Introduction
Cock-tailed Tyrant Alectrurus tricolor (Vieillot, 1816), is a small passerine bird belonging to the family Tyrannidae. This species is notable for its striking appearance, with males exhibiting a unique black-and-white plumage and a long, cocked tail that is characteristic of the species (Herzog et al. Reference Herzog, Terrill, Jahn, Remsen, Maillard and García-Solíz2017). Females, on the other hand, have a more subdued coloration, often brownish, which provides camouflage in their natural habitats (Smith et al. Reference Smith, Güller and del Castillo2017). Cock-tailed Tyrant is primarily a grassland species, making it an important indicator of the ecosystem quality in which it resides (Codesido and Fraga Reference Codesido and Fraga2009; Kanegae et al. Reference Kanegae, Levy and Freitas2012). The distribution of Cock-tailed Tyrant spans across parts of South America, and its range extends from the grasslands of Brazil, particularly in the Cerrado biome, into parts of Paraguay, northern Argentina, and Bolivia (Farnsworth et al. Reference Farnsworth, Langham, de Juana, del Hoyo, Elliott, Sargatal, Christie and de Juana2020). The species is typically found in open, grassy savannas and wetlands, preferring areas with tall grasses where it can nest and forage (López-Lanús et al. Reference López-Lanús, Di Giácomo, Azpiroz, Haynes, Galimberti, Keyel, Marino, Miñarro, Zaccagnini and López-Lanús2013b). However, its distribution is increasingly fragmented due to habitat loss, and it is considered “Vulnerable” by the International Union for Conservation of Nature (IUCN) (BirdLife International 2017). The conversion of grasslands into agricultural lands, forestation (Eucalyptus and Pinus), and the expansion of cattle ranching have severely reduced the available habitat for this species (BirdLife International 2017; Bencke et al. Reference Bencke, Dias, Fontana, Overbeck, De Patta Pillar, Müller and Bencke2024; López-Lanús et al. Reference López-Lanús, Di Giácomo, Azpiroz, Haynes, Galimberti, Keyel, Marino, Miñarro, Zaccagnini and López-Lanús2013b).
Understanding the distribution range of a species is crucial for conservation planning, biodiversity monitoring, ecological research, assessing climate change impacts, managing invasive species, and studying evolutionary processes (Mota-Vargas and Rojas-Soto Reference Mota-Vargas and Rojas-Soto2012). Knowing where a species is found helps to identify critical habitats for protection, monitor population changes, and understand ecological interactions (Lamoreux et al. Reference Lamoreux, Morrison, Ricketts, Olson, Dinerstein and McKnight2006). It also aids in predicting shifts due to climate change, managing invasive species, and revealing historical biogeographical events (Powers and Jetz Reference Powers and Jetz2019; Tan et al. Reference Tan, Ferguson and Yang2024). This comprehensive knowledge is essential for effective conservation strategies and maintaining ecosystem health.
In this context, species distribution models (SDMs) are vital tools in biodiversity conservation for several reasons. They help predict the distribution of species based on environmental parameters, which is crucial for identifying suitable habitats and planning conservation actions (Fajardo et al. Reference Fajardo, Lessmann, Bonaccorso, Devenish and Muñoz2014). These models are particularly useful in assessing the impacts of climate change, guiding restoration efforts, and managing invasive species, guiding conservation strategies across various ecological realms (Rathore and Sharma Reference Rathore and Sharma2023). Additionally, SDMs link scientific knowledge to policy and decision-making processes, making them powerful tools for conveying the consequences of environmental changes to stakeholders (McShea Reference McShea2014). Also, by combining systematic conservation planning with SDMs it is possible to represent the geographical ranges of threatened species, helping to mitigate the effects of biased sampling and incomplete knowledge (Porfirio et al. Reference Porfirio, Harris, Lefroy, Hugh, Gould and Lee2014).
Ecologically, Cock-tailed Tyrant plays a key role in grassland ecosystems, where it feeds on a variety of insects and other invertebrates. Its foraging behaviour involves scanning the grasslands from perches and making short flights to capture prey. The species’ reliance on undisturbed grassland habitats makes it particularly sensitive to SDMs. The use of SDMs on Cock-tailed Tyrant was first implemented by Marini et al. (Reference Marini, Barbet-Massin, Lopes and Jiguet2013). The models were based on data from the literature and showed that the species appears to be non-migratory, as its breeding and winter niches are similar, with no evidence of latitudinal or altitudinal migration. The authors indicated that there may be climatic niche switching, as the predicted distributions for breeding and winter seasons do not accurately forecast occurrences in the opposite season, concluding that Cock-tailed Tyrant is likely resident throughout its range, and that the observed movements may reflect partial population migration, seasonal shifts in social behaviour or nomadism (Marini et al. Reference Marini, Barbet-Massin, Lopes and Jiguet2013). Our study expands on the work of Marini et al. (Reference Marini, Barbet-Massin, Lopes and Jiguet2013), who provided the first detailed analysis of the migratory status of Cock-tailed Tyrant using ecological niche modelling based on 141 occurrence records. While their results clarified that the species is largely non-migratory, we extended this perspective by incorporating a larger data set and by combining SDMs with land-cover and land-use data. This approach refined the understanding of where suitable habitat currently persists and highlights the degree to which habitat loss and fragmentation may affect the species. In doing so, our study complements the ecological insights of Marini et al. (Reference Marini, Barbet-Massin, Lopes and Jiguet2013) by adding a conservation-focused assessment that underscores the urgent need for targeted management actions.
Methods
We acquired occurrence data for Cock-tailed Tyrant from the Global Biodiversity Information Facility (GBIF) and records are available for download via DOI:10.15468/dl.t4at8n. The data set initially contained 1,644 records, of which 1,583 entries included valid geographical coordinates and were retained for analysis. We excluded records lacking coordinates, containing obvious georeferencing errors, or falling outside the known South American distribution. The temporal coverage of the data set spanned the late nineteenth century to 2024, but since our modelling relied on long-term climatic averages (WorldClim 1950–2000), the inclusion of older records remained valid for representing potential distribution. We also included recent records (post-2000) to ensure that contemporary occurrences were represented in the analysis and to assess whether the species continues to occupy areas predicted as suitable. In terms of comparability with the IUCN Red List map, the IUCN range polygon represents the species’ extant distribution irrespective of date and thus serves as a broad reference of known extent of occurrence. By contrast, our analysis refined this by highlighting suitable habitat within that extent using environmental variables and land-cover layers. Therefore, although our records spanned a wider temporal range, the comparison with the IUCN map remained valid because both approaches aim to represent the species’ overall geographical distribution, rather than abundance at specific time periods.
We classified observations according to the Western Hemisphere seasonal definitions, i.e. summer (December–February), autumn (March–May), winter (June–August), and spring (September– November), to analyse seasonality. This classification was included to test whether Cock-tailed Tyrant shows seasonal differences in its geographical distribution, since the species has historically been described as migratory or partially migratory in parts of its range. We plotted geographical distributions for each season by latitude and longitude to explore seasonal patterns in observations. Additionally, a heatmap of spatial distribution was produced in R (version 4.4.1, using the ggplot2 package) by dividing the data into geographical zones with 10-degree latitude and longitude bins, facilitating an examination of seasonality within each zone. A chi-square test was applied to evaluate if the seasonal occurrence distribution differed significantly from an even distribution.
We used MaxEnt (version 3.4.4) (Phillips et al. Reference Phillips, Anderson, Dudík, Schapire and Blair2017, Reference Phillips, Dudík and Schapire2024) to model potential distribution for Cock-tailed Tyrant. A total of 231 presence records were used for training, and 10,230 background points were drawn from the study region to define the available environment. We used the default MaxEnt settings for feature classes (linear, quadratic, hinge, and product) and logistic output. Regularisation multipliers followed program defaults (linear/quadratic/product = 0.05; categorical = 0.25; hinge = 0.5; threshold = 1.0). We evaluated the model performance using the Area Under the Curve (AUC) of the Receiver Operating Characteristic (ROC) (Fielding and Bell Reference Fielding and Bell1997). Model quality was further evaluated by examining omission rates at different thresholds. Cross-validation was carried out internally by MaxEnt through random partitioning of training and background data, and variable importance was assessed via percentage contribution and permutation importance scores.
Modelling was based on climatic predictors obtained from WorldClim (version 2.1) at 30 arc-second resolution (Fick and Hijmans Reference Fick and Hijmans2017). These layers represent long-term averages for temperature and precipitation variables calculated over the period 1950–2000 and are widely used as baseline climatic conditions in SDMs. The bioclimatic variables were directly used as input into MaxEnt without additional averaging, consisting of 18 continuous and one categorical variable, along with an altitude variable. To determine each environmental variable’s contribution to the species’ distribution, we calculated permutation importance, revealing the impact of each variable when its values were permuted. These variables were included because they capture both broad climatic gradients and environmental extremes relevant to grassland bird ecology. We did not apply a priori reduction based on multicollinearity, as MaxEnt is relatively robust to correlated predictors by constraining variable contributions during model fitting (Phillips et al. Reference Phillips, Anderson and Schapire2006). Instead, we relied on permutation importance and percentage contribution to interpret which variables most influenced the model.
We compared the predicted habitat polygons generated by the SDM to the occurrence records to identify vegetation structure with favourable conditions. Since Cock-tailed Tyrant is a grassland specialist (Azpiroz et al. Reference Azpiroz, Isacch, Dias, Di Giacomo, Fontana and Palarea2012), we overlaid the habitat polygons with 30-m land-use raster data from the Global Land Analysis & Discovery (GLAD) project (Hansen et al. Reference Hansen, Potapov, Pickens, Tyukavina, Hernandez Serna and Zalles2021). This allows the identification of natural vegetation structures within the species’ range and to refine the representation of the best scheme of suitable habitat available. Occurrence records falling within these polygons were examined to determine which vegetation categories coincided with known presence points, and categories corresponding to natural grasslands, savannas, and wetlands were considered favourable vegetation structures. In contrast, predicted areas dominated by croplands, plantations or urban cover were excluded when estimating the final extent of suitable habitat. To assess suitable habitat, we used land-cover and vegetation data from MapBiomas for Argentina (v. 1.0), Bolivia (v. 2.0), Brazil (v. 9.0), and Paraguay (v. 1.0) to classify the biomes and ecoregions containing Cock-tailed Tyrant. As MapBiomas classification levels, versions, and precision vary by country, it was used only as a secondary resource for identifying vegetation units, land-use change, and habitat loss.
We used QGIS (version 3.34.11) to perform spatial analyses, comparing polygons for Cock-tailed Tyrant to the geographical range published on the IUCN Red List website (BirdLife International 2017; map updated October 2016).
Results
Locality records of Cock-tailed Tyrant show three major groups of plots. A large section exclusively present in Brazil near the Atlantic coast, another group more in the centre of the distribution, which includes the State of Mato Grosso do Sul in Brazil, the eastern area of Paraguay, and some old (and undated) records in Misiones in Argentina, and a last group in Bolivia, in lowlands north-east of the Andes and separated >1,110 km from the central distribution group (Figure 1). Most of the observations (1,498) were recorded from the year 2000, between July and December, and the species was more frequently observed in spring and observed less in autumn (Figure 2).
Figure 1. Geographical occurrences of Cock-tailed Tyrant including all the records available in the Global Biodiversity Information Facility (GBIF).
Figure 2. Number of records of Cock-tailed Tyrant classified by year (left above), month (left below), and season (right).
Our analysis indicated that the distribution of records was consistent across seasons. Each geographical group contained records in all four seasons (Figure 3). Likewise, the heatmaps revealed similar geographical frequencies among seasons (Figure 4). The chi-square test indicated (χ² = 1.08, df = 3, P = 1.0) that the differences in occurrence counts across seasons are not statistically significant. This suggests that there is no strong evidence that Cock-tailed Tyrant is more frequently observed in any particular season based on the available data.
Figure 3. Scatter plots displaying the geographical distribution of Cock-tailed Tyrant occurrences for each season, based on latitude and longitude.
Figure 4. Heatmaps displaying the seasonal occurrence of Cock-tailed Tyrant by geographical region. The colour intensity represents the number of occurrences, with darker colours indicating higher frequencies.
The SDM showed some of the better predicted conditions where the highest frequency of records is present. The AUC value is 0.974 with which the model holds a strong support. As the WorldClim v2.1 layers represent long-term averages from 1950 to 2000, our model predictions should be interpreted as the species’ potential distribution under baseline climatic conditions, rather than year-to-year variability. From MaxEnt results, the variables that contributed the most to the model included: ecoregion (33.5% contribution), precipitation in different seasons (pre6190_l7, pre6190_l10, and pre6190_l1 with 14–15% contribution), and vapour pressure (vap6190_ann, 7.8% contribution). Permutation importance values show precipitation and vapour pressure as significant, with pre6190_ann having the highest permutation importance (19.8%), and thus probably highly influential in predicting the distribution of Cock-tailed Tyrant. (See Supplementary material Appendix S1 for the complete list.)
The minimum predicted conditions that include most of the new records (leaving two outside) was 0.25 (Figure 5). We tested different thresholds, and 0.25 was the minimum value that retained most recent occurrence records, while leaving outside only a few historical records (1938, 1947, 1977, and 1990) and a couple of observations without specific dates. The final suitable habitat polygons were derived from the 0.25 logistic threshold output, but with additional refinements to improve ecological realism. Specifically, we removed regions where the species has never been recorded (e.g. Uruguay, Peru, and small scattered polygons distant from core areas) and applied minor smoothing to eliminate artificial “zig-zag” edges created by raster resolution effects. These adjustments did not expand the range beyond the 0.25 threshold polygons; rather, they reduced overpredicted areas while slightly improving the fit to observational records (Figure 6). In this suitable habitat polygons were included two observations that were not included in the 0.25 predicted conditions by MaxEnt in Brazil (gbifID 4192282828, latitude – 19.881493, longitude – 56.25914) and Bolivia (gbifID 2229454277, latitude – 13.701913, longitude – 63.73282).
Figure 5. Species distribution model (SDM) for Cock-tailed Tyrant. The darker areas represent the better predicted conditions. The green polygon highlights the contour of 0.25. Note that some historical and “undated” (black arrows) records lie outside the 0.25 area.
Figure 6. Potential distribution for Cock-tailed Tyrant, compared with IUCN maps (above), and land cover/use (below). Strata identified with * indicate potential suitable habitat for Beni population in Bolivia (wetland bare ground), and Paraguay and Brazil (dense short vegetation). Land use extracted from Global Land Analysis & Discovery (GLAD).
The three resulting areas of predicted conditions at the 0.25 limit for Cock-tailed Tyrant sum 946,974 km2 (Figure 6). Within the predicted habitat polygons, land cover was highly heterogeneous. The largest share corresponded to dense short vegetation (38.9%), which includes natural grasslands and savannas suitable for Cock-tailed Tyrant. Other natural categories included wetland bare ground (13.1%), open tree cover (11.1%), dense tree cover (11.4%), wetland dense tree cover (3.4%), wetland open tree cover (2.4%), semi-arid formations (0.1%), and water-bodies (0.8%). Anthropogenic categories comprised a substantial proportion of the polygons, with cropland (13.1%), tree-cover gain (1.0%), and built-up areas (1.4%). Additionally, tree-cover loss (2.8%) and wetland-cover loss (0.3%) highlighted on-going land-use changes within the predicted range.
The area in Bolivia (144,908 km2) is the smallest and is separated from other areas by the dry ecosystems of the Dry Chaco and Bosques Chiquitanos (Figure 6). The other two areas (B and C) are separated by the Parana River. About 70% of area A comprises Beni savanna (suitable habitat identified as “wetland bare ground” in Figure 6) and includes areas of south-west Amazon moist forests. A large portion of the region has an equatorial climate with dry winters with monsoonal rainfall. Most of the region comprises tropical and subtropical grasslands, savannas, and shrublands. It also includes areas of tropical and subtropical moist broadleaf forests. Area B is mostly composed of southern cone Mesopotamian grasslands in the south and Cerrado in the north (suitable habitat for Cock-tailed Tyrant identified as “dense short vegetation” in Figure 6) and includes areas of the Alto Paraná Atlantic forests. Most of the region has an equatorial climate with dry winters and areas with a warm and temperate climate, characterised by high humidity and hot summers. It is part of the tropical and subtropical grasslands, savannas, and shrublands biome. Most of area C comprises Cerrado (suitable habitat for Cock-tailed Tyrant identified as “dense short vegetation” in Figure 6), with areas of Alto Paraná Atlantic forests. Much of the region has an equatorial climate with dry winters, and areas of warm and temperate climate with dry winters and hot summers. The area has tropical and subtropical grasslands, savannas, and shrublands, also including areas of tropical and subtropical moist broadleaf forests. Percentage values are presented in Table 1.
Table 1. Percentage of land cover according to the classification of Figure 6
The resulting areas of predicted conditions at the 0.25 limit match rather well with the IUCN extant distribution map (Figure 6), which has 873,995 km2, smaller than the area suggested by the SDM. These predicted areas have different land covers (Figure 6), of which not all are suitable for the presence of Cock-tailed Tyrant. Based on the suitable habitat for Bolivian populations (Beni savanna), and suitable natural grasslands in Paraguay and Brazil (Southern Cone Mesopotamian grasslands and Cerrado), the final suitable habitat estimated for Cock-tailed Tyrant is shown in Figure 7 and consists of 177,753 km2.
Figure 7. Final overlap of suitable habitat and grassland ecosystems estimated for Cock-tailed Tyrant. Raster available at https://doi.org/10.6084/m9.figshare.30070465.
Discussion
Our analysis provides a refined picture of the distribution of Cock-tailed Tyrant by combining SDM with current land-cover data, suggesting areas of suitable habitat that remain under present-day conditions. This approach does not attempt to reconstruct historical distributions or evaluate differences with expert-based range maps, but instead highlights how much of the species’ extent is currently under pressure from agricultural expansion, afforestation, and exotic pasture conversion. In this context, it is known that the Cerrado, one of the most biodiverse savanna regions in the world, has experienced extensive degradation over recent decades (Myers et al. Reference Myers, Mittermeier, Mittermeier, da Fonseca and Kent2000). Agricultural expansion, particularly for soybean production, afforestation, and exotic pasture plantations for cattle ranching, has led to the conversion of approximately half of the biome’s original area into farmland (Klink and Machado Reference Klink and Machado2005). As a result, species that depend on the savanna structure, including Cock-tailed Tyrant, face dwindling habitats and heightened vulnerability. Conservation efforts aimed at curbing afforestation, promoting sustainable agricultural practices, and restoring native vegetation are essential to mitigate further degradation and protect the Cerrado’s ecological integrity (Borges and Loyola Reference Borges and Loyola2020; Strassburg et al. Reference Strassburg, Brooks, Feltran-Barbieri, Iribarrem, Crouzeilles and Loyola2017).
This reduction aligns with documented patterns in other grassland-specialist species facing similar pressures, particularly those linked to the rapid expansion of agriculture, afforestation (i.e. Eucalyptus plantations), and cattle ranching using exotic grasses in South American ecosystems (Azpiroz et al. Reference Azpiroz, Isacch, Dias, Di Giacomo, Fontana and Palarea2012; López-Lanús et al. Reference López-Lanús, Aramburú, Narosky, Vergara, Vasquez and Bustamante2013a). These findings are consistent with IUCN’s designation of Cock-tailed Tyrant as Vulnerable, emphasising the species’ heightened sensitivity to habitat fragmentation and loss within savanna and grassland environments (BirdLife International 2017). When comparing our suitability habitat findings with the existing IUCN distribution map, we found a generally congruent overlap, albeit with some extensions in the predicted range. The suitability analysis indicated an area of 177,753 km² for Cock-tailed Tyrant.
The SDM results in this study, with an AUC value of 0.974, indicate robust predictive performance, largely influenced by precipitation and vapour pressure, as well as by ecoregion type. These variables reflect the species’ reliance on regions characterised by specific climatic regimes, consistent with the observations of Marini et al. (Reference Marini, Barbet-Massin, Lopes and Jiguet2013), who also found Cock-tailed Tyrant to be largely resident with minimal seasonal distributional shifts. Our data further affirm that this species exhibits negligible seasonal movement, aligning with Marini et al.’s conclusion that Cock-tailed Tyrant may engage in partial nomadism or localised shifts rather than long-distance migrations. It is important to highlight that Cock-tailed Tyrant might rely on certain levels of humidity or rainfall for its habitat, breeding or feeding patterns as suggested by four of the five contribution variables.
Regarding the estimation of suitable habitat for species, it must be reviewed periodically as new versions of MapBiomas and other land-cover data sets become available. This is essential because on-going land-use change continuously alters habitat availability, and outdated classifications can lead to overestimations of suitable areas. For example, some areas classified as natural grassland by MapBiomas are in fact dominated by Eucalyptus plantations (Figure 8). Such misclassifications are directly relevant to Cock-tailed Tyrant, as they may inflate estimates of available habitat and highlight the importance of updating habitat assessments with the most recent and accurate land-use information.
Figure 8. Maps illustrate inconsistencies between land-cover classification and actual land use in southern Paraguay. Red areas represent “suitable habitat” as classified by MapBiomas, corresponding to natural grassland categories. However, high-resolution satellite imagery (panels 1–3) reveals that these areas are in fact occupied by Eucalyptus plantations, not native grasslands. Yellow polygons delineate the extent of plantation blocks.
This limited area of optimal habitat reinforces the need for targeted conservation within these specific grassland and savanna regions, as emphasised by Codesido and Fraga (Reference Codesido and Fraga2009) and Brazeiro et al. (Reference Brazeiro, Achkar, Toranza and Bartesaghi2020). Habitat restoration within these areas, coupled with stringent land-use policies, could help to mitigate the on-going habitat degradation that threatens Cock-tailed Tyrant and similar species across South America’s grassland biomes.
Given that grassland birds often serve as indicators of ecosystem health due to their sensitivity to habitat changes (Brazeiro et al. Reference Brazeiro, Achkar, Toranza and Bartesaghi2020), our findings have broader implications. Our analysis, combining SDM with vegetation cover and land-use data, provides a precise identification of suitable habitat that is directly relevant for conservation planning. Although we do not project future scenarios, this approach highlights areas where Cock-tailed Tyrant is most vulnerable to decline because suitable grasslands already overlap with intense land-use pressures, such as conversion to cropland and expanding tree plantations (e.g. Eucalyptus). By identifying where high-quality habitat persists and where it is under greatest threat, the results can guide immediate conservation priorities and management interventions. Conservation measures for Cock-tailed Tyrant and similar species may benefit from incorporating SDM to identify critical habitats, public policies, guide restoration efforts, and international environment standards (e.g. Verified Carbon Standards, Forest Stewardship Council, etc.). Additionally, policies promoting sustainable productive practices and the protection of native grasslands are essential to maintaining viable populations of Cock-tailed Tyrant and other grassland-dependent birds. Implementing sustainable cattle ranching practices on natural grasslands is the only productive activity that supports and maintains the ecological health and biodiversity of grassland ecosystems, providing benefits for the grassland structure and the fauna that depend on it, and giving an economic input to landowners (Parera and Carriquiry Reference Parera and Carriquiry2014).
As with any SDM approach, our analysis has some limitations that should be considered when interpreting the results. MaxEnt relies on presence-only data, which can be affected by spatial sampling bias and uneven survey effort across the species’ range (Yackulic et al. Reference Yackulic, Chandler, Zipkin, Royle, Nichols and Campbell Grant2013). Although MaxEnt is generally robust to these issues, presence-only models do not incorporate true absence information, which may lead to overprediction in poorly sampled areas. Another caveat is that the climatic predictors used (WorldClim averages for 1950–2000) represent baseline conditions and may not fully reflect recent or on-going climate shifts. In addition, we did not explicitly reduce multicollinearity among predictors, which can inflate the importance of correlated variables; we addressed this by interpreting model results primarily through permutation importance rather than raw contributions. Finally, our integration of land-cover data relies on available classification schemes (GLAD and MapBiomas), which may include uncertainties or mismatches with current land use at fine spatial scales, as shown in Figure 8. Despite these warnings, we consider our combined approach appropriate for identifying broad patterns of habitat suitability and for informing conservation necessities for Cock-tailed Tyrant.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0959270925100269.
Acknowledgements
The authors want to thank the staff of Guyra Paraguay for their support and friendliness, and Edder Ortiz and Glayson Bencke for advice at different stages of the manuscript. This work was possible thanks to the financial support of BirdLife International through the Small Grants Program for Grasslands Conservation, second edition (CV-015-2023). PC thanks the Consejo Nacional de Ciencia y Tecnología (CONACYT) for financial aid through the Sistema Nacional de Investigadores Program.
Open access
