Introduction
Important aspects of the feeding strategies and dietary patterns of Quaternary megafauna in the Brazilian Intertropical Region (BIR; Figure 1; sensu Cartelle, Reference Cartelle, Eisenberg and Redford1999) have been investigated based on carbon and oxygen isotopic data (e.g., Sanchez et al., Reference Sánchez, Prado and Alberdi2004; MacFadden, Reference MacFadden2005; Dantas et al., Reference Dantas, Dutra, Cherkinsky, Fortier, Kamino, Cozzuol, Ribeiro and Vieira2013, Reference Dantas, Cherkinsky, Bocherens, Drefahl, Bernardes and França2017; França et al., Reference França, Dantas, Bocchiglieri, Cherkinsky, Ribeiro and Bocherens2014; Pansani et al., Reference Pansani, Muniz, Cherkinsky, Pacheco and Dantas2019; Asevedo et al., Reference Asevedo, Ranzi, Kalliola, Pärsinen, Ruokolainen, Cozzuol and Nascimento2021; Faria et al., Reference Faria, Carvalho and Araújo-Júnior2021). Three dietary guilds have been identified: browsers, mixed feeders, and grazers (Sanchez et al., Reference Sánchez, Prado and Alberdi2004; MacFadden, Reference MacFadden2005; Dantas et al., Reference Dantas, Dutra, Cherkinsky, Fortier, Kamino, Cozzuol, Ribeiro and Vieira2013, Reference Dantas, Cherkinsky, Bocherens, Drefahl, Bernardes and França2017). In general, the species Eremotherium laurillardi, Notiomastodon platensis, and Toxodon platensis have been interpreted as generalist feeders (Dantas et al., Reference Dantas, Dutra, Cherkinsky, Fortier, Kamino, Cozzuol, Ribeiro and Vieira2013, Reference Dantas, Cherkinsky, Bocherens, Drefahl, Bernardes and França2017; Faria et al., Reference Faria, Carvalho and Araújo-Júnior2021).
Location map of the Brazilian Intertropical Region in South America, showing the location of Itapipoca (Ceará State) and Miranda River (Mato Grosso do Sul State). Adapted from Pansani et al. (Reference Pansani, Muniz, Cherkinsky, Pacheco and Dantas2019).
Figure 1 Long description
The map shows the Brazilian Intertropical Region (BIR) in South America, highlighting the locations of Itapipoca in Ceara State and the Miranda River in Mato Grosso do Sul State. The BIR is shaded, covering parts of several states including AM, PA, RO, MT, MS, GO, DF, MG, ES, RJ, SP, PR, SC, RS, BA, SE, AL, PE, PB, RN, CE, PI and MA. A red star marks the Miranda River and a yellow star marks the Jirau tank deposit in Itapipoca. The map includes a scale bar and compass for orientation. Insets show Brazil′s location in South America and the BIR within Brazil.
Over the last decade, several important studies have examined the feeding habits of megafauna species from the Late Pleistocene and Early Holocene in the BIR, based on isotopic paleoecology using stable carbon and oxygen isotopes (e.g., Dantas et al., Reference Dantas, Dutra, Cherkinsky, Fortier, Kamino, Cozzuol, Ribeiro and Vieira2013, Reference Dantas, Cherkinsky, Bocherens, Drefahl, Bernardes and França2017; França et al., Reference França, Dantas, Bocchiglieri, Cherkinsky, Ribeiro and Bocherens2014; Pansani et al., Reference Pansani, Muniz, Cherkinsky, Pacheco and Dantas2019; Faria et al., Reference Faria, Carvalho and Araújo-Júnior2021). In general, these studies indicate that megafauna species did not exhibit significant dietary differences over time, suggesting that, despite changes in vegetation physiognomy during the late Quaternary, their feeding habits remained stable. Although the vegetation composition of the BIR did not change drastically from the last glacial maximum (Late Pleistocene) to the onset of the current interglacial (Early Holocene), the persistence of open environments favorable to megafauna species during the Early Holocene provides evidence that the BIR served as an environmental refuge for these taxa during this period.
Here, we present isotopic paleoecological data based on stable carbon isotope analyses of Holocene megafauna from the BIR (Faria et al., Reference Faria, Carvalho, Araújo-Júnior, Ximenes and Facincani2025) and compare them with isotopic records from other regions of the Americas. This study provides the first δ13C isotopic analysis of Mid–Late Holocene megafaunal fossils from the BIR, encompassing the taxa E. laurillardi, N. platensis, S. populator, X. bahiense, and P. major, recovered from the Jirau Paleontological Site (Itapipoca, Ceará State, Brazil) and the Miranda River (Mato Grosso do Sul State, Brazil). The results aim to demonstrate that the BIR functioned as an environmental refuge for mammalian megafauna during the Holocene.
Material and methods
Materials
The carbon isotopic composition (δ13C) was analyzed for eight fossil tooth samples from the Jirau Paleontological Site (Itapipoca, Ceará State, Brazil), comprising two specimens of E. laurillardi, and one specimen each of N. platensis, S. populator, T. platensis, X. bahiense, and P. major. An additional E. laurillardi sample from the Miranda River (Mato Grosso do Sul State, Brazil) was also included. All analyses were performed at the Laboratory of Radioecology and Environmental Changes (LARA-UFF), Universidade Federal Fluminense. Geochronological data and their implications are presented in Faria et al. (Reference Faria, Carvalho, Araújo-Júnior, Ximenes and Facincani2025).
Isotopic paleoecology (δ13C)
The fossils were examined under a microscope and sub-sampled to assess possible diagenetic alterations. Subsequently, chemical treatments with sodium hypochlorite and hydrochloric acid were applied to remove potential contaminants (Oliveira et al., Reference Oliveira, Macario, Carvalho, Moreira, Alves, Chanca and Fernandes2021a; Oliveira et al., Reference Oliveira, Macario, Silva, Pereira, Chanca, Alves and Pessenda2021b). To convert the carbon present in the hydroxyapatite of the analyzed tooth samples into CO₂, the material was hydrolyzed by creating a vacuum in a septum-sealed tube, followed by the injection of 85% phosphoric acid.
Spectrometric analyses were performed on the CO₂ released from carbonate digestion at 72°C with orthophosphoric acid, using an online configuration with the GasBench II preparation system. The results were calibrated to the VPDB scale using the reference materials NBS-19, IAEA–CO–8, and IAEA–CO–9. The reported standard deviation (1σ) corresponds to the internal precision of 8–10 replicate readings from the same preparation. The δ13C data were used to infer the diets of the analyzed fossil mammals.
The analyzed taxa exhibit different trophic enrichment factors, varying between 13.36‰ and 14.35‰ (Tejada-Lara et al., Reference Tejada-Lara, MacFadden, Bermudez, Rojas, Salas-Gismondi and Flynn2018), indicating that body mass influences physiological carbon isotope values. As the analyzed species are extinct and lack living analogs – and given the large variation in body mass estimates reported in the literature – we adopted an enrichment factor of 14‰ for medium- and large-sized herbivores, following Cerling and Harris (Reference Cerling and Harris1999).
Dietary categories were defined according to δ13C values as follows: dense-forest browsers (−22‰ to −16‰), deciduous/semideciduous forest browsers (−16‰ to −11‰), wooded-savannah browsers (−11‰ to −8‰), high-latitude and high-altitude grazers feeding on C₃ plants (−11‰ to −8‰), grazers feeding on C₄ plants (−3‰ to +5‰), and mixed feeders consuming both C₃ and C₄ plants (−8‰ to −3‰) (Domingo et al., Reference Domingo, Prado and Alberdi2012).
The relative proportion of C₃ and C₄ plants consumed was estimated using the equation proposed by Koch et al. (Reference Koch, Hoppe and Webb1998):
${{\delta }^{13}}{{\text{C}}_{{\text{enamel}}}} = \left[ {{{\delta }^{13}}{{\text{C}}_{{\text{C}}4}} \times X\left] + \right[{{\delta }^{13}}{{\text{C}}_{{\text{C}}3}} \times \left( {1{\text{--}}X} \right)} \right]{\text{ }}$
where X represents the proportion of C₄ vegetation in the diet.
For carnivorous taxa, δ13C values below −16‰ indicate a diet composed primarily of prey feeding on C₃ plants, whereas values above −6‰ reflect the consumption of prey that fed predominantly on C₄ plants (Pushkina et al., Reference Pushkina, Bocherens, Chaimanee and Jaeger2010).
Results
The stable carbon isotope values (Table 1) from Itapipoca, Ceará State, range from δ13C = −16.93‰ (P. major) to −9.36‰ (E. laurillardi). The analyzed herbivorous taxa exhibit browsing diets within different vegetation settings, including deciduous to semideciduous forests and wooded savannahs (Figure 2A), following the classification of Domingo et al. (Reference Domingo, Prado and Alberdi2012). S. populator, the only carnivorous taxon analyzed in this study (Table 1), shows isotopic values consistent with a diet composed of prey that consumed both C₃ and C₄ plants, indicating that it inhabited wooded savannah environments. At the Miranda River site, E. laurillardi displays a similar isotopic value (δ13C = −9.36‰), also suggesting a browsing diet within a wooded savannah environment (Figure 2B).
Scatter-plot of δ13C values of megafauna species from Itapipoca and Miranda River.
Figure 2 Long description
Two scatter plots labeled A and B. Both share the same horizontal axis labeled delta 13 C VPDB, ranging from minus 20 to plus 3. Vertical position of points carries no numerical meaning and serves only to separate symbols. Vertical dashed lines divide each plot into four vegetation bands labeled closed canopy forest, deciduous and semideciduous forest, wooded savannah and savannah, from left to right. Background illustrations of trees and grass mark each vegetation zone but carry no data value. Image A is titled Itapipoca. Multiple data points appear across the deciduous and semideciduous forest band and the wooded savannah band. Points fall approximately between minus 16 and minus 9. One point appears near minus 16 in the deciduous and semideciduous forest zone. Several points cluster between minus 14 and minus 12 in the same zone. Additional points appear near minus 11 and minus 10, also within the deciduous and semideciduous forest band. One point falls near minus 9 within the wooded savannah band. A star symbol appears near minus 9 in the wooded savannah zone. One square symbol appears near minus 10 at the boundary of wooded savannah. Image B is titled Miranda River. A single filled circle appears near minus 9 in the wooded savannah band. No other data points are visible in this panel. Species are identified by symbol: filled circle for Eremotherium laurillardi, filled hexagon for Notiomastodon platensis, filled diamond for Toxodon platensis, filled square for Xenorhinotherium bahiense, filled triangle for Paleolama major and star for Smilodon populator.
Ages obtained for the megafauna of Itapipoca and the respective isotope values and the proportion of consumed C3 and C4 plants. Radiocarbon ages were obtained by Faria et al. (Reference Faria, Carvalho, Araújo-Júnior, Ximenes and Facincani2025).
Table 1 Long description
Columns list Sample, Taxa, Locality, Radiocarbon Age in BP, Calibrated age in cal yr BP, delta 13 C in per mil, X percent C4, and X percent C3. Seven samples are from Jirau, CE and one sample is from Miranda, MS. Radiocarbon Age ranges from 3,492 plus or minus 165 BP for PDR 09 to 7,940 plus or minus 502 BP for PDR 07, with calibrated age ranges spanning 3353 to 4231 cal yr BP up to 7836 to 10159 cal yr BP. Delta 13 C values range from minus 16.93 for PDR 09 to minus 9.36 for PDR 11. X percent C4 and X percent C3 are reported for all rows except PDR 03, where both fields are blank. Values for X percent C4 range from 0 to 0.29, and X percent C3 range from 0.71 to 1.00, including 1.00 for PDR 05, PDR 06, and PDR 09.
Discussion
Isotopic paleoecology (δ13C)
The isotopic data presented in Table 2 for S. populator from different localities in South America indicate that it preyed upon animals feeding on both C₃ and C₄ plants, suggesting that it inhabited wooded savannah environments in the BIR and the Pampas region of Argentina. This pattern reflects a broad dietary spectrum and a wide range of prey resources. P. major exhibits isotopic values (Table 2) indicating that this species occupied wooded savannah habitats as well as the ecotone margins of deciduous to semideciduous tropical forests within the BIR.
The limited isotopic data available for X. bahiense in the BIR (Table 2) suggest a browsing diet within deciduous to semideciduous tropical forest environments. Its sister taxon, M. patachonica, which inhabited southern Brazil and north-central Argentina within the Pampas biome under subtropical climatic conditions, exhibited a browsing diet in wooded steppe environments of southern South America (Table 2). Therefore, M. patachonica appears to have feeding strategies very similar to those of X. bahiense, which occupied warmer and drier regions of South America (BIR), suggesting that these taxa may represent vicariant species adapted to different climatic regimes.
E. laurillardi, N. platensis, and T. platensis exhibit generalist feeding habits both in the BIR and across the Americas (Table 2), with browsing, mixed, and grazing diets recorded from different localities during the Late Pleistocene and Early Holocene. This pattern indicates that these taxa employed ecological plasticity as an adaptive strategy, allowing them to inhabit a variety of environmental settings, including seasonal dry tropical forests, wooded savannahs, open savannas, prairies, and steppes – conditions that supported their broad Pan-American distribution. Thus, the ecological plasticity of these taxa, as evidenced by the isotopic values presented in Table 2, accounts for their extensive geographic distribution across the Americas documented in the fossil record.
The correlation between the isotopic and geochronological data presented in Table 2 is illustrated in Figure 3. These results indicate that E. laurillardi, N. platensis, and T. platensis from the Pleistocene–Holocene transition in the BIR exhibited mixed diets rather than diets composed exclusively of grasses. This shift in feeding strategy likely represents a response to the increase in temperature and humidity at the onset of the Holocene, which promoted the transformation of extensive savannah areas into wooded savannahs and deciduous to semideciduous forests. We attribute this dietary change to ecological pressures associated with the climatic and environmental transitions that marked the Pleistocene–Holocene boundary.
Graph showing the relationship between isotopic data and the respective ages of the Late Pleistocene to Early Holocene megafauna of the BIR shown in Table 2.
Figure 3 Long description
The horizontal axis label reads, delta 13C, percent. The horizontal axis tick labels run from negative 20 to 3 in increments of 1. The vertical axis label reads, Age, years. The vertical axis tick labels are 3000, 5000, 10000, 15000, 20000 and 25000. Text along the top of the plot reads, closed canopy forest, deciduous or semideciduous forest, wooded savannah and savannah. A horizontal line is drawn at 3000 on the vertical axis. Most plotted points lie above 5000 years and below 20000 years, with many points concentrated between about negative 9 and 0 on the horizontal axis and between about 9000 and 16000 on the vertical axis. A smaller set of points lies between about negative 15 and negative 10 on the horizontal axis, with ages mostly between about 5000 and 12000 years. Points are sparse at the most negative horizontal axis values near negative 20 and at the most positive values near 3. The highest ages are near 25000 years, with points around horizontal axis values near negative 3 to negative 1. The lowest ages are just above 3000 years, including a point near horizontal axis value around negative 17. Across the full set of points, older ages appear at multiple horizontal axis values and the plotted points do not form a single straight rising or falling pattern. A legend at the lower left lists five categories with distinct marker shapes: Smilodon populator, Notiomastodon platensis, Toxodon platensis, Xenorhinotherium bahiense and Palaeolama major.
Therefore, the ecological plasticity demonstrated by E. laurillardi, N. platensis, and T. platensis enabled their persistence in the BIR during the Holocene, as supported by the geochronological data reported by Faria et al. (Reference Faria, Carvalho, Araújo-Júnior, Ximenes and Facincani2025). The localities of Itapipoca and the Miranda River valley likely served as environmental refuges for these taxa during this period.
The ecological plasticity of the analyzed taxa, as shown in Figure 3, spans the interval between 25,000 and 3000 years BP, indicating the persistence of environments favorable to the survival of megafauna species until the onset of the Late Holocene in the BIR. Accordingly, the chronological and stable carbon isotope data presented in this study demonstrate that the Itapipoca and Rio Miranda regions functioned as environmental refugies for South American megafauna during the Holocene.
Itapipoca and Miranda River as environmental refuges
The environmental refuge theory is grounded in the understanding that Quaternary climatic and environmental changes influenced the distribution – through both expansion and contraction – of tropical forest areas relative to open vegetation formations such as the Caatinga and Cerrado in Brazil (Ab’Saber, Reference Ab’Saber1992). The cenral premise is that each “environmental refuge” once represented a broader mosaic of environmental and landscape conditions that became spatially restricted as a result of Quaternary climatic oscillations (Ab’Saber, Reference Ab’Saber1992). From this perspective, we propose that certain areas within the BIR acted as environmental refuges for megafauna taxa during the Holocene.
Paleodistribution maps of the main vegetation formations of South America. (A) STDFs on narrow and broad definition for the LGM, Mid Holocene and present (Werneck et al., Reference Werneck, Costa, Colli, Prado and Jack2011), the warmer colors correspond to regions with STDFs in South America; (B) map of vegetation distribution during the HCO and LGM in South America (modified from Vivo and Carmignotto, Reference Vivo and Carmignotto2004).
Figure 4 Long description
A) Maps of South America display seasonally dry tropical forests (SDTFs) under narrow and broad definitions for current climate, Holocene 6,000 years ago and Last Glacial Maximum (LGM) 21,000 years ago. Warmer colors indicate higher probability of SDTF presence. North is oriented at the top. B) Vegetation distribution during the Holocene Climate Optimum (HCO) and LGM shows regions like evergreen forests, dense savannas, semi-evergreen forests, open savannas, grassland/steppe and desert/cold steppe. Key areas include the Amazon basin and northeast Brazil. Stars mark Jirau tank deposit in Itapipoca and Miranda River in Miranda. The maps illustrate changes in vegetation formations over time, with specific areas labeled for different vegetation types.
Climatic and environmental changes between the last glacial maximum (LGM) and the Holocene Climatic Optimum (HCO) profoundly altered the distribution of open and closed habitats across South America. The HCO, which occurred approximately between 6500 and 4000 years BP, was a global event characterized by significantly higher humidity levels compared to the LGM. This climatic shift led to the replacement of extensive open vegetation formations by forests or other closed-canopy landscapes in tropical regions situated within the 30° latitude belt of South America (Vivo and Carmignotto, Reference Vivo and Carmignotto2004). Nevertheless, some areas within the BIR remained relatively stable during this period, providing suitable environmental conditions that enabled the persistence of megafaunal species until the Late Holocene (Figure 3).
Itapipoca and Miranda are located in the Caatinga and Cerradão regions, respectively, both of which are broadly classified as seasonally dry tropical forests (SDTFs; Figure 4), according to paleodistribution maps by Werneck et al. (Reference Werneck, Costa, Colli, Prado and Jack2011). SDTFs are characterized by a markedly seasonal climate, with several months of intense drought and a diverse vegetation mosaic that includes deciduous and semideciduous forests, shrubs, cacti, woodland savannahs, and open savannahs (Mooney et al., Reference Mooney, Bullock, Medina, Bullock, Mooney and Medina1995). The Cerradão corresponds to deciduous and semideciduous forest formations within the Cerrado biome (savannah) and is also classified as an SDTF (Werneck et al., Reference Werneck, Costa, Colli, Prado and Jack2011). The current interglacial period is marked by increased humidity and tropical conditions, as evidenced by pollen and paleoclimatic data (De Oliveira et al., Reference De Oliveira, Barreto and Suguio1999a, Reference De Oliveira, Bezerra, Mozeto and Aravena1999b; Behling et al., Reference Behling, Arz, Tzold and Wefer2000; Cruz et al., Reference Cruz, Vuille, Burns, Wang, Cheng, Werner, Lawrence- Edwards, Karmann, Auler and Nguyen2009), which led to the reduction and fragmentation of open environments in the BIR (Figure 4).
Pollen data from the marine core GeoB 3104-1, located approximately 70 km from the city of Fortaleza (Ceará State) and 137 km from Itapipoca, obtained by Behling et al. (Reference Behling, Arz, Tzold and Wefer2000), indicate a predominance of Caatinga and Cerrado vegetation between 42,000 and 8500 years BP in northeast Brazil. This dominance was interrupted by brief humid intervals around 40,000, 33,000, and 24,000 years BP, as well as between 15,500 and 11,800 years BP, during which tropical and gallery forests expanded over the Caatinga and Cerrado areas (Behling et al., Reference Behling, Arz, Tzold and Wefer2000). Between 11,800 and 8500 years BP, a return to drier climatic conditions is recorded, marked by the re-expansion of Caatinga and Cerrado vegetation.
Pollen data from the middle course of the São Francisco River (Bahia), in the Icatu River valley – approximately 700 km from Itapipoca – reveal the vegetational dynamics of the Caatinga biome from 10,990 years BP to the present (De Oliveira et al., Reference De Oliveira, Barreto and Suguio1999a). Five pollen zones with distinct vegetation characteristics were identified: SA1 (10,990–10,540 years BP), containing pollen from Amazonian and montane forest taxa, suggesting humid conditions and low temperatures; SA2 (10,540–6790 years BP), marked by an abundance of Mauritia pollen, indicating a progressive increase in humidity and temperature, with a subsequent decline in rainforest taxa and a gradual rise in Caatinga and Cerrado taxa between 8910 and 6790 years BP; SA3 (6790–6230 years BP), showing no pollen record, which indicates arid or semi-arid conditions; SA4 (6230–4535 years BP), reflecting the return of a vegetational mosaic of gallery forest, Cerrado, and Caatinga under more humid climatic conditions; and SA5 (4535 years BP to the present), characterized by the establishment of the current semi-arid climate and vegetation typical of the Caatinga.
Isotopic data obtained from soil organic matter in Lake Caçó, located in Barreirinhas (Maranhão State), covering the period from the Late Pleistocene (15,000 years BP) to the present, reveal the vegetation dynamics of the region (Pessenda et al., Reference Pessenda, Gouveia, Ribeiro, De Oliveira and Avarena2010). The study identified the presence of tropical forest between the Late Pleistocene and Early Holocene, indicating climatic conditions wetter than those of the present. Between 9000 and 3000 years BP, an expansion of Cerrado vegetation over tropical forest areas was recorded, corresponding to the establishment of a drier climate. From 3000 years BP to the present, isotopic data show a depletion in 13C values, associated with the re-expansion of tropical forests over Cerrado areas, reflecting the return of a humid climate similar to current conditions.
Considering the vegetation dynamics described above, along with the vegetation distribution model shown in Figure 4, it is evident that the Itapipoca region and other areas of northeastern Brazil maintained relatively stable habitats until the onset of the Late Holocene. This environmental stability likely supported the continued survival of megafaunal mammal species, indicating that northeastern Brazil functioned as an environmental refuge during the Holocene.
Lagoa Negra, located approximately 200 km from the Miranda River basin, provides the only available pollen record for this region (De Oliveira et al., Reference De Oliveira, Bezerra, Mozeto and Aravena1999b). Between 10,200 and 8770 years BP, the pollen spectrum indicates the predominance of extensive Cerrado (savanna) vegetation, reflecting dry climatic conditions. From 8870 to 7840 years BP, this pattern was reversed, with evidence of forest expansion over grassland areas, corresponding to a period of increased humidity. After 7840 years BP, the pollen record suggests a reduction in precipitation, indicated by a rise in grass and herbaceous pollen, reflecting the expansion of open vegetation over forested areas. From 5190 years BP to the present, a marked increase in arboreal and algal elements, along with a decrease in grass and herbaceous pollen, indicates the establishment of extensive forest cover with limited pasture areas. Therefore, the vegetation dynamics in the Miranda River basin, together with the isotopic values of E. laurillardi (Table 1), support the interpretation that suitable habitats for megafaunal species persisted in this region until approximately 5000 years BP.
Figures 3 and 4, corroborated by the data presented in Table 1, indicate that the regions of Itapipoca and the Miranda River valley maintained favorable environmental conditions for the survival of megafaunal species, characterizing these areas as refugia from the Late Pleistocene to the Mid Holocene. The extinction of megafauna in various parts of South America during the Pleistocene–Holocene transition has been well documented in the literature (e.g., Peyre et al., Reference Peyre, Guérin, Guidon and Coppens1998; Auler et al., Reference Auler, Piló, Smart, Wang, Hoffmann, Richards, Edwards, Neves and Cheng2006; Oliveira et al., Reference Oliveira, Porpino and Barreto2010; Dantas et al., Reference Dantas, Dutra, Cherkinsky, Fortier, Kamino, Cozzuol, Ribeiro and Vieira2013; Hubbe et al., Reference Hubbe, Hubbe, Karmann, Cruz and Neves2013; França et al., Reference França, Dantas, Bocchiglieri, Cherkinsky, Ribeiro and Bocherens2014). However, several studies also indicate that some large mammal species persisted longer, becoming extinct during the Early, Mid, and Late Holocene across the American continent (Guthrie et al., Reference Guthrie2004; Cione and Tonni, Reference Cione, Tonni, Barrio, Etcheverry, Caballé and Llambías2005; Steadman et al., Reference Steadman, Martin, MacPhee, Jull, McDonald, Woods, Iturralde-Vinent and Hodgins2005; Johnson et al., Reference Johnson, Holliday, Martínez, Gutiérrez and Politis2012; Prado et al., Reference Prado, Martinez-Maza and Alberdi2015; Faria et al., Reference Faria, Carvalho, Araújo-Júnior, Ximenes and Facincani2025).
The climatic changes that occurred between the LGM and the HCO profoundly affected the distribution of open and closed environments in South America. The increased humidity during the HCO favored the expansion of closed habitats, such as tropical forests, which in turn reduced the extent of open environments like the Cerrado and Caatinga – ecosystems that were more favorable to the survival of megafaunal species in the BIR. According to Vivo and Carmignotto (Reference Vivo and Carmignotto2004), these climatic shifts promoted the diversification of megafauna in Africa while contributing to their extinction in South America. The chronological data obtained by Faria et al. (Reference Faria, Carvalho, Araújo-Júnior, Ximenes and Facincani2025), together with the results presented in this study, indicate the persistence of large mammals in Brazil until the beginning of the Late Holocene. This finding challenges previous interpretations that placed the extinction of Brazilian megafauna at the Pleistocene–Holocene transition and suggests that comparable faunal patterns between South America and Africa persisted beyond the Late Pleistocene (Vivo and Carmignotto, Reference Vivo and Carmignotto2004).
When comparing the present-day faunas of South America and Africa, it is evident that they differ markedly in the number and diversity of large and megamammal species. According to Vivo and Carmignotto (Reference Vivo and Carmignotto2004), these distinctions emerged from the Mid Holocene onward, coinciding with the reduction and fragmentation of open vegetation areas and the expansion of tropical forests driven by increased humidity. The same authors argue that the reduction in the extent of various plant formations in South America and Africa explains the divergent faunal patterns observed between the two continents. In South America, the loss of open vegetation within the 30° latitude belt was particularly significant (Figure 5), whereas in Africa, extensive areas of semi-arid open vegetation, such as savannahs, persisted and continued to sustain large and megamammalian faunas. The climatic oscillations that affected both continents during the Late Pleistocene and Holocene thus contributed to the current disparity between the persistence of megafaunal communities in Africa and the dramatic (∼83%) loss of megafaunal species in South America. Consequently, the maintenance of large tracts of open vegetation in the Itapipoca and Miranda regions likely provided environmental refuges for megafaunal species until the onset of the Late Holocene.
(A) Current vegetation formations of South America and Africa; (B) vegetation changes on both continents during the LGM; (C) vegetation changes during the HCO; (D) summary of the changes described. In dark gray, stable areas for megafauna during the LGM and unstable habitats during the HCO, and light gray, stable habitats during the HCO and unstable habitats during the LGM (Vivo and Carmignoto, Reference Vivo and Carmignotto2004).
Figure 5 Long description
Thematic map set with four sub images labeled A, B, C and D. Each sub image shows South America on the left and Africa on the right, with north at the top. A latitude scale at left marks 30 degree north, Equator and 30 degree south. The image A showing present day vegetation formations. Labels include Desert, Savanna, Woodland, Evergreen forest, Dry forest and evergreen forest, Grassland, Grassland or Steppe, Mixed grass and shrub and Temperate formations. The label Equator appears across both continents. Desert is labeled across the northern part of Africa and also at the far southern end of Africa. Evergreen forest is labeled in the equatorial belt of South America and in the equatorial belt of Africa. Savanna is labeled on both continents on either side of the equatorial evergreen forest belt. Woodland is labeled in southern Africa. Grassland or Steppe is labeled in the southern part of South America. Temperate formations is labeled at the far southern end of South America. The image B showing vegetation changes during the LGM. Labels include Desert, Open savannas, Mosaic of open forest and savanna, Grassland or Steppe and Desert or Cold Steppe. Desert is labeled across the northern part of Africa and also at the far southern end of Africa. Open savannas is labeled across large areas of Africa south of the northern desert and also in parts of South America. Mosaic of open forest and savanna is labeled in northern South America. Grassland or Steppe is labeled in southern South America. Desert or Cold Steppe is labeled at the far southern end of South America. The image C showing vegetation changes during the HCO. Labels include Savanna, Dense savanna, Evergreen forest and Temperate formations. Evergreen forest is labeled in the equatorial belt of South America and Africa. Dense savanna is labeled across a broad band in Africa south of the equatorial evergreen forest. Savanna is labeled in parts of Africa and in parts of South America outside the evergreen forest belt. Temperate formations is labeled at the far southern end of South America. The image D showing a two class suitability overlay with a legend at upper right. The legend text reads, Suitable during HCO, Unsuitable during LGM and Suitable during LGM, Unsuitable during HCO. On South America, one class forms a broad area across much of the continent and the other class appears as smaller areas including a band in the southern part and additional patches. On Africa, one class covers a large central and southern area, while the other class appears in the northern desert region and in additional areas toward the south.
Conclusions
The isotopic data (δ13C) presented indicate an exclusively browser diet for P. major and X. bahiense, indicating that they inhabited areas of deciduous and semideciduous forest in Itapipoca at the Mid and beginning of the Late Holocene. E. laurillardi, T. platensis and N. platensis have a mixed diet with small proportions of grasses and herbaceous plants, indicating the presence of wooded savannas in Itapipoca. The generalist strategy (grazing, mixed, and browsing) indicates the ability to adapt to different ecosystems in the BIR and South America. S. populator, the only carnivore analyzed, also indicates the presence of wooded savannahs in Itapipoca. E. laurillardi presents a mixed diet for the region of the Miranda River basin, indicating that it inhabited a wooded savanna. Therefore, through the data obtained (Table 1) and also the pollen and paleoclimatic data, we corroborate the presence of environments favorable to the survival of megafauna until the end of the Mid Holocene in the BIR, indicating the region as the last environmental refuge for the magnificent fauna of large and megamammals in South America.
The regions of Itapipoca and the Miranda River basin, through the data and discussions presented, served as environmental refuges during the climatic/environmental transformations between the LGM and HCO, due to the presence of vast regions of different savannah-type environments in the BIR. The disappearance of vast areas of savannah, steppes and prairies in South America at the end of the HCO, replaced by areas of tropical and subtropical forests, made the survival of meso- and megamammal species in South America unviable. The Holocene ages obtained for the analyzed megafauna species coincide chronologically with the environmental/climatic transformations responsible for the disappearance of vast areas favorable to the survival of megafauna in the BIR.
When we compare the diversity of large and megamammals currently found in South America and Africa, we observe that all large mammals adapted to open environments (savannah type) in South America became extinct at the end of the Mid Holocene during the HCO. The persistence of vast areas of savannah environments in Africa at the end of the HCO allowed the survival of several species of megafauna on the African continent. Therefore, the disappearance of open environments due to replacement by tropical and subtropical forests in South America is the main agent of the extinction of megafauna on this continent.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/qua.2026.10077
Acknowledgments
To FAPERJ (Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro) (FAPERJ: E-26/202.047/2020; E-26/202.048/2020. E-26/200.828/2021; E-26/201.371/2021; E-26/200.998/2024), Conselho Nacional de Desenvolvimento Científico e Tecnológica (CNPq 304246/2024-7; 305576/2021-6; 174860/2023-4) and Universidade do Estado do Rio de Janeiro (Prociencia grant nº 38.944-5) for funding the research. We also thank the Museum of Prehistory of Itapipoca and the Zoology Laboratory of the Federal University of Mato Grosso do Sul, for providing the fossils for carrying out the research.
Declaration of competing interests
The authors declare that they have no known competing financial interest or personal relationships that could have appeared to influence the work reported in this manucript.
