Introduction

With a diversity of 1470 recognized species1, 184 of which occur in Brazil2,3,4, the order Chiroptera is the second most diverse among mammals. The considerable taxonomic diversity is accompanied by a variety of habits, both in terms of feeding and the use of environments, whether for shelter or foraging5. Individuals of this taxon interact with a wide range of organisms, playing important ecological roles in forest ecosystems and in the reforestation of degraded environments6. They act as controllers of pest insects7, pollinators of plants with high economic, social, and ecological value8, seed dispersers9, of plants including of pioneer species10, and as biocontrol agents with significant economic value for agricultural activities, as highlighted in recent years11,12.

However, the degradation of natural areas due to deforestation for agricultural and livestock activities13,14, resulting in changes in climate and land use, mining activities causing the suppression and prevention of vegetation regeneration, alteration and fragmentation of habitats, and other disturbances caused by anthropogenic activities are intensifying processes of extinction, rapid modifications of species’ geographical and phenological distributions15. Similarly, knowledge gaps, including uncertainties about the true distribution of species, contribute to the incorrect identification of ecological and evolutionary processes, as well as the inefficient use of data for the development of conservation strategies16.

In this context, species distribution models (SDMs) have been employed as method to address a variety of issues related to biological conservation, such as: biogeographic analyses17; identification of priority areas for conservation18,19; conservation of rare or endangered species20,21 and analysis of knowledge gaps22. Additionally, the creation and establishment of Conservation Units (UC) are crucial, as the fundamental purpose of these spaces is to preserve the environment and its biodiversity, aiming to reduce the destruction of natural areas23,24,25. In Brazil, the UC are classified into two levels of protection: Full Protection Units (UPI) and Sustainable Use Units (UUS) by the National System of Conservation Units (SNUC)26. The former is designated for the basic objective of nature preservation, allowing only indirect use of natural resources, such as environmental education, ecotourism, and scientific research. The latter, on the other hand, reconciles nature conservation with the sustainable use of natural resources27.

Even though they are not considered UC, Brazilian environmental legislation defines a third level of natural area with restricted use, known as Indigenous Lands (TI), as outlined in the Federal Constitution of 198828. These are spaces traditionally occupied by indigenous peoples, where they permanently reside and carry out their productive activities, essential for the preservation of environmental resources and cultural production, according to their customs and traditions. Although not specifically designed for biodiversity conservation, they have a direct influence on the process, as indigenous peoples demand greater environmental integrity in their areas, consequently leading to lower deforestation rates29,30.

However, biodiversity knowledge within conservation areas in Brazil remains scarce, a problem attributed to the recent creation of these areas and their legislative tool (SNUC)31. In addition, the UC are often located with them often being located in areas of low economic interest or focused on specific taxonomic groups, thus not protecting rare species or those with limited distribution32. It is estimated that species with distributions restricted to 750 km2 for bats of the Cerrado are not adequately protected by Brazil’s current network of reserves33.

Thus, this study aims to assess the contribution of Brazilian UC Full Protection Units (UPI) and Sustainable Use Units (UUS), as well as Indigenous Lands (TI) to the protection of bat species occurring in Brazil and ecosystem services they provide. Additionally, this study provides maps with total species richness values and by trophic guilds34, as well as potential distribution maps of species classified as Data Deficient by the International Union for Conservation of Nature (IUCN).

Results

A total of 371,474 occurrence points of bats were obtained for the entire neotropical region. Following data cleaning and treatment, 36,556 points (Fig. 2, Table S1) were used for the modeling of 165 species out of 184 with occurrences in Brazil (Appendix 1). All models presented AUC and TSS values ​​greater than 0.9 and Sorensen above 0.8 (Table S2).

The contribution of the units to conservation.

The ANOVA revealed interaction between the Type of Protection and Biome, Type of Protection and Trophic Guild, and Type of Protection and Threat Level according to the IUCN (Table 1). Overall, it is observed that bats have a significant portion of their distribution area outside the designated protection units (Fig. 1A), with the Cerrado biome showing the highest average, with over 50% of the distribution area outside these protected areas (Fig. 1A). Regarding the UPI, low values (less of them 20%) of areas within these units are observed, with the Pampa and Pantanal biomes having the lowest values of areas protected by UPIs and by all other units, in addition to TIs (Fig. 1A).

Table 1 Factorial Analysis of Variance (ANOVA) with the percentage of distribution area (area %) by species according to the type of Protection (outside, UPI, UUS and TI), Biome (Amazon, Cerrado, Caatinga, Atlantic Forest, Pampas and Pantanal) and Trophic Guild (insectivores, carnivores, Nectarivores, omnivores and Hematophagous) and the percentage of distribution area by species according to the Type of Protection, Biome and degree of threat according to IUCN (DD, LC, EN and NT).
Full size table
Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Graph of the factorial Analysis of Variance (ANOVA) between (A) percentage of distribution area (Area %), Protection Type (Outside, Strict Protection Units (UPI), Sustainable Use Units (UUS), and Indigenous Lands (TI)), and Biome (Amazon, Cerrado, Caatinga, Atlantic Forest, Pampas, and Pantanal), (B) percentage of distribution area, Protection Type, and Trophic Guilds (Insectivores, Carnivores, Nectarivores, Omnivores, and Hematophages), and (C) (B) percentage of potential distribution area, Protection Type, and Trophic Guilds (Insectivores, Carnivores, Nectarivores, Omnivores, and Hematophages) and threat level according to the IUCN (DD, LC, EN, and NT). The central points are the means, and the vertical lines represent the 95% confidence interval.

For trophic guilds, insectivorous bats have approximately 50% of their distribution area outside conservation areas, with approximately 10% of the distribution areas included in UPIs, 20% in UUS, and 20% in TIs (Fig. 1B). This same pattern is observed in the other trophic guilds, with carnivores and hematophages being the guilds that occur mostly outside of protected areas (Fig. 1B).

For the analyses conducted for degree of threat according to the IUCN, species not evaluated (NE) hold 30% of their distribution area outside of protected areas, with approximately 10% in UPIs, UUS, and TIs (Fig. 1C). For species classified as Data Deficient (DD), it is estimated that 50% of their area is outside of protection, with only 10% included in UPIs, UUS, or TIs (Fig. 1C). This same pattern follows for species classified as Endangered (EN) and Near Threatened (NT), with 70% of suitable areas for NT species located outside of protected areas (Fig. 1C).

Species richness of bats

There is a greater richness of fruit-eating bat species in the northern part of the country, in the Amazon biome (Fig. S1). Meanwhile, hematophagous bats show higher richness in the northwest of the country, also in the Amazon biome, and in the southeastern coast of Brazil, in the Atlantic Forest biome (Fig. S2). Nectarivores, omnivores, insectivores, non-gleaning insectivores, and overall species richness show a clear pattern, with higher values observed across the Amazon and Atlantic Forest biomes. (Figs. S3, S4, S5, S6, and S8). The exception were the gleaning insectivores (Fig. S7), which showed the highest richness in the Amazon biome.

Regarding the IUCN degree of threat, it is noted that DD species (Fig. S9) are distributed throughout the Brazilian territory, although they are more conspicuous in the Amazon. The species classified as LC species (Fig. S10) are concentrated in the Amazon, with EN species (Fig. S11) being conspicuous in the Atlantic Forest and Cerrado, while NT species (Fig. S12) are distributed throughout the Atlantic Forest, Caatinga, and Cerrado, from the northeast coast to the southeast coast of the country. The potential distribution maps of the 21 species classified as DD, can be observed separately (Figs. S13 to S33).

Potential distribution maps of DD (Data Deficient) species

For the 21 species classified as Data Deficient (DD) with potential occurrence, some exhibit wide distribution, such as Aeroestes egregius (Peters, 1870) (Fig. S13) and Tonatia bidens (Spix, 1823) (Fig. S32), which have potential ranges throughout the entire Brazilian territory. Similarly, Eumops delticus (Thomas, 1923) (Fig. S19) shows a distribution spanning the Amazon, Cerrado, and Caatinga biomes. Conversely, other species display more restricted distributions, including Cynomops mastivus (Thomas, 1911) (Fig. S16), Diclidurus ingens (Hernández-Camacho, 1955) (Fig. S18), Lasiurus castaneus (Handley, 1960) (Fig. S22), Lonchorhina inusitata (Handley and Ochoa, 1997) (Fig. S23), Myotis simus (Thomas, 1901) (Fig. S25), Peropteryx trinitatis (Miller, 1899) (Fig. S26), Saccopteryx gymnura (Thomas, 1901) (Fig. S28), and Thyroptera lavali (Pine, 1993) (Fig. S30), all of which have potential distribution areas restricted to the northern region of Brazil, primarily in the Amazon.

The species Chiroderma doriae (Thomas, 1891) has its potential distribution primarily in the Midwest and Southeast regions of Brazil, encompassing areas of the Cerrado and Atlantic Forest (Fig. S14). Cynomops abrasus (Temminck, 1826) shows a potential distribution across the Amazon, Cerrado, and southern regions of Brazil, including the Brazilian Pampas (Fig. S15). The potential distribution of Cynomops planiceps (Peters, 1866) extends through the Amazon, Caatinga, Cerrado, and Atlantic Forest, following the Brazilian coast (Fig. S17). Eumops maurus (Thomas, 1901) is found mainly in the Amazon, with its range extending into the Midwest, particularly in the Brazilian Cerrado (Fig. S20). Histiotus velatus (I. Geoffroy, 1824) has a potential distribution in the Cerrado, Atlantic Forest, and Pampas regions of southern Brazil (Fig. S21). Meanwhile, Molossops neglectus (Williams and Genoways, 1980) potentially occurs mostly in the Atlantic Forest, with possible distribution in the Caatinga as well (Fig. S24).

The species Thyroptera wynneae (Velazco, Gregorin, Voss, and Simmons, 2014) has a potential distribution in the Amazon and Cerrado, extending into areas of the Atlantic Forest (Fig. S31). Similarly, Vampyressa pusilla (Wagner, 1843) is observed to have potential occurrence in the Amazon, Cerrado, and Atlantic Forest (Fig. S33). The species Rhogeessa hussoni (Genoways and Baker, 1996) and Thyroptera devivoi (Gregorin, Gonçalves, Lim, and Engstrom, 2006) also show potential distributions in the Cerrado, Amazon, and Caatinga regions (Figs. S27 and S29, respectively).

Discussion

The species richness of bats in Brazil, analyzed by guilds and degree of threat by the IUCN, indicates that biomes considered biodiversity hotspots, such as the Amazon and Atlantic Forest35, are also areas of diversity for bats30. However, it is estimated that only up to 20% of the species in these biomes are likely to be protected by the country’s conservation units. This context further jeopardizes the ecosystem services provided by bats, given that bats have their distribution area largely outside conservation areas. Consequently, this also hinders conservation strategies for these animals, as species classified as Data Deficient by the IUCN have a high estimated distribution outside protected area.

In general, the proportion of each biome covered by conservation units in Brazil is unevenly distributed. For instance, the Amazon biome has 28% of its area within conservation units, while other biomes such as the Atlantic Forest (9.5%), Caatinga (8.8%), Cerrado (8.3%), Pantanal (4.6%), and Pampas (3%)36. have significantly lower coverage. Although the Amazon is often highlighted for having the largest percentage of Full Protection Units (UPIs) compared to other biomes29, with the highest conservation area coverage in the country, studies show that, on average, only 20% of bat species in this region are effectively protected. These results are similar to recent studies have shown that fish and turtle species are also not adequately protected37,38. In the Cerrado, approximately 60% of bat occurrence areas are outside any protected area, also threatened by anthropogenic actions, mainly due to agribusiness expansion39, exacerbating the preservation of life for endemic and endangered species such as the pollinating bats Lonchophylla bokermanni (Taddei, Vizotto, and Sazima, 1983) and Lonchophylla dekeyseri (Taddei, Vizotto, and Sazima, 1983), pollinating bats.

High values and areas with the potential distribution of bats outside any protected area compromise the provision of ecosystem services offered by this taxon, especially to seed dispersing, pollinating, and insect pest-controlling bats. Both seed dispersal and pollination are the main ways in which bats contribute to ecosystem succession, facilitated by their tendency to defecate and spit seeds while flying, allowing for long-distance dispersal39. Genera of bats such as Carollia, Sturnira, and Artibeus are important in the seed rain of pioneer species such as Cecropia spp., Piper spp., Solanum spp., and Vismia spp., which are among the most abundant plants during the early stages of primary and secondary succession in the Neotropical region6,9,40,41. Similarly, 528 species of angiosperms use pollination services in about worldwide, with approximately 360 of these pollinated by individuals of the Phyllostomidae family in the Neotropical region6. The role of insectivorous bats in controlling agricultural pests is also recognized7,42. For example, the species Tadarida brasiliensis (I. Geoffroy, 1824) consumes a variety of arthropods, including agricultural pests such as the cotton bollworm (Helicoverpa zea) and the tobacco budworm (Heliothis virescens)43.

In terms of richness, studies already published indicate the occurrence of at least 146 species, distributed in nine families and 64 genera solely in the Amazon44, and in the Atlantic Forest, the species list compiled9 presents richness of seven families, 40 genera, and 59 species, with the majority belonging to the Phyllostomidae family. This can be explained by the vast distribution of the Amazon biome, which covers two-thirds of Brazil’s territory, along with the diversity of environments and phytophysiognomies that contribute to its high biodiversity. However, both the Amazon and the Atlantic Forest have been facing intense exploitation in recent years, driven by activities such as agriculture, logging, and soybean cultivation. This exploitation has been particularly severe in the Amazon since 201945, besides the fact that the Atlantic Forest holds only 13% of intact primary vegetation46,47. These factors are pointed out as main drivers of biodiversity loss48,49, including bats, especially for species dependent on vegetation cover or that respond to fragmentation9,50, leading to a reduction in shelter and food supply for the bat community.

Tropical regions harbor high biological diversity, but with limited knowledge of the real distribution of species (Wallacean Deficit)51. In this context, it becomes impossible to have a general protocol that is sufficient to represent all species. However, SDMs can play an important role in the conservation of Chiroptera species and in determining priority conservation areas through Systematic Conservation Planning52.

It is important to consider that part of the pattern we identified may stem from a sampling bias in the bat records available for Brazil. Consistent with other studies53,54, as observed here, we emphasize that investments in basic research, such as field data collection, are still necessary. This is especially true for the Amazon and Caatinga biomes, particularly when aimed at small, emerging research groups located outside state capitals. We also highlight the large number of species classified as Data Deficient (DD), which have a broad geographic distribution, likely due to insufficient sampling of this group in certain areas. Additionally, our findings further reinforce the importance of Indigenous Territories (ITs) in the conservation of species and the ecosystem services provided by bats, a pattern already observed in turtles55. Notably, the conservation of bat diversity has direct impacts on human health, as the decline in bat populations leads to a reduction in ecosystem services such as insect control (agricultural pests), which results in increased pesticide use and, consequently, higher infant mortality rates56.

Methods

Occurrence data

The taxonomic classification used in this study followed the Brazilian Society for the Study of Bats (SBEQ)57, with the inclusion of the species Choeroniscus godmani (Thomas, 1903) and the synonymization of the species Micronycteris homezorum (Pirlot, 1967) with Micronycteris minuta (Siles and Baker, 2020). Additionally, the species were classified according to the degree of threat presented by the International Union for Conservation of Nature (IUCN) (https://www.iucnredlist.org/): DD (Data Deficient); LC (Least Concern); NT (Near Threatened); VU (Vulnerable); EN (Endangered); CR (Critically Endangered); NE (Not Evaluated) and by trophic guild, namely: frugivores, nectarivores, sanguivores, omnivores, carnivores, insectivores (gleaners and non-gleaner)34. A total of 371,474 occurrence points of bats were obtained for the entire neotropical region. Coordinates obtained after cleaning the occurrence data were used to create Species Distribution Models (SDM) for all species with a minimum of five unique occurrence points (Table S01). Following data cleaning and treatment, 36,556 points (Fig. 2, Table S1) were used for the modeling of 165 species out of 184 with occurrences in Brazil, as models were created only for species with a minimum of five unique occurrence points (Table S1).

Fig. 2
Fig. 2The alternative text for this image may have been generated using AI.
Full size image

Occurrences points of bats with distribution across Brazil obtained for the Neotropical region. To create the figures, the Qgis software version 3.34.3 (https://qgis.org/pt_BR/site/) was used.

Occurrences records of bat species from 1900 to 2024 from the digital collections of GBIF (https://www.gbif.org/), SpeciesLink (https://specieslink.net/), and Map of Life (https://mol.org/), were used for the entire Neotropical region. Additionally, to obtain occurrence records from the literature, searches for articles were conducted in the digital databases ISI Web of Science (http://www.webofknowledge.com), Google Scholar (https://scholar.google.com.br/), Scopus (https://www.scopus.com), and Scientific Electronic Library Online (Scielo, http://www.scielo.org), with bat*, specie* list* and Chiroptera as keyword. After compiling the occurrence data, records of species with incorrect or missing coordinates, as well as duplicate records, were removed from the analyses. A comprehensive list of all articles used for point-of-occurrence data extraction has been compiled (Appendix 02).

Environmental variables.

We utilized 19 bioclimatic variables, with a resolution of 9.4 × 9.4 km, for the Neotropical region, obtained from the WorldClim database (http://www.worldclim.org/) and the digital elevation model (DEM). These variables include: Mean annual temperature; Monthly mean diurnal temperature range; Isothermality; Temperature seasonality; Maximum temperature of the warmest month; Minimum temperature of the coldest month; Annual temperature range; Mean temperature of the wettest quarter; Mean temperature of the warmest quarter; Mean temperature of the coldest quarter; Annual precipitation; Precipitation of the wettest month; Precipitation of the driest month; Precipitation seasonality; Precipitation of the driest quarter; Precipitation of the wettest quarter; Precipitation of the warmest quarter; Precipitation of the coldest quarter.

These variables are part of a dataset of monthly climate data sampled between 1970 and 2000, provided by the WorldClim 2.1 version58. This dataset is commonly used for Species Distribution Modeling (SDM) to predict the potential distribution of species59 and is often considered a better choice for studies aimed at conservation planning60, such as ours. To address multicollinearity among the variables, we conducted a Principal Component Analysis (PCA)61 and used the resulting eigenvalues as environmental variables. We then selected the principal components (axes) that accounted for 95%62 or more of the explained variance, using these axes as input variables for the model.

Algorithms

To create the Species Distribution Models (SDMs), four algorithms were employed: Maxent (MXE)63, Random Forest (RDF)64​, Support Vector Machine (SVM)65​, and Gaussian-Bayesian (GAU)​66. The final suitability maps produced by these algorithms were combined into an ensemble model to reduce uncertainties inherent in individual models67. To minimize uncertainties inherent to each model, we adopted this ensemble approach as the final model, ensuring a more robust and reliable prediction of species distributions60,68. The ensemble model represents the average suitability of models where the Jaccard threshold values60 exceeded the average thresholds calculated for each species68. The Jaccard threshold was applied to minimize omission and commission errors in the models60. In addition, spatial restrictions were applied to the models to prevent overprediction in the species distribution60,69.

Maps of richness and potential distribution

To achieve this, a binary occurrence map was created, where suitability values higher than the Jaccard threshold indicated species presence, and the map was partitioned into pixels with species occurrence and pixels without species occurrence. For partitioning the binary map, two methods were used: (i) for species with more than 30 occurrence points, the chessboard method was employed.(ii) for species with fewer than 30 points, a random selection of points was made, using 70% of the points for modeling and 30% for evaluation60. Subsequently, we apply an spatial restrictions, only pixels where the species was predicted and had occurrence records, or pixels near areas with both predictions and occurrence points, were retained in the species’ potential distribution map60. Since spatial restrictions generate more conservative maps, limiting occurrence areas to locations near or with species records, a second model was generated without these spatial restrictions. Thus, two models were produced: a more restrictive and conservative one (with spatial restrictions) and a less conservative model, which includes areas of environmental suitability without considering species presence. All procedures were carried out using the enmtml function, available in the ENMTL package70 for the R environmen71. All scripts used in this study are provided in the supplementary material (Appendix 03).

Refers to a type of map that visualizes the distribution of richness in a given area or population. The species richness of bats was calculated as the sum of all binary occurrence maps generated by the SDM procedure. These maps were aggregated to represent species richness in three categories: all species combined, species grouped by trophic guild, and species grouped by degree of threat according to the IUCN. Additionally, the occurrences of species classified as Data Deficient (DD) by the IUCN were spatially represented and presented separately. All maps were created using QGIS software (http://www.qgis.org/pt_PT/site/forusers/download.html). For this analysis, the omnivorous and carnivorous trophic guilds were analyzed together, while the insectivorous guild was analyzed in three ways: (i) all insectivores, regardless of foraging strategy, (ii) gleaning insectivores, and (iii) all insectivores excluding gleaners.

Identification of the contribution of each type of unit

To assess the contribution of Conservation Units (UCs) and Indigenous Territories (TIs) to the conservation of bat species, we generated a grid with a resolution of 0.05° covering the entire territory of Brazil. This grid was then overlaid onto the species distribution maps, from which we extracted presence and absence data for each species in the corresponding grid cells. Next, the grid cells were overlaid onto maps of UCs and TIs, and each cell was classified according to its overlap as Unprotected Areas (UPI), Sustainable Use Conservation Units (UUS), or Indigenous Territories (TI).

For a grid cell to be classified within a Conservation Unit (UC) or Indigenous Territory (TI), at least 75% of its area needed to fall within the boundaries of a UC or TI. Cells that did not meet this threshold or were entirely outside UCs and TIs were classified as “unprotected.” Additionally, each cell was classified based on its location within the historical distribution of Brazilian biomes (Amazon, Cerrado, Caatinga, Atlantic Forest, Pampas, and Pantanal). After applying these classifications, the percentage of each species’ distribution area falling within Unprotected Areas (UPI), Sustainable Use Conservation Units (UUS), Indigenous Territories (TI), and “unprotected” areas (outside of protected regions) was calculated.

With these data, a factorial Analysis of Variance (ANOVA) was conducted to analyze the percentage of distribution area per species as a function of the Type of Protection (Outside, UPI, UUS, and TI), with Biomes (Amazon, Cerrado, Caatinga, Atlantic Forest, Pampas, and Pantanal) and Trophic Guilds (insectivores, carnivores, nectarivores, omnivores, and hematophagous) used as covariates. Additionally, a second factorial ANOVA was performed to assess the percentage of distribution area per species based on the Type of Protection (Outside, UPI, UUS, and TI), using Biomes and IUCN threat level as covariates. Data on Conservation Units (UCs) and the historical distribution of Brazilian biomes were sourced from the Ministry of the Environment (MMA) (http://mapas.mma.gov.br/i3geo/datadownload.htm), and Indigenous Territories (TIs) data were obtained from the National Indian Foundation (FUNAI, http://www.funai.gov.br/index.php/shape).