Journal of South American Earth Sciences 109 (2021) 103296

Contents lists available at ScienceDirect

Journal of South American Earth Sciences

journal homepage: www.elsevier.com/locate/jsames

Paleoenvironments and paleoecology of the Santa Cruz Formation (early-middle Miocene) along the Río Santa Cruz, Patagonia (Argentina)

Richard F. Kay a,*, Sergio F. Vizcaíno b,c, M. Susana Bargo b,d, Jackson P. Spradley e, Jos´e I. Cuitino˜ f a Department of Evolutionary Anthropology and Division of Earth and Ocean Sciences, Duke University, Durham, NC, 27708, USA b Division´ Paleontología Vertebrados, Facultad de Ciencias Naturales y Museo, Unidades de Investigacion,´ Anexo Museo, Av. 60 y 122, 1900, La Plata, Argentina c Consejo Nacional de Investigaciones Científicas y T´ecnicas (CONICET), Argentina d Comision´ de Investigaciones Científicas, Provincia de Buenos Aires (CICPBA), Argentina e North Carolina State University - College of Veterinary Medicine, Department of Molecular Biomedical Sciences. Raleigh, NC, USA f Instituto Patagonico´ de Geología y Paleontología Centro Nacional Patagonico,´ Puerto Madryn (U9120), Argentina

ARTICLE INFO ABSTRACT

Keywords: The continental early-middle Miocene Santa Cruz Formation (SCF) in Austral Patagonia contains the best record Ecometrics of South American mammalian faunas prior to the Great American Biotic Interchange (GABI) and is of particular Paleobiology interest because it is the best preserved high-latitude continental biotic record in the Southern Hemisphere Paleoclimate spanning the mid-Miocene Climatic Optimum. Through intensive fieldwork we recovered numerous fossil ver­ Mammals tebrates, mostly mammals, from the SCF along the Río Santa Cruz (RSC), the type area for the formation and its Neogene South America fauna. We examine whether the SCF fauna differed among three distinct temporal intervals of the SCF spanning, Frugivore problem from the oldest to youngest, the Atlantic coastal suite of localities Fossil Levels (FL) 1–7, at about 17.4 Ma, through localities in the RSC Barrancas Blancas (BB), between ~17.2 and ~16.3 Ma, and Segundas Barrancas Blancas (SBB), between ~16.5 and ~15.6 Ma. With the objective of reconstructing paleoenvironmental and community structure of these RSC faunas, we compared them with 55 extant lowland mammalian localities ◦ ◦ across South America from 8 N to 55 S latitude representing a wide range of seasonality and, annual rainfall and temperature, as well as canopy height and net primary productivity, sampling communities ranging from tropical rainforest to semi-arid steppe. Extant nonvolant mammalian genera at each locality were assigned a body size interval and niche parameters reflecting diet and substrate use, from behavioral data in the literature. Extinct genera were assigned similar niche metrics on the basis of their morphology. From the generic niche parameters, we compiled indices and ratios that express vectors of the community structure of each fauna, including the total number of genera, the pervasiveness of arboreality, frugivory, and browsing, and the relative richness of predators to their prey. The community structure variables were used to model community structure of the fossil localities based on uniformitarian principles. The fossil sample includes 44 genera of mammals from FL 1–7, 38 genera from BB, and 44 genera from SBB. The Simpson Coefficients of faunal similarity among the fossil localities are no greater than expected on the basis of the geographic distances among them, and do not suggest any apparent climatic differences. Based on the models we obtained no significant differences in MAP (Mean Annual Precipitation) for FL 1–7, BB and SBB, with mean estimates of 1635 mm, 1451 mm, and 1504 mm, with the confidence intervals for the estimates overlapping widely. MAT (Mean Annual Temperature) estimates ◦ ◦ ◦ are between ~21 C and ~22 C for FL 1–7 and SBB, possibly lower at 16 C for BB, but with a wide and ◦ ◦ overlapping range of estimates. Temperature seasonality is modest (3 C to 4 C) and similar for all localities. Canopy heights exceed 20 m for all sites. Despite these geographic and inferred climatic similarities, the presence of certain key taxa (e.g., the caviomorph rodent Prolagostomus and the typothere Pachyrukhos) together with an increased overall abundance and richness of rodents with ever-growing cheek teeth suggests a trend to aridifi­ cation in the upper part of the SCF at SBB compared with FL 1–7 and BB. Taken together, we propose that the SCF paleoenvironment consisted largely of semi-deciduous forests ranging into savannas with gallery-forest com­ ponents. This range of habitats occurs today where the mesic inland Atlantic forests of Southern Brazil,

* Corresponding author. E-mail addresses: [email protected] (R.F. Kay), [email protected] (S.F. Vizcaíno), [email protected] (M.S. Bargo), [email protected] (J.P. Spradley), [email protected] (J.I. Cuitino).˜ https://doi.org/10.1016/j.jsames.2021.103296 Received 9 December 2020; Received in revised form 23 March 2021; Accepted 23 March 2021 Available online 29 March 2021 0895-9811/© 2021 Elsevier Ltd. All rights reserved. R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

northeastern Argentina and eastern Paraguay give way northwestward into the more xeric Paraguayan Gran Chaco. These interpretations are in general agreement with other sources of evidence from sedimentology, paleosols, isotopes, paleobotany and other faunal elements. We highlight the value of focusing paleoenvir­ onmental and paleocological studies of the SFC on stratigraphically and geographically confinedsamples instead of on the entire temporal and geographic distribution of the SCF based on historical collections with limited provenance. The Santacrucian can be considered a model to the study of South American faunas after the arrival of hystricomorph rodents and anthropoid primates but before GABI.

1. Introduction a renewed background and vision. From a paleoecological perspective, in the last decade there have The continental early-middle Miocene (Burdigalian-early Langhian) been two different approaches for the understanding of the biota and fossil record of the Santa Cruz Formation (SCF) from Austral Patagonia, environments of the SCF: either considering stratigraphically and represents the biota that has most impacted historically and conceptu­ geographically restricted samples obtained through exhaustive field ally the understanding of the Cenozoic biotic evolution of South America work efforts (Kay et al., 2012; Vizcaíno et al., 2010) or globally along the prior to the Great American Biotic Interchange (GABI), when South entire temporal and geographic distribution of the formation based on America was mostly isolated from other Continents . Moreover, the SCF bibliography and historical collections (Croft, 2013). contains the best preserved high-latitude continental biotic record in the The approach by Vizcaíno et al. (2010) and Kay et al. (2012) was Southern Hemisphere providing further insights into mid-Miocene based on fossils collected from geographically and stratigraphically temperature and precipitation. It is well known that floras and faunas restricted sets of localities of the Atlantic Coast outcrops, which are often dependent on warm, wet conditions expanded to higher latitudes in in an excellent state of preservation, including partial or complete ar­ South America at that time (Frenguelli, 1953; Ortiz-Jaureguizar and ticulated skeletons, offering a unique opportunity to perform paleobio­ Cladera, 2006; Pascual and Odreman Rivas, 1971; Pascual and logical studies based on a form-function approach (Vizcaíno et al., Ortiz-Jaureguizar, 1990). The analysis of Hinojosa (2005) of early and 2012d). For these authors, the fact that many taxa come from certain ◦ middle Miocene floras at 33–34 S latitude between 21 and 13 Ma, levels deposited in a restricted time frame provides a narrow temporal suggests an increase in mean annual temperatures (MATs) from the late window that allows reliable paleoecological analysis (Kay et al., 2012; ◦ ◦ Oligocene-early Miocene from 16–17 C to MATs exceeding 22 C. Perkins et al., 2012). Vizcaíno et al. (2010) used the relationship be­ Likewise, at the beginning of the Miocene, the floras indicate an even tween population density and body size for estimating the on-crop more abrupt rise in rainfall, which culminated in values exceeding 1000 biomass (in kg/km2) of the species of these paleocommunities and mm. During the middle and late Miocene, rainfall subsided, reaching calculated their metabolic requirements. Kay et al. (2012) reconstructed minimum values of ~440 mm by around 10 Ma (Hinojosa, 2005). The the niche structure by identifying the number of species present, the expanded faunal samples and precise dating of our early-middle body size, locomotion, and diet of the mammalian genera at the suite of Miocene SCF localities (Trayler et al., 2020b) make them ideal candi­ localities FL 1–7 compared with similar kinds of data for extant faunas of ◦ dates for evaluating Patagonian climate and biota nearly 20 farther South America; using that data they interpreted the paleoclimate and south that Hinojosa’s localities, practically at the southern end of the paleoenvironment of FL 1–7. The FL 1–7 fauna was later studied by continent. Spradley et al. (2019) updating the approach of Kay et al. (2012) to Historically, the record from the Río Santa Cruz (RSC) represents the derive paleoecological predictive models for the same Atlantic Coast first systematic and exhaustive effort for collecting and studying verte­ fauna and the Miocene La Venta fauna of Colombia. Finally, Rodrí­ brate fossils from Patagonia, planned by Francisco P. Moreno and guez-Gomez´ et al. (2020) estimated the biomass of the primary and executed by Carlos and Florentino Ameghino in 1887 (Brinkman and secondary consumers of the paleoecosystem to assess if the resources Vizcaíno, 2014; Fernicola et al., 2014, 2019b; Vizcaíno et al., 2013). For available would satisfy the nutritional requirements of all species of the following three or four decades, the material and intellectual results secondary consumers. of this first and subsequent expeditions undertaken by Carlos and pub­ The results of the foundational work by Kay et al. (2012) were pre­ lished by Florentino stimulated important academic institutions of the sented as the concluding chapter of a volume on the paleobiology of the world to obtain their own collections of fossils from the SCF (Vizcaíno Santacrucian biota recorded on the Atlantic Coast edited by some of the et al., 2013, 2016, 2017b). Especially during that period, the Santa Cruz authors of this contribution (Vizcaíno et al., 2012a). When considering fossils became an inescapable reference for comparing other vertebrate the future direction of their research program, Vizcaíno et al. (2012d) continental faunas, either older or younger. proposed to expand this approach to a more complete geographic and Conceptually, the Ameghino collections were crucial for under­ chronologic range of the SCF, recording the different assemblages at standing and setting the succession of Cenozoic faunas from Patagonia different levels and evaluating the ecological changes that occurred (Ameghino, 1906) and constituted the basis for the establishment of the during the time of deposition of the formation in different areas. The Santacrucian South American Land Mammal Age (SALMA) (Pascual aforementioned update of the RSC stratigraphy and fossil record pro­ et al., 1965). With the addition of fossils from other SCF localities, vides an invaluable opportunity to achieve that undertaking in the his­ especially those from the Atlantic coast outcrops, their abundance and torically and scientifically most significant location of the SCF and its quality makes them the best material for interpreting the taxonomic fossil record. richness and biological diversity of mammals in Patagonia after the The main goal of this contribution is to reconstruct the paleoecology mid-Cenozoic arrival of primates and rodents, but before the arrival of (paleoclimate and paleoenvironment) of two temporally-restricted and North American immigrants as part of the Great American Biotic distinct, mostly non-overlapping fossil faunas recovered by us along the Interchange (GABI) (Vizcaíno et al., 2012d). Recent intensive fieldwork RSC compared with the temporally constrained and slightly older FL 1–7 has refined the stratigraphy and the vertebrate fossil record of the SCF fauna of the Atlantic coast presented in Kay et al. (2012). We examine from the right bank of the RSC compiled in a volume edited by Fernicola whether the SCF fauna differed among three distinct temporal intervals et al. (2019a). In that volume, Fernicola et al. (2019b) proposed that the of the SCF spanning, from the oldest in the Atlantic coast FL 1–7, at exposures along the RSC should be considered the type area for the SCF about 17.4 Ma (Trayler et al., 2020b), to the youngest localities in the and its fauna. The work constitutes a starting point for comparisons with RSC Barrancas Blancas, between ~17.2 and ~16.3 Ma, and Segundas other early and middle Miocene exposures and faunas in Patagonia with Barrancas Blancas, between ~16.5 and ~15.6 Ma (Cuitino˜ et al., 2016,

2 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

2019a). The SCF is superbly exposed along coastal cliffs and wave-cut plat­ forms south of Río Coyle, including sites such as Anfiteatro,Estancia La 2. Background Costa, Canad˜ on´ Silva and Puesto Estancia La Costa (Fig. 1). In this area, the SCF is lithologicaly similar to that of the RSC, including thicknesses 2.1. Geological setting of a maximum of 100 m (Matheos and Raigemborn, 2012; Tauber, 1997a). Fossil levels (FL) 1 to 7 were defined in laterally continuous The SCF is an early to middle Miocene lithostratigraphic unit widely package about 35 m thick of exposures in the Estancia La Costa Member, distributed in the Austral Basin and mostly composed of fine-grained (Kay et al., 2012; Tauber, 1997a). The age of this composite level is fluvial sediments, showing thickness variations from about 500 m based on the age of two radiometrically dated tuffs that bracket it: the close to the Andean foothills to about 200 m at the Atlantic coast. In CO tuff of Perkins et al. (2012), and the CV-13 tuff of Trayler et al. terms of stratigraphy and sedimentology, the best known localities are (2020b). The former as refinedby high-precision U–Pb age modeling is along the Atlantic coast between Río Coyle and Río Gallegos (Matheos 17.31 Ma and the latter is 17.62 Ma (Trayler et al., 2020b). and Raigemborn, 2012; Tauber, 1997a, 1997b; Trayler et al., 2020b; Zapata, 2018), inland along Río Santa Cruz and Río Chalía (Cuitino˜ et al., 2016, 2019a, 2021; Fernicola et al., 2014), and in the Andes south 2.2. Santacrucian fauna of Lago Posadas (Cuitino˜ et al., 2019b) (Fig. 1). Age assignments for the SCF in western (Andean) localities suggest a time of accumulation be­ In addition to what was mentioned above, the fossils of the SCF are tween 18.5 and 14 Ma (Blisniuk et al., 2005), whereas in the eastern part the best record for interpreting the biological diversity of mammals in of the basin this time span is restricted between 18 and 15 Ma (Cuitino˜ the southern part of South America (Patagonia) prior to the GABI et al., 2016, 2021; Perkins et al., 2012; Trayler et al., 2020b). (Vizcaíno et al., 2012d). As noted by Simpson (1980) this fauna is The SCF crops out in several localities along the escarpments of the particularly important for understanding a phase in the history in which east-west oriented valley of the RSC. Most of the best exposures are the communities of South American mammals consisted of a complex located along the southern margin of this valley, from which recent mixture of descendants of ancient lineages of the continent (meta­ surveys provided novel sedimentologic, geochronologic and paleonto­ therians, xenarthrans, astrapotheres, notoungulates and litopterns) and logic data (Fernicola et al., 2019a). Following the original nomenclature new immigrants (primates and rodents) from other land masses (prob­ of Carlos Ameghino, these localities are known from east to west as ably Africa) (Antoine et al., 2012; Arnal et al., 2020; Boivin et al., 2017; Barrancas Blancas (BB), Segundas Barrancas Blancas (SBB) and Yaten´ Bond et al., 2015; Kay, 2015a, 2015b; Lavocat, 1976; Seiffert et al., Huageno (YH) (Fernicola et al., 2014). In BB the SCF lies transitionally 2020). Fossil mammals include small paucituberculatans and caenoles­ above the early Miocene shallow marine deposits of the Monte Leon´ tids, medium to large carnivorous sparassodonts (Metatheria), several Formation, while its base is covered in western localities. In turn, the armadillos, some medium-sized glyptodonts, a large diversity of me­ SCF is overlaid everywhere by fluvial terrace conglomerates of late dium to large-bodied sloths, and one small anteater (Xenarthra), a large Miocene-Pliocene age showing a sharp, regionally extensive erosional astrapothere (Astrapotheria), several small typotheres, two medium to surface. large toxodontids, a large homalodothere (Notoungulata), some pro­ In the RSC valley, the SCF is mostly composed of partly pedogenized terotheriids and a medium-sized macraucheniid (Litopterna), as well as fine-grained fluvial deposits accumulated on fluvial floodplains. Sparse many caviomorph rodents (Rodentia), and a medium-sized platyrrhine lenticular sandstones represent accumulation within fluvial channels. monkey (Primates). Tabular, fine-grained tuff and tuffaceous horizons are also common There is also a rich assemblage of birds (Rheiformes, Tinamiformes, (Fig. 2). Zircon U–Pb ages from these tuffs, in combination to calculation Gruiformes, Anseriformes, Pelecaniformes, Ciconiiformes, Falconi­ of sedimentation rates, constrain the time span of the SCF to between formes, and Cariamiformes) (Degrange et al., 2012; Diederle and Nor­ ~17.2 and ~16.3 Ma for Barrancas Blancas, and between ~16.5 and iega, 2019). Unaccountably, no crocodilians and turtles are recorded, ~15.6 Ma for Segundas Barrancas Blancas (Cuitino˜ et al., 2016, 2019a). but among the herpetofauna there are anurans and several squamates, including a tupinambine teiid and some iguanians and colubrids (Albino

Fig. 1. Map of the southern region of Santa Cruz Province showing the distribution of the Santa Cruz Formation (in yellow) and the localities studied. On the Atlantic coast, FL 1–7 refers to the penecontemporaneus localities Anfiteatro(1), Estancia La Costa (2), Canad˜ on´ Silva (3), and Puesto Estancia La Costa (4) (Kay et al., 2012). In the Río Santa Cruz, Barrancas Blancas (BB), Segundas Barrancas Blancas (SBB) and Yaten´ Huageno (YH). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Regarding the fossil mammals recently collected in the RSC, Ferni­ cola et al. (2019c) reported 1828 specimens from the three localities (540 specimens in Barrancas Blancas, 1267 in Segundas Barrancas Blancas, and 21 in Yat´en Huageno), comprising 64 species: 10 species of metatherians (four Sparassodonta, five Paucituberculata, and one Microbiotheria), 12 species of xenarthrans (five Folivora and seven Cingulata), one astrapothere, nine notoungulates (three Toxodontia and six Typotheria), seven litopterns (six Proterotheriidae and one Macraucheniidae), 24 rodents (11 Octodontoidea, two Erethizontoidea, fiveCavioidea, and six Chinchilloidea), and one new species of primate (Homunculidae) (Fig. 3). Combining the list of species of the new collection with other species previously reported for the RSC and recognized by the specialists in Fernicola et al. (2019a) the taxonomic richness rises to 95 taxa (16 species of metatherians, 22 xenarthrans, two astrapotheres, 16 notoungulates, 7 litopterns, 31 rodents, and one pri­ mate (see Appendix 2 in Fernicola et al., 2019c). Following the work of Fernandez´ (2020), in gthe present contribution we identify three species not reported by Fernicola et al. (2019c): the typotheres Protypotherium compressidens Ameghino (1891), Interatherium rodens Ameghino (1887) and I. extensus Ameghino, 1895.

3. Materials and methods

3.1. Selection of samples, localities, and levels

Fossil faunas. The study presented here is based on fossil vertebrate collections recovered during fieldwork by the authors and their team during 2013 and 2014 in the RSC at BB, and SBB. The RSC collection includes more than 1900 specimens that are cataloged in the permanent collections of the Museo Regional Provincial Padre M.J. Molina (MPM- PV), Río Gallegos, Santa Cruz Province, Argentina. In the results section we tabulate the number of specimens broken down by taxon. Most of these specimens were reported in the volume edited by Fernicola et al. (2019a). As in Kay et al. (2012), we based our analysis solely on the collection made by us (i.e. not considering the old collections of un­ certain provenance and stratigraphic level). Additional faunal comparisons are based on specimen records from our collecting expeditions in the SCF at FL 1–7 in Vizcaíno et al. (2012a). The FL 1–7 fauna is a composite of specimens collected at pene­ contemporaneous localities (Anfiteatro, Estancia La Costa, Canad˜ on´ Silva, Puesto Estancia La Costa; Fig. 1) from the Estancia La Costa ◦ Member of the SCF on the Atlantic coast south of Río Coyle, between 51 ′ ◦ 03 and 51 11’ S (Vizcaíno et al., 2012d). The fossil species lists are updated from those reported by Vizcaíno et al. (2012a) and Kay et al. (2012) based on systematic revisions and fieldwork work undertaken since the collecting season of 2011. Each fauna samples a different non-overlapping age range: FL 1–7 is younger than a tuff designated CV-13 dated at 17.62 Ma and older than the CO tuff (17.31 Ma) (Trayler et al., 2020b); BB is dated at between 17 Fig. 2. Age-correlated sedimentary sections of the Santa Cruz Formation at Ma and 16.5 Ma, and SBB between 16.3 Ma and 15.6 Ma. A brief interval Barrancas Blancas and Segundas Barrancas Blancas localities at the Río Santa Cruz. Each section represents the integration from several closely-spaced partial of temporal overlap between BB and SBB contains no fossils (Fig. 2). sections reported by (Cuitino˜ et al., 2019a). Zircon U–Pb ages from tuffs are For each fauna we present taxonomic identifications (genera and indicated in red; all other ages are estimated upon sedimentation rates (Cuitino˜ species presence or absence), and the stratigraphic ranges at BB and SBB et al., 2016, 2019a). (For interpretation of the references to color in this figure (Supplementary document S1). Stratigraphic ranges are adjusted by legend, the reader is referred to the Web version of this article.) correlation based on reported radiometric ages, supplemented by depositional rates. et al., 2017; Fernicola and Albino, 2012; Muzzopappa, 2019). Our reports of absolute and relative abundance are based on spec­ The 1887 expedition of Carlos Ameghino to the RSC produced more imen records in the MPM-PV collection catalogs. Species abundances are than 2000 fossil specimens, on the basis of which his brother Florentino roughly comparable between most components of the fauna (meta­ erected 110 new species of mammals (Ameghino, 1887), dramatically therians, notoungulates, primates, litopterns) because the identifiable increasing the number of Santacrucian taxa from the 12 described parts are mainly dentitions, but xenarthran records are not comparable earlier (see references in Fernicola et al., 2019b). Between 1887 and because they are based on other anatomical parts. This is especially the 1894, approximately 500 added taxa from the SCF were proposed by case for armored taxa for which the identification is based on dermal Ameghino and Mercerat, of which about 120 type specimens came from scutes. the RSC (Fernicola et al., 2019b). Extant Faunas. For the analysis of extant mammalian faunas, we used a subset of 55 of the 85 faunas enumerated by Spradley et al. (2019)

4 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Fig. 3. Exposures of the SCF at the Río Santa Cruz. A view of the river and exposures. B, Barrancas Blancas, and C, Segundas Barrancas Blancas. Life reconstructions of some of the genera recorded at the Río Santa Cruz. Taxa >5 kg are approximately scaled and those < 5 kg (Abderites, Palaeothentes, Pachyrukhos and Homunculus) are not. Modified from Chapters 10, 11, 12, 13, 14, 15 and 16 of Vizcaíno et al. (2012a,b,c,d). from localities with restricted geographic areas across South America tropics and high latitudes (Ojeda and Mares, 1989; Patterson et al., (Fig. 4). We consider only the faunas of low elevation because the SCF’s 1998). depositional environment indicates a lowland with little topographic The list of extant localities in Appendix A provides some of the relief (Cuitino˜ et al., 2019a; Raigemborn et al., 2015, 2018a) and accompanying information about elevation above sea level, mean because faunal composition is affected by altitude. Our upper-limit annual precipitation (MAP), mean annual temperature (MAT), as well as cutoff is based conservatively on data for elevational gradients of summaries of niche metrics for modern and Santacrucian localities (see vegetation, which are particularly significantabove our cutoff elevation, below). Other data, including precipitation and temperature season­ although the elevation at which the changes occur differs between the ality, canopy height, and net primary productivity are provided in

5 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Fig. 4. The South American continent with the position of the localities evaluated. Black dots are extant faunas listed in Appendix A. Three overlapping red stars are the Miocene localities of the Santa Cruz Formation. The four panels represent the mean annual tem­ ◦ ◦ perature ( C), temperature seasonality ( C), mean annual precipitation (mm), and pre­ cipitation seasonality (Coefficient of Varia­ tion of monthly rainfall). Data from NOAA Physical Sciences Laboratory (PSL) (https:// psl.noaa.gov/data/gridded/data.UDel_Air T_Precip.html). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Supplementary document S2. The full list of 208 genera of extant non- Global Land Analysis and Discovery group of the University of Maryland volant mammalian genera is found in Supplementary document S3. A (https://glad.umd.edu/dataset/gedi). Spatial maps of 30 arc-seconds few updates were made to Spradley et al.’s faunal lists and based on resolution for MAP, MAT, and temperature seasonality and a resolu­ other highly corroborated and complete lists in the literature (see notes tion of 30 arc-minutes for canopy height were then uploaded into QGIS® in SD S1). The most significantdifference between the lists reported here (QGIS.org, 2021). In order to extract the data for each locality. The and those in Spradley et al. (2019) is that we count genera, not species sampling areas of the extant faunas represent a wide range of mean ◦ ◦ (see Section 3.2). While it is likely that a few taxa go unrecorded in annual rainfall and the faunas range in latitude from 10 N to 55 S. Both Spradley’s list and ours, this procedure provides a more reliable repre­ floraldiversity and the complexity of vegetation in the lowland tropics is sentation of any given fauna at a locality than does the use of distribu­ strongly correlated with annual rainfall (Gentry, 1988). At one extreme, tion maps (Spradley et al., 2015). at Río Caura, Estado Bolívar, Venezuela, rainfall exceeds 3.5 m per year MAP, MAT, and temperature seasonality data for all localities were with no appreciable dry season. At the other extreme at Parque Nacional obtained through the online database WorldClim.org (https://www.wor Bosques Petrificados de Jaramillo, Santa Cruz Province, Patagonia, ldclim.org/data/bioclim.html) as high-resolution spatial data. MAP is Argentina, with rainfall ~200 mm per annum and dry the year around. ◦ definedas the sum annual precipitation in millimeters averaged over the Mean annual temperature for the extant ranges from ~28 C at Puerto ◦ course of a thirty-year span (1970–2000) (O’Donnell and Ignizio, 2012). Paez,´ Estado Apure, Venezuela to ~4 C at Parque Nacional Tierra del MAT is defined as the annual mean temperature in degrees Celsius as Fuego, Provincia de Tierra del Fuego, Argentina. In environments where calculated from the mean monthly temperatures for a given year. rainfall exceeds 2000 mm/year and a dry season lasts fewer than 4 Temperature seasonality is defined by O’Donnell and Ignizio (2012) as months, evergreen rainforest predominates. In regions with less than “the amount of temperature variation over a given year (or averaged 1000 mm of rainfall and dry intervals longer than 6 months, the domi­ years) based on the standard deviation (variation) of monthly temper­ nant vegetation is drought-resistant and deciduous. Areas of interme­ ature averages”, and is also measured in degrees Celsius. Precipitation diate rainfall between 1000 and 2000 mm/year with 4–6 months of dry seasonality (expressed as a percentage) is the coefficientof variation of season tend to exhibit semideciduous forests, often as riparian gallery the monthly total precipitation to the mean monthly total precipitation. forests with interspersed savannas. Spatial data for canopy height (in meters) were downloaded from the

6 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

3.1.1. Use of genus as the level of interest Basic ecological information for each extant genus was compiled Kay et al. (2012) analyzed the SCF fauna at the generic level whereas from the literature (Supplemental Document S3). These include cate­ Spradley et al. (2019) used species rather than genera as the unit of gorical variables for average body mass, primary diet, and preferred comparison. We have transformed Spradley’s dataset into one that uses locomotion, which, in combination, permit a general evaluation of the genera. In choosing the generic level for our analysis, we recognize that role of the genus in niche space (Vermillion et al., 2018), as described in we may be sacrificingsensitivity in detecting differences in community Table 1. The niche structure of the fossil mammalian genera from FL structure but this choice is made for several reasons. First is the problem 1–7, BB, and SBB given in Appendix B, is based upon ecomorphological of detection bias. The likelihood of recovering a high percentage of the reconstructions presented in Vizcaíno et al. (2012a) with updates ´ ´ species present in a fossil community must be far lower than that for the (Alvarez and Arnal, 2015; Alvarez and Perez,´ 2019; Arnal and Vucetich, recovery of most of its genera. 2015b; Cassini, 2013; Munoz˜ et al., 2019; Toledo, 2016; Toledo et al., Second, the recognition of fossil species is much more problematic 2014). than the recognition of genera. Hapalops Ameghino, 1887 is a classic To evaluate the community structure of the extant faunas, we use example (Bargo et al., 2019; Kay et al., 2012). Ameghino (1887, 1891, total generic richness or total herbivore richness, and the indices devised 1894) named a vast number of species. Scott (1903–1904) recognized 22 by Kay and Madden (1997a; 1997b) to express the number of genera species of Ameghino’s as likely valid and listed another 15 for which he within a guild (that is, with a particular niche specialization or within a “could arrive at no definite conclusion (page 258)”. Bargo et al. (2019) body size range) relative to total number of genera. would further reduce the number of Hapalops species at BB and SBB to one, with certainty, and an uncertain number of others for which there is 1. Frugivore Index expresses the proportion of frugivorous and seed- insufficient information. Such uncertainly would render any effort to eating species to the total number of plant-eating species in the reconstruct paleoecology based on species numbers problematic fauna: whereas the number of genera is more stable. 100*(F(I) + S + F(L))/(F(I) + S + F(L) + L + G + Sc (Tu) + Sc(L)) Third, morphologically-defined genera sensu Mayr (1950) 1 have been used with success as the primary analytical units for a wide range of large-scale paleontological analyses in paleoecology. The justification 2. Browsing Index expresses the proportion of browsing or leaf-eating is that analyses of genus-level operational units generally capture (i.e., shrub and tree leaves and forbs) species to the total number of species-level patterns (Jablonski and Finarelli, 2009). The genus taxon grazing and browsing species in the fauna: also was preferred by for an analysis on the evolution of body size in 100*(L)/ (L + G) Cenozoic mammalian herbivores from South America “because they are discrete taxonomic units accepted by most paleontologists, and they are less affected by the problems of evaluating intraspecific variation in fossils” (pg. 82 in Vizcaíno et al., 2012b). Table 1 Definitionsof the ecological categories used in this study. Each genus is assigned 3.1.2. Should we limit the analysis to taxa greater than 500 g? a number from each variable. Locomotion/substrate preference (six categories) Often, comparison among faunas for the purpose of assessing following Fleming (1973) and Andrews et al. (1979). paleoecology, paleoclimate or the driving factors of mammalian biodi­ Body Definition Locomotor Definition Dietary Definition versity restrict their comparisons to species that exceed 500 g in body Mass Category Category Category size because data for smaller taxa are sensitive to detection bias (Reed, 1998; Robinson et al., 2017; Rowan et al., 2016, 2020). We include these 1 (I) 10–100 g 1: LT large 1: Vert vertebrate terrestrial prey smaller taxa for two reasons: as already noted, using genera rather than (>1 kg) species should compensate for detection bias. Second, at least in the 2 (II 100 g-1 2: ST small 2: Sc(I) scavenging Neotropics, species (and generic) richness of small mammal taxa (pri­ kg terrestrial and insects mates, small rodents, metatherians) varies from place to place whereas (<1 kg) – that of large mammals is virtually constant (Emmons, 1984, 1999). Any 3 (III) 1 10 kg 3: A arboreal 3: Sc(Tu) scavenging (including and tubers loss of discriminatory power by using the genus rather than the species is gliding) compensated for by the added biotic information gained from the in­ 4 (IV) 10–100 4: A(T) arboreal/ 4: Sc(L) scavenging clusion of small mammals. kg terrestrial and leaves (scansorial) 5 (V) 100–500 5: S(Aq) semi- 5: MYR ants and 3.2. Niche parameters kg aquatic termites 6 (VI) >500 kg 6: T(F) fossorial 6: I(F) insects (with (including some fruit or For reconstructing the niche structure of all faunas, we identify the semi- nectar) number of genera present at the modern localities and fossil sites, the fossorial) body size, substrate preference and use, and diet of the taxa, as deter­ 7: F(I) fruit (or gum), with mined by behavioral studies of the living genera and as inferred from some animal ecomorphology in the case of the fossil genera. protein 8: S small seeds of grasses (and other 1 Morphologically-defined genera (morphogenera) are defined by Mayr plants or “… (1950) as: one [or several] species of common ancestry, which differ in a insects) pronounced manner from other groups of species and are separated from them 9: F(L) fruit with by a decided morphological gap,” Genera are considered by Mayr to occupy leaves adaptive plateaus “based on a more fundamental difference in ecology than that 10: L leaves between the ecological niches of species.” To this is sometimes added the caveat (browse) that paraphyletic taxa are inadmissible (Wood and Collard, 1999). In any event, 11: G stems and as noted by Jablonski and Finarelli (2009), a significant percentage of mor­ leaves of phogenera are monophyletic (Jablonski and Finarelli 2009) and species and grasses (graze) genus-level diversificationdynamics are comparable (Liow and Finarelli, 2014).

7 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

3. Arboreality Index is used to express the proportion of arboreal The Simpson Coefficient of faunal similarity is 76.3% between FL 1–7 species to the total number of non-volant species: and BB, 81.6% between FL 1–7 and SBB, and 92.1% between BB and SBB. 100*(A + 0.5*A(T)/ (sum of all locomotor categories) To interpret the comparison of SI between the fossil localities, we examined a cluster of seven extant localities: Reserva de la Biosfera y 4. Predator/prey ratio expresses the proportion of secondary Estacion´ Biologica´ del Beni, Bolivia, Parque Nacional Madidi, Bolivia, consumers Parque Nacional Noel Kempff Mercado, Bolivia, Estacion´ Biologica´ Cocha Cashu, Peru, Reserva Nacional Tambopata, Peru, Parque Nacio­ 100*((Vert + Sc(I) + Sc(Tu) + Sc(L) + MYR + I(F))/all herbivores nal del Manú, Peru, Otishi National Park, Peru, and Manaus, Brazil (Appendix A). These localities are at low elevation (less than 600 masl), ◦ ◦ 5. Size Index: expresses the proportion of species in size classes II and between 2.5 and 14.6 S, separate by an average of 184 km, and have an ± III relative to those in class IV an V where the size classes are II = average rainfall of 2142 561 mm. There are 28 pairwise comparisons ’ 100–1000 g; III = 1–10 kg; IV = 10–500 kg; V = >500 Kg: possible among these localities. They have an average Simpson s Index of 67% with a range of 34%–90%. There is significant correlation be­ 100*(Class II + Class III)/ Class IV + Class V tween SI and distance between the localities in this sample (R2 = 0.70, P < 0.0001). On the other hand, SI and rainfall are not correlated signif­ In comparing the rodents of the three fossil faunas we present a icantly between the localities (R2 = 0.010, P = 0.62). The SI values of the discrete variable for cheek-tooth crown height (hypsodonty). Hyp­ three SCF localities, separated by a maximum of 125 km, fall within the sodonty is based on the relationship between the height of the crown and range expected for such a distance. In other words, in terms of faunal the anteroposterior length of the crown of the lower molars (Janis and similarity, the three SCF faunas, despite their temporal separation, do Fortelius, 1988). Arnal (2011) evaluated the grade of hypsodonty as not differ more than would be expected for localities with similar follows: brachydont: ratio of mesiodistal length/crown height < 0.5; geographic distance of separation. mesodont: ratio < 1.0; protohypsodont ratio >1.0 but forming a root; euhypsodont (rootless crown). From this we derive a hypsodonty score 4.1.1. Meridiolestida (HS), HS is the percentage of specimens of each crown height weighted The early Miocene meridiolestidan Necrolestes (Rougier et al., 2012; by crown height with euhypsodont taxa counting for 3: protohypsodont, Wible and Rougier, 2017) has been recorded in the SCF, but it is un­ 2: mesodont, 1: and brachydont, 0. The total score is divided by the common. All remains come from coastal SCF. The type specimen is from maximum possible score if all taxa were euhypsodont, i.e., 300, and is Monte Observacion´ (Ameghino, 1891) and the specimens from Prince­ expressed as a percentage. If all specimens are euhypsodont, the score is ton University/Yale Peabody Museum Collection (YPM-VPPU; JB 100%; if all are brachydont the weighted score is 0%. Hatcher’s old collections) are from south of Río Coyle along the Atlantic The Simpson Coefficientof faunal similarity (SI) (Simpson, 1943) is coast and Killik Aike Norte on the Río Gallegos. We have recovered three used to compare some of the faunas. SI is 100*(C/N1) where C is the specimens of Necrolestes (cranial and mandibular remains) in FL 1–7, number of taxa in common between two faunas and N1 is the total one in Canad˜ on´ de las Vacas and one in Rincon´ del Buque (Fig. 1), but number of taxa in the smaller sample. have not recorded it so far in the RSC localities. Necrolestes has anatomical specializations found in extant subterra­ 3.3. Statistical analyses nean mammals. Recent studies proposed that it was as a head-lift digger, by analogy with extant African golden moles and Australian marsupial Least-squares regressions using linear and second degree polynomial moles; analysis of the ear region supports the inference that it was models and Principal Components analysis (using a correlation matrix) adapted for subterranean habits (Wible and Rougier, 2017). were carried out with JMP® Pro 15.0.0 for Mac. We also estimated environmental parameters using two machine- 4.1.2. Metatheria learning models, random forest (RF) and Gaussian process regression Metatherians (Sparassodonta and Paucituberculata) are relatively (GPR), previously used in paleoenvironmental reconstruction by infrequent elements of the SCF. We identified95 metatherian specimens Spradley et al. (2019). These machine-learning models take advantage in our collections from FL 1–7, BB, and SBB: 43, 15, and 37, respectively. of the same multivariate approach as a linear regression, while also At FL 1–7, 11 genera and 14 species are recorded; at BB, there are 5 incorporating relationships between geographically closely related data genera and 7 species and; at SBB, 8 genera and 10 species (Table 2). points in order to model even non-linear relationships between variables There seems to be a taphonomic bias in the collections made at the (see Spradley et al., 2019 for further explanation of RF and GPR and different localities. Sparassodonts (45 records) are clearly more abun­ their utility in paleoenvironmental reconstruction). The models were dant in FL 1–7 (where more complete and articulated specimens appear) derived from the ecological indices discussed above, and were used to than in the RSC localities: 31 in FL 1–7, 4 in BB, and 10 in SBB. We estimate mean annual precipitation (MAP), mean annual temperature recorded 8 genera and 8 species in FL 1–7, 2 genera and species in BB (MAT), temperature seasonality, and canopy height at FL 1–7, BB, and and 4 genera and species in SBB. Borhyaena and Sypalocion are present in SBB. As demonstrated in Spradley et al. (2019), these machine learning the three localities; Cladosictis and Perathereutes appear in the coast and models have the ability to effectively estimate temperature seasonality SBB, and Prothylacynus, Lycopsis and Acrocyon are recorded only in FL and canopy height, in addition to MAP and MAT, in an extant 1–7. However in the old collections of RSC, there are records of Acyon, species-level dataset from Australia as well as South America. Error es­ Acrocyon, and Lycopsis (Fernicola et al., 2019c). This highlights the need timates are presented as mean absolute error (MAE). to use caution when evaluating predator-prey ratios in these faunas. Sparassodonts were part of the predator guild, being mainly hyper­ 4. Results carnivores. They exhibited different locomotor abilities (from scansorial to terrestrial) and a wide range of body masses, from 1 kg to ~40 kg). 4.1. Review of mammalian occurrences Prevosti et al. (2012) reconstructed the Santacrucian predator guild and suggested that there was a good niche partition within these Our sample of fossil mammals includes 44 genera from FL 1–7, 38 sparassodonts. genera from BB, and 44 genera from SBB (Appendix B). These faunas are Paucituberculatans (50 records) are more abundant in the RSC, very similar to one another at the generic level. FL 1–7 and BB share 29 particularly in SBB (12 in FL 1–7; 11 in BB; 27 in SBB), where fossils are genera; FL 1–7 and BB share 31 genera and BB and SBB share 35 genera. more fragmentary and more small remains are found. At FL 1–7 we

8 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Table 2 Occurrence and frequency of metatherian specimens at FL 1–7, Barrancas Blancas, and Segundas Barrancas Blancas.

Taxon FL 1-7 Barrancas Blancas (BB) Segundas Barrancas Blancas (SBB) ◦ ◦ ◦ Higher level classification Genus P/A N sp Na Freq. (%)b P/A N sp. Na Freq. (%)b P/A N sp. Na Freq. (%)b

SPARASSODONTA Hathliacynidae Cladosictis 1 1 8 28.6 0 0 0.0 1 1 4 40.0 Sipalocyon 1 1 4 14.3 1 1 3 75.0 1 1 2 20.0 Perathereutes 1 1 1 3.6 0 0 0.0 1 1 1 10.0 Prothylacynus 1 1 2 7.1 0 0 0.0 0 0 0.0 Hathliacynidae indet. 5 17.9 0 0.0 0 0.0 Borhyaenidae Borhyaena 1 1 8 28.6 1 1 1 25.0 1 1 1 10.0 Lycopsis 1 1 1 3.6 0 0 0 0.0 0 0 0 0.0 Arctodictis 1 1 1 3.6 0 0 0.0 0 0 0.0 Acrocyon 1 1 1 3.6 0 0 0.0 0 0 0.0 Borhyaenidae indet. 0 0.0 0 0.0 2 20.0 Summary 8 8 31 2 2 4 4 4 10

PAUCITUBERCULATA Palaeothentiidae Palaeothentes 1 4 10 90.9 1 3 9 81.8 1 3 22 81.5 Acdestis 1 1 1 9.1 1 1 1 9.1 1 1 2 7.4 Microbiotheriidae Microbiotherium 1 1 1 0 0 0.0 1 1 2 7.4 Abderitidae Abderites 0 0 0.0 1 1 1 9.1 1 1 1 3.7 Summary 3 6 12 3 5 11 4 6 27

SUMMARY ALL TAXA 11 14 43 5 7 15 8 10 37

a number of catalog entries. b Frequency of occurrence (number of subordinal records) divided by the total records at this locality. P/A, presence/absence; Freq., frequency. recorded 3 genera and 6 species, 3 genera and 5 species in BB, and 4 features in the specimens we collected (Fernicola and Vizcaíno, 2019). genera and 6 species in SBB (Table 2). Santacrucian glyptodonts are moderately large (up to 120 kg) and Palaeothentes is, by far, the most abundantly recorded everywhere. ambulatory selective feeders in relatively closed to strictly closed hab­ Palaeothentes recorded at RSC are medium to large-sized curso-saltato­ itats (Vizcaíno et al., 2011, 2012c). rial insectivorous species ranging from ~80 g to 900 g; (Abello et al., The armadillos Proeutatus and Prozaedyus are the most abundant and 2012); the larger, more frugivorous Palaeothentes aratae (Strait et al., frequent cingulates at all localities, while Stenotatus is more frequent/ 1990) was not identified at the RSC. The medium-sized insectivore-­ abundant in the Río Santa Cruz. Peltephilines are more common in the frugivore Acdestis oweni and the frugivorous/scansorial Abderites mer­ coastal localities than in the RSC, but they are less frequent than the idionalis were both recorded at BB and SBB (Chornogubsky et al., 2019). other three taxa. We did not record Stegotherium at FL 1–7, and in the We have no records of Stilotherium, which is found in the old collections RSC we found only three osteoderms at BB, suggesting that it was quite at RSC (Fernicola et al., 2019c). Stilotherium likely had an insectivorous rare. All Santacrucian armadillos were diggers and the variation of the diet based on the morphology of its cheek teeth illustrated by Abello masticatory apparatus shows a broad range of specializations from et al. (2021). herbivory and strict myrmecophagy. The taxonomic richness and di­ versity of armadillos supports the environmental interpretation of a 4.1.3. Xenarthra mixture of open and relatively closed vegetation in relatively dry con­ Among xenarthrans, Cingulata and Folivora are disparately repre­ ditions (Vizcaíno et al., 2012c). sented among the moderate to large mammals of the SCF. We recorded Folivora (sloths) are also abundant elements of the SCF. However, as 550 specimens of xenarthrans in our collections (129 in FL 1–7, 191 in with cingulates, skeletal remains such as skull and mandibles are rare BB, and 230 in SBB; Table 3). At FL1-7 we recorded 13 genera and 15 and postcranial elements are particularly abundant, which makes species, 9 genera and 9 species at BB, and 10 genera and 10 species at taxonomic allocations difficult (Bargo et al., 2019). We recorded 123 SBB. specimens of sloths (57 in FL 1–7, 20 in BB and 46 in SBB). At FL 1–7 we Cingulates (armadillos and glyptodonts) remains are probably the recorded 6 genera and 8 species, one genus and species in BB and 5 most frequently found in the field, due to the more than 1000 osteo­ genera and species in SBB. The most common sloth remains are un­ derms that constitute an individual carapace. Other skeletal remains are identifiablepostcranial elements of Megatherioidea indet. (FL1-7: 53%, uncommon, especially skulls, mandibles, and teeth. Particularly for the BB: 80% and SBB: 67%, Table 3). Hapalops is the only genus recorded in glyptodonts, precise specific or even generic allocation of osteoderms the three localities, with the highest abundance, and Schismotherium and may depend on the association with cranial remains (Fernicola and Nematherium are recorded on the coast and at SBB. While Pelecyodon, Vizcaíno, 2019). Beyond those specimens with carapace associated with Hyperleptus and Eucholoeops record only in the coast, Xyophorus records endoskeleton, our collecting strategy for cingulates was to collect a few are only in the RSC (SBB). Remarkably, we recorded megatheriids only osteoderms from each locality and level to document taxon pre­ in the highest levels of the RSC (at SBB), but they are not recorded so far sence/absence. We recorded 426 specimens or lots of osteoderms: 121 BB or FL1-7. The Planopinae indet at Segundas Barrancas Blancas is glyptodonts (13 in FL 1–7, 48 in BB and 60 in SBB), and 305 armadillos either Planops or Prepotherium, see Bargo et al. (2012). Planopinae are (58 in FL 1–7, 123 in BB, and 124 in SBB). At FL 1–7 and BB two genera the largest of the Santacrucian sloths (100–120 kg (Toledo et al., and species of glyptodonts were recorded, while in SBB only one. 2014);). Megatherioid sloths have body masses from ~40 to 80 kg, while Armadillos are represented by four genera and species at FL 1–7, 5 in BB, mylodontids reached ~80–90 kg. According to Toledo (2016) the and 4 in SBB (Table 3). mid-sized sloths (Hapalops, Pelecyodon, Schismotherium, Hyperleptus and Glyptodonts (Propalaehoplophoridae) are very common components Eucholoeops) are members of the arboreal folivore paleoguild. The of the SCF. mylodontid Nematherium may have been semiarboreal consumers of Cochlops and Eucinepeltus are recorded in the three localities. The leaves, fruits and tubers due to their digging capabilities, while absence of Propalaehoplophorus in our collections but present in the old megatheriids were the most terrestrial sloths, and folivores. collections at RSC may be an artifact due to the lack of diagnostic Vermilingua (anteaters) are poorly represented in the SCF (Bargo

9 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Table 3 Occurrence and frequency of xenarthran specimens at FL 1–7, Barrancas Blancas, and Segundas Barrancas Blancas.

Taxon FL 1-7 Barrancas Blancas (BB) Segundas Barrancas Blancas (SBB) ◦ ◦ ◦ Higher level classification Genus P/A N sp. Na Freq. (%)b P/A N sp. Na Freq. (%)b P/A N sp. Na Freq. (%)b

CINGULATA Propalaehoplophoridae Cochlops 1 1 2 15.4 1 1 12 25.0 0 0 0 0.0 Eucinepeltus 0 0 0 0.0 1 1 2 4.2 1 1 11 18.3 Propalaehoplophorus 1 1 4 30.8 0 0 0 0.0 0 0 0 0.0 Propalaehoplophori- dae indet. 7 53.8 34 70.8 49 81.7 Summary 2 2 13 2 2 48 1 1 60

Peltephilidae Peltephilus 1 1 12 20.7 1 1 7 5.7 1 1 7 5.6 Dasypodidae Proeutatus 1 1 26 44.8 1 1 42 34.1 1 1 43 34.7 Stenotatus 1 1 2 3.4 1 1 26 21.1 1 1 22 17.7 Prozaedyus 1 1 15 25.9 1 1 47 38.2 1 1 52 41.9 Stegotherium 0 0 0 0.0 1 1 1 0.8 0 0 0 0.0 Dasypodidae indet 3 5.2 0.0 0.0 Summary 4 4 58 5 5 123 4 4 124

VERMILINGUA Myrmecophagidae Protamandua 1 1 1 100.0 Summary 1 1 1 0 0 0 0 0 0

FOLIVORA Megatherioidea Hapalops 1 2 12 21.1 1 1 3 15.0 1 1 6 13.0 Pelecyodon 1 1 2 3.5 0 0 0 0.0 0 0 0 0.0 Hyperleptus 1 1 1 1.8 0 0 0 0.0 0 0 0 0.0 Schismotherium 1 1 3 5.3 0 0 0 0.0 1 1 1 2.2 Xyophorus 0 0 0 0,0 0 0 0 0,0 1 1 1 2,2 Megatherioidea indet. 30 52.6 16 80.0 31 67.4 Megatheriidae Planopinae indet. 0 0 0 0.0 0 0 0 0.0 1 1 2 4.3 Megatheriidae indet. 0.0 0.0 4 8.7 Megalonychidae Eucholoeops 1 2 6 10.5 0 0 0 0.0 0 0 0 0.0 Mylodontidae Nematherium 1 1 2 3.5 0 0 0 0.0 1 1 1 2.2 Mylodontidae indet. 1 1.8 1 5.0 0.0 Summary 6 8 57 1 1 20 5 5 46

SUMMARY ALL TAXA 13 15 129 9 9 191 10 10 230

a Number of catalog entries. b Frequency of occurrence (number of subordinal records) divided by the total records at this locality. P/A, presence/absence; Freq., frequency.

Table 4 Occurrence and frequency of typothere, toxodont, astrapothere and litoptern specimens at FL 1–7, Barrancas Blancas, and Segundas Barrancas Blancas.

Taxon FL 1-7 Barrancas Blancas (BB) Segundas Barrancas Blancas (SBB) ◦ ◦ ◦ Higher level classification Genus P/ N of sp. Na Freq. (%)b P/ N of sp. Na Freq. (%)b P/ N of sp. Na Freq. (%)b A A A

TYPOTHERIA Hegetotheriidae Hegetotherium 1 1 40 23.3 1 1 11 22.4 1 1 10 3.6 Pachyrukhos 0 0 0 0.0 0 0 0 0 1 1 86 30.8 Interatheriidae Interatheriun 1 2 70 40.7 1 2 17 34.7 1 2 59 21.1 Protypotherium 1 3 62 36.0 1 3 21 42.9 1 4 124 44.4 Interatheriidae indet. 10 5.8 0 0 Summary 3 6 182 4 7 49 4 8 279

TOXODONTIA Toxodontidae Nesodon 1 1 14 31.1 1 1 12 31.6 1 1 18 39.1 Adinotherium 1 1 30 66.7 1 1 24 63.2 1 1 18 39.1 Toxodontidae indet. 11 24.4 0 0 Homolodotheriidae Homalodotherium 1 1 1 2.2 1 1 2 5.3 1 1 10 21.7 Summary 3 3 56 3 3 38 3 3 46

ASTRAPOTHERIA Astrapotheriidae Astrapotherium 1 1 8 100.0 1 1 4 100.0 1 1 12 100.0 LITOPTERNA Proterotheriidae Anisolophus 0 0 0 0 1 1 1 7.7 1 1 15 51.7 Diadiaphorus 1 1 4 22.2 1 1 3 23.1 1 1 1 3.4 Thoatherium 1 1 8 44.4 1 1 4 30.8 1 1 6 20.7 Tetramerorhinus 1 1 2 11.1 1 2 3 23.1 1 2 3 10.3 Proterotheriidae indet. 13 72.2 0 0.0 1 3 10.3 Macraucheniidae Theosodon 1 1 4 22.2 1 1 2 15.4 1 1 1 3.4 Summary 4 4 31 5 6 13 6 6 29

SUMMARY ALL TAXA 11 14 277 13 17 104 14 18 366

a Number of catalog entries. b Frequency of occurrence (number of subordinal records) divided by the total records at this locality. P/A. presence/absence; Freq.. frequency.

10 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296 et al., 2012). They are only recorded at FL1-7 by one specimen of Pro­ toxodontids. Homalodotherium is large (about 400 kg) brachyodont tamandua, and there are no records at the RSC. clawed “ungulate” that may have been a browser that inhabited rela­ tively closed habitats (Elissamburu, 2010). 4.1.4. Astrapotheria Astrapotheres, the largest South American native ungulates, were 4.1.7. Litopterna represented in the SCF by only one genus. We identified only 12 speci­ Litopterns are much less abundant than notoungulates in the SCF. We mens as belonging to Astrapotherium: 8 in FL1-7, 4 in BB and 12 in SBB identified73 specimens in our collections: 31 from FL 1–7, 13 from BB, (Table 4. Fig. 5). Astrapotherium was a one-ton browser that inhabited and 29 from SBB (Table 4). Only 7 of them correspond to the macrau­ closed habitats (Cassini, 2013). cheniid Theosodon (FL 1–7: 4 from, BB: 2, SBB: 1). We recorded 3 genera and species of protherotheriids from FL 1–7, 4 genera and 5 species from 4.1.5. Typotheria both BB and SBB. Typotheres are common elements of the SCF. We identified 510 According to Cassini et al. (2012) litopterns are brachyodont typothere specimens in our collections from FL 1–7, BB, and SBB: 182, mixed-feeders (Theosodon; 100–150 kg) or browsers (protherotheriids; 49, and 279, respectively (Table 4). At FL 1–7, 3 genera are recorded 25–50 kg) in closed habitats. (exact species counts are pending revisions). At BB, there are 3 genera and 6 typothere species; at SBB 4 genera and 8 species. The composite 4.1.8. Rodentia stratigraphic ranges for typothere genera are depicted in Fig. 5. Caviomorph rodents are by far the most abundant mammals in the Most of the taxa are represented in similar abundances at FL 1–7 and SCF. We recovered more than 900 identifiable specimens from FL 1–7, BB. Interatherium and Protypotherium have similar abundances and BB, and SBB: 250, 173, and 507, respectively. At FL 1–7, nine genera are Hegetotherium is common but less abundant than the other two. The recorded (species identificationsare pending revisions); at BB, there are situation at SBB is quite different. While Interatherium and Protypothe­ 14 genera and 19 species; at SBB, 17 genera and 23 species (Table 5, rium remain common, Hegetotherium declines in abundance with 4% of Fig. 6). all typothere records versus 23% and 22% at FL 1–7 and BB. The likelihood of recording a taxon at a particular locality may The clearest difference in typothere abundance is seen in Pachyr­ depend on its rarity, and be a consequence of sampling probability. In ukhos. This taxon is at least very rare (but more probably absent) in our such cases, one cannot say with any certainty that a taxon would not also material from FL 1–7 (one possible record out of 183 typothere speci­ have occurred at the other localities had we collected a sufficient mens), nor does (Tauber, 1997a) record this taxon in the Ea. La Costa number of specimens. For example, we record one specimen of Dudumus Member of coastal SCF, whereas it does occur in the overlying Ea. La at SBB (0.2%) but none occur at either FL 1–7 or BB. The same is true of Angelina Member, and is common in some places, e.g., at Killik Aike Acaremys (none at FL 1–7, 1.2% at BB, 0.4% at SBB), "Eocardia" excavata Norte2; more details are provided in (Vizcaíno et al., 2021). Pachyrukhos (0.4%, 2.9%, 0.8%), Schistomys (0%, 0% and 0.6%), Scleromys (0.8%, is absent at BB but is by far the most common typothere at SBB (44%). 4.2%, 0.8%), and Prospaniomys (0%, 0.6%, 0.4%). Befittingtheir rarity, As all the typotheres are characterized by euhypsodont cheek teeth, either Dudumus nor Prospaniomys were recorded in the old collections of there is no evident trend in this respect. The appearance of Pachyrukhos RSC while Adelphomys, Pseudoacaremys are found in the old collections in SBB is notable. This species is usually depicted as having been capable but not in our samples (Fernicola et al., 2019c). The more common ro­ of moving rapidly using saltatorial locomotion. It had a short tail, limbs dent taxa (at least 5% of specimens occur at least in one locality) are arranged in a parasagittal orientation; a long hindlimb with the prox­ represented in Fig. 6. In these commonly occurring taxa we may be on imal segment much shorter than the distal all suggesting that it could safer ground to infer that the absence, or presence at a different level of move rapidly using saltatorial locomotion (Cassini et al., 2012; Sinclair, frequency may indicate some difference in community structure. 1908). Its presence at SBB has been interpreted as indicative of more Most of the more common taxa are represented in similar abundance open areas in comparison with BB, FL 1–7 and BB (Fernandez´ and at all three localities. These include Sciamys (5.2% at Fl 1–7, 5.2% at BB, Munoz,˜ 2019). 3.0% at SBB), Spaniomys (6.4%, 11.0%, 8.8%), Steiromys (2.0%, 6.9%, 1.6%). Eocardia has abundances as 15.4% at FL 1–7, 11.6% at BB, and 4.1.6. Toxodontia 11.4% at SBB. Toxodontia are the most common large ungulates (around 100 kg or In a number of cases, a genus at SBB occurs far more frequently than more) and probably mammals of the SCF but its two main groups at FL 1–7 or BB. Acarechimys, a brachydont taxon, appears 8.0% of the (toxodontids and homalodotheriids) are very disparately represented. time at SBB but only 2.0% at FL 1–7 and 1.7% at BB. The common We identified140 Toxodontia specimens in our collections from FL 1–7, euhysodont taxon Prolagostomus accounts for 26.2% of all rodent records BB, and SBB: 56, 38, and 46, respectively (Table 4, Fig. 5). at SBB (131 specimens) but we have no records at FL 1–7 and only 2 Among toxodontids, two genera and two species are represented in specimens at BB. Pliolagostomus is absent at FL 1–7 and BB whereas at the three localities. The remains of the Adinotherium are twofold those of SBB it is quite common (7.6%). Stichomys shows a gradual increase in Nesodon in the FL 1–7 and BB, but both are equally recorded at SBB. Both abundance (absent at FL 1–7, 4.1% at BB, 9.2% at SBB). are considered ambulatory grazing ungulates of moderate (Adinotherium Several taxa decrease in numbers, especially Perimys (19.0%, 11.6%, about 100 kg) to large size (Nesodon, about 500 kg) (Cassini et al., 2012). 6.8%), and Neoreomys (46.8%, 28% 14%). Respecting Perimys, there are Another toxodont Hyperotoxodon is recorded from the old collections of also a difference in body size of the species (Arnal et al., 2019). Perimys RSC but Fernandez´ and Munoz˜ (2019) did not report it from our erutus (a small species) and Perimys onustus (a larger species) both occur collections. at BB and SBB but the former is more common is more common at BB (11 Homalodotheriids are represented by a single genus and species specimens to 1 specimen) and the latter is more abundant at SBB (1 usually quite rare, at least at FL 1–7 (1 specimen) and BB (2 specimens). specimen to 23 specimens). We identified more remains from SBB (10), but still far fewer than Table 6 lists the niche characteristics of commonly occurring rodents in the Santa Cruz Formation at FL 1–7, BB, and SBB. Among the more common rodent taxa, there are no discernible trends in body size other 2 than in species of Perimys. Locomotor habits are poorly documented but Killik Aike Norte (formerly Felton’s Estancia) on the north shore of the — estuary of Río Gallegos was previously suggested to be temporally equivalent to no obvious trends are apparent the more common rodent taxa were ˜ FL 1–7 (Ea. La Costa Member) (Tejedor et al., 2006) but reevaluation of the ground-dwellers (Munoz et al., 2019). The one possibly scansorial tephra indicate a much younger age and allocation to the upper levels of the (semi-arboreal) taxon Steiromys remains stable in numbers. Acarechimys, coastal SCF (Trayler et al., 2020b). with some fossorial tendencies, increases in abundance at SBB Perimys,

11 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296 of records specimen of 5% > comprise that taxa (i.e., illustrated are genera common the Only Toxodontia). (Typotheria, Notoungulata and Astrapotheria, of genera the for ranges stratigraphic rank). composite The ordinal 5. Fig. each

12 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Table 5 Occurrence, frequency and hypsodonty scores of rodent specimens at FL 1–7, Barrancas Blancas, and Segundas Barrancas Blancas.

Taxon and hypsodonty score FL 1-7 Barrancas Blancas (BB) Segundas Barrancas Blancas (SBB) ◦ ◦ ◦ Higher level Genus Hysodonty score P/ N of Na freq. P/ N of Na freq. P/ N of Na freq. classification (Note 1) A sp. (%)b A sp. (%)b A sp. (%)b

CAVIOIDEA Eocardia Euhypsodont 1 2 40 15.44 1 1 20 11.56 1 1 57 11.40 "Eocardia" Euhypsodont 1 1 1 0.39 1 1 5 2.89 1 1 4 0.80 excavata Phanomys Protohypsodont 0 0 0 0.00 1 1 14 8.09 1 1 1 0.20 Schistomys Euhypsodont 0 0 0 0.00 0 0 0 0.00 1 1 3 0.60 Neoreomys Protohypsodont 1 1 117 45.17 1 1 51 29.48 1 1 71 14.20 CHINCHILLOIDEA Neoepiblemidae Perimys Euhypsodont 1 1 48 18.53 1 3 20 11.56 1 2 34 6.80 Chinchillidae Prolagostomus Euhypsodont 0 0 0 0.00 1 1 2 1.16 1 1 131 26.20 Pliolagostomus Euhypsodont 0 0 0 0.00 0 0 0 0.00 1 1 37 7.40 Dinomyidae Scleromys Protohypsodont 1 1 2 0.77 1 1 8 4.62 1 1 4 0.80 OCTODONTOIDEA Prospaniomys Mesodont 0 0 0 0.00 1 1 1 0.58 1 1 2 0.40 Spaniomys Mesodont 1 2 16 6.18 1 1 19 10.98 1 1 44 8.80 Stichomys Mesodont 0 0 0 0.00 1 1 7 4.05 1 1 46 9.20 Dudumus Brachydont 0 0 0 0.00 0 0 0 0.00 1 1 1 0.20 Acaremyidae Acarechimys Brachydont 1 2 5 1.93 1 3 3 1.73 1 4 40 8.00 Acaremys Mesodont 0 1 3 1.16 1 1 2 1.16 1 1 2 0.40 Sciamys Protohypsodont 1 1 13 5.02 1 1 9 5.20 1 2 15 3.00 ERETHIZONTOIDEA Erethizontidae Steiromys Brachydont 1 2 5 1.93 1 2 12 6.94 1 2 8 1.60 Summary 9 14 250 14 19 173 17 23 500

Notes: Hypsodonty score based on the following references: (Arnal, 2011; Arnal et al., 2019; Arnal and Vucetich, 2015a; Candela et al., 2012; Kramarz, 2002; Kramarz et al., 2015; P´erez, 2010; Rasia and Candela, 2019; Vucetich et al., 2015). a Number of catalog entries. b Frequency of occurrence (number of subordinal records) divided by the total records at this locality. P/A. presence/absence; Freq.. frequency.

Table 6 Niche characteristics of commonly occurring rodents in the Santa Cruz Forma­ tion at FL 1–7, Barrancas Blancas, and Segundas Barrancas Blancas.

Taxon Abundance Crown height Body Locomotion increase or size decrease class

Acarechimys ↑ ↓ brachydont I, II terrestrial (fossorial) Eocardia ↔ euhypsodont III good runner ( Munoz˜ et al., 2019) Neoreomys ↓ protohypsodont III ambulatory and an occasional runner Munoz˜ et al., (2019) Perimys ↓ euhypsodont I Digger (Munoz˜ Fig. 6. The composite stratigraphic ranges for the genera of Rodentia. Only the et al., 2019) ↔ common genera are illustrated (i.e., taxa that comprise > 5% of spec­ Phanomys protohypsodont II cursorial ↑ imen records). Prolagostomus euhypsodont II cursorial Pliolagostomus ↑ euhypsodont II cursorial Sciamys ↔ protohypsodont I ? with digging tendencies decreases in abundance. Spaniomys ↔ mesodont II ? ˜ The most notable trend in rodents with evident implications for Steiromys ↔ brachydont IV a climber (Munoz paleobiology is that of cheek-tooth crown height of the rodents et al., 2019) semiarboreal (Table 7). Specimens of high-crowned rodent taxa predominate at all Stichomys ↑ mesodont II ? localities; taxa with evergrowing (euhypsodont), rooted but very high Up arrow indicates a trend towards increased abundance; down arrow is a crowns (protohypsodont), and with crowns higher than tooth length decrease in abundance; up-down arrow is an increase then a decrease; side- (mesodont) account for 90%–96% of all specimens. This also is reflected pointing arrow means no change. in the weighted hypsodonty score (HS) showing overall high levels of hypsodonty with a slight dip at BB compared with FL 1–7 and SBB. The are restricted today to an arboreal niche. A possible occurrence of a percentage of euhypsodont specimen counts decline in between FL 1–7 primate in the old collections was Ecphantodon ceboides, based on a and BB but are much larger at SBB. The bulk of the euhypsodont spec­ specimen now lost (Kay and Perry, 2019; Mercerat, 1891). We record imens at SBB (34%) are from the chinchillids Prolagostomus and Pliola­ one species, Homunculus vizcainoi, from BB and SBB. At FL 1–7 there are gostomus. These taxa are rare or absent at the other localities. two: H. patagonicus and another unnamed species. The number of primate species found at any locality today is corre­ 4.1.9. Primates lated to mean annual rainfall (Fig. 7). One genus of primate is present in Platyrrhine primates are a relatively rare occurrence in our sample several of our low-elevation localities in the Gran Chaco, of Paraguay, but especially important for reconstructing paleoclimate because they

13 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Table 7 Frequency of occurrence of rodent crown height categories.

FL 1-7 Barrancas Blancas Segundas Barrancas Blancas

Crown Height Specimen % Genera % Species % Specimen % Genera % Species % Specimen % Genera % Species %

Euhypsodont 37.8 33.3 NR 27.2 28.6 31.6 53.2 35.3 30.4 Protohypsodont 51.0 33.3 NR 47.4 28.6 21.1 18.2 17.6 21.7 Mesodont 7.3 11.1 NR 16.2 21.4 15.8 18.4 23.5 13.0 Brachydont 3.9 22.2 NR 9.3 21.4 31.6 10.2 23.5 34.8 Weighted hypsodonty score (maximum is 100) 74.3 64.2 71.5

Specimen number, generic and species percentages: sum of all specimens, genera or species in each tooth type at a site divided by the total number of specimen records at that site; notes: NR: not recorded. To calculate the weighted hypsodonty score the specimen percentages are weighted by crown height with euhypsodont taxa counting for 3, protohypsodont, 2; mesodont, 1; and brachydont 0. The total score is divided by the maximum possible score if all taxa were euhypsodont, i.e., 300, and expressed as a percentage.

◦ with MAP ranging from ~650 to ~1400 mm (mean = 1082 mm). To the richness show strong increases peaking at about 25 C MAT, above west, rainfall is ~650 mm. Here, there is a dry-forest vegetation of spiny, which the indices level off or decline. Among our variables, the Arbor­ thorny shrubs and low trees. In the eastern Chaco with rainfall closer to eality Index are most highly correlated with MAT. In least-squares re­ 1400 mm, there is a park-like landscape of clustered trees and shrubs gressions with a second order polynomial fit, the Arboreality Index interspersed with tall, herbaceous savannas. Two genera are found at accounts for 70% of the variance in MAT. Therefore, we consider the various localities ranging between ~625 and ~2200 mm (mean = 1394 Arboreality Index to be the best predictor of MAT in our sample of extant mm). faunas (Table 8). Temperature seasonality also exhibits significant correlations with 4.2. Estimating climate variables the five factors. As with MAT, generic richness, Arboreality, Frugivory, and Browsing indices all decline with temperature seasonality whereas 4.2.1. Assessment of extant faunas the Predator-Prey Index exhibits an increase. All correlations are sig­ < In our sample, mean annual precipitation (MAP) is significantly nificant at the 0.0001 level of probability. Arboreality and tempera­ — 2 correlated (p < 0.0001) with the generic richness of nonvolant mammals ture seasonality are particularly highly correlated R is 0.78 between and the four faunal indices and ratios (Arboreality, Frugivory, Browsing, these variables. In none of the niche variables is there a significant and Predator-Prey) (Supplemental Document S1; Fig. 8). The Predator- correlation with precipitation seasonality. Prey ratio shows a steady decline with increased rainfall. The other For several reasons we expect that some of the indices, while they are indices (Frugivore, Browsing, and Arboreality) as well as generic rich­ promising in principle for estimating MAP, MAT, and temperature sea­ – ness show strong increases with MAP peaking at about 2500 mm rainfall sonality at FL 1 7, BB, and SBB may be unreliable in practice. The po­ per year, above which the indices level off or decline. Notably, this tential challenges relate to two factors: (1) Species rarity may ‘hockey-stick’ curve mirrors trends of net primary productivity and differentially affect the rarity of primary versus secondary consumers by rainfall, which also levels off or declines above 2500 mm of rainfall underestimating the number of predatory taxa, and therefore the (Currie, 1991; Kay et al., 1997; McCain et al., 2018). When localities Predator-Prey ratio. (2) Irrespective of MAP, the synchroneity in fruiting above 2500 mm are removed, the Browsing Index is the variable most at high latitudes makes it unlikely that non-migrating frugivorous tightly correlated with MAP (R2 = 0.78; p < 0.0001; Table 8). mammals would exist in as great numbers in high-latitude paleofaunas. The relationships between Mean Annual Temperature (MAT) and the indices also are highly significant (p < 0.0001) (Fig. 9). As with MAP, 4.2.1.1. Predator-Prey Index. There exist likely sampling errors in the the Predator-Prey ratio shows a steady decline with increased MAT. The proportions of predator and prey genera. For energetic reasons, the other indices (Frugivore, Browsing, and Arboreality) and generic biomass of predators is always smaller than that of their prey. Across a wide range of habitats, in general, there are threefold fewer mammalian predators per kilogram of mammalian prey (Hatton et al., 2015). As one might expect, the number of individual predator specimens collected will reflect this imbalance. For example, at SBB, only 0.8% of all mammal specimens are carnivorous species. Specimen rarity no doubt also accounts for the absence of the hypercarnivorous sparassodonts Acyon, Lycopsis, Acrocyon and the insectivorous paucituberculate Stilo­ therium, present in the old collections of the RSC but absent in the new collections. In practice, unless we have very large samples, we likely will under-sample predator richness in our localities, which will make any prediction of MAT or MAP from the Predator-Prey ratio unreliable.

4.2.1.2. Frugivore Index. The second effect has to do with the overall composition of the mammalian frugivore guild. Kay et al. (2012) called attention to the surprisingly low number of frugivorous taxa at FL 1–7. As they noted, if FL 1–7 were a tropical fauna, the rainfall would be predicted to be less than one-third that expected from the proportions of browsing or arboreal species. BB and SBB repeat this pattern with very few frugivores. We hypothesize that this is an effect of temperature seasonality (and indirectly with latitude, itself highly correlated with temperature seasonality). This low number of ’frugivores’ is a likely consequence of the seasonality of fruit production (Smith-Ramirez and Fig. 7. Box and whisker plot of the number of primate genera versus MAP Armesto, 1994), possibly mediated by synchronicity of flowering( Tutin (mm/yr). Data from Kay et al. (1997).

14 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Fig. 8. MAP vs all niche metrics (all levels of rainfall for all localities of low elevation). A, Predator-Prey Ratio; B, Frugivore Index; C, generic richness; D, Browsing Index; E, Arboreality Index; F, Net Primary Productivity. The line is the second-degree polynomial best fit. MAP units are mm/yr. NPP units are mg of carbon/M2/ day. Rainfall color coded for localities: Vermilion, <500 mm; burnt orange, 500–1000 mm; green, 1000–1500 mm; aqua, 1500–2000; dark blue, 2000–2500 mm; plum, 2500–3000 mm; black, >3500 mm. (For interpretation of the references to color in this figurelegend, the reader is referred to the Web version of this article.) and Fernandez, 1993) at high latitudes. Even a committed non-volant 2008). So, a low mammalian Frugivore Index could be explained if mammalian frugivore cannot find fruit in the cold season. In support migratory birds and bats were the dominant seed predators and of this, we find a strong negative correlation between temperature dispersers. seasonality and the Frugivore Index (R2 = 0.58; n = 46). Alternatively, fruit may have been abundant but instead of nonvolant mammalian 4.2.1.3. Principal Components Analysis. Informed by the potential frugivores this feeding guild was occupied to a greater extent by birds or challenges of using the Predator-Prey and Frugivore indices at high bats rather than nonvolant mammals: it has been noted that more than latitudes, and considering the reduced predictive value of all indices 50% of the woody species of the temperate forest produce fleshyfruits, above 2500 mm rainfall on account of the ‘hockey-stick’ phenomenon, but most of the seed dispersal is accounted for by just a few opportunistic we undertook a Principal Components Analysis (PCA) using the 46 low- species, mostly migratory birds (Aizen and Ezcurra, 1998). Frugivorous altitude localities with <2500 mm of rainfall and 3 variables: generic bats also migrate, following fruit abundance (Richter and Cumming, richness and indices of Arboreality and Browsing. In PC1, all variables

15 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Table 8 Least-squares regressions of MAP and MAT with indices and counts of genera for 46 extant South American faunas with elevation <750 masl and rainfall <2500 mm. For MAT predictions, the Arboreality Index is fit with a second-degree polynomial. Variable Model parameters for MAP Model parameters for MAT Model parameters for Temperature Seasonality

Browsing Index MAP = 448.97 + 22.37*Browsing – Index R2 = 0.779 P-Value=<0.0001 ◦ Arboreality Index MAT = 133.69 + 3.94*Arboreality Index - 0.119* Temperature Seasonality ( C) = (Arboreality Index-23.93) 2 R2 = 0.698 P- 5.28–0.11*Arboreality Index R2 = 0.724 P- Value=<0.0001 Value=<0.0001 ◦ Principal Components MAP = 1371.02 + 315.9 *PC1 R2 = MAT = 23.03 + 2.99 *PC1 - 0.82*PC2 R2= 0.724 P- Temperature seasonality ( C) = 2.785–0.816*PC1 of 3 variables1 0.684 P-Value=<0.0001 Value=<0.0001 R2 = 0.708 P-Value=<0.0001

Notes: 1: Eigenvectors for 3-factor Principal Components analysis given in Table 9.

◦ Fig. 9. MAT in C versus all niche metrics (all levels of rainfall for all localities of low elevation). A, Predator-Prey Ratio; B, Frugivore Index; C, generic richness; D, Browsing Index; E, Arboreality Index. The line is the second-degree polynomial best fit.MAP units are mm/yr. Rainfall color coded as for Fig. 8. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

16 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

◦ have large and roughly equal positive factor loadings. PC1 explains degree quadratic fit.This yields MAT values of 22.6 ± 6.1 C for FL 1–7, ◦ ◦ 81.4% of the overall variance. Factor loadings on PC2 (12.3% of vari­ 21.7 ± 6.1 C for BB, and 22.9 ± 6.1 C for SBB, respectively (Table 10). ance) separates localities with a high generic richness from those with a In extant faunas, the number of herbivorous taxa3 increases with high Browsing Index. Factor loadings on PC3 accounting for 6% of the MAT (Fig. 10A) and decreases with latitude (Fig. 10B). The numbers of ◦ variance separate localities with high Browsing Index from those with a herbivores in SCF faunas, localities that were south of 50 latitude at the high Arboreality Index (Table 9). In least-squares regressions with a time of deposition (Kay et al., 2012), are unlike those of extant faunas of second order polynomial fit, PC1 accounts for 68% of the variance in that latitude but fall within the range of taxa with much higher MAT: all MAP, 72% of the variance in MAT, and 71% of the variance in tem­ extant faunas with >22 herbivorous genera have MAT values above ◦ perature seasonality (Table 8). 20 C. The three SCF faunas have 25, 26, and 32 herbivorous genera, ◦ respectively, giving much warmer predicted temperatures of 24.8 C (FL ◦ ◦ 4.2.2. Estimating climate variables in the Santa Cruz Formation 1-7), 24.9 ± 7.7 C (BB), and 24.3 ± 7.7 C (SBB) (Fig. 10). From the formulae derived by comparison of climate variables in In short, MAT estimates from the Arboreality Index do not differ Table 8, we estimated those parameters in the three levels of SCF. significantlybetween FL 1–7 and SBB and indicate warm conditions. The Table 10 summarizes the predictions with ± 2 sigma from the various estimates for BB based on the Arboreality Index are very anomalous and models. we suspect that they are a consequence of poor sampling of the sloth fauna. This conclusion is supported by the PC analysis and a consider­ 4.2.2.1. Mean annual precipitation. Based on the Browsing Index, which ation of the herbivore richness. ◦ explains 77.8% of the variance of MAP in the extant localities, we esti­ For MAT, machine learning, MAT estimates are ~22 C for FL 1–7, ◦ ◦ mate MAP to have ranged from 1682 to 1860 mm/yr in our Miocene 16 C for BB, and 21 C for SBB. The estimate for BB is significantlylower faunas. While there is a gradual decline in the mean estimates, with the than that for FL 1–7 and SBB (i.e., the 95% confidenceinterval for the BB highest value at FL 1–7 and the lowest at SBB, the differences are slight estimate falls outside the range of estimates for FL 1–7 and SBB). As with and none is significantly different from any other. the other estimates at BB, however, this could be an artifact of the With all three principal components included in a multivariate limited sampling and low generic count of sloths from BB. regression to estimate MAP at the Miocene localities, 79% of the vari­ ance is explained. Removal of PC3 as a variable reduces the explained 4.2.2.4. Temperature seasonality. The Arboreality Indices for SCF faunas variance to 77%. Removing PC2 reduces the explained variance to 57%. FL 1–7, BB, and SBB give us temperature seasonality estimates of ◦ ◦ Therefore, we used both PC1 and PC2 to make the estimates. Combined, ranging from 3.1 to 4.6 C, with the high being the estimate at BB. It is a PC1 and PC2 variable yields MAP estimates of 1484, 984, and 1168 likely that the seasonality value for BB would be lower had we better mm/yr for FL 1–7, BB, and SBB, respectively. Once again, although these representation of semi-arboreal sloth taxa. values fluctuate, the differences are not statistically significant. For temperature seasonality, machine learning, MAT estimates are ◦ ◦ Estimated MAP using the machine learning approaches (Gaussian 3.0 C for FL 1–7, 3.7 for BB, and 3.4 C for SBB. None is significantly process regression and random forest models Spradley et al., 2019) different from the others (i.e., the 95% confidence intervals for all es­ based on generic richness and Arboreality and Browsing indices give timates overlap). similar results: 1600 mm/yr for FL 1–7, 1560 for BB, and 1580 for SBB. 4.2.2.5. Canopy height. Using machine learning, canopy height was 4.2.2.2. Precipitation seasonality. None of our variables does a good job estimated to be ~23 m for FL 1–7, 27 m for BB, and 20 m for SBB. of estimating seasonality of precipitation. Only 17% of the total variance Confidence intervals overlap among all estimates. in precipitation seasonality is explained by the four indices and generic richness, taken individually or in combination. The ‘best’ predictor variable is the arboreality Index, but it explains only 9.6% of the vari­ 4.3. Comparing the SCF paleobiota and paleoenvironments with modern ance. We conclude that our niche variables offer little in the way of localities explanatory power. Similarly uninformative results are found with the machine-learning approaches. A PCA of the variables, generic richness, the Browsing Index, and the Arboreality Index, is described in Table 9 and illustrated in Fig. 11A. PC1 4.2.2.3. Mean annual temperature. Based upon the values of the has evenly balanced factor loadings for all three variables, so any Arboreality Index, we reconstruct the MAT at FL 1–7 and SBB as 21.5 ± combination of high scores will move PC1 in a positive direction. Factor ◦ ◦ 6.6 C and 18.9 ± 6.5 C, respectively. Our estimate for BB is much lower loadings on PC2 separate localities with high numbers of genera from ◦ at 11.9 ± 6.9 C. There is good reason to question this last estimate those with a high Browsing Index. PC3 separates localities with high because BB has a notable scarcity of semi-arboreal sloths: we have no Browsing Index from those with a high Arboreality index. records of Hapalops, Pelecyodon, Hyperleptus, Eucholoeops, and Twelve extant localities in the data set have similar values to the Nematherium. Miocene localities for the first two principal components. All 12 are ◦ ◦ In the PCA, PC1 explains 75% of the variance in MAT using a second between 17 and 27 South, of which eight are found in the Upper Parana´ Atlantic Forest and the humid and dry Chaco ecoregions (Olson et al., 2001) (Fig. 11B and C). Three of these localities (the blue stars in Table 9 Fig. 11B and C) especially resemble the Miocene localities in all three Factor loadings (Eigenvectors) for the first three principal components of a Principal Components Analysis using 3 variables for 46 extant low-altitude lo­ Principal Components: Parque Nacional Iguazú, Misiones, Argentina, calities with rainfall <2500 mm. Reserva Natural del Bosque Mbaracayú, Paraguay, and Reserva de Recursos Manejados San Rafael, Paraguay. These three have generic Variable Principal Principal Principal Component 1 Component 2 Component 3 counts and indices very similar to the Miocene localities. The average (81.4%) (12.3%) (6.3%) number of genera for the three extant localities is 43 (modern range from 9 to 70 genera); the 55% Browsing index is moderate (modern Generic 0.553 0.824 0.123 Richness range 0%–100%), and the 17% Arboreality Index is relatively low Browsing 0.584 0,489 0.648 Index Arboreality 0.594 0.286 0.751 3 The number of herbivorous taxa is the sum of all herbivorous categories (i. Index e., excluding primarily insectivorous, carnivorous, or scavenging taxa.

17 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Table 10 Comparing estimates of environmental variables from this and prior studies of the Santa Cruz Formation.

Variable Study FL 1–7 (~17.6–~17.3 Ma) Barrancas Blancas (~17.2 and ~16.3 Ma) Segundas Barrancas Blancas (~16.5 and ~15.6 Ma)

MAP (mm) Kay et al. 1250–1750 mm–4 estimates none none (2012) Spradley 840–1344 mm —4 estimates none none et al. (2019) Trayler et al. ~1000 ± 235 mm at 17.4 Ma none none (2020a,b) This Paper Regression: 1860 ± 561 mm Principal Regression: 1705 ± 591 mm Principal Regression: 1682 ± 614 mm Principal Components: 1484 ± 594 mm; GPR: 1592 ± Components: 984 ± 600 mm; GPR: 1569 Components:1168 ± 596 mm; GPR: 1540 ± 415 mm RF:1606 ± 279 mm ± 415 mm RF:1546 ± 279 415 mm; RF: 1626 ± 279 mm Precipitation Trayler et al. small to moderate seasonal change in seasonality (2020a,b) precipitation and vegetation composition ◦ MAT ( C) Kay et al. warm– based on climate tolerances of none none (2012) reptiles and amphibians Spradley 15.0-23.4–4 estimates none none et al. (2019) ◦ Trayler et al. 20 ± 4 C at 17.4 Ma none none (2020a,b) ◦ ◦ This Paper Regression: 21.5 ± 6.6 CPrincipal Regression: 11.9 ± 6.9 Principal Regression: 18.9 ± 6.5 C Principal ◦ ◦ ◦ Components: 22.6 ± 6.1 C; GPR: 22.6 ± Components: 21.7 ± 6.1 C GPR: 15.4 ± Components: 22.9 ± 6.1 C; GPR: 21.0 ± 2.7 ◦ ◦ ◦ ◦ ◦ ◦ 2.7 C; RF:22.5 ± 2.7 C C 2.7; RF:14.9 ± 2.7 C C; RF:21.4 ± 2.7 C ◦ Temperature Spradley 3.21 to 4.40 C –4 estimates – – ◦ seasonality ( C) et al. (2019) Trayler et al. – – (2020a,b) ◦ ◦ ◦ This Paper Regression: 3.05 ± 1.64 C GPR: 3.0 ± 0.7 Regression: 4.63 ± 1.68 C; GPR: 3.7 ± Regression: 3.63 ± 1.65 C; ◦ ◦ ◦ ◦ ◦ ◦ C; RF: 3.01 ± 0.8 C 0.7 C; RF: 3.7 ± 0.8 C GPR: 3.3 ± 0.7 C; RF: 3.4 ± 0.8 C Canopy height Spradley 16- 26 meters–4 estimates – – (meters) et al. (2019) This Paper GPR: 22.5 ±6.4 m; RF:24.0 ±7.6 m GPR: 28.1 ± 6.4 m RF:25.97.6 m GPR: 20.7 ± 6.4 m; RF:19.3 ± 7.6 m

Notes: Formulae for the regression models are in Table 8. GPR: Estimate results from Gaussian process regression model. RF: Results from random forest model.

(modern range 0%–51%). By comparison, the mean number of genera in The other localities illustrated in Fig. 11 sample a cline from south­ the Miocene localities is 43 (range 38–47); the mean Browsing Index is east to northwest of decreased rainfall and increased temperature. The 58% (Range 55%–63%), and the mean Arboreality Index is 14% (range localities in the southeast have more continuous forest canopy with a 6%–21%). The remarkably low Arboreality Index at BB has already been minor part savanna; The continuous canopy gives way to the wet Chaco mentioned as being a possible artifact the apparent paucity of generi­ and dry Chaco regions with progressively lower and more discontinuous cally identifiable sloths material. woodlands of semideciduous, xeromorphic trees and shrubs, with gal­ The three above-mentioned extant localities are at low elevation. lery forest along river courses, interspersed with savannas of grasses or ◦ They have an MAT between 21 and 23 C and temperature seasonality forbes. (Hayes and Scharf, 1995; Mereles, 2005). The riverine forests ◦ ◦ between 2 and 4 C. MAP is between 1600 and 1800 mm/yr with low to contain species typical of the Atlantic Forest (Mares et al., 1989). moderate precipitation seasonality and they have an average canopy All the extant localities illustrated in Fig. 11B and C resemble the SCF height up to 25 m. These parameters closely match our reconstruction faunas in having low primate richness: Iguazú, Mbaracayú and San for the SCF. These are located in the Upper Parana,´ or ‘interior’ Atlantic Rafael have two genera, each with one species. Such a low richness need Forest ecoregion extending from southern Brazil into eastern Paraguay not preclude an even more xeric environment than that described above, and the Province of Misiones of Argentina (Fig. 11). These Atlantic forest however. Primates occur in even more arid environments, as in the dry localities have characteristically semi-deciduous sub-tropical tall forests Chaco of western Paraguay at Parque Nacional Defensores del Chaco (3 with closed canopies reaching or exceeding 20 m. However, these forests genera) and Parque Nacional Tinfunque´ (3 genera) with 660 and 680 never form a continuous block but rather are interspersed with a variety mm MAP. of savanna types (low savannas, high savannas, and "dirty" savannas) The armadillo fauna of Atlantic Forest localities is less rich than that (Eiten, 1972) and other types of forest formations (medium forest, gal­ in the SCF, which might suggest a more xeric environment in the SCF lery forest) (Cartes, 2003). Where rainfall is at the lower limit of the than the above environmental reconstruction suggests. There are three target range, these mesic forests transition into woodland savanna genera of armadillos at Iguazú and San Rafael and four at Mbaracayú. (Cerrado denso) in which the herbaceous layer is the ecologically pre­ Vizcaíno et al. (2006) advocated for the use of armadillos as a proxy to dominant component of the ecosystem in terms of biomass production. estimate environmental conditions for the coastal SCF. They note that Shrubs or low trees, may or may not be present, but their crowns never the greatest similarity between SCF and extant faunas in terms of form a closed canopy (Huber, 1987). The localities represented by aqua abundance, richness, and niche diversity of armadillos is found further and purple stars in Fig. 11C show a gradation from mesic to xeric west in the dryer and more seasonal Chaquena˜ Province. This province vegetation. encompasses our extant localities Tinfunqu´e (seven genera) and Parque Nacional Iguazú, located in Misiones Province, Argentina and Defensores del Chaco and Pantanal (five genera). the Reserva de Recursos Manejados San Rafael and the Reserva Natural The above-mentioned extant and the SCF faunas have some del Bosque Mbaracayú are located in Paraguay in the Upper Parana´ specialized myrmecophagous xenarthran species. In the extant faunas, Atlantic Forest ecoregion. The predominant vegetation is semi- they are represented by two anteaters Myrmecophaga and Tamandua and deciduous subtropical forest. Variations in the local environment and at least one armadillo (Priodontes). Myrmecophagous taxa have similar ◦ ◦ type of soil allow for the occurrence of other plant communities—gallery southern range limits between 28 and 34 S, as do the most common forests, bamboo forests, palmito (Euterpe edulis) forests, and araucaria termites upon which they prey (Constantino, 2002; Redford, 1984) the forests (Cacciali et al., 2015; Di Bitetti et al., 2003). most southerly distributions of which occur in southeast Brazil and

18 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Fig. 10. Number of herbivorous taxa versus ◦ A, Mean Annual Temperature ( C) and B, latitude (absolute degrees from equator). The line is the second-degree polynomial best fit. The shaded area is the 95% confi­ dence interval for the mean. Blue, red and green stars are the fossil localities placed on the curve of extant localities based on esti­ mated paleolatitude and herbivore richness. (For interpretation of the references to color in this figurelegend, the reader is referred to the Web version of this article.)

◦ ◦ Uruguay between 27 S and 34.5 S. The specialized feeding habits of common, suggesting forested habits, or at least the occurrence of trees vermilinguans supports our reconstruction of the SCF as a subtropical (Candela et al., 2012). and warm temperate environments (see Kay et al., 2012 for a discus­ The appearance and prevalence in the upper levels of SCF of sion). In the SCF is also notable the presence of the semi-arboreal ant­ euhypsodont rodents, especially Prolagostomus and Pliolagostomus and eater Protamandua (at least in the coastal area) and the strictly the typothere Pachyrukhos suggests a shift away from more closed- myrmecophagous armadillo Stegotherium. canopy habitat to a more savanna-like environment, at least in Another particular similarity between these extant faunas and the patches: increasing crown height may be an evolutionary response to the SCF faunas is the presence of one species of scansorial porcupine (Ere­ silica phytoliths in plant (especially grass) leaves, or to increases in thizontidae, Rodentia). The Santacrucian erethizontid Steiromys is exogenous dietary grit in the form of wind-blown dust, or to both factors

19 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Fig. 11. A. PCA of generic richness and the niche variables Browsing Index and Arboreality Index for extant localities with the position of the three Miocene faunas interpolated from the factor loadings. Localities represented by open triangles are similar to the Miocene localities on the firstand second PCs; those encompassed by the box are similar in all three PCs. B, C. Distribution of MAT and MAP in Paraguay and surrounding regions of Brazil, Argentina, and Bolivia with key localies ◦ ◦ indicated between 19 and 29 S. Localities with red star are in the Dry Chaco; aqua, Wet Chaco; green, Pantanal; blue, Atlantic Forest; and purple, in the transitional between Atlantic Forest and Mesopotamian Grasslands. The three extant localities that closely match the Miocene localities in all three PCs (open triangles encompassed by the box in part A) are in the Atlantic forest block. All named and identified localities in B and C are similar in the first and second PCs (the green triangles in part A). Ecoregions are based on a map of Cacciali (2010). Colors of the dots in part A are coded as for Fig. 8. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

(Covert and Kay, 1981; Kay et al., 1999). Dolichotis, a caviomorph rodent found in open and semi-open habitats in Pachyrukhos is usually depicted as an agile terrestrial animal that Argentina (Mares and Ojeda, 1982; Patton et al., 2015) Maras have moved using saltatorial locomotion. In many ways it resembles a hare pronounced postcranial specializations for fast locomotion– a slender (Lagomorpha), with a short tail, limbs arranged in a parasagittal body and the relatively longest limbs among caviomorph rodents (Cli­ orientation, a long hindlimb with the proximal segment much shorter maco das Chagas et al., 2019; Elissamburu and Vizcaíno, 2004). than the distal, and the tibia and fibula slender and broadly syn­ desmosed (Sinclair, 1908). In South America today lagomorphs are not 5. Discussion very diverse. Despite the name, the Tapeti or Brazilian forest cottontail Sylvilagus brasiliensis (Leporidae) generally occupies transition zones in 5.1. What does the mammal record say about paleoenvironment? the periphery of the Brazilian Amazon, often in intergrading heteroge­ neous forests and savanna (Jos´e De Sousa et al., 2005). The introduced It is important to separate two questions related to the mammalian European hare Lepus europeaus occupies a range of environments, from fossil record of the SCF: what does it tell us about the paleoenvironment the bushy steppes and Andean deserts of Bolivia and Peru to the dry and and what does it indicate about structure of the biotic community (p.333 humid forests and wooded savannas of Paraguay and Brazil (Bonino in Kay et al., 2012). In other words, estimating MAP or MAT on one hand et al., 2010). Pachyrukhos may also be comparable with the mara, and identifying the biotic environment on the other hand are distinct but

20 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296 related questions. This distinction was not always made in the older and using extant South American faunas as analogs. We found that literature. mammalian community structure remained relatively stable throughout There is general agreement in the older literature that temperate the temporal interval from FL 1–7, through BB, to SBB. Nevertheless, conditions with sufficient rainfall to support a mixture of forested and despite great overall similarity of SBB to the two older faunas, the per­ open environments prevailed in the time interval represented by the SCF centage of rodent specimens with euhypsodont cheek teeth is higher at (Pascual and Ortiz-Jaureguizar, 1990). Tauber (1997a) divided the SBB than at FL1-7 or BB, suggesting a trend towards increased aridity. coastal exposures of SFC south of Río Coyle into two members and The euhypsodont chinchillid rodents Prolagostomus and Pliolagostomus evaluated the mammal faunas of each separately. He concluded that the are abundant in SBB whereas they are rare or absent in the earlier SCF climate became progressively less humid and more “open”, that is with levels. Prolagostomus evidently reappears as an immigrant, because a less tree cover, in the upper stratigraphic member when compared with species of Prolagostomus is recorded in the slightly older early Miocene the lower. Comparing faunas from the two levels using the cenogram Pinturas Formation, where it reportedly occurs in a xeric context (Kra­ approach of Legendre (1986), Croft (2001) found the two levels to be marz and Bellosi, 2005); its absence at FL 1–7, and rarity at BB (just two very similar and concluded that no obvious environmental changes had partial teeth; ~1% of rodent specimens) suggests that FL 1–7 and BB occurred. Noting, as had others, that the fauna in both members of the environments were more mesic. Its reappearance would then indicate a coastal SCF includes a combination of arboreal (the primate Homun­ range extension from a nearby more xeric source area. Pliolagostomus culus) and savanna-adapted mammals, he suggested that the region was makes its firstappearance in the Santacrucian at SBB and is much more moister and less open (i.e., having a more continuous canopy in places) common in the succeeding middle Miocene Colloncuran (Bucher et al., than proposed by Pascual and Ortiz-Jaureguizar (1990). Whereas 2020; Vucetich, 1984). Tauber (1997b) suggested that the SCF climate had become less humid Likewise, the reappearance of the euhypsodont cursorial hegetothere and more open during this interval, Croft’s interpretation was that there Pachyrukhos in SBB (it is recorded in the Pinturas Formation (Kramarz might even have been slightly more rainfall in the upper levels. Later et al., 2010) but not at the lower levels of the SCF) suggests a trend Vizcaíno et al. (2006) used armadillos as a proxy to estimate environ­ towards more open environments—i.e., to a more savanna-influenced mental conditions for the coastal SCF and suggested that the armadillo landscape from one where closed-canopy forests were predominate. diversity is consistent with the environmental interpretation of Tauber We hasten to point out that such a trend is not meant to suggest that of open vegetation in relatively dry conditions with marked rainfall grasses came to be more important component of the vegetation, for the seasonality for the upper levels. near-ground herbaceous layer might equally have been composed of forbs, as it is some vegetative communities in the Gran Chaco today. 5.1.1. Estimating MAP and MAT and did it change? The presence of the primate Homunculus has often been taken as an Our estimates of MAP at FL 1–7 broadly overlap the estimates of indicator of forested environments but it need not indicate a closed- other studies by Kay et al. (2012) and Spradley et al. (2019) (Table 10). canopy humid forest. Up to three genera of primates occur in the xeric Our estimates of ~1500–1800 mm/yr are higher than the mean estimate Gran Chaco of Paraguay in gallery forests near rivers and in a dry-forest of ~1000 by Trayler et al. (2020a) for the lower levels of the SCF at vegetation of spiny, thorny shrubs and low trees (Stallings, 1985). Canad˜ on´ de las Vacas, based on tooth stable isotope analysis, but the confidenceintervals of our estimates and those of Trayler et al. (2020a) 5.2. Other sources of evidence overlap broadly. Our comparison of FL 1–7 with the two Río Santa Cruz faunas reveals no significant differences in estimated MAP among the 5.2.1. Geologic considerations localities. Sedimentological studies in the SCF at FL 1–7, BB and SBB suggest ◦ ◦ We estimate an MAT between ~21 C and 25 C, similar to the es­ the unit accumulated in broad floodplains dissected by networks of ◦ ◦ timate of Spradley et al. (2019) of ~15 C to 23 C and in agreement small fluvial channels draining eastward to the Atlantic (Cuitino˜ et al., ◦ with the >20 C estimate of Trayler et al. (2020a). None of our estimates 2019a; Matheos and Raigemborn, 2012). Frequent volcanic explosive show a consistent trend with all values overlapping in their 95% con­ activity resulted in the deposition of ash sediments forming single tuff fidence intervals. Similarly, we find no clear trends in temperature horizons, or intermingled with epiclastic sediments throughout the SCF. seasonality. Thick tuff horizons are more prominent in the lower part of the unit. This sedimentary system remained relatively constant showing minor verti­ 5.1.2. Taxonomic structure versus community structure cal (temporal) changes, and experienced constant aggradation, resulting The mammalian generic richness of the three localities is very similar in the development of abundant poorly developed paleosol horizons and the overall generic faunal similarity among our three fossil localities (Raigemborn et al., 2018a). A slight westward grain size increase of is very high, as indicated by the Simpson Index of Faunal Similarity (SI). channel sandstones was highlighted by Cuitino˜ et al. (2019a) for the RSC But similarity in community structure must not be equated with change localities. Overall, general sedimentologic studies seem to be insensitive in taxonomic composition. As noted by Maas and Krause (1994): to those MAP or MAT variations suggested by other proxies. “The two are equivalent only if taxonomic turnover alters the rela­ tive proportions of ecologically definedgroups (such as trophic groups), 5.2.2. Estimates of paleoenvironment from pedogenic features, isotopes, and thus has an impact on the ecological interactions among species. In and paleobotany contrast, if taxonomic composition changes without altering the relative The mineral composition of the sediments at FL 1–7 (absence of proportions of ecological groups, the result is community stasis (pp. kaolinite and the preservation of weatherable minerals such as volcanic 97–98).” glass and feldspars) suggest a temperate and subhumid climate with In our dataset of extant faunas, we found SI to be more highly precipitation in the range of <1000–1200 mm/yr (Raigemborn et al., correlated with geographic distance then with rainfall, so we should not 2018a). This MAP estimate is in agreement with that of Brea et al. assume a priori that a change in taxonomic composition of a fauna (2017) whose paleobotanical studies of a nearby coastal Estancia la necessarily indicates any fundamental change in community structure. Costa Member locality estimated MAP at ~870 mm/yr, albeit with a Instead, we might better consider whether there are changes in wide range of variation. Using carbon isotope records, Trayler et al. ecologically-defined groups as we have done with locomotion and di­ (2020a) proposed a MAP of ~1000 ± 235 mm/yr at nearby Canad˜ on´ de etary differences. las Vacas at contemporaneous stratigraphic levels (~17.4 Ma) to FL 1–7. To explore how the composition of SCF mammalian faunas might A marked rainfall seasonality (or at least a seasonally fluctuating have been shaped by abiotic factors, we examined substrate preference water table) was inferred from the development of pedogenic calcretes and trophic structure (diet) as inferred from ecomorphological studies, and other features of the paleosols that can form in during soil-drying

21 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296 episodes of long duration (Raigemborn et al., 2018b). Brea et al. (2017) feeding on insects and small animals in the same fashion as living rheas also suggested that there was a dry season of ~7 months, based on (Folch, 1992). Living seriemas range over a variety of semi-open and dry xeromorphic features of wood anatomy. These findings are at variance landscapes in South America, such as the thorny scrub and other with those of Trayler et al. (2020a) whose study of isotope zoning in the semi-arid woodland areas of the Brazilian “caatinga,” the grassy teeth of large SCF herbivores suggested small to moderate seasonal savanna-like “cerrados,” and the “monte” and “chaco” forests in Bolivia, change in precipitation and vegetation compositions. Paraguay, and Argentina (Gonzaga, 1996). Both species are distributed Previously published estimates are in general agreement that MAT in in central Argentina and Paraguay, but the geographic range of the the Estancia La Costa Member was deposited in warm conditions. Brea Black-legged Seriema (Chunga burmeisteri) is not entirely sympatric with ◦ et al. (2012) offers two estimates from wood anatomy: 19.3 ± 1.7 C and the Red-legged Seriema (Cariama cristata) having a more westward ◦ 9.3 ± 1.7 C. Trayler et al. (2020a) used oxygen isotope values to esti­ distribution in Argentina. The Red-legged Seriema (Cariama cristata) is ◦ mate temperatures of at least 20 C for FL 1–7. widely distributed in South America, occurring in Argentina, Uruguay, There are no prior estimates for precipitation, temperature, or sea­ Paraguay, Bolivia, and in Brazil, from the northeast to the southeast sonality at BB or SBB. The closest MAP and MAT estimates for similar- (Redford and Peters, 1986; Sick, 1997). Overlapping the geographic age levels as BB come from the SCF at Canad˜ on´ de las Vacas and ranges of both species (BirdLife International NatureServe, 2014) with Rincon´ del Buque, north of Río Coyle. Trayler et al. (2020a) report a associated habitats of biomes (Olson et al., 2001) corroborates that decreasing MAP between ~17.4 and ~16.9 Ma, from ~1000 mm/yr to Chunga are restricted to drier forested areas, whereas Cariama are 525 mm/yr with a recovery to ~840 ± 270 mm/yr by ~16.4 Ma. At the adapted to more open often mesic environments but they are spatially level of temporal resolution we can achieve, (17.2 Ma to 16.3 Ma for the sympatric in transitional patches between xeric forest and open habitats BB fauna) we would be unlikely to pick up this ephemeral (geologically in the Paraguayan Chaco (Brooks, 2014). speaking) drop in MAP, if it occurred. The BB locality lies more inland from the above-mentioned localities, probably more than 150 km away 5.3. Considering models to analyze paleoenvironments and paleoecology from the Miocene coastline. One would not necessarily expect to finda of Patagonia similar pattern of rainfall in the two regions. Trayler et al. (2020a) reported decreasing MAT to a minimum at 5.3.1. Restricted versus global sampling ◦ ◦ ~17.1 Ma (~17 C), with a rebound to 20–23 C at ~17.0 Ma, and an As mentioned in Section 1, in the last decade there have been two ◦ increase to a maximum of 26 C at about 16.4 Ma. For the temporally approaches for the understanding of the biota and environments of the ◦ equivalent interval at BB, we estimate a MAT of 21.6 ± 6.1 C, which SCF. One focuses on stratigraphically and geographically restricted fossil falls within the range estimated by Trayler et al., once the reduced samples obtained on recent field collections (Rodríguez-Gomez´ et al., temporal resolution is accounted for. 2020; Spradley et al., 2019; Vizcaíno et al., 2010, 2012a), while the The temporal span of SBB (16.5 Ma to 15.6 Ma) with a slight increase other considers a composite list of fossils recorded from the entire in MAT and a stable MAP represents a geologically younger interval not temporal and geographic distribution of the formation based on his­ sampled by Trayler et al. (2020a), but they do report a temperature torical collections (Croft, 2013). increase towards the end of their record at about 16.5 Ma. Croft (2013) identifiedthree main factors that hindered the study of Stable isotope and paleobotany studies are in general agreement Santacrucian faunas from a synecologic perspective: (1) the unknown about the probable biotic environment of the SCF. Brea et al. (2017; precise geographic and/or stratigraphic provenance of many specimens 2012) reconstructed the biota as being a mixture of open temperate and from the old collections, which made it difficultto assess to what degree semiarid forests in the lower part of coastal SCF. That agrees with the fossil assemblages of the SCF had been affected by time averaging 13 Trayler et al. (2020a) suggestion, based on δ C values, that and other factors; (2) taxon inflation– many named species and even large-bodied herbivores were likely mixed feeders or browsers, moving genera are likely invalid due to the fragmentary nature of the type between wooded and open areas but not occupying fully closed-canopy specimens and lack of appreciation of the range of variation likely systems. encountered in a species, thereby potentially inflating the number of valid taxa; and (3) absence of a comprehensive stratigraphic framework 5.2.3. Evidence from other faunal elements for seriating various known localities of the SCF. Croft’s publication does Non-mammalian elements of the fauna also contribute to our un­ not consider the results of a recently published volume on a long-term derstanding of Santacrucian paleoenvironments. The frog Calyp­ comprehensive research program on the geology and paleontology of tocephalella, and the lizard Tupinambis (Salvator), indicates that the the SCF along the Atlantic coast (Vizcaíno et al., 2012a). Coincidently, climate of FL 1–7 was much warmer and wetter than in this region today the Vizcaíno et al. volume addressed the same issues mentioned by (Kay et al., 2012). The presence of Calyptocephalella seems to indicate Croft, and initiated efforts to solve them: it was based on specimens with permanent lowland lakes, ponds, and quiet streams, possibly developed precise geographic and/or stratigraphic information and a revised syn­ in a forested area (Fernicola and Albino, 2012). Calyptocephalella was thetic taxonomy at the generic level, to place multiple widespread lo­ recently reported for the first time in the SCF at Barrancas Blancas calities into a geochronological framework (Fleagle et al., 2012; Perkins (Muzzopappa, 2019), extending the same environmental interpretations et al., 2012). to the central area of Santa Cruz, which is also supported by the record of Croft (2013) critiqued an article by Vizcaíno et al. (2010), who the freshwater bivalve Diplodon (Perez´ et al., 2019). analyzed mammal richness based on faunal lists from recently collected Based on the extrapolation of habitat preferences of extinct birds in fossil samples from restricted Atlantic coastal localities of the SCF, the SCF from their living analogs, Degrange et al. (2012) identified the excluding otherwise valid Santacrucian taxa of uncertain provenance. probable presence of alternating areas of herbaceous vegetation with Croft (2013) noted that Vizcaíno et al. (2010) omitted some well-known shrubby or wooded areas. The habitat preferences of extant rheas, Santacrucian taxa. He endeavored to test whether such absences could tinamids and seriemas (analogs to SCF flightless Santacrucian phor­ be a consequence of the smaller sample sizes by using a family-level usrhacids) are consistent with open areas because of their rarefaction analysis of the YPM-VPPU Collection. He attempted to ground-dwelling cursorial capabilities. The presence of waterfowls, reconcile absences using distributional data from recent and historical limpkins, spoonbills and darters would indicate the existence of SCF collections, many with uncertain stratigraphic information. The temporarily flooded savannas or permanent water bodies in forested analysis indicated that most single-locality samples from the SCF have areas (Degrange et al., 2012). lower familial diversity than expected based on the taxonomic distri­ Degrange et al. (2012) considered Santacrucian rheas as mainly bution of specimens in the YPM-VPPU collection as a whole. He grazers, roaming open grasslands, scrub forests, or chaparral, as well as concluded that, given the large regional extent of the SCF and the large

22 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296 geographic area encompassed by most modern communities used for of extant species to model the structure-function and infer the behavior comparison, single-locality samples are not necessarily more appro­ of extinct species) have been and continue to be one of the most robust priate than multi-locality samples for paleoecological analyses of San­ paradigms of Earth sciences. It is often expressed with the mantra ‘the tacrucian mammal communities. present is the key to the past’. In paleontology, it is common to infer We do not findthis argument to be persuasive because it substitutes paleoclimatic and paleoenvironmental conditions from the distribution one possible source of error for another. On one hand, smaller samples of extant close relatives of extinct taxa whose paleobiology is to be risk not recovering rare taxa that may well have been identifiedif more inferred, as has been done for the Santacrucian fauna by many authors specimens had been collected—the presence of sparassodont genera in (Croft, 2001; Kay et al., 2012; Tauber, 1997b; Vizcaíno et al., 2010). the old collections that we did not find in our collections are a case in However, although this approach is appropriate in general, its restricted point. On the other hand, inclusion of more geographically disparate and uncritical application may result in hypotheses that are not well samples collected without well-corroborated stratigraphic control supported and have little or no heuristic value. Vizcaíno et al. (2017a) makes it all but certain that there will be stratigraphic (temporal) called this " naïve actualism." An obvious example is the naïve inference averaging. For instance, we know now that two of the purported missing that would be drawn for the records of the Pleistocene in Patagonia and taxa at FL 1–7 mentioned by Croft, Pachyrukhos and Prolagostomus, are the pampas region of the Jaguar Felis onca, a species that today lives recorded only from stratigraphically higher parts of the SCF along the exclusively in tropical closed environments but that in historical times RSC (Fernicola et al., 2019c) at localities other than those studied by had a much wider distribution including in arid environments where it is Vizcaíno et al. (2010), and along the Río Gallegos (Raigemborn et al., now extirpated. As mentioned by Kay et al. (p. 361 in 2012), another 2018b), and at Rincon´ del Buque and in the upper levels of Canad˜ on´ de example would be the assignment of equivalent conditions to those of las Vacas north to the Río Coyle (Fig. 1), an issue that we are expanding the Andean-Patagonian forests for the early Miocene of Patagonia, due in another contribution. to the present distributional limits of microbiotheriid marsupials, or an unwarranted inference that the paleoenvironment of the Santacrucian 5.3.2. The problems of modern analogs might have resembled a high-elevation cold and wet environment like SCF assemblages evaluated here may represent mammalian com­ that inhabited by extant Andean caenolestid marsupials owing to the munities without close modern analogs. First, the SCF faunas are presence of the caenolestid paucituberculatan Stilotherium dissimile. dominated by representatives of mammalian orders or suborders with Paleoenvironmental and paleoclimatic reconstructions require a more few or no living survivors (Litopterna, Astrapotheria, Notoungulata, or comprehensive analysis that involves the weighing of many strands of Folivora and Cingulata with reduced numbers), by taxa that are wholly biotic and abiotic evidence because intrinsic and extrinsic factors con­ or partially replaced by members of other orders, e.g. Sparassodonta by ditioning the current distribution of taxa do not necessarily reflecttheir Carnivora (Carrillo et al., 2020) and by the arrival and explosive radi­ maximum ranges of environmental and climatic tolerances (Vizcaíno ation of sigmodontine rodents, of which there are extant 74 genera et al., 2017a). (D’Elía et al., 2006). Nevertheless, we are confident that our extant Also, the morphology of an extinct taxon is conditioned by a complex South American localities are suitable analogs for paleoenvironmental interplay of phylogeny and adaptation. In this way, it is not safe to as­ reconstructions because they have been shown to be generalizable to sume that an animal’s morphology is solely a function of the niche and widely differing mammalian faunas. Spradley et al. (2019), using habitat that its relatives occupy today. This is precisely the case for many species-level data (versus our use of generic data) and similar trophic, extinct xenarthrans, a group abundant in Santacrucian times and and locomotor/substrate preference categories found that a model throughout the Cenozoic of South America. Large-bodied sloths and derived from extant South American species data gives reliable results glyptodonts are morphologically so different from their living sloth and when applied to the geographically and phylogenetically distinct faunas armadillo relatives as to readily suggest they had very different modes of of Australia. life; they have no modern analogs and a simplistic application of an A second and unavoidable deficiency in our approach is that the actualistic approach may produce nonsensical reconstructions (Viz­ fossil localities of SCF exist in habitat zones that do not exist today: caíno, 2014; Vizcaíno et al., 2006, 2018). Again, this does not invalidate remembering that the SCF latitude in the early Miocene was similar to actualism. In such situations, more extended comparisons must be made today, there are no modern equivalents of warm regions so far from the with other mammals using biomechanical approaches that address equator. This challenge is exemplified by what could be called “the form-function relationships (Vizcaíno et al., 2018). frugivore problem.” There are surprisingly few frugivorous genera in our The "the present is the key to the past" is not the only paradigm in SCF localities. The Frugivore Index (FI) in our lowland faunas, including historical branches of Earth sciences. Different disciplines have also those with rainfall exceeding 2500 mm, has a median of 73% with a proposed that the "the past is the key to the present" or even “for the range of 20%–91%, whereas the FI of the three SCF faunas is 14%, 14%, future”. For instance, within ichnology, the fact that some structures and 17% (Fig. 7B). This would predict a very low rainfall, less than one- were originally described and interpreted from the fossil record, and third that predicted by the proportions of browsing or arboreal species. only later recorded in the modern deep sea, has been regarded as It is more likely, however, that the low number of mammalian frugivores “reverse uniformitarianism” (Frey and Seilacher, 1980). In geology, a in our sample is a consequence of the seasonality of fruit production at major frontier of geological research was initiated in the 1970s, that high latitudes, irrespective of rainfall or habitat. Much of the variance involves predicting future geologic trends or events through study of the (62%) in FI is explained by latitude alone whereas only 44% of the present and past, rather than trying to understand the past, often using variance in FI is explained by rainfall. In this sense, having a temperate what one knows about the present (Doe, 1983). ◦ ◦ climate with MAT exceeding 20 C at 50 S latitude, when the day length It may be the case that the past can also be a key to understanding the must be highly seasonal, is outside the worldwide range of available past. Vizcaíno et al. (2004) suggested the possibility of fossil forms modern terrestrial environments. Even if fruit were plentiful, it would acting as biomechanical analogs for other fossils belonging to different have been seasonal and the non-volant mammalian frugivore niche lineages and without living representatives. The example they used was would likely have been occupied by volant mammals and birds that precisely a Santacrucian form, the armadillo Peltephilus, as a model for could migrate. interpreting the enigmatic Palaeocene Ernanodon antelios from Asia. For Peltephilus, a biomechanical study of its masticatory apparatus (Vizcaíno 5.3.3. The Santacrucian as a reference model for other Cenozoic biotas of and Farina,˜ 1997) revealed a pattern not recorded among extant mam­ Patagonia mals but that would be similar to that of Ernanodon. Uniformitarianism (the constancy of physical variables and processes This approach can be expanded beyond the specific level. As through Earth history) and its derivative actualism (in this case the use mentioned in Section 2.2, Simpson (1980) noted the Santacrucian is

23 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296 particularly important for understanding a phase in the history of South abundance of euhypsodont caviomorph rodents and small typotheres American mammals in which the communities consisted of a complex between the lower (BB) and upper (SBB) parts of the Santa Cruz For­ mixture of descendants of ancient lineages of the continent and new taxa mation. These differences may indicate somewhat more arid conditions (primates and rodents) from other land masses. Due to the abundance prevailed in the upper levels of the formation. Further calibration of and quality of its specimens, it also represents a unique opportunity for a these appearances and of potentially cyclic environmental changes profound and comprehensive understanding of the paleobiology and would require of exhaustive collections of in situ remains of different paleoecology of other, less well documented Oligocene and Miocene mammalian taxa, especially of smaller mammals, which seem to yield South American mammal communities. Here and in previous contribu­ the most obvious differences in community structure, as well as other tions (Kay et al., 2012; Vizcaíno et al., 2010), we have selected a series of evidence, through the whole stratigraphic column. localities representing a sufficiently restricted time interval and This contribution also highlights the value of focusing on strati­ geographic spread to develop a paleosinecological analysis. The Santa­ graphically well-controlled fossil samples from restricted geographic crucian now becomes a model that can be applied to the study of areas for reconstructing paleoecological aspects of extinct communities. post-rodent/primate arrival but pre-GABI South American faunas, be Once that has been done, one can move to a broader scenario to make a they savannas or forests. Particularly meaningful features of the model real contribution to the understanding of the paleoecology of a partic­ are that the taxa are closer phylogenetically, morphologically and in ular geologic formation. This applies to the Santa Cruz Formation in time to members of older faunas. Vizcaíno (2014) proposed this particular, but also for other mammal-dominated faunas prior to the approach, with fossil forms acting as analogs for other fossils belonging Great American Biotic Interchange, at least. to lineages with distant or without living representatives, as an inter­ esting methodological tool for the study of evolutionary paleobiology. It Author statement may be called “fosilism” or, more specificallyin this case “Santacrucism”. Richard. F. Kay: Conceptualization, fieldwork, Methodology, Data 6. Summary and conclusions curation, Writing – original draft, reviewing, and editing. Sergio Viz­ caíno: Conceptualization, fieldwork, Methodology, Data curation, Kay et al. (2012, pp 358–359) concluded that the biome in which the Writing – original draft preparation, reviewing, and editing. M. Susana Santacrucian fauna of FL 1–7 lived consisted of a “mosaic of open Bargo: Conceptualization, fieldwork, Methodology, Data curation, temperate humid and semi-arid forests, with lakes in some areas and Writing – original draft preparation, reviewing, and editing. Jackson seasonal flooding in others, no doubt promoting the formation of Spradley: Fieldwork, Data curation, Methodology, reviewing and edit­ marshlands with a mixture of grass and forbes”. As already mentioned, ing. Jose´ Cuitino:˜ Conceptualization, fieldwork, Methodology, Data this description fits the definition of a savanna, as described by Huber curation, Writing – original draft preparation, reviewing, and editing. (1987) as a tropical or subtropical biome in which the herbaceous layer is the ecologically predominant compartment of the ecosystem in terms Declaration of competing interest of biomass production. Shrubs or low trees, may or may not be present, but their crowns never form a closed canopy. The authors declare that they have no known competing financial Based upon our analysis, the extant faunas with greatest similarity to interests or personal relationships that could have appeared to influence the SCF faunas are compatible with our reconstruction of MAP an MAT the work reported in this paper. for semi-deciduous forests with savanna components and that can form at elevations not exceeding 1000–1200 masl when frost rarely occurs, Acknowledgements ◦ MAT exceeds 18 C, and MAP exceeds 500 mm, whether seasonally distributed over the year or not (Huber, 1987). To the extent that MAT We thank the Direccion´ de Patrimonio, Secretaría de Cultura de ◦ remained above 18 C, which all models predict, but precipitation was Estado of the Santa Cruz Province, and the Museo Regional Provincial P. greater, as our new analysis suggests, the proportions of closed-canopy M.J. Molina, Río Gallegos for permissions and assistance for fieldwork. semi-deciduous forests would have increased in proportion to the We thank specially L. Acosta, L. Chornogubsky, J. Fernicola, L. Gonza­ savanna component. This range of habitats occurs today where the lez, S. Hernandez´ del Pino, V. Krapovickas, N. Munoz,˜ A. Racco, and N. mesic inland Atlantic forests of northeastern Argentina give way Toledo for their collaboration during field work. We thank reviewers northwestward into the more xeric Paraguayan wet and dry Chaco María Encarnacion´ Perez´ and David Polly for their insightful comments. (Fig. 11). This interpretations are in general agreement with the other Dr. Shineng Hu kindly provided the maps of climate variables used in sources of evidence from sedimentology, paleosols, isotopes, paleo­ Figs. 4 and 11. This work was supported by the Universidad Nacional de botany and other faunal elements discussed above. La Plata (grant UNLP11/N867); Agencia Nacional de Promocion´ Cien­ The broad picture of our analysis shows persistent faunal stasis tífica y T´ecnica (grant PICT 2017-1081), National Geographic Society during almost the entire time interval of deposition of the Santa Cruz (grant number 9167-12), and the National Science Foundation (grants Formation. A close analysis reveals differences in the presence and 0851272 and 1348259).

Appendix C. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.jsames.2021.103296.

Appendix A. Generic richness, predator and prey richness and niche parameters for 55 extant and three fossil faunas

Locality Latitude Longitude Elevation Mean Annual Mean Annual Total Frugivore Browsing Arboreality Predator/ (masl) Temperature Precipitation number of Index Index Index prey Ratio ◦ ( C) (mm) genera

Parque Nacional 47.67 68.17 232 10.5 222 21 40 0 5 110 Bosques (continued on next page)

24 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

(continued )

Locality Latitude Longitude Elevation Mean Annual Mean Annual Total Frugivore Browsing Arboreality Predator/ (masl) Temperature Precipitation number of Index Index Index prey Ratio ◦ ( C) (mm) genera

Petrificados de Jaramillo, Santa Cruz, Argentina Reserva de Biosfera 34.03 67.92 568 16.1 289 23 27 0 11 109 de Nacu˜ n˜an,´ Mendoza, Argentina Parque Nacional 37.95 65.65 305 14.9 371 20 40 0 15 100 Lihu´e Calel, La Pampa, Argentina Parque Nacional 21.25 61.50 234 23.8 624 32 63 14 16 68 Teniente Enciso, Gran Chaco, Paraguay Parque Nacional 20.22 60.29 170 24.9 660 40 81 25 24 81 Defensores del Chaco, Gran Chaco, Paraguay Parque Nacional 54.83 68.50 227 4.1 664 10 50 0 10 150 Tierra del Fuego, Tierra del Fuego, Argentina Parque Nacional 23.78 60.22 138 23.1 681 48 60 30 18 84 Tinfunque, Paraguay Low Montane, Salta, 23.50 64.00 337 22.5 782 23 58 40 24 92 Argentina1 Chaco, Salta, 22.40 63.00 283 23.1 817 34 44 11 12 113 Argentina1 La Poligonal, Buenos 37.33 59.17 286 13.2 857 24 27 25 10 118 Aires, Argentina Laguna de Mar 30.50 62.67 69 18.6 866 19 38 20 11 138 Chiquita, Buenos Aires, Argentina Transitional Forest, 22.50 64.00 562 21.8 1008 39 65 50 22 129 Salta, Argentina1 Yungas Piedmont, 23.00 64.33 332 22.6 1012 44 73 33 23 100 Salta, Argentina2 Parque Nacional 40.16 71.36 647 9.8 1068 29 36 22 9 107 Lanin, Neuquen,´ Argentina Parque Nacional 22.50 42.25 20 23.2 1114 43 82 25 28 95 Serra do Cipo, Minas Gerais, Brasil Parque Estadual do 19.65 42.55 329 23.4 1149 38 75 50 32 58 Rio Doce, Minas Gerais, Brasil Parque Nacional Lago 42.18 71.68 584 8.7 1154 9 25 33 0 125 Puelo, Chubut, Argentina Parque Nacional Los 50.00 73.25 242 7.4 1173 14 20 25 11 180 Glaciares, Santa Cruz, Argentina Parque Nacional El 31.85 58.32 27 18.4 1181 19 44 20 13 111 Palmar, Entre Ríos, Argentina Reserva de biosfera 33.50 54.00 17 17.1 1189 23 40 17 11 130 Banados˜ del Este, Uruguay Acurizal, Mato 17.75 57.62 96 26 1200 39 77 40 33 73 Grosso, Brazil1 Pantanal, Mato 17.72 57.37 95 26.1 1205 50 68 38 22 100 Grosso, Brazil Parque Nacional 28.02 58.02 73 21.5 1271 20 38 40 13 150 Mburucuya,´ Corrientes, Argentina Hato Masaguaral, 8.57 67.58 67 27.4 1373 28 73 50 32 87 Guarico, Venezuela1 Parque Nacional 22.65 56.18 265 23.1 1397 26 85 50 23 100 Cerro Cora,´ Amambay, Paraguay (continued on next page)

25 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

(continued )

Locality Latitude Longitude Elevation Mean Annual Mean Annual Total Frugivore Browsing Arboreality Predator/ (masl) Temperature Precipitation number of Index Index Index prey Ratio ◦ ( C) (mm) genera

Parque Nacional 26.25 56.80 142 21.5 1418 42 76 50 29 68 Ybycuí, Paraguarí, Paraguay Parque Estadual 24.13 47.97 537 18.1 1429 42 78 40 25 83 Carlos Botelho, Sao˜ Paulo, Brazil Cerrado, Mato Grosso 20.00 54.50 419 24.1 1458 53 74 43 25 93 do Sul, Brazil Pampa, Rio Grande 29.26 52.34 569 17.6 1460 44 58 25 10 121 do Sul, Brazil Parque Nacional Noel 14.27 60.87 252 23.8 1536 55 73 44 34 67 Kempff Mercado, Santa Cruz, Bolivia Valle de Cuna, 27.09 54.95 198 20.2 1575 41 75 50 24 71 Misiones, Argentina Reserva Natural del 24.11 55.46 200 22.3 1631 51 75 57 20 82 Bosque Mbaracayú, Paraguay Reserva de Recursos 26.62 55.63 311 20.6 1705 43 73 50 16 95 Manejados San Rafael, Paraguay Otishi National Park, 11.67 73.08 428 25.5 1745 57 86 60 41 63 Cuzco and Junin, Peru Parque Nacional 25.62 54.33 223 20.7 1759 35 53 57 16 133 Iguazú, Misiones, Argentina Parque Nacional 13.80 67.63 201 25.8 1759 65 72 64 38 64 Madidi, La Paz, Bolivia Reserva de la Biosfera 14.64 66.29 184 25.2 1788 39 65 63 37 70 y Estacion´ Biologica´ del Beni, Beni, Bolivia Parque Nacional da 23.00 44.52 96 23 1858 35 84 33 30 84 Serra da Bocaina, Río de Janeiro and Sao˜ Paulo, Brazil Bosque de Iwokrama, 4.50 59.00 97 26.8 2021 53 83 80 36 77 Guyana Puerto Paez,´ Apure, 6.38 67.48 48 28.2 2162 21 69 50 36 62 Venezuela1 Reserva Nacional 12.33 69.30 283 25 2241 65 79 75 42 67 Tambopata, Peru Manaus, Amazonas, 2.50 60.00 133 27 2319 44 89 100 44 63 Brazil1 Brownsberg Nature 4.94 55.18 335 24.7 2447 51 83 80 39 70 Park, Suriname Puerto Ayacucho, 5.25 67.67 131 27.8 2464 39 84 75 46 56 Amazonas, Venezuela1 Parque Nacional 6.17 62.50 589 25.1 2470 61 77 71 35 94 Canaima, Bolivar, Venezuela Estacion´ Biologica´ 12.00 70.00 278 24.9 2495 60 82 71 50 54 Cocha Cashu, Peru1 Belem,´ Para, Brazil1 1.45 48.48 25 26.8 2528 52 79 67 36 79 Paracou, French 5.38 52.90 12 26 2650 52 87 100 41 68 Guiana Rio Cenepa (Alto 4.78 78.28 253 25.1 2780 55 81 67 41 72 Maranon), Amazonas, Peru1 Esmeralda, 3.08 65.58 148 26.7 2860 49 91 67 48 48 Amazonas, Venezuela1 Parque Nacional 1.08 75.92 226 25.3 2871 59 80 71 43 66 Yasuní, Ecuador Ecuador Tropical, 1.00 76.50 271 24.9 3149 65 81 71 44 81 Oriente, Ecuador1 Parque Nacional del 12.78 71.22 618 24.2 3259 46 77 57 51 48 Manú, Madre de Dios and Cuzco, Peru (continued on next page)

26 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

(continued )

Locality Latitude Longitude Elevation Mean Annual Mean Annual Total Frugivore Browsing Arboreality Predator/ (masl) Temperature Precipitation number of Index Index Index prey Ratio ◦ ( C) (mm) genera

Estaçao˜ Ecologica´ de 2.03 50.43 8 26.6 3424 40 70 57 39 61 Maraca,´ Roraima, Brazil Río Caura, Bolivar, 5.00 64.00 376 25 3632 65 81 71 40 76 Venezuela Santa Cruz Formation 51.00 69.14 <200 44 14 63 21 57 FL 1-7 Santa Cruz Formation 50.27 70.28 <200 38 17 56 6 27 Barrancas Blancas Santa Cruz Formation 50.17 70.68 <200 47 14 55 16 31 Segundas Barrancas Blancas Notes: Except as noted refer to supplemental documents in Spradley et al., 2019. Note 1: details in Kay and Madden (1997a,b). Note 2.: Finca El Carmen, (Jayat and Ortiz, 2010)

Appendix B. Niche assignments for the mammalian genera of the Santa Cruz Formation

TAXON PRESENCE/ NICHE ASSIGNMENT NICHE ASSIGNMENT ABSENCE (Numeric- see Table 1)

Higher level Genus FL BB SBB Body mass BM category Diet Diet Locomotion Substrate Body Locomo- Primary classification 1-7 (BM) kg category preference size tion diet

MERIDIOLESTIDA, Necrolestes 1 0 0 very small I insectivorous I(F) fossorial T(F) 1 6 6 Necrolestidae Ameghino, 1891 SPARASSODONTA, Cladosictis Ameghino, 1 0 1 6.60 III hypercanivorous V scansorial ? A(T) 3 4 1 Hathliacynidae 1887 Sipalocyon 1 1 1 2.11 III hypercanivorous V scansorial A(T) 3 4 1 Ameghino, 1887 Perathereutes 1 0 1 1.00 II hypercanivorous V scansorial ? A(T) 2 4 1 Ameghino, 1891 Borhyaenoidea Prothylacynus 1 0 0 31.79 IV hypercanivorous V scansorial A(T) 4 4 1 Ameghino, 1891 Lycopsis Cabrera, 1 0 0 20.07 IV hypercanivorous V scansorial A(T) 4 4 1 1927 Borhyaenidae Arctodictis Mercerat 1 0 0 50.00 IV hypercanivorous/ V the most T(A) 4 1 1 1891 scavenger terrestrial Borhyaena Ameghino, 1 1 1 36.40 IV hypercanivorous/ V the most T(C) 4 1 1 1887 scavenger terrestrial Acrocyon Ameghino, 1 0 0 11.49 IV hypercanivorous V scansorial A(T) 4 4 1 1887 PAUCITUBERCULATA, Palaeothentes 1 1 1 0.082 I Insectivorous I(F) terrestrial/ T(A) 1 2 6 Palaeothentiidae Ameghino, 1887 cursorial/ saltatorial Abderites Ameghino, 1 1 1 0.487 II frugivorous F(I) ? ? 2 ? 7 1887 Acdestis Ameghino, 1 1 1 0.344 II frugivorous F(I) ? ? 2 ? 7 1887 Microbiotheriidae Microbiotherium 1 0 1 0.061 I insectivorous I(F) arboreal A 1 3 6 Ameghino, 1887 CINGULATA, Peltephilus Ameghino, 1 1 1 11.00 IV herbivorous: roots, S(Tu) terrestrial T(F) 4 6 3 Peltephilidae 1887 tubers, scavenging? diggers Dasypodidae Proeutatus Ameghino, 1 1 1 15.00 IV omnivorous/ S(L) terrestrial T(F) 4 6 4 1891 herbivorous diggers Stegotherium 0 1 0 11.5 IV myrmecophagous MYR digger T(F) 4 6 5 Ameghino, 1887 Stenotatus Ameghino, 1 1 1 4.00 III omnivorous/ S(I) terrestrial T(F) 3 6 2 1891 insectivorous diggers Prozaedyus 1 1 1 1.00 II omnivorous/ S(I) terrestrial T(F) 2 6 2 Ameghino, 1891 insectivorous diggers Propalaehoplophoridae Cochlops Ameghino, 1 1 0 82.99 IV selective browser L terrestrial T(A) 4 1 10 1889 ambulatory Propalaehoplophorus 1 0 0 81.64 IV selective browser L terrestrial T(A) 4 1 10 Ameghino, 1887 ambulatory Eucinepeltus 0 1 1 115 kg V selective browser L relatively T(A) 5 1 10 Ameghino, 1891 closed environments VERMILINGUA, Protamandua 1 0 0 4.00–5.00 III myrmecophagous MYR terrestrial A(T) 3 4 5 Myrmecophagidae Ameghino, 1904 semiarboreal FOLIVORA, Hapalops Ameghino 1 1 1 46.29 IV herbivorous L Terrestrial/ A(T) 4 4 10 Megatherioidea 1887 folivorous Semiarboreal Pelecyodon Ameghino 1 0 0 50.00 IV herbivorous L terrestrial/ A(T) 4 4 10 1891 Folivorous semiarboreal (continued on next page)

27 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

(continued )

TAXON PRESENCE/ NICHE ASSIGNMENT NICHE ASSIGNMENT ABSENCE (Numeric- see Table 1)

Higher level Genus FL BB SBB Body mass BM category Diet Diet Locomotion Substrate Body Locomo- Primary classification 1-7 (BM) kg category preference size tion diet

Hyperleptus 1 0 0 IV herbivorous L terrestrial/ A(T) 4 4 10 Ameghino 1891 folivorous semiarboreal Megalonychidae Eucholoeops 1 0 0 78.00 IV herbivorous L terrestrial/ A(T) 4 4 10 Ameghino, 1887 folivorous semiarboreal Mylodontidae Nematherium 1 0 1 95.02 IV herbivorous, tubers, G terrestrial/ A(T) 4 4 11 Ameghino 1887 roots Semiarboreal Schismotherium 0 0 1 37.99 IV herbivorous L terrestrial/ A(T) 4 4 10 Ameghino, 1887 Folivorous Semiarboreal Xyophorus Ameghino 0 0 1 IV herbivorous L terrestrial/ A(T) 4 4 10 1887 folivorous Semiarboreal Megatheriidae Planopinae 0 0 1 108 V herbivorous L Terrestrial T 5 10 folivorous ASTRAPOTHERIA, Astrapotherium 1 1 1 921.32 VI browser/mixed- L- G ambulatory T(A) 4 5 11 Astrapotheriidae Burmeister, 1879 feeder/closed (semiacuatic?) habitats TYPOTHERIA, Interatherium 1 1 1 2.38 III grazers in open G cursorial/ T(C) 3 5 11 Intheratheriidae Ameghino 1887 habitats semifossorial (semiacuatic?) Protypotherium 1 1 1 7.73 III grazers in open G cursorial T(C) 3 1 11 Ameghino (1885) habitats Hegetotheriidae Hegetotherium 1 1 1 7.71 III grazers in open G cursorial/ T(C) 3 1 11 Ameghino, 1887 habitats semifossorial Pachyrukhos 0 0 1 2,5 III grass and leaves G/L open habitat/ T(C) 3 1 11 Ameghino, 1885 (note1) Cursorial TOXODONTIA, Homalodotherium 1 1 1 405.08 V browser L ambulatory T(A) 5 1 10 Homalodotheriidae Flower, 1873 Toxodontidae Adinotherium 1 1 1 100.29 V grazers in open G ambulatory T(A) 5 1 11 Ameghino, 1887 habitats Nesodon Owen, 1847 1 1 1 637.51 VI intermediate G/L ambulatory T(A) 6 1 11 between grazing & browsing (note 2) LITOPTERNA, Diadiaphorus 1 1 1 82.05 IV browsers in closed L cursorial T(C) 4 1 10 Proterotheriidae Ameghino, 1887 habitat Tetramerorhinus 1 1 1 35.06 IV browsers in closed L cursorial T(C) 4 1 10 Ameghino, 1894 habitat Thoatherium 1 1 1 24.20 IV browsers in closed L cursorial T(C) 4 1 10 Ameghino, 1887 habitat Anisolophus 0 1 1 36.61 IV browsers in closed L cursorial T(C) ? ? 10 Burmeister, 1885 habitat Macraucheniidae Theosodon Ameghino, 1 1 1 121.55 V browsers in closed L cursorial T(C) 5 1 10 1887 habitat CHINCHILLOIDEA, Perimys Ameghino, 1 1 1 0.32 I mixed grazer/ G/L cursorial T(C) 1 2 11 Neoepiblemidae 1887 browser (note 3) Chinchillidae Pliolagostomus 0 0 1 100–1000 ?grasses (note 4) G/L T(C) 2 2 11 Ameghino, 1887 Prolagostomus 0 1 1 100–1000 ?grasses (note 4) G/L T(C) 2 2 11 Ameghino, 1887 Dinomyidae Scleromys Ameghino, 0 1 1 ~5 kg (III) III Browser (note 5) L(F) per ? ? 3 ? 10 1887 Neoreomys CAVIOIDEA, Neoreomys 1 1 1 7.12 III buds, leaves, fruits, L(F) cursorial T(C) 3 2 10 Dasyproctidae Ameghino, 1887 seeds (note 6) Eocardiidae Eocardia Ameghino, 1 1 1 1.4 III grasses, herbs, leaves G cursorial T(C) 3 2 11 1887 of shrubs and trees (more abbrassive than extant forms) (note 7) "Eocardia" excavata 1 1 1 1.5 III grasses, herbs, leaves G cursorial T(C) 3 2 11 Ameghino, 1891 of shrubs and trees (more abbrassive than extant forms) (note 7) Phanomys Ameghino, 0 1 1 0.80 II Figerous plants (note L inhabited open T(C) 2 2 10 1887 8) environments (cursorial?). Schistomys 0 0 1 ~1 kg (II/ II/III grazer G/L Cursorial T(C) 2 2 11 Ameghino, 1887 III) as for Phanomys) OCTODONTOIDEA Stichomys Ameghino, 0 1 1 0,79 Arnal: 100gr- Browser, fed on L(F) ? ? 2 ? 10 1887 1kg plants somewhat more fibrous than the brachiodont taxa (note 9) (continued on next page)

28 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

(continued )

TAXON PRESENCE/ NICHE ASSIGNMENT NICHE ASSIGNMENT ABSENCE (Numeric- see Table 1)

Higher level Genus FL BB SBB Body mass BM category Diet Diet Locomotion Substrate Body Locomo- Primary classification 1-7 (BM) kg category preference size tion diet

Prospaniomys 0 1 1 I (by analogy frugivore/ F(l) T(F) 1 6 7 Ameghino (1902) with insectivore (note 7) Spaniomys) Spaniomys Ameghino, 1 1 1 0,65 Arnal: 100gr- Plant eater (Note 10) F(L) ? ? 2 ? 7 1887 1kg Dudumus Arnal et al. 0 0 1 I (by analogy Frugivorous- F(I) ? ? 1 ? 6 (2014) with insectivorous (note Spaniomys) 11) Acaremyidae Acarechimys 1 1 1 100-1000gr, Frugivorous- F(I) T(F) (note 13) T(F) 1 6 7 Patterson in except for A. insectivorous (note Kraglievich, 1965 minutissimus 12) (smaller) Acaremys Ameghino, 0 1 1 ? I L(F) (note 14) L(F) 1887 Sciamys Ameghino, 1 1 1 I L(F) (note 14) L(F) note 15 ? 1 ? 10 1887 ERETHIZONTOIDEA, Steiromys Ameghino, 1 1 1 14.17 IV bark, coniferous L semiarboreal A(T) 4 4 10 Erethizontidae 1887 leaves and seeds PRIMATES, Homunculus 1 1 1 2.70 II frugivorous, F(L) A 2 3 9 Homunculidae Ameghino, 1891 folivorous Notes: 1: Euhypsodont; grass and leaves depending on their availability (Cassini, 2013); 2, (Cassini, 2013); 3, Diet like a capybaras, grasses but also plants associated to water bodies, seeds, fruits (G Vucetich personal Communication). Notes 4–15 are from M. Arnal, personal communication. 4, Euhypsodont taxa considered to be grazers; 5, Protohypsodont teeth, taller than Sciamys but lower than Phanomys (similar to Neoreomys), (M. Arnal, personal communication); 6, harder than extant Dasyprocta (M. Arnal, personal communication); 7, brachydont; 8, Protohypsodont teeth, so it should already be a higher consumer of fibrousplants; 9, tendency to hypsodonty-fed on plants somewhat more fibrousthan the brachydont taxa; 10. Tendency to hypsodontia-fed on plants somewhat more fibrousthan the brachydont taxa; 11, brachydont; A Like extant Caviocricetus, included insects in its diet; 12, brachydont, could be similar to Prospaniomys because it has the same type of teeth.13, Very large tympanic bullae but with brachydont teeth, digging but not fully fossorial; 14, Protohypsodont tooth crowns like Adelphomys and Stichomys; 15, bullae are small, not fossorial.

References Ameghino, F., 1906. Les formations sedimentaires´ du Cr´etac´e sup´erieur et du Tertiaire de Patagonie avec un parallele entre leurs faunes mammalogiques et celles de l’ancien continent. Anales del Museo Nacional de Buenos Aires, S´erie 3 (15), 1–568. Abello, M.A., Martin, G.M., Cardoso, Y., 2021. Review of the extinct ‘shrew- Andrews, P., Lord, J.M., Nesbit-Evans, E.M., 1979. Patterns of ecological diversity in opossums’(Marsupialia: Caenolestidae), with descriptions of two new genera and fossil and modern mammalian faunas. Biol. J. Linn. Soc. 11, 177–205. three new species from the Early Miocene of southern South America. Zool. J. Linn. Antoine, P.-O., Marivaux, L., Croft, D.A., Billet, G., Ganerød, M., Jaramillo, C., Martin, T., Soc. https://doi.org/10.1093/zoolinnean/zlaa165. Orliac, M.J., Tejada, J., Altamirano, A.J., Duranthon, F., Fanjat, G., Rousse, S., Abello, M.A., Ortiz-Jaureguizar, E., Candela, A.M., 2012. Paleoecology of the Gismondi, R.S., 2012. Middle Eocene rodents from Peruvian Amazonia reveal the Paucituberculata and Microbiotheria (Mammalia, Marsupialia) from the late early pattern and timing of caviomorph origins and biogeography. Proc. Royal Soc. B 279, Miocene of Patagonia. In: Vizcaíno, S.F., Kay, R.F., Bargo, M.S. (Eds.), Early Miocene 1319–1326. Paleobiology in Patagonia: High-Latitude Paleocommunities of the Santa Cruz Arnal, M., 2011. Sistematica,´ filogenia e historia evolutiva de roedores Octodontoidea Formation. Cambridge University Press, Cambridge, UK, pp. 156–172. (Caviomorpha, Hystricognathi) del Oligoceno tardío-Mioceno medio vinculados al Aizen, M.A., Ezcurra, C., 1998. High incidence of plant-animal mutualisms in the woody origen de la familia Octodontidae. Universidad de La Plata. flora of the temperate forest of southern South America: biogeographicical origin Arnal, M., Kramarz, A.G., Vucetich, M.G., Frailey, C.D., Campbell Jr., K.E., 2020. New and present ecological significance. Ecol. Austral 8, 217–236. Palaeogene caviomorphs (Rodentia, Hystricognathi) from Santa Rosa, Peru: Albino, A., Brizuela, S., Vizcaíno, S., 2017. The southernmost fossil record of Squamates. systematics, biochronology, biogeography and early evolutionary trends. Papers in Amphib-Reptilia 38, 15–30. ´ Palaeontology 6, 193–216. Alvarez, A., Arnal, M., 2015. First approach to the paleobiology of extinct Prospaniomys Arnal, M., Kramarz, A.G., Vucetich, M.G., Vieytes, E.C., 2014. A new early Miocene (Rodentia, Hystricognathi, Octodontoidea) through head muscle reconstruction and octodontoid rodent (Hystricognathi, Caviomorpha) from Patagonia (Argentina) and the study of craniomandibular shape variation. J. Mamm. Evol. 22, 519–533. ´ a reassessment of the early evolution of Octodontoidea. J Vert Paleo 34, 397–406. Alvarez, A., P´erez, M.E., 2019. Deep changes in masticatory patterns and masseteric Arnal, M., Perez,´ M.E., Deschamps, C., 2019. Revision of the Miocene caviomorph musculature configurationsaccompanied the eco-morphological evolution of cavioid rodents from the Río Santa Cruz (argentinean Patagonia). Publicacion´ Electronica´ de rodents (Hystricognathi, Caviomorpha). Mamm. Biol. 96, 53–60. la Asociacion´ Paleontologica´ Argentina 19, 193–229. Ameghino, F, 1885. Nuevos restos de mamíferos fosiles´ oligocenos, recogidos por el Arnal, M., Vucetich, M., 2015a. Revision of the fossil rodent Acaremys Ameghino, 1887 profesor Pedro Scalabrini y pertenecientes al Museo Provincial de la Ciudad del (Hystricognathi, Octodontoidea, Acaremyidae) from the Miocene of Patagonia Parana.´ Boletín de la Academia Nacional de Ciencias de Cordoba´ 8, 5–207. (Argentina) and the description of a new acaremyid. Hist. Biol. 27, 42–59. Ameghino, F., 1887. Enumeracion´ sistematica´ de las especies de mamíferos fosiles´ Arnal, M., Vucetich, M.G., 2015b. Main radiation events in Pan-Octodontoidea coleccionados por Carlos Ameghino en los terrenos eocenos de la Patagonia austral. (Rodentia, Caviomorpha). Zool. J. Linn. Soc. 175, 587–606. Boletin del Museo de La Plata 1, 1–26. Bargo, M.S., De Iuliis, G., Toledo, N., 2019. Early Miocene sloths (Xenarthra, Folivora) Ameghino, F, 1889. Contribucion´ al conocimiento de los mamíferos fosiles´ de la from the Río Santa Cruz valley (southern Patagonia, Argentina). Ameghino, 1887 República Argentina. Academia Nacional de Ciencias de Cordoba,´ Argentina 6, revisited. Publicacion´ Electronica´ de la Asociacion´ Paleontologica´ Argentina 19, 1–1027. 102–137. Ameghino, F., 1891. Nuevos restos de mamíferos fosiles´ descubiertos por Carlos Bargo, M.S., Toledo, N., Vizcaíno, S.F., 2012. Paleobiology of the santacrucian sloths and Ameghino en el Eoceno inferior de la Patagonia austral. Especies nuevas, adiciones y anteaters (Xenarthra, Pilosa). In: Vizcaíno, S.F., Kay, R.F., Bargo, M.S. (Eds.), Early correcciones. Revista Argentina de Historia Natural 1, 289–328. Miocene Paleobiology in Patagonia: High-Latitude Paleocommunities of the Santa Ameghino, F., 1894. Enum´eration synoptique des especes` de mammiferes` fossiles des Cruz Formation. Cambridge University Press, Cambridge, UK, pp. 216–242. formations eoc´ enes` de Patagonie. Boletín del Academia Nacional de Ciencias BirdLife International NatureServe, 2014. Bird Species Distribution Maps of the World. Cordoba´ 13, 259–451. BirdLife International, Cambridge, UK. Ameghino, F, 1902. Premi`ere contribution a` la conaissance de la faune mammalogique Blisniuk, P.M., Stern, L.A., Chamberlain, C.P., Idelman, B., Zeitler, P.K., 2005. Climatic des couches a` Colpodon. Boletín del Academia Nacional de Ciencias Cordoba´ 17, and ecologic changes during Miocene surface uplift in the southern patagonian 71–141. Andes. Earth Planet Sci. Lett. 230, 125–142. Ameghino, F, 1904. Nuevas especies de mamíferos Cretaceos´ y Terciarios de la República Boivin, M., Marivaux, L., Orliac, M.J., Pujos, F., Salas-Gismondi, R., Tejada-Lara, J.V., Argentina. Anales de la Sociedad Científica Argentina 56-58, 1–142. Antoine, P.O., 2017. Late middle Eocene caviomorph rodents from Contamana, Peruvian Amazonia. Palaeontol. Electron. 20, 19A.

29 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Bond, M., Tejedor, M.F., Campbell, K.J., Chornogubsky, L., Novo, N., Goin, F.J., 2015. Cuitino,˜ J.I., Vizcaíno, S.F., Bargo, M.S., Aramendía, I., 2019b. Sedimentology and fossil Eocene primates of South America and the African origins of New World monkeys. vertebrates of the Santa Cruz Formation (early Miocene) in Lago Posadas, a sub- Nature 520, 538–541. andean locality of southwestern Patagonia. Andean Geol. 46, 383–420. Bonino, N., Cossíos, D., Menegheti, J., 2010. Dispersal of the European hare, Lepus Currie, D.J., 1991. Energy and large-scale patterns of animal-and plant-species richness. europaeus in South America. J Vert Biol 59, 9–15. Am. Nat. 137, 27–49. Brea, M., Zucol, A.F., Bargo, M.S., Fernicola, J.C., Vizcaíno, S.F., 2017. First Miocene D’Elía, G., Luna, L., Gonzalez,´ E.M., Patterson, B.D., 2006. On the Sigmodontinae record of Akaniaceae in Patagonia (Argentina): a fossil wood from the early Miocene radiation (Rodentia, Cricetidae): an appraisal of the phylogenetic position of Santa Cruz Formation and its palaeobiogeographical implications. Bot. J. Linn. Soc. Rhagomys. Mol. Phylogenet. Evol. 38, 558–564. 183, 334–347. Degrange, F.J., Noriega, J.I., Areta, J.I., 2012. Diversity and paleobiology of the Brea, M., Zucol, A.F., Iglesias, A., 2012. Fossil plant studies from late Early Miocene of Santacrucian birds. In: Vizcaíno, S.F., Kay, R.F., Bargo, M.S. (Eds.), Early Miocene the Santa Cruz Formation: paleoecology and paleoclimatology at the passive margin Paleobiology in Patagonia: High-Latitude Paleocommunities of the Santa Cruz of Patagonia, Argentina. In: Vizcaíno, S.F., Kay, R.F., Bargo, M.S. (Eds.), Early Formation. Cambridge University Press, Cambridge, UK, pp. 138–155. Miocene Paleobiology in Patagonia: High-Latitude Paleocommunities of the Santa Di Bitetti, M.S., Placci, G., Dietz, L.A., 2003. A Biodiversity Vision for the Upper Parana´ Cruz Formation. Cambridge University Press, Cambridge, UK, pp. 104–128. Atlantic Forest Eco-Region: Designing a Biodiversity Conservation Landscape and Brinkman, P., Vizcaíno, S.F., 2014. Clemente Onelli’s sketch map and his first-hand, Setting Priorities for Conservation Action. World Wildlife Fund, Washington, DC, retrospective account of an early fossil-hunting expedition along the Río Santa Cruz, p. 116. southern Patagonia, 1888–1889. Arch. Nat. Hist. 41, 326–337. Diederle, J.M., Noriega, J.I., 2019. New records of birds in the Santa Cruz Formation Brooks, D.M., 2014. Ecological notes on Seriema species in the Paraguayan Chaco, with (early-middle Miocene) at the Río Santa Cruz valley, Patagonia, Argentina. observations on Chunga biology. Ornithology Research 22, 234–237. Publicacion´ Electronica´ de la Asociacion´ Paleontologica´ Argentina 19, 55–61. Bucher, J., P´erez, M.E., Gonzalez´ Ruiz, L.R., D’Elía, L., Bilmes, A., 2020. New middle Doe, B.R., 1983. The past is the key to the future. Geochem. Cosmochim. Acta 47, Miocene (Langhian - Serravallian) vertebrate localities in northwestern Patagonia, 1341–1354. Argentina: a contribution to high latitude South American Land Mammal age Eiten, G., 1972. The cerrado vegetation of Brazil. Bot. Rev. 38, 201–341. sequence. J. S. Am. Earth Sci. https://doi.org/10.1016/j.jsames.2020.103024. Elissamburu, A., 2010. Estudio biomecanico´ y morfofuncional del esqueleto apendicular Burmeister, G, 1879. Description physique de la Republique´ Argentine d’apr`es des de Homalodotherium Flower 1873 (Mammalia, Notoungulata). Ameghiniana 47, observations personnelles et ´etrang`eres. 3 (animaux vertebr´ ´es, 1: Mammif`eres 25–43. vivants et ´eteints). P.E. Coni, Buenos Aires. Elissamburu, A., Vizcaíno, S., 2004. Limb proportions and adaptations in caviomorph Burmeister, G, 1885. Examen´ crítico de los mamíferos y reptiles fosiles´ denominados por rodents (Rodentia: Caviomorpha). J. Zool. 262, 145–159. D. Augusto Gravard y mencionados en su obra precedente. Anales del Museo Emmons, L.H., 1984. Geographic variation in densities and diversities of non-flying Nacional de Historia Natural, Buenos Aires 3, 95–174. mammals in Amazonia. Biotropica, pp. 210–222. Cabrera, A, 1927. Datos para el conocimiento de las dasiuroideos fosiles argentinos. Emmons, L.H., 1999. Of mice and monkeys: primates as predictors of mammal Revista del Museo de La Plata 30, 271–315. community richness. In: Fleagle, J.G., Janson, C., Reed, K.E. (Eds.), Primate Cacciali, P., 2010. Chromatic variation in populations of Xenodon merremi (Serpentes: Communities. Cambridge University Press, Cambridge, UK, pp. 237–267. Dipsadidae) in Paraguay. Acta Herpetol. 5, 107–112. Fernandez,´ M., 2020. Sistematica´ y Filogenia de los Interatheriinae (Notoungulata, Cacciali, P., Bauer, F., Martínez, N., 2015. Herpetofauna de la Reserva Natural del Typotheria, Interatheriidae): implicancias bioestratigraficas´ en la Formacion´ Santa Bosque Mbaracayú, Paraguay. Kempffiana 11, 29–47. Cruz (Mioceno Temprano-Medio) de la Provincia de Santa Cruz, Argentina. Candela, A.M., Rasia, L.L., Perez, M.E., 2012. Paleobiology of Santacrucian caviomorph Unpublished PhD. Dissertation. Universidad Nacional de Lujan,´ Argentina, Lujan,´ rodents: a morphofunctional approach. In: Vizcaíno, S.F., Kay, R.F., Bargo, M.S. Argentina, p. 645. (Eds.), Early Miocene Paleobiology in Patagonia: High-latitude paleocommunities of Fernandez,´ M., Munoz,˜ N.A., 2019. Notoungulata and Astrapotheria (Mammalia, the Santa Cruz Formation. Cambridge University Press, Cambridge, UK, Meridiungulata) of the Santa Cruz Formation (Early-Middle Miocene) along the Río pp. 287–305. Santa Cruz, Argentine Patagonia. Publicacion´ Electronica´ de la Asociacion´ Carrillo, J.D., Faurby, S., Silvestro, D., Zizka, A., Jaramillo, C., Bacon, C.D., Antonelli, A., Paleontologica´ Argentina 19, 138–169. 2020. Disproportionate extinction of South American mammals drove the Fernicola, J.C., Albino, A., 2012. Amphibians and squamate reptiles from the Santa Cruz asymmetry of the Great American Biotic Interchange. Proc. Natn. Acad. Sci 117, Formation (late early Miocene), Santa Cruz province, Argentina: paleoenvironmental 26281–26287. and paleobiological considerations. In: Vizcaíno, S.F., Kay, R.F., Bargo, M.S. (Eds.), Cartes, J.L., 2003. Brief history of conservation in the Interior Atlantic Forest. In: Early Miocene Paleobiology in Patagonia: High-Latitude Paleocommunities of the Galindo-Leal, C., Ibsen de Gusmao,˜ C. (Eds.), The Atlantic Forest of South America. Santa Cruz Formation. Cambridge University Press, Cambridge, UK, pp. 129–137. Island Press, Washington, pp. 269–287. Fernicola, J.C., Bargo, M.S., Vizcaíno, S.F., Kay, R.F., 2019. Early–Middle Miocene Cassini, G.H., 2013. Skull geometric morphometrics and paleoecology of Santacrucian Paleontology in the Río Santa Cruz, Southern Patagonia, Argentina. 130 years since (late early Miocene; Patagonia) native ungulates (Astrapotheria, Litopterna, and Ameghino, 1887. Publicacion´ Electronica´ de la Asociacion´ Paleontologica´ Argentina, Notoungulata). Ameghiniana 50, 193–216. pp. 1–259, 19. Cassini, G.H., Cerdeno,˜ E., Villafane,˜ A.L., Munoz,˜ N.A., 2012. Paleobiology of Fernicola, J.C., Bargo, M.S., Vizcaíno, S.F., Kay, R.F., 2019. Historical background for a santacrucian native ungulates (Meridiungulata: Astrapotheria, Litopterna and revision of the paleontology of the Santa Cruz Formation (Early–Middle Miocene) Notoungulata). In: Vizcaíno, S.F., Kay, R.F., Bargo, M.S. (Eds.), Early Miocene along the Río Santa Cruz, Patagonia, Argentina, 19. Publicacion´ Electronica´ de la Paleobiology in Patagonia: High-Latitude Paleocommunities of the Santa Cruz Asociacion´ Paleontologica´ Argentina, pp. 1–13. Formation. Cambridge University Press, Cambridge, UK, pp. 243–286. Fernicola, J.C., Cuitino,˜ J.I., Vizcaíno, S.F., Bargo, M.S., Kay, R.F., 2014. Fossil localities Chornogubsky, L., Abello, M.A., Barmak, G., 2019. The Metatheria from the Río Santa of the Santa Cruz Formation (early Miocene, Patagonia, Argentina) prospected by Cruz (Santa Cruz Formation, early-middle Miocene, Argentina): history and new Carlos Ameghino in 1887 revisited and the location of the notohippidian. J. S. Am. records. Publicacion´ Electronica´ de la Asociacion´ Paleontologica´ Argentina, Earth Sci. 52, 94–107. pp. 62–84, 19. Fernicola, J.C., Vizcaíno, S.F., 2019. Cingulates (Mammalia, Xenarthra) of the Santa Cruz Climaco das Chagas, K.S., Vassallo, A.I., Becerra, F., Echeverría, A., de Castro Formation (early-middle Miocene) from the Río Santa Cruz, Argentine Patagonia. Loguercio, M.F., Rocha-Barbosa, O., 2019. Locomotion in the fastest rodent, the Publicacion´ Electronica´ de la Asociacion´ Paleontologica´ Argentina 19, 85–101. mara Dolichotis patagonum (Caviomorpha; Caviidae; Dolichotinae). Mastozool. Fernicola, J.C., Vizcaíno, S.F., Bargo, M.S., Kay, R.F., Cuitino,˜ J.I., 2019c. Analysis of the Neotrop. 26. early- middle Miocene mammal associations at the Río Santa Cruz (Patagonia, Constantino, R., 2002. The pest termites of South America: taxonomy, distribution and Argentina). Publicacion´ Electronica´ de la Asociacion´ Paleontologica´ Argentina 19, status. J. Appl. Entomol. 126, 355–365. 239–259. Covert, H.H., Kay, R.F., 1981. Dental microwear and diet: implications for determining Fleagle, J.G., Perkins, M.E., Heizler, M.T., Nash, B., Bown, T.M., Tauber, A.A., Dozo, M. the feeding behaviors of extinct primates, with a comment on the dietary pattern of T., Tejedor, M.F., 2012. Absolute and relative ages of fossil localities in the Santa Sivapithecus. Am. J. Phys. Anthropol. 55, 331–336. Cruz and Pinturas formations. In: Vizcaíno, S.F., Kay, R.F., Bargo, M.S. (Eds.), Early Croft, D.A., 2001. Cenozoic environmental change in South America as indicated by Miocene Paleobiology in Patagonia: High-Latitude Paleocommunities of the Santa mammalian body size distributions (cenograms). Divers. Distrib. 7, 271–287. Cruz Formation. Cambridge University Press, Cambridge, UK, pp. 41–58. Croft, D.A., 2013. What constitutes a fossil mammal community in the early Miocene Fleming, T.H., 1973. Numbers of mammalian species in north and central American Santa Cruz Formation? J. Vertebr. Paleontol. 33, 401–409. forest communities. Ecology 54, 555–563. Cuitino,˜ J.I., Fernicola, J.C., Kohn, M., Naipauer, M., Bargo, M.S., Kay, R.F., Vizcaíno, S. Flower, W.H., 1873. On a newly discovered extinct ungulate mammal from Patagonia, F., 2016. U-Pb geochronology of the Santa Cruz Formation at the Santa Cruz and Homalodontotherium cunninghami. Philosophical Transactions of the Royal Society of Bote rivers (southernmost Patagonia, Argentina) and its implications for the fossil London 164, 173–182. vertebrate communities. J. S. Am. Earth Sci. 70, 198–210. Folch, A., 1992. Order Struthioniformes, ostrich to ducks. In: Del Hoyo, J., Elliott, A., Cuitino,˜ J.I., Fernicola, J.C., Raigemborn, M.S., Krapovickas, V., 2019a. Stratigraphy and Sargatal, J. (Eds.), Handbook of the Birds of the World. Lynx Edicions, Barcelona, depositional environments of the Santa Cruz Formation (early Miocene) along the pp. 75–110, 1. Santa Cruz river, southern Patagonia. Argentina Publicacion´ Electronica´ de la Frenguelli, J., 1953. La flora Fosil´ de la region´ del alto Río Chalía en Santa Cruz Asociacion´ Paleontologica´ Argentina 19, 14–33. (Patagonia). Notas del Museo de La Plata. Seccion´ Paleontología 16. Cuitino,˜ J.I., Raigemborn, M.S., Bargo, M.S., Vizcaíno, S.F., Munoz,˜ N.A., Kohn, M.J., Frey, R.W., Seilacher, A., 1980. Uniformity in marine invertebrate ichnology. Lethaia 13, Kay, R.F., 2021. Insights on the controls on floodplain-dominatedfluvial successions: 183–207. a perspective from the early-middle Miocene Santa Cruz Formation in Río Chalía Gentry, A.H., 1988. Changes in plant community diversity and floristic composition on (Patagonia, Argentina). J. Geol. Soc. jgs2020–2188. environmental and geographical gradients. Ann. Mo. Bot. Gard. 1–34. Gonzaga, L., 1996. Family Cariamidae (seriemas). Handbook of the Birds of the World 3, 234–239.

30 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

Hatton, I.A., McCann, K.S., Fryxell, J.M., Davies, T.J., Smerlak, M., Sinclair, A.R., Munoz,˜ N.A., Toledo, N., Candela, A.M., Vizcaíno, S.F., 2019. Functional morphology of Loreau, M., 2015. The predator-prey power law: biomass scaling across terrestrial the forelimb of Early Miocene caviomorph rodents from Patagonia. Lethaia 52, and aquatic biomes. Science 349. 91–106. Hayes, F.E., Scharf, P.A., 1995. The birds of Parque Nacional Cerro Cora,´ Paraguay. Muzzopappa, P., 2019. Calyptocephalella (Anura, Australobatrachia) remains from the COTINGA 4, 20–24. Río Santa Cruz (early-middle Miocene, Santa Cruz Formation), Santa Cruz province, Hinojosa, L.F., 2005. Cambios climaticos´ y vegetacionales inferidos a partir de Argentina. Publicacion´ Electronica´ de la Asociacion´ Paleontologica´ Argentina 19, paleofloras Cenozoicas del sur de Sudam´erica. Rev. Geol. Chile 32, 95–115. 48–54. Huber, O., 1987. Neotropical savannas: their flora and vegetation. Trends Ecol. Evol. 2, O’Donnell, M.S., Ignizio, D.A., 2012. Bioclimatic predictors for supporting ecological 67–71. applications in the conterminous United States. US Geological Survey Data Series Jablonski, D., Finarelli, J.A., 2009. Congruence of morphologically-defined genera with 691, 4–9. molecular phylogenies. Proc. Natn. Acad. Sci 106, 8262–8266. Ojeda, F.A., Mares, M.A., 1989. A biogeographic analysis of the mammals of Salta Janis, C.M., Fortelius, M., 1988. On the means whereby mammals achieve increased Province, Argentina. Special Publications, Museum of Texas Tech University 27, functional durability of their dentitions, with special reference to limiting factors. 1–66. Biol. Rev. 63, 197–230. Olson, D.M., Dinerstein, E., Wikramanayake, E.D., Burgess, N.D., Powell, G.V., Jos´e De Sousa, E., Oliveira, J.A., Dias, P.A., De Oliveira, T.G., 2005. Update on the Underwood, E.C., D’amico, J.A., Itoua, I., Strand, H.E., Morrison, J.C., 2001. geographical distribution and habitat of the tapiti (Sylvilagus brasiliensis: Terrestrial Ecoregions of the World: a New Map of Life on Earth. Bioscience 51, Lagomorpha, Leporidae) in the Brazilian Amazon. Mammalia 69, 245–250. 933–938. Kay, R.F., 2015a. Biogeography in deep time— what do phylogenetics, geology, and Ortiz-Jaureguizar, E., Cladera, G.A., 2006. Paleoenvironmental evolution of southern paleoclimate tell us about early platyrrhine evolution? Mol. Phylogenet. Evol. 82, South America during the Cenozoic. J. Arid Environ. 66, 498–532. 358–374. Pascual, R., Ortega Hinojosa, E.J., Gondar, D., Tonni, E., 1965. Las edades del Cenozoico Kay, R.F., 2015b. New World monkey origins. Science 347, 1067–1068. mamalífero de la Argentina, con especial atencion´ a aquellas del territorio Kay, R.F., Madden M., R., Vucetich, M.G., Carlini, A.A., Mazzoni, M.M., R´e, G.H., Bonaerense. Anales de la Comision´ de Investigaciones Científicas de la Provincia de Heizler, M., Sandeman, H., 1999. Revised age of the Casamayoran South American Buenos Aires 6, 165–193. Land Mammal ‘Age’- climatic and biotic implications. Proc. Natn. Acad. Sci 96, Owen, R, 1847. Notices of some fossil Mammalia of South America. Reports of the British 13235–13240. Association of Advancements in Science (Southampton) 16, 65–67. Kay, R.F., Madden, R.H., 1997a. Mammals and rainfall: paleoecology of the middle Pascual, R., Odreman Rivas, O., 1971. Evolucion´ de las comunidades de los vertebrados Miocene at La venta (Colombia, south America). J. Hum. Evol. 32, 161–199. del Terciario Argentina, los aspectos paleozoogeograficos´ y paleoclimaticos´ Kay, R.F., Madden, R.H., 1997b. Paleogeography and paleoecology. In: Kay, R.F., relacionados. Ameghiniana 7, 372–412. Madden, R.H., Cifelli, R.L., Flynn, J.J. (Eds.), Mammalian Evolution in the Pascual, R., Ortiz-Jaureguizar, E., 1990. Evolving climates and mammal faunas in Neotropics. Smithsonian Institution Press, Washington, D. C., pp. 520–550 Cenozoic South America. J. Hum. Evol. 19, 23–60. Kay, R.F., Madden, R.H., Van Schaik, C., Higdon, D., 1997. Primate species richness is Patterson, B.D., Stotz, D.F., Solari, S., Fitzpatrick, J.W., Pacheco, V., 1998. Contrasting determined by plant productivity: implications for conservation. Proc. Natn. Acad. patterns of elevational zonation for birds and mammals in the Andes of southeastern Sci 94, 13023–13027. Peru. J. Biogeogr. 25, 593–607. Kay, R.F., Perry, J.M.G., 2019. Primates from the valley of the Río Santa Cruz, Santa Cruz Patton, J.L., Pardinas,˜ U.F., D’Elía, G., 2015. Mammals of South America, vol. 2. province, Argentina. Publicacion´ Electronica´ de la Asociacion´ Paleontologica´ University of Chicago Press, Rodents. Argentina 19, 230–238. Perez,´ L.M., Cuitino,˜ J.I., Varela, A.N., Munoz,˜ N.A., 2019. Diplodon cf. colhuapiensis Kay, R.F., Vizcaíno, S.F., Bargo, M.S., 2012. A review of the paleoenvironment and (Bivalvia, Hyriidae) in the Santa Cruz Formation (early–middle Miocene) at the Río paleoecology of the Miocene Santa Cruz Formation. In: Vizcaíno, S.F., Kay, R.F., Santa Cruz, Patagonia, Argentina. Stratigraphic and paleoenvironmental Bargo, M.S. (Eds.), Early Miocene Paleobiology in Patagonia: High-Latitude considerations. Publicacion´ Electronica´ de la Asociacion´ Paleontologica´ Argentina Paleocommunities of the Santa Cruz Formation. Cambridge University Press, 19, 34–47. Cambridge, UK, pp. 331–364. Perez,´ M.E., 2010. Sistematica,´ ecología y bioestratigrafía de Eocardiidae (Rodentia: Kraglievich, J.L., 1965. Speciation phyletique dans les rongeurs fossiles du genre Hystricognathi: Cavioidea) del Mioceno temprano y medio de Patagonia. Eumysops Ameghino (Echimyidae, Heteropsomyinae). Extrait de Mammalia 29, Universidad Nacional de La Plata. 258–267. Perkins, M.E., Fleagle, J.G., Heizler, M.T., Nash, B., Bown, T.M., Tauber, A.A., Dozo, M. Kramarz, A., 2002. Roedores chinchilloideos (Hystricognathi) de la Formacion´ Pinturas, T., 2012. Tephrochronology of the Miocene Santa Cruz and Pinturas formations, Mioceno temprano-medio de la provincia de Santa Cruz, Argentina. Rev. Mus. Argentina. In: Vizcaíno, S.F., Kay, R.F., Bargo, M.S. (Eds.), Early Miocene Argent. Ciencias Nat. Nueva Ser. 4, 167–180. Paleobiology in Patagonia: High-Latitude Paleocommunities of the Santa Cruz Kramarz, A.G., Vucetich, M.G., Carlini, A.A., Ciancio, M.R., Abello, M.A., Deschamps, C. Formation. Cambridge University Press, Cambridge, UK, pp. 23–40. M., Gelfo, J.N., 2010. A new mammal fauna at the top of the Gran Barranca sequence Prevosti, F.J., Forasiepi, A.M., Ercoli, M.D., Turazzini, G.F., 2012. Paleoecology of the and its biochronological significance. In: Madden, R.H., Carlini, A.A., Vucetich, M. mammalian carnivores (Metatheria, Sparassodonta) of the Santa Cruz Formation G., Kay, R.F. (Eds.), The Paleontology of Gran Barranca. Cambridge University Press, (late early Miocene). In: Vizcaíno, S.F., Kay, R.F., Bargo, M.S. (Eds.), Early Miocene Cambridge, UK, pp. 264–277. Paleobiology in Patagonia: High-Latitude Paleocommunities of the Santa Cruz Kramarz, A.G., Bellosi, E.S., 2005. Hystricognath rodents from the Pinturas Formation, Formation. Cambridge University Press, Cambridge, UK, pp. 173–193. early–middle Miocene of Patagonia, biostratigraphic and paleoenvironmental QGIS.org, 2021. QGIS Geographic Information System. QGIS Association. http://www. implications. J. S. Am. Earth Sci. 18, 199–212. qgis.org. Kramarz, A.G., Bond, M., Arnal, M., 2015. Systematic description of three new mammals Raigemborn, M.S., Krapovickas, V., Beilinson, E., Peral, L.E.G., Zucol, A.F., Zapata, L., (Notoungulata and Rodentia) from the early Miocene Cerro Bandera Formation, Kay, R.F., Bargo, M.S., Vizcaíno, S.F., Sial, A.N., 2018a. Multiproxy studies of Early northern Patagonia, Argentina. Ameghiniana 52, 585–597. Miocene pedogenic calcretes in the Santa Cruz Formation of southern Patagonia, Lavocat, R., 1976. Rongeurs caviomorphies de l’Oligoc`ene de Bolivie. II. Rongeurs du Argentina indicate the existence of a temperate warm vegetation adapted to a Bassin D´es´edien de Salla-Luribay. Palaeovertebrata 7, 15–90. fluctuating water table. Palaeogeogr. Palaeoclimatol. Palaeoecol. 500, 1–23. Legendre, S., 1986. Analysis of mammalian communities from the late Eocene and Raigemborn, M.S., Krapovickas, V., Zucol, A.F., Zapata, L., Beilinson, E., Toledo, N., Oligocene of southern France. Palaeovertebrata 16, 191–212. Perry, J., Lizzoli, S., Martegani, L., Tineo, D.E., 2018b. Paleosols and related soil- Liow, L.H., Finarelli, J.A., 2014. A dynamic global equilibrium in carnivoran biota of the early Miocene Santa Cruz Formation (Austral-Magallanes basin, diversification over 20 million years. Proc. Royal Soc. B 281, 20132312. Argentina). Lat. Am. J. Sedimentol. Basin Anal. 25, 117–148. Maas, M.C., Krause, D.W., 1994. Mammalian turnover and community structure in the Raigemborn, M.S., Matheos, S.D., Krapovickas, V., Vizcaíno, S.F., Bargo, M.S., Kay, R.F., Paleocene of North America. Hist. Biol. 8, 91–128. Fernicola, J.C., Zapata, L., 2015. Paleoenvironmental reconstruction of the coastal Mares, M.A., Braun, J.K., Gettinger, D., 1989. Observations on the distribution and Monte L´eon and Santa Cruz formations (early Miocene) at Rincon´ del Buque, ecology of the mammals of the cerrado grasslands of central Brazil. Ann. Carnegie southern Patagonia: a revisited locality. J. S. Am. Earth Sci. 60, 31–55. Mus. 58, 1–60. Rasia, L.L., Candela, A.M., 2019. Upper molar morphology, homologies and evolutionary Mares, M.A., Ojeda, R.A., 1982. Patterns of diversity and adaptation in South American patterns of chinchilloid rodents (Mammalia, Caviomorpha). J. Anat. 234, 50–65. hystricognath rodents. In: Mares, M.A., Genoways, H.H. (Eds.), Mammalian Biology Redford, 1984Redford, K.H., 1984. The termitaria of Cornitermes cumulans (Isoptera, in South America. University of Pittsburg, Pittsburgh, pp. 393–432. Termitidae) and their role in determining a potential keystone species. Biotropica 16, Matheos, S., Raigemborn, M.S., 2012. Sedimentology and paleoenvironment of the Santa 112–119. Cruz Formation. In: Vizcaíno, S.F., Kay, R.F., Bargo, M.S. (Eds.), Early Miocene Redford, K.H., Peters, G., 1986. Notes on the biology and song of the red-legged seriema Paleobiology in Patagonia: High-Latitude Paleocommunities of the Santa Cruz (Cariama cristata). J. Field Ornithol. 261–269. Formation. Cambridge University Press, Cambridge, UK, pp. 59–82. Reed, K.E., 1998. Using large mammal communities to examine ecological and Mayr, E., 1950. Taxonomic categories in fossil hominids. Cold Spring Harbor Laboratory taxonomic structure and predict vegetation in extant and extinct assemblages. Press, pp. 109–118. Paleobiology 24, 384–408. McCain, C.M., King, S.R., Szewczyk, T., Beck, J., 2018. Small mammal species richness is Richter, H., Cumming, G., 2008. First application of satellite telemetry to track African directly linked to regional productivity, but decoupled from food resources, straw-coloured fruit bat migration. J. Zool. 275, 172–176. abundance, or habitat complexity. J. Biogeogr. 45, 2533–2545. Robinson, J.R., Rowan, J., Campisano, C.J., Wynn, J.G., Reed, K.E., 2017. Late Pliocene Mercerat, A., 1891. Sobre la presencia de restos de monos en el Eoceno de Patagonia. environmental change during the transition from Australopithecus to Homo. Nature Rev. Mus. La Plata 2, 73–74. Ecol Evol 1, 1–7. Mereles, M.F., 2005. Una aproximacion´ al conocimiento de las formaciones vegetales del Rodríguez-Gomez,´ G., Cassini, G.H., Palmqvist, P., Bargo, M.S., Toledo, N., Martín- Chaco boreal, Paraguay. Rojasiana 6, 5–48. Gonzalez,´ J.A., Munoz,˜ N.A., Kay, R.F., Vizcaíno, S.F., 2020. Testing the hypothesis of an impoverished predator guild in the early Miocene ecosystems of Patagonia: an

31 R.F. Kay et al. Journal of South American Earth Sciences 109 (2021) 103296

analysis of meat availability and competition intensity among carnivores. Vizcaíno, S., Bargo, M., Kay, R., Farina,˜ R., DiGiacomo, M., Perry, J., Prevosti, F., Palaeogeogr. Palaeoclimatol. Palaeoecol. https://doi.org/10.1016/j. Toledo, N., Cassini, G., Fernicola, J., 2010. A baseline paleoecological study for the palaeo.2020.109805, 109805. Santa Cruz Formation (late-early Miocene) at the Atlantic coast of Patagonia, Rougier, G.W., Wible, J.R., Beck, R.M., Apesteguía, S., 2012. The Miocene mammal Argentina. Palaeogeogr. Palaeoclimatol. Palaeoecol. 292, 507–519. Necrolestes demonstrates the survival of a Mesozoic nontherian lineage into the late Vizcaíno, S., Kay, R.F., Bargo, M., 2012a. Early Miocene Paleobiology in Patagonia: High- Cenozoic of South America. Proc. Natn. Acad. Sci 109, 20053–20058. Latitude Paleocommunities of the Santa Cruz Formation. Cambridge University Rowan, J., Beaudrot, L., Franklin, J., Reed, K.E., Smail, I.E., Zamora, A., Kamilar, J.M., Press, Cambridge, UK, p. 370. 2020. Geographically divergent evolutionary and ecological legacies shape mammal Vizcaíno, S.F., 2014. Interview on paleobiology. In: Sanchez-Villagra,´ M.R., Seilac biodiversity in the global tropics and subtropics. Proc. Natn. Acad. Sci 117, MacLeod, N. (Eds.), Issues in Palaeontology: A Global View. Interviews and Essays. 1559–1565. Scidinge Hall Verlag, Zürich. Rowan, J., Kamilar, J.M., Beaudrot, L., Reed, K.E., 2016. Strong influence of Vizcaíno, S.F., Bargo, M.S., Kay, R.F., Milne, N., 2006. The armadillos (Mammalia, palaeoclimate on the structure of modern African mammal communities. Proc. R. Xenarthra, Dasypodidae) of the Santa Cruz formation (early-middle Miocene). An Soc. Lond. [Biol.] 283, 20161207. approach to their paleobiology. Palaeogeogr. Palaeoclimatol. Palaeoecol. 237, Scott, W.B., 1903-1904. Mammalia of the Santa Cruz Beds. Parts 1-3, Edentata, Reports 255–269. of the Princeton University Expeditions to Patagonia, vol. 5. Princeton University, Vizcaíno, S.F., Bargo, M.S., Kay, R.F., Raigemborn, M.S., 2021. The Record of the Princeton, pp. 1–364, 1896-1899. Typothere Pachyrukhos (Mammalia, Notoungulata) and the Chinchillid Seiffert, E.R., Tejedor, M.F., Fleagle, J.G., Novo, N.M., Cornejo, F.M., Bond, M., de Prolagostomus (Mammalia, Rodentia) in the Santa Cruz Formation (Early-middle Vries, D., Campbell, K.E., 2020. A parapithecid stem anthropoid of African origin in Miocene) South to the Río Coyle, Patagonia, Argentina. Publicacion´ Electronica´ de the Paleogene of South America. Science 368, 194–197. La Asociacion´ Paleontologica´ Argentina. Sick, H., 1997. Ornitologia Brasileira. Nova Fronteira, Rio de Janeiro. Vizcaíno, S.F., Bargo, M.S., Toledo, N., 2017a. Revaluacion´ crítica de los mamíferos Simpson, G.G., 1943. Mammals and the nature of continents. Am. J. Sci. 241, 1–31. actuales como indicadores paleoambientales. Ejemplos de la mastofauna Simpson, G.G., 1980. Splendid Isolation: the Curious History of South American neotropical. XXX Jornadas Argentinas de Mastozoología, p. 41. Abstract book. Mammals. Yale Univ. Press, New Haven. Vizcaíno, S.F., Brinkman, P.D., Kay, R.F., 2016. On the objectives and results of the Sinclair, W.J., 1908. The Santa Cruz Typotheria. Proc. Am. Phil. Soc. 47, 64–78. Handel T. Martin paleontological expedition (1903-04) to the Santa Cruz Formation Smith-Ramirez, C., Armesto, J.J., 1994. Flowering and fruiting patterns in the temperate in southern Patagonia. Rev. Mus. La Plata 1, 316–333. rainforest of Chiloe, Chile–ecologies and climatic constraints. J. Ecol. 353–365. Vizcaíno, S.F., Cassini, G.H., Toledo, N., Bargo, M.S., 2012b. On the evolution of large Spradley, J.P., Glazer, B., Kay, R.F., 2019. Mammalian faunas, ecological indices, and size in mammalian herbivores of Cenozoic faunas of Southern South America. In: machine-learning regression for the purpose of paleoenvironment reconstruction in Patterson, B., Costa, L. (Eds.), Bones, Clones and Biomes: an 80-million Year History the Miocene of South America. Palaeogeogr. Palaeoclimatol. Palaeoecol. 518, of Recent Neotropical Mammals. University of Chicago Press, Chicago, pp. 76–101. 155–171. Vizcaíno, S.F., Cassini, G.H., Fernicola, J.C., Bargo, M.S., 2011. Evaluating habitats and Spradley, J.P., Williams, B.A., Kay, R.F., 2015. Environmental variables affecting primate feeding habits through ecomorphological features in glyptodonts (Mammalia, species richness in the Neotropics. Am. J. Phys. Anthropol. 56, 294. Xenarthra). Ameghiniana 48 (315), 305–319. Stallings, J.R., 1985. Distribution and status of primates in Paraguay. Primate Conserv. 6, Vizcaíno, S.F., De Iuliis, G., Brinkman, P., Kay, R.F., Brinkman, D.D., 2017b. On an album 51–58. of photographs recording fossils in the “Old Collections" of the Museo de La Plata Strait, S.G., Fleagle, J.G., Bown, T.M., Dumont, E.R., 1990. Diversity in body size and and Ameghino’s private collection at the beginning of the XXth Century. Publicacion´ dietary habits of fossil caenolestid marsupials from the Miocene of Argentina. Electronica´ de la Asociacion´ Paleontologica´ Argentina 17, 14–23. J. Vertebr. Paleontol. 10, 44A. Vizcaíno, S.F., Farina,˜ R.A., 1997. Diet and locomotion of the armadillo Peltephilus: a new Tauber, A.A., 1997a. Bioestratigtrafía de la Formacion´ Santa Cruz (Mioceno inferior) en view. Lethaia 30, 79–86. el extremo sudeste de la Patagonia. Ameghiniana 34, 413–426. Vizcaíno, S.F., Farina,˜ R.A., Bargo, M.S., De Iullis, G., 2004. Functional and phylogenetic Tauber, A.A., 1997b. Paleoecología de la Formacion´ Santa Cruz (Mioceno inferior) en el assessment of the masticatory adaptations in Cingulata (Mammalia, Xenarthra). extremo sudeste de la Patagonia. Ameghiniana 34, 517–529. Ameghiniana 41, 651–664. Tejedor, M.F., Tauber, A.A., Rosenberger, A.L., Swisher III, C.C., Palacios, M.E., 2006. Vizcaíno, S.F., Fernicola, J.C., Bargo, M.S., 2012c. Paleobiology of santacrucian New primate genus from the Miocene of Argentina. Proc. Natn. Acad. Sci 103, glyptodonts and armadillos (xenarthra, Cingulata). In: Vizcaíno, S.F, Kay, R.F, 5437–5441. Bargo, M.S (Eds.), Early Miocene Paleobiology in Patagonia: High-f tlatitude Toledo, N., 2016. Paleobiological integration of Santacrucian sloths (early Miocene of paleocommunities ohe Santa Cruz Formation. Cambridge University Press, Patagonia). Ameghiniana 53, 100–141. Cambridge, UK, pp. 194–215. Toledo, N., Cassini, G.H., Vizcaíno, S.F., Bargo, M.S., 2014. Mass estimation of Vizcaíno, S.F., Kay, R.F., Bargo, M.S., 2012d. Background for a paleoecological study of Santacrucian sloths from the early Miocene Santa Cruz Formation of Patagonia, the Santa Cruz Formation (late early Miocene) on the Atlantic coast of Patagonia. In: Argentina. Acta Palaeontol. Pol. 59, 267–280. Vizcaíno, S.F., Kay, R.F., Bargo, M.S. (Eds.), Early Miocene Paleobiology in Trayler, R.B., Kohn, M.J., Bargo, M.S., Cuitino,˜ J.I., Kay, R.F., Stromberg, C.A.E., Patagonia: High-Latitude Paleocommunities of the Santa Cruz Formation. Vizcaíno, S.F., 2020. Patagonian aridification at the onset of the mid-Miocene Cambridge University Press, Cambridge, UK, pp. 1–22. Climate Optimum. Paleoceanography and Paleoclimatology, 35; Special Issue: “The Vizcaíno, S.F., Toledo, N., Bargo, M.S., 2018. Advantages and limitations in the use of Miocene: The Future of the Past ” e2020PA003956. extant xenarthrans (Mammalia) as morphological models for paleobiological Trayler, R.B., Schmitz, M.D., Cuitino,˜ J.I., Kohn, M.J., Bargo, M.S., Kay, R.F., reconstruction. J. Mamm. Evol. 25, 495–505. Stromberg, C.A.E., Vizcaíno, S.F., 2020. An improved approach to age-modeling in Vucetich, M., Arnal, M., Deschamps, C., Perez,´ M., Vieytes, E., 2015. A brief history of deep time: implications for the Santa Cruz Formation, Argentina. Geol. Soc. Am. caviomorph rodents as told by the fossil record. In: Vassallo, A.I., Antenucci, D. Bull. 132, 233–244. (Eds.), Biology of Caviomorph Rodents: Diversity and Evolution. SAREM Series A, Tutin, C.E., Fernandez, M., 1993. Relationships between minimum temperature and fruit Mendoza, pp. 11–62. production in some tropical forest trees in Gabon. J. Trop. Ecol. 241–248. Vucetich, M.G., 1984. Los roedores de la Edad Friasense (Mioceno medio) de Patagonia. Vermillion, W.A., Polly, P.D., Head, J.J., Eronen, J.T., Lawing, A.M., 2018. Ecometrics: a Rev. Museo de La Plata (nueva s´erie), VIII. Paleontología 50, 47–126. trait-based approach to paleoclimate and paleoenvironmental reconstruction. In: Wible, J.R., Rougier, G.W., 2017. Craniomandibular anatomy of the subterranean Croft, D., Su, D.F, Simpson, S.W. (Eds.), Methods in Paleoecology: Reconstructing meridiolestidan Necrolestes patagonensis Ameghino, 1891 (Mammalia, Cladotheria) Cenozoic Terrestrial Environments and Ecological Communities. Springer from the early Miocene of Patagonia. Ann. Carnegie Mus. 84, 183–252. International Piblishing, pp. 373–394. Wood, B., Collard, M., 1999. The human genus. Science 284, 65–71. Vizcaíno, S., Bargo, M., Fernicola, J., 2013. Expediciones paleontologicas´ durante los Zapata, L., 2018. Estudio paleoambiental de la formacion´ Santa Cruz (Mioceno inferio- Siglos XIX y XX A la Formacion´ Santa Cruz (Mioceno Inferior, Patagonia) y destino medio) entre los ríos Coyle y Gallegos, Patagonia Austral. Universidad Nacional de de los fosiles.´ In: Alonso, R.N. (Ed.), III Congreso Argentino de Historia de la La Plata, Argentina. Geología, Actas. Mundo Grafico´ Salta Editorial, Salta, pp. 231–246.

32