The spatial overlap between ITs and bats across the entire Amazon biome was assessed based on a cross-sectional comparative analysis of two main spatial datasets (i.e., ITs and bat species distribution range data), which were analysed both under the biogeographic boundaries of Amazonia and for each specific country.

The extent and distribution of the ITs were obtained from the Rede Amazônica de Informação Socioambiental Georreferenciada (RAISG, 2019). Since this data has been publicly accessible for many years (with the consent of Indigenous Peoples), we do not foresee any ethical problems in including and presenting the geographical boundaries of the ITs in this article. On the contrary, we believe this might encourage researchers to further recognize the conservation values of these territories and increase the likelihood of engaging Indigenous Peoples in land and ecosystem management. Blank areas in our maps do not necessarily indicate an absence of Indigenous Peoples or their lands, but rather areas for which officially recognized Indigenous land tenure cannot be inferred based on publicly available geospatial data sourced from RAISG. As such, our analyses only include ITs that have been officially recognized following the standard categorization of RAISG (i.e. titled, homologated, demarcated and/or approved by decree).

Bat species distribution ranges were sourced from the IUCN Red List of Threatened Species distribution maps (IUCN, 2018). We considered a total of 223 species that had at least part of their distribution range within the Amazonian biome. All distributions were carefully reviewed and validated by three bat researchers with extensive expertise in Neotropical bat studies (A. L.-B., P. M. V. and R. R.) and, when needed, corrected using ArcGIS v. 10.3.1 (ESRI, 2013). All modifications were based on new species descriptions or taxonomical re-assessments, as well as on several bat distribution range expansions recently published in scientific literature (e.g., Mantilla-Meluk and Montenegro, 2016; López-Baucells et al., 2014, 2018; Velazco et al., 2017). This dataset is available from author R. R. upon request.

Geospatial data for each of the Amazonian country surface areas were sourced from the Global Administrative Areas spatial database (GADM, 2016). The extent of the Amazon biome was determined following its standard biogeographic definition, as defined and provided by RAISG (see a comprehensive description in RAISG, 2017). We stress that while RAISG mostly uses a hydrographic definition of the Amazon (including territories up to the watershed of the headwaters of Amazonian rivers), in this paper we restrict our analysis to the biogeographic boundaries of the Amazon, based on the extent of the Amazon biome (see Eva and Huber, 2005 for further details). We note that some of the figures presented in this paper might vary from the ones reported by RAISG, due to the differences in the definitions used.

For each bat species, we calculated the proportion of the distribution range that overlaps with ITs based on the spatial intersection of the different geographical datasets. We first compiled bat species distribution ranges (validated and corrected) for a total of 223 species into a single map, which we overlapped with the geographic distribution of ITs. The spatial overlay operation was carried out using the geo-algorithm “intersect” in ArcGIS v. 10.3.1 (ESRI, 2013), which enables the calculation of the surface of overlap between ITs and bat distribution ranges.

The first part of the analysis was performed at the biome-wide scale. Considering that around 25% of all the Amazon biome is covered by officially recognized ITs, we adopted this threshold (i.e., 25%) to identify bat species with a substantial section of their distribution range intercepting with ITs. As such, we identified those species with more than 25% of their global distribution range within Amazonian ITs across the entire biome. Additionally, for each of these selected species (those with >25% of their distribution overlapping with Amazonian ITs), we also calculated the percentage of its global distribution within the Amazon section of each of the nine countries (e.g., percentage within Peruvian Amazonia).

Furthermore, the effect of the ITs’ size, their average altitude and latitudinal differences on total bat diversity was modelled using a generalised linear model with Gaussian distribution. All three fixed factors were centred and scaled and effects plots were created using the allEffects function from the 'effects' package in R (Fox, 2003; Fox and Weisberg, 2019). We also created two histograms describing the rarity distribution of bat species across ITs and the distribution of richness in rare bat species across all ITs (see Supplementary Materials for further details). In the context of this study, we defined as rare those species occurring across less than 10% of the ITs of the Amazon biome.

The second part of the analysis was performed at the national scale. The rationale for evaluating national-level conservation values of ITs is that many conservation plans and targets are designed on the sole basis of species distribution range data at the national level (e.g., Montesino-Pouzols et al., 2014; Dallimer and Strange, 2015). As such, we calculated for each species the overlap between the national distribution and the national IT cover within Amazonia. This allowed us to identify those species with more than 25% of their national distribution range within Amazonian ITs in each of the nine countries analysed.

Based on their IUCN Red List status, bat species were classified as Threatened (i.e., species listed as Vulnerable, Endangered and Critically Endangered), Data Deficient (hereinafter DD) or Not Evaluated (hereinafter NE). We then calculated the total number of bat species within each category for each IT and each of the nine considered countries. In order to analyse and visualize how the proportion of Threatened vs DD species are distributed across the Amazon, we also created a bivariate map showing both types of conservation status (see Fig. S1 in the Supplementary Materials). We classified ITs into three classes using quantile classification – based on the distribution of number of threatened species. For Threatened species, we classified ITs as Low (one species), Medium (two species) and High (three to five species). For DD species, we classified ITs as Low (zero to three species), Medium (four to five species) and High (six to 11 species). We identified megadiverse ITs as those ITs with the highest number of bats with overlapping distribution ranges (i.e., first quartile, 103–137 species). Additionally, all bats were also classified into different functional guilds based on Kalko (1998), using the type of diet (i.e., insectivorous, frugivorous, nectarivorous, sanguivorous or carnivorous) and diet strategy (i.e., aerial and gleaning) as the separation criteria.

All the geospatial analyses were carried out in ArcGIS v. 10.3.1 (ESRI, 2013) under the World Cylindrical Equal Area coordinate system.

We found a total of 22 species from four different families and 17 genera that had more than 25% of their global distribution ranges overlapping with ITs (Table 1). Of these, overlapping proportions oscillated between 100% for Dashe’s nectar-feeding bat Hsunycteris dashe and 28.6% for the Carriker's round-eared bat Lophostoma carrikeri (see Fig. 1 for examples of these distribution ranges). Three of the species with over 25% of their global distribution ranges overlapping with ITs are classified as Threatened (i.e., one as Endangered and two as Vulnerable), seven are classified as DD species and four have not yet been evaluated. The remaining seven are currently listed as either Least Concern or Near Threatened (Table 1).

Fig. 1.

Distribution of Amazonian Indigenous Territories (ITs) classified according to the total number of species occurring within their boundaries. Shadowed areas in red correspond to officially recognized ITs. Distribution ranges of some bat species with >25% of overlap with ITs have been added to illustrate some of our results.

(0.72MB).

Gleaning animalivores (seven species) and frugivores (six species) of the family Phyllostomidae were the most represented of the bat species with over 25% of their distribution range overlapping with ITs (Table 1). Aerial insectivorous bats on the other hand were only represented by three species of molossids, the Guianan bonneted bat Eumops maurus, the Colombian bonneted bat Eumops trumbulli and the recently described Cynomops mastivus (Moras et al., 2016), one species of emballonurid, the Pale-winged dog-like bat Peropteryx pallidoptera, and the LaVal’s disk-winged bat Thyroptera lavali (family Thyropteridae). Some species such as Dashe’s nectar-feeding bat Hsunycteris dashe are known from only one type locality, which is inside of the IT Comunidad Nativa Matsés in northeastern Peru (Velazco et al., 2017).

Some ITs were found to harbour up to 137 different bat species (i.e., more than 50% of all the Amazonian bat species; Fig. 1). The 22 species with more than 25% of their global distribution within Amazonian ITs are predominantly present in Colombia (19 species), Brazil (18) and Peru (17), with up to 13 species being distributed across the three countries (Table S1 in the Supplementary Materials). In fact, some species had between 70 and 100% of their Amazonian distribution within a single country (e.g. Scleronycteris ega or Phyllostomus latifolius in Brazil, the three Lonchorhina spp in Colombia, or both Sturnira nana and Hsunycteris dashe in Peru).

Finally, our results also show that a large fraction of the bat species analysed (i.e., 32%) are in fact rare (occurring across less than 10% of the ITs of the Amazon biome), and that those territories richer in total number of species are also ranking high in terms of number of rare species (see Figs. S2 and S3 in the Supplementary Materials).

The number of species with more than 25% of their distribution range overlapping with ITs was consistently higher when conducting analyses at the national level (Table 2). Peru was the country with the highest number of bat species with over 25% of their national Amazonian distribution range overlapping with ITs (145), closely followed by Brazil (136), Colombia (134) and Ecuador (134). In contrast, the other five Amazonian countries (i.e., Bolivia, Guyana, French Guiana, Suriname and Venezuela) had fewer than 10 species with more than 25% of their Amazonian distribution range within ITs (Table 2). Here it is important to note that countries with the highest number of bat species on ITs (i.e., Peru, Brazil, Colombia and Ecuador) correspond to those with a highest percentage of officially recognized IT land cover (i.e., ranging from 25 to 55%), whereas countries with a lowest number of bat species overlapping with ITs match those with lower officially recognized IT land cover (i.e., 0–20%). At the national level, the number of Threatened and DD species with more than 25% of their national distribution range overlapping with ITs varied respectively between one (Ecuador, Guyana, French Guiana and Suriname), seven (Colombia), five (French Guiana) and twenty-one (Brazil).

At the biome-wide scale, most of the ITs that harbour the highest levels of bat diversity are found in the North-Northwest region of the Amazon, with ITs becoming gradually poorer following a latitudinal gradient towards the South (Fig. 2). Moreover, transboundary areas in the North-Northwest Amazon count with the highest density of ITs, as well as the ITs with the largest areas. Most of the species that were identified as having more than 25% of their global distribution ranges within ITs were generally found in the transboundary areas between Brazil, Colombia and Peru, in the region around Três Fronteiras and close to the Putumayo River (Fig. 2). Some of the ITs with the highest levels of bat diversity correspond to the Alto Río Negro Indigenous Territory (Brazil), the Yanomami Indigenous Territory (Brazil), the Raposa Serra do Sol (Brazil) and the Resguardo Predio Putumayo (Colombia).

Fig. 2.

Distribution of Amazonian Indigenous Territories classified according to A) the number of Threatened (T) bat species; B) the number of Data Deficient (DD) bat species; and C) the number of Not Evaluated (NE) bat species occurring within their boundaries. Shadowed areas in red correspond to recognised Indigenous Territories (ITs).

(0.87MB).

We also found that ITs at higher altitude had lower bat species richness than those in the lowlands (b = −0.068 ± 0.002; z = 28.59; p < 0.0001; see Fig. S4 in the Supplementary Materials). In addition, bat species richness in ITs was found to increase towards northern latitudes (b = 0.025 ± 0.002; z = 11.46; p < 0.0001; see Fig. S5 in the Supplementary Materials). However, IT size did not have any significant effect on bat species richness (b = 0.002 ± 0.002; z = 1.17; p = 0.242; see Table S2 and Fig. S6 in the Supplementary Materials).

The analysis of geospatial patterns for specific conservation categories showed clear differences between the distribution of Threatened, DD and NE species. The first ones have a relatively even distribution range across all Amazonian ITs, although they are remarkably prevalent in the ITs in Southeast Brazilian Amazonia. In contrast, DD and NE species tend to be more abundant in ITs within the North-western Amazon (Fig. 2). ITs located in two transboundary areas at Colombia-Brazil and Colombia-Ecuador frontiers harbour both high numbers of threatened and DD bat species (Fig. 2).

It is important to note that the numbers highlighted in this study are most likely conservative estimates. While we only used officially recognized ITs for our analysis, we know that many Indigenous lands remain unrecognized to this day. We therefore consider that the amount of biodiversity that de facto depends on Indigenous stewardship across the Amazon should be reasonably higher. Moreover, we know that bat research within Amazonian ITs has been meagre at best (e.g., Bernard et al., 2011; Delgado‐Jaramillo et al., 2020), especially for some guilds such as the aerial insectivorous bats, for which available information is rather poor in almost all the Amazon. The fact that research permits to sample bats –as well as other taxa– in many Amazonian countries do not include authorization to enter ITs (e.g., Bolivia, Brazil) adds to this challenge. The stringent legislation controlling access and activities within ITs has been pinpointed as a potential barrier to carry out conservation-related research in these areas, given the increased bureaucratic load entailed (Guedes dos Santos et al., 2015).

We acknowledge that the use of IUCN distribution maps has some limitations. IUCN distribution maps represent the known or inferred limits of a species’ distribution range as a minimum convex polygon shape. Therefore, the polygon represents those areas where certain bat species might occur, although this does not mean that the distribution of the species is even within the whole area, or that it does not expand beyond it. It is important to note that these distribution polygons are, in practice, positioned somewhere between the extent of occurrence and the true area of occupancy of the species (see Rondinini et al., 2006 and Gaston and Fuller, 2009). Even after our careful, expert revision of all the IUCN distribution maps, we note that our study findings should be treated with caution.

While some authors highlight that IUCN distribution data possibly underestimate the extent of occurrence and overestimate the true area of occupancy (see Montesino-Pouzols et al., 2014), others have observed that these data do not appear to be subject to high omission errors (Venter et al., 2014). For amphibians, for instance, 95% of the known occurrence of 4500 amphibian species fall within or immediately adjacent to their mapped distribution (Ficetola et al., 2014). Moreover, these maps have been shown to be relatively robust to commission errors (i.e., the species is mapped as present in locations where it is in fact not present). Venter et al. (2014) simulated commission errors in the IUCN data to see how it would influence their results, and they actually found out that their results were robust to randomly simulated commission errors on the maps.

Nonetheless, despite recognised data gaps and limited data accuracy for certain range-restricted species (particularly in some remote areas), the IUCN Red List of Threatened Species represents insofar the most frequently used and updated dataset on the distribution range of vertebrate species and their conservation status (Schipper and Chanson, 2008; Le Saout et al., 2013; Montesino-Pouzols et al., 2014). They have been extensively used in conservation research, particularly in global-level overviews at coarse-scale resolution (e.g., Strassburg et al., 2012; Davidson and Dulvy, 2017). In fact, their use is ubiquitous in analyses on the effectiveness of protected areas in representing biodiversity (Rodrigues et al., 2004; González-Maya et al., 2015; Klein et al., 2015) and incipient in studies assessing the biodiversity harboured in ITs (see Schuster et al., 2019). As such, it constitutes the most tenable geospatial resource to carry out exploratory analyses at coarse-scale resolution on the intersections between bat distribution ranges and ITs (see also Conenna et al., 2017), in the very same way in which they have been used in studies on protected area effectiveness (Jenkins et al., 2015; Klein et al., 2015).

Future studies on this vein could benefit from the use of species distribution models (SDM), predictive modelling, and/or species’ area of habitat data (e.g., Rondinini et al., 2011; Santini et al., 2019; Brooks et al., 2019). For instance, Delgado‐Jaramillo et al. (2020) have modelled the distribution of several Amazonian bat species using Maxent. Although we have opted not to use predictive modelling due to the scarce number of records of several of our target species, we emphasise that for some better known Amazonian bat species, predictive models could be used to identify the potential overlap of their distribution and ITs. Such an approach would greatly benefit from a curated database of bat distribution across the Amazon (see Muylaert et al., 2017 for an example of such a database for the Atlantic Forests of South America). As for SDMs, it is important to note that their limitations are substantial when applied to microendemics or to small numbers of occurrences, and when accurate corresponding covariates are not available (Wisz et al., 2008; Synes and Osborne, 2011; van Proosdij et al., 2015).