Introduction
The Amazon biome contains naturally fragmented savanna enclaves embedded within a predominantly forested landscape, forming highly heterogeneous ecosystems with distinct environmental conditions and biogeographic histories (Carvalho & Mustin Reference Carvalho and Mustin2017). Although smaller and more isolated than the core South American savannas, these Amazonian savannas harbour unique assemblages of flora and fauna, including exclusive and habitat-specialist species (Barbosa et al. Reference Barbosa, Campos, Pinto and Fearnside2007, Plotkin & Riding Reference Plotkin and Riding2011). These enclaves are composed of a mosaic of open vegetation formations associated with both Amazonian and Cerrado biotas, resulting in singular ecological communities shaped by environmental filtering, climatic stability and long-term landscape isolation (Silva & Bates Reference Silva and Bates2002, Miranda et al. Reference Miranda, Absy and Rebêlo2003, Ratter et al. Reference Ratter, Bridgewater and Ribeiro2003, Aleixo et al. Reference Aleixo, Rossetti, Moreno, Gabriel, Valdes, Dino and de Toledo2010, Werneck Reference Werneck2011).
Despite their biological importance, Amazonian savannas remain among the least studied and least protected ecosystems in South America (Carvalho & Mustin 2017, Carvalho et al. Reference Carvalho, Resende, Barlow, Moura, Maciel and Alves-Martins2023). There are substantial geographic and taxonomic gaps in Amazonian ecological research, particularly in open vegetation ecosystems and transitional habitats (Oliveira et al. Reference Oliveira, Soares-Filho, Paglia, Brescovit, de Carvalho and Silva2017, Carvalho et al. Reference Carvalho, Resende, Barlow, Moura, Maciel and Alves-Martins2023). At the same time, these enclaves are increasingly threatened by deforestation, agricultural expansion, infrastructure development, uncontrolled fires and land-use conversion (Barbosa et al. Reference Barbosa, Campos, Pinto and Fearnside2007, Hilário et al. Reference Hilário, Toledo, Mustin, Castro, Costa-Neto and Kauano2017, Carvalho et al. Reference Carvalho, Mustin, Hilário, Vasconcelos, Eilers and Fearnside2019). Such pressures intensify habitat degradation and fragmentation, altering vegetation structure, ecological connectivity and species persistence across naturally isolated landscapes (Dirzo et al. Reference Dirzo, Young, Galetti, Ceballos, Isaac and Collen2014, Newbold et al. Reference Newbold, Hudson, Hill, Contu, Lysenko and Senior2015, Lovejoy & Nobre Reference Lovejoy and Nobre2018).
Small non-volant mammals, particularly marsupials (Didelphimorphia) and rodents (Rodentia), are among the vertebrate groups most strongly associated with habitat structure and environmental heterogeneity, making them valuable indicators of environmental change and habitat quality (Vieira & Monteiro-Filho Reference Vieira and Monteiro-Filho2003, Lambert et al. Reference Lambert, Malcolm and Zimmerman2005). In Brazil, 339 species of non-volant small mammals are currently recognized (Abreu et al. Reference Abreu, Casali, Costa-Araújo, Garbino, Libardi and Loretto2024), many of which occur within the Amazon biome. In Amazonian savanna enclaves, these assemblages often include a combination of open-habitat species associated with the Cerrado and species typically associated with adjacent Amazonian forests, forming unique ecological communities (Pires & Prance Reference Pires, Prance, Prance and Lovejoy1985, Fine & Bruna Reference Fine and Bruna2016). Although Amazonian savanna enclaves generally harbour fewer species than adjacent forest ecosystems, they often support distinct taxonomic and functional assemblages, including habitat specialists and geographically restricted species (Oliveira et al. Reference Oliveira, Soares-Filho, Paglia, Brescovit, de Carvalho and Silva2017). Beyond their taxonomic diversity, non-volant small mammals perform essential ecological functions, including seed dispersal, pollination, soil bioturbation and energy transfer across trophic networks (Lacher et al. Reference Lacher, Murphy, Rogan, Smith, Upham, Wilson, Lacher and Mittermeier2016). Several species also respond rapidly to habitat disturbance, making them important indicators of environmental integrity and landscape transformation. Nevertheless, knowledge regarding their distribution, ecology and conservation status in Amazonian savanna enclaves remains limited, especially when compared to adjacent forest ecosystems (Carvalho & Mustin 2017, Carvalho et al. Reference Carvalho, Resende, Barlow, Moura, Maciel and Alves-Martins2023). Consequently, little is known about whether these enclaves function as priority areas for endemic or functionally distinct assemblages of non-volant small mammals, or how ongoing anthropogenic pressures may affect the conservation value of these ecosystems (Carvalho & Mustin 2017, Carvalho et al. Reference Carvalho, Resende, Barlow, Moura, Maciel and Alves-Martins2023).
Protected areas (PAs) and Indigenous territories (ITs) are widely recognized as fundamental strategies for biodiversity conservation in the Brazilian Amazon. Together, these areas cover approximately a third of the Brazilian territory, and they play an important role in reducing deforestation and maintaining ecosystem functions (Soares-Filho et al. Reference Soares-Filho, Moutinho, Nepstad, Anderson and Rodrigues2010, Rorato et al. Reference Rorato, Picoli, Verstegen, Camara, Silva Bezerra and Escada2021). Under the Brazilian National System of Conservation Units (SNUC), PAs encompass multiple objectives, including biodiversity conservation, scientific research, ecosystem maintenance and sustainable resource use. Nevertheless, conservation planning and management often prioritize charismatic or better-known taxa, potentially neglecting less conspicuous groups such as small mammals (Galetti et al. Reference Galetti, Giacomini, Bueno, Bernardo, Marques and Bovendorp2009, Di Marco et al. Reference Di Marco, Venter, Possingham and Watson2017, Oliveira et al. Reference Oliveira, Soares-Filho, Paglia, Brescovit, de Carvalho and Silva2017). Furthermore, many Amazonian savanna enclaves remain outside strictly protected regions and are increasingly exposed to anthropogenic degradation.
In this context, gap analysis combined with species distribution models (SDMs) has become an important tool for identifying conservation priorities, evaluating species representativeness within PA networks and detecting regions of high environmental suitability that remain unprotected (Rodrigues et al. Reference Rodrigues, Andelman, Bakarr, Boitani, Brooks and Cowling2003, Elith & Leathwick Reference Elith and Leathwick2009, Franklin Reference Franklin2010, Booth et al. Reference Booth, Nix, Busby and Hutchinson2014, Carvalho et al. Reference Carvalho, Resende, Barlow, Moura, Maciel and Alves-Martins2023, Guedes et al. Reference Guedes, Diniz-Filho and Moura2024, Pena et al. Reference Pena, Silva, Alexandre and Vieira2026). These approaches are particularly relevant for poorly sampled regions and taxonomic groups, where field data remain spatially limited (Hortal et al. Reference Hortal, de Bello, Diniz-Filho, Lewinsohn, Lobo and Ladle2015). Nevertheless, large-scale assessments focusing specifically on non-volant small mammals in Amazonian savanna enclaves are still lacking.
Here, we assessed the conservation importance of Amazonian savanna enclaves for non-volant small mammals in the Brazilian Amazon. Specifically, we aimed to: (1) evaluate whether PAs and ITs overlap with priority regions for the conservation of small non-volant mammals; (2) assess the contribution of Amazonian savanna enclaves to the conservation of this group; and (3) determine whether these priority enclaves remain preserved or are associated with vegetation degradation.
Methods
Species list, classification and distribution maps used
We utilized the updated species list of non-volant small mammals (orders Didelphimorphia and Rodentia) exclusive to the Brazilian Amazon biome (Fig. 1) of Pena et al. (Reference Pena, Silva, Alexandre and Vieira2026). This dataset was carefully curated to exclude synonymized species, dubious records and exotic taxa. General information regarding taxonomic determinations, dietary classifications (trophic guild), behavioural characteristics (behavioural guild) and locomotion characteristics was compared with the most recent list of Brazilian mammals (Abreu et al. Reference Abreu, Casali, Costa-Araújo, Garbino, Libardi and Loretto2024; see Appendix S1 Table S2). Additionally, the list incorporates International Union for Conservation of Nature (IUCN) threat classifications for both marsupials and rodents, providing a robust foundation for evaluating conservation priorities. We provide supplementary material with the data included in the study (Appendix S1 Tables S2 & S3), indicating how they were categorized in relation to trophic and behavioural groups, locomotion and other relevant characteristics.
Diagrammatic map of the study area, highlighting the different uses and land covers existing in the Amazon. The upper right corner highlights the location of the Brazilian Amazon in green. In the lower right corner, the areas protected by Brazilian legislation are highlighted: dark blue represents the strictly protected areas (SPAs); red represents the sustainable use areas (SUAs); and black squares represent the Indigenous territories (ITs).
Figure 1. Long description
A map of the Amazon region showing various land uses and covers. The map includes state boundaries, strictly protected areas (SPAs) in dark blue, sustainable use areas (SUAs) in red, and Indigenous territories (ITs) marked with black squares. Different land covers are represented by various colors: savannah in orange, forest in green, forest plantation in dark green, pasture in light brown, rivers or lakes in blue, and soybean fields in yellow. Mining areas are indicated by black stars. The upper right corner highlights the Brazilian Amazon in green, and the lower right corner emphasizes areas protected by Brazilian legislation.
For SDMs, we combined the four algorithms MaxEnt (MXE), Random Forest (RDF), Support Vector Machine (SVM) and Gaussian–Bayesian (GAU). Model inputs included occurrence records obtained from multiple sources: species occurrence data were obtained from digitized mammal collections available through the Brazilian Biodiversity Information System (SiBBr), SpeciesLink, the Global Biodiversity Information Facility (GBIF) and the American Museum of Natural History (AMNH). To complement these records, a systematic literature search was performed in ISI Web of Knowledge, Scopus and Google Scholar using combinations of the keywords ‘small mammals’, ‘non-volant’, ‘Rodentia’, ‘rodents’, ‘Didelphimorphia’, ‘marsupials’ and ‘Amazon’. Environmental predictors consisted of 19 bioclimatic variables (at ∼9.4 × 9.4 km resolution) from WorldClim supplemented with an altitudinal layer encompassing the entire Neotropical region. These data are derived from monthly climate averages sampled between 1970 and 2000 and are part of WorldClim version 2.1 (Fick & Hijmans Reference Fick and Hijmans2017). They are typically used for SDMs to assess potential species distributions (Lee et al. Reference Lee S, Yoo Y and Lee J2026) and are also considered good models for species conservation planning (Pimenta et al. Reference Pimenta, Andrade, Fernandes, Melo Amboni, Almeida and Bello Soares2022). To address multicollinearity among the variables, we performed a principal component analysis (PCA) (Legendre & Legendre Reference Legendre and Legendre2012), retaining the principal components that together explained at least 95% of the total variance in the dataset (Júnior & Nóbrega Reference Júnior and Nóbrega2018); these axes were subsequently used as environmental predictors in the models. Only species with more than five occurrence records were modelled, and spatial autocorrelation was controlled. Absence data were generated where required. Models were built with and without spatial restrictions, and binary maps were derived using Jaccard thresholds (for details on the small mammal potential distribution model’s protocol, see Appendix S1).
The final ensemble model, representing mean suitability across all algorithms, incorporated only predictions that surpassed the mean Jaccard threshold for each species, with additional spatial constraints implemented to minimize overprediction artifacts. Our modelling protocol strictly adhered to the methodology outlined in Pimenta et al. (Reference Pimenta, Andrade, Fernandes, Melo Amboni, Almeida and Bello Soares2022). Binary maps, evaluation metrics, thresholds and detailed methodological procedures can be accessed in Pena et al. (Reference Pena, Silva, Alexandre and Vieira2026).
Conservation priority areas and the importance of protected areas and Indigenous territories
To identify priority areas for the conservation of small non-volant mammals, we used Zonation software (Moilanen Reference Moilanen2007), which is based on the principles of complementarity, irreplaceability and representativeness (Moilanen Reference Moilanen2007, Lehtomaki & Moilanen Reference Lehtomäki and Moilanen2013, Di Minin et al. Reference Di Minin, Veach, Lehtomaki, Montesino and Moilanen2014). In the Zonation framework, areas of lesser importance for conservation were assigned the lowest scores (closer to 0), while the most valuable areas for biodiversity received the highest scores (closer to 1). For this study, the Additive Benefit Function algorithm was applied, which is suitable for analyses involving many species (Lehtomaki & Moilanen Reference Lehtomäki and Moilanen2013). This quantitative approach prioritized conservation areas based on species distribution maps, generating a single output map (matching the resolution of our input data). We applied this methodology comprehensively across multiple taxonomic and ecological groupings: all marsupials; all rodents; each marsupial trophic guild; each rodent trophic guild; each marsupial behavioural guild; each rodent behavioural guild; and each IUCN threat classification category (see Appendix S1).
To assess whether PAs (strictly protected areas (SPAs) and sustainable use areas (SUAs)) and ITs effectively encompass regions of high importance for the conservation of small non-volant mammals in the Brazilian Amazon biome, we analysed the importance values derived from Zonation, comparing cells inside and outside the PAs and calculating the average importance of protected and unprotected cells. The statistical significance of these values was evaluated using Monte Carlo randomization tests (10 000 iterations). The complete methodology used in Zonation and how the randomizations were performed is explained in de Oliveira et al. (Reference De Oliveira, Alexandre, Pena, Correia, Luz and Bernardi2025).
This analytical framework was applied across all Zonation output maps (species, guilds and IUCN threat categories). PA boundaries were obtained from official sources: federal and state-level SPAs/SUAs from the Ministério do Meio Ambiente (MMA) spatial database and ITs from Fundação Nacional dos Povos Indígenas (FUNAI), ensuring our analysis incorporated the most authoritative spatial data available for the Brazilian Amazon.
Importance of savannas
To evaluate the conservation role of savanna enclaves for non-volant small mammals in the Amazon forest, we analysed the Zonation-generated importance maps by classifying cells into savanna enclaves and non-enclave areas. We overlaid the importance maps with historical savanna distribution data and classified cells as enclaves if ≥50% of their area fell within the historical savanna range, with all other cells considered non-enclaves. For each prioritization map produced by Zonation, we calculated the average importance value for both enclave and non-enclave cells and assessed statistical significance using Monte Carlo randomization tests (10 000 iterations) following the same procedure applied to our PA analysis. The historical distribution data for Amazonian savannas were obtained from the TerraBrasilis platform (https://terrabrasilis.dpi.inpe.br/downloads/), ensuring our enclave classification was based on authoritative vegetation records. This approach allowed us to quantitatively compare the conservation value of these unique savanna habitats against surrounding forest areas.
Moreover, we examined the relationship between habitat integrity and conservation value by correlating Zonation-derived importance scores with remaining natural vegetation cover. Using MapBiomas collection 9 (Souza et al. Reference Souza, Shimbo, Rosa, Parente, Alencar and Rudorff2020), we created binary land-cover maps where anthropogenic areas (mining, pasture, etc.) were coded as 0 and natural vegetation (including forests and gallery forests) were coded as 1. These maps were overlaid with historical savanna distribution data, classifying cells as enclaves if ≥50% fell within historical savanna boundaries. We then superimposed this classification onto a 10-km × 10-km grid (matching our importance maps’ resolution) to calculate the percentage of natural vegetation in each cell. By converting both datasets to raster format with identical resolution and extent, we performed Pearson correlation analyses between vegetation-cover percentages and conservation importance values across all Zonation output maps. A positive correlation indicated well-preserved vegetation in high-importance areas, while a negative correlation suggested loss of vegetation in high-importance areas or vegetation presence in low-importance areas.
Results
Marsupials
Priority conservation areas for all marsupial species were highlighted in the central part of the state of Pará and eastern Amazonas (Fig. 2). SPAs and ITs were highly important for the conservation of these species, and the savanna areas were also of great importance for marsupial species; however, the areas of importance were within already-degraded areas (Appendix S1 Table S1).
Diagrammatic map of priority areas for total marsupials (Didelphimorphia: Didelphidae) in the Brazilian Amazon, showing protected areas and Amazonian savannas. IT = Indigenous territory; SPA = strictly protected area; SUA = sustainable use area.
For all species of frugivorous/omnivorous marsupials (Appendix S1 Fig. S1A), ITs presented low conservation importance; however, areas outside the PAs (SPAs and SUAs) and ITs had some importance (Appendix S1 Table S1). For insectivorous/omnivorous marsupial species (Appendix S1 Fig. S1B), the SPAs and ITs were of great importance for conservation (Appendix S1 Table S1). Similarly, savanna areas were of great importance for the marsupial species of both feeding guilds; however, these were within areas that are already degraded (Appendix S1 Table S1).
All PAs (SPAs and SUAs) and ITs showed high conservation importance for arboreal marsupial species and scansorial marsupial species (Appendix S1 Fig. S2A,B & Table S1). For terrestrial marsupials, only ITs showed importance for conservation (Appendix S1 Fig. S2C & Table S1). Arboreal and terrestrial species showed a positive correlation; that is, the areas of importance were within non-degraded areas. However, the same was not observed for scansorial species (Appendix S1 Table S1).
None of the PAs showed relevant importance for the conservation of species classified as Least Concern (LC) by the IUCN; these were perhaps within already-degraded areas (Appendix Fig. S3A & Table S1). The SPAs and ITs presented considerable importance for species that do not fit into any threat classification (Appendix S1 Fig. S3B & Table S1).
Rodents
The priority conservation areas for the conservation of total rodent species (Fig. 3) highlighted areas with high conservation importance in several parts of the states of the Amazon biome, such as in the western part of the state of Roraima and in a large part of the state of Amapá. However, no PAs (SPAs or SUAs) or ITs showed importance for the conservation of the species (Appendix S1 Table S1). The savanna areas were highly important for conserving the studied small mammal species, particularly the arboreal and terrestrial marsupials, as well as the rodent groups (frugivores/granivores, frugivores/seed-predators, arboreal and scansorial), with importance values ranging from 0.837 to 0.876. Generally, these areas were positively associated with natural vegetation cover and native savanna presence. However, arboreal rodents were an exception, displaying a negative correlation, which indicates that their priority areas are located in regions with low natural cover (Appendix S1 Table S2).
Diagrammatic map of priority areas for total rodents (Rodentia: Caviidae, Cricetidae, Etenomyidae and Echimyidae) in the Brazilian Amazon, showing the protected areas and Amazonian savannas. IT = Indigenous territory; SPA = strictly protected area; SUA = sustainable use area.
For folivorous rodent species, SPAs, ITs and savanna areas were important for conservation (Appendix S1 Fig. S4A & Table S1). The same pattern was observed for frugivore/folivore rodents and for frugivore/omnivore rodents; however, both savanna and non-savanna areas were important for their conservation, although this correlation was negative (Appendix S1 Figs S4B & S4D & Table S1). None of the PAs (SPAs or SUAs) or ITs were important for the conservation of the frugivore/granivore species (Appendix S1 Fig. S4C & Table S1). For frugivore/seed predator species, only ITs showed low relevant importance for species conservation (Appendix S1 Fig. S4E & Table S1). For both herbivores and insectivores/omnivores, PAs showed high conservation importance, as did savanna areas still within vegetated areas, both showing positive correlations (Appendix S1 Fig. S4F,G & Table S1).
Arboreal rodent species showed Importance Index values significantly higher than those expected by chance across the evaluated protected areas (SPAs, SUAs and ITs), ranging from 0.837 to 0.885 (Appendix S1 Fig. S5A & Table S1). Nevertheless, none of these species achieved high conservation priority according to the Importance Index ranking. For semiaquatic, scansorial and semifossorial species, PAs (SPAs and SUAs) and ITs showed high conservation importance, although only for scansorial and semifossorial species were the PAs still within vegetated areas (Appendix S1 Fig. S5B–D & Table S1). For terrestrial species, no PA or IT showed conservation importance; it is worth noting that the observed importance was greater than the randomized importance, ranging from 0.862 and 0.837 (Appendix S1 Fig. S5E & Table S1). The savanna areas were important for conservation, but the correlation was negative.
PAs (SPAs and SUAs) and ITs showed importance for Data Deficient (DD) species, with values ranging from 0.872 to 0.901 among the PAs evaluated (Appendix S1 Fig. S6A & Table S1); both savanna and non-savanna areas showed high importance for these species, but the correlation was negative. None of the PAs evaluated showed importance for the conservation of species of Least Concern (LC; Appendix S1 Fig. S6B & Table S1). PAs (SPAs and SUAs) and ITs showed high importance for the conservation of the species without an applicable IUCN threat classification, so much so that the correlation was positive; that is, these areas of importance were still within vegetated areas.
Discussion
Our study reveals important conservation gaps for non-volant small mammals in Amazonian savanna enclaves, highlighting a mismatch between areas of high conservation importance and current protection efforts. Although ITs and SPAs contribute substantially to the conservation of several taxonomic and ecological groups, many priority areas identified by our analyses remain outside PA networks. This pattern is particularly concerning in Amazonian savannas, one of the least protected ecosystems in the Amazon biome (Carvalho & Mustin Reference Carvalho and Mustin2017).
The relationship between importance and vegetation cover was not consistent across all taxonomic groups and conservation categories. Approximately half of the correlations were negative, with contrasting responses observed between marsupials and rodents. For total species richness, habitat importance was negatively correlated with vegetation cover for marsupials but was positively correlated for rodents, indicating distinct responses to landscape structure and habitat availability. Despite this variation, several areas identified as conservation priorities were associated with lower levels of native vegetation cover, suggesting that ecologically important habitats are already experiencing landscape modification.
While our correlation analysis revealed a negative trend between conservation priority and vegetation cover in some groups, the low coefficients indicate that this relationship has limited predictive power at the landscape scale. Rather than relying on this weak correlation, we base our main conservation recommendations on two lines of evidence: (1) the disproportionately high Zonation priority scores assigned to savanna enclaves, which identify them as irreplaceable habitats for endemic and DD species, and (2) the grossly inadequate legal protection of these same enclaves, particularly in Amapá, where <1% falls under any strict protection category. These findings, independently of the correlation analysis, underscore the need for expanded conservation action in Amazonian savannas. We therefore interpret the weak negative correlation as a modest supporting observation but not as a primary driver of our conclusions.
These findings are especially relevant in Amazonian savanna enclaves, where ongoing agricultural expansion, infrastructure development, altered fire regimes and habitat fragmentation continue to threaten native biodiversity (Barbosa et al. Reference Barbosa, Campos, Pinto and Fearnside2007, Carvalho & Mustin 2017). In Amapá, for example, most savanna areas occur outside SPAs and remain vulnerable to anthropogenic pressures. Habitat fragmentation can reduce landscape connectivity, limit dispersal and recolonization processes and increase the vulnerability of species with specialized ecological requirements (Ewers & Didham Reference Ewers and Didham2006, Dirzo et al. Reference Dirzo, Young, Galetti, Ceballos, Isaac and Collen2014). Consequently, the persistence of many small-mammal populations may depend on the maintenance of remaining habitat patches and landscape connectivity.
PAs showed significant conservation importance for most ecological guilds and taxonomic groups analysed, reinforcing their role in maintaining the diversity of non-volant small mammals in Amazonian savannas. Although the magnitude of this importance varied among guilds, the overall pattern suggests that PAs provide suitable conditions for species with different ecological strategies and habitat-use patterns. Nevertheless, variation among the groups indicates that conservation effectiveness is not uniform, emphasizing the need for a representative network capable of encompassing the diversity of habitats and ecological functions associated with these species.
Our analysis of IUCN categories revealed that PAs exhibited particularly high conservation importance for DD rodent species. DD species are not considered threatened; rather, they lack sufficient information for a reliable assessment of extinction risk (IUCN 2026). In the Amazon, many small mammals remain poorly known due to limited sampling, unresolved taxonomy and substantial geographic knowledge gaps (Patton et al. Reference Patton, Pardiñas and D’Elía2015, Costa-Araújo et al. Reference Costa-Araújo, Rossi, Pavan, Schneider, Sampaio, Pieczarka and Silva2020). Therefore, the high conservation importance of PAs for DD rodents suggests that these areas may encompass habitats that are critical for species whose distributions and ecological requirements remain inadequately documented. This finding reinforces the importance of PAs as reservoirs not only of known biodiversity but also of poorly studied taxa requiring further research and monitoring.
PAs are established through complex ecological, social, political and economic processes, and their spatial distribution does not necessarily reflect biodiversity patterns alone. Nevertheless, conservation planning has historically relied on better-known vertebrate groups, while less conspicuous taxa frequently remain underrepresented in conservation assessments and planning frameworks (Rodrigues et al. Reference Rodrigues, Andelman, Bakarr, Boitani, Brooks and Cowling2004, Jenkins et al. Reference Jenkins, Pimm and Joppa2013). This issue may be particularly relevant in Amazonian savanna enclaves, where several areas identified as highly important for non-volant small mammals currently fall outside PA networks.
Although PAs and ITs remain among the most effective mechanisms for biodiversity conservation in the Amazon, important knowledge gaps persist regarding the distribution and conservation status of many small-mammal species. These limitations hinder the identification of priority areas and the development of effective conservation strategies. Expanding biodiversity inventories, improving taxonomic knowledge and strengthening long-term monitoring programmes within PAs and ITs are therefore essential to support evidence-based conservation planning.
Our findings are intrinsically linked to and build substantially upon prior works (e.g. Pena et al. Reference Pena, Silva, Alexandre and Vieira2026), which identified broad conservation gaps for non-volant small mammals based on species richness and the proportion of potential distribution areas falling within PAs and ITs via traditional gap analysis. While that earlier study quantified the extent of protection coverage (i.e., how much of each species’ distribution is legally safeguarded), it did not evaluate the internal hierarchy of conservation value across the landscape or the current integrity of those habitats. Here, we advance this framework by employing Zonation spatial prioritization to rank cells according to their additive conservation importance, moving beyond static coverage metrics to pinpoint where conservation efforts are most urgently needed. Crucially, we incorporate a novel degradation assessment by directly correlating these Zonation-derived priority scores with actual vegetation cover derived from MapBiomas. This integration allows us to determine the highest-priority areas, particularly in Amazonian savanna enclaves – regions with severe habitat loss, a critical dynamic that our previous analysis could not reveal. Therefore, while our earlier paper highlighted the extent of the protection deficit (the static conservation gap), the current study exposes the intensity of the threat occurring within irreplaceable priority zones (the dynamic degradation risk). Taken together, both studies provide a comprehensive and complementary conservation roadmap: the first establishes what and where to protect based on species distribution, while the second adds details regarding the essential urgency of when and how to act by linking priority areas to their current degradation status. These findings reinforce the need for expanded PA networks, habitat restoration and the integration of Indigenous-led management specifically tailored to the under-protected and rapidly degrading savanna enclaves of the Brazilian Amazon.
Overall, our findings demonstrate that the current PA network plays a fundamental role in conserving non-volant small mammals in the Brazilian Amazon, but important gaps remain, particularly in Amazonian savanna enclaves. The limited protection afforded to these ecosystems, combined with increasing anthropogenic pressures, highlights the urgent need to expand conservation efforts, to strengthen the management of existing PAs and ITs and to incorporate poorly known taxa into future conservation planning. Without such actions, unique components of Amazonian biodiversity may remain inadequately protected despite their high conservation value.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S037689292610054X.
Data accessibility
All data are in the article.
Acknowledgements
SAP and RJRA thank the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – Brasil (CAPES; Finance Code 001) for their doctoral fellowships. TBV thanks the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for a Research Productivity Grant (Process No. 319204/2025-1). We also thank the UFPA/EMBRAPA Postgraduate Program in Ecology and the members of the IEPA Mastozoology Laboratory and the ChiruXingu Group for their assistance in interpreting the data and for the friendships forged during this study.
Author contributions
Simone Almeida Pena: writing of the original version; conceptualization or the acquisition, analysis and interpretation of data. Cláudia Regina Silva: editing and review for important intellectual content. Rafaela Jemely Rodrigues Alexandre: analysis and review for important intellectual content and interpretation of data. Letícia Lima Corrêa: Analysis and review for important intellectual content and interpretation of data. Thiago Bernardi Vieira: supervision; data design and analysis; substantial contributions to the conception and design of the work.
Financial support
The authors declare that no funds, grants or other support were received during the preparation of this manuscript.
Competing interests
The authors have no relevant financial or non-financial interests to disclose.
Ethical standards
Not applicable.
