolithic in Central Western France. Lithic industries are associated with faunal and macrofaunal remains in all layers [46]. Seventeen OSL samples are under processing so the preliminary chronology relies only on geomorphology, fauna and industry record. The bulk of the samples analyzed for DNA content come from stratigraphic unit 401, a sandy loam from the first phase of slope deposits potentially associated with intra-MIS 5 climate degradation, either the MIS 5d or 5b. Mixed fauna recovered in this unit is consistent with the mosaic environments described in deposits from the Early Weichselian in Northern France [47]. It is associated with a Middle Paleolithic industry characterized by recurrent Levallois debitage. The DNA studies of Bovinae and Cervidae remains were performed in a house next to the site during the 2018 excavation campaign.

Archeological specimens
Enlène cave. We studied 11 archeological samples from the Enlène cave. The samples, registered under numbers Enlène 6170 to 6180, are stored in a permanent repository in the Laboratoire de Préhistoire de Pujol, 09200 Montesquieu-Avantès, France. The samples are accessible by others in this private repository upon request to Robert Bégouën at the address indicated above. The permit (reference: MV/FM/MB/2015/20302) obtained for all aspects of the study was delivered by the Service Régional de l'Archéologie (32 rue de la Dalbade, BP811, 31080 Toulouse cedex 6 France), and approved by the collection owner (Robert Bégouën).
Portel-Ouest cave. We studied 14 archeological samples from the Portel-Ouest cave. The samples, registered under numbers Portel T1 to T14, are stored in a permanent repository in the Museum of Prehistory reserves, Établissement Publique de Cooperation Culturelle-Centre Européen de Recherches Préhistoriques (EPCC-CERP), 1 avenue Léon-Jean Grégory, 66720 Tautavel, France. The samples are accessible by others upon request at the address indicated above. The permit (reference: 20170328) obtained for all aspects of the study was delivered under the control of the Service Régional de l'Archéologie Occitanie (5 rue Salle-l'Évêque, 34000 Montpellier, France), the collection owner (Régis Vézian), and the administrator of EPCC-CERP.
Roc-en-Pail. We studied 30 archeological samples from the Roc-en-Pail site. The samples are registered under numbers Roc-en-Pail 4, 10,11,12,18,29,32,33,37,44,49,54,75,118,181,182,196,206,231,260,297,318,321,337,398,409,410,464,528,529. They are temporarily stored in the research laboratory of the archeological excavation permit holder (Sylvain Soriano, MSH Mondes, UMR7041, 21 allée de l'université, 92000 Nanterre, France). After a period which may not exceed five years after the excavation, the archaeological remains, including the specimens described in this paper, will be deposited and accessible by others in a public repository of the Pays-de-la-Loire region. The location of the permanent repository will be available upon request to the Service Régional de l'Archéologie Pays-de-la-Loire (1 rue Stanislas Baudry, BP 63518, 44035 Nantes Cedex 1, France). The permits (references: 134/2016, 193/2017, 556/2018) for all aspects of the study were delivered by the Prefecture de la region des Pays-de-la-Loire.

Overview of experimental procedures
The equipment used for DNA studies was carried to museum and archeological sites in the luggage compartment of a mid-size car. In addition to the machines and kits described below that were used for DNA extraction and analysis, small equipment and consumables consisted of tubes, pipettes, pipette filter tips, disposable clothes, surgical blades, Petri dishes, aluminum foil, bench coat protector, biohazard bags, buffers and ultrapure distilled water. The full list of laboratory material is available from the authors. Temperature-sensitive reagents were transported in a cool box and stored at -20˚C upon arrival.
Investigators responsible for DNA studies (JME, JU) wore masks, hair nets, disposable lab coats and gloves. Working surfaces were covered with versi-dry protection paper sheets (Nalge Nunc; Rochester, NY, USA) that were changed after each experiment. All material related to the dissection of archeological samples was changed between samples. There was no post-PCR processing of the samples in the museum and archeological sites. Wastes were brought back to the laboratory.

DNA extraction
For bone samples, we scraped off the outer surface using a single-use surgical blade to delineate a sampling area (S1A and S1B Fig). Then, a clean blade was used to recover bone powder from which DNA extraction was performed. For coprolites, the cortex was removed to allow recovery of material from the coprolite core (S1C and S1D Fig). The samples recovered from coprolites ranged from a fine powder to small granules. When granules were obtained, the sample was crushed between two aluminum foils. For bone and coprolites samples, the powder was transferred into a 2-ml Eppendorf tube (Eppendorf, Hamburg, Germany) until filling the bottom of the tube (S1E Fig). This amount of material corresponds to 30-50 mg. Five hundred μl of DNA extraction buffer (0.25 M EDTA, 10 mM Tris-EDTA (pH 8), 0.2% N-lauryl-sarcosyl, 200 μg/ml proteinase K (Thermo Fisher Scientific (Walthman, MA, USA)) were added to each experimental or mock sample, and the tubes were incubated 2-16 h at 42˚C under constant agitation (1,200 rpm) in an Eppendorf thermomixer. The tubes were then centrifuged (5,000 g; 5 min) in an Eppendorf microcentrifuge, and the supernatant was transferred into an Amicon ultra-0.5 ml 30 kDa centrifugal filter unit (Millipore; Burlington, MA, USA). The filter unit was centrifuged (14,000 g; 7 min), rinsed 4 times by adding 500 μl of ultrapure DNase/ RNase-free distilled water (Thermo Fisher Scientific) and centrifugation (14,000 g; 7 min), and the DNA extract was recovered as a~50-μl sample volume by centrifugation (1,000 g; 1 min) of the filter unit in the reverse position. The extract was further purified using Qiagen minelute PCR purification kit (Qiagen; Venlo, Netherlands) according to manufacturer's instructions, and eluted in 50 μl of 10 mM Tris, pH 8.

Real-time PCR analysis
Real-time PCR analysis was carried out using TaqMan MGB probes obtained from Thermo Fisher Scientific. Primers and probes for the TaqMan assays (Table 1) were designed with the help of Primer Express software 3.0.1 (Thermo Fisher Scientific) or the custom assay design tool available on the Thermo Fisher Scientific web site using reference sequences for the mitochondrial genomes of Bison priscus (NC_027233), Bison schoetensacki (NC_033873), Rangifer tarandus (NC_007703), Cervus elaphus (NC_007704), and Crocuta crocuta spelaea (NC_020670). Real-time PCR was carried out in a 20-μl reaction volume containing 10 μl of 2X TaqMan fast advanced master mix (Thermo Fisher Scientific), 900 nM of forward and reverse primers, 250 nM of TaqMan probe, and water (PCR blank), mock or DNA extracts. The TaqMan fast advanced master mix contains dUTP (instead of dTTP) and uracil-N-glycosylase to prevent from carryover contamination from one experiment to another. For each

Kinetic analysis of DNA recovery
These experiments were performed in the Paleogenomic and Genetic analysis platform of the Musée de l'Homme. This platform has a clean room facility for pre-PCR steps carried out on ancient DNA, and modern DNA laboratories for PCR and post-PCR steps.
To set up a rapid DNA extraction procedure, we used the methods described above to analyze two samples previously published: a steppe bison rib [48], and a cave hyena coprolite [11]. Briefly, 300 mg of bone or coprolite powder was recovered using a single-use surgical blade, divided in two 150 mg samples that were transferred into 2-ml Eppendorf tube, and 1.5 ml of DNA extraction buffer was added. The samples were incubated under constant agitation in an Eppendorf Thermomixer, and aliquots of 500 μl were retrieved after a 1-h, 2-h, or 16-h (sample 1) or a 1-h, 3-h, or 16-h (sample 2) incubation time. DNA was then extracted using Amicon and Qiagen columns and eluted in 50 μl of 10 mM Tris, pH 8. Sizing of the DNA recovered in extracts was performed by electrophoresis using the LabChip GX touch nucleic acid analyzer (PerkinElmer; Waltham, MA, USA). TaqMan assays for steppe bison, cave hyena, and red deer DNA were performed using the Mic PCR cycler.

DNA sequencing
To confirm that correct species identification was achieved using TaqMan assays, a subset of DNA extracts were further analyzed by DNA sequencing. In order to avoid dissemination of amplified DNA in museum and archeological sites lacking an appropriate laboratory, these studies were performed in the Paleogenomic and Genetic analysis platform of the Musée de l'Homme.
Aliquots (0.3 to 0.9 μl) of the DNA extracts were PCR-amplified in a 50-μl reaction volume containing 300 nM of forward and reverse primers (Table 1), 200 μM dNTP, 2.5 mM MgCl 2 , 5 μl of GeneAmp AmpliTaq Gold DNA polymerase buffer II, and 2.5 U of AmpliTaq Gold DNA polymerase (Thermo Fisher Scientific). To disclose the DNA damages expected for authentic ancient DNA, uracil-N-glycosylase was not present in the PCR mix used for these experiments. PCR procedures consisted in an enzyme activation step (95˚C, 10 min), followed by a single round of 45 PCR cycles (95˚C, 15 s; 53˚C, 20 s; 70˚C, 1 min) performed in a Veriti thermal cycler (Thermo Fisher Scientific). The full reaction volume was loaded onto an 8% polyacrylamide gel stained with Sybr Green I (Thermo Fisher Scientific). PCR blanks and mock extracts always failed to yield any amplification products. PCR amplicons were eluted from the gel, recovered in 10 mM Tris (pH 8), and ligated to Illumina (San Diego, CA, USA) adapters. Libraries of DNA fragments were generated using the Illumina TruSeq Nano DNA LT sample kit following the manufacturer's recommendations, except that 5 PCR cycles were performed after the ligation step. Illumina sequencing of pooled amplified libraries was performed on the MiSeq platform by Fasteris (Geneva, Switzerland). The sequence reads were quality filtered, the adapter sequences were removed using Fastx-Toolkit 0.0.13 (http:// hannonlab.cshl.edu/fastx_toolkit/download.html) and Cutadapt (version 2.7) [49] with a quality score of 20 and above, and minimum read length of 50 nucleotides. Sequence reads were aligned using CodonCode Aligner v.8.0 (http://www.codoncode.com/aligner/), and DNA sequence logos were generated using the seqLogo R package [50]. The DNA reads of this study have been deposited at Dryad (doi:10.5061/dryad.8gtht76jz). For each sample, the consensus sequence derived from DNA reads was used for species identification by BlastN alignment to the sequences recorded in the GenBank nr/nt database. In this study, we analyzed Pleistocene remains of herbivore (Bovidae, Cervidae) and carnivore (Hyaenidae) species. Late Pleistocene Eurasian specimens of the cave hyena display some particular features, including different body size and proportions as compared to the most closely related extant species, the African spotted hyena (Crocuta crocuta) [51,52]. Consequently, the species/subspecies status of the cave hyena is debated and has been alternatively referred to as Crocuta crocuta, Crocuta spelaea, or Crocuta crocuta spelaea. However, genetic studies revealed that the Eurasian cave hyena and the extant spotted African hyena are so closely related that the cave hyena should be considered as a chronospecies rather than a distinct species or even subspecies [11,53]. We therefore refer below to cave hyena remains as Crocuta crocuta samples.

Validation of experimental procedures
To demonstrate the potential of our method to perform ancient DNA studies, we analyzed two samples previously published: a steppe bison rib (SGE2) from which a complete mitochondrial genome sequence for the steppe bison has been deciphered using shotgun DNA sequencing [48]; and a cave hyena coprolite (CC8) from which shotgun DNA sequencing yielded a complete cave hyena and a partial red deer (Cervus elaphus) mitochondrial genome sequence [11]. For each sample, we performed two experiments: one experiment aiming to compare the DNA extracts recovered after a 1-h, 3-h, or 16-h incubation period; and one experiment aiming to compare extracts recovered after a 1-h, 2-h, or 16-h incubation period. Fig 3A and 3B show that for the Bison priscus bone sample, there were no clear-cut differences between the amplification plots for extracts obtained after a long (2 h or more) or a short (1 h) incubation period. For the Crocuta crocuta coprolite, C T values for DNA amplification were similar for extracts obtained after a 3-to 16-h (Fig 3C) or a 2-to 16-h ( Fig 3D) incubation period, and slightly but systematically lower than C T recorded after a 1-h incubation period. The data therefore indicate a much better yield of DNA from the coprolite after a long than a short (1 h) incubation period, and no improvement in yield for incubation times longer than 2 h. Importantly, the TaqMan assays carried out on the coprolite extracts made it possible to detect DNA originating from the defecator (Crocuta crocuta) and from its diet (Cervus elaphus). We also characterized the DNA extracts by analyzing the size of the native (i.e. unamplified) DNA fragments recovered using our extraction procedure. As expected for ancient DNA, the bulk of the material recovered corresponded to DNA fragments shorter than 100-bp (Fig 3E and 3F). Overall, the data reported in Fig 3 indicate that our extraction procedure and TaqMan assays are suitable for analyzing ancient DNA, and that a 2-h incubation time is long enough for ancient DNA recovery from bone and coprolite samples. Based on these data, we decided to perform incubation times lasting 2 to 16 h for DNA extraction.

Analysis of Enlène cave samples
The first floor of the Bégouën museum (Montesquieu-Avantès, Ariège, France) includes a room for archiving archeological and paleontological remains excavated from the Volp caverns, and a laboratory in which no ancient DNA studies have ever been performed and where we installed our mobile laboratory. We analyzed 11 Bovinae bone samples originating from the Magdalenian layer of the Salle-du-Fond. The samples were studied using TaqMan assays for Bison priscus, Bison schoetensacki, and Bos primigenius mitochondrial DNA. Fig 4 shows real-time PCR data for the Enlène 6178 specimen. Using the Bison priscus probe, amplification was recorded for all three replicate samples, with C T values ranging between 30 and 34 cycles, depending on the amount (from 0.9 to 0.1 μl) of DNA extract introduced in the assay. These results indicate robust amplification data for the detection of Bison priscus DNA. Fig 4 also shows that, using the same DNA extract, the Bison schoetensacki assay yielded either no amplification or a very faint signal, with a C T of 39 cycles for the largest amount of extract being used. The large difference between the C T values (9 PCR cycles) recorded with the two probes for the same DNA amount indicates efficiencies that differ by more than two orders of magnitude to detect DNA in the extract, thus validating the design of the Bison priscus assay. As a whole, for the 11 Enlène cave samples tested, we only obtained evidence for Bison priscus DNA, with a total of six positive samples (S2 Fig, S1 Table). For one sample (E6179), two DNA extracts were prepared and studied on separate days, and both of them were positive for Bison priscus DNA content.
To confirm that correct species identification was achieved using the TaqMan assay, a subset of Enlène samples were analyzed using high-throughput sequencing of PCR amplicons. The DNA sequencing approach also offers the opportunity to demonstrate that the extracts contain genuine ancient DNA. Indeed, amplicons obtained from ancient DNA display typical patterns of nucleotide misincorporations, that mostly consist in transitions, especially C/G-to- T/A transitions [54]. Fig 5 shows that for each of the Enlène samples analyzed, the consensus sequence perfectly aligned with the Bison priscus reference sequence, and that the patterns of nucleotide misincorporations fit with those expected for PCR fragments obtained from ancient DNA. Thus, the sequence data support the conclusions derived from the TaqMan studies.

Analysis of cave hyena coprolites from the Portel-Ouest cave
We performed the analysis of coprolites from the Portel-Ouest cave in the CERP of Tautavel. CERP, a building contiguous to the Musée de Préhistoire de Tautavel, houses a variety of devices for the analysis of archeological material, but no DNA studies have been performed previously in this laboratory. MIS 3 layers D (end of Middle Paleolithic) and B (Upper Paleolithic) of the Portel-Ouest cave yielded several dozens of Crocuta crocuta coprolites, of which 14 (seven from each layer) were selected for our studies.
To test for DNA preservation in the cave hyena coprolites, we analyzed the DNA extracts using the Crocuta crocuta mitochondrial genome TaqMan assay. As shown in Fig 6A, Fig, S1 Table), including 7 samples from layer B1, and a single sample from layer D. This indicates better DNA preservation in the Upper than in the Middle Paleolithic layer.
To gain some information on the cave hyena diet, we next analyzed all 8 coprolites positive for Crocuta crocuta DNA with TaqMan assays for five Cetartiodactyla species, including two Cervidae and three Bovinae species. Whereas no evidence was obtained for the presence of Cervus elaphus, Rangifer tarandus, Bison priscus or Bos primigenius in these Portel cave coprolites, amplification of Bison schoetensacki DNA was demonstrated in two of them (Fig 6B, S1  Table). Considering the close genetic proximity between Bison schoetensacki and Bison bonasus [55], additional PCR studies were performed to characterize by DNA sequencing the fragments amplified with the Bovinae primers. In the Bison schoetensacki mitochondrial genome sequence, the sequence located between the TaqMan primers displays a C residue at a position (nucleotide 15,273 in the reference sequence) where a T residue is recorded in all extant and extinct specimens of the Bison bonasus Bb2 [56,57] lineage analyzed so far. The consensus sequences obtained for coprolites DNA amplified using Bovinae primers display the distinctive Bison schoetensaki C residue (Fig 7). At other positions, abundant C to T substitutions guarantee that amplification was initiated from ancient DNA. The detection of Bison schoetensaki DNA in two coprolites, one (T2) recovered from layer D, and one (T14) from layer B1, indicates that this Bovinae species was present by the end of the Middle Paleolithic and the beginning of the Upper Paleolithic (i.e. about 40,000 years ago) in the vicinity of the Portel cave.

Analysis of Roc-en-Pail samples
The fieldwork carried out at Roc-en-Pail gave us the opportunity to perform ancient DNA studies on freshly excavated samples. Based on paleontological criteria, the bulk of excavated remains are ascribed to Bovinae and Cervidae specimens. As usual for Bovinae remains, species identification was hardly feasible using morphometric data, so most samples were referred to as Bos/ Bison remains. These samples were systematically analyzed using the three Bovinae TaqMan assays. On the other hand, the Cervidae remains were expected to correspond to reindeer specimens. They were nevertheless all studied using the Rangifer tarandus and Cervus elaphus mitochondrial genome TaqMan assays. Fig 8A shows results obtained for two Bovinae samples using the Bison schoetensacki assay. C T values with this assay were 32.7 (ReP529) and 33.5 (ReP32) PCR cycles. Of the 16 Bovinae samples surveyed, 4 (3 bone fragments and a tooth sample) yielded DNA that could be amplified using the Bison schoetensacki assay (S4 Fig), but none of them provided evidence for the presence of Bison priscus or Bos primigenius DNA (S1 Table). In line with studies performed on Le Portel cave samples (see above), we checked by DNA sequence analysis the correct identification of the DNA fragments for two samples.  Turning to Roc-en-Pail Cervidae remains, Fig 8B shows examples of samples for which successful DNA amplification was achieved using the Rangifer tarandus assay. For the whole series of Cervidae remains tested for DNA content (14 samples), TaqMan analysis disclosed the presence of Rangifer tarandus DNA in 7 bone fragments (S7 Fig, S1 Table), whereas Cervus elaphus DNA was never detected. ReP 18 specimen was studied using two DNA extracts obtained on separate days, and both of them indicated the presence of Rangifer tarandus DNA. Sequencing analysis confirmed the correct identification of Rangifer tarandus DNA using our TaqMan assay (S5B Fig).

Discussion
This work evaluated the feasibility of using a mobile platform to perform ancient DNA analysis in museums lacking facilities for genetic studies, and as a complementary approach to archeological fieldwork during an excavation campaign. The samples surveyed originate from cave and open-air sites spanning a 60,000-year time period, from MIS 5 (Roc-en-Pail) to MIS 2 (Enlène). Successful real-time DNA analysis was achieved for several animal species using a variety of remains, including coprolite, bone and tooth samples.
Several lines of evidence support the notion that our method provides reliable data, notably that contamination and specificity issues were adequately addressed. First, negative controls (i. e. PCR blanks and mock extracts) always failed to yield an amplification signal. Second, the species for which DNA amplification was obtained are either extinct or no longer present in the vicinity of the sites where the studies were conducted. Third, when a sample evaluated positive for DNA content was studied through two extracts, the second experiment corroborated the first one. Fourth, DNA sequencing carried out using the TaqMan assay primers or different primers confirmed that correct species identification had been achieved with the real-time PCR assays.

PLOS ONE
Our mobile platform consists of two devices for DNA extraction (heater-mixer, minicentrifuge) and of a compact real-time PCR machine. In order to make easily accessible the on-field approach, we paid special attention to set-up a platform that only includes commercially available devices. The full set of instruments and pipettes of our mobile platform amounts to 15,000 €, and the cost to perform DNA extraction and real-time PCR analysis (4 replicates) with one TaqMan assay is 15 €. The methods described here allow a single investigator to perform DNA extraction and analysis of 12 samples within 4 hours. The procedures for sample preparation and DNA extraction were designed to limit the number of instruments in the platform and the working time. Thus, instead of a drilling machine we used single-use surgical blades to scrap the bone surface and recover bone powder. This sampling procedure is consistent with the recommendation to retrieve material from the compact bone surface for best endogenous DNA yield [17]. To speed up the DNA extraction process and limit sample handling, we did not perform a bleach wash nor repeated incubation steps [14,15]. Instead, we used a single incubation step procedure and a similar extraction protocol for bone, tooth and coprolite samples. Thus, there is room for improving the on-field procedures for DNA extraction, while keeping in mind that sample handling and working time should be reduced as much as possible. For studying samples from animal species still present in the site environment, bleaching of the archeological material and implementing the platform with a portable clean room is recommended. For human samples, it would not be reasonable to perform on-field ancient DNA studies.
The samples we screened originate from sites that have not been studied previously for ancient DNA. In Enlène, we focused on Bovinae remains from the deep cave sector that was inhabited by Magdalenian people 18,000 years ago. We detected Bison priscus DNA in several bone samples, which provides a molecular support to confirm the notion that Magdalenian individuals from Enlène hunted the extinct steppe bison. For the Portel-Ouest cave, the analysis of cave hyena coprolites indicated better DNA preservation in the~35,000-year-old (Upper Paleolithic) than in the~40,000-year-old (Middle Paleolithic) layer. Moreover, we were able to detect Crocuta as well as Bison DNA in some coprolites, indicating that our approach is useful to obtain genetic data on a carnivore and its diet. Interestingly, the bison DNA present in coprolites points to a species different from the one detected in Enlène. Paleontological data argue for the presence of Bison priscus mediator remains in the Portel-Ouest deposits, especially in the~45,000-year-old F layer [35]. Thus, comparison of the paleontological and genetic data suggests the presence of several bison species in the cave paleoenvironment, possibly at different time periods (see below). Finally, Roc-en-Pail challenged our method for the analysis of MIS 5 (i.e. more than 71,000-year-old) samples from an open-air site. While the presence of reindeer remains indicates good agreement between the molecular and archeozoological data, the bison species disclosed at Roc-en-Pail by genetic studies was not fully anticipated and extends the observation made on the Portel cave samples.
Recent studies attempted to elucidate the descent of the European bison or wisent (Bison bonasus) by genomic analysis of remains sampled across Eurasia. Their complete mitochondrial genome sequences revealed two clades [55][56][57]. One clade, referred to as Bb2 by Massilani et al. [56], includes the extant wisent as well as all historical and Holocene specimens and part of the Pleistocene ones. The other clade only includes Pleistocene specimens. The status of this clade is debated. It has been named clade Bb1 or X by the two groups of investigators who consider it as an extinct Bison bonasus lineage [56,57]. We proposed that this clade corresponds to the extinct woodland bison (Bison schoetensacki) but failed to convince others who rather adhere to the notion that Bison schoetensacki became extinct during the Middle Pleistocene [58]. Whatever the exact status of the intriguing Bison schoetensacki/Bb1-X clade, in the present studies we recorded it for time periods and paleoenvironments that fit with previous expectations [56,57]. Thus, it was present in the MIS 5 Roc-en-Pail and in the MIS 3 Portel-Ouest layers. Mixed faunal record in these layers suggest the presence of non-continuous deciduous forest as well as a steppe environment that hosted "non-analogue" faunal communities as compared to those of present day [59]. By contrast, in the Enlène Magdalenian habitat (MIS 2 glacial period), we recorded Bison priscus specimens.
In conclusion, this work demonstrates the feasibility to perform ancient DNA analysis under a variety of working condition, including nearby an excavation site. The real-time PCR approach we used provides ancient DNA data within a few hours without on-site post-PCR processing of the samples, a prerequisite to avoid the dissemination of amplified DNA in sensitive environments. The robustness of the data was corroborated by subsequent DNA sequencing analysis. With the ongoing progress of methods for DNA analysis, including portable sequencing devices [60], we anticipate great interest in the near future for on-field ancient DNA studies.