Introduction
Recent global assessments have analyzed and synthesized data showing the staggering loss of biodiversity and ecosystem deterioration due to human actions (Díaz et al. Reference Díaz, Settele, Brondízio, Ngo, Agard, Arneth, Balvanera, Brauman, Butchart, Chan and Garibaldi2019; Secretariat of the Convention on Biological Diversity 2020; WWF Reference Almond, Grooten, Juffe-Bignoli and Petersen2022). Latin America has shown the greatest regional decline in mean vertebrate population abundance, at 94% between 1970 and 2018 (WWF Reference Almond, Grooten, Juffe-Bignoli and Petersen2022). The majority of globally-threatened species are declining due to land use change (Díaz et al. Reference Díaz, Settele, Brondízio, Ngo, Agard, Arneth, Balvanera, Brauman, Butchart, Chan and Garibaldi2019; Harfoot et al. Reference Harfoot, Johnston, Balmford, Burgess, Butchart, Dias, Hazin, Hilton-Taylor, Hoffmann, Isaac and Iversen2021; IUCN 2025; Jaureguiberry et al. Reference Jaureguiberry, Titeux, Wiemers, Bowler, Coscieme, Golden, Guerra, Jacob, Takahashi, Settele and Díaz2022), and protecting terrestrial biodiversity and preventing extinctions therefore requires protection and management of habitat. Numerous conservation organizations and global initiatives are working towards this goal. Target 3 of the Kunming-Montreal Global Biodiversity Framework (KMGBF) aims to “Ensure and enable that by 2030 at least 30 per cent of terrestrial, inland water, and of coastal and marine areas, especially areas of particular importance for biodiversity and ecosystem functions and services, are effectively conserved and managed,” commonly known as 30x30 initiative (Kunming-Montreal Global Biodiversity Framework; CBD 2022), and many governments and private organizations are working towards fulfilling this target, as well as other national and local goals for habitat protection.
Identifying biodiversity conservation priority areas to prevent species extinction is an important step needed to inform strategic investment in conservation. Key Biodiversity Areas (KBAs, sites of significance for the global persistence of biodiversity; IUCN 2016), including Alliance for Zero Extinction sites (AZE, sites supporting the overwhelming majority of the remaining population of one or more Critically Endangered or Endangered species; Alliance for Zero Extinction 2013) have been identified for this purpose. Efforts are underway to update and increase the comprehensiveness of these site networks (KBA Partnership 2023). Conserving key sites is a critical tool to help deliver Target 4 of the KMGBF on undertaking urgent management actions to prevent species extinctions (Bolam et al. Reference Bolam, Ahumada, Akçakaya, Brooks, Elliott, Hoban, Mair, Mallon, McGowan, Raimondo and Rodríguez2023; Boyd et al. Reference Boyd, Brooks, Butchart, Edgar, Da Fonseca, Hawkins, Hoffmann, Sechrest, Stuart and Van Dijk2008). By understanding the distribution of specific species and their overlap with existing protected areas, conservation practitioners can pinpoint which highly threatened species are under-protected and where to invest to fill the gaps in the existing protected area network to prevent species extinctions.
Birds are the best known of any class of vertebrates, and previous studies have overlaid the distributions of bird species with protected area data to determine the proportion of each species’ range that is already protected (Butchart et al. Reference Butchart, Clarke, Smith, Sykes, Scharlemann, Harfoot, Buchanan, Angulo, Balmford, Bertzky and Brooks2015; Venter et al. Reference Venter, Fuller, Segan, Carwardine, Brooks, Butchart, Di Marco, Iwamura, Joseph, O’Grady and Possingham2014). However, these global assessments have used the species’ mapped ranges, rather than the area of habitat (AOH) within them, which provides a finer-resolution indication of species’ potential distributions (Beresford et al. Reference Beresford, Buchanan, Donald, Butchart, Fishpool and Rondinini2011; Brooks et al. Reference Brooks, Pimm, Akçakaya, Buchanan, Butchart, Foden, Hilton-Taylor, Hoffmann, Jenkins, Joppa and Li2019). The main sources of range maps for this purpose are the International Union for Conservation of Nature Red List of Threatened Species’ (the “Red List”) distribution maps, produced by BirdLife International in its role as the IUCN Red List Authority for birds (hereafter Red List range maps). Recent efforts to improve these maps use high-resolution remote sensing and citizen science data (e.g., Huang et al. Reference Huang, Medina, Brooks, Butchart, Fitzpatrick, Hermes, Jenkins, Johnston, Lebbin, Li and Ocampo-Peñuela2021; Palacio et al. Reference Palacio, Negret, Velásquez-Tibatá and Jacobson2021). Such citizen science data allow the refinement of bird distribution maps, including the creation of AOH maps using various data sources and methods (Beresford et al. Reference Beresford, Buchanan, Donald, Butchart, Fishpool and Rondinini2011; Huang et al. Reference Huang, Medina, Brooks, Butchart, Fitzpatrick, Hermes, Jenkins, Johnston, Lebbin, Li and Ocampo-Peñuela2021; Lumbierres et al. Reference Lumbierres, Dahal, Soria, Di Marco, Butchart, Donald and Rondinini2022; Medina et al. Reference Medina, Pimm and Huang2024; Ocampo-Peñuela et al. Reference Ocampo-Peñuela, Jenkins, Vijay, Li and Pimm2016; Palacio et al. Reference Palacio, Negret, Velásquez-Tibatá and Jacobson2021; Warudkar et al. Reference Warudkar, Goyal, Kher, Vinay, Chanda, Bandi, Jenkins, Robin and Pimm2022).
Some recent range-mapping efforts have relied on the use of occurrence data (i.e., point localities from museum specimen records and/or citizen science projects such as eBird, iNaturalist, etc., often through the Global Biodiversity Information Facility data infrastructure network) to update or improve the existing Red List range maps (Huang et al. Reference Huang, Medina, Brooks, Butchart, Fitzpatrick, Hermes, Jenkins, Johnston, Lebbin, Li and Ocampo-Peñuela2021; Warudkar et al. Reference Warudkar, Goyal, Kher, Vinay, Chanda, Bandi, Jenkins, Robin and Pimm2022) or to derive new ranges based on species’ extent of occurrence (EOO; Palacio et al. Reference Palacio, Negret, Velásquez-Tibatá and Jacobson2021). However, most of these efforts have focused only on subsets of species, such as forest birds. We assessed these approaches to standardize the process across all species and habitats and determine the most appropriate approach for threatened birds of the Americas.
Given the issues and lessons learned from these efforts, we aimed to systematize the generation of AOH maps with minimal introduction of errors. Our approach includes some of the features of the other methods, tempered with expert opinion. Here we use publicly available bird occurrence data to develop a method to map the Area of Habitat (AOH), of 129 terrestrial land birds in Latin America, and overlay these with existing protected areas as compiled by the World Database of Protected Areas (WDPA, UNEP-WCMC and IUCN 2025), plus additional private protected areas not gazetted. We aim to quantify how much habitat is protected and unprotected, identify which species are under-protected, set habitat protection goals for under-protected species, and map areas where additional habitat protection projects could be developed and implemented to fill in the gaps of existing protected areas and reduce the extinction risk for these species.
Methods
Species data
We focused on species of land birds from Latin America (excluding the Caribbean and other outlying islands) listed in the Red List categories “Critically Endangered” (CR), “Endangered” (EN), and “Vulnerable” (VU D1 and D2; IUCN 2022) as of January 2026. We included only VU species listed under Red List criterion D (i.e., D1, and D2; see IUCN 2012 for definition of criteria) because these are species that have populations of <1,000 mature individuals and/or have very restricted distributions and can become CR or Extinct in a very short period of time. We also excluded seven species that are considered “lost” (i.e., they have not been reported in at least ten years; https://searchforlostbirds.org/), and Great-billed Seed-finch Sporophila maximiliani, because its taxonomy is currently so unclear that its occurrence and distribution cannot be properly determined. After applying these filters, the final set included 129 threatened bird species.
We obtained point locality data for each of the 129 selected bird species from eBird (eBird Basic Dataset. Version: EBD_relDec-2022. Cornell Lab of Ornithology, Ithaca, New York. Dec 2022, accessed: February 14, 2023). We filtered the eBird data to occurrences from the year 2000 onward to have only recent data and likely extant populations; and only those using the “Traveling” Protocol Type with a distance covered by the observer of <5 km, with the Protocol Type of “Stationary,” or with the Protocol Type “Incidental.” This limits the data to recent records, likely to have the most precise location information.
Habitat, elevation, land conversion pressure, and protected area data
We obtained land cover data from the European Space Agency – Climate Change Initiative (ESA-CCI) Land Cover map for the year 2020 (v. 2.1.1.; http://maps.elie.ucl.ac.be/CCI/viewer/index.php; accessed on May 6, 2023). The layer consists of 22 primary classes and 15 sub-classes of land cover at 300 m spatial resolution (https://maps.elie.ucl.ac.be/CCI/viewer/download/CCI-LC_Maps_Legend.pdf). Moreover, given the challenge that represents accurately mapping and differentiating natural and semi-natural grasslands versus cultivated areas, we used two recently developed layers to account for areas that are croplands: the Global Cropland-Extent Product (GCEP30, Thenkabail et al. Reference Thenkabail, Teluguntla, Xiong, Oliphant, Congalton, Ozdogan, Gumma, Tilton, Giri, Milesi and Phalke2021) and the Global Grassland Class and Extent maps (Parente et al. Reference Parente, Sloat, Mesquita, Consoli, Stanimirova, Hengl, Bonannella, Teles, Wheeler, Hunter and Ehrmann2024). The native resolution of both these layers is 30 m, so they were resampled to 300 m using nearest neighbor interpolation (best suited for categorical or discrete data) to use the same resolution across all layers.
We used elevation data from the Shuttle Radar Topography Mission (SRTM 4.1) digital elevation model layer (downloaded from https://srtm.csi.cgiar.org/srtmdata/; accessed on June 16, 2023; 250 m resolution). The elevation layer was resampled to 300 m using bilinear interpolation (recommended for continuous data) to use the same spatial resolution as the habitat layer.
To account for local habitat degradation in our prioritization analysis, we used the Conversion Pressure Index (CPI) layer developed by Oakleaf et al. (Reference Oakleaf, Kennedy, Wolff, Terasaki Hart, Ellis, Theobald, Fariss, Burkart and Kiesecker2024). The CPI is a global map of land conversion that combines past rates of anthropogenic pressure (derived from multiple drivers) with suitability maps for potential future expansion by large-scale development projected up to 2030 (Oakleaf et al. Reference Oakleaf, Kennedy, Wolff, Terasaki Hart, Ellis, Theobald, Fariss, Burkart and Kiesecker2024). CPI values range from 0, which indicates no conversion pressure, to 1, which represents the highest conversion pressure. According to the authors, this information can be used to identify areas for proactive conservation to ensure national commitments under the KMGBF. The CPI layer was resampled to 300 m using bilinear interpolation to use the same spatial resolution as the other layers.
We downloaded protected area data from the World Database of Protected Areas (WDPA; UNEP-WCMC and IUCN 2025; www.protectedplanet.net; accessed on May 15, 2025). The WDPA includes different protected area management categories developed by the IUCN, but we included all categories in our analysis as the reporting of categories and their enforcement is widely variable among countries. We also added several private protected areas and/or protected area expansions that are owned and managed by American Bird Conservancy (ABC) partners, but which are not yet recognized in the WDPA because they have been established recently. We merged the overlapping protected area polygons into a single polygon to avoid overestimation of the protected area coverage. We did not include Other Effective area-based Conservation Measures (OECMs; CBD 2018) as few Latin American countries have recognized any sites as OECMs yet. Colombia is the only country to have done so to date, and the effectiveness of these sites has not yet been assessed.
Developing Area of Habitat (AOH) maps
The method for estimating AOH involves selecting the available suitable habitat(s) and elevation range within the species’ mapped geographic range (Brooks et al. Reference Brooks, Pimm, Akçakaya, Buchanan, Butchart, Foden, Hilton-Taylor, Hoffmann, Jenkins, Joppa and Li2019). The geographic range is defined as the “current known limits of distribution of a species, accounting for all known, inferred or projected sites of occurrence” (IUCN 2016) and is depicted as a polygon map in the IUCN Red List (Red List range maps). These maps are derived from a variety of sources, including specimen localities obtained from museum data, observer records documented in Red Data Books, published literature, survey reports and other unpublished sources; records of species documented in Important Bird and Biodiversity Area assessments, distribution atlases derived from systematic surveys, distribution maps in field guides and other handbooks, and expert opinion (Buchanan et al. Reference Buchanan, Donald and Butchart2011).
While eBird and other citizen science data have been used in some cases to refine IUCN Red List range maps, this has not been done systematically across all species. As a result, many available range maps are outdated, as they are typically revised only during Red List reassessments. In our dataset, nearly half of the species had Red List assessments at least five years old, and their range maps are therefore likely to be of similar age. To address this limitation, we overlaid the existing Red List range maps with eBird locality data and used these combined inputs to propose refinements to species’ distributions for Area of Habitat (AOH) generation (Figure 1). In cases where the Red List range adequately encompassed the eBird records, no modification was applied.
Example map, for Chestnut-capped Piha Lipaugus weberi in western Colombia. The map shows the eBird localities (turquoise dots), the original IUCN/BirdLife range map (showing the Extant and Possibly Extinct portions of the range as green and red polygons respectively), the new range created in this study (blue polygons), and the AOH (purple) generated within the new range map by selecting the species’ suitable habitat and elevation range.
Figure 1. Long description
A topographical relief map of a mountainous region in Colombia featuring several data layers.
* Background: A grayscale shaded relief map showing rugged terrain with labels for towns including Valdivia, Briceño, Yarumal, Campamento, Angostura, Guadalupe, Carolina del Príncipe, Gómez Plata, Anorí, Amalfi, and Yalí.
* I U C N / BirdLife range map: Large red-outlined polygons labeled Possibly Extinct cover the central and southern portions of the map. Small green-outlined polygons labeled Extant are located in the North near Anorí and East near Amalfi.
* Expert redrawn range map: Three distinct blue-outlined polygons. One horizontal polygon in the Northwest, one large vertical polygon in the North-Central region near Anorí, and one rounded polygon in the Southeast near Amalfi.
* A O H (Area of Habitat): Solid purple shaded regions located strictly within the blue expert-redrawn boundaries, representing suitable habitat for Lipaugus weberi.
* eBird localities: Numerous turquoise dots clustered primarily within the purple A O H zones, particularly concentrated around the towns of Anorí and Amalfi.
* Scale and Orientation: A North arrow and a 5 kilometer scale bar are located in the bottom-left quadrant.
All range maps were subsequently reviewed by David Wiedenfeld to ensure internal consistency and ecological plausibility. The review primarily focused on correcting clear omission errors that were strongly supported by occurrence data and, to a lesser extent, on identifying outlier records of questionable origin or likely georeferencing errors, as well as potential commission errors (i.e., areas unlikely to be occupied given species’ habitat preferences, ecology, and biogeographic context). Importantly, range modifications were driven primarily by the spatial pattern of eBird records rather than by expert opinion alone.
To determine the elevation range of the species, we used the range of elevations from the existing eBird data points, excluding the highest and lowest 1% to reduce the impact of outliers. Once the new ranges were delimited, we overlaid the point locality data with the habitat (land cover) layer to extract the corresponding values for each point. We checked the habitat classes against the habitat coding and descriptions in the IUCN Red List assessments and specific publications and excluded those that were not listed or mentioned there. Also, because the land cover map may misclassify some pixels, the land cover classes for all presence data were inspected visually and those that were obvious mistakes were excluded (a common mistake in satellite image classification is classifying pastures or certain crops as natural grasslands). Based on this information, we proceeded to derive AOH maps by intersecting the elevation and habitat layers within the new range map and removing the pixels of unsuitable habitats (including the cropland layers) and outside the elevation range. All analyses and maps were done with QGIS (ver. 3.44; www.qgis.org).
Minimum conservation targets and gap analysis
To guide future avian conservation priorities in the Americas, we defined two minimum conservation targets (i.e., the minimum proportion of a species’ AOH that should be included in protected areas to be considered adequately represented and safeguarded) for protected area coverage for each focal species. The first area-based target estimated the proportion of the AOH needed to be represented in protected areas in order to sustain the conservation of those species. The second population-based target estimated the area needed to protect 1,000 mature individuals of each species (or the maximum population size for those species whose estimated maximum population is < 1,000 mature individuals) to prevent imminent extinction of these species.
The minimum conservation target based on the proportion of AOH was set to 100% for species with AOH ≤50 km2, whereas for species with AOH ≥20,000 km2, the minimum conservation target was set to 4% of the range. These values and cutoffs were selected to ensure that the entire population is protected for species with small ranges and to ensure an area adequate for conservation to protect species with large ranges. For all other species (in the range of >50 km2 and <20,000 km2), the target was interpolated using a linear regression on the log-transformed AOH (Rodrigues et al. Reference Rodrigues, Akcakaya, Andelman, Bakarr, Boitani, Brooks, Chanson, Fishpool, Da Fonseca, Gaston and Hoffmann2004).
The population-based target to protect was selected using the population size threshold for a species to avoid qualifying for the Vulnerable category under criterion D1 of the IUCN Red List of Threatened Species (IUCN 2012), which is 1,000 mature individuals. To calculate the area that is needed to include 1,000 mature individuals, we estimated the territory sizes or area requirements for each species in our list as described in Table 1. Estimates were obtained from the literature (see Supplementary material). Whenever there was no direct information for the species in our list, we used data from congeneric species of similar sizes and habits. These values are rough estimates, because very few of the rarest birds, that is, most species in our analysis, have published estimates of territory size. In addition, “territory size” is not applicable to species such as parrots, which do not defend a territory but rather range daily over significant distances in flocks. We used estimates from congeners or related species where possible and estimates of foraging ranges to develop these numbers.
Territory size estimates for birds, by size class. Estimates are per individual; therefore, the area must be doubled for a pair or multiplied by average flock size for species such as parrots that travel in flocks
Table 1. Long description
The table consists of four columns: Group, Size class, Habitat, and Territory size. Data is organized by bird group:
* Cracids: Large (curassow) at 50.0 ha; Medium (guan, chachalaca) at 20.0 ha.
* Columbids: Pigeon at 10.0 ha; Dove at 5.0 ha.
* Hummingbirds: All at 4.0 ha.
* Raptors: Large (hawk, falcon) at 100 ha; Medium (Sharp-shinned Hawk) at 20.0 ha; Small (pygmy owl) at 10.0 ha.
* Parrots: Large (Amazons, macaws) at 44 ha; Medium (Pyrrhura) at 22 ha; Small (Forpus) at 11 ha.
* Passerines (Small): Forest habitat at 2.0 ha; Open habitat at 3.0 ha; Grassland habitat at 5.0 ha.
* Passerines (Large): Forest (piha) at 5.0 ha; Open (Cinclodes) at 4.0 ha.
To assess the level of protection of each species, we intersected the individual AOH maps with the World Database of Protected Areas layer and estimated the area and proportion within protected areas. We were then able to assess the conservation gap (measured in square kilometers) for each focal species and considering the two conservation targets for each species.
To find the main priority areas for conservation, we ran the Zonation 5 v2.4 software (Moilanen et al. Reference Moilanen, Lehtinen, Kohonen, Jalkanen, Virtanen and Kujala2022). Zonation is spatial prioritization software for conservation planning that identifies which locations in a landscape are more important for retaining biodiversity. Given the resolution of our data and the number of pixels involved (over 450 million of total number pixels for the entire area, and over 1.3 million of pixels with data across all features), it was not possible to use prioritizr (Hanson et al. Reference Hanson, Schuster, Strimas‐Mackey, Morrell, Edwards, Arcese, Bennett and Possingham2025) or other integer linear optimization tool on a regular computer with 32 Gb of RAM. The advantage of Zonation (versus other prioritization programs) is that it can be used with high-resolution data on a continental scale on a modest computer system. The main output is a priority rank map where grid cells are ranked from highest to lowest priority in terms of conservation value. Therefore, the user can select the top X% ranked cells of the entire study area, depending on the objective of the prioritization. Because the new version of Zonation does not include a target-based planning algorithm, instead of selecting a predetermined percentage of ranked cells for all species, we selected the percentage of area represented for each individual species, using their minimum conservation targets as a cut-off criterion. For instance, if the minimum conservation target of a particular species is 27% of its AOH, we selected the top X% ranked cells that included 27% of that species’ AOH, and so forth for all priority species. As part of the prioritization with Zonation, it is possible to assign a weight to each conservation feature by assigning a numerical value proportional to the importance of the layer. For instance, different weights can be assigned to species depending on their Red List Threatened category. Here, we used weights from 0 to 4 to account for the area-based conservation target for each species, assigning 0 to species that had their conservation target met, and weighing the other species from 1 to 4 depending on the proportion of their AOH needed to reach their targets. We used the CAZPW (Core Area Zonation with Positive and Negative Weights) marginal loss rule which prioritizes areas that are core habitats for features (such as species or habitats), considering both positively and negatively weighted layers (Moilanen et al. Reference Moilanen, Kohonen, Lehtinen, Kivistö, Jalkanen, Virtanen and Kujala2024).
We used the protected areas layer as a hierarchical mask to lock in those cells in the solution and identify the priority areas outside the present protected area network that will be needed to reach the species’ minimum conservation targets. We also used the Conversion Pressure Index (described above) as a condition layer (Moilanen et al. Reference Moilanen, Kohonen, Lehtinen, Kivistö, Jalkanen, Virtanen and Kujala2024), that accounts for local habitat deterioration and its influence on the prioritization process. The purpose here is to identify preliminary areas for conservation at a large scale. Actual implementation of habitat protection for the focal species should refine this analysis with more detailed information at the landscape scale (including existing OECMs).
Results
We redrew the range maps and mapped the Area of Habitat for the 129 priority threatened bird species in Latin America. See Figure 1 depicting an example map for the Chestnut-capped Piha Lipaugus weberi, endemic to northwestern Colombia. Of the 129 species included in our analysis, 18 are in the IUCN Red List Category Vulnerable (VU D1 or D2), 79 Endangered (EN), and 32 Critically Endangered (CR).
When assessing the minimum conservation targets based on the proportion of AOH to be protected, over half (57%) of the species (74) have their targets met (i.e., 100% of the proposed minimum conservation area is currently protected), and 55 (43%) need to increase the area under protection to reach their targets (Supplementary material Table S1). On average, 39.9% of the AOH of all species analyzed is currently under protection, ranging from five species with less than 1% protected (Little Inca-finch Incaspiza watkinsi, Chilean Woodstar Eulidia yarrelli, Oaxaca Hummingbird Eupherusa cyanophrys, Blue-throated Hillstar Oreotrochilus cyanolaemus, and Marañon Antshrike Thamnophilus shumbae) to eight with >99% of their AOH protected (Supplementary material Table S1). The average conservation gap for the 55 species that do not have their proportion of AOH target protected is 27.9%.
The main priority areas for covering the conservation gaps of the 55 species whose targets have not been met are in the Tropical Andes of Peru, Ecuador, and Colombia; northeastern Brazil; the Caribbean coast of Colombia; and central and northern Mexico (Figure 2). The area that would be needed to meet the minimum conservation targets for these species, without considering the overlap in multiple species distributions, is 17,443 km2. However, in some cases, multiple species may occur in the same area; therefore, protecting a specific area may protect more than one species. Because the Zonation algorithm produces priority solutions that are well-balanced and where priority sites have high complementarity (Moilanen et al. Reference Moilanen, Lehtinen, Kohonen, Jalkanen, Virtanen and Kujala2022), the total area needed to meet the minimum conservation target for the 55 focal species is reduced to 14,363 km2.
The existing protected area network (green), the range maps of all focal species (light lilac), and the priority areas (dark purple) needed to meet the proposed minimum conservation targets for all 55 species for which the minimum conservation targets are not already met. Protected areas are excluded in the enlarged inset maps for better visualization.
Figure 2. Long description
The main map covers Mexico, Central America, and South America. Green regions representing protected areas are widely distributed, with large concentrations in the Amazon basin and the Yucatan Peninsula. Light lilac regions representing species ranges cover much of Mexico, the Andes, and the Atlantic Forest. Dark purple priority areas are scattered, with notable clusters in the Mexican highlands, the Northern Andes, and the Brazilian coast.
Four inset maps provide magnified views of priority areas without the green protected area overlay:
* Top-left inset: Focuses on the Gulf of Mexico coastline and Central Mexico, showing small dark purple clusters.
* Top-right inset: Shows the northern coast of South America and the Caribbean islands, with priority areas along the coastline.
* Far-right inset: Displays the eastern coast of Brazil, highlighting a dense concentration of dark purple priority areas in the Atlantic Forest region.
* Bottom-left inset: Focuses on the Andes region through Ecuador and Peru, showing a linear string of dark purple priority areas following the mountain range.
A legend in the bottom-left corner identifies the three color categories: dark purple for Priority areas, light lilac for Ranges of all species, and green for Protected areas.
An example of the Zonation software results identifying areas with overlap of several species can be seen in the Murici area, in the State of Alagoas, northeastern Brazil (Figure 3), where a conservation priority area of 5,158 ha contains parts of the AOH of ten of the species comprised in this study, including nine that do not have their minimum conservation targets met (Forbes’s Blackbird Anumara forbesi, Pernambuco Foliage-gleaner Automolus lammi, White-collared Kite Leptodon forbesi, Scalloped Antbird Myrmoderus ruficauda, Alagoas Antwren Myrmotherula snowi, Alagoas Tyrannulet Phylloscartes ceciliae, Pinto’s Spinetail Synallaxis infuscata, Orange-bellied Antwren Terenura sicki, and Long-tailed Woodnymph Thalurania watertonii). This relatively small area would protect an average of 9.4% (range 0.5–50.4 %) of the AOH of these nine species.
Priority conservation area of 5,158 ha identified by the Zonation software in the Murici area, northeastern Brazil (inset map). This area would protect an average of 9.4% of the AOH of nine priority bird species.
Figure 3. Long description
A satellite map displays a mosaic of forest and agricultural land.
* Inset Map: Located in the top right, showing the outline of Brazil with a red square on the northeastern coast and a star marking Brasilia.
* Main Map View:
- West: The urban area of Uniao dos Palmares is visible.
- Center-North: A large, jagged cyan outline identifies the Priority area identified, covering a dense dark green forest patch.
- East and South: Bright green outlines mark Existing protected areas, which partially overlap and extend Southeast from the cyan region.
- Southwest: The town of Branquinha is labeled near a river.
- South: The town of Murici is labeled at the bottom center.
* Map Elements:
- A North arrow and a 5 kilometer scale bar are located in the bottom left.
- A legend in the bottom left corner defines the green outline as Existing protected areas and the cyan outline as Priority area identified.
- Attribution text at the bottom right reads Contains modified Copernicus Sentinel data 2025 processed by Sentinel Hub.
When assessing the minimum conservation targets based on the area needed to protect up to 1,000 mature individuals of each species, 119 (92%) of the species have their targets met, and 10 (8%) need to increase the area under protection to reach their targets (Supplementary material Table S1). The average conservation gap for these 10 species is 8.3% (range 0.2%–36.3%), and the total area needed to cover these gaps is 661.3 km2 (range 0.9 km2–236.9 km2). Because these 10 species are a subset of the 55 species identified using the proportion of AOH conservation target, the priority areas for them can be selected using the broader scale prioritization exercise.
Discussion
Our results identified 55 under-protected bird species that require 14,363 km2 of additional habitat protection within very discrete areas to meet their minimum conservation targets (i.e., the minimum proportion of a species’ AOH that should be included in protected areas to be considered adequately represented and safeguarded). This is a tiny fraction (<0.1%) of the total area of approximately 20,000,000 km2 of Central and South America. The distribution of the lands needing protection is not even across the Neotropical Region, and is concentrated mostly in five countries (Mexico, Colombia, Ecuador, Peru, and Brazil), but even within these countries, the area is not a significant fraction of their total area. For example, the priority sites within each country represent approximately 0.17% of the total area of Mexico, 0.17% of Colombia, 0.78 % of Ecuador, 0.29% of Peru, and 0.02% of Brazil.
The area required to meet the area-based conservation targets for the 55 under-protected species may appear relatively small; however, this result is likely influenced by the high spatial resolution of our analysis. Whereas most regional, national, and global efforts to map Area of Habitat (AOH), including many cited in this study, have relied on pixel sizes of 1 km² or coarser, our analysis was conducted at a resolution of approximately 0.09 km². This finer resolution was selected because many focal species occur in the Tropical Andes region and other mountainous areas, where steep elevational gradients and pronounced habitat heterogeneity can be obscured or averaged out when using coarser (1×1 km) spatial units. As a result, the finer grain of our maps yields smaller AOH estimates and reduces the likelihood of including unsuitable habitat or elevations.
A notable and encouraging result of our analysis is that 74 of the 129 species in the analysis already meet their minimum proportional area conservation target. Nevertheless, continued conservation efforts remain essential to ensure that the protected areas where these occur, and the portions of the ranges of the 55 under-protected species that are already protected, are effectively managed to secure long-term habitat integrity and species persistence.
An important feature of our analysis is that it illustrates protection gaps for globally-threatened birds, providing useful information to ensure minimum protection for and prevent extinctions of this important subset of species. The area identified represents less than 0.1% of the region’s 30x30 protection goals. Increased public commitment to habitat protection, exemplified by national commitments to the 30x30 Initiative, are providing the enabling conditions to allow these lands to become conserved. Therefore, we recommend that the areas identified in this study are prioritized for conservation within the 30x30 Initiative, as well as included in countries’ biodiversity-inclusive spatial plans (Target 1 of the KMGBF) and be considered in plans for restoration under Target 2. They can also be included in countries’ National Biodiversity Strategies and Action Plans (NBSAPs), and other biodiversity conservation efforts. Further implementation of habitat protection within these priority areas to meet the conservation thresholds for individual species will require additional planning and project development at a local landscape scale, and must consider land tenure, local stakeholders, and ensure Free, Prior, and Informed Consent (FPIC) of such local stakeholders, particularly Indigenous peoples. Although the area required to meet the targets is not so large and spread across several countries, local impacts on the current occupants and owners of the land must be carefully considered and impacts allayed.
As expected, the priority sites identified in our study overlap with sites identified as internationally significant for biodiversity. For instance, 40.5% of the 14,363 km2 falls within the global network of KBAs, overlapping with 122 KBAs and covering an average 9.7% (range 0.002–75.4%) of the area of each of these 122 sites. Most importantly, besides including 667 bird KBA-trigger species (i.e. species for which the site meets KBA criteria), these sites also include 373 KBA-trigger species from other taxonomic groups, such as 148 amphibians, 110 plants, 58 reptiles, 30 mammals, 16 freshwater fish, and 11 arthropods. Moreover, 41 of the 122 KBAs are AZE sites that harbour the majority of the remaining population of one or more Critically Endangered or Endangered species, including 22 bird species and 56 non-bird species.
Area of Habitat (AOH) is a deductive model that substantially improves the spatial resolution of species’ distribution maps by reducing commission errors (i.e. the inclusion of areas that are unsuitable for the species). However, although our population-based targets aim to promote species persistence, the mere availability of suitable habitat does not necessarily ensure the presence of viable populations, which may also depend on factors such as population dynamics, dispersal capacity, habitat connectivity, and other ecological constraints. Nevertheless, as stated above, the objective of this study is to identify preliminary conservation priorities at broad spatial scales. The effective implementation of habitat protection should therefore focus on these priority areas and further refine the analysis using more detailed, landscape-scale data and species-specific information.
The recent analysis for the Northern Andes by Medina et al. (Reference Medina, Pimm and Huang2024) is complementary to our work. Our analysis focuses on species already considered threatened from all of Latin America outside the Caribbean, whereas their analysis includes all range-restricted (<50,000 km2) species and focuses on a relatively small, but important area. Medina et al. (Reference Medina, Pimm and Huang2024) use a method similar to that of Huang et al. (Reference Huang, Medina, Brooks, Butchart, Fitzpatrick, Hermes, Jenkins, Johnston, Lebbin, Li and Ocampo-Peñuela2021), with its reliance on the original IUCN/BirdLife maps and treatment of absence data. Although our two analyses differ, they are nonetheless similar in important ways, identifying some of the same regions as key areas. This is a significant result, indicating that these areas are important in a fundamental biological way, at least in the Northern Andes region, an area that has been highlighted as a conservation priority for threatened resident species of mammals, birds, reptiles, and amphibians (Wilson et al. Reference Wilson, Lin, Schuster, González, Gómez, Botero‐Delgadillo, Bayly, Bennett, Rodewald, Roehrdanz and Ruiz Gutierrez2022).
The 129 bird species in our analysis represent 12 orders of birds, with diverse life histories and ecologies, occupying landscapes ranging from sea level and relatively unbroken lands to high-elevation and highly dissected mountainous regions. They occur in habitats ranging from very humid forests to semi-arid deciduous woodlands. Developing methods to produce range maps that can account for the myriad differences among species and conditions has proven difficult, as seen from the fact that several new methods for producing range maps for birds have been proposed in just the last three years. An additional challenge for producing maps for many species in the Neotropics, especially ones that are threatened and therefore already rare, is the lack of data and the biased distribution of data that do exist. Citizen science programs such as eBird and scientific data collection efforts such as GBIF are very important sources of bird distribution information but often highly biased toward accessible locations and places where rare and sought-after species are already known to occur. They rarely provide sufficient data from across a representative portion of a species’ range to develop highly accurate range maps. Therefore, review using expert opinion to identify areas where birds likely occur or not occur remains important in the production of bird range maps.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0959270926100690.
Acknowledgments
We kindly thank David Luther, Stuart Butchart, Ryan Huang, Wilderson Medina, James Lewis, and Jorge Velásquez-Tibatá for providing feedback on early versions of this manuscript. Our thinking on this paper also benefitted from numerous discussions with Stuart Pimm, Ryan Huang, Wil Medina, and others in their working group over the last several years. Thanks also to two anonymous reviewers and Richard Schuster that helped improved the manuscript with their comments and suggestions.
Data availability
The AOH maps for all 129 bird species included in this study can be downloaded from Zenodo using the following link https://doi.org/10.5281/zenodo.20430340.
Open access
