Data from eleven of the twelve frog species of the family Ceratophryidae are included. Ceratophrys testudo was excluded, as this species has not been collected since the holotype (Andersson 1945). DNA sequences from several specimens per species were available for most species on GenBank. Based on the results from Feng et al. (2017), Calyptocephallela gayi (Calyptocephalellidae), Gastrotheca weinlandii (Hemiphractidae), Pseudis paradoxa (Hylidae) and Rhinoderma darwinii (Hemiphractidae) were selected as outgroup species where Calyptocephallela gayi was used to root the tree. The GenBank accession numbers can be found in Suppl. material 3: table S1.

The phylogenetic tree of Ceratophryidae was constructed by first analysing a concatenated matrix with 8172 bp DNA sequences from GenBank. The six mitochondrial gene sequences used include portions of cytochrome oxidase subunit I (COI) gene, cytochrome b (cytb) gene, 12S rRNA gene, the intermediate partition sequence tRNAVal gene, 16S rRNA gene and NADH dehydrogenase subunit 1 (ND1) gene. The eight nuclear gene sequences include the Seven In Absentia Homolog 1 (SIAH1) gene, Tyrosinase precursor (TYR) gene, Recombination activating protein 1 (RAG-1) gene, Ribosomal protein L3 (RPL3) gene, Proopiomelanocortin A (POMC) gene, Chemokine receptor 4 (CXCR4) gene, lactose dehydrogenase beta chain gene (lactose_dh) and Fibrinogen A alpha polypeptide gene (fib_A) (Suppl. material 3: table S1). In addition, 256 morphological characters from Gómez and Turazzini (2021) were added as additional characters in the mixed matrix to enrich the dataset (Suppl. material 3: table S1).

We employed BEAST v. 2.7.4 (Bouckaert et al. 2014) to estimate the divergence time of Ceratophryidae based on the genetic matrix. The analysis ran for one hundred million generations utilizing a relaxed clock log-normal model, with log-data sampled every 10,000^th iteration. A calibrated Yule Model with a 25% burn-in and a randomly generated starting tree were utilized. We estimated a TIM2+F+I+G4 substitution model for the genetic matrix in IQ-TREE v.1.6.12. Subsequently, a GTR model was set in BEAUti 2, incorporating rate parameters estimated as follows: A-C: 5.1986, A-G: 15.2835, A-T: 5.1986, C-G: 1.0000, C-T: 34.0549, G-T: 1.0000 (see Suppl. material 7: appendix S1). For the BEAST analysis, we employed the fossils Ceratophrys ameghinorum (Fernicola 2001) and Lepidobatrachus australis (Moreno 1889), along with the nodes 37, 41, 51, 52, and 53 from the time-calibrated phylogenetic tree of Feng et al. (2017), as calibration points. C. ameghinorum was used to establish a minimum bound between 5.33 and 4 million years ago (Ma) for the divergence of the clade containing C. aurita and C. joazeirensis from its sister group (Tomassini and Montalvo 2013). Similarly, L. australis was used to establish a minimum bound between 5.33 and 4 Ma, for the divergence of Lepidobatrachus from its sister group (Tomassini and Montalvo 2013; Gómez and Turazzini 2021).

Two independent runs were made, and sampled values were joined and analysed for convergence. Tracer v.1.7.2 (Rambaut et al. 2018) was used to assess the best run based on the likelihood effective sample size (ESS). All parameters in the model had an ESS higher than 900 (Suppl. material 4: table S2). TreeAnnotator (Bouckaert et al. 2014) was used to obtain a maximum clade credibility tree with divergence times from the best run. Chronogram tree visualization and editing was done with FigTree v.1.4.2.

The divergence time reconstruction inferred for the phylogenies is the key input for biogeographical analysis (Canitz et al. 2022). We utilized the BioGeoBears package (Matske 2013) in RASP v.4.4 (Yu et al. 2020) to compare biogeographical models based on AICc and AICw-weighted statistics. RASP v.4.4 includes a likelihood version of the dispersal-vicariance analysis (DIVALIKE), the dispersal-extinction-cladogenesis model (DEC), and the Bayesian analysis of biogeography (BAYAREALIKE). Furthermore, it includes the free parameter “j” (“jump dispersal”) to simulate founder-event speciation (Matske 2013), additional to the dispersal (d) and extinction (e) rates. Every species in the analysis was coded according to the biogeographic regions within South America outlined by Hutter et al. (2017): (A) Cerrado, (B) Temperate, (C) Atlantic Forest, (D) Amazonia and the Guiana Lowlands province, and (E) Choco. As no species from the Ceratophryidae are distributed in the Tropical Andes and Guiana Shield, we excluded those regions from the analyses. It was decided to include the Guiana Lowlands province within the Amazonia region to simplify the analysis, as some occurrences of C. cornuta were found in this region. We verified the species-specific geographical data acquired from the biodiversity information system GBIF (https://www.gbif.org/; accessed in May 2023) and cross-referenced them with the distribution polygons per species obtained from the IUCN Red List website (https://www.iucnredlist.org/). A total of 6744 occurrences for all the extant Ceratophryidae species (except for C. testudo) and the four outgroup species were gained from the GBIF. Errors, doubtful, and duplicated occurrences were removed, resulting in a final dataset of 2860 records: 1383 for the outgroup and 1477 for Ceratophryidae species (Fig. 1, Suppl. material 5: table S3).

The maximum number of areas in ancestral ranges was restricted to three, in accordance with the current distributional range codified for each species across the designated regions. We estimated the probabilities of ancestral areas for each node, drawing from a subset comprising 1000 randomly selected trees in RASP v.4.4., utilizing the chronogram and trees derived from BEAST2 software as input files (Yu et al. 2015).

Figure 1. 

Biogeographic map for species in the Ceratophryidae family. (A) Map of South America with convex hull polygons representing the occurrences for every species of Ceratophrys. (B) The Gran Chaco region with convex hull polygons that represent the occurrences for species of Lepidobatrachus and Chacophrys.

The nineteen bioclimatic variables were downloaded from the WorldClim2 database (http://www.worldclim.com/version2; accessed in December 2023), with a 30 seconds (~1 km²) spatial resolution (Hijmans et al. 2005). We generated a database of environmental data extracted for each location point per species using ARCMAP 10.5 (ESRI 2015). We performed three methods to analyse the relatedness of environmental variables to phylogenetic groups in Ceratophryidae: (a) We used the Jackknife method (Sinharay 2010) to estimate the relative contributions of environmental variables for each species, ran in Maxent v.3.4.3 software (Phillips et al. 2017); (b) A principal components analysis (PCA) was performed using PAST v.4.03 (Hammer et al. 2001) to determine the environmental variables with the highest loading per phylogenetic groups; and (c) a phylogenetic Principal Components Analysis (pPCA) was performed using the mean environmental data from 19 variables and the chronogram tree to analyse the evolutionarily covariance of phylogenetic species related to environmental niche differentiation by region. We utilized the phyl.pca function to run the pPCA, and the phylomorphospace function to project the phylogeny and the biogeographic regions onto the first two environmental principal components. Both functions were applied with the phytools package (Revell 2012; Revell and Harmon 2022) in R software (R-Core-Team 2023).

The genetic and the combined analysis (genetic + morphology) show a similar phylogeny, where the monophyly of Ceratophrys, Chacophrys and Lepidobatrachus is well supported for Ceratophryidae (Fig. 2 and Suppl. material 1: fig. S1). The phylogeny yielded Ceratophrys to be the sister clade of Chacophrys + Lepidobatrachus with a 100% Ultra Bootstrap Approximation (UFB). Ceratophrys is divided into major two clades, where the first clade contains C. calcarata and C. cornuta (UFB = 100%), and the second clade includes C. stolzmanni as sister (UFB = 82%) of the clade containing C. aurita, C. joazeirensis, C. cranwelli and C. ornata (UFB = 100%).

Figure 2. 

Phylogenetic relationships of Ceratophryidae based on genetic (Faivovich et al. 2014) and morphological data (Gómez and Turazzini 2021) with Ultrafast Bootstrap Approximation percentages. Outgroup relationships can be found in Suppl. material 3: table S1.

The divergence times are shown in Table 2 and the chronogram is provided in Suppl. material 2: fig. S2 and Suppl. material 8: appendix S2. The chronogram estimated with BEAST suggests that the crown group of Ceratophryidae originated 19.2 Ma in the early Miocene (95% Highest Posterior Density [HPD] 14.7–24.1 Ma, node 5). Within the Ceratophryidae family, the crown group of Ceratophrys is estimated at 17.2 Ma (95% HPD 12.5–21.7 Ma, node 9) in the Miocene; the crown group of Chacophrys diverged from Lepidobatrachus near to 13.7 Ma (95% HPD 8.2–19.1 Ma, node 6); and the crown group of Lepidobatrachus originated 5.0 Ma in the Pliocene (95% HPD 4.0–6.5 Ma, node 7). All divergence times within Ceratophryidae are estimated to fall in the Neogene (23.0–2.6 Ma; Fig. 3, Suppl. material 2: fig. S2).

Figure 3. 

Ancestral Area Reconstruction for extant species of Ceratophryidae in South America. A dispersal-extinction-cladogenesis model (DEC) for Ceratophryidae reconstructed with RASP v.4.4. software. The condensed tree is a chronogram based on 10,001 trees derived from genetic data reconstructed by Bayesian inference analysis in BEAST2 for Ceratophryidae. Biogeographical region categories are based on Hutter et al. (2017). The colored circles on internal nodes visualize the marginal probabilities of ancestral area for the corresponding clade of that node. Letters inside each node represents the most probable ancestral area, whereas support values and node number are located on left and right of each node, respectively. Biogeographic events are highlighted for dispersal event (blue) and vicariance events (green) surrounding each node. The geographical distribution of taxa across regions is represented using color coding, with each extant taxon depicted at the tip nodes. Symbology in the map of South America: Estimated time event of speciation are detailed in million years ago (Ma), arrows denote dispersal events, dashed lines indicate vicariance events, a star symbol marks the estimated origin region for Ceratophryidae. Biogeographic information on speciation events per node can be found in Table 2 and Suppl. material 8: appendix S2. Photograph credits are detailed in Suppl. material 10: appendix S4.

The Dispersal-extinction-cladogenesis model (DEC = –30.38 lnL; AICc = 65.76; AICc_wt = 39%) was the best among other biogeographical models (Table 3). We identified five dispersals followed by five vicariance events in the biogeographical history of the Ceratophryidae clade to explain distribution of extant species in South American regions (Fig. 3; Suppl. material 9: appendix S3). The ancestral area reconstruction suggests an origin around 19.8 million years ago (95% HPD 14.72–24.05 Ma, node 5) in the Cerrado region for Ceratophryidae (Fig. 3, Table 2). Speciation of Chacophrys pierottii (node 6, 13.70 Ma [95% HPD 8.15–19.50 Ma]) and the crown group Lepidobatrachus (node 7, 4.95 Ma [4.00–6.47 Ma]) also has its origin in the Cerrado region around 13.70 Ma to 3.46 Ma (Fig. 3). Independent dispersion and vicariance events occurred between approximately 17 Ma to 10 Ma, leading to the migration from the Cerrado region to both the Choco and Amazonia. This migration marked the diversification of Ceratophrys cornuta, C. calcarata, and C. stolzmanni (Fig. 3, nodes 9–11). Consequently, C. cornuta inhabited the Amazonia region (area D), while C. calcarata colonized the Choco region (area E), in a vicariance event estimated around 10.4 Ma (95% HPD, 5.1–15.8 Ma, see Fig. 3). The distribution of C. stolzmanni (14.71 Ma, Cerrado → Choco|Cerrado) and C. calcarata (17.2 Ma, Cerrado → Choco; 10.3 Ma, Choco → Amazonia|Choco), suggests an independent origin due to sequential dispersal and vicariance events from Cerrado to colonize the Choco region (area E, see Fig. 3). The most recent speciation event within Ceratophryidae are estimated to have occurred in the Pliocene, producing the split of C. ornata and C. cranwelli (4.8 Ma [95% HPD 2.1–8.2 Ma]; node 13), as well as C. joazeirensis and C. aurita (4.2 Ma [95% HPD 4.0–4.7 Ma]; node 14). These splits reflect, respectively dispersal and vicariance events from the Cerrado towards the Temperate and the Atlantic Forest (Fig. 3).

The Jackknife resampling test shows the environmental importance of Temperature Seasonality (BIO4) for several species of Ceratophryidae (C. aurita, C. cranwelli, C. ornata, Ch. pierottii, L. asper, L. laevis and L. llanensis). Mean Temperature of Coldest Quarter (BIO11) is environmentally important for C. calcarata, Precipitation of Driest month (BIO14) for C. cornuta, Annual Precipitation (BIO12) for C. joazeirensis and Precipitation of Coldest Quarter (BIO19) for C. stolzmanni (Table 1).

The environmental PCA analysis for Chacophrys and Lepidobatrachus displays overlap between the four species (Fig. 4A). Annual Precipitation (BIO12) accumulates 89.7% of the total variance on Principal Component 1 (PC1), while Temperature Seasonality (BIO4) and Precipitation of Warmest Quarter (BIO18) accumulates 7.4% of the variance in PC2. Both components include 97.1% of variance. The PCA analysis for the Ceratophrys group shows that Temperature Seasonality (BIO4) and Annual Precipitation (BIO12) together accumulate 89.9% of the total variance in PC1. This analysis reveals a separation between C. stolzmanni from C. cornuta and C. aurita in the environmental space (Fig. 5). Precipitation of Coldest Quarter (BIO19) and Precipitation of the Warmest Quarter (BIO18) accumulates 5.4% of the variation in the PC2 (Fig. 4B, Suppl. material 6: table S4).

The pPCA shows that the first principal component has the greatest multivariate load from Precipitation Seasonality (BIO15, positive load) and overall precipitation (BIO12; negative load). The second PC is mostly accumulated by Temperature Seasonality (BIO4; negative load) and Isothermality (BIO3; positive load) and. C. cornuta is highly related with precipitation variables and Temperature seasonality, located far from all other members of this family in the environmental space (Fig. 5). There is a group which include all species of Lepidobatrachus, Ch. pierottii and C. ornata highly related to Precipitation seasonality (BIO15) and Isothermality (BIO3). Ceratophrys stolzmanni and Ceratophrys joazeirensis are mostly related to Precipitation seasonality (BIO15). Vicariance events are associated with variation in niche differentiation of extant Ceratophrys species in the environmental space by regions (Fig. 5). Species from Chacophrys and Lepidobatrachus have evolved and originated within the Cerrado region (Fig. 3) where they inhabit similar environmental niches characterized by arid ecosystems. This stands in contrast to the diverse environmental niches observed within the Ceratophrys clade (Fig. 4 and Fig. 5).

Figure 4. 

Scatter plot of the Principal Components Analysis (PCA) and correlation matrix for the nineteen environmental variables from WorldClim2 for Ceratophryidae. PCA for extant (A) species of Chacophrys, Lepidobatrachus, and (B) Ceratophrys in South America. Values in the x-axis are the percentage of variance explained mostly by the principal component 1 (PC1) and on the y-axis principal component 2 (PC2). The colored dots represent individual occurrences per species in the PCA and are surrounded by 95% ellipses. (C) Correlation plot with positive (blue) or negative (red) direction that show the collinearity between environmental variables. Variables names are detailed in Table 1.

Figure 5. 

Phylomorphospace of Phylogenetic Principal Components for extant Ceratophrys species in South America. (A) Phylogenetic PCA for Ceratophryidae species to represent the environmental niche differentiation data based on the phylogenetic chronogram. Tip colors indicate (B) Biogeographical region categories based on Hutter et al. (2017): A: Cerrado, B: Temperate, C: Atlantic Forest, D: Amazonia and the Guiana Lowlands province, E: Choco; because none of the species from Ceratophryidae are distributed in the Tropical Andes and Guiana Shield, we excluded those regions from the analyses (Other). (C) Precipitation annual mean (mm) data from BIO12 variable. (D) Temperature seasonality (Standard Deviation × 100) data from BIO4 variable. Species codes: Ceraur = Ceratophrys aurita; Cercal = Ceratophrys calcarata; Cercor = Ceratophrys cornuta; Cercra = Ceratophrys cranwelli; Cerjoa = Ceratophrys joazeirensis; Cerorn = Ceratophrys ornata; Cersto = Ceratophrys stolzmanni; Chapie = Chacophrys pierottii; Lepasp = Lepidobatrachus asper; Leplae = Lepidobatrachus laevis; Leplla = Lepidobatrachus llanensis. Biogeographic events are highlighted with letters (D, yellow) for dispersal event and (V, green) for vicariance events based on the DEC developed in RASP software. Statistic groups assessed by Tukey test in (C) and (D) are represented by letters, whereas species that have significative differences (p < 0.05) among all other species are represented by asterisk (*).

The phylogenetic tree (Fig. 2) supports previous studies about the phylogeny of the family Ceratophryidae. Nonetheless, there are still differences. The comparison between the combined (DNA + morphology) and the DNA based data set resulted in the same topology. However, the combined dataset gave a higher node support in the topology compared with the DNA based data set (Fig. 2, Suppl. material 1: fig. S1). Morphological characters could exhibit phylogenetic signal that complements genetic data, which tend to increase node support, especially in cases where genetic variation is low, when molecular markers fail to resolve deep phylogenetic relationships (Wahlberg et al. 2005) or when distantly related taxa share similar genetic sequences due to convergent evolution (Edwards et al. 2012). Chacophrys was expected to be the sister group of Ceratophrys + Lepidobatrachus based on the 93% Jackknife support from Barcelos et al. (2022). However, this is inconsistent with our phylogeny, which exhibits a high node support for the clade which includes Ceratophrys as sister group to Chacophrys + Lepidobatrachus (Fig. 2). On the other hand, other studies have hypothesized a similar phylogenetic relationship (Fabrezi 2006; Faivovich et al. 2014; Frazao et al. 2015; Brusquetti et al. 2018; Hime et al. 2021), despite Faivovich et al. (2014) recovering low support for this group (<50%).

All nodes within Lepidobatrachus in our phylogeny (Fig. 2) are well supported and agree with the results from previous studies about this genus (Faivovich et al. 2014; Brusquetti et al. 2018; Gómez and Turazzini 2021; Barcelos et al. 2022). In the Ceratophrys clade, we found C. stolzmanni to be phylogenetically linked to two clades formed by C. aurita + C. joazeirensis and C. cranwelli + C. ornata, with relative high support values (82% UFB, 100% UFB in the inner clades). This contrasts with previous studies (Faivovich et al. 2014; Barcelos et al. 2022), where C. stolzmanni was found to belong to a clade also including C. cornuta + C. calcarata, but with lower support values (<60% parsimony jackknife frequencies) compared to our study. It was hypothesized that C. cranwelli and C. ornata were sister taxa, as detailed by Faivovich et al. (2014). However, in the phylogeny presented by Barcelos et al. (2022), the Jackknife frequency node support was <50%. Our phylogeny has a 100% node support on this clade confirming that C. cranwelli and C. ornata are sister taxa (Fig. 2). These different results can be explained by the fact that Barcelos et al. (2022) utilized morphological data, whereas the analysis of Faivovich et al. (2014) relied exclusively on genetic data. Our study supplements these analyses by incorporating both genetic and morphological characters (Fig. 2).

Our chronogram mostly agreed with previous studies in estimating divergence times in Anura, despite the inclusion of only a few representatives of Ceratophryidae in those analyses (Fabrezi et al. 2009; Faivovich et al. 2014; Brusquetti et al. 2018). We present the first genetically calibrated chronogram of the Ceratophryidae family (excluding C. testudo). The divergence of the crown group of Ceratophryidae is estimated to have occurred ~19.2 Ma (95% HPD 14.7–24.1 Ma) at the beginning of the Neogene (Fig. 3, Suppl. material 2: fig. S2). The divergence time estimation of the MRCA of Chacophrys and Lepidobatrachus is congruent with the ML tree of Roelants et al. (2007), at around 18.8 Ma (95% HPD 12.7–25.1 Ma). These results agree with the divergence time estimations within Lepidobatrachus (Ruane et al. 2011; Frazao et al. 2015; Brusquetti et al. 2018; Hime et al. 2021).

The ancestral area for Ceratophryidae reveals that the Cerrado (area A; Fig. 3) is the core area estimated to be the biogeographic origin for this group. The high diversity and closely related fossil species of this clade documented in the Cerrado region, such as Baurubatrachus pricei and Beelzebufo ampinga (Barcelos et al. 2022), support the hypothesis of the origin of this clade in this region in South America (Fig. 3). This agrees with the estimated ancestral area for Chacophrys and Lepidobatrachus frogs, which is also suggested to be in the Cerrado region (Figs 3–5). Faivovich et al. (2014) predicted that most diversification in Ceratophryidae occurred in semi-arid environments with three independent transitions to different humid environments in the Choco, Amazonia and Atlantic Forest. We found that temperature and precipitation seasonality are important factors for extant species located in the Cerrado region (i.e. Ch. pierottii, Lepidobatrachus species, C. cranwelli and C. joazeirensis), whereas main precipitation is related to humid tropical inhabitants like C. cornuta and C. calcarata (Fig. 5).

The node related with extant Ceratophrys species (node 9) has a high marginal probability to assign the Cerrado region as the ancestral area (A = 61.89%), but low probability to support the RASP route in the speciation events (10.5%; Table 2; Suppl. material 8: appendix S2). The most probable scenario entails independent dispersions originating from the Cerrado to other regions, succeeded by vicariance events in the Choco, Amazonia, Atlantic Forest, and Temperate Forest (Fig. 3, Suppl. material 8: appendix S2). Extant species of this group show limited dispersal capabilities and a strong dependence of their habitat (Zeisset and Beebee 2008). Inside this clade, a vicariance event caused the divergence of C. cornuta and C. calcarata ~10.4 Ma, where they inhabited the Amazonia and Choco regions, respectively (Fig. 3). The Eastern Andes Cordillera experienced surface uplifts during the middle Miocene period (Leier et al. 2013; Carrapa et al. 2014), and Gregory-Wodzicki (2000) suggested that not more than half of the modern elevation of the North-eastern Cordillera was present by ~10 Ma (Hoorn et al. 2010) . This uplift may explain the vicariance event causing the divergence of C. calcarata isolated in the Choco region and C. cornuta in the Amazonian Basin. It this case, the MRCA of both species has a hypothetical wider distribution in the Choco and the Amazonia regions, which were isolated by the rise of the young Eastern Andes Cordillera after the Pebas system (23 to 10 Ma) transitioned to the Acre system (10–7 Ma) in the late Miocene (Fig. 3). On the other hand, the divergence time estimation data indicates a vicariance event in the distribution of C. stolzmanni, occurring ~14.4 Ma (95% HPD 10.5–18.5; node 11; Fig. 3), which was caused by the formation of the Andes in southern Ecuador (Evenstar et al. 2015). The divergence of C. aurita from C. joazeirensis and C. cranwelli + C. ornata (node 12) is probably caused by the epeirogenic uplift of the Brazilian Shield during Middle- Late Tertiary elevations (Ab’sáber 2000; Klöcking et al. 2020).

Cocoon production is observed in all species of Ceratophryidae, but is unknown for the species C. testudo, C. calcarata and C. cornuta (Faivovich et al. 2014). In agreement with these findings, cocoon formation originated in the MRCA of the clade, estimated to have occurred 19.2 Ma in the early Miocene (Faivovich et al. 2014). The formation of a cocoon, which acts as protection against water loss in dry environments, could have been lost in C. cornuta and C. calcarata, both inhabitants of the wet tropical forest in the Amazonia and the Choco regions, around 10.4 Ma (node 10, Fig. 3).

Three species of Ceratophrys (C. cornuta, C. calcarata and C. aurita) inhabit areas with a high average precipitation. C. cornuta is distributed in tropical humid environments, separated from all other species in the environmental space (Fig. 5). C. aurita is an inhabitant of the Atlantic Forest, characterized by a higher temperature seasonality compared with the former two species (Fig. 5). All other species from Ceratophrys, Lepidobatrachus and Chacophrys are associated with more semi-arid environments characterized by higher and variable temperature seasonality (Fig. 5, Table 1). Biogeographically, environmental niche differentiation is shown even in phylogenetic sister species (e.g., C. aurita–C. joazeirensis; C. ornata–C. cranwelli; Figs 3–5).

Faivovich et al. (2014) hypothesised that most diversification in Ceratophryidae occurred in semi-arid environments, with three independent transitions to humid environments by C. cornuta, C. aurita and C. ornata. Our results suggest two independent environmental niche differentiation shifts to wet environments, one associated with C. cornuta and C. calcarata, and the other associated with C. aurita (Figs 3–5). Nevertheless, Faivovich et al. (2014) also concluded an independent transition from C. ornata to a more humid environment of the Pampean region in central Argentina. However, our pPCA and PCA tests showed a relationship of this species with highly variable Temperature seasonality and precipitation of the driest month (Figs 4, 5, Table 1). The niche differentiation observed between sister species C. joazeirensis and C. aurita can be attributed to variances in average precipitation (Figs 4, 5). C. joazeirensis primarily inhabits the Cerrado Brazilian forest (area A, Fig. 3), characterized by lower precipitation levels in contrast to the Atlantic Forest (area C), where C. aurita is found.