The area delineated for the ENM includes the SATG and the Espinal and Delta ecoregions covering the estimated distribution area in Barrio (1980) and IUCN (2008) (Fig. 1) (for more details, see Supplementary material). We used 19 bioclimatic variables from CHELSA V2.1 at the spatial resolution of 30 arc seconds (Karger et al., 2017). Since C. ornata has burrowing habits (Gallardo, 1970), we also used soil layers from SoilGrids (Hengl et al., 2017). LULC layers were not included in the ENM due to the unavailability of layers covering a wide range of record dates. Given the high conversion rate of SATG, modeling the species potential distribution with non-contemporary layers would introduce biases into the model estimation. All variables were clipped to the extent of the selected calibration area (Fig. 1).

Fig. 1.

Model calibration area used to assess the potential distribution of C. ornata, showing the South American Temperate Grasslands (SATG) and its subdivisions and the Espinal and Delta Ecoregions. AR: Argentina. BR: Brazil. UY: Uruguay.

(1.06MB).

We gathered occurrence data from herpetological collections, citizen science, and long-term fieldwork monitoring. To obtain records from herpetological collections, we extensively searched online databases (Supplementary material) and contacted relevant museums, institutes, and universities. We confirmed the record accuracy by visiting the collections or asking curators for photos. Data from citizen science were obtained through a program applied in Argentina, Brazil, and Uruguay (see methodology in Supplementary material). Long-term monitoring in Argentina has been conducted since 2008. Survey techniques included auditory surveys (Heyer et al., 1994) after heavy rainfalls coinciding with high calling activity (Gallardo, 1972). We conducted five auditory surveys (Heyer et al., 1994) per night (1900–0200, 10 min each) over two consecutive nights after heavy rainfall in 240 sites. Some of the records obtained were published in previous studies (Agostini et al., 2012, 2016, 2020, 2021; Deutsch et al., 2017; Perrone et al., 2022).

To model the species potential distribution and suitable areas, we used Maxent 3.4.1 algorithm (Phillips et al., 2017). Niche modeling procedures (i.e., database filtering, correlation analysis, model selection and model conversion to binary) are detailed in the Supplementary material. All statistical analyses were conducted in the R environment (https://www.r-project.org/).

In areas predicted by the model to be suitable for C. ornata but with no recent records (after 1995), we performed targeted field surveys employing active sampling and passive acoustic monitoring (PAM) (Sugai et al., 2019). The site selection and methodological procedure are detailed in Supplementary material. Active sampling, PAM locations and settings of each automatic recorder are detailed in Appendix 1 in Supplementary material. Sites with unsuccessful records were considered as 'absences'.

To evaluate how differently LULC and landscape configuration influence C. ornata occurrence, we employed Generalized Linear Models with binomial family distribution (GLMs) (Crawley, 2007). Due to the lack of high-quality regional LULC layers predating 1985 and considering that our database contains records dating back almost 100 years, we opted for a model incorporating current records and contemporary LULC layers. Hence, C. ornata occurrence was included in a single model as a response variable with two states: presence (records from 2015 onwards) and absence. From each point of presence and absence, we extracted LULC in a buffer of a 1,000 m radius. To define the buffer size, we based on home range studies of C. ornata (Ibáñez, 2023). LULC layers for 2019 were downloaded from Map Biomas Pampa v.1.0. (Baeza et al., 2022). We considered the percentage of six cover categories (Flooded Grassland, Permanent Water Bodies, Agriculture and Livestock, Forest Plantation, Open Forest, and Closed Forest) and of Percentage of Sand in the soil (downloaded from Soilgrids, Hengl et al., 2017), and two landscape configuration variables (Number of Flooded Grassland Patches and Distance to the Nearest Flooded Grassland Patch) as fixed effects. We excluded the Unvegetated Area category from the model since it does not distinguish between urbanized areas, newly planted crop fields and sandy coast. Methodological details of GLM analysis are provided in Supplementary material.

We obtained 991 confirmed records (932 for Argentina, 34 for Brazil, and 25 for Uruguay) (Appendix 2 in Supplementary material). To avoid sampling bias and spatial correlation when running the ENM, we used a 3 km buffer, resulting in 317 records.

A total of 9 bioclimatic and soil variables were selected for the model (Table 1). Model building resulted in 48 candidate models with different feature classes–regularization multiplier combinations. Following the sequential criteria (see Supplementary material), the feature class Hinge with a regularization multiplier of 1 was selected as the optimal setting (hereafter H1; MTP OR: 0.00; test AUC: 0.9048; ΔAICc: 89.45). The predicted suitable areas extend to the northwest of Buenos Aires province, east and north of the Atlantic coast (Buenos Aires province), south of Córdoba and Santa Fe provinces, and northeast of La Pampa in Argentina. Suitable areas in Uruguay comprise the southern and eastern coasts and, in Brazil, southern Rio Grande do Sul (Fig. 2A). The binary map predicts that suitable habitat for C. ornata covers about 413,204 km2 (Fig. 2B).

Fig. 2.

(A) Final Maxent prediction of suitable areas for C. ornata. The black circles correspond to the presence data used in the ENM. (B) Final binary model showing suitable and unsuitable areas for the species. AR: Argentina. BR: Brazil. UY: Uruguay.

(0.71MB).

Maxent showed that four bioclimatic variables emerged as the most relevant to the model and contributed to 93% of the variation: isothermality, mean annual air temperature, annual precipitation amount, and precipitation seasonality (Table 1).

We obtained 120 absences in areas predicted to be suitable but with no recent records (Appendix 2 in Supplementary material). Active surveys covered ∼368 days/307 h, and ∼808 h of recordings were obtained from passive acoustic monitoring.

GLM results indicated that the current occurrence of C. ornata is associated with four variables (R2 Nagelkerke = 0.455): Distance to the Nearest Patch, Forest Plantations, Permanent Water Bodies, and Percentage of Sand. Percentage of Sand (Z = 5.850; p < .05) and Percentage of Permanent Water Bodies (Z = 2.121; p < .05) significantly and positively influenced the species probability of occurrence. Conversely, the percentage of Forest Plantations (Z = −2.216; p < .05) and Distance to the Nearest Patch (Z = −2.297; p < .05) negatively influenced the presence of C. ornata (Table 2).

The impact of habitat loss has been widely identified as the most direct cause of global biodiversity annihilation (Ceballos et al., 2017). However, it has been challenging to determine the association between this threat and decline events, particularly for species with wide distribution ranges (Staude et al., 2020). Furthermore, several studies on amphibian populations in SATG revealed that various factors proposed to explain population declines exhibit a high degree of temporal and spatial variation (Agostini et al., 2020, 2021; Perrone et al., 2022), and this may also be the case for C. ornata. Therefore, given the different scenarios, distinct conservation actions are needed in each country of the species distributional range.

In Uruguay, the historical localities of C. ornata have been extensively transformed, mainly by forest plantations and urbanization (Baeza et al., 2022). Despite our intensive search efforts conducted in key areas, added to previous studies (Gambarotta et al., 1999; Bardier and Maneyro, 2015), the species has not been found since 1982. We recommend intensifying search efforts in coastal wetlands on the west coast of Uruguay, which has not yet been properly explored (Langone pers. comm.). We suggest updating the national conservation status and up-listing the category from Vulnerable to Critically Endangered (Possibly Extinct).

In Brazil, the citizen science program obtained more recent documented records than the last scientifically reported for this country by Braun and Braun (1980) and Gayer et al. (1988). In addition, in the southern State of Rio Grande do Sul, there are extensive suitable areas for C. ornata with some degree of protection (e.g., Taim Ecological Station) that still could harbor remnant populations. We recommend intensifying search efforts in strategic remote areas using passive acoustic monitoring.

To discuss the status of C. ornata in Argentina, it is necessary to perform a regional analysis. The northeastern Buenos Aires Province (also known as the Rolling Pampas subregion, see Fig. 1) emerged as a suitable area. Moreover, museum collections contain many specimens collected in this subregion before 1975, indicating the species was relatively abundant in the past. Nonetheless, the absence of records during field monitoring and the scarcity of recent citizen science records strongly suggest a local population decline in the Rolling Pampas. It is worth mentioning that the Rolling Pampas is one of the most modified areas of Argentina and one of the world's most impacted, where GMO glyphosate-resistant soy crops have replaced more than 75% of the native vegetation (Baeza and Paruelo, 2020). Under this scenario, habitat loss could have affected C. ornata populations and likely explain the remarkable disparity between historical and current occurrences (see Appendix 2 in Supplementary material).

In agreement with previous studies (Deutsch et al., 2017), we identified two areas in the Argentinean territory harboring the best-preserved C. ornata populations and emerging as key areas for ensuring the viability of the species. The two priority conservation areas belong to the Inland Pampas and the Flooding Pampas, respectively (see Fig. 1). The fact that these areas have received moderate modification pressure, conserving the last and largest remnants of native grassland (Baeza and Paruelo, 2020) could explain the high number of current records. Nonetheless, different human activities may compromise the future of the grassland remnants and their biodiversity. In the Inland Pampas, the traditional land use consisting of extensive cattle ranching has been rapidly converted to crops in recent years, reducing the number and extension of natural grassland patches (Baeza et al., 2022). In the eastern part of the Flooding Pampas, urbanization and intensive cattle-raising have recently expanded over the grassland remnants (Athor and Celsi, 2016). Urbanization and tourism development promote the existence of forestations of invasive exotic species, road and trail construction, and wetland drainage, which are carried out without ecosystem planning (Athor and Celsi, 2016). We argue that applying in-situ conservation actions, combining research, mitigation activities and educational and awareness-raising programs could present a viable and practical approach to conserving C. ornata in Argentina. Remnants of natural grasslands and associated wetlands must receive conservation priority and increasing protected areas coverage (private and public) could offer an opportunity to conserve grassland patches. This is particularly suitable for the Flooding Pampas, where other conservation strategies that recognize and encourage alternative economic uses like sustainable cattle-raising (e.g., Alianza de Pastizal; Altmann and Berger Filho, 2020; Vaccaro et al., 2020) have been extensively working in the same direction. Furthermore, we recommend moving forward on population genetics studies that will allow us to evaluate future in-situ and ex-situ population management. Finally, productive activities and urbanization will continue to be in constant conflict with the conservation needs of temperate grassland relicts. In agreement with the conservationist and academic community, we highlight the need for better policy legislation to control human activities advancing on the remnants of native grassland and associated wetlands. These actions will undoubtedly benefit not only C. ornata populations but also the threatened biodiversity of the South American temperate grasslands.