δ13C change through the GABI temporal interval
Bioapatite δ13C values for Rodentia, Notoungulata, Litopterna, Pilosa, Cingulata, Proboscidea, Perissodactyla, Artiodactyla, Sparassodonta and Carnivora are shown in Fig. 2 and Supplementary Tables S1 and S2 As for the dietary-bioapatite δ13C enrichment (ε*diet-enamel), we use ε*diet-enamel of +14.1‰21 for all orders (except for rodents and xenarthrans), +12.8‰ for rodents22, and +15.6‰ for xenarthrans23. Carnivorans´ trophic fractionation was also accounted for24 (see Supplementary Text).
Bioapatite δ13C (‰, VPDB) values of 10 orders of herbivore and carnivore mammals from the La Pampa and Buenos Aires provinces between the Late Miocene (Chasicoan) and the Late Pleistocene (Lujanian). (A) Rodentia, (B) Notoungulata, (C) Litopterna, (D) Pilosa, (E) Cingulata, (F) Proboscidea, (G) Perissodactyla, (H) Artiodactyla, (I) Sparassodonta, (J) Carnivora. The grey bars represent the vegetation δ13C cut-off values between a C3-dominated diet, an intermediate C3-C4 diet, and a C4-dominated diet. The lightest grey denotes a ε*diet-enamel of +14.1‰21, whereas the darkest one corresponds to a ε*diet-enamel of +12.8‰ for rodents22, and ε*diet-enamel of +15.6‰ for xenarthrans23. Sparassodonts and carnivorans show raw δ13C data (solid symbols) and δ13C data corrected for the trophic offset (open symbols), adding 1.3‰ to the raw values24. Locality abbreviations are the same as in Fig. 1. *Late Pleistocene (Lujanian) perissodactyl samples come from Luján, Río Salado, Arroyo Tapalqué, Cascada del Paleolama, Zanjón Seco, Paso Otero and Arroyo Chasicó localities (Supplementary Table S1). **Late Pleistocene (Lujanian) artiodactyl samples come from Playa del Barco, Santa Rosa, Cascada del Paleolama and Paso Otero localities (Supplementary Table S1). Zig-zag lines in the chronology tables denote non-represented ages: the Montehermosan is followed by the Chapadmalalan, the Marplatan, the Ensenadan, the Bonaerian, and the Lujanian.
The most noticeable feature is an overall increase of δ13C values for some of the endemic herbivore mammals (rodents, notoungulates, pilosians) recorded between the Late Miocene and the Pleistocene (Fig. 2). The increase in the bioapatite δ13C values of these taxa is intrinsically related to a shift in the use of plant resources from C3-dominated vegetation to mixed C3-C4 vegetation. The expansion of C4-dominated grasslands in Argentina has been recorded at ~8–7 Ma11. The rising of C4 plants at the Pampean region is framed within the timing of the “Edad de las Planicies Australes” (“Age of Southern Plains”), a period between 11 and 3 Ma characterized by an increase in aridity, a gradual decrease in temperatures and the establishment of vast grasslands after the retreat of the Paranean Sea18. The shift in our stable isotope data support the premise posed by MacFadden et al.10, who stated that after ~8 Ma, the Pampean ecosystem was characterized by a mix of C3 trees, shrubs and grasses, as well as C4 grasses. Unfortunately, there is not a good paleobotanical/palynological record in the Pampean deposits2, which makes the study of stable isotopes on vertebrate bioapatite of utmost interest in the understanding of vegetation evolution in the Pampean area during the late Cenozoic.
Representatives of the orders Rodentia and Notoungulata show relatively high bioapatite δ13C values at the Chasicoan (∼9.5 Ma, Late Miocene), including Arroyo Chasicó and Cerro La Bota25 localities, that remain close to the δ13C cut-off value between a C3-dominated diet and a C3-C4 diet (Fig. 2A,B). As the proportion of C4 plants within Pampean ecosystems may not have been significant until the latest Miocene, these high bioapatite δ13C values may be pointing to the consumption of C3 plants from xeric environments. However, we cannot fully discard the fact that C4 plants may have already found favorable conditions to thrive at this early age in the Pampean area as also suggested by Hynek et al.12 for the Northwest of Argentina. Rodentia, Notoungulata and Pilosa fed on a C3-C4 environment later in the Late Miocene/Early Pliocene Caleufú26 and Early Pliocene Farola Monte Hermoso27 localities (Fig. 2A,B,D). This trend towards an increase in the use of C4 resources is not as marked in Litopterna and Cingulata, which show δ13C values that remain mostly within a C3-dominated diet throughout the studied temporal interval (Fig. 2C,E). During the Pleistocene, notoungulates and pilosians show intermediate C3-C4 diets (Fig. 2B,D), whereas endemic litopterns and cingulates along with immigrant proboscideans, perissodactyls and artiodactyls show diets at the interplay between a C3-dominated diet and a mixed C3-C4 diet (Fig. 2C,E–H).
Rodentia
The dietary change recorded in rodent δ13C values is detected at the Late Miocene/Early Pliocene Caleufú and Early Pliocene Farola Monte Hermoso localities (Fig. 2A). This shift is statistically significant when compared to older localities from the Late Miocene (Arroyo Chasicó and Salinas Grandes de Hidalgo) (Supplementary Table S3). In our study, evaluated remains from Caleufú include Caviidae and Dinomyidae, whereas Farola Monte Hermoso is represented by Caviidae and Chinchillidae. All of them have a wide range of δ13C values that was not observed among these same families in older localities (Fig. 2A, Supplementary Table S2). Therefore, studied rodents were able to incorporate C4 resources to their previous C3-dominated diet as soon as C4 vegetation was available in the region. This is in agreement with Hynek et al.12, who observed a wide δ13C range on rodent tooth enamel in the Pliocene deposits from Northwestern Argentina, evidencing the consumption of C4 vegetation at this time.
Notoungulata
Chasicoan notoungulates from the Arroyo Chasicó (Late Miocene) locality are represented herein by different taxa of the families Toxodontidae, Hegetotheriidae, Mesotheriidae and Homalodotheriidae (Supplementary Table S2). Only the first three are represented in Late Miocene Cerro La Bota (Chasicoan), Telén, Quehué and Salinas Grandes de Hidalgo (Huayquerian). Chasicoan notoungulates show a wide δ13C range indicative of dietary flexibility from more forested-arboreal areas to open grasslands within a C3-continuum (Fig. 2B, Supplementary Table S2). Some of the Arroyo Chasicó and Cerro La Bota notoungulate δ13C values surpass the vegetation δ13C threshold value between a C3-dominated diet and a C3-C4 diet. It is likely that they may be recording a C3 diet within open xeric conditions as C4 plants were not still present in significant proportions at this time11, although this result may as well point to the existence of favorable climatic and environmental conditions for C4 photosynthesis at this early time as suggested by Hynek et al.12. Huayquerian notoungulate δ13C record starts with a decrease indicating a pure-C3 diet at this time period (Fig. 2B, Supplementary Table S2). Therefore, the shift in tooth enamel δ13C values represents a real change in resource use (not family-biased) for the faunas assigned to the Huayquerian. As observed in rodents, notoungulates record a significant shift in their δ13C values from the most modern Neogene faunas onwards (Fig. 2B, Supplementary Tables S2 and S3). In this context, the notoungulates from the Late Miocene/Early Pliocene Caleufú and Early Pliocene Farola Monte Hermoso localities, represented in our study by Toxodontidae, Hegetotheriidae and Mesotheriidae, depict a wide range of tooth enamel δ13C values around the boundary between C3-dominated and C3-C4 diets (Fig. 2B, Supplementary Table S2). Our results suggest that notoungulates changed their dietary preferences at the end of Miocene-beginning of the Pliocene, taking advantage of the expansion of C4 vegetation in the region, with hegetotherids and mesotherids likely developing a more flexible resource use and toxodontids adopting a more specialised intermediate C3-C4 diet. Toxodontids (Toxodon) from the Quaternary Late Pleistocene Playa del Barco and Santa Rosa sites retained this newly adopted dietary behavior becoming important dwellers of intermediate C3-C4 areas (Fig. 2B, Supplementary Tables S2 and S3). Our results agree well with previous stable isotope analyses on notoungulates from other fossil sites in South America12,20,28. These studies argued that the notoungulate dietary shift is pinpointed between 7 and 5.6 Ma and that this shift was independent from the notoungulate family, as also suggested by our results.
Litopterna
Litopterns from the Late Miocene Arroyo Chasicó (Chasicoan) and Salinas Grandes de Hidalgo (Huayquerian) localities show a C3-dominated diet, with a wide range of δ13C values indicative of a continuum from forested areas to more open conditions (Fig. 2C, Supplementary Table S2). These results agree well with the δ13C values from Salinas Grandes de Hidalgo provided by MacFadden et al.10. Litopterns from the Early Pliocene Farola Monte Hermoso locality, represented by Macraucheniidae and Proterotheriidae, show a narrower range of tooth enamel δ13C values, but still within a C3-dominated diet (Fig. 2C, Supplementary Table S2). Overall, our litoptern δ13C record does not track the expansion of C4 plants in the region, contrary to rodents and notoungulates. With respect to the Quaternary taxa, Macraucheniidae from the Early Pleistocene Toscas del Río de La Plata locality (Ensenadan) still had a C3-dominated diet, although with some values close to the C3-C4 transition, whereas Late Pleistocene (Lujanian) Macrauchenia from Arroyo Chasicó, Paso Otero and Cascada de Paleolama shows a slight shift towards more mixed C3-C4 diet, although the low number of samples from this time period precludes us from making a more solid statement (Fig. 2C, Supplementary Table S2). Our results are supported by previous studies from the Quaternary localities of Camet Norte (Argentina)15 and Tarija (Bolivia)29.
Pilosa and Cingulata
A diagenesis test was performed to check whether the analysed xenarthran orthodentine retained the original isotopic signal (Supplementary Text, Supplementary Table S4). Bocherens et al.16 measured the difference between collagen and bioapatite δ13C values within a same specimen and suggested that glyptodonts (Cingulata) and ground sloths (Pilosa) would have been herbivorous. Although in our record Xenarthra data points are scattered and scarce, it is notable the shift observed in the pilosian δ13C values from a C3-pure diet at the Late Miocene to a C4-based diet at the Pliocene and Pleistocene (Fig. 2D). Nothrotheriidae from the Late Miocene Arroyo Chasicó (Chasicoan) and Mylodontidae from Salinas Grandes de Hidalgo (Huayquerian) localities depict a δ13C value indicative of a diet based on C3 items. Later on, Mylodontidae from the Late Miocene/Early Pliocene Caleufú and Early Pliocene Farola Monte Hermoso localities record higher δ13C values pointing to a shift towards the consumption of C3-C4 vegetation from more open spaces, tracking the expansion of C4 plants in the region (Fig. 2D, Supplementary Table S2). Finally, pilosians from the Late Pleistocene Playa del Barco and Santa Rosa localities, represented by Mylodontidae and Megatheriidae, show a marked dietary change, with an overall consumption of C3-C4 resources (Fig. 2D, Supplementary Table S2). Cingulata orthodentine δ13C record (Fig. 2E, Supplementary Table S2) does not show such a marked shift through time: pampatheriids from the Late Miocene Arroyo Chasicó (Chasicoan) and glyptodontids and dasypodids from Telén, Quehué and Salinas Grandes de Hidalgo (Huayquerian) localities depict δ13C values indicative of a diet based on C3 items, whereas glyptodontids from localities from the Late Pleistocene Buenos Aires city area show a wide range of δ13C values, with most of individuals consuming C3-dominated resources and others incorporating mixed C3-C4 resources (Fig. 2E, Supplementary Table S2). Our Quaternary xenarthran results are in line with those reported by Bocherens et al.15, whose collagen δ13C analyses on Late Pleistocene cingulates and pilosians from the Buenos Aires province point to a greater consumption of C4 resources in the latter.
Proboscideans
Gomphotheriidae (Stegomastodon) from the Early Pleistocene Toscas del Río de La Plata and Late Pleistocene Playa del Barco and Buenos Aires city area localities shows wide ranges of δ13C values pointing to a flexible diet, which would have included pure-C3 and mixed C3-C4 plant resources (Fig. 2F, Supplementary Table S2).
Perissodactyla
Equids from the Early Pleistocene Toscas del Río de La Plata locality have δ13C values indicative of a C3-dominated diet. On the other hand, Equus and Hippidion from different Late Pleistocene (Lujanian) sites show a broad variability of δ13C values pointing to consumption of C3 and mixed C3-C4 vegetation (Fig. 2G, Supplementary Table S2), similarly to what is observed in the case of gomphotherids.
Artiodactyla
Artiodactyls from the Early Pleistocene Toscas del Río de La Plata locality are represented by Camelidae. Lama sp. records significantly lower δ13C values indicative of a browsing diet in a more wooded area, when compared to other camelids, assigned to?Palaeolama sp., with higher δ13C values pointing to a more intermediate diet (t = −1.361, p = 0.042) (Fig. 2H, Supplementary Table S2). At the Late Pleistocene Playa del Barco site, artiodactyls are represented by the cervid Morenelaphus, which shows the lowest δ13C values among our artiodactyl record (Supplementary Table S2), pointing to a browsing C3-diet and occupation of wooded areas. At the Late Pleistocene Santa Rosa locality, the camelids Lama and Hemiauchenia seem to have partitioned resources to some extent, with the former showing significantly lower δ13C values and the latter recording higher δ13C values (t = −3.492, p = 0.025) (Supplementary Table S2).
Dietary flexibility evidenced by northern immigrants (proboscideans, perissodactyls and artiodactyls) as shown by our δ13C values and other isotopic studies from North and South America13,15,30,31,32,33, and by extension, their plasticity in niche occupation, is in agreement with the habitat theory proposed by Vrba34 according to which generalists and open biome specialists from North America experienced a successful radiation throughout South America34,35.
Sparassodonta
Endemic sparassodonts are represented by the carnivorous and scansorial borhyaenoid Lycopsis from the Late Miocene Arroyo Chasicó site, and the hypercarnivorous thylacosmilid Thylacosmilus from the Late Miocene Quehué and Salinas Grandes de Hidalgo localities. Sparassodont δ13C record reveals a consumption of prey from C3 open areas (Fig. 2I, Supplementary Table S2). We used MixSIAR mixing model to estimate the proportions of source (prey) contributions to a consumer (predator)36 (see the Materials and methods section and the Supplementary Text). Late Miocene sparassodonts preferentially preyed on notoungulates (Supplementary Table S5).
Carnivora
Immigrant Carnivora from the Late Pleistocene Playa del Barco and Santa Rosa sites are represented by the hyper-carnivorous felid Smilodon, whose δ13C values point to the ingestion of prey species from mixed C3-C4 areas (Fig. 2J, Supplementary Table S2). MixSIAR results indicate that the analysed specimens from Playa del Barco preferentially preyed on notoungulates, whereas analysed specimens from Santa Rosa preferentially preyed on perissodactyls (Supplementary Table S5). In our study, Smilodon potential prey from mixed C3-C4 areas point to an intermediate situation between the ones provided by Cotte et al.37 and Bocherens et al.15. Cotte et al.37 carried out δ13C analyses on Late Pleistocene Smilodon tooth enamel from the Buenos Aires province and suggested that it mainly preyed on species from wet areas with C3 vegetation. Bocherens et al.15 performed δ13C analyses on Late Pleistocene Smilodon collagen from the same province, but their results point to Smilodon’s consumption of prey from open landscapes. This disparity may be indicative of its ability to hunt on a wide range of habitats and species.
Differences between Sparassodonta and Carnivora δ13C values are statistically significant (t = −17.746, p < 0.001) evidencing a different resource use and indirectly mirroring the expansion of C4 vegetation.
Niche occupation through the GABI
Isotopic niches were assessed by using a Bayesian approach described in the Materials and methods section and the Supplementary Text. In the Pre-GABI Late Miocene Arroyo Chasicó locality, rodents, notoungulates and litopterns show some degree of ellipse (niche) overlap (Fig. 3A, Supplementary Text). Overall, taxa ingested vegetation from C3 areas, although notoungulates seems to have significantly incorporated C3 plants from more open spaces, whereas rodents and litopterns consumed C3 vegetation from more wooded areas (Supplementary Table S6). δ18O are driven by particular water use, ecologies and ecosystem preferences among different taxa and therefore, we consider this isotopic system in the assessment of habitat use, which along with the evaluation of resource use can help us depict niche occupation and width. In this sense, fitted loess δ18O curves along time depicted in Fig. S1 show different patterns for each order probably reflecting particular water use, rather than recording global climatic trends. At Arroyo Chasicó, rodents show significantly lower δ18O values (Fig. 3A, Supplementary Table S6) when compared to notoungulates and litopterns, which may point to ingestion of water (either via drinking or food) subject to a lower degree of evaporation. Rodents have narrower home ranges when compared to large mammals, so they could be tracking more local conditions, whereas notoungulates and litopterns would provide a more integrated signal of their broader home ranges38. In this sense, notoungulates and liptoterns do not show significantly different δ18O values (Fig. 3A, Supplementary Table S6), which point to ingestion of water subject to similar hydrological conditions. The δ13C-δ18O ellipses could not be retrieved for xenarthrans and sparassodonts since they have fewer than three datapoints, although most of them fall within a C3-dominated area (sparassodonts would have used prey from these areas). They show overall low δ18O values indicative of water ingestion with a lower degree of evaporation (Fig. 3A).
Standard ellipse areas (SEAs) generated by SIBER representing isotopic niches for selected Neogene and Quaternary localities. (A) Arroyo Chasicó, (B) Salinas Grandes de Hidalgo, (C) Farola Monte Hermoso, (D) Playa del Barco, and (E) Santa Rosa. The grey bars represent the transition between a C3-dominated diet, an intermediate C3-C4 diet and a C4-dominated diet. Sparassodonta and Carnivora ellipses show raw δ13C data (solid ellipse) and δ13C data corrected for the trophic offset (open ellipse), adding 1.3‰ to the raw values24.
At the Late Miocene Salinas Grandes de Hidalgo locality, which evidences the first pulse of the GABI through the presence of the procyonid Cyonasua (not sampled in this study), there is a more intense ellipse (niche) overlap among the analyzed taxa (Fig. 3B). As observed in the case of the Arroyo Chasicó locality, all taxa consumed C3-dominated vegetation (or preyed on taxa with a pure C3 diet in the case of sparassodonts), but their δ13C values are more condensed (Supplementary Table S6) and reflect their coexistence in a C3 woodland to open C3 grassland (Fig. 3B). Also δ18O values show a more intense overlap (Supplementary Table S6), pointing to ingestion of water subject to similar hydrological conditions. The more intense ellipse (niche) overlap and the contraction of ellipse (niche) breadth observed at this locality when compared to the Arroyo Chasicó locality may be indicative of tougher conditions related to increasing aridity reported previously for the end of the Miocene26,39.
During the latest Miocene and Early Pliocene, the expansion of C4 plants at the Pampean area occurred10,12. If at this time some of the endemic herbivore taxa were able to incorporate C4 vegetation, this may have eased the competition among fauna and favored their sympatry. This is illustrated by the significant change in the δ13C values of rodents, notoungulates and pilosians at the Late Miocene/Early Pliocene Caleufú and Early Pliocene Farola Monte Hermoso localities (Fig. 2A,B,D). Farola Monte Hermoso rodents, notoungulates and pilosians retrieved ellipses that evidence a more intense use of open C3 and/or C4 vegetation when compared with previous times (Fig. 3C). Rodents show the widest range of δ13C values (Fig. 3C) pointing to the consumption of a broad array of resources that include both C3-dominated and mixed C3-C4 vegetation. Rodents also show significantly different δ18O values when compared to notoungulates and litopterns (Supplementary Table S6), pointing to ingestion of water subject to different hydrological conditions (less intense evaporation) probably related to more local conditions and the ecology of some groups (e.g., fossil capybaras-Phugatherium, Hydrochoeriinae, Caviidae, -are recorded associated to water bodies). Notoungulates and pilosians show high δ13C values indicative of the use of C3 and C4 plants from more open spaces when compared to litopterns (Fig. 3C; Supplementary Table S6). It is noticeable the fact that this relationship between notoungulates and litopterns is maintained since the Chasicoan, something that has been supported by other lines of evidence, such as dental and body morphology19. Overlapping δ18O values among notoungulates, litopterns and pilosians point to water use from sources subject to similar hydrological regime (Fig. 3C; Supplementary Table S6). Another interesting observation that is consistent at the Chasicoan, Huayquerian and Montehermosan (Fig. 3A–C) lies in the fact that rodent δ13C and δ18O values show a positive covariance. Lower resource δ13C values correspond to lower (drinking or food) ingested water δ18O and vice versa pointing to more closed or more open areas, respectively. This relationship is not observed in the case of larger taxa supporting the idea that rodents may be better recorders of ecosystem variability at a more local level.
North American herald taxa such as cricetids26 and procyonids40 are recorded in the Pampean region since the Late Miocene/Early Pliocene (Huayquerian). Then, the massive entrance of northern fauna took place once the Panama Isthmus was fully established. Therefore, the shift in vegetation use by endemic fauna evidenced by our isotopic data at the Caleufú (late Huayquerian) and Farola Monte Hermoso (Montehermosan) sites was not a consequence of pressure caused by new arrivals.
Global records point to optimal climatic conditions at this interval, when reorganized ocean circulation, perhaps associated with initial restriction of circulation between the Pacific and Atlantic, contributed to the Pliocene Warm Period between ~4.7 and 3.1 Ma41. Based on the faunal assemblage, warm conditions, with open xerophytic woodlands and grasslands, similar to today´s Chacoan phytogeographic province, but probably with greater humidity, were proposed for Farola Monte Hermoso locality. A global cooling coincident with expansion of ice in Antarctica happened at ~3.0 and 2.7 Ma, marking the beginning of the Quaternary42. The main GABI events are contemporary to this climatic deterioration, which accompanied by more arid conditions favored the development of open areas in South America and the establishment of a savanna corridor between North and South America facilitating the colonization by generalists and open biome specialists from North America34,35. South American endemic fauna experienced a large number of extinctions at the Late Pliocene (Sanandresian) and Early Pleistocene (Ensenadan)18. Although initial studies pointed to the arrival of northern immigrants as the trigger of this extinction events43, more recent research suggests that the entrance into the Pampean region was gradual and that the biomass represented by these new taxa was insignificant at this moment and therefore, the global cooling is proposed as the main triggering factor of the extinction18,44,45. The interaction with northern immigrants became more important since the Early Pleistocene (Ensenadan) and Middle Pleistocene (Bonaerian) and it was particularly keen during the Late Pleistocene (Lujanian). We lack isotopic data from the Bonaerian fauna, but our record of the Ensenadan and Lujanian endemic taxa, such as notoungulates and pilosians, shows a modification in their vegetation ingestion towards a more generalist intermediate C3-C4 diet (Figs. 2 and 3).
At the Late Pleistocene Playa del Barco locality, the faunal configuration is mainly characterized by the presence of large and megamammals. The analyzed taxa show a marked change in the resource use, with the full incorporation of mixed C3-C4 vegetation typical of grasslands at the Pampean area (Fig. 3D). Notoungulates, pilosians and perissodactyls almost exclusively consumed C3-C4, whereas proboscideans also incorporated plants from C3-dominated open areas (Fig. 3D, Supplementary Table S6). Felids (Smilodon) preyed on taxa that consumed mixed C3-C4 vegetation with mixing models pointing to notoungulates as their possible preferred prey (Fig. 3D, Supplementary Tables S5 and S6). Artiodactyla, exclusively represented by the cervid Morenelaphus, is the only order that shows a clear resource partitioning. Their significantly lower δ13C values evidences the consumption of C3 plants from more wooded spaces (Fig. 3D, Supplementary Table S6). Their higher δ18O values (Fig. 3D, Supplementary Table S6) may be related to either ingestion of drinking water or plant water subject to a higher degree of evaporation.
At the Late Pleistocene Santa Rosa locality, also large and megamammals are represented (Fig. 3E). Notoungulates and pilosians show similar dietary preferences (mixed C3-C4 vegetation) to that depicted at Playa del Barco. Artiodactyls are represented by camelids, whereas perissodactyls are represented by equids, both showing δ13C values indicative of a C3 diet from open areas and an intermediate C3-C4 diet (Fig. 3E). Similarly to that observed at Playa del Barco, felids (Smilodon) preyed on individuals that incorporated mixed C3-C4 vegetation, although the mixing model indicates that representatives of this locality probably preyed on perissodactyls (Fig. 3E, Supplementary Table S5). Santa Rosa taxa show overlapping and non-significantly different δ18O values (Fig. 3E, Supplementary Table S6), similar to those shown at Playa del Barco pointing to a similar hydrological regime.
Northern newcomers such as probocisdeans, perissodactyls and camelids had a flexible dietary behavior (from pure C3 to mixed C3-C4 diet, Fig. 2), which may have facilitated their successful radiation in South America as supported by their abundant fossil record in the subcontinent (see 13). The case of cervids is noticeable since they showed a more restricted diet based on plant resources from C3-dominated areas (Figs. 2 and 3D, Supplementary Table S2). Cervids may have successfully radiated into South America by taking advantage of resources from wooded areas within a savanna habitat. According to Cione et al.18, although the mammalian taxonomic diversity at the Late Pleistocene was high, the number of individuals per species and the total biomass were not probably elevated as most of the South American glacial ecosystems may not have been very productive. Following this statement, Playa del Barco and Santa Rosa localities may have represented low-productive ecosystems and therefore, resource partitioning would have been key to support endemic and non-endemic large mammalian coexistence. This seems to be the case for both localities, where notoungulates and xenarthrans seem to have relied more on mixed C3-C4 vegetation, gomphotheres, equids and camelids fed on C3 plants from open areas and intermediate C3-C4 vegetation, and cervids consumed C3 plants from wooded areas (Fig. 3D,E).
Source: Ecology - nature.com
