New perspectives on human subsistence during the Magdalenian in the Swabian Jura, Germany

The Swabian Jura of southwestern Germany is famous for its Paleolithic sites which have been studied since the 1860s. While there is a rich tradition of research on the Magdalenian, many of the best-known sites were not excavated using modern methods, and recently, few discoveries of new sites have been made. Thus, much of the information on this period comes from sites lacking data collected using modern standards. This has left open questions regarding the recolonization of the Swabian Jura and hunter-gatherer subsistence and settlement during the Magdalenian in the region. Langmahdhalde is a recently discovered rock shelter in the Lone Valley of the Swabian Jura that has intact, well-stratified horizons dating to the Magdalenian with associated lithic artifacts, faunal remains, and combustion features. In this paper, we present a study of the faunal materials from this new site. We use the macrofaunal remains to discuss human subsistence and a taphonomic analysis of the microfaunal remains to determine the spatial scale of our previously published paleoenvironmental interpretations. Our results on human subsistence support previous interpretations from other Magdalenian assemblages in Central Europe. Further, our taphonomic study of the microfauna suggests that our paleoenvironmental analyses are relatively local, within a maximum of 70 km2 from the rock shelter. We place these results within the larger context of human paleoecology in the region and suggest that the successful resettlement of the Swabian Jura by Magdalenian peoples during the Late Glacial was probably facilitated by the presence of a higher diversity of resources on the landscape compared to regions to the west.


Introduction
During the Last Glacial Maximum (LGM; 27,200 to 23,500 cal yr BP; Sanchez Goñi and Harrison 2010), Central Europe was largely uninhabited by humans. Later, as the glaciers retreated and climates began to ameliorate during the Late Glacial (~18,000 to 11,600 cal yr BP), Magdalenian people recolonized Central Europe. By approximately 16,500 cal yr BP, it was largely resettled, except for areas further north that remained close to glacial fronts (Kretschmer 2015). The archaeological record suggests that the Swabian Jura of southwestern Germany was recolonized by populations from the west (Taller et al. 2014;Maier 2015Maier , 2017 by 16,300 cal yr BP, before the Meiendorf-Interstadial (Taller et al. 2014) or GI-1e of the NGRIP record (Litt et al. 2001;Lowe et al. 2008). The timing of this recolonization indicates that Magdalenian peoples entered the region not at the onset of an interstadial, as originally thought, but during the cold and dry conditions of the late Pleniglacial or GS-2 on the NGRIP record (Litt et al. 2001;Lowe et al. 2008). Taller et al. (2014) argued that the This article is part of the Topical Collection on Post-glacial human subsistence and settlement patterns Electronic supplementary material The online version of this article (https://doi.org/10.1007/s12520-020-01119-w) contains supplementary material, which is available to authorized users. resettlement of the Swabian Jura during the Magdalenian was, therefore, not driven by ameliorating climates, but by population growth and these populations' adaptations to specific environmental conditions.
There is a long history of archaeological research in the Swabian Jura of southwestern Germany. One of the first Magdalenian sites discovered in this region, indeed one of the first systematically excavated Paleolithic sites in the Swabian Jura, is Schussenquelle, an open air site that was excavated by Oscar Fraas in the mid-1800s (Fraas 1867;Schuler 1994). Local artifact assemblages dated to the Magdalenian are characterized by backed-bladelet industries, bone tools, reindeer antler, and fewer art objects compared to the earlier Upper Paleolithic (Schmidt 1912;Eriksen 1991;Taller 2014;Maier 2015). Studies of the faunal assemblages of these sites have indicated that the most commonly hunted animals were large migratory game, such as reindeer (Rangifer tarandus) and horse (Equus ferus), but ibex (Capra ibex), hare (genus Lepus), fox (genus Vulpes), and ptarmigan (Lagopus lagopus) are also commonly found in these assemblages (Riek 1973a;Eriksen 1996;Gaudzinski and Street 2003;Napierala et al. 2014;Maier 2015). Studies of artifact assemblages and raw material sourcing found that the Magdalenian sites in Central Europe can be broken into distinct regional groups (Maier 2012(Maier , 2015Kretschmer 2015). These studies group the sites of the Swabian Jura with other sites near the Federsee and in the Franconian Jura (Kretschmer 2015;Maier 2015). Further, Maier (2012) argues that the social networks of Magdalenian peoples in Central Europe were primarily within these groups. Weniger (1987Weniger ( , 1989 conducted what is still the most indepth study of settlement patterns in southwestern Germany during the Magdalenian. He described a pattern of seasonal mobility that includes the use of small, medium, and large sites. Small sites, he argued, are short-term field camps that were occupied by smaller bands of people during the spring and/or summer and used for short periods of time for hunting ibex and horse. Assemblages from these sites are thus characterized by few cores, stone tools, and organic artifacts, no portable art, and small faunal assemblages in which reindeer and horse are equally represented or horse is the dominant taxon. He called his medium sites "residential camps of local groups" (Weniger 1987(Weniger , 1989. These occur in both the lowlands and hills, are occupied during spring and summer, have hearths, and have approximately equal amounts of horse and reindeer remains. Large sites represent areas occupied during the winter by larger aggregations of people. Here, hunting focused on reindeer and more intensive activities such as food storage and hide working. These sites therefore have portable art and many more cores and stone tools. Notable archaeological sites dating to the Magdalenian in the Swabian Jura include the cave sites of Schussenquelle (Fraas 1867;Schuler 1994), Brillenhöhle (Riek 1973b), and Hohle Fels (Napierala et al. 2014;Taller 2014), and the rock shelter Felsställe (Kind 1987). However, several Magdalenian sites in the Swabian Jura are poorly preserved, severely affected by taphonomic processes, or are in contexts that are mixed with early Holocene materials. Further, until the past few years, no new archaeological sites with intact and well-stratified Magdalenian horizons have been discovered in the Swabian Jura since Kind's (1987) work at Felsställe. Newly excavated sites with Magdalenian deposits in this region are, therefore, necessary to address current research questions, such as whether there were local patterns in human subsistence and what factors drove the recolonization of the Swabian Jura.
Recent work in the Swabian Jura has led to the discovery of new Magdalenian sites (e.g., Conard et al. , 2019Kind and Beutelspacher 2018;Floss 2019). In 2016, the University of Tübingen began excavations at Langmahdhalde, a rock shelter in the Lone Valley, which revealed intact archaeological horizons dating to the Magdalenian, prompting annual excavations . Our recent study of the faunal remains from this site provided the first look at what faunal data from a new site with intact horizons dating to this time period can contribute to our understanding of human subsistence behavior and past environmental conditions (Wong et al. 2017). This previous study, though, presented results from only a portion of the faunal assemblage and a more comprehensive exploration of the remains from the site is necessary.
Following our initial study, we produced the first highresolution paleoenvironmental reconstructions for the Late Glacial in the Swabian Jura based on microfaunal remains and stable isotope data from horse and reindeer bone collagen (Wong et al. 2020). With this work, we demonstrated that Late Glacial environments in the Swabian Jura were more heterogeneous than modern tundra environments and likely even included stands of trees. In general, more heterogeneous environments have a higher species diversity (MacArthur and MacArthur 1961;Ganzhorn et al. 1997;Ceballos et al. 1999;Southwell et al. 1999;Cramer and Willig 2002;Williams et al. 2002;Ricklefs and Relyea 2014, p. 426), indicating that Magdalenian hunter-gatherers in the Swabian Jura had access to a larger diversity of animal and plant resources than are available in modern tundra environments (Wong et al. 2020).
Taphonomic analyses of microfaunal remains further refine paleoenvironmental interpretations made based on microfaunal data because they can indicate which nonhuman predator(s), such as predatory birds or small mammalian carnivores, probably deposited these remains at a site (Andrews 1990;). An understanding of that predator's behavior, such as hunting range and diet breadth, can have implications for the scale of paleoenvironmental interpretations that are based on the assemblage (Andrews 1990;Fernández-Jalvo and Andrews 1992;Fernández-Jalvo et al. 2011. Here, we present our latest work with the Langmahdhalde faunal assemblage which includes (1) a complete analysis of the macrofaunal remains from the site dating to the Magdalenian and (2) a taphonomic analysis of the microfaunal remains. We use the results of these two analyses, coupled with previous paleoenvironmental results, to discuss how this new record of human subsistence from the Swabian Jura adds to our understanding of the Magdalenian and to suggest factors influencing the resettlement of the Swabian Jura during the Late Glacial.
As this study focuses on the Magdalenian, we concentrate on GH3/AHIII to GH6/AHVI, the Late Glacial horizons. In GH4/AHIV, GH5/AHV, and GH6/AHVI, excavators have found lithics from one horizon that refit with those from another horizon. These three horizons are also characterized by more lithic and faunal artifacts than the layers above, including sections of reindeer antler, several almost complete horse elements, some small pieces of mammoth (Mammuthus primigenius) ivory, and a needle blank made on a goose humerus (Anser sp.; Wong et al. 2017). GH5/AHV includes six combustion features and burnt limestone rocks in association with these features.
Following the University of Tübingen system, workers excavated the site by 1/4m2 within 1x1 m quadrants named by the coordinates of their southwestern corner (Fig. 3). The team excavated GH5/AHV and GH6/AHVI in 2 to 3 cm deep sublayers that follow the geology of the site. Following the methods of our paleoenvironmental reconstructions (Wong et al. 2020), we present the microfaunal results for GH5/ AHV and GH6/AHVI by sublayer. Because GH3/AHIII has very few archaeological remains, excavators did not waterscreen all sediment from this horizon; they only waterscreened the southwestern 50 × 50 cm subquadrant from each 1 × 1 m quadrant. All other horizons discussed in this study had all sediment water-screened. This may result in the underrepresentation of small taxa and elements that are usually recovered during water-screening, as opposed to during excavation, in GH3/AHIII.

Methods
We define macrofauna as those specimens that belong to medium-to large-sized animals and were likely deposited at the site as a result of human or large carnivore activity. In the Upper Paleolithic of the Swabian Jura, these could, therefore, include taxa ranging in size from hare or ptarmigan to large ungulates, like bison (Bison sp.). In the current context, we define microfauna as small animals that have strong potential to reconstruct local environments. For our study, this includes only taxa from the orders Rodentia and Eulipotyphla (shrews, moles, and hedgehogs; we refer to this order as "insectivores" Fig. 1 Map of Langmahdhalde and other archeological sites mentioned in the text. 1 = Langmahdhalde; 2 = Vogelherd; 3 = Felsställe; 4 = Schussenquelle; 5 = Hohle Fels; 6 = Brillenhöhle. Map made in QGIS version 3.4 with topographic data from SRTM NASA version 3, hydrology data from the Landesanstalt für Umwelt Baden-Württemberg, administrative boundaries from ©EuroGeographics, and ocean data from Natural Earth throughout the paper). Of course, these two groups are not mutually exclusive, and there is likely to be overlap in the species brought to the site by different agents. However, micromammal assemblages (including rodents and insectivores) are almost always deposited by nonhuman predators, such as small mammalian carnivores or predatory birds ).

Macrofauna
Our taxonomic identifications of the macrofaunal remains from Langmahdhalde were based on the vertebrate comparative collection in the Institute for Archaeological Sciences at the University of Tübingen and several osteological atlases (e.g., Pales and Lambert 1971;Schmid 1972;Gilbert 1990;Hillson 2005;Gilbert et al. 2006). We identified all specimens to the lowest taxonomic level possible and recorded specimens using Stiner's (2005) landmark system with some modifications. If specimens were not identifiable to a specific taxon, we assigned them to a body size category, such as "medium mammal." Species-level identifications of hare remains were based on tooth morphology (Donard 1982;Callou 1997;Niethammer and Krapp 2003) and post-cranial measurements (Donard 1982;Pelletier et al. 2015). Species-level identifications of fox remains were based on tooth measurements (Baumann 2016). We calculated the number of identified specimens (NISP) for each taxonomic category and the minimum number of individuals (MNI) for each specific taxon (Grayson 1984;Lyman 2008). Our MNI calculations used the most common element and took side into account when possible. If specimens articulate, we gave them a total NISP value of one. We included long bone shaft fragments in NISP counts, but we did not include small unidentifiable ungulate tooth fragments.
We documented taphonomic signatures on all faunal remains, such as burning (Stiner et al. 1995), weathering (Behrensmeyer 1978), mineral staining, breakage, tooth marks, and any human modifications (Lyman 1994;Fig. 2 The stratigraphy of Langmahdhalde from excavation years 2016 to 2018. This is the southern profile of the site at the y-coordinate 37 (which runs east to west). Combustion features, stones, and roots can be identified using the key in the upper right. Further information can be found in Table 1. GH = geological horizon; AH = archaeological horizon.  Fernández-Jalvo and Andrews 2016). We define spiral fractures following Shipman et al. (1981). When discussing "cone fractures," we are referring to fracturing as a result of dynamic loading and direct percussion, referred to as "flake scars" by Lyman (1987) and the "impact point" by Johnson (1987). This type of fracture often causes pieces of bone to flake off; we refer to these as "cone negatives." There has been much discussion regarding the reliability of identifications of cut marks in the archaeological record (see Lyman 1994, p. 297-299 andDomínguez-Rodrigo et al. 2017 for discussions). We identified cut marks based on their placement and orientation and the following morphological attributes: V-shaped in crosssection, elongate, and the presence of "shoulder effects." Potential cut marks were examined under a Zeiss SteREO Discovery V8 microscope before being classified as such.
When possible, we also recorded indicators of age, such as epiphyseal fusion, tooth eruption, and tooth wear (Severinghaus 1949;Silver 1969;Payne 1973;Miller 1974;Levine 1979Levine , 1982Hufthammer 1995). When aging reindeer remains, we used tooth wear stages based on Miller (1974) and Severinghaus (1949) and epiphyseal fusion information from Hufthammer (1995). We follow Miller's (1974) reindeer age categories and define juveniles as 0 to 25 months old, subadults as 27 to 39 months old, and adults as 41 months or older. When aging horse remains, we used tooth eruption and long bone fusion information from Silver (1969) and tooth wear information from Levine (1979Levine ( , 1982. We follow Turner's (2002) horse age categories and define juvenile horses as 0 to 2 years old, prime adults as 3 to 6 years old, and old horses as 7 years old or older.
To evaluate whether density-mediated attrition affected the representation of skeletal elements in the macrofaunal assemblage at the site, we used two methods. First, we conducted a Spearman's rank-order correlation between bone density and survivorship (percent minimum animal units or %MAU) for scan sites for which bone density values are available (following Lyman 1994). The MNI values we used to calculate survivorship for each horizon can be found in Supplementary materials 1. We ran the correlation using the Stats package in RStudio version 1.2.1335. A significant positive correlation between survivorship and bone density would imply that density-mediated attrition might have affected the preservation of the assemblage. We calculated %MAU following Binford (1978Binford ( , 1984. Due to sample size, we conducted this analysis for only two taxonomic groups: hare (all specimens identified to mountain hare, Lepus timidus; European hare, Lepus europaeus; or the genus Lepus) and medium ungulate. We define medium ungulates as those ungulates weighing between approximately 40 and 250 kg which, at this site, includes specimens assigned to ibex, red deer (Cervus elaphus), reindeer, "large deer" (e.g., the genera Cervus and Rangifer), and the broader taxonomic category "medium ungulate." For our calculations with medium ungulates, we used bone density data for reindeer from Lam et al. (1999), without corrections for marrow cavities (called "BMD 1 "), as Lam et al. (1999) found that variations in relative bone density across different taxa are low enough to allow for accurate interpretations using density data from similar species. In our examination of hares, we used bone density values of snowshoe hare from Pavao and Stahl (1999; referred to as Lepus canadensis in the text).
The second method we used to examine whether densitymediated attrition impacted the assemblage is the ratio of teeth to cranial bone (cranial elements and mandibles). In general, we assume that as carcasses were transported from kill sites, the teeth remained inside the crania; thus, cranial bone and teeth would arrive at the rock shelter together and have equal minimum number of element (MNE; Lyman 1994) values. Tooth enamel, though, has a higher mineral content than bone, so teeth are more likely to preserve in the archeological record (Lyman 1994;Hillson 2005). If the ratio of tooth MNE to cranial bone MNE is higher than one, teeth are more common in the record than bone and it is possible that this is a result of their higher mineral content, not the actual number of teeth and bones deposited at the site. If this is the case, densitymediated attrition may be impacting the assemblage.
To evaluate potential decisions made by the huntergatherers at Langmahdhalde regarding carcass transport, we examine the representation of different skeletal elements at the site for the major taxonomic groups: hare, fox, horse, and medium ungulates. The hare and medium ungulate groups are defined above and the fox group includes specimens identified as arctic fox (Vulpes lagopus) and to the genus Vulpes. Based on Stiner (1991), we divide the skeleton into nine anatomical regions. We calculated the MAU for each region by dividing MNE by the number of times the elements of the region occur in the skeleton (Binford 1978(Binford , 1984.

Microfauna
For all horizons of the site, the microfaunal remains are more densely distributed along the dripline of the rock shelter, which generally corresponds to the units with east-west values of 49, 50, and 51 (see Fig. 3). The microfaunal assemblage at Langmahdhalde is large (n > 400,000) and, as such, we sampled the assemblage and analyzed only the remains from the southwestern corner of quadrant 50/38 (Fig. 3). There is one exception to this: excavators did not save any sediment from GH3/AHIII for water-screening from quadrant 50/38, and we therefore analyzed the taphonomy of the microfauna from the southwest corner of quadrant 49/37 ( Fig. 3) for GH3/AHIII. We included GH2/AHII and GH2a/AH2a in this analysis to provide context for GH3/AHIII to GH6/AHVI. We also present the taxonomic identifications from this sample (Wong et al. 2020 includes further taxonomic identifications from all subquadrants of quadrants 49/37, 50/ 38, and 50/39). We used several identification guides (e.g., Repenning 1967;von Koenigswald et al. 1974;Agadjanian et al. 1977;Niethammer and Krapp 1978Nadachowski 1982) and the vertebrate comparative collection in the Institute for Archaeological Sciences at the University of Tübingen to make these identifications. In order to classify specimens in the genera Sorex and Apodemus to the specieslevel, we followed measuring criteria from Niethammer and Krapp (1978, p. 326, 338, 361), Ziegler (1995), andMaul (2001). We took measurements at the Institute for Archaeological Sciences at the University of Tübingen using a Keyence Digital Microscope VHX-500F. We calculated the NISP and MNI for each taxonomic category (Grayson 1984;Lyman 2008). We did not include specimens identifiable only to Rodentia in our taxonomic identifications and we gave specimens that articulate a total NISP value of one. We do differentiate common and field voles (Microtus arvalis and agrestis) in this study, although, as we reported in Wong et al. (2020), both species are present in every horizon and sublayer of the site.
We identified burning in the microfaunal assemblage following visual criteria discussed in Stiner et al. (1995). We differentiated between burning and oxide staining with a visual inspection using a Zeiss Stemi 305 EDU microscope, where burning is actual blackening and calcination, as opposed to shades of brown that are more likely to be staining.
Our taphonomic analysis of the microfauna from Langmahdhalde follows the methodology established by Andrews (1990) and expanded on by other researchers (e.g., Fernández-Jalvo and Andrews 1992; Fernández-Jalvo et al. 2016 and references cited within). These and other studies have demonstrated that accumulations of micromammalian bones left by avian and mammalian predators show patterns in species representation, breakage, digestion, and skeletal element representation that are unique to certain categories of predator (Dodson and Wexlar 1979;Korth 1979;Andrews and Evans 1983;Andrews 1990;Stewart et al. 1999). Andrews (1990, p. 88-90) defines five predator categories, each associated with specific predators, based on the level of digestion on molars, incisors, and long bones, and the level of breakage observed in the assemblage. Category 1 predators leave little modification on their prey assemblages, meaning digestion is either absent or light and there is little breakage. Category 2 predators leave intermediate levels of modification. Category 3 predators leave moderate levels of modification. Category 4 predators greatly modify their prey assemblages. Finally, category 5 predators leave extreme levels of digestion on prey teeth and long bones and high levels of breakage in prey bone assemblages.
We recorded skeletal element representation, breakage, and digestion in the microfaunal remains from Langmahdhalde. We calculated the relative abundance of each skeletal element in the assemblage by horizon and sublayer based on the MNI value of the entire horizon or sublayer. We calculated MNI following Andrews (1990) and used all rodent and insectivore remains from the horizon. We calculated the percent relative abundance of each skeletal element by horizon by multiplying the actual number of the element represented in that horizon or sublayer by 100 and dividing by the number of that element that we would expect in the horizon based on the MNI (Andrews 1990, p. 46-47). For example, if a horizon has an MNI of 10, we would expect there to be 20 femora in the horizon because each individual rodent or insectivore has two femora. In Andrews (1990), vertebrae are grouped together as one "skeletal element." We did the same here and included only cervical, thoracic, lumbar, and sacral vertebrae in this category and set the expected number of vertebrae for one individual as 32, following Andrews (1990). We did not calculate the relative abundance of in situ teeth (teeth in the alveolar socket of the maxilla or mandible).
We recorded breakage in long bones, mandibles, and maxillae using Andrews ' (1990, p. 51, 53, and 56) portion and breakage categories. We also calculated the relative proportion of isolated molars and incisors using the following equation: where the number of missing teeth is equal to the number of in situ teeth present in the assemblage subtracted from the number of expected teeth. We use the relative proportion of isolated molars and incisors to discuss mandible and maxilla destruction during consumption and post-depositional taphonomic processes, such as trampling or screening. This index assumes that teeth are in their alveolar sockets when an individual is consumed by a predator. A proportion of isolated teeth over 100% would indicate that there are more isolated teeth in the assemblage than can be explained by the number of intact alveolar sockets in the assemblage, thus suggesting that destruction of the mandibles and maxillae occurred. To further evaluate the level of breakage in cranial elements, we also calculated percent molar and incisor loss using the following equation: where the number of expected teeth is defined above and tooth loss is the number of empty alveolar sockets preserved in the mandibles and maxillae of the assemblage. Finally, we recorded digestion on molars, incisors, proximal femora, proximal ulnae, and distal humeri. When categorizing levels of digestion on teeth and the ends of long bones, we used Andrews' (1990) categories: light, moderate, heavy, and extreme but also added the categories "light/moderate" and "moderate/heavy" (following Rhodes et al. 2018Rhodes et al. , 2019. To avoid overcounting, we recorded digestion on incisor tips only and did not include small fragmented pieces of incisor tips in our counts. We also only recorded digestion on molars that are ≥ 60% complete.

Taxonomic representation
We present the NISP and MNI values for the Langmahdhalde macrofaunal remains in Table 2. GH3/AHIII has the smallest amount of macrofaunal remains, followed by GH6/AHVI, whereas GH4/AHIV and GH5/AHV have the highest NISP values. Across all four archeological horizons, hare, reindeer, horse, ptarmigan, and small carnivores are the most common groups represented. Several hare teeth and one innominate are identifiable to the species-level in GH4/AHIV through GH6/ AHVI, demonstrating that both European and mountain hare are present in the assemblage. The innominate is from GH4/ AHIV and we assigned it to mountain hare based on measurements (diameter of transverse condyle = 12.6 mm; maximum diameter of anterior-posterior condyle = 13.1 mm ;Donard 1982;Pelletier et al. 2015). Small carnivores are mostly represented by foxes and Mustelids which are present across all horizons except GH3/AHIII which has no Mustelid remains.
Ungulates, in particular deer, are quite common in the assemblages of all four AHs. Reindeer NISP and MNI values include several pieces of antler, one of which is attached to a fragment of the cranium (Fig. 4) and another has an intact base, indicating that it was collected after it was shed. Both are from AHV. The macrofaunal assemblage also includes three pieces of mammoth ivory, one in GH5/AHV and two in GH6/AHVI. Large carnivore remains are rare in the assemblage except for nine cave lion (Panthera spelea) remains. The mammal categories based on body size have much higher NISP values than any other taxonomic category (Table 2). These numbers are primarily driven by long bone shaft fragments which, in GH3/AHIII, GH4/AHIV, and GH5/AHV, make up over 80% of the NISP values for the categories small, medium, and large mammal and in GH6/AHVI make up between 50% and 65% of the NISP values for these categories.
Very few bird remains are present in GH3/AHIII. In GH4/ AHIV, GH5/AHV, and GH6/AHVI, most of the bird specimens identifiable to the species-level are ptarmigan and other medium-sized birds, including ducks and other Phasianids (Table 2). There are also several small bird remains, such as Passeriforms, in the assemblages from AHIV and AHV (Table 2).

Density-mediated attrition
In Table 3, we present the results of a Spearman's rank-order correlation between bone density values and survivorship (%MAU) for the hare and medium ungulates at Langmahdhalde. A Spearman's rank-order correlation found no significant p values for any horizon or any strong correlations. We show the tooth and bone MNE values and ratios for horse, medium ungulate, and hare by horizon in Table 4. In GH3/AHIII, the sample sizes are small, but the tooth to bone ratio suggests that both are equally preserved, with ratio values of 1. The same is generally true in GH5/AHV, where the ratio of the total MNEs is 1.14. In GH6/AHVI, there are very little data, only an MNE of three but the total tooth to bone ratio for the horizon is larger than one. In contrast to the other horizons, in GH4/AHIV, teeth are present more often than cranial bones and every tooth to bone ratio is above one. The total tooth to bone ratio in this horizon is primarily driven by hare remains. Overall, we find no evidence that densitymediated attrition has affected the assemblage, except in GH4/ AHIV.

Taphonomy
We summarize the taphonomic modifications on the macrofaunal remains from Langmahdhalde in Tables 5 and 6. Many  (Table 5). We recorded several specimens in the macrofaunal assemblage with spiral fractures, cone fractures, cut marks, and other signs of human modification. Four long bone shaft fragments  display cone fractures and two specimens are negatives of cone fractures (Table 5). Between approximately 3 and 7% of the specimens have spiral fractures, depending on the horizon. In terms of evidence of butchering, we observed cut marks on 0.6 to 1.8% of the specimens, depending on the horizon ( Table 5). The majority of the specimens displaying  The final row for each horizon ("Total") presents these MNE values summed. The medium ungulate group includes specimens identified as ibex, red deer, reindeer, large deer, and medium ungulate. The hare group includes specimens identified as European hare and mountain hare, as well as specimens assigned to the genus Lepus. The fox group includes specimens identified as artic fox and assigned to the genus Vulpes  Pavao and Stahl (1999). Medium ungulate density values from reindeer values reported in Lam et al. (1999) without corrections for marrow cavities All features are in GH5/AHV. Specimens included in NSP counts are both diagnostic and nondiagnostic cut marks are long bone shaft fragments of medium mammals, but in GH4/AHIV, horse, reindeer, and fox remains have cut marks, as does one medium bird specimen. In this horizon, there is one fox (Vulpes sp.) mandible with two parallel cut marks on the buccal side that suggests it was skinned, probably for its fur (Binford 1981, p. 47;Wong et al. 2017). We observed cut marks on horse and reindeer specimens in GH5/ AHV and on horse specimens in GH6/AHVI. Another observed human modification on the faunal remains is a small (approximately 3.5 cm long) antler fragment, from GH6/ AHVI, that has been grooved. The abovementioned reindeer antler from GH5/AHV that is connected to the braincase has a notch close to the base (Fig. 4) that, based on its placement and morphology, is likely anthropogenic, although the poor preservation of the antler's surface makes this difficult to confirm.
In general, there are very few diagnostic burned remains at the site; most burned faunal remains are small, unidentifiable fragments weighing less than 0.1 g (9.5% of the specimens in AHIII are burned and 0 to 3.2% of specimens in GH4/AHIV through GH6/AHVI are burned; Table 5). Although there are six combustion features in AHV, there are very few burned faunal remains associated with these features and those that are associated are small fragments of teeth, long bone shafts, or unidentifiable elements, most of which are not diagnostic (Table 6). Further, these burned fragments are mostly from feature 1, the largest combustion feature at the site, and feature 6 ( Table 6).
We observed carnivore modifications on all categories of taxa, excluding carnivores. Less than 1% of the specimens in each horizon of the assemblage have evidence of carnivore bite marks or gnawing. In GH6/AHVI, there are three digested specimens. On five hare (Lepus sp.) specimens, we observed bite marks or punctures. Four of these specimens are from GH4/AHIV: one is a calcaneus with carnivore gnawing and a puncture mark approximately 4.3 mm in diameter, one is a first phalanx with a small circular puncture mark, one is a second phalanx with a small circular puncture mark, and one is a humerus fragment with a bite mark and carnivore gnawing. The final specimen is from GH6/AHVI and has five bite marks that measure an average of 1.4 mm in diameter, as well as evidence of carnivore gnawing on one end.

Aging
Some specimens from the macrofaunal assemblage at Langmahdhalde provide information on the age at which the animal died, and we provide a complete summary of this information in Supplementary materials 2. In GH4/AHIV through GH6/AHVI, there are six juvenile hare specimens, six juvenile fox specimens, and 12 medium bird specimens that are unfused. There is a fetal pelvis in GH3/AHIII, both the left and right sides, that is probably ungulate. Most specimens that provide aging information are reindeer or horse. Of these, there are nine that can be assigned to specific age categories, none of which are from GH3/AHIII or GH5/AHV. In GH4/AHIV, there are two juvenile and one adult reindeer and one old horse. In GH6/AHVI, we identified one reindeer specimen as juvenile.

Skeletal element representation
In Fig. 5, we display the representation of each anatomical region of the body by AH for hare, fox, horse, reindeer, and medium ungulate at Langmahdhalde. The samples are small but some patterns are visible. For hare, most of the body is represented in GH4/AHIV and GH5/AHV, and the feet, lower hind limb, and front limb are almost always present. There are few fox remains in the assemblage, resulting in few clear trends in the skeletal element representation for this taxon except that feet are present in every horizon. For horse, both upper front limbs and lower hind limbs are almost always present, and in GH5/ AHV, head elements are overwhelmingly represented compared to other elements and other horizons. For reindeer, antler and crania are present and are the most common elements for every horizon except GH3/AHIII, which has neither. The lower hind limb of reindeer is present in every horizon. In the category "medium ungulates" (which includes reindeer), cranial elements are always present, and otherwise, the majority of the body is present in every horizon except GH3/AHIII, although axial elements are nonexistent.

Microfauna taphonomy
The taphonomic analysis of the microfaunal remains included 7921 specimens. In Table 7, we report the total number of microfaunal specimens in our taphonomic analysis by horizon and sublayer. Table 7 also includes the relative abundance of each skeletal element and the overall MNI value for each horizon and sublayer. No microfaunal remains that are identifiable to element were observed in GH5/AHV sublayer 1 of our sample, so we exclude this sublayer from our taphonomic analysis. We present the taxonomic identifications in Table 8. All horizons of the assemblage are dominated by Arvicolids, with common and field voles present in every horizon and sublayer. In the Holocene horizons, red-backed voles (Myodes glareolus) are the most common taxon, while in the Late Glacial horizons, collared lemmings (Dicrostonyx sp.) dominate and narrow-headed voles (Lasiopodomys gregalis) are represented in numbers similar to common and field voles. Insectivores are rare in the assemblage and are primarily in GH2/AHII, although there are three shrew specimens in GH5/AHV sublayer 4.
In Table 9, we report the number of specimens in the microfaunal assemblage that show evidence of burning or oxide staining. Only three of the microfaunal remains included in the taphonomic analysis are burned, all of which are upper molars. Between 13 and 40% of the microfaunal specimens exhibit oxide staining, depending on horizon. The greatest percentage of oxide-stained specimens is in GH4/AHIV (40.4%).

Breakage
We present the representation of long bone portions in the microfaunal assemblage from Langmahdhalde in Table 10.
In general, few complete specimens are present in the assemblage; GH5/AHV sublayer 2 has the highest percentage of complete long bones (humeri, 20%). There are no complete long bones in the Holocene horizons of GH3/AHIII. In GH2/     AHII and GH2a/AHIIa, distal portions are the most common. In GH3/AHIII, the proximal portion is the most common, and in GH4/AHIV to GH6/AHVI, proximal and distal ends are similarly represented, making up about 28 to 46% of the long bone specimens.
In Table 11, we report information on maxilla and mandible breakage, including Andrews' (1990) breakage categories, molar and incisor loss, and the relative proportion of isolated molars and incisors (% isolated molars and incisors). As with long bone breakage, we find very few complete mandibles in the Langmahdhalde microfaunal assemblage. Most of the mandibles are missing their ascending rami and have broken inferior borders. Across all horizons and sublayers where maxillae are present, they are highly fragmented; most maxillae no longer retain their zygomatic arch and none are present in skulls. The percent molar and incisor loss values for all horizons and sublayers at Langmahdhalde are quite low (almost all values fall below 50%), further indicating that mandibles and maxillae are quite fragmented in the assemblage.
We could not calculate the relative proportion of isolated teeth for GH2a/AHIIa because there are no mandibles or maxillae in this horizon. Only in GH4/AHIV is there a value less than 100% for the relative proportion of isolated molars, indicating that there was less tooth row breakage in this horizon; all other horizons have values over 140% for the relative proportion of isolated molars. GH2/AHII has the highest value with 288.9% isolated molars, indicating that this horizon likely had the highest amount of tooth row breakage. Finally, all horizons (except GH2a/AHIIa) have relative proportions of isolated incisors over 110%, indicating higher levels of breakage for anterior mandibles and maxillae. GH6/AHVI sublayers 1 and 2 have the highest values (428.6% and 339.1%, respectively).

Digestion
In Table 12, we report all recorded evidence of digestion on long bones in the Langmahdhalde microfaunal assemblage. We provide all incisor digestion data in Table 13 and all molar digestion data in Table 14. The majority of the molars, incisors, and long bones in the assemblage are not digested, except for the long bones in GH2/AHII which most commonly show moderate levels of digestion. In GH2a/AHIIa and GH3/ AHIII, there are very few long bones that we could include in the analysis; of these, only two are digested, both from GH2a/AHIIa. In all horizons and sublayers from GH4/AHIV to GH6/AHVI, except GH5/AHV sublayer 2, there are specimens that have light, medium, and heavy digestion. In GH4/ AHIV, approximately 40% of the long bones are digested, with light and medium digestion being the most common. All sublayers of GH5/AHV and GH6/AHVI have remarkably similar proportions of digested long bones. The majority of the long bones in these horizons that are digested have similar -MNI values for each horizon or sublayer are based on all rodent and insectivore remains from the horizon or sublayer and are reported in the table. We did not calculate the % relative abundance of in situ (in bone) teeth. Actual # = the actual number of that element represented in the assemblage of that horizon or sublayer. Expected # = the number of that element we would expect in the horizon based on the MNI of the horizon. % relative abundance is calculated by multiplying "actual #" by 100 and dividing the result by the "expected #" (Andrews 1990, p. 46-47)   proportions of digestion-level categories (light to extreme), across all sublayers: light digestion is the most common, followed by medium. Most incisors and molars show no evidence of digestion and those that do usually have low levels.
No in situ incisors are digested. We identified four isolated incisors (one from GH4/AHIV and three from GH5/AHV sublayer 6) and three molars (from GH5/AHV sublayers 4 and 6 and GH6/AHVI sublayer 2) as extremely digested. Because very few molars were recovered in mandibles and maxillae, we can say little regarding digestion on in situ versus isolated molars.

Accumulators of the microfauna
In the Langmahdhalde microfaunal assemblage, most teeth and long bones are not digested (the only exception being the long bones of GH2/AHII), implying that a category 1 predator, which modifies prey remains very little prior to their deposition, was responsible for this assemblage. This interpretation, though, does not explain the high levels of breakage we observed in mandibles, maxillae, and long bones across the sample or the percentages of long bones and teeth that are digested in the assemblage. It is therefore probable that more than one type of predator accumulated the microfaunal assemblage at the site. However, we cannot eliminate the possibility that the high levels of breakage in the assemblage could also be the result of trampling or breakage during water-screening. Our discussion, therefore, focuses primarily on using the percentage of long bones, molars, and incisors that have been digested to present the likely predator(s) responsible for depositing the assemblage in each horizon and sublayer. To do so, we rely on Andrews ' (1990, p. 88-90) summary of predator modifications on micromammal assemblages, as well as Rhodes et al.'s (2019)  predator categories according to digestive modification. When discussing predators based on these two sources, we only use those species distributed in Central Europe during the Pleistocene and/or Holocene. In GH2/AHII, the percentage of incisors that are digested (24.6%) falls in line with a category 2 predator, whereas the percentage of digested molars (40.8%) and long bones (73.3%) indicate a category 3 predator. This suggests that several predators likely deposited the microfauna from this horizon. Predators associated with category 2 incisor digestion are long-eared owls (Asio otus) and great gray owls (Strix nebulosa); predators associated with category 3 molar digestion are Eurasian eagle owls (Bubo bubo) and tawny owls (Strix aluco); and predators associated with category 3 long bone digestion are little owls (Athene noctua), kestrels (Falco tinnunculus), hen harriers ( Circus cyaneus), and peregrine falcons (Falco peregrinus). Long-eared owls, great gray owls, and hen harriers prefer to feed on Arvicolids (Andrews 1990, p. 184, 190-191, 197), while Eurasian eagle owls and tawny owls are opportunist feeders (Andrews 1990, p. 189 and 193). Little owls, kestrels, and peregrine falcons are unlikely to greatly influence micromammal assemblages. Both little owls and peregrine falcons consume very few mammals; little owls prefer insects and peregrine falcons prefer birds (Andrews 1990, p. 193-196). Although kestrels prefer to consume voles, their pellets are not necessarily accurate representations of the individuals they have consumed as there is much loss of individual prey items due to digestion or the style of eating (Andrews 1990, p. 196). Arvicolids are, therefore, probably overrepresented in this horizon compared to their natural occurrence because of contributions from long-eared owls, great gray owls, and hen harriers, but the contributions of opportunistic predators as well indicate that the presence or absence of microfauna taxa in the assemblage is likely a good indicator of their presence or absence on the natural landscape.
GH2a/AHIIa and GH5/AHV sublayer 2 have similar digestion results. In both, the percentage of digested incisors and long bones implies a category 2 predator, while the percentage of digested molars suggests a category 3 predator. As mentioned above, predators associated with category 2 incisor digestion are long-eared owls and great gray owls, and predators associated with category 3 molar digestion are Eurasian eagle owls and tawny owls. Eurasian eagle owls and tawny owls are also associated with category 2 long bone digestion. These results, then, indicate that both opportunistic and specialist feeders likely accumulated the microfaunal assemblage in these horizons. We therefore find that, like in GH2/AHII, Arvicolids may be overrepresented in these two horizons compared to their relative abundance on the landscape but that presence/absence data for the taxa in the assemblage will The burn stages follow Stiner et al. (1995). The burned specimen in GH5/ AHV is from sublayer 3 Table 10 Representation of long bone portions of microfauna from Langmahdhalde by horizon and sublayer, following Andrews (1990, p. 50-52) GH2   probably reflect the presence/absence of most taxa on the landscape. GH3/AHIII and GH6AHVI sublayer 2 have similar results in terms of which category of predator the levels of digestion on the microfaunal remains indicate. Both show low percentages of digestion on incisors and long bones, indicating a category 1 predator and higher percentages of molar digestion (30 to 40%) that are more characteristic of a category 3 predator. Predators associated with category 1 digestion on incisors are barn owls (Tyto alba), short-eared owls (Asio flammeus), and snowy owls (Bubo scandiacus); predators associated with category 1 digestion on long bones are barn owls, short-eared owls, snowy owls, long-eared owls, and great gray owls. Again, Eurasian eagle owls and tawny owls are associated with category 3 digestion on molars. Like longeared owls and great gray owls, short-eared owls and snowy owls prefer Arvicolid prey (Andrews 1990, p. 181-182, 185). Barn owls are more similar to Eurasian eagle owls and tawny owls in this respect; they prefer Arvicolids, Murids, and Soricids as prey items (Andrews 1990, p. 179-180) and are, therefore, for the purposes of this study, considered generalists.
GH4/AHIV and GH5/AHV sublayers 4 and 6 have similar proportions of digestion on the teeth and long bones in their samples. The percentage of digested incisors suggests a category 1 predator, while the percentage of digested molars and long bones are indicative of a category 2 predator. As discussed above, barn owls, short-eared owls, and snowy owls are category 1 predators for percent digested incisors, while Eurasian eagle owls and tawny owls are category 2 predators for percent digestion on long bones. Snowy owls and great gray owls are category 2 predators associated with percent digestion on molars.
In GH5/AHV sublayer 3 and GH6/AHVI sublayers 1 and 3, there are few digested incisors (less than 20%), supporting the actions of a category 1 predator, such as a barn owl, shorteared owl, or snowy owl. Further, the percentage of digested molars in these sublayers indicates a category 3 predator and the percentage of digested long bones indicates a category 2 predator, both associated with Eurasian eagle owls and tawny owls. Lastly, in GH5/AHV sublayer 5, the percentage of digested molars, incisors, and long bones all indicate a category 2 predator: long-eared owls, great gray owls, snowy owls, Eurasian eagle owls, and tawny owls.
There are teeth with extreme levels of digestion in GH4/ AHIV, GH5/AHV sublayer 4, and GH6/AHVI sublayer 2. Predators associated with extreme digestion are usually mammalian carnivores or hen harriers (Andrews 1990), but juvenile barn owls also produce higher levels of digestion in their pellet assemblages (Andrews 1990;. As the Langmahdhalde assemblage was deposited by several different owl species and there are no other indications that mammalian predators or hen harriers contributed to the  No maxillae were present in skulls in the sampled Langmahdhalde microfaunal material. Molar and incisor loss is defined by the number of empty alveolar sockets. The % molar or incisor loss is calculated by multiplying tooth loss by 100 and dividing by the expected number of teeth. The expected number of teeth is the total number of teeth that should be present in the number of mandibles or maxillae present in the assemblage; we use three molars and one incisor for each quadrant on the mouth to calculate the expected number. The relative proportion of isolated teeth ("% isolated molars" and "% isolated incisors") is calculated by multiplying the number of isolated teeth by 100 and dividing by the number of missing teeth, where the number of missing teeth is calculated by subtracting the number of in situ teeth from the number of expected teeth     assemblage otherwise, we find that it is unlikely that these predators were involved (except in GH2/AHII where hen harriers are suggested by the percentage of digested long bones). This taphonomic analysis indicates that the small birds and young small mammals in the macrofaunal assemblage could have been deposited by the avian and mammalian predators responsible for the microfaunal remains at the site. The diets of long-eared, short-eared, and tawny owls include birds, and peregrine falcons, potential contributors to GH2/AHII, eat primarily birds (Andrews 1990). Further, several owls, including the Eurasian eagle owl, hunt Lagomorphs (Andrews 1990). The hare specimens from GH4/AHIV and GH6/AHVI that  The %NISP for each level of digestion is calculated using the total number of incisors in the horizon or sublayer. The %NISP of isolated or in situ incisors is calculated using the total number of isolated or in situ incisors in the horizon or sublayer. The %NISP for total isolated or in situ incisors is calculated using the total number of incisors in the horizon or sublayer have bite marks or punctures on them suggest that, at least in these horizons, a small mammalian carnivore likely contributed to the Langmahdhalde faunal assemblage. Based on the taphonomic analysis of the microfaunal assemblage from Langmahdhalde, we hypothesize that in all of the horizons and sublayers of the site, generalist hunters, such as Eurasian eagle owls and tawny owls, contributed to the assemblage, making the presence or absence of taxa a good reflection of their presence or absence on the landscape. But because birds which strongly prefer Arvicolids as prey were also probably responsible for the deposition of these materials, the relative abundances of these species in the assemblage are unlikely to be accurate reflections of the situation in the natural environment. Analytical methods that rely on the presence or  The %NISP for each level of digestion is calculated using the total number of molars in the horizon or sublayer. The %NISP of isolated or in situ molars is calculated using the total number of isolated or in situ molars. The %NISP for total isolated or in situ molars is calculated using the total number of molars absence of taxa in the Langmahdhalde microfaunal assemblage to reconstruct past environmental conditions are therefore more accurate than those that use taxonomic abundance data. Finally, our reconstructions of the primary accumulators of the microfaunal assemblage at Langmahdhalde have implications for the scale of paleoenvironmental reconstructions that use the microfaunal remains from the site. Long-eared owls, tawny owls, and short-eared owls have home ranges that are less than 10 km 2 (Andrews 1990, p. 182, 192), while hen harriers, Eurasian eagle owls, and barn owls have ranges between approximately 10 and 30 km 2 (Haller 1978;Picozzi 1978;Mikkola 1983;Arroyo et al. 2005;Martin et al. 2008). Finally, the predators with the largest home ranges are snowy owls and great gray owls, with ranges between 50 and 70 km 2 (Bull et al. 1988;Chang and Wiebe 2018). As all horizons and sublayers in the assemblage, except GH5/AH5 sublayer 3 and GH6/AHVI sublayers 1 and 3, seem to be associated with great gray owls, paleoenvironmental reconstructions from Langmahdhalde apply to within 70 km 2 of the rock shelter for these horizons. For GH5/AH5 sublayer 3 and GH6/ AHVI sublayers 1 and 3, barn owls have the largest hunting ranges of the likely accumulators, meaning that paleoenvironmental reconstructions probably apply to within 55 km 2 . For context, we note that the Lone Valley is approximately 44 km long.

Human use of the rock shelter
Overall, the macrofaunal remains from Langmahdhalde support current understandings of human subsistence behavior in southwestern Germany during the Magdalenian. The main prey animals were horse and reindeer, large migratory game that travel in herds and were readily available on the landscape. The presence of collected and worked reindeer antler at the site suggests that reindeer were not only a source of food, but also of raw materials. The people using Langmahdhalde also took smaller game, like ptarmigan, hare, and fox. In the case of fox, Langmahdhalde provides evidence that these animals were also used as a source of fur. Based on the presence of cut marks, cone fractures, and cone negatives in the assemblage, we find evidence that butchering and marrow extraction took place at the site. Further, needle-making and antler working were also probably performed, as demonstrated by the needle blank (Wong et al. 2017) and the worked antler pieces. As we have found no evidence of human modifications on the cave lion remains in Langmahdhalde, it is unclear whether lions used the site when humans were not present, or if their deposition at the site was anthropogenic.
The macrofaunal assemblage provides little evidence for which season(s) humans occupied the rock shelter, although the ungulate fetal pelvis in GH3/AHIII suggests that it may have been during late winter and/or spring, as most ungulates give birth during spring or summer. Based on element representation of the major human prey taxa at the site (Fig.  5), it is possible that for all horizons hunter-gatherers transported only certain portions of reindeer and horse to the site. However, sample sizes for the MAU calculations are quite small and we are cautious of making interpretations based on skeletal element representation in GH4/AHIV because this horizon could be affected by density-mediated attrition. In GH4/AHIV and GH5/AHV, though, the majority of the anatomical regions of the body are present for medium ungulates and hare. Based on the reindeer and horse results, it seems unlikely that medium ungulates (which includes reindeer) were generally transported whole to the site, but again, sample sizes are small. Looking at the age of reindeer and horse specimens at the site, we see that people exploited a mix of ages, with more juveniles being represented in the assemblage, although this is difficult to interpret due to small sample sizes.
The small sample size in the macrofaunal remains from GH3/AHIII, representing the end of the Late Glacial sequence at Langmahdhalde, may be the result of a decline in the use or abandonment of the rock shelter by humans. The absence of significant amounts of lithics and other artifacts in this horizon suggests this is a likely scenario, although a geoarcheological examination of the site is necessary to rule out natural processes, such as erosion or increased sedimentation, that would decrease artifact abundances. It is possible that the low number of small birds and small carnivores in this horizon is a result of the sampling bias in water-screening, as most of the specimens belonging to these taxa are recovered from screened materials. As GH4/AHIV and GH5/AHV have the highest number of remains, and GH5/AHV has several combustion features, it is possible that these horizons represent a more intensive use of the rock shelter, although we cannot say whether this means use for longer periods of time, more repeated use of the rock shelter, or larger group sizes in residence.
The faunal assemblages and combustion features in GH3/AHIII to GH6/AHVI at Langmahdhalde suggest that the site falls within Weniger's (1987Weniger's ( , 1989) medium site category, classifying it as a "residential camp of local groups." Further, thus far, excavators at Langmahdhalde have recorded over 100 stone tools and approximately 30 cores from these horizons, both classifying Langmahdhalde as more of a medium site in Weniger's system. Overall, Weniger (1987Weniger ( , 1989) classifies most of the Magdalenian sites of southwestern Germany as small sites, including Vogelherd and Hohlenstein-Stadel, the only sites from the Lone Valley included in his study. He classifies Brillenhöhle and Hohle Fels as medium sites and Felsställe and Schussenquelle as large sites (Weniger 1987). Recent excavations at Hohle Fels and work with the assemblages from both Hohle Fels and its neighboring site Helga Abri, though, suggest that the Magdalenian occupation of Hohle Fels was larger than previously thought (Taller 2014; Hess 2019).
Compared to archeological sites to the west of the Swabian Jura, such as Petersfels, Kesslerloch, or Champréveyres in Switzerland, Langmahdhalde has greater relative proportions of horse remains and fewer examples of worked antler and bone (Albrecht 1979;Albrecht et al. 1983;Leesch 1997;Morel and Müller 1997;Napierala 2008). It also lacks the mobile art present at these sites. Within the context of the Swabian Jura, Langmahdhalde continues to offer evidence of larger or, perhaps, more repeated, use of the region by Magdalenian peoples, especially in the Lone Valley, where only smaller Magdalenian sites were known previously.

Paleoecology
Our previously published reconstructions of the environment during the Late Glacial, based on the bioclimatic analysis and stable isotope analysis of bone collagen of reindeer and horse from the site, suggest that the environment surrounding Langmahdhalde was generally open tundra (Wong et al. 2020). With the addition of the above taphonomic analysis of the microfauna, we can say, additionally, that this reconstruction probably applies to within 70 km 2 of the site for all horizons and sublayers except GH5/AH5 sublayer 3 and GH6/AHVI sublayers 1 and 3, where reconstructions are likely within 55 km 2 of the site.
In Wong et al. (2020), we found that the tundra in this area likely had warmer winter temperatures, more annual precipitation, and longer vegetative activity periods than modern tundra environments, allowing for a more mosaic environment to develop that included patches of trees (Wong et al. 2020). This may explain the presence of a European elk/moose (Alces alces) phalanx in the macrofaunal assemblage of GH4/ AHIV. Modern European elk/moose prefer deciduous and mixed forests but, in the summer, can penetrate deep into the tundra (Nygrén 1986). The more heterogeneous environment may have allowed some individuals of this species to occupy the region, at least temporarily. There is also one roe deer (Capreolus capreolus) specimen in the macrofaunal assemblage (in GH3/AHIII). This species is usually associated with wooded areas that provide some cover (Walker 1968(Walker , p. 1404, and is rare in Central European assemblages dating to the Magdalenian. We suggest two possible explanations for the presence of this specimen in the assemblage. First, it is possible that, like the European elk/moose, a few roe deer were occasionally present in the Swabian Jura during this time. Another possibility is that the specimen is younger than the date from GH3/AHIII, as excavators recovered it from the westernmost part of the excavation, where the reduced stratigraphy is less clear. The ecological complexity of the Late Glacial of the Swabian Jura is further highlighted by the situation among different deer taxa in the region. Stable isotope results from Langmahdhalde and other sites demonstrate that, during the Late Glacial, the preferred habitats of reindeer shifted northward (Drucker et al. 2012;Immel et al. 2015;Wong et al. 2020), but reindeer, known to subsist on a variety of vegetation (Walker 1968(Walker , p. 1402Spiess 1979, p. 31), remained in the region. As a result, there is more and more evidence that their niches overlapped with the other deer species on the landscape. This is visible in the δ 13 C and δ 15 N values from bone collagen of horse and deer specimens. In Fig. 6, we compare these stable isotope results from several sites in the Swabian Jura across the Upper Paleolithic and find that the Magdalenian results, including those from Langmahdhalde, overlap much more than those from the Aurignacian and Gravettian (see Supplementary materials 3 for the stable isotope data and sources and Supplementary materials 4 and 5 for the methods associated with these values). This probably meant that these species were in increasingly direct competition with each other for resources.

Conclusions
Previously known sites in the Lone Valley of the Swabian Jura have demonstrated the presence of only small field camps that were likely occupied for short periods of time during the Magdalenian (Weniger 1987(Weniger , 1989. Although the size of the groups remains unclear, Langmahdhalde documents therepeated occupation of the Lone Valley by Magdalenian people. groups of Magdalenian people coming together at residential camps in this valley. Our work with the faunal remains from the site also demonstrates that these people were hunting large game, such as horse and deer, whose dietary ni ches were shif ting and begi nni ng to overlap. Paleoenvironmental reconstructions based on faunal remains from Langmahdhalde have shown that the tundra environments in the region were more heterogeneous and diverse than previously thought. Additionally, our taphonomic analysis of the microfauna from the site indicates that this environmental reconstruction applies specifically to the Lone Valley and its surroundings (within 55 to 70 km 2 of the site); further studies are necessary to determine whether this reconstruction is true throughout the Swabian Jura.
Overall, our work with the faunal remains from Langmahdhalde has combined traditional zooarcheological analysis, microfaunal analyses, and stable isotope analysis of bone collagen (in this paper and in Wong et al. 2017, Wong et al. 2020. This combination of studies has allowed us to examine human use of the rock shelter and the Lone Valley within the context of the local ecology. Based on all of our work, we suggest that the vegetative situation of the Swabian Jura during the Late Glacial may have made it an attractive place to settle for Magdalenian peoples moving into the area from the west. As current work indicates that the Swabian Jura was recolonized during the colder Pleniglacial (Taller et al. 2014), this region may have provided a greater diversity of vegetative and faunal resources for human populations compared to surrounding regions. This may have been especially true since humans living to the southwest of the Swabian Jura, at sites such as Kesslerloch, Champréveyres, and Petersfels, were in closer proximity to glacial fronts (Becker et al. 2016;Wong et al. 2020). Environmental reconstructions from the Swiss Jura also find that temperatures were colder than those reconstructed for the Swabian Jura and environments were open grassland with alpine and steppe vegetation (Coope and Elias 2000;Leesch et al. 2012). Our work suggests, then, that expansion into the Swabian Jura may not have been driven exclusively by adaptations to specific environments (Taller et al. 2014), but also by the availability of comparatively diverse resources in this region, further emphasizing the need for regional-and local-scale paleoenvironmental data.

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflicts of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.