Peer Review #2 of "Direct evidence of megamammal-carnivore interaction decoded from bone marks in historical fossil collections from the Pampean region (v0.2)"

1

pulses, fluvial process and various pedogenetic events influenced this period (Tonni et al., 2003;Fucks & Deschamps, 2008;Cione, Tonni & Soibelzon, 2009). When the collections analysed in this study were originally collected, these units were included in the "Pampean Formation" (Tonni, 2011). Current biostratigraphical information (Tonni, 2009) allows the material from MCNV to be assigned to the Ensenadan to Lujanian, Stage/Age and the material from MNHN and NMW to the Bonaerian and Lujanian Stages/Ages. Furthermore, in the NMW collection, the old reference to Upper Pampean is currently equivalent to the Bonarian Stage/Age (Tonni, 2011).
The last record of these mammal groups comes from the Guerrero Member of the Luján Formation, deposited between 21,000 and 10,000 14 C years BP. (Tonni, 2009). In the case of the MLP assemblage, the presence of the notoungulate Mesotherium cristatum (Notoungulata, Mesotheriidae) among the identified species means this material can be dated as Ensenadan (Cione, Tonni & Soibelzon, 2009) (Fig. 2 and Table 1).
To understand the natural burial conditions of the remains, we considered different types of bone surface modifications such as post-depositional fractures, the presence of original sediment or concretions, fluvial erosion, trampling, weathering, root growth, manganese spots and burning traces (e.g., Behrensmeyer, 1978;Binford, 1981;Shipman, 1981;Olsen & Shipman, 1988;Lyman, 1994;Fernández-Jalvo & Andrews, 2003, 2016. These allowed us to discard any type of intervention that could simulate carnivore activity or, if superimposed onto carnivore marks, could have indicated a previous carnivore intervention. We follow the literature to identify whether bone marks were the result of carnivore activity (e.g., Haynes, 1980Haynes, , 1982Haynes, , 1983Binford, 1981;Capaldo & Blumenschine, 1994;Lyman, 1994;Domínguez-Rodrigo & Piqueras, 2003;Pickering et al., 2004;Domínguez-Rodrigo et al., 2012;Delaney-Rivera et al., 2009;Sala, Arsuaga & Haynes, 2014;Sala & Arsuaga, 2016). As large mammal bones are too large to be ingested (Fernández-Jalvo & Andrews, 2016), we did not considered this effect as a possible agent of the marks. Furthermore, small bones tend to be splintered by the teeth of predators, making them impossible to classify either anatomically or taxonomically (Fernández-Jalvo & Andrews, 2016). Therefore, this type of fragmented material was not included in our review. The only exception was the case of the indeterminate and smaller bones from the MLP collection where part of the original association was conserved. Coprolites were absent in the reviewed collections.
We classified the bone marks potentially produced by carnivores into four categories (Table S1): (i) pitting and/or punctures, (ii) u-shaped elongated scratches or scores, (iii) furrowing; and (iv) spiral fractures. To investigate the body size of the potential carnivores that inflicted the marks, we used a box-plot diagram (Hammer, Harper & Ryan, 2001) to compare the size of the pitting and/or punctures from the MCNV, MNHN and MLP specimens with those published by Pickering et al. (2004) (various bones), de los  (bone specimen Xen 30-12), and Martin (2016) (various bones); the material from NMW was excluded for the small sample size (Table S2 to S5). We follow the studies mentioned above as they allowed us to compare palaeontological and archaeological cases from the Pampean region, Patagonia, and one African case, and appreciate any similarities and/or differences with African ecosystems. Even though this information was still statistically poor, it allowed us to make some preliminary assumptions.
Additionally, assigning a pit or puncture to a specific taxa is always problematic given the different factors involved (e.g., the part of the bone marked and the bite force of an animal) (Delaney-Rivera et al., 2009). Nevertheless, the overlapping of our data with the comparative cases allowed us to ascribe the marked bones to general carnivore size categories. Even though some authors have also included scores in their studies of body size (Delaney-Rivera et al., 2009;Labarca et al., 2013, de Araújo Júnior, de Oliveira Porpino & Paglarelli Bergqvist, 2011, we agree with Domínguez-Rodrigo & Piqueras (2003) that score marks relate not only with teeth size, but also the effect of the teeth being dragged over the bone surface; variability can therefore be expected from this type of marks.
We examined the fossil remains of the megaherbivores present in the collections with 3.5x and 12x magnifying glasses. We also used a Dino-Lite Microscope AD4113T (at magnifications of 20x to 45x) and the software Dino-Lite 2.0. Both the length and breadth (major and minor axes) of the scores, pits and punctures were measured. Larger marks were measured using a caliper, and smaller ones were recorded with the measurement tool installed in the Dino-Lite. For each collection, high-resolution digital images were taken, in each museum, using a Panasonic Lumix DMC-TZ35 camera.
For the MLP assemblage we also applied the well-established archaeozoological variables MNI (Minimum Number of Individuals) and NISP (Number of Identified Specimens), as all the specimens are part of the same taphocoenosis (Lyman, 1994). While MNI was used to account for the minimum number of mammals with carnivore marks represented in the sample, the second informed the counting per taxa or skeletal part categories.

Results
We found four bones (0.2% of the total) of megaherbivores and 24 bones (1.24% of the total) of smaller and indeterminate species with potential carnivore intervention. In addition, a detailed description of the marks is given in the supplementary information (Data S1). Below, we give a general overview of the most important damage found in each collection (Table 2 and Table S5) and provide general observations from the box-plot diagram ( Fig. 3): (i) A right tibia from the MCNV (nº 64-492) that corresponds to the ground sloth cf.
Scelidotheriinae gen (Tardigrada, Mylodontidae). This bone presents important furrowing on both epiphyses and pits and scores on the distal epiphysis, as well as on the posterior and medial faces of the diaphysis (Fig. 4). In the box plot diagram it can be observed that the measurements of these pits slightly overlaps with the maximum sizes of large carnivores (and outliers) from Pickering et al. (2004) and falls within the measurements presented by de los , but are slightly bigger than the Pampean case (de los . Nevertheless, this discrepancy could be due the bigger pit from MCNV that seems to be enlarged by postdepositional process (Data S1 and Fig. S1). They also coincide with the smaller sizes from Cueva  Table S5 shows that the log area coincides with the range for the rest of the sampled material; and (iv) At the MLP, one femur condyle from the notoungulate Toxodontidae (MLP 15-I-20-32) (Notoungulata; Toxodonta) was found with scratches ( Fig. 7). Moreover, in this collection 22 long bones of smaller species and two further indeterminate bones have fresh fractures, scratches, punctures/pits and crenulated edges (details of these marks are shown in Table S6) ( Fig. 8 to 10).
The box plot reveals the same trend for these pits and punctures as seen in the other cases.

Discussion
The information presented above suggests that the different types of bone marks found on both megamammal and small mammal remains were most likely inflicted by some large-sized carnivores that inhabited the Pampean region during the Pleistocene. Considering the limited evidence available from this region, the data presented here is crucial for exploring different predator-prey and/or scavenging scenarios, at a coarse scale.
It is clear that carnivores with an important capacity for bone modification and/or consumption would have been responsible for the various marks observed. Even though felids such as Smilodon or Puma could have produced some bone-damage, as observed in some studies (Van Valkeburgh & Hertel, 1993;Marean & Ehrhardt, 1995;Muñoz et al., 2008;Kaufmann et al., 2016), their reduced bone-breaking potential rules them out as the principal generator of the feeding traces recorded. Furthermore, it is worth mentioning that the highly specialised visceraeating dentition of the dirk-toothed Smilodon would have prevented this animal from feeding on carrion unlike other scimitar-toothed predators (e.g., Homotherium) (Palmqvist et al., 2007).
Identifying potential agents of the megamammal tooth-marks Based on the box plot comparisons (Fig. 3), the marks on the samples in this study best match those made by the giant Pampean Chapalmalania (de los . This procyon had previously been compared with a bear, although according to  the cranial configuration is more similar to that of hyenas. From the information presented by Pickering et al. (2004), it seems that the damage inflicted also coincides to some degree with that made by large African carnivores, such as large canids, spotted hyenas and lions, or the smaller marks realised by Panthera onca mesembrina (Martin, 2016). These African species correspond to sizes 2 or 3 in the Bunn ranking (1986). Cross-referencing these sizes with the Pampean carnivores, they coincide with several ursids, felids and canids, although some Pampean species were larger, such as Smilodon populator, size 4, and Arctotherium angustidens, size 5 (Table 3).
Moreover, the reports from the various South American sites involving pitting and/or punctures show a similar range of values as seen in this study ( Table 4). Most of this information could not be plotted, as the number of marks found at each site was too low to be able to input them into the calculation. Nevertheless, it can be observed that the majority range from 5 to 10 mm in size (those from Cueva del Milodón are larger, as shown in the box-plot). According to this data, different members of the Pampean large-carnivore guild would have produced the bone damage observed on the samples from the various museums. To determine which carnivores were involved, we must relate the marks to the types of bone damage generated by the potential ursid, felid and canid taxa.
The MCNV cf. Scelidotheriinae gen. tibia is the bone that presents the most significant carnivore interventions. A combination of pitting, scratches and important furrowing was observed, on both the epiphyses and medial faces. Even though all three groups of carnivores were capable of leaving these types of marks, certain characteristics allow us to relate this damage to ursids. In particular, the group of aligned pits imprinted on the medial rim (Fig. 4A) of the distal epiphysis is planar that could indeed have been made by the premolars or molars of ursids (Haynes, 1983).
In contrast, the parallel, V-shaped tooth marks on the posterior face ( Fig. 4C and Fig. 4E) could be related to a series of incisors and canines and would coincide with the dragging action of a straight incisor arcade (Biknevicius, Van Valkenburgh & Walker, 1996). On the other side, the parallel scores, like those seen on the distal metadiaphysis (Fig. 4B), are also generally characteristic of ursids (Haynes, 1983;Saladié et al., 2013). In addition, the intensive furrowing coincides with the bone-breaking capacity of this animal (Soibelzon et al., 2014). Other damage typical of ursids observed on the tibia includes the elongated gouge on the lateral side of the articular face (Fig. 4A) and the quadrangular-shaped grooves on the medial face of the diaphysis ( Fig. 4D) (Burke, 2013;Saladié et al., 2013). That being said, these grooves, and the gouges  (Saladié et al., 2013). Also, according to current research, this damage should be superficial, a feature not observed on this bone (Haynes, 1983;Saladié et al., 2013). To this respect, some authors suggest that the damage produced by ursids is less intense than that inflicted by other groups (Haynes, 1983;Arilla et al., 2014;Sala & Arsuaga, 2016), a pattern not observed here. Consequently, more than one animal may have participated in imprinting the complex and producing the marks observed on this tibia. If that is the case, Panthera onca could have been involved, too. This species also possessed straight incisive arcades (Biknevicius, Van Valkenburgh & Walker, 1996) that could have produced the elongated V-shape marks (Haynes, 1983) on the posterior face. The important furrowing noticed at both ends of the bone is also consistent with this felid's damage-producing capacity (Martin, 2008(Martin, , 2016Domínguez-Rodrigo et al., 2015). The humerus of Glossotherium robustum housed in the MNHN has suffered less bone loss than the tibia. Feeding marks on this element have several characteristics that could indicate its consumption by Arctotherium. As observed on the tibia, the short, wide scratches present on the condyle and the wide, elongated, superficial pitting, agree with actualistic studies of ursid marks ( Fig. 5B and Fig. 5C) (Haynes, 1983;Burke, 2013;Saladié et al., 2013). Nevertheless, the presence of V-shape punctures in the trochlea (Fig. 5B), characteristic of felids rather than ursids, means that other taxa, such as Panthera onca, cannot be ruled out (Haynes, 1983). Both groups were capable of furrowing the epiphysis (Martin, 2008;Arilla et al., 2014;Domínguez-Rodrigo et al., 2015) as observed on the trocheal part of the bone (Fig. 5D). The furrowing on the MNW Glossotherium robustum humerus is more ambiguous than the marks on the other two bones, since various taxa could have inflicted this type of damage on cancellous bone ( Fig. 6A to 6D). The cusp that made the puncture could have been on a secodont tooth from a felid or canid (Fig. 6B). Both these groups have the capacity to damage and destroy cancellous Carnivore-marked bones represent only 18.11% (NISP: 25) of the total assemblage. The low proportion found at this site could have been influenced by its location in running water. As explained by Ameghino, ([1889] 1916) the material from this site was scattered along a 20 m stretch on both sides of a channel. Therefore, the current may not only have dispersed the primary association, but also mixed it with bony remains not originally consumed by the carnivore/s involved. This may also have influenced the skeletal assemblage, including the paucity of axial parts, resulting from density-mediated destruction or the winnowing of lighter axial bones.
Nevertheless, the fact that carnivores mark 18.11% of the bones also indicates that a basic level of primary association remained when this material was collected. The presence of the Toxodontidae femur and other smaller bones with carnivore marks indicates that a MNI of 2 animals were consumed in the location itself. In addition, the dominance of fractured long bones could, partly, have been the result of carnivore activities that transported limbs to this area.
Consequently, the carnivore/s involved in the formation of the collected assemblage must have had the capacity to break long bones and/or the ability to predate upon megamammals. In this sense, given the absence of specialised bone-crushers in the Americas, some type of canid may have been responsible for the described interventions. It is likely that either Theriodictis platensis or Protocyon scagliorum from the Ensenadan Stage/Age generated these marks, as also inferred In any event, although the proportion of carnivore marks that we have found on bones of megamammals is relatively low, this precludes the conclusion that the sites where the remains were originally collected represented the den of a hypercarnivore or bone-cracking species.
Other potential carnivores specialising in medium-sized and/or small taxa, such as Canis nehringui or Dusicyon avus, could have fed on the megaherbivore community during the late Pleistocene (Prevosti & Vizcaíno, 2006;Prevosti, Tonni & Bidegain, 2009). At ca. 14.000 cal yrs BP  Homo sapiens also became part of the carnivore guild. Humans not only Manuscript to be reviewed scavenged megamammal carcasses , but were also more successful hunters of these animals than the existing carnivores (Cione, Tonni & Soibelzon, 2009).

Megamammal carcass consumption during the Pleistocene
Considering the skeletal elements, bone mark locations, and the level of use of the bones, it seems most likely that these marks represent the final stages of megamammal carcass consumption.
(i) Marks on the tibia and the humeri are situated on the epiphysis, both the articular surface and metadiaphyses. In a hunting event, carnivores that have access to a large mammal usually begin to feed on the abdominal part, later moving to femoral muscle masses, leaving some marks on the distal epiphyses and diaphyses (Haynes & Klimowicz, 2015). Forelimbs are usually consumed later, since the skin is harder in these areas (Haynes, 1982;Haynes & Klimowicz, 2015). The same usually happens with lower limb bones, such as the tibia, due to their smaller quantities of meat (Haynes, 1982;Blumenschine, 1986;Haynes & Klimowicz, 2015). The intense gnawing of the cf. Scelidotheriinae gen. tibia, both on the distal epiphysis and medial face of the diaphysis, as well as, to a lesser degree, on the proximal epiphysis, implies that this element was fully exploited. The presence of marks on the diaphysis indicates that even the hardest part of the shaft was utilised. The same is true for both Glossotherium robustum humeri. The damage to the distal epiphyses was inflicted in subsequent stages and not at the beginning of the consumption sequence. The presence of furrowing on the three elements implies that the various carnivores involved were consuming a substantial amount of bone. In the case of the MLP assemblage, the dominance of broken long bone diaphyses indicates access to within-bone nutrients, relating to the last stages in the consumption sequence (Binford, 1981;Haynes, 1982;Blumenschine, 1987;Capaldo & Blumenschine, 1994).
(ii) Intensity of carcass use is related to resource availability (Haynes, 1980(Haynes, , 1982Van Valkeburgh & Hertel, 1993;Delaney-Rivera et al., 2009), the size of the hunting pack (Van Valkenburgh et al., 2016), or multiple carnivore taxa involvement (Pobiner & Blumenschine, 2003;Delaney-Rivera et al., 2009). In general terms, large animal tissue is usually conserved for longer once dead (Blumenschine, 1987) and their bones have fewer marks than seen on bones of smaller species (Yravedra, Lagos & Bárcena, 2011;Domínguez-Rodrigo et al., 2015). As the easy-to-access meat is consumed, carnivores tend to eat the remaining parts of the carcass and inflict more significant damage to the bones (Binford, 1981;Haynes, 1982;Blumenschine, 1986;  The described feeding traces therefore appear to indicate that during the Pleistocene, different species within the large carnivore guild would have accessed and consumed megamammal bones and/or the marrow of smaller animals, in the final stages of a consumption sequence. Although discussion of how the animals were predated is difficult without more contextual information, given the multiple possibilities for carnivore exploitation of megamammal carcases (Pobiner & Blumenschine, 2003), two possible extreme scenarios are considered here: the marks described resulted from a first access (hunting) event and/or secondary access (scavenging) activity. The first case would involve the same group of carnivores killing and consuming the edible muscle tissues and then exploiting bones and within-bone nutrients. Early access to the carcass of an animal that had died a natural death by the same carnivore group can be also included in this situation (Blumenschine, 1986). Alternatively, after the death of the animal (either from natural causes or hunting activities), various carnivore taxa could have fed on a single carcass. In this second situation, one group would have consumed the primary edible tissues of the bony elements, and, at a later stage, the bones and marrow would have been exploited by other carnivores.
These interventions resulting from hunting and/or scavenging events indicate that in both cases, megamammal carcasses were completely exploited by various members of the large-sized carnivore guild in the region. Our samples belong to different time periods within the Pleistocene ( Fig. 2 and Table 1). This provides weak but positive evidence suggesting that consumption of edible tissues as well as the bony elements and/or marrow by different carnivore groups was a pattern that occurred repeatedly throughout that period. Full exploitation of carcasses is expected, at least periodically when food is scarce and/or more carnivore species are present, as has been              Manuscript to be reviewed  Table 2.
Measurements of pits, punctures and scores. Presence of furrowing or crenulated edges was also indicated.
1 Table 2. Measurements of pits, punctures and scores. Presence of furrowing or crenulated edges 2 was also indicated.