Species identification of Late Pleistocene bat bones using collagen fingerprinting

Bats form the second most diverse mammalian order (Chiroptera), after rodents, and vary widely in their physiology and ecology. Those species that live in temperate climates are generally insectivorous and nocturnal or crepuscular, sheltering in tree hollows, caves, or buildings during the day. They are potentially valuable ecological indicators, due to their dependence on suitable roosting sites and arthropod food, both of which are commonly affected by human activities. Identification of bats from ancient assemblages that are found in caves could therefore provide useful data for palaeoenvironmental reconstructions and show the effect of habitat loss. Here, we apply the recently developed approach of collagen fingerprinting by soft ionisation mass spectrometry to the identification of ancient bat remains in an archaeological assemblage from Pin Hole Cave (Derbyshire, England). Our results show that a simple set of markers can distinguish all seven genera of bats known to be present in either modern or ancient Britain (Myotis, Nyctalus, Pipistrellus, Barbastella, Plecotus, Eptesicus, and Rhinolophus). Further analysis indicates that species‐level determination is possible in some of these taxa, but it would more readily be achieved using the more advanced methods of collagen sequence analysis by liquid chromatography coupled with tandem mass spectrometry. Within our assemblage yielding ~6,800 ancient bone collagen fingerprints, we identified only ~1% that derived from chiropterans, and these were predominantly derived from Myotis (two apparent Brandt's bat fingerprints but the majority indistinguishable between the whiskered, Daubenton's and Natterer's bats), Barbastella (the western barbastelle being the only member of this genus known within Europe), and Rhinolophus (identified as the lesser horsehoe bat R. hipposideros rather than the rare greater horseshoe bat R. ferrumequinum). We infer that the site was likely used by roosting bats throughout the year, and the accumulation of these remains was probably not the result of predator activity. More importantly, the peptide biomarkers provided here could proove valuable in the more systematic analysis of microfaunal remains across many European archaeological and palaeontological sites, preferably those that are collected with well curated stratigraphical information and chronological frameworks.


| INTRODUCTION
Bats are the second most diverse order of mammals, after rodents, with over a thousand extant species recognised and found throughout much of the world (Wilson & Reeder, 2005). They are the only mammals that are capable of active flight, made possible by modification of the limbs to form a membranous wing, most of it supported by the elongated finger bones. Although some species of megachiropteran bats can weigh as much as a kilogram when adult, for example, the golden-capped fruit bat Acerodon jubatus (Stier & Mildenstein, 2005), most bats are considerably smaller, typically weighing less than 50 g (Barclay & Brigham, 1991). However, despite their limited size, they still show great ecological and physiological diversity (Hutson, Mickleburgh, & Racey, 2001). Their ability to fly, together with their presence on every continent but Antarctica, has led to a wide range of feeding and roosting habits, as well as social behaviours (Jones, Jacobs, Kunz, Willig, & Racey, 2009).
Bats from the temperate regions of the Northern Hemisphere feed exclusively on insects and other arthropods, consuming a third or more of their own body mass each night in summer (Schober & Grimmberger, 1989). This level of consumption suggests that some species of bat may have a role in the suppression of arthropod populations (Kunz, Braun de Torrez, Bauer, Lobova, & Fleming, 2011).
Bats from temperate regions roost during the day time in natural enclosures such as tree hollows, rock crevices, and caves (Lewis, 1995), as well as man-made structures such as kilns and buildings (Jones et al., 2009). Different types of roost site are occupied in different seasons, due the changing needs of the bats (Schober & Grimmberger, 1989). For example, caves are frequently occupied in winter as the stable microclimates of these sites are favourable for hibernation. Given their reliance on suitable roosting sites, habitat infrastructure and invertebrate prey, all of which are likely to be affected by human activities, bats are sensitive to environmental changes and are therefore potentially important bioindicators (Jones et al., 2009). Many approaches are available for the study of modern bat populations, and this has led to a wealth of information regarding their present status; however, less is known about their past distributions and numbers. Information about prehistoric populations of bats can be obtained through the study of microfaunal assemblages, such as bone remains recovered from caves, and can provide useful comparative palaeoenvironmental data (e.g., Royer et al., 2017;Stoetzel et al., 2018).
Accumulations of bat remains from caves may be the result of natural deaths of the animals, especially when they are likely to be at their weakest towards the end of hibernation, but may also be deposited there by predators. A range of other vertebrates feed on bats, but the main predators of temperate bats are owls, with some species that are also vulnerable to falcons and hawks (Speakman, 1991). Given that owls may also roost in caves, the presence of bat remains in microfaunal assemblages could well be attributed to predation, with limestone cave environments offering better preservation of skeletal remains than open sites for various reasons that include more alkaline conditions and more stable typically cooler temperatures (Geigl, 2002;Weiner & Bar-Yosef, 1990).
Many of the morphological features used to identify bats are based on features of soft tissue, particularly the ears and muzzle (Schober & Grimmberger, 1989). These characters are not present in the case of ancient bat remains, as soft tissues do not typically survive for as long as skeletal elements. Although all but one of the 17 species that are presently found in the British Isles can generally be identified from skeletal characteristics, this relies mostly on features of the dentition, and only a few of the postcranial elements can be identified to genus or in a very few cases to species (Stebbings, Yalden, & Herman, 2007).
The analysis of DNA sequences has been widely applied in wildlife studies (e.g., Cronin, Palmisciano, Vyse, & Cameron, 1991;Foran, Crooks, & Minta, 1997) with even guano being used as a resource for bat identifications (Clare, Lim, Fenton, & Hebert, 2011). However, ancient DNA studies are relatively expensive and time-consuming and have limited success rates. This is particularly the case with smaller sample sizes and for those dating back to the Pleistocene (Höss, Jaruga, Zastawny, Dizdaroglu, & Paabo, 1996). An alternative biomolecular approach to species identification, collagen fingerprinting was more recently developed and circumvents some of these limitations.
Collagen fingerprinting is where type 1 collagen, the dominant protein in bone, is extracted into solution and enzymatically digested into peptides, which can be measured by soft ionisation mass spectrometry. The simplest approach of this is through matrix-assisted laser desorption/ionisation time-of-flight (MALDI-ToF) mass spectrometry and has been applied to the study of domesticated animals (Buckley et al., 2010;Buckley, Collins, Thomas-Oates, & Wilson, 2009) as well as wild fauna (Buckley & Kansa, 2011), including marine mammals (Buckley et al., 2014) and rodents (Buckley, Gu, Shameer, Patel, & Chamberlain, 2016). Since the initial publications using peptide mass fingerprinting (PMF) of archaeological bone (Buckley et al., 2009), several subsequent studies have been conducted that establish the validity of this approach to species identification; these are not limited to bone (e.g., Pin Hole Cave (SK533742) is a particularly important archaeological site in the British Isles because it acts as the type site for the mammal assemblage of marine isotope stage 3 fauna, so it is of potential importance in palaeobiostratigraphy (Currant & Jacobi, 2001). The Creswell Crags, a limestone gorge within which Pin Hole Cave is located, is also known for Britain's earliest cave art, yielding insights into one of the three known phases of ancient human occupations in the area, beginning with Neanderthals. The Creswell area is domi- there were at least two principal sediment bodies dating to the Pleistocene period, an upper red cave earth and a lower yellow cave earth but with faunal remains and lithic artefacts found throughout both (Jenkinson et al., 1984). Excavations were carried out in the early 1980s, focussing on two small areas approximately 30 m into the cave, with one~1.5 × 1.0 m at the top of the sequence and the other 1.0 × 0.5 m in much earlier deposits at the base (see Buckley et al., 2017) to more carefully obtain microfaunal remains through the use of sieving. These remains from the 1980s excavations were the source of the collagen fingerprints analysed here.

| MATERIALS AND METHODS
The collagen fingerprints of 6,805 specimens were collected previously as part of an earlier study that focussed on megafaunal bone fragments (Buckley et al., 2017). All of the bone remains were analysed as intact specimens that were small enough to fit within the 96-well microtiter plates used for high-throughput analyses. Given this constraint, megafaunal remains were clearly fragments, but microvertebrate remains mostly comprised intact skeletal elements. They were acquired using a relatively nondestructive approach in which 0.3 M hydrochloric acid (HCl) was added to the samples for only 3 hr prior to being removed and filtered into 50 mM ammonium bicarbonate (ABC) in which the collagen extracted from each sample was digested with the enzyme trypsin overnight for 18 hr. Spotted with 2 μl α-cyano hydroxycinnamic acid matrix, the air-dried droplets were analysed using a Bruker Ultraflex II MALDI-ToF mass spectrometer (see Buckley et al., 2017). Modern reference samples were also analysed following the above criteria. These included specimens of Elite tandem mass spectrometry following Buckley et al. (2015) in order to assist with peptide sequence identification. For sequence comparison, the little brown bat (Myotis lucifugus) sequences for both COL1A1 (G1QDY4_MYOLU) and COL1A2 (G1PSJ6_MYOLU) were BLAST searched against "Chiroptera" using National Center for Biotechnology Information, with only taxa that resulted in sequence matches retained for analysis here. Peptides of interest in the PMF, determined through visual comparison of spectra from the collagen digests of different species, were associated with peptide sequence information where possible, evaluating them against expected precursor mass without the proton (i.e., −1 Da). Phylogenetic analyses were also carried out following Buckley et al. (2015), in which maximum T A B L E 1 Collagen peptide mass fingerprint biomarkers observed through matrix-assisted laser desorption/ionisation analysis Value observed in others (e.g., Rhinolophus, Myotis, and Eptesicus) but useful for separating Barbastella from Plecotus due to absence in the latter. Peptide labels following Buckley et al. (2016); see Table S1 for peptide biomarker sequences. d By sequence similarity based on E. fuscus.
likelihood analyses were performed using PhyML with a JTT + I + G amino acid substitution model used and 10,000 generations/bootstrap reiterations to give branch support.

| Taxon discrimination
A combination of markers that we found useful for distinguishing bats from other taxa were m/z 1,435.7, 1,453.7, and 1,459.7. All seven genera could also be discriminated using a combination of other 2t67 is also particularly useful, this latter being homologous to a marker used to separate sheep from goat bone (Buckley et al., 2010).
Another marker that we have previously described for other taxa (2t85 reflecting marker A elsewhere; Buckley et al., 2009) remains useful for separating at the family level, along with a further marker described here (2t21). Intriguingly, the previously published "D" marker (2t69) was useful at separating the noctule bats from the pipistrelles.

| Sequence analysis
In addition to Myotis, the only taxa with sequence information relevant to this study are Eptesicus and Rhinolophus (Table 2 and Appendix S1 resulting in an expected peak at m/z 1,283.7. However, interestingly, we observe this at m/z 1,267.7 due to the substitution at one of the proline hydroxylation sites. Perhaps one of the most valuable regions of the spectra is at m/z 1,500-1,650, within which the 2t76 markers are pre-

Rhinolophus, reflected by an increase in 30 Da; the increase by 28 Da in
Nyctalus and Pipistrellus is likely attributable to a transition of one of the alanine residues to a valine but was not able to be confirmed through the liquid chromatography with tandem mass spectrometry analyses carried out here.
In a maximum likelihood tree that was inferred from the concatenated collagen sequences (COL1A1 and COL1A2), the genera Hipposideros and Rhinolophus form a well-supported clade with the fruit bats of the genera Rousettus and Pteropus, whereas the genera Myotis, Eptesicus, and Miniopterus form another wellsupported clade (Figure 1). We also observed a sufficient number of substitutions within the two genera from which more than one sequence were available, to yield species-specific information, with at least 4-10 substitutions within the three representatives of Myotis and six within Pteropus (Table 2). However, we have previously described the apparent biases against the observation of such biomarkers in the MALDI fingerprint (Buckley et al., 2016). Consistent with previous observations (Buckley et al., 2016), the collagen alpha-2(I) chain was observed to be typically 3-4 times more variable than the alpha-1(I) chain (twice this for M. brandtii in comparison with M. davidii).

| Identifications from Pin Hole Cave
The bats ( Figure 2)  There is also one additional spectrum that does not match any of F I G U R E 1 Maximum likelihood tree of the concatenated bat COL1A1 and COL1A2 sequences (showing bootstrap support for 10,000 replicates); see Appendix S1 for sequences our reference material but most closely resembles Nyctalus. Representative spectra for the bat taxa that were identified are shown in Figure 2, together with spectra from the putative representatives of M. brandtii and Nyctalus leisleri. Note that it is difficult to distinguish Plecotus from Barbastella, as this is reliant upon the presence of the 2t76 marker (see Tables 1 and S2). The differences observed at the species level in some taxa are supported by differences observed in both Rhinolophus species analysed (Figures S1 and S2) in addition to the sequence differences discussed above. Myotis, Eptesicus, and Miniopterus) conform with the division of the bats into the Yinpterochiroptera and Yangochiroptera (e.g., Teeling et al., 2002Teeling et al., , 2005. Although this split of the order Chiroptera was initially controversial, and rejects the longstanding view that the Microchiroptera are a monophyletic group, it has subsequently been confirmed from genomic data and is now generally accepted (Lei & Dong, 2016;Tsagkogeorga, Parker, Stupka, Cotton, & Rossiter, 2013).
The relative dates of the divergence between Yinpterochiroptera and Yangochiroptera,~52-58 Ma (Jones, Bininda-Emonds, & Gittleman, 2005;Teeling et al., 2005), are reflected in the differences between With regard to the bat remains from the Pin Hole Cave assemblage, although these are relatively few in number they show some interesting similarities and contrasts with the bat fauna nowadays present in this part of the British Isles. Horseshoe bats (Rhinolopus hipposideros) were most common, constituting slightly more than half of the good quality fingerprints that were obtained (Figure 3). Cave sites are frequently used by overwintering horseshoe bats (Schober & Grimmberger, 1989;Schofield & McAney, 2008), and the accumulation is in that sense not surprising. Although the species does not appear to be present in this area now (National Biodiversity Network, 2018), its range is known to have contracted substantially towards the south and west, since the beginning of the last century (Schofield & McAney, 2008). Lesser horseshoe bats were previously dominant among the bat species identified from Pin Hole Cave (Jenkinson, 1984) and were also numerous among the remains identified from the Neolithic material of Dowel Cave, further west in Derbyshire (Yalden, 1986), in keeping with this observation. The contraction in the species' range has been attributed to the loss of both woodland and roosting sites in abandoned mines (Harris, Morris, Wray, & Yalden, 1995), and the disappearance of the lesser horseshoe bats from these cave sites may offer some tentative support to the former explanation.
Mouse-eared bats (Myotis spp.) were also relatively common in the assemblage (Figure 3) and M. nattereri) and generally use cave sites both for winter hibernation and for social interaction in the autumn (Berge & Jones, 2008a, 2008bJan et al., 2010;Richardson, Waters, & Waters, 2008;Smith & Rivers, 2008). It is therefore not surprising that many of the remains were identified as Myotis, with at least two species apparently present.
Indeed, three species of mouse-eared bat (M. mystacinus, M. nattereri, and M. daubentonii) were identified from the earlier excavation of Pin Hole Cave, as were two species (M. mystacinus and M. nattereri) from Dog Hole Fissure, another cave in the Creswell Crags (Jenkinson, 1984), and at least two species (M. nattereri and M. daubentonii) from Dowel Cave (Yalden, 1986). In addition, although Bechstein's bat (Myotis bechsteinii) has only recently been recorded so far north in England, around 70 km to the west of Creswell (National Biodiversity Network, 2018), we cannot rule out that this species may have been more common earlier in the Holocene, when its deciduous woodland habitat was more extensive (Yalden, 1986). Given that this is the one species listed above that we have not been able to rule out with either reference material or available sequences, it is a plausible candidate for our unconfirmed identification of two specimens.
Remains from barbastelles (Barbastella) also appeared to be present in the assemblage (Figure 2), where a few individuals of this species were previously identified among the Pin Hole Cave and Dog Hole Fissure microfauna (Jenkinson, 1984). Although nowadays rare and generally restricted to more southerly parts of Britain, this species has recently been recorded in the area of the site (Cook, 2018;National Biodiversity Network, 2018). Overwintering barbastelles will enter cave sites during colder weather and are therefore not unexpected here (Schober & Grimmberger, 1989;Greenaway, 2008). The presence of relatively numerous barbastelle remains at the site, despite the current rarity of these bats, may again relate to their preference for old woodland (Greenaway, 2008), much of which has now been lost through human activity. On the other hand, it might equally reflect the severity of the winter weather at the time, as these bats respond to such conditions by entering caves (Harris et al., 1995).
Long-eared bats (Plecotus) were notably absent from the identified microfaunal remains. Brown long-eared bats (P. auritus) often hibernate in caves, and the species is common throughout most of Britain (Schober & Grimmberger, 1989;Entwistle & Swift, 2008;National Biodiversity Network, 2018). These bats are generally associated with tree cover (Entwistle & Swift, 2008), so they are likely to have been at least as common in the past, when deciduous woodland was more widespread (Yalden, 1986). Furthermore, long-eared bats have previously been recorded from both Pin Hole Cave and Dog Hole Fissure at Creswell (Jenkinson, 1984); however, this material was from the earlier excavations that encompassed much of the length of Pin Hole Cave. It may therefore originate from bats that occupied more exposed situations near the cave entrance, whereas the material identified in the present study was excavated from a single location at the rear of the cave.
No bats from the genera Pipistrellus, Eptesicus, and Nyctalus were confidently identified among the fingerprints, with the possible exception of N. leisleri. A single uncertain bat spectrum was found to most closely match N. noctula but had several peak differences, although the site is within the known range of both of these Nyctalus species (National Biodiversity Network, 2018). However, none of the above bats are generally associated with underground sites in Britain (although they may roost in rock crevices in some locations; Schober & Grimmberger, 1989;Hutson, 2008;Jones & Racey, 2008;Mackie & Racey, 2008;Shiel, Jones, & Waters, 2008), so it is unlikely that they would be present in the assemblage unless they were the victims of predators. Notably, one pipistrelle and two Leisler's bats (N. leisleri) were previously recorded from the Pin Hole Cave microfauna (Jenkinson, 1984) making our interpretation of the unknown fingerprint more plausible, but it should be emphasized that these were from the earlier excavations and therefore may have come from more exposed locations near to the entrance or be the result of predator activity. Likewise, a single Leisler's bat that was identified from Dowel Cave was attributed to avian predation (Yalden, 1986).
Overall, the bat fauna recovered from the cave suggests that it was used for hibernation, as all of the species that were present use underground sites in winter. It is therefore conceivable that the remains are from bats that died there, when they were at their weakest towards the end of hibernation.

| CONCLUSIONS
Past microfauna provide an opportunity to infer the effects of changing environments and, in this context, the potential importance of bat remains from archaeological sites in Britain has been recognised for some time (Yalden, 1986). However, their study has been hampered by the challenges in identifying isolated bones from their morphological attributes alone. Here, we demonstrate the successful use of collagen peptide mass fingerprinting to distinguish genera of bats that are present in Britain. The application of the procedure to an assemblage from a cave site in England highlights changes that have occurred in the local bat fauna, which can to some extent be related to the effects of past human activity on the landscape.