Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2025 Nov 19;31(11):e70603. doi: 10.1111/gcb.70603

Rapid Global Deforestation Leaves Forest‐Dependent Raptors With Half of Their Suitable Habitat Remaining

Christopher J O'Bryan 1, Zunyi Xie 2,3,4,✉, Hongli Li 2,✉, Evan R Buechley 5, Martin K‐F Bader 1, James R Allan 6, Ralph Buij 5,7
PMCID: PMC12628116  PMID: 41255350

ABSTRACT

Raptors—a group that includes hawks, eagles, owls, and vultures—are among the most imperiled vertebrates, with over half of species declining and one‐fifth threatened with extinction. Most raptors depend on forests for survival, yet the extent and consequences of recent deforestation within their ranges remain poorly understood. We analyzed high‐resolution global data on forest change between 2001 and 2023 within the ranges of 369 forest‐dependent raptor species. On average, these species have lost 10% of forest within their ranges since 2001. Seventy‐seven species had already lost at least 15% of additional forest cover within their ranges in the latter half of the 20th century. Currently, only 52% of forest cover remains within forest‐dependent species' ranges on average, with many forest specialists retaining less than 25% cover. Forest‐dependent raptors listed as Critically Endangered by the IUCN have lost the most forest and now retain the least forest cover within their ranges. Shifting agriculture is the primary driver of forest loss—particularly in tropical Africa and South America—followed by commodity‐driven deforestation in Southeast Asia. Because forest loss is frequently used as a proxy for extinction risk for raptors, our findings have direct applications for species threat assessments, regional conservation priorities, and global biodiversity commitments.

Keywords: bird, extinction risk, forest loss, habitat loss, land use change, species distribution

We analyzed high‐resolution global data on forest change between 2001 and 2023 within the ranges of 369 forest‐dependent raptor species. On average, these species have lost 10% of forest within their ranges since 2001. Seventy‐seven species had already lost at least 15% of additional forest cover within their ranges in the latter half of the 20th century. Currently, only 52% of forest cover remains within forest‐dependent species' ranges on average, with many forest specialists retaining less than 25% cover. Forest‐dependent raptors listed as Critically Endangered by the IUCN have lost the most forest and now retain the least forest cover within their ranges. Shifting agriculture is the primary driver of forest loss—particularly in tropical Africa and South America—followed by commodity‐driven deforestation in Southeast Asia. Because forest loss is frequently used as a proxy for extinction risk for raptors, our findings have direct applications for species threat assessments, regional conservation priorities, and global biodiversity commitments.

graphic file with name GCB-31-e70603-g005.jpg

1. Introduction

Raptors—birds within the taxonomic orders Accipitriformes, Cathartiformes, Falconiformes, Strigiformes, and Cariamiformes (McClure et al. 2019)—are sentinels for planetary health. They regulate ecosystems through predation and scavenging, which aid ecosystem stability and provide benefits to human society (Bonetti et al. 2024; Donázar et al. 2016; O'Bryan et al. 2018). However, raptors are among the most imperiled vertebrates, with over half of species declining and one‐fifth threatened with extinction (McClure et al. 2018). Most raptor species depend on forests for survival (McClure et al. 2018), yet deforestation has rapidly increased over the past two decades due to urbanization, expanding and shifting agriculture, resource extraction, and increased wildfires linked to climate change (Curtis et al. 2018; Hansen et al. 2013).

Forest‐dependent raptors are one of the most sensitive groups of vertebrates to forest loss due to their position at the top of the forest food chain and their large area requirements (Newton 2010; J. M. Thiollay 1989). Forest loss decreases nesting opportunities and the availability of prey (Miranda et al. 2021), often forcing individuals to form larger territories to meet energetic demands (Martínez‐Ruiz et al. 2016). In addition, direct persecution by humans increases when forest raptors switch to livestock (McPherson et al. 2016; Miranda et al. 2019; Zuluaga and Echeverry‐Galvis 2016) and where raptors are hunted for consumption (Whytock et al. 2016) in fragmented habitats. As such, many forest‐dependent raptors have vanished completely in heavily human‐altered forests or forest patches in tropical (Jullien and Thiollay 1996; J.‐M. Thiollay and Meyburg 1988) and temperate (Bruggeman et al. 2023; Mozgeris et al. 2021) regions.

Despite the importance of forest loss to raptor populations, we have limited information on recent deforestation within the ranges of forest‐dependent raptors at a global scale. Forest raptors are often elusive and difficult to monitor, leading to limited knowledge about their population status and trends (Buechley et al. 2019). In the absence of population data, Red List assessments are often inferred from the extent of forest loss within ranges (IUCN 2024), but such assessments have not been made for most forest‐dependent raptors. This knowledge gap presents challenges for countries that are committed to reducing species extinctions and to protecting and restoring degraded habitats (Allan et al. 2022; Hering et al. 2023; United Nations Environment Programme 2022). We aim to help address these challenges by generating estimates of contemporary forest loss within the distribution ranges of Earth's forest‐dependent raptors.

In this study, we harness high‐resolution data on forest change from 2001 to 2023 to quantify forest loss (Hansen et al. 2013) and gain (Du et al. 2023) within 369 forest‐dependent raptor species' geographic distributions, which represent 66% of all raptors. For each species, we also determine the dominant drivers of forest loss (Curtis et al. 2018) and calculate the percentage of forest habitat remaining within their ranges. We map patterns of forest loss across raptor ranges and scale by species richness to determine areas of greatest concern.

2. Materials and Methods

2.1. Spatial Data on Raptor Geographic Distributions

We obtained extent‐of‐occurrence maps on the native and reintroduced distribution ranges of raptor species from BirdLife International (v2022.2). We define raptors as birds in the orders Accipitriformes, Cathartiformes, Falconiformes, Strigiformes, and Cariamiformes (n = 561 species) (McClure et al. 2019). We only included the extant parts of each species' geographic distribution in our analysis, including the migratory, breeding, and wintering distributions, and excluding ranges of possibly extinct or extinct species. We excluded parts of species' ranges with uncertain or no current records of species presence as assessed by the IUCN, as well as introduced, reintroduced, vagrants, and species with unknown origin. We only included raptors that were considered forest‐dependent (n = 374 species), i.e., raptors classified as forest generalists (medium dependency) or specialists (high dependency) according to BirdLife International (see Table S1). As such, because forest‐dependent raptors rely on forest habitat for survival, we identified the functional range for each raptor as the area of baseline forest cover within their BirdLife International distribution range polygons. To determine this, we calculated forest cover for each raptor species at 30 m resolution for the year 2000 from Global Forest Watch (Hansen et al. 2013). After removing species that did not intersect with the layers used in this analysis (five species), the final sample size was 369 species (66% of all raptors).

2.2. Spatial Data on Forest Cover Change

We obtained spatial data on the distribution of forest cover change, including forest loss (Hansen et al. 2013) and forest gain (Du et al. 2023). Forest loss is defined as a change from a forest (all vegetation taller than 5 m in height) to a non‐forest state—that is, a stand‐replacement disturbance or the complete removal of tree cover canopy—within the years 2001 through 2023 at a 30 m resolution (Hansen et al. 2013). Forest gain is defined as the inverse of loss, or a non‐forest to forest change within the years 2001 through 2021 (Du et al. 2023). We also obtained spatial information on six dominant drivers of forest loss at a 10 km resolution: (1) commodity‐driven deforestation (permanent, long‐term forest loss due to agriculture, mining, or energy infrastructure); (2) shifting agriculture (small‐ to medium‐scale forest loss due to agriculture that is later abandoned and followed by subsequent forest regrowth); (3) forestry (large‐scale forestry operations within managed forests and tree plantations); (4) wildfire (large‐scale forest loss resulting from the burning of forest vegetation with no visible human conversion or agricultural activity afterward); and (5) urbanization (forest and shrubland conversion for the expansion and intensification of existing urban centers) (Curtis et al. 2018). We resampled the spatial resolution of the dominant driver layer using the nearest neighbor assignment to match the resolution of the forest loss and forest gain layers (30 m). We then identified the dominant driver of deforestation within each cell of forest loss. We conducted all geospatial analyses in ArcGIS 10.8 (ESRI) and Google Earth Engine under a Mollweide equal area projection.

2.3. Analyzing Forest Loss Within Raptor Ranges

We estimated the area and proportion of forest loss within raptor functional ranges by intersecting the forest loss layer, the forest gain layer, and the dominant drivers layer with each raptor's range polygons. We report forest loss and forest gain separately as primary and successional forests are ecologically distinct (e.g., Bowen et al. 2009). Furthermore, we calculated the forest cover remaining within each raptor range by subtracting total forest loss from the baseline forest cover for the year 2000. We also conducted a post hoc comparison of forest cover from 2000 with historical cover dating back to 1960. To do this, we obtained spatial information on forest cover for 1960 from the HILDA+ landcover data set at a 1 km resolution (Winkler et al. 2021) and intersected it with raptor ranges. We map our results of recent forest loss using a 30 × 30 km2 grid, which is a standard approach for visualizing patterns across species distributions at a global scale, with limited commission errors (Di Marco et al. 2017; O'Bryan et al. 2022).

2.4. Analyzing Predictive Importance of Forest Loss and Raptor Traits for Threat Status

To identify predictors of the IUCN classification of forest‐dependent raptors, we trained a random forest model (R packages caret [Kuhn 2008] and randomForest [Liaw and Wiener 2002]) using the variables forest loss, drivers of forest loss, range size, baseline forest cover, percent forest cover remaining, body size, and forest dependency. The “Data‐Deficient” category comprised only two observations, which were excluded from the data prior to analysis to avoid biased learning and misclassification caused by an extremely underrepresented class. Given the limited number of observations (n = 8) in the minority class (“Critically Endangered” category), we manually performed a random split to allocate three observations to the test set and five to the training set. For the remaining data, we used a stratified 70/30 split to address class imbalance and ensure representative class distributions in the training and test sets. Model training was based on repeated 10‐fold cross‐validation (10 repeats) and included a hyperparameter tuning routine suggesting 500 trees and two randomly sampled features (predictors) at each split (mtry = 2). We also applied the synthetic minority over‐sampling technique (SMOTE) during training to account for class imbalance in our dataset. Feature importance was assessed using a model‐agnostic, permutation‐based approach, where the change in cross‐entropy loss is calculated after permuting each predictor individually (R package DALEX [Biecek 2018]). The importance of a feature was expressed as the ratio of the cross‐entropy loss of the reduced model (with the feature in question permuted) to that of the full model. A higher cross‐entropy loss ratio indicates greater feature importance, as the model's predictive performance declines more strongly when the feature is permuted. We ran 1000 permutation rounds to ensure stable importance rankings and derived quantile‐based 95% confidence intervals from these iterations.

3. Results

We find that forest‐dependent raptors have experienced an average loss of 10% (±SD 6%) forest habitat within their ranges globally since 2001, with 75 species (20%) having lost at least 15%. Moreover, 77 of the assessed species (21%) have already experienced a reduction of at least 15% forest cover from 1960 to 2000. Across all forest‐dependent raptor ranges, an average of 196,560 km2 of forest has been lost since 2001 (±SD 375,535 km2; Table S2). The species with the greatest forest loss include many currently listed as Least Concern by the IUCN. For example, the Wallace's hawk‐eagle ( Nisaetus nanus ), a forest specialist of Southeast Asia, has lost 35% of its forest habitat since 2001.

Our analysis further shows that raptors have lost at least 10% of their forested ranges since 2001 in each of the 30 × 30 km2 grid cells of 58 countries, the majority of which are in the Caribbean (12 countries) and Southeast Asia (9 countries), followed by Northern and Western Europe (7 countries each), Central America (5 countries), and East Asia (4 countries). The countries containing the highest average percentage of forest loss across raptor ranges include Madagascar (24%), Singapore and Brunei Darussalam (22% each), Malaysia (21%), Comoros (17%), and Indonesia (16%) (Figure 1A). When scaled by species richness (richness multiplied by mean percentage forest loss in a grid cell) to reveal areas of forest loss that are particularly important for the conservation of multiple raptor species, priority hotspots of forest loss occur primarily in Southeast Asia, as well as parts of Central and South America (Figure 1B).

FIGURE 1.

FIGURE 1

Open in a new tab

Patterns of forest loss across forest‐dependent raptor ranges. (A) Percentage of forest loss from 2001 to 2023 averaged across forest‐dependent raptor species' distributions within 30 × 30 km grid cells. Inlays of 13 species examples show their approximate location, percent forest loss, and pie charts indicating the proportion of forest loss attributed to commodity‐driven deforestation, shifting agriculture, forestry, wildfire, and urbanization. (B) Percentage forest loss as in panel (A) but multiplied by forest‐dependent raptor species richness to reveal hotspots of impact. Inlays of 20 country examples show their forest‐dependent raptor species richness along with pie charts indicating the average proportion of forest loss attributed to each of the five drivers averaged across grid cells in the country. Maps were generated under a Mollweide equal‐area projection. Silhouettes were obtained from www.phylopic.org and are used under the Creative Commons Attribution‐Non Commercial 3.0 Unported license. Silhouette credits are as follows: Otus sp. = Nix Illustration, Nisaetus sp. and Buteo sp. = modified from Andy Wilson, Asio sp. = Lauren McLean.

The dominant driver of recent forest loss across raptor ranges was shifting agriculture (i.e., small‐ to medium‐scale forest loss due to agriculture that is later abandoned and followed by subsequent forest regrowth), which accounted for 39% of loss on average (mean area of 56,113 km2), followed by commodity‐driven deforestation (i.e., permanent loss of forest habitat due to agriculture, mining, or energy infrastructure; 35% of loss and 52,438 km2 on average), and forestry (i.e., large‐scale forestry operations within managed forests and tree plantations; 26% of loss and 57,329 km2 on average) (Table S3). A plurality of forest in South America, Southeast Asia, and Melanesia was lost from commodity‐driven deforestation, and the same was true for Sub‐Saharan Africa, Melanesia, South America, and Central America from shifting agriculture. Furthermore, North America, Europe, and southern Asia are hotspots for forest loss from forestry, and Oceania, Europe, North America, and central Asia from wildfire (Figure 2). Indeed, forestry was the leading driver of total forest loss (i.e., accounting for > 50% when averaged across all grid cells) for raptors within 73 countries, mostly in Europe and western Asia, while shifting agriculture was the leading driver for 52 countries, mostly in Africa. Commodity‐driven deforestation was the dominant driver within 11 countries, all from Southeast Asia, and wildfire was dominant only in Australia, accounting for 56% of total forest loss within raptor ranges on average. Urbanization was not a dominant driver in any country, with the highest being in the United States and in seven Caribbean nations (all at 3% each).

FIGURE 2.

FIGURE 2

Open in a new tab

Drivers of forest loss within forest‐dependent raptor ranges. Percentage of lost forest from the years 2001 to 2023 averaged across forest‐dependent raptor species' distributions within 30 × 30 km grid cells attributed to (A) commodity‐driven deforestation, (B) shifting agriculture, (C) forestry, (D) wildfire, and (E) urbanization. Maps are presented under a Mollweide equal‐area projection.

Among raptor families, the two forest‐dependent New World vulture species (Cathartidae) had the highest percentage of forest loss within their ranges on average over the last two decades (22% ± SD 15%). Falcons and caracaras (Falconidae) had the second highest loss (13% ± SD 8%), followed by barn‐owls (Tytonidae; 13% ± SD 7%), hawks and eagles (Accipitridae; 11% ± SD 6%), and typical owls (Strigidae; 10% ± SD 6%) (Table S4). Large raptors (> 1300 g; n = 30; Shaw et al. 2024) had a slightly higher percentage of forest loss within their ranges (11% ± SD 7%) than all other species (n = 339; 10% ± SD 7%) on average (Table S5). Shifting agriculture was the dominant driver of forest loss for all raptors, including large (37% of loss ±SD 28%) and small‐to‐medium species (39% of loss ±SD 38%) (Table S6).

The assessed forest‐dependent raptors now have an average of 52% forest cover remaining across their ranges globally (±SD 25%). Ten species of particular concern have both the highest percentage of recent forest loss (i.e., at least 15% loss) and the smallest absolute forest cover area remaining in their range (i.e., within the first quartile). Of these species, nearly all (7) have a declining population trend according to the IUCN Red List of Threatened Species (IUCN 2024) (Figure 3). For example, the Critically Endangered Ridgway's hawk ( Buteo ridgwayi ), endemic to the island of Hispaniola in the Caribbean, has approximately 163 km2 of forest cover remaining in its range following 29% loss since the year 2001. Similarly, the Vulnerable Pemba scops‐owl ( Otus pembaensis ), a species endemic to Pemba Island off the coast of Tanzania, has lost 17% of forest habitat since 2001 and has only 365 km2 of forest remaining in its range.

FIGURE 3.

FIGURE 3

Open in a new tab

Percentage of forest loss and remaining forest cover within the ranges of forest‐dependent raptors that have lost at least 15% of their forested range (n = 75). (A) Percentage of forested range loss from 2001 to 2023, and (B) percentage forest cover remaining. Red bars indicate species of high forest‐dependency (forest specialists), and green bars represent species of medium forest‐dependency (forest generalist; see Table S1 for definitions). Asterisks identify species (n = 10) that have the smallest forest cover area remaining in their ranges (i.e., within the first quartile).

3.1. Forest Loss Across Threat Categories

Among imperiled forest‐dependent raptors, Critically Endangered species have lost a disproportionate amount of forest habitat over the last two decades on average (14%; ±SD 11%) compared to Vulnerable (10%; ±SD 7%) and Endangered species (7%; ±SD 6%) (Table S7; Figure 4A). Species listed as Least Concern and Near Threatened both lost 11% of forest within their ranges on average (±SD 6% and 7%, respectively) (Table S7; Figure 4A). Shifting agriculture was the leading driver of forest loss for threatened (46% of loss; ±SD 43%) and non‐threatened raptors (41% of loss; ±SD 38%) on average, followed by commodity‐driven deforestation (28% of loss; ±SD 37% for threatened and 30% of loss; ±SD 34% for non‐threatened) (Table S8; Figure 4B). For example, the Near Threatened barred eagle‐owl ( Bubo sumatranus ) of the Indomalayan region has lost a third of its forested range within the study period almost exclusively from commodity‐driven deforestation. For a complete list of species and their results, see Data S1.

FIGURE 4.

FIGURE 4

Open in a new tab

Percentage forest loss and drivers of loss within the ranges of forest‐dependent raptors by IUCN threat category. (A) Mean percentage of forest lost across each IUCN threat category, along with standard errors; and (B) Stacked bar plot of the mean percentage of lost forest from five dominant drivers across each IUCN threat category.

Species listed as Critically Endangered have the lowest percentage of forested habitat remaining in their ranges (42%; ±SD 25%), followed by Least Concern (50%; ±SD 25%), Endangered (53%; ±SD 24%), Near Threatened (58%; ±SD 21%), Vulnerable (61%; ±SD 25%), and Data Deficient (82%; ±SD 10%) species.

3.2. Forest Gain Within Ranges

Forests are not static and in addition to forest loss, forest‐dependent raptors have also experienced an average 4% (±SD 6%) forest cover gain from 2001 to 2021. This amounts to an average forest gain area of 51,899 km2 across all species' ranges (Table S2). Of this forest gain, 27% overlaps with areas where forestry is a dominant driver of loss (Table S3), suggesting that plantation forests, which may be less ecologically valuable, may explain nearly a third of the regrowth. Across threat classes, species listed as Endangered experienced the highest average forest gain during the study period (8%; ±SD 21%) followed by Vulnerable species (5%; ±SD 7%). Critically Endangered and Near Threatened species all experienced an average 4% forest gain across their ranges (±SD 4%), with Least Concern species experiencing 3% (±SD 3%).

3.3. Predictors of Extinction Risk

A post hoc analysis using a random forest algorithm identified body size and forest dependency as the most important predictors of IUCN threat status of the assessed raptors, with nearly equal importance scores (cross‐entropy loss ratio; Figure S1). The next most important predictor was the percent forested range after loss (for the year 2023), followed by baseline forest cover (for the year 2000) and range size. The percentage of forest loss and proportion of loss from each of the five drivers followed in the ranking of importance (Figure S1).

4. Discussion

We provide a global assessment of recent forest loss (from 2001 to 2023) within the ranges of 369 forest‐dependent raptor species, which represent 66% of all raptors. We find that these species have lost an average of 10% of forest within their ranges, with 77 forest‐dependent raptors (21%) having already lost at least 15% of additional forest cover from 1960 to 2000. Undoubtedly, forest loss was even more substantial prior to this and during the industrial revolution, resulting in profound extinction debts (Liao et al. 2022). For example, many parts of North America and Europe experienced > 80% gross forest loss from 1500 to 1992 (Liao et al. 2022), of which forest undoubtedly contained forest‐dependent raptors. As of 2023, forest‐dependent raptors have just half of their range containing forest on average, according to our analysis. Forests are important for breeding, nesting, and food acquisition for forest‐dependent raptors, and forest cover below certain thresholds may negatively impact population viability. For example, the harpy eagle ( Harpia harpyja ) feeding rate was shown to decrease with deforestation to the point that they could not effectively reproduce once half of their forested range was lost within a breeding territory (Miranda et al. 2021). Similarly, populations of the Critically Endangered Philippine eagle ( Pithecophaga jefferyi ) are now facing a severe genetic bottleneck most likely from deforestation (Luczon et al. 2014). In the late 1960s, the Philippines had approximately 35% forest cover remaining, which led to the first public warning about the declining population of the Philippine eagle due to forest loss and hunting (Rabor 1968). Our analysis suggests that the Philippine eagle has experienced a further 9% forest loss in the last two decades, mostly from shifting agriculture.

Our post hoc random forest analysis indicated that body size and forest dependency are important predictors of extinction risk. These results follow similar research assessing extinction risk of raptors. For example, forest dependency was an important predictor of extinction risk for 557 extant raptor species globally (Buechley et al. 2019). Smaller ranged species tend to be more at risk given small population sizes and high vulnerability to changes in habitat (Harris and Pimm 2008). Our results indicate that the amount of forest cover within a forest‐dependent raptor species' range and overall range size are also important predictors of extinction risk. This shows the critical role that deforestation plays in driving extinction risk for Earth's raptors.

We found that forest raptors with the greatest forest loss over the last two decades include many currently listed as Least Concern by the IUCN, and these Least Concern species have lost more forest within their ranges than Vulnerable and Endangered ones on average. Contrary to open‐country raptors whose trends over large portions of their range have been assessed from counts (e.g., along roads [Shaw et al. 2024]), forest‐dependent raptors are often elusive and rarely monitored, especially in the tropics (Buechley et al. 2019). As such, populations may be suspected to be stable in the absence of evidence for any declines or substantial threats. Some of the Least Concern forest‐dependent raptors with substantial loss of forest within their range may however warrant reclassification based on IUCN's Red List data criterium B2, which refers to the relationship between population reduction and habitat loss over three generations. Our analysis highlights Least Concern species affected by a high percentage of forest loss and a small extent of remaining forest cover within their ranges (Figure 3). For example, the bearded screech‐owl ( Megascops barbarus ) and the Puerto Rican screech‐owl ( Megascops nudipes ) merit review. Furthermore, case studies in the literature point to a link between forest loss and population trends. For example, a long‐term study (1973–2018) conducted in Finland on the boreal owl ( Aegolius funereus ), a forest specialist listed as Least Concern by the IUCN, showed that the loss of mature and old‐growth forests led to a decline in the availability of primary habitats and alternative food sources, causing fledgling starvation and long‐term population decline (Kouba et al. 2020). Similarly, from 2002 to 2005, forest habitat loss and fragmentation in Canada reduced the foraging efficiency and reproductive success of the northern saw‐whet owl ( Aegolius acadicus ) by reducing home range size and provisioning rates (Hinam and Clair 2008). Our analysis shows that these species have lost 13% and 16% of their forested range, respectively, from 2001 to 2023, suggesting potential concomitant population declines.

Our results point to shifting agriculture, commodity‐driven deforestation, and forestry being the top three drivers of forest loss within the assessed raptor ranges. Commodity‐driven deforestation—that is, permanent loss of forest habitat due to agriculture, mining, or energy infrastructure—accounted for nearly a third of the deforestation in our analysis on average and was the dominant driver of loss in Southeast Asia and South America. Forestry was the leading driver of total forest loss in our analysis (i.e., accounting for > 50% when averaged across all grid cells) and occurred mostly in Europe and western Asia, whereas in Africa the leading driver was shifting agriculture, and wildfire was dominant only in Australia. In South America, a combination of shifting agriculture and commodity‐driven deforestation was the key drivers of forest loss within raptor ranges.

Permanent loss of forest from many of these drivers has been linked to extirpation of forest‐dependent species. For example, the African goshawk ( Accipiter tachiro ), the southern banded snake‐eagle ( Circaetus fasciolatus ), and the African wood owl ( Strix woodfordii ) went locally extinct across five sites in South Africa where indigenous forest was lost from 1990 to 2014 (Cooper et al. 2017). While permanent forest loss is costly for forest‐dependent raptors, strategic loss (e.g., from managed forestry or shifting agricultural operations) may minimize impacts. For example, a study on forest‐dependent raptors in Estonia suggests that if old‐growth stand structure is preserved and buffer zones around nests are maintained, some timber harvesting outside of the breeding season has limited impacts on reproduction (Lõhmus 2005). Furthermore, some drivers of forest loss may be more easily reversible than others, such as shifting agriculture or wildfire versus forestry or commodity‐driven deforestation. But for some raptor species, replacement of natural forests with plantations is not as detrimental as other drivers of loss given plantations can provide breeding sites (Rodríguez et al. 2021). Nevertheless, forest loss or conversion has several immediate direct impacts on raptor populations including reduced breeding opportunities, limited shelter, and reduced prey base (Buechley et al. 2019; McClure et al. 2018).

Beyond the direct impacts on raptor population viability, forest loss may exacerbate persecution of raptors by human communities that can lead to increased extinction risk. For example, in the rural Andean landscape of Colombia, black‐and‐chestnut eagles ( Spizaetus isidori ) prey on more domestic fowl in areas where deforestation had occurred (Zuluaga and Echeverry‐Galvis 2016), which led to retaliatory killings via shooting and trapping that exacerbated extinction risk (Restrepo‐Cardona et al. 2020). Similarly, in southern Chile the lower the proportion of forest in the region the higher the risk of chicken predation by diurnal raptors and subsequent conflict (Almuna et al. 2020). As forest loss often coincides with road development and increased access to forests by hunters (Ziegler et al. 2016), it may also lead to higher subsistence hunting pressure where forest raptors are targeted for consumption, such as in Central and West Africa (Buij et al. 2016; Whytock et al. 2016). Therefore, forest loss can result in indirect impacts to raptor populations via increased persecution by humans.

Separately from forest loss, we also found that the assessed raptors experienced an average 4% forest gain across their ranges during the study period. Forest gain can mitigate some of the impact of forest loss, as many forest‐dependent raptors are known to utilize secondary regrowth, exotic stands or plantations (Santander et al. 2021; Seaton et al. 2013; Suárez et al. 2000). Indeed, exotic tree plantations can provide suitable nesting conditions and support raptor populations with breeding densities and success comparable to or better than that found in native forests in some cases (e.g., Martínez‐Hesterkamp et al. 2018; Rodríguez et al. 2021). However, forest raptor community richness often declines when native forest is replaced by regrowth (e.g., Jullien and Thiollay 1996). Also, forest maturation and related canopy height are of importance, as mature forest often takes a long time to regenerate. In our analysis, forest is defined as vegetation over 5 m, but mature forest over 15–20 m is often preferred by raptors, especially for breeding (Jiménez‐Franco et al. 2018; Saga and Selås 2012; Santangeli et al. 2012). Given a preference for nesting in taller trees that disappear with logging, raptors may be impacted by forest loss irrespective of short‐term forest gain. The forest gain in our analysis only spans the last 20 years, meaning that for many species the forest may not have matured enough to be accurately considered important habitat. For example, an analysis of woodland bird assemblages in fragmented sub‐tropical brigalow landscapes in southern Queensland, Australia, found that only regrown forest at least 30 years old had a strong influence on species richness and abundance compared to regrown forest less than 30 years old (Bowen et al. 2009). And a recent expert elicitation of the minimum requirements for the endangered red goshawk ( Erythrotriorchis radiatus ) in Queensland, Australia revealed that the species needs regrown forest to be at least 29 years old for healthy populations (Thomas et al. 2025). Consequently, while forest gain can be important for raptors in some cases, its importance is highly variable and often requires multiple decades to meaningfully contribute to population viability.

Our results point to concerning levels of forest loss across forest‐dependent raptor ranges, but there is opportunity for improving our understanding of forest loss and its implications. For example, we assume that the forested portion of a raptor species' range is generally occupied by the species, but a species will not usually occur throughout its extent of occurrence. This area may contain unsuitable or unoccupied forest habitat (e.g., at the wrong altitude, the wrong forest type, etc.). Future work could incorporate habitat suitability maps to improve the accuracy of range estimates, and also the estimates of percentage forest loss within a raptor's area of occupancy (Sutton et al. 2021). Also, while we differentiate between medium and high forest‐dependent species in our analysis, medium forest‐dependent species can cope with a degree of loss better than forest specialists. And even for forest specialists, sensitivity to deforestation may differ across the distribution range of the species. For example, the northern goshawk ( Accipiter gentilis ), which is classified as a highly forest‐dependent species, has also colonized open and urban landscapes, breeding in city parks and hunting in urban areas (Rutz et al. 2006). BirdLife International's classification of forest dependency, while highly valuable, merits further expert review. Lastly, while our analysis gives important insight into contemporary forest loss for raptors, historical context is valuable to illuminate shifting baseline syndrome (Papworth et al. 2009). We have attempted to incorporate more historical estimates of forest cover within raptor geographic distributions, dating back to the year 1960. However, these data are coarse and are not methodologically consistent with contemporary datasets. Therefore, as improved geospatial methods account for historical forest loss, so can analyses of forest loss within species' geographic ranges.

Our results have important applications for regional and global environmental commitments. For example, the Kunming‐Montreal Global Biodiversity Framework of the Convention on Biological Diversity seeks to reach a global vision of a world living in harmony with nature by 2050 (United Nations Environment Programme 2022), including 23 action‐oriented targets for the year 2030. Relevant targets include restoring 30% of degraded ecosystems (including forests; target 3) and halting species extinctions and managing human–wildlife conflicts (target 4). Our results imply several countries and regions where long‐term forest protection and restoration are needed within raptor ranges to ensure population persistence and reduce conflict with humans. Given that forest raptors are linchpins for ecosystem integrity that can have cascading effects on human health and wellbeing (Donázar et al. 2016), now is the time to halt further loss and promote restoration of forest habitat.

Author Contributions

C.J.O. and R.B. framed the study. C.J.O., Z.X., H.L., and M.K.‐F.B. performed the analyses. C.J.O. wrote the manuscript with input and support from all authors. All authors discussed and interpreted the results and helped write the manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Permutation‐based feature importance (variable importance) of a random forest model for predicting IUCN threat status of forest‐dependent raptors.

Table S1: Forest dependency categories and descriptions from Birdlife International used in this study.

Table S2: Mean, standard deviation, minimum, and maximum area (km2) of forest change within the geographic distributions of 369 forest‐dependent raptor species.

Table S3: Mean area (km2) and standard deviation (SD) of forest change across five different drivers of forest loss within the geographic distributions of 369 forest‐dependent raptor species.

Table S4: Mean, standard deviation, minimum, and maximum area (km2) of forest change within the geographic distributions of 369 forest‐dependent raptor species, organized by family.

Table S5: Mean, standard deviation, minimum, and maximum area (km2) of forest change within the geographic distributions of 369 forest‐dependent raptor species, organized by raptor body size.

Table S6: Mean and standard deviation area (km2) of forest change within the geographic distributions of 369 forest‐dependent raptor species, organized by raptor body size and driver of loss.

Table S7: Mean, standard deviation, minimum, and maximum area (km2) of forest change within the geographic distributions of 369 forest‐dependent raptor species, organized by IUCN status and driver of loss.

Table S8: Mean and standard deviation area (km2) of forest change within the geographic distributions of 369 forest‐dependent raptor species, organized by driver of loss and IUCN status.

Data S1: gcb70603‐sup‐0001‐FigureS1‐TableS1‐S8‐DataS1.docx.

GCB-31-e70603-s001.docx (386.2KB, docx)

Acknowledgments

We thank R. Erkens for his helpful comments on the figures. C.J.O. would like to thank J. Wallace Coffey for his guidance. This work was partly supported by the Natural Science Foundation of China (42371311).

O'Bryan, C. J. , Xie Z., Li H., et al. 2025. “Rapid Global Deforestation Leaves Forest‐Dependent Raptors With Half of Their Suitable Habitat Remaining.” Global Change Biology 31, no. 11: e70603. 10.1111/gcb.70603.

Funding: This work was supported by the National Natural Science Foundation of China (42371311).

Contributor Information

Zunyi Xie, Email: zunyixie@henu.edu.cn.

Hongli Li, Email: lihongli@henu.edu.cn.

Data Availability Statement

The supplementary data file can be obtained from https://doi.org/10.5281/zenodo.17440465. Spatial data for analyzing forest change and drivers of loss can be obtained from Global Forest Watch at https://data.globalforestwatch.org/documents/941f17325a494ed78c4817f9bb20f33a/explore and https://data.globalforestwatch.org/documents/ff304784a9f04ac4a45a40f60bae5b26/about. The forest gain data can be obtained from https://code.earthengine.google.com/?asset=projects/ee‐tufangbobo/assets/ForestGain/forestGainFinal. The HILDA+ data can be obtained from https://doi.pangaea.de/10.1594/PANGAEA.921846. Raptor geographic distributions can be requested from Birdlife International at https://datazone.birdlife.org/contact‐us/request‐our‐data. All spatial analyses were performed in ArcGIS (v10.8) and Google Earth Engine (GEE). The random forest algorithm was conducted in R (v4.5.0). The GEE code for analyzing baseline forest cover within species' ranges for the year 2000 can be found here https://code.earthengine.google.com/72b34c0ca6547bf89d40c8f5d69c98a6, for forest loss can be found here https://code.earthengine.google.com/9c8f73aac1635a3857327b849e940550, for forest gain can be found here https://code.earthengine.google.com/?asset=projects/ee‐tufangbobo/assets/ForestGain/forestGainFinal, and for the drivers of forest loss can be found here https://code.earthengine.google.com/45c32a82aa39025848a82f47aea59393.

References

  1. Allan, J. R. , Possingham H. P., Atkinson S. C., et al. 2022. “The Minimum Land Area Requiring Conservation Attention to Safeguard Biodiversity.” Science 376, no. 6597: 1094–1101. 10.1126/science.abl9127. [DOI] [PubMed] [Google Scholar]
  2. Almuna, R. , Cruz J. M., Vargas F. H., and Ibarra J. T.. 2020. “Landscapes of Coexistence: Generating Predictive Risk Models to Mitigate Human‐Raptor Conflicts in Forest Socio‐Ecosystems.” Biological Conservation 251: 108795. 10.1016/j.biocon.2020.108795. [DOI] [Google Scholar]
  3. Biecek, P. 2018. “DALEX: Explainers for Complex Predictive Models in R.” Journal of Machine Learning Research 19, no. 84: 1–5. [Google Scholar]
  4. Bonetti, E. A. , Isacch J. P., Paterlini C. A., Bó M. S., and Biondi L. M.. 2024. “Global Review of Services and Conflicts Provided by Raptors in Urbanized Habitats.” Urban Ecosystems 27, no. 6: 2101–2120. 10.1007/s11252-024-01579-1. [DOI] [Google Scholar]
  5. Bowen, M. E. , McAlpine C. A., Seabrook L. M., House A. P. N., and Smith G. C.. 2009. “The Age and Amount of Regrowth Forest in Fragmented Brigalow Landscapes Are Both Important for Woodland Dependent Birds.” Biological Conservation 142, no. 12: 3051–3059. 10.1016/j.biocon.2009.08.005. [DOI] [Google Scholar]
  6. Bruggeman, J. E. , Kennedy P. L., Andersen D. E., Deisch S., and Stukel E. D.. 2023. “Declining American Goshawk (Accipiter atricapillus) Nest Site Habitat Suitability in a Timber Production Landscape: Effects of Abiotic, Biotic, and Forest Management Factors.” Journal of Raptor Research 57, no. 4: 595–616. 10.3356/JRR-22-116. [DOI] [Google Scholar]
  7. Buechley, E. R. , Santangeli A., Girardello M., et al. 2019. “Global Raptor Research and Conservation Priorities: Tropical Raptors Fall Prey to Knowledge Gaps.” Diversity and Distributions 25, no. 6: 856–869. 10.1111/ddi.12901. [DOI] [Google Scholar]
  8. Buij, R. , Nikolaus G., Whytock R., Ingram D. J., and Ogada D.. 2016. “Trade of Threatened Vultures and Other Raptors for Fetish and Bushmeat in West and Central Africa.” Oryx 50, no. 4: 606–616. 10.1017/S0030605315000514. [DOI] [Google Scholar]
  9. Cooper, T. J. G. , Wannenburgh A. M., and Cherry M. I.. 2017. “Atlas Data Indicate Forest Dependent Bird Species Declines in South Africa.” Bird Conservation International 27, no. 3: 337–354. 10.1017/S095927091600040X. [DOI] [Google Scholar]
  10. Curtis, P. G. , Slay C. M., Harris N. L., Tyukavina A., and Hansen M. C.. 2018. “Classifying Drivers of Global Forest Loss.” Science 361, no. 6407: 1108–1111. 10.1126/science.aau3445. [DOI] [PubMed] [Google Scholar]
  11. Di Marco, M. , Watson J. E. M., Possingham H. P., and Venter O.. 2017. “Limitations and Trade‐Offs in the Use of Species Distribution Maps for Protected Area Planning.” Journal of Applied Ecology 54, no. 2: 402–411. 10.1111/1365-2664.12771. [DOI] [Google Scholar]
  12. Donázar, J. A. , Cortés‐Avizanda A., Fargallo J. A., et al. 2016. “Roles of Raptors in a Changing World: From Flagships to Providers of Key Ecosystem Services.” Ardeola 63, no. 1: 181–234. 10.13157/arla.63.1.2016.rp8. [DOI] [Google Scholar]
  13. Du, Z. , Yu L., Yang J., et al. 2023. “Mapping Annual Global Forest Gain From 1983 to 2021 With Landsat Imagery.” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 16: 4195–4204. 10.1109/JSTARS.2023.3267796. [DOI] [Google Scholar]
  14. Hansen, M. C. , Potapov P. V., Moore R., et al. 2013. “High‐Resolution Global Maps of 21st‐Century Forest Cover Change.” Science 342, no. 6160: 850–853. 10.1126/science.1244693. [DOI] [PubMed] [Google Scholar]
  15. Harris, G. , and Pimm S. L.. 2008. “Range Size and Extinction Risk in Forest Birds.” Conservation Biology 22, no. 1: 163–171. 10.1111/j.1523-1739.2007.00798.x. [DOI] [PubMed] [Google Scholar]
  16. Hering, D. , Schürings C., Wenskus F., et al. 2023. “Securing Success for the Nature Restoration Law.” Science 382, no. 6676: 1248–1250. 10.1126/science.adk1658. [DOI] [PubMed] [Google Scholar]
  17. Hinam, H. L. , and Clair C. C. S.. 2008. “High Levels of Habitat Loss and Fragmentation Limit Reproductive Success by Reducing Home Range Size and Provisioning Rates of Northern Saw‐Whet Owls.” Biological Conservation 141, no. 2: 524–535. 10.1016/j.biocon.2007.11.011. [DOI] [Google Scholar]
  18. IUCN . 2024. “The IUCN Red List of Threatened Species. Version 2024‐2 [Computer Software].”
  19. Jiménez‐Franco, M. V. , Martínez‐Fernández J., Martínez J. E., Pagán I., Calvo J. F., and Esteve M. A.. 2018. “Nest Sites as a Key Resource for Population Persistence: A Case Study Modelling Nest Occupancy Under Forestry Practices.” PLoS One 13, no. 10: e0205404. 10.1371/journal.pone.0205404. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Jullien, M. , and Thiollay J.. 1996. “Effects of Rain Forest Disturbance and Fragmentation: Comparative Changes of the Raptor Community Along Natural and Human‐Made Gradients in French Guiana.” Journal of Biogeography 23, no. 1: 7–25. 10.1046/j.1365-2699.1996.00963.x. [DOI] [Google Scholar]
  21. Kouba, M. , Bartoš L., Bartošová J., Hongisto K., and Korpimäki E.. 2020. “Interactive Influences of Fluctuations of Main Food Resources and Climate Change on Long‐Term Population Decline of Tengmalm's Owls in the Boreal Forest.” Scientific Reports 10, no. 1: 20429. 10.1038/s41598-020-77531-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kuhn, M. 2008. “Building Predictive Models in R Using the Caret Package.” Journal of Statistical Software 28, no. 5: 1–26. 10.18637/jss.v028.i05.27774042 [DOI] [Google Scholar]
  23. Liao, Z. , Peng S., and Chen Y.. 2022. “Half‐Millennium Evidence Suggests That Extinction Debts of Global Vertebrates Started in the Second Industrial Revolution.” Communications Biology 5, no. 1: 1311. 10.1038/s42003-022-04277-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Liaw, A. , and Wiener M.. 2002. “Classification and Regression by randomForest.” R News 2, no. 3: 18–22. [Google Scholar]
  25. Lõhmus, A. 2005. “Are Timber Harvesting and Conservation of Nest Sites of Forest‐Dwelling Raptors Always Mutually Exclusive?” Animal Conservation 8, no. 4: 443–450. 10.1017/S1367943005002349. [DOI] [Google Scholar]
  26. Luczon, A. U. , Fontanilla I. K. C., Ong P. S., Basiao Z. U., Sumaya A. M. T., and Quilang J. P.. 2014. “Genetic Diversity of the Critically Endangered Philippine Eagle Pithecophaga jefferyi (Aves: Accipitridae) and Notes on Its Conservation.” Journal of Threatened Taxa 6, no. 10: 6335–6344. 10.11609/JoTT.o3748.6335-44. [DOI] [Google Scholar]
  27. Martínez‐Hesterkamp, S. , Rebollo S., Pérez‐Camacho L., García‐Salgado G., and Fernández‐Pereira J. M.. 2018. “Assessing the Ability of Novel Ecosystems to Support Animal Wildlife Through Analysis of Diurnal Raptor Territoriality.” PLoS One 13, no. 10: e0205799. 10.1371/journal.pone.0205799. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Martínez‐Ruiz, M. , Escalante P., and Renton K.. 2016. “Forest Cover Influences Territoriality of Collared Forest‐Falcons in a Modified Landscape of Tropical Moist Forest.” Journal of Raptor Research 50, no. 4: 404–415. 10.3356/JRR-16-12.1. [DOI] [Google Scholar]
  29. McClure, C. J. W. , Schulwitz S. E., Anderson D. L., et al. 2019. “Commentary: Defining Raptors and Birds of Prey.” Journal of Raptor Research 53, no. 4: 419. 10.3356/0892-1016-53.4.419. [DOI] [Google Scholar]
  30. McClure, C. J. W. , Westrip J. R. S., Johnson J. A., et al. 2018. “State of the World's Raptors: Distributions, Threats, and Conservation Recommendations.” Biological Conservation 227: 390–402. 10.1016/j.biocon.2018.08.012. [DOI] [Google Scholar]
  31. McPherson, S. C. , Brown M., and Downs C. T.. 2016. “Diet of the Crowned Eagle ( Stephanoaetus coronatus ) in an Urban Landscape: Potential for Human‐Wildlife Conflict?” Urban Ecosystems 19, no. 1: 383–396. 10.1007/s11252-015-0500-6. [DOI] [Google Scholar]
  32. Miranda, E. B. P. , Menezes J. F. S., Farias C. C. L., Munn C., and Peres C. A.. 2019. “Species Distribution Modeling Reveals Strongholds and Potential Reintroduction Areas for the World's Largest Eagle.” PLoS One 14, no. 5: e0216323. 10.1371/journal.pone.0216323. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Miranda, E. B. P. , Peres C. A., Carvalho‐Rocha V., et al. 2021. “Tropical Deforestation Induces Thresholds of Reproductive Viability and Habitat Suitability in Earth's Largest Eagles.” Scientific Reports 11, no. 1: 13048. 10.1038/s41598-021-92372-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Mozgeris, G. , Mörtberg U., Pang X.‐L., Trubins R., and Treinys R.. 2021. “Future Projection for Forest Management Suggests a Decrease in the Availability of Nesting Habitats for a Mature‐Forest‐Nesting Raptor.” Forest Ecology and Management 491: 119168. 10.1016/j.foreco.2021.119168. [DOI] [Google Scholar]
  35. Newton, I. 2010. Population Ecology of Raptors. Bloomsbury Publishing Plc. [Google Scholar]
  36. O'Bryan, C. J. , Allan J. R., Suarez‐Castro A. F., et al. 2022. “Human Impacts on the World's Raptors.” Frontiers in Ecology and Evolution 10: 624896. 10.3389/fevo.2022.624896. [DOI] [Google Scholar]
  37. O'Bryan, C. J. , Braczkowski A. R., Beyer H. L., Carter N. H., Watson J. E. M., and McDonald‐Madden E.. 2018. “The Contribution of Predators and Scavengers to Human Well‐Being.” Nature Ecology & Evolution 2, no. 2: 2. 10.1038/s41559-017-0421-2. [DOI] [PubMed] [Google Scholar]
  38. Papworth, S. K. , Rist J., Coad L., and Milner‐Gulland E. J.. 2009. “Evidence for Shifting Baseline Syndrome in Conservation.” Conservation Letters 2, no. 2: 93–100. 10.1111/j.1755-263X.2009.00049.x. [DOI] [Google Scholar]
  39. Rabor, D. 1968. The Present Status of the Monkey‐Eating Eagle, Pithecophaga jefferyi Ogilvie‐Grant, of the Philippines. International Union for Conservation of Nature and Natural Resources, Publications, New Series. [Google Scholar]
  40. Restrepo‐Cardona, J. S. , Echeverry‐Galvis M. Á., Maya D. L., Vargas F. H., Tapasco O., and Renjifo L. M.. 2020. “Human‐Raptor Conflict in Rural Settlements of Colombia.” PLoS One 15, no. 1: e0227704. 10.1371/journal.pone.0227704. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Rodríguez, B. , Rodríguez A., Lorenzo J. A., and Martínez J. M.. 2021. “Exotic Tree Plantations as Alternative Breeding Habitat for an Endemic Avian Predator.” Journal of Avian Biology 52, no. 4: jav.02527. 10.1111/jav.02527. [DOI] [Google Scholar]
  42. Rutz, C. , Bijlsma R. G., Marquiss M., and Kenward R. E.. 2006. “Population Limitation in the Northern Goshawk in Europe: A Review With Case Studies.” Studies in Avian Biology 31: 158–197. [Google Scholar]
  43. Saga, Ø. , and Selås V.. 2012. “Nest Reuse by Goshawks After Timber Harvesting: Importance of Distance to Logging, Remaining Mature Forest Area and Tree Species Composition.” Forest Ecology and Management 270: 66–70. 10.1016/j.foreco.2012.01.015. [DOI] [Google Scholar]
  44. Santander, F. , Alvarado O. S., and Estades C. F.. 2021. “Effect of Forest Cover on Raptor Abundance in Exotic Forest Plantations in Chile.” Ardeola 68, no. 2: 391–408. 10.13157/arla.68.2.2021.ra5. [DOI] [Google Scholar]
  45. Santangeli, A. , Lehtoranta H., and Laaksonen T.. 2012. “Successful Voluntary Conservation of Raptor Nests Under Intensive Forestry Pressure in a Boreal Landscape.” Animal Conservation 15, no. 6: 571–578. 10.1111/j.1469-1795.2012.00551.x. [DOI] [Google Scholar]
  46. Seaton, R. , Minot E. O., and Holland J. D.. 2013. “Home Range and Habitat Use of New Zealand Falcons ( Falco novaeseelandiae ) in an Exotic Plantation Forest During the Breeding Season.” Journal of Raptor Research 47, no. 3: 223–233. 10.3356/JRR-12-26.1. [DOI] [Google Scholar]
  47. Shaw, P. , Ogada D., Dunn L., et al. 2024. “African Savanna Raptors Show Evidence of Widespread Population Collapse and a Growing Dependence on Protected Areas.” Nature Ecology & Evolution 8, no. 1: 45–56. 10.1038/s41559-023-02236-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Suárez, S. , Balbontín J., and Ferrer M.. 2000. “Nesting Habitat Selection by Booted Eagles Hieraaetus pennatus and Implications for Management.” Journal of Applied Ecology 37, no. 2: 215–223. 10.1046/j.1365-2664.2000.00503.x. [DOI] [Google Scholar]
  49. Sutton, L. J. , Anderson D. L., Franco M., et al. 2021. “Geographic Range Estimates and Environmental Requirements for the Harpy Eagle Derived From Spatial Models of Current and Past Distribution.” Ecology and Evolution 11, no. 1: 481–497. 10.1002/ece3.7068. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Thiollay, J. M. 1989. “Area Requirements for the Conservation of Rain Forest Raptors and Game Birds in French Guiana.” Conservation Biology 3, no. 2: 128–137. 10.1111/j.1523-1739.1989.tb00065.x. [DOI] [Google Scholar]
  51. Thiollay, J.‐M. , and Meyburg B. U.. 1988. “Forest Fragmentation and the Conservation of Raptors: A Survey on the Island of Java.” Biological Conservation 44, no. 4: 229–250. 10.1016/0006-3207(88)90018-3. [DOI] [Google Scholar]
  52. Thomas, H. , Simmonds J. S., Ward M., Eyre T. J., and Maron M.. 2025. “The Value of Regrowth Forests and Woodlands for Threatened Fauna Species.” Biological Conservation 307: 111131. 10.1016/j.biocon.2025.111131. [DOI] [Google Scholar]
  53. United Nations Environment Programme . 2022. Kunming‐Montreal Global Biodiversity Framework. United Nations Environment Programme. [Google Scholar]
  54. Whytock, R. C. , Buij R., Virani M. Z., and Morgan B. J.. 2016. “Do Large Birds Experience Previously Undetected Levels of Hunting Pressure in the Forests of Central and West Africa?” Oryx 50, no. 1: 76–83. 10.1017/S0030605314000064. [DOI] [Google Scholar]
  55. Winkler, K. , Fuchs R., Rounsevell M., and Herold M.. 2021. “Global Land Use Changes Are Four Times Greater Than Previously Estimated.” Nature Communications 12, no. 1: 2501. 10.1038/s41467-021-22702-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Ziegler, S. , Fa J. E., Wohlfart C., Streit B., Jacob S., and Wegmann M.. 2016. “Mapping Bushmeat Hunting Pressure in Central Africa.” Biotropica 48, no. 3: 405–412. 10.1111/btp.12286. [DOI] [Google Scholar]
  57. Zuluaga, S. , and Echeverry‐Galvis M. Á.. 2016. “Domestic Fowl in the Diet of the Black‐and‐Chestnut Eagle ( Spizaetus isidori ) in the Eastern Andes of Colombia: A Potential Conflict With Humans?” Ornitología Neotropical 27: 113–120. 10.58843/ornneo.v27i0.86. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Figure S1: Permutation‐based feature importance (variable importance) of a random forest model for predicting IUCN threat status of forest‐dependent raptors.

Table S1: Forest dependency categories and descriptions from Birdlife International used in this study.

Table S2: Mean, standard deviation, minimum, and maximum area (km2) of forest change within the geographic distributions of 369 forest‐dependent raptor species.

Table S3: Mean area (km2) and standard deviation (SD) of forest change across five different drivers of forest loss within the geographic distributions of 369 forest‐dependent raptor species.

Table S4: Mean, standard deviation, minimum, and maximum area (km2) of forest change within the geographic distributions of 369 forest‐dependent raptor species, organized by family.

Table S5: Mean, standard deviation, minimum, and maximum area (km2) of forest change within the geographic distributions of 369 forest‐dependent raptor species, organized by raptor body size.

Table S6: Mean and standard deviation area (km2) of forest change within the geographic distributions of 369 forest‐dependent raptor species, organized by raptor body size and driver of loss.

Table S7: Mean, standard deviation, minimum, and maximum area (km2) of forest change within the geographic distributions of 369 forest‐dependent raptor species, organized by IUCN status and driver of loss.

Table S8: Mean and standard deviation area (km2) of forest change within the geographic distributions of 369 forest‐dependent raptor species, organized by driver of loss and IUCN status.

Data S1: gcb70603‐sup‐0001‐FigureS1‐TableS1‐S8‐DataS1.docx.

GCB-31-e70603-s001.docx (386.2KB, docx)

Data Availability Statement

The supplementary data file can be obtained from https://doi.org/10.5281/zenodo.17440465. Spatial data for analyzing forest change and drivers of loss can be obtained from Global Forest Watch at https://data.globalforestwatch.org/documents/941f17325a494ed78c4817f9bb20f33a/explore and https://data.globalforestwatch.org/documents/ff304784a9f04ac4a45a40f60bae5b26/about. The forest gain data can be obtained from https://code.earthengine.google.com/?asset=projects/ee‐tufangbobo/assets/ForestGain/forestGainFinal. The HILDA+ data can be obtained from https://doi.pangaea.de/10.1594/PANGAEA.921846. Raptor geographic distributions can be requested from Birdlife International at https://datazone.birdlife.org/contact‐us/request‐our‐data. All spatial analyses were performed in ArcGIS (v10.8) and Google Earth Engine (GEE). The random forest algorithm was conducted in R (v4.5.0). The GEE code for analyzing baseline forest cover within species' ranges for the year 2000 can be found here https://code.earthengine.google.com/72b34c0ca6547bf89d40c8f5d69c98a6, for forest loss can be found here https://code.earthengine.google.com/9c8f73aac1635a3857327b849e940550, for forest gain can be found here https://code.earthengine.google.com/?asset=projects/ee‐tufangbobo/assets/ForestGain/forestGainFinal, and for the drivers of forest loss can be found here https://code.earthengine.google.com/45c32a82aa39025848a82f47aea59393.

Articles from Global Change Biology are provided here courtesy of Wiley

RESOURCES