香京julia种子在线播放

Amud Cave is located in the Nahal Amud valley, 5 km northwest of the Sea of Galilee on the edge of the Jordan Valley (Figure 1). It is situated at the top of a steep cliff, ca. 30 m above the present valley floor. The site was first excavated between 1961 and 1964 by a Japanese expedition (Suzuki and Takai, 1970) and re-opened later between 1991 and 1994 by a joint Israeli-American team. The sequence of Amud Cave displays two main stratigraphic units: the uppermost unit A, which mostly consists of mixed Holocene sediments, and, underlying it, unit B formed by anthropogenic Middle Paleolithic sediments (Shahack-Gross et al., 2008; Zeigen et al., 2019; see Supplementary Figure 1). Within this latter stratigraphic unit, four sub-units were identified, well dated by Thermoluminescence and ESR-U series (Valladas et al., 1999; Rink et al., 2001) and numbered B1–B4 from surface to bottom (See Supplementary Data 1.1 for further detail). The uppermost sub-units, B1 and B2, are thought to represent a continuous deposition phase and were dated to ca. 55 Ka BP. The deepest and oldest sub-unit, B4, dated to 68.5–70 Ka BP, is separated from B2 by sub-unit B3, a sterile layer resulting from the collapse of the cave's roof. The sub-units are not all uniformly distributed among the two main excavated areas of the cave. In particular, sub-unit B1 can be found exclusively in Area (A) along the northern wall of the cave, while sub-unit B4 was encountered mainly in the central part of the cave in the two contingent Areas B and C. All the deposits yielded a large amount of lithic and faunal material, in association with ash, hearths, and other combustion features, especially in the central part of the cave. Taken together, these suggest an intense occupation (Shahack-Gross et al., 2008). In Areas B–C, organized lithic production is associated with hearths, whereas Area A is characterized by higher frequencies of cortical flakes, exhausted, and broken tools, as well as bones in different burning states found within the cemented ash. This led Area A to be interpreted as a discard area (Alperson-Afil and Hovers, 2005; Zeigen et al., 2019). Neanderthal remains were retrieved exclusively from Area A (Hovers et al., 1995, 2000).

In a sample of 5,340 specimens published by Rabinovich and Hovers (2004), one of the main observations was the high degree of fragmentation of the remains resulting in the dominance of small shaft fragments and splinters. This fragmentation pattern considerably constrained taxonomic analyses and the number of identifiable bones available for study. Nonetheless, results from this Rabinovich and Hovers' (2004) study identified the mountain gazelle (Gazella gazella) as the most-represented species (12%−14% of the assemblage), which is also part of the most abundant ungulate body size group (BSG) represented in the assemblage (BSG-D: body mass of 15–40 kg, as defined in Rabinovich and Hovers, 2004). Depending on the sub-unit, this size group is between 58% and 60% of the assemblage. Gazelle remains were mostly of juveniles and adults (up to 4–5 years-old). While amounting to <5% of the NISP, the next most represented species are the fallow deer (Dama mesopotamica), wild goat (Capra sp.), and red deer (Cervus elaphus). These taxa were all classified as falling in BSG B and C (80–250 kg and 40–80 kg, respectively, ranging between 5% and 10% of the NISP depending on the sub-units). Other, less abundant taxa included wild boar (Sus scrofa) and roe deer (Capreolus capreolus). Represented by only a few isolated remains are the common fox (Vulpes vulpes) and aurochs (Bos primigenius). Rabinovich and Hovers (2004) also suggested the possible presence of larger taxa such as rhinoceros (Stephanorhinus sp.), whose remains (mostly long bone splinters) might be misidentified as those of the aurochs, with both taxa falling into the same BSG A (>1,000 kg). Gazelle and fallow deer were shown to be represented by both cranial and post-cranial elements, suggesting that carcasses were likely transported complete to the site, in contrast to larger animals (body mass > 1,000 kg) which were only represented by long bone shafts, fragmented limbs and a few teeth (Rabinovich and Hovers, 2004).

The presence and impact of carnivores was observed to be negligible at Amud cave (Rabinovich and Hovers, 2004), with a notable absence of bone surface alterations caused by carnivores (tooth marks, scratching marks and gnawing marks) on the fauna, a rare phenomenon in Middle Paleolithic cave sites. Burnt bones were reported to represent up to 40% of the identified remains, and the burning colors exhibited were interpreted as likely indicating indirect exposure to fire (Rabinovich and Hovers, 2004). Cut-marks were observed on long bone shafts on 1%−3% of the total assemblage (Rabinovich and Hovers, 2004).

Kebara cave is located on the western flank of Mt Carmel (Figure 1), ca. 60 m above current sea level and 2.5 km from the modern seashore. The excavation campaigns at Kebara were initiated by F. Turville-Petre in 1931 and continued by M. Stekelis between 1951 and 1965. Later campaigns were undertaken between 1982 and 1990 (Bar-Yosef and Meignen, 2007; Meignen and Bar-Yosef, 2019, and references therein). Kebara cave displays a long Mousterian sequence, subdivided into 12 units dated from around 60–48 Ka BP (Valladas et al., 1987; See Supplementary Data 1.2 for further details). Large amounts of lithic material, animal remains as well as laminated hearths were uncovered in most of these units. Given the nature of the sediments, the concentration of combustion features and the density of finds, it has been suggested that the cave served mainly as a base-camp during the Mousterian occupation phases (Meignen et al., 2006, 2017).

Two major Middle Paleolithic occupation episodes were identified on the basis of the faunal remains found at Kebara. The first one, referred to as the “midden period,” corresponds to units XII to VIII and is characterized by an abundance of fauna and lithic material, as well as by high concentrations of burning evidence, such as cemented hearths and charred/burnt bones (Speth, 2019; Supplementary Figure 2). The bones are heterogeneously distributed, with a higher concentration along the northern wall of the cave and a very low density over the cave floor in the other zones, which has been shown to be independent from diagenetic processes (Weiner et al., 1993, 2007). The bone fragments concentrated in the northern zone show a higher incidence of burning, lower fragmentation rate and lower economic utility [according to Binford's (1978) criteria], suggesting that they were intentionally deposited there as a discard midden over successive human occupations of the cave (Speth, 2019). The over-representation of exhausted cores and cortical elements in this zone in comparison to the rest of the lithic assemblage reinforces this interpretation (Bar-Yosef et al., 1992). In contrast, the “post-midden period” (Units VII–V) shows a lower density of bone remains, with a significantly decreased contrast between the central and northernmost areas of the cave (Speth and Tchernov, 2007; Speth, 2019).

The faunal assemblage collected during the latest excavation campaigns (1982–1990) has been studied extensively by Speth (2019, and references therein), who assessed over 20,670 specimens across the seven Middle Paleolithic units. He identified the mountain gazelle (Gazella gazella) as being invariably the most abundant taxa, representing 45.6% of the total NISP, the fallow deer (Dama mesopotamica) representing 24.7% of the NISP, aurochs (Bos primigenius) 14.3% of the NISP, and red deer (Cervus elaphus) 8% of the NISP. Other represented taxa are bears (Ursus sp., 3.9% of the NISP) and Equids (2.1% of the NISP). Finally, wild goat (Capra sp.), roe deer (Capreolus capreolus), hartebeest (Alcelaphus bucephalus), and steppe rhinoceros (Stephanorhinus hemitoechus) represent <1% of the total number of remains.

The taphonomic analyses conducted on this assemblage showed no marked difference in the representation of cranial vs. postcranial elements. Evidence of carnivore damage was observed on 9.09% of the NISP in unit IX (Speth, 2019: Table 3.9), and burnt bones were found to represent 9% of the NISP, with burning traces observed more frequently on smaller taxa. The incidence of these burning traces was higher on limbs than on other skeletal elements, and more frequent on diaphysis fragments than on epiphyses, which was interpreted as indication that the burning events were likely related to cooking processes. Cut-marked bones were reported to represent up to 15% of the NISP (excluding dental elements; Speth, 2019).

Both Amud and Kebara faunal collections are stored and curated at the National Natural History Collections of the Hebrew University of Jerusalem. A comprehensive faunal study of the main stratigraphic units of Amud is currently in progress by one of the authors (A.J.). Data from previous zooarcheological studies (Rabinovich and Hovers, 2004; Speth, 2019) as well as unpublished databases, were combined to these newly collected data to form our faunal assemblage (N, Table 1). Within the assemblage, the specimens recorded as bearing cut-marks (n_{cut − marked} in Table 1) were identified and re-examined to confirm the presence of cut-marks, following published work on the differentiation of butchery marks from taphonomic damage (e.g., Shipman and Rose, 1983; Behrensmeyer et al., 1986; Andrews, 1995; Blasco et al., 2008; Domínguez-Rodrigo et al., 2009; Fernández-Jalvo and Andrews, 2016). For Amud, the studied specimens derive from stratigraphic sub-units B4, B2, and B1, dated respectively to ca. 70 Ka BP and to ca. 56 Ka BP (Valladas et al., 1999; Rink et al., 2001). Across the faunal remains assessed to date within these stratigraphic sub-units (N = 11,485), 249 bones were identified as bearing cut-marks, accounting for 2.2% of this assemblage (Table 1). For Kebara, we selected specimens from unit IX (N = 1,226), which is dated to 58.4 ± 0.4 Ka BP (Valladas et al., 1987) and is therefore broadly contemporaneous with the younger sub-units from Amud Cave. Cut-marks were identified on 95 bone fragments from this unit, representing 8.4% of the identified specimens (Table 1).

The specimens bearing cut-marks (n_{cut − marked}, Table 1) were first examined and described to record the following parameters for each cut-marked specimen:

Identifications and prey size classification. Osteological and taxonomic identification was conducted for each specimen using previous identifications provided in the collections databases as well as the comparative osteological and archaeozoological collections housed at the National Natural History Collections at the Hebrew University of Jerusalem. To facilitate comparisons between sites and stratigraphic units and evaluate the influence of prey size on the butchery patterns observed, the specimens were grouped using the following body size classification:

General taphonomic assessment. The degree of preservation and alteration of the cut-marked specimens was evaluated using the following parameters:

A sub-sample of specimens bearing cut-marks was further selected for detailed macroscopic assessment and micro-morphometric analyses (hereafter “detailed sample,” Table 1). For the Amud material, 43 specimens were selected in such a way that all of the studied sub-units and spatial areas of the cave would be equally represented, while excluding specimens from disturbed contexts or uncertain stratigraphic attribution (Supplementary Figures 1B, C). For Kebara, 34 specimens were selected for detailed analyses and originate from three squares situated in proximity to the northern cave wall (F19, G19, H19, and I13; Supplementary Figure 2B). Specimens in this sub-sample were measured and analyzed using macro- and microscopic techniques to quantify, characterize, and measure the cut-marks. They represent, respectively, 17.3% and 35.8% of the specimens identified as bearing cut-marks from Amud and Kebara, and account, respectively, for 0.4% and 3% of the total assemblages.

Macroscopic analyses. The selected specimens were first cleaned by applying acetone with a soft brush to remove the sediment embedded in the incisions or overlaying glue residues masking the marks. With the aid of a binocular lens (magnification: 10x−30x), the bone surfaces were assessed, drawn to record the location and morphology of the cut-marks, and measured with a caliper. The following parameters were recorded or calculated:

Micro-morphometric analyses. The selected specimens were studied using two Focus Variation microscopes, the Alicona InfiniteFocus G5+ (AIF) and the Portable Alicona RL (ARL) optical surface measurement systems (Optimax Ltd, Market Harborough, UK). These instruments are housed at the Imaging and Analysis Centre, Science Innovation Platforms, at the Natural History Museum (London, UK). The portable ARL system was transported and used directly at the National Natural History Collections of the Hebrew University of Jerusalem for the first phase of data collection. These optical systems allow for the non-destructive and non-invasive three-dimensional (3D) analysis of microscopic surface features. They create a series of individual image planes and overlapping focus levels to produce a virtual reproduction of the object in 3D. The recorded x, y, and z coordinates of each reconstructed pixel can be then used to conduct linear measurements of the surface features using the AIF software IF-MeasureSuite (Bello and Soligo, 2008; Bello and Galway-Witham, 2019). For both microscopes, a 10x lens was used to capture the finer detail of the cut-marks (AIF: working distance = 17.5 mm; numerical aperture = 0.3; vertical resolution = 100 nm; ARL: working distance = 17.5 mm; numerical aperture = 0.3; vertical resolution = 150 nm). Both systems offer the same level of accuracy and precision [finest lateral topographic resolution, AIF = 1.76 μm; ARL = 2 μm; Minimum measurable profile roughness (Ra), AIF = 0.5 μm, ARL = 0.55 μm], and are therefore expected to yield comparable outputs.

Linear and profile variables of the cut-marks were recorded following the methodology proposed by Bello and Soligo (2008), Bello et al. (2009), and Bello et al. (2013), by extracting a 2D profile from the mid-point of each incision (Figure 2). The following variables were considered:

The data were analyzed across the sites and stratigraphic units to evaluate whether specific cut-marks patterns could be observed. Results were compared between the Amud and Kebara samples, as well as between and within the Amud sub-units to assess intra-site spatial and chronological variability for this site. The spatial analysis of cut-marked specimens from Amud focused on sub-units B1–B2, to compare specimens from Area A to specimens from Areas B–C (Supplementary Figure 1).

Descriptive statistics and plots were computed using R Studio (version 2022.12.0+253, Posit Team, 2022; “ggplot” Wickham, 2006), and statistical tests were carried out using Past (version 4.05; Hammer et al., 2001) or R Studio. Statistical comparisons between groups were only carried out for sample sizes ≥ 5. For quantitative variables, the data was analyzed using non-parametric tests as the data was skewed and did not follow a normal distribution: Mann-Whitney-Wilcoxon or Mann-Whitney U tests (α 0.05) were used for pairwise comparisons, and Kruskal-Wallis tests followed by Dunn tests (where relevant) were used for comparisons between three groups or more. Qualitative data was analyzed using Chi square tests for independence (α 0.05) to determine whether any observed difference between samples were statistically significant. Chi square post-hoc tests include residuals analyses and, when significant differences were revealed for a contingency table of a matrix larger than 2 × 2, Fisher's exact test was applied to the collapsed contingency tables following DeViva (2014; see Sharpe, 2015). In addition, the micro-morphometric measurements were further tested across samples through a series of Principal Component Analyses (PCA) performed in R studio (“factoextra”; Kassambara and Mundt, 2017) on a covariance matrix, using all five variables (length, WIS, depth, OA, and Rd) as a way to assess the overall micro-morphometric variation of cut-marks across samples when analyzed as three-dimensional features. To prepare the data for PCA, the variable OA was converted from circular to linear data (OA_linear) following Courtenay et al. (2021) and Valtierra et al. (2024), to ensure all the variables considered were linear.

Species and prey size representation. While the same taxa were exploited at both sites, taxon representation is highly heterogeneous between the two sites (Figure 3A). This is reflected in our cut-marked samples. The Amud sample comprises almost exclusively taxa falling into prey size 1 (73.9%), and fewer from prey size 2 (7.2%) and includes a few unidentified specimens (not included in prey size comparisons). The sample from Kebara is mostly represented by specimens from prey size 1 and 2 (40% and 42.1%, respectively), and comprises also large ungulates (i.e., prey size 3, 17.9%).

Body parts representation. Long bone shafts are the most represented anatomical elements across Amud sub-units, where they constitute 60%−91.6% of the cut-marked specimen. They represent only 35.8% of the Kebara sample, where the distribution of body parts in the Kebara sample is relatively balanced and all body parts are well-represented (Figure 3B).

These proportions are roughly equivalent in the detailed samples (Supplementary Figure 4D). Long bone shafts represent up to 78.6% of the selected specimens from Amud and 29.4% of the Kebara detailed sample. Prey size 1 is the most prominent category represented in the detailed samples of Amud, and represents 50% of the Kebara detailed sample. However, it is worth mentioning that only three specimens of aurochs and equid are present in the detailed sample from Kebara (Supplementary Figure 4).

Exposure to fire. The majority of Amud cut-marked specimens was identified as likely burnt (n = 172, 68.5%). Another 52 were recognized as “burnt” (20.7%), and 27 as unburnt (10.7%). The majority of Kebara specimens was classified as unburnt (n = 90, 89.1%). An additional seven specimens were identified as heavily burnt (6.9%), and four (<0.1%) as likely burnt (Table 2 and Figure 3C). The differences between the two sites are statistically significant, as well as intra-site differences within Amud, with sub-layer B1 showing less burnt specimens than the other sub-units (χ² = 202.42, df = 2, p < 0.001; Supplementary Table 1).

Element completeness. Out of the 249 Amud cut-marked specimens, the greater majority (97.8%) represent less than half of a complete skeletal element; one was complete (0.4%), three were preserved to more than 50% (1.2%), and four were preserving half of the element (1.6%). Out of the 95 cut-marked specimens from Kebara, 16 items were complete (15.8%), the rest of the specimens (84.2%) preserved less than half of the skeletal element (Table 2 and Figure 3D). Chi² tests confirmed these inter-sites differences were significant (χ² = 19.682, df = 2, p < 0.001), and reveal further differences within the Amud cut-marked assemblage, with B1 presenting more complete elements than the other sub-units (χ² = 9.615, df = 2, p = 0.008; Supplementary Table 1).

Taphonomic bone surface alterations. Carnivore impact on the faunal remains is practically non-existent at both sites. Other major fossildiagenetic alterations, such as striations due to trampling, root-marks, water dissolution and drying cracks are also minimal and, when present, did not affect the reading of the butchery marks (Table 3).

Anthropogenic bone surface modifications. Considering all sub-units together, specimens bearing cut-marks represent 2.2% of the assessed assemblage for Amud, while they account for 7.8% of the Kebara layer IX assemblage. These numbers are in line with those provided in previous publications of both assemblages, i.e., 1%−3% of the remains studied from Amud depending on the sub-units (Rabinovich and Hovers, 2004), and 15% of the identified remains from Kebara (Speth, 2019). For Amud, we recognized 10 cut-marked bones bearing impact notches. In Kebara, two cut-marked specimens displayed impact notches. Both specimens from Kebara and four from Amud were included in our detailed sample.

The difference observed in the frequencies of burnt and fragmented specimens in both samples, as well as the scarcity of other anthropogenic and non-anthropogenic bone surface modifications, are mirrored in the detailed samples of both sites (Supplementary Table 2 and Supplementary Figure 4).

In the framework of the present study, the difference in fragment size between the two sites is significant, and it may have influenced several of the variables interpreted in the context of butchery behaviors. In particular, fragment size limits the ability to achieve taxonomic identifications and therefore to link butchery marks with prey choice. The strong difference in fragmentation of the two assemblages also likely distorted the comparison of the amounts of cut-marks present in each sample (see Abe et al., 2002; Domínguez-Rodrigo and Yravedra, 2009), despite the small difference observed between the two sites when evaluating the relative proportion of complete cut-marks observed in each sample. This potential bias was corrected in the present study by scaling the number of cut-marks per specimen relative to surface area, shifting the focus to the density of cut-marks per fragment, rather than cut-mark raw counts. Additionally, fragmentation affects the range of cut-mark lengths that could be measured in their entirety potentially leading to an underestimation of cut-mark length at Amud in comparison to Kebara, and thus explaining our results regarding this characteristic.

It might be tempting to argue that the difference in fragment size stems from different butchery strategies. Indeed, when comparing bone surface area in the two sites for only small ungulates (prey size 1), the two samples differ significantly, suggesting that the difference is not related to prey size. One could then argue that the smaller surface area of bone fragments in the Amud detailed sample and its lower size variability (Figure 4C) resulted from specialized butchery strategies meant to break bones into very small fragments. This interpretation, however, is not consistent with the limited amount of percussion damage observed at Amud relative to the size and number of fragments recovered. Another possible explanation is that the high frequency of burning at Amud (Section 4.1) led to its extreme bone fragmentation. Indeed, as suggested by previous studies, the bone fragments in Amud, burned in large proportions, would have been more friable and thus more susceptible to fragmentation due to post-depositional agents such as sediment compaction (Stiner et al., 1995; Rabinovich and Hovers, 2004; Reidsma, 2022). This topic would require further detailed study.

Despite the higher frequencies of burning, and potentially more post-depositional stresses on the bones of Amud in comparison to Kebara, it would seem that cut-mark morphology was preserved at both sites. At both sites, cut-marks outlines appeared pristine, with visible shoulder effects, internal microstriations and Hertzian cones visible in the SEM images (Supplementary Figure 5).

Cutting tools. One of the most prominent trends observed in the lithic assemblages of both open-air and cave sites in the late Middle Paleolithic in the Levant is the production of subtriangular short blanks, such as triangular flakes and points, often removed from unidirectional convergent Levallois cores. At Amud and Kebara, this similarity in morphology is obtained through slightly different site-specific knapping procedures (Meignen, 1995, 2019; Hovers and Belfer-Cohen, 2013; Krakovsky, 2017). Still, flint was the main raw material used for tool production at both sites. The two lithic assemblages are generally similar also in the preponderance of Levallois flaking strategies and in the production of flakes including points and triangular flakes and the low frequencies of retouched items (Goder-Goldberger, 1997; Hovers, 1998, 2004; Alperson-Afil and Hovers, 2005; Ekshtain et al., 2017; Meignen, 2019 and references therein). The general overlap in cut-mark micro-morphometrics across the two detailed samples (in particular the width and opening angle of the cut) is therefore not surprising when considering the broad similarities of the lithic assemblages (Section 2.1). The similarities in raw material used and in toolkit is a plausible explanation of the inter-assemblage similarities in micro-morphometric characteristics of the cut-marks.

Faunal spectrum, body parts representation and butchering activities. Most of our Amud sample consists of fragments of long-bone diaphysis from small ungulates. These anatomical areas typically undergo a reduced range of butchery activities (e.g., fileting), which would result in a high number of small, clustered cut-marks (as shown in Figure 6; Soulier and Morin, 2016; Soulier and Costamagno, 2017). In contrast, given the higher proportion of medium-sized and large ungulates, as well as the wider range of skeletal elements represented in the Kebara sample, we would expect a wide range of butchering techniques in this sample (such as skinning, disarticulating, and fileting; Speth, 2012 and references therein; Gifford-Gonzalez, 2018). This may explain the greater variability observed in cut-marks length and depth (greater depth of cut as been associated with the cutting of larger muscles; Bello et al., 2009; Wallduck and Bello, 2018) at the two sites.

However, when comparing only small ungulates (prey size 1), we still observe a greater density of incisions, higher proportion of overlapping marks and lower proportion of linear marks in the Amud sample compared to Kebara (Sections 4.3.1). These macroscopic differences cannot be related to the presence of different prey sizes. Moreover, at both sites, prey size 1 includes nearly exclusively mountain gazelles. Thus, taxonomic variation within this group can be considered minimal and unlikely to account for the observed differences between assemblages. Therefore, these patterns may suggest behavioral differences in the processing of small ungulates (see Section 5.4).

The Amud sample in our study consists of three stratigraphic sub-units. Given their chronologies, we expected that if modes of faunal exploitation changed through time, B1 and B2 (both within MIS 3) would be more similar to one another and differ from the earlier B4 (dated to the end of MIS 4). However, the variations observed in the detailed sample across the various sub-units were contrary to our expectations, with some aspects of the B1 sample differing from the two other sub-units. Firstly, B1 fragments are larger, show a lower density of incisions and are less frequently burnt than in any of the other sub-units. Secondly, specimens from B1 are the most variable in terms of cut-mark micro-morphometrics, bearing generally wider cut-marks with a larger floor radius compared to specimens from B2 and B4. Interestingly, it also appears that within the Amud sample, the sub-sample from sub-unit B1 appears to be the most similar to the Kebara unit IX sample in its micro-morphometric measurements, as well as in terms of cut-marks density (Figure 4B).

It is important to note however that sub-unit B1 is only present in the peripheral area of the cave (Area A; Supplementary Figure 2). Hence the intra-site differences might not necessarily reflect chronological variation, but might instead relate to differential use of the space in each cave, with different butchery activities being carried out in the central occupation area (Areas B–C; represented in our sample by specimens attributed to sub-unit B2) compared to the peripheral area of the cave (Area A; represented in our sample by specimens attributed to sub-unit B1). Indeed, Alperson-Afil and Hovers (2005) suggested that in sub-unit B2, knapping and living activities were carried out in Area C, while Area A was used as a refuse area. This interpretation was further supported by archaeomagnetic data (Zeigen et al., 2019) documenting variable heating intensities in this area. Thus, the presence of relatively large bone fragments variably exposed to heat (based on bone coloration) in sub-unit B1 may be explained as the result of cleaning of the central area. Interestingly, previous publications have suggested that the concentrations of bones along the northern wall during accumulation of unit IX in Kebara likely represented a discard midden (Speth, 2019). It is therefore possible that the similarities between Kebara IX and Amud B1 stem from the similar use of the two areas as discard area. Still, we are unable to explain the differences in cut-mark densities within the Amud sample through this spatial functional perspective.

Experimental work combined with ethnographic and archeological case studies have shown that cut-marks may attest to behavioral variations. While the processes leading to these phenomena are not well-understood (Domínguez-Rodrigo et al., 2017), several authors noted behavioral variations related to modes of exploitation and processing of the hunted fauna. For example, humans may opt for different timing and ways of obtaining meat off carcasses (i.e., preparing drying, boiled or rotten meat as opposed to fresh meat; e.g., Abe, 2005; Soulier and Morin, 2016; Soulier, 2021; Wallduck and Bello, 2018; Speth and Morin, 2022). Differences in group organization and social modes of food sharing were also suggested as underlying drivers of variation in the patterning of cut-marks (Stiner et al., 2009, 2011). Interestingly, experimental work on differences between expert vs. novice butchers suggested that skill cannot be correlated with the amount and morphologies of cut-marks (Pobiner et al., 2018; Soulier, 2021). For this reason, we consider that an interpretation suggesting that butchers were generally less skilled at Amud compared to butchers at Kebara does not sufficiently explain the higher density, clustered appearance and lower linearity of the cut-marks in Amud compared to Kebara.

Possibly, differences between cut-marks in the Amud and Kebara samples could stem from a more intensive exploitation of the carcasses at Amud cave. Such behavior could have resulted from a higher pressure on resources in a somewhat drier environment at Amud compared to Kebara (Section 1). Alternatively, more intensive occupations in Amud Cave, estimated through lithic frequencies per volume per duration (cf. Hovers, 2001, p. 133), could have necessitated a higher intensity of carcass use. However, experimental work has suggested that the frequencies of cut-marks are poorly correlated with the intensity of butchering activities (Egeland, 2003; Pobiner et al., 2018).

Another possible explanation for differences in cut-mark density and linearity between the two sites is that the butchering of meat in more advanced states of decomposition took place in Amud, but less frequently in Kebara. It has been shown that decaying carcasses tend to be more difficult to process, often resulting in the production of haphazard, deep, and sinuous cut-marks (Speth, 2017; Wallduck and Bello, 2018). The higher frequencies of non-linear marks in the Amud samples, compared to Kebara, could therefore suggest that the acquisition of meat off the prey was approached differently in the two sites, with, for example, decaying carcasses being processed more often at Amud Cave than at Kebara. Further experimental research and comparative work is needed to better assess the influence of these factors on cut-mark patterns. Additional variables, such as group organization (e.g., number of individuals involved simultaneously in butchering one carcass, see Egeland et al., 2014) and the types of gestures employed by the butcher(s) would also benefit from further investigation, which could help identifying the range of factors that could result in high frequencies of cut-marks.