Most tropical forests are experiencing high rates of deforestation, negatively affecting biodiversity and ecosystems processes (Theobald et al., 2020). As a result, an increasing proportion of tropical landscapes has been transformed into heterogeneous mosaics, where forest remnants are surrounded by anthropogenic land covers (e.g., agricultural and urban areas; Melo et al., 2013). Understanding how species respond to changes in landscape composition, namely the types and proportions of each land cover, is fundamental to inform public policies and guide biodiversity conservation in these human-modified landscapes (Arroyo-Rodriguez et al., 2020; McGarigal et al., 2023). Among these changes, forest loss is one of the most important means of landscape modification, as it reduces resource availability, disrupts ecological functions and drives substantial biodiversity loss. These effects, however, are partially mitigated when the surrounding anthropogenic matrix retains forest habitats, which are key to maintain forest-dwelling species (Arroyo-Rodriguez et al., 2020). For instance, both old-growth and secondary forests, as well as scattered trees, can provide a variety of resources that organisms may use throughout the landscape (Arce-Peña et al., 2022; Larson and Sander, 2024). Although old-growth and secondary forests differ in structure and composition, they can play complementary roles in providing shelter, food and connectivity for wildlife. In fact, as each land cover type may offer different complementary and/or supplementary roles in providing shelter, food and connectivity for wildlife resources (Dunning et al., 1992; Fahrig et al., 2011), it is reasonable to expect that landscapes with more heterogeneous matrices should support a higher number of species when compared to landscapes with more homogeneous matrices (Fahrig et al., 2011; Willig et al., 2019). Therefore, assessing the relative importance of landscape composition and heterogeneity is critical to better evaluate species responses of ecologically relevant taxa such as bats and rodents.

Bats and small rodents are known to respond in complex ways to landscape changes (Presley et al., 2019). For example, studies have shown that the taxonomic diversity and community composition of both groups can be shaped by landscape composition (Arroyo-Rodríguez et al., 2016; Arce-Peña et al., 2022; Morales-Díaz et al., 2019; Carvalho et al., 2020; Paniccia et al., 2022). However, to date the effects of landscape composition on the functional dimension of bat and small rodent diversity remain poorly understood (Farneda et al., 2024; Paniccia et al., 2022). Functional diversity measures the variation in ecological traits within assemblages, including traits that are critical to species’ performance in specific habitats (Cadotte et al., 2011). In tropical environments, for instance, bats and small rodents exhibit a wide range of functional traits that underpin their essential ecological roles, such as herbivory, predation, soil aeration, pollination, and seed dispersal (Kunz et al., 2011; Lacher et al., 2019; Presley et al., 2019; Godó et al., 2022). Therefore, it is crucial to assess their functional response to landscape changes, which can provide valuable and complementary information for preserving these taxa and their ecological functions in human-modified landscapes (Magioli et al., 2016; Paniccia et al., 2022). However, it is important to note that, depending on the species’ vagility, each taxon may be affected by landscape compositional variables measured at different spatial scales (Miguet et al., 2016). Hence, although bats and small rodents interact with compositional variables of the surrounding landscape, the scale at which such interaction is strongest (the scale of effect) will vary in association with different traits, among which vagility and specialization may play a significant role (see Miguet et al., 2016).

The Yucatán peninsula, Mexico, is a biodiversity hotspot with ≈60 bat species and 15 small rodent species (hereafter referred only as rodents; Sosa-Escalante et al., 2013; Zaragoza Quintana et al., 2016). However, this region currently presents one of the highest deforestation rates in Mexico, having lost nearly 286,000 ha of forest cover in the last four years (Ellis et al., 2023). In this study we assessed the influence of four landscape compositional variables (i.e., forest cover, urban cover, agricultural cover, and compositional heterogeneity) on the taxonomic and functional diversity of bats and rodents in three regions with different patterns of land-use change within the Yucatán peninsula.

Following the predictions postulated by the landscape complementation and supplementation hypotheses (Dunning et al., 1992; Fahrig et al., 2011), we expect that forest loss and increasing urban and agricultural cover will decrease the taxonomic and functional diversity of bats and rodents. Yet, as the provision of complementary and/or supplementary resources can be higher in more heterogeneous landscapes, the taxonomic and functional diversity of both taxa should be positively related to compositional heterogeneity (Tscharntke et al., 2012). Finally, we expect the effects of land-use change will positively affect rodent taxonomic diversity (Godó et al., 2022), while negatively affecting the functional diversity of forest-specialized bats (Jung and Threlfall, 2018).

A total 284 bats belonging to 13 native species were captured (Tables S3 and S4 provide a complete species list, their distribution and their traits). Species richness encompassed 41.9% (13/31 species) of the Phyllostomidae and Mormoopidae species reported for the Yucatán peninsula. Artibeus jamaicensis (94 individuals), Artibeus lituratus (56), and Glossophaga mutica (48) were the most abundant species. In contrast, three species (Carollia perspicillata, Mimon cozumelae, and Micronycteris microtis) were recorded only once in Tizimín, and one species (Chiroderma villosum) only once in Calakmul. NMDS ordinations showed that sampling landscapes from Tizimín and Mérida clustered together for bats, indicating higher similarity in community composition compared to Calakmul (Fig. 3a). Despite some compositional differences among regions and landscapes, the region only explained 36.2% of the variation in bat composition (p > 0.05). The forest cover gradient explained a larger portion (50%) of community dissimilarities, although this effect was not statistically significant (p > 0.05). Complementary ANOSIM analysis showed no significant differences in bat community composition among regions (R = 0.13; p = 0.16).

Fig. 3.

Non-metric Multidimensional Scaling (NMDS) ordination based on species composition for (a) bats (stress value = 0.05) and (b) rodents (stress value = 0.06) at the nine sampling landscapes located in three regions (Mérida, Tizimín, and Calakmul) from the Yucatán peninsula, Mexico. Landscape names include a regional code (Me, Ti, Ca) followed by a letter indicating low (L), intermediate (I), and high (H) landscape modification.

Regarding rodent assemblages, we captured 99 rodents from 10 native species and two exotics (Mus musculus captured in MeI landscape, Rattus rattus in TiH). The most abundant species varied by region: Peromyscus yucatanicus in Mérida (27 individuals), Sigmodon toltecus in Tizimín (22), and Heteromys gaumeri in Calakmul (9). Five species were recorded only in Calakmul (Tables S3, S4). The NMDS indicated significant differences in rodent composition among regions (p = 0.003; Fig. 3b). Forest cover accounted for 50% of the variation in rodent composition, but its effect was not statistically significant (p = 0.13). ANOSIM results indicated significant differences among regions for rodent community composition (R = 0.58, p = 0.01).

Landscape predictors of taxonomic and functional diversity

The taxonomic and functional diversity of both taxa varied among landscapes. Bat species diversity varied from 1 to 11 species, being highest in TiL, the landscape with the lowest modification in Tizimín, and lowest in MeH, the most modified landscape from Mérida (Fig. 4, Tables S3, S5). Rodent species diversity ranged from 1 to 6 species, being highest in CaI (intermediate modification from Calakmul) and lowest also in MeH (Fig. 4). Most bat species were frugivorous (7 species, 78.9% of all individuals recorded); also, in terms of habitat use, 50% of bat individuals used both the understory and canopy strata (Table S4). Although other habitat use (semiaquatic, semiarboreal and arboreal) was also present, rodents were primarily terrestrial, representing 79.8% of individuals (7 species). Our data confirmed the landscape composition at the regional scale, showing a gradient of landscape modification from Calakmul with the highest forest cover to Mérida with the highest urban cover (Appendix S4; Fig. S1).

Fig. 4.

Diversity profiles based on Hill numbers of order q0, q1 and q2 for taxonomic and functional diversity for bats and rodents from three regions on the Yucatán peninsula, Mexico. Colors depict each region and the sampling landscapes within (Mérida in purple, Tizimín in orange, Calakmul in green). Landscape names include a regional code (Me, Ti, Ca) followed by a letter indicating low (L), intermediate (I), and high (H) landscape modification.

The scale of effect varied among landscape and response variables (Table S6) but was similar in bats (mean ± 95%CI = 1135 ± 253 m) and rodents (1167 ± 155 m) (W = 342.5, p = 0.250; Fig. S2). However, bats showed a broader range and confidence interval, likely reflecting interspecific differences in vagility, which is expected in community-level analyses. The multi-model averaging approach showed that urban cover was the most important landscape predictor, as it was included in the best models for most response variables, being negatively associated with taxonomic and functional diversity metrics (q0 and q1) for bats and rodents (Fig. 5, Tables S7, S8, Fig. S3). Urban cover was the only variable influencing diversity in rodents, while forest cover was positively associated with bat functional diversity (q0 and q1; Fig. 5). Taxonomic diversity of order 2 (q2) followed similar patterns for both taxa. The rest of landscape variables were much less important (Fig. S3, Tables S7, S8).

Fig. 5.

Relative importance value (sum of Akaike weights Σwi) of each landscape compositional variable (UC: urban cover; FC: forest cover, AC: agriculture and cattle farming cover, CH: compositional heterogeneity) for predicting the taxonomic and functional diversity (Hill numbers of order 0 and 1) of bats and rodents. Black bars indicate predictor variables included in models within ΔAICc <2 of the best models; gray bars indicate variables that did not perform better than the null model. The sign (+/−) indicates if the response was positive or negative. The percentage of deviance explained by each complete model is also indicated. Diversity responses for rodents do not include AC results due to high collinearity with urban and forest cover. In all cases, each landscape variable was measured at their respective scales of effect (see scales in Table S6).

To our knowledge, this is the first study to assess small mammals’ taxonomic and functional responses to landscape composition, using variables measured across different scales. Despite our relatively small sample size (n = 9 landscapes), we analyzed three regions with contrasting gradients of landscape modification and found strong evidence that urban cover is the main compositional variable negatively impacting the diversity of bats and rodents. As discussed below, this finding has critical ecological and applied implications.

The importance of urban cover for bats and rodents is not surprising. Both taxonomic and functional diversity decreased with increasing urban cover in the landscape. This can be explained by the fact that urban expansion occurs at the expense of the original forest cover, reducing availability of resources such as food and shelters, essential to maintaining species and their functional traits (Tscharntke et al., 2012; Meyer et al., 2015). Urban expansion also contributes to landscape fragmentation, intensifying edge effects and reducing connectivity between forest remnants and scattered trees. This further limits the movement of individuals and constrains landscape supplementation dynamics in human-modified landscapes (Dunning et al., 1992; Fahrig et al., 2011; Willig et al., 2019). Among the landscape variables evaluated, urban cover was the only factor influencing rodent diversity patterns, not only as a consequence of their low dispersal capacity and small home ranges (Godó et al., 2022), but also because arboreal rodents are highly dependent of forested habitats. The influence of urban cover on bat functional diversity affected the more specialized species, such as insectivores that are restricted to forested habitats, more than the highly mobile frugivores bats (Avila-Cabadilla et al., 2012; Ramírez-Mejía et al., 2020). Urban cover also impacted bat community composition, with landscapes MeL, TiI, and TiH showing high community similarity despite their different forest cover percentages. For example, MeL is mainly composed of forest (79.9%), but it is located on Mérida’s outskirts which is experiencing rapid urban growth (Islebe et al., 2015). These findings reinforce that urban cover negatively influence community responses, even in landscapes with substantial forest cover, potentially driving biotic homogenization. This could explain the convergence of bat communities, where sensitive species are replaced by disturbance-adapted ones like Artibeus jamaicensis, A. lituratus and Glossophaga mutica (Avila-Cabadilla et al., 2012; Ramírez-Mejía et al., 2020). Indeed, A. jamaicensis and G. mutica were dominant species in human-modified landscapes, where the synanthropic rodent species Mus musculus and Rattus rattus were exclusively found.

Contrary to our expectations, forest cover was weakly related to the taxonomic diversity of both taxa. However, it was strongly and positively related to bat functional diversity, where specialized insectivorous bats that rely on dense vegetation were mostly found in landscapes with >47% forest cover. Indeed, forested landscapes in our study area harbored bat species with unique functional attributes, like gleaning and aerial feeding, specialist frugivores, sanguinivores, and with preferences for specific tree strata (e.g., understory) (García-Morales et al., 2016; Ramírez-Mejía et al., 2020; Carvalho et al., 2021; Farneda et al., 2024). As forest cover provides a structurally complex environment with abundant roosting and feeding resources, our findings suggests that these attributes are favored in landscapes with higher forest cover. Also, some frugivorous bats that can forage in modified landscapes also depend on forest cover for feeding and roosting (Ripperger et al., 2015). Interestingly, the low and intermediate modified landscapes in Mérida exhibited low taxonomic and functional diversity despite having a relatively high forest cover. It is important to note that these landscapes are mainly composed of secondary forests. However, due to limitations in consistently distinguishing old-growth from secondary forests using satellite imagery, we combined them into a single forest class. While both provide key resources for bats and rodents, their ecological differences may influence species responses. For example, successional plants common in secondary forests, such as Solanum spp. and Cecropia spp, support generalist frugivores like Artibeus and Carollia (Carvalho et al., 2021; Saldaña-Vázquez et al., 2013). This pattern is reflected in our results on functional diversity for bats—specifically common and dominant species (q1 and q2)—where forest cover was the most important variable. The prevalence of successional vegetation in these secondary forests likely promotes higher abundances of generalist frugivores, even if overall diversity remains low.

Landscape compositional heterogeneity and agricultural cover had weak effects on bat and rodent diversity. A heterogeneous landscape matrix (i.e., secondary forest, scattered trees, arboreal crops) can provide supplementary resources, while also facilitating movement between forest remnants (Martínez-Ferreira et al., 2020). Therefore, it is important to assess not only landscape compositional heterogeneity but also local variables that reflect the heterogeneity of different land covers in the landscape (Avila-Cabadilla et al., 2012). On the other hand, we found that landscapes with specific land covers also differed in their most abundant species. For instance, Sigmodon toltecus, a generalist and disturbance-tolerant rodent commonly found in grasslands, crops and disturbed areas dominated the landscape with highest agricultural cover (TiI). In contrast, Heteromys gaumeri, a forest-associated granivore dominated the landscape with highest forest cover (CaL). These patterns may reflect differences in species and trait filtering driven by the land cover types that dominated each landscape (Paniccia et al., 2022).

It is known that changes in rodent and bat communities can disrupt food chains, seed dispersal, and insect control, affecting forest succession and ecological stability (Kunz et al., 2011; Morales-Díaz et al., 2019; Godó et al., 2022). The loss of these functions can also affect human well-being, disrupting species interactions and increasing the risk of emergence and transmission of infectious diseases (Sánchez-Soto et al., 2024). Taken together, our results indicate that human-mediated landscape modifications are a threat to different dimensions of diversity for both bats and rodents, even in landscapes with higher forest cover. Considering the negative role that urban cover plays on the taxonomic and functional diversity of both taxa, preserving biodiversity in the heterogeneous landscape mosaics of the Yucatán peninsula is a challenging task. Therefore, effective management strategies are needed to mitigate the negative impacts of human-landscape modification. Maintaining forest remnants and expanding overall native vegetation cover are crucial actions for increasing landscape connectivity and, consequently, species and communities' persistence. Other effective actions for biodiversity conservation in the Yucatán peninsula include reducing deforestation rates and establishing reserves both away from and connecting urban areas. Preventing deforestation is critical, especially for preserving bat and rodent assemblages and their key functions in forest ecosystems, including herbivory, predation, seed dispersal and pollination. Safeguarding these ecological processes supports ecosystem resilience and functionality, which are essential for sustaining biodiversity and the provision of ecosystem services within human-modified landscapes.

All data produced from this study are provided in the manuscript and the electronic supplementary material.