Introduction

Loss of biodiversity as a consequence of loss and fragmentation of native habitats has been observed frequently in tropical forests (Saunders et al. 1991; Myers et al. 2000; Haddad et al. 2015). Regarding habitat fragmentation, fragment size is one of the most important drivers of change in ecological communities, and smaller fragments are more prone to rapid decreases in biodiversity, compared to larger forest patches (Santos et al. 2008; Leal et al. 2012; Haddad et al. 2015). With the reduction of native areas, patch attributes (e.g., fragment size) and landscape parameters (e.g., connectivity and matrix type) affect environmental structure of the forest patches and therefore modify the microclimatic conditions (Bennet and Saunders 2010; Haddad et al. 2015). Habitat configuration of forest patches may threaten the native communities (Nichols et al. 2007; Haddad et al. 2015; Boscolo et al. 2017), and litter volume, tree diameter, height and density, and canopy openness are important characteristics that vary between forest fragments (Lira et al. 2015; Ezcurra 2016). Furthermore, abiotic parameters such as temperature, humidity, and light incidence also differ between patches, mainly due to forest edge effects, which are stronger in smaller than in larger forest fragments (Turner 1996; Ranta et al. 1998; Harper et al. 2005; Herrmann et al. 2005; Magnano et al. 2015).

Among insects that inhabit forest fragments, edaphic beetles represent a group that present strict associations with microclimatic conditions and that are highly susceptible to habitat changes (Pearson and Cassola 1992; Barbosa et al. 2002; Nichols et al. 2007; Fagundes et al. 2011; Wardhaugh et al. 2013). Edaphic beetles exhibit distinct feeding habits and can be classified as detritivores, saprophages, predators, or herbivores, and are represented mostly by species of the families Carabidae, Scarabaeidae, and Staphylinidae (Didham et al. 1998a; Barbosa et al. 2002; Lassau et al. 2005). Responses of such insects are associated with their environmental tolerance and their foraging strategies (Didham et al. 1998a; Larsen et al. 2006; Fagundes et al. 2011; Wardhaugh et al. 2013). Some species of edaphic beetles thrive in either disturbed or conserved habitats whereas other species are habitat generalists (Rodríguez et al. 1998; Fagundes et al. 2011). Moreover, some edaphic beetles prefer wet conditions and inhabit sites with closed canopy cover and higher leaf litter amount, whereas other species occur at dry and more exposed sites (Barbosa et al. 2002; Grimbacher and Catterall 2007; Fagundes et al. 2011).

Among edaphic beetles, numerous families are considered reliable bioindicators (Halffter and Favila 1993; Rodríguez et al. 1998; Maleque et al. 2009), which help understand the trends of ecological communities under habitat disturbances (Noss 1990). However, the responses of beetle assemblages facing habitat disturbances may differ between indicator groups at different taxonomic levels (Basset et al. 2004). Studies on indicator species and populations can provide important information for decisions on the conservation status of an ecosystem (Noss 1999; Stem et al. 2005; Beaupre and Douglas 2009). When analyzing effects of disturbances on ecological communities, assessments on a broader scale (e.g., a set of various indicator families) may produce a more informative picture than that by assessments based on a single family or an entire phylum (Basset et al. 2004).

Through the understanding of community structures based on indicator groups, it is possible to predict trends in biodiversity (Noss 1990; Rimbach et al. 2013; Alves and Hernández 2017). Furthermore, alterations in community structure due to habitat changes can directly and indirectly affect the ecological stability and, consequently, influence ecosystem services for human well-being (Bennet and Saunders 2010; Haddad et al. 2015). Evidence suggests that edaphic beetles perform crucial ecological services that are undoubtedly associated with ecosystem functioning. For example, decomposers and predators are involved in several trophic interactions, which makes them important for nutrient cycling and parasite control (Didham et al. 1998a; Lassau et al. 2005; Losey and Vaughan 2010).

The Atlantic rainforests are neotropical ecosystems that occur mainly along the coastal regions of Brazil. The northern Atlantic rainforest is an important endemism center (Pernambuco Endemism Center—CEPE) and is considered a hotspot for global conservation because of the high levels of endemism and the substantial loss of native vegetation (Prance 1987; Myers et al. 2000; Silva and Casteleti 2003). Intensive agriculture in the north Atlantic rainforest since the sixteenth century produced a highly fragmented landscape, with a coverage reduction to currently ca. 12% (Ribeiro et al. 2009). Fragment configuration, edge effects, and canopy openness are known to affect dung beetle assemblages in the CEPE (Filgueiras et al. 2011; Costa et al. 2013). Based on these findings, our main objective was to analyze how environmental parameters and fragment size affect edaphic beetle assemblages. Specifically, we assessed the effect of micro-environmental characteristics (i.e., tree density and diameter, canopy openness, and litter depth and dry weight) and fragment size on beetle abundance, species richness, species composition and diversity. Regarding the specific habitat requirements of Coleoptera species that inhabit tropical forests, we hypothesized that environmental parameters and fragment size should affect beetle assemblage structure. Ecosystems that have complex vegetation structures with high plant density and high amounts of litter may indicate high resource availability, which should positively affect the abundance of animal assemblages that depend on the vegetation environment to feed and to find food (e.g., herbivores, predators, and detritivores) (Cuevas-Reyes et al. 2004; Sayad et al. 2012; Lira et al. 2015). Therefore, we predicted that increasing tree density and diameter and increasing litter depth and dry weight should positively affect edaphic beetle abundance. A more open canopy may be associated with a less complex vegetation structure (Sarges et al. 2012); however, it also favors the establishment of beetle species that depend on greater light exposure (Didham et al. 1998a; Karen et al. 2008). Based on the intermediate disturbance hypothesis (Connel 1978), we expected higher species richness, abundance, and diversity of edaphic beetles in forest fragments with a more open canopy than in fragments with closed canopy. As larger fragments may comprise more microhabitats than that in smaller fragments (Fleury and Galetti 2006) and, thus, present higher resource availability, we expected fragment size to positively affect edaphic beetle abundance, species richness, species composition and diversity.

Materials and methods

Study site

This study was conducted in 12 forest fragments (from 6 to 469 ha) (Fig. 1, Table in the Supplementary Material), all of which were surrounded by sugarcane plantations, which were subjected to annual management cycles involving slash-and-burn practices. Forest patches are located in Usina Trapiche (08°35′S; 35°06′W), which is situated in the municipality of Sirinhaém, Pernambuco state, Brazil (Fig. 1). The region has a tropical rainy climate with a mean annual precipitation of 2400 mm and a mean annual temperature of 25 °C (Silva et al. 2010). The forest fragments had dense ombrophilous vegetation and are situated on top of hills below 100 m a.s.l.

Fig. 1
figure 1

Map of the sampled forest fragments (in dark-gray) and the location of the sampling site in Usina Trapiche, Pernambuco, Brazil. UB Ubaca, BA Baquinha, XA Xanguá, XA2 Xanguá2, SP São Pedro, CE Canto Escuro, CE2 Canto Escuro2, FR Franco, FR2 Franco2, MC Mata das Cobras, PC Pedra do Cão, TA Tauá

Full size image

Beetle sampling

Beetles were sampled during the dry season of the year (December 2012 and January 2013), period when there were no burning activities in the sugarcane plantations. Samples were collected only during the dry season because there is no clear distinct seasonality in beetles (specifically Scarabaeidae) in this region of the Atlantic rainforest (Filgueiras et al. 2009; Iannuzzi et al. 2016). Beetles were collected using pitfall traps, which consisted of a cylindrical plastic container (15 cm diameter × 13 cm height) containing 70% alcohol (250 ml) to preserve the collected material. In each fragment, four 40-m long linear transects were stablished at 20 m distance from each other, located ca. 200 m from the forest edge toward the forest core in order to reduce edge effects (Hill et al. 2011). In each transect, five traps were placed at 10 m distance from each other, and thus, 20 traps were used per forest fragment. Traps were left for 5 days after which beetles were collected and stored in 70% alcohol. In the laboratory, all beetles were identified at family level (taxonomic keys from Downie and Arnett 1995; Arnett et al. 2002; Rafael et al. 2012). After this, the beetles were morphotyped and, if possible, identified to species level. The identified specimens were deposited in the Coleção Entomológica da Universidade Federal de Pernambuco (CE-UFPE).

Environmental parameters

Habitat structure was analyzed based on the methods used by Lira et al. (2015), considering five micro-environmental characteristics: tree density, tree diameter, litter depth, litter dry weight, and canopy cover. In addition, the respective size of each forest fragment was recorded using digital maps. Tree density was measured as the number of trees and shrubs that touched the linear sampling transects. Tree diameter was measured at breast height (DBH), using trees that were included in the tree density measurements. For both tree density and tree diameter, mean values were produced for each transect. Litter depth was measured in three random 25 cm2 squares per fragment, placed at the soil surface, and the litter depth of each peak of the square was measured using a ruler. Litter dry weight was recorded using three random litter samples (25 × 25 cm) from sites at 10 m distance from each other in each transect. The litter samples were dried at 50 °C for 48 h, and the dry mass was weighted using a balance DCR CL-LB, with a precision of 100 g. Canopy cover was estimated using hemispherical photographs taken using a Nikon Coolpix E4500 camera with a Nikon FC-E8 lens. One photography was taken in the center of each transect, and the photographs were analyzed using Gap Light Analyzer software version 2.0 (Frazer et al. 1999). The photographs were taken at times of diffuse light, between 5:00 and 6:00 h.

Data analyses

To assess sampling efficiency in the studied forest fragments, we used species accumulation curves. We used the estimators of species richness Jackknife 1, Jackknife 2, Chao 1, and Chao 2 based on the number of samples (i.e., forest fragments) to compare beetle species richness between forest fragments. Rarefaction and extrapolation curves were produced using the nonparametric methods of Colwell et al. (2012). These analyses were performed using Estimates software version 9.1.0 (Colwell 2013).

Dichotomius iannuzziae Valois et al. (2017) as dominant in the Coleoptera assemblage, representing more than half of the collected beetles. Therefore, we performed data transformation on similarity matrices. In addition, we fitted models with and without this species to test its importance for the effect of environmental parameters and fragment size on beetle assemblage.

To analyze beetle diversity in forest fragments, we used components of diversity, considering alpha (Dα), beta (Dβ), and gamma (Dγ) diversities (Jost 2006). We evaluated the components of diversity using Hill numbers, using the number of species (0D) and Shannon diversity (1D) (Hill 1973; Jost 2006). 0D assigns higher weight to rare species than 1D, which considers the relative species abundance. 0Dα indicates the mean number of species per forest fragment, whereas 1Dα represents the number of abundant species of the patches. Beta diversity indicates the species turnover within the landscape, based on species richness (0Dβ) and relative species abundance (1Dβ). 0Dγ indicates the total number of species in the study, and 1Dγ indicates the number of abundant species. We calculated diversity numbers according to Marcon and Hérault (2015), which provides a bias correction for incomplete sampling. Statistical analyses were conducted using the entropart package in R software version 3.2.0 (R Core Team 2015; Marcon and Hérault 2015).

To analyze whether environmental parameters and fragment size affected observed and expected beetle species richness and abundance, we fitted generalized linear models (GLM) with a negative binomial distribution (to analyze abundance), a Poisson distribution (to analyze species richness), and a Gaussian distribution (to analyze estimated species richness). Estimated species richness was obtained using the estimator Chao 1, which is recommended for assemblages with many rare species, similar to what we found in this study. The effects of tree density, DBH, canopy openness, litter depth, litter dry weight, and forest fragment size on beetle abundance and species richness were tested. The models used to analyze the dependent variables were reduced based on the Akaike Information Criteria (AIC). The variables that were excluded in the most parsimonious models were considered non-significant. The significance of the dependent variables was tested using a likelihood ratio test between the full and the reduced model using the lmtest package (Zuur et al. 2009; Hothorn et al. 2018). Normality of the residuals was visually assessed from normal q–q plots, and the presence of outliers was tested using Cook’s distance. GLMs for beetle abundance were conducted with and without the data of the dung beetle D. iannuzziae. All analyses were performed using R software version 3.2.0 (R Core Team 2015).

We used principal coordinates of neighbour matrices (PCNM) from the geographical distance matrix (coordinates in decimal degrees) to generate spatial variables. This method partitions the spatial dataset structure in eigenfunctions that can be used as variables of the spatial structure (Bocard and Legendre 2002; Dray et al. 2006). These spatial variables were used to assess the effects of the spatial arrangement of forest fragments (connectivity) on the beetle assemblage similarity (species richness and composition) using a redundancy analysis (RDA). The significance of each variable was calculated using a partial Monte Carlo test with 9999 permutations. All analyses were performed using the vegan package in R software version 3.2.0 (R Core Team 2015; Oksanen et al. 2018).

Correlations between beetle assemblage compositions (with and without D. iannuzziae data) and the environmental factors were calculated using an RDA with a ‘manual selection’ of independent variables in the CANOCO software version 4.5 (ter Braak and Smilauer 2003). The RDA allows testing the effect of each factor on species composition as the only explanatory variable (Leps and Smilauer 2003). In this analysis, linear combination of the species abundance and the environmental variables were fitted, and the significance of each variable was calculated using a partial Monte Carlo test with 9999 permutations (Legendre and Legendre 1998; McCune and Grace 2002; Leps and Smilauer 2003). Linear responses of changes in the species composition were confirmed using a detrended correspondence analysis prior to the RDA (lengths of gradients = 2.77) (Leps and Smilauer 2003).

In the RDA and GLMs, correlations between environmental variables were assessed using the variance inflation factor (VIF) calculated using CANOCO software version 4.5 (ter Braak and Smilauer 2003). VIF values > 10 indicated that the variance of the canonical coefficients was inflated by multicollinearity of the explanatory variables and produced instability of the regression model (ter Braak and Smilauer 2003; Kutner et al. 2005; Zuur et al. 2009).

Results

A total of 1701 beetles belonging to 65 species and 14 families were collected. According to Chao 1 and Chao 2, the samplings performed in forest fragments recorded 46.30% and 33.52%, respectively, of the expected species. According to Jackknife 1 and Jackknife 2, the samplings recorded 65.11% and 50.59%, respectively, of the expected species. Staphylinidae and Carabidae were the families with most species (s = 14 and 10, respectively), which together comprised 36.92% of the total number of observed species (Table 1). Scarabaeidae and Nitidulidae were the families with the highest abundance of collected beetles (n = 1184 and 225), together representing 82.83% of the total number of collected individuals. Erotylidae and Mordellidae were represented by only one species each, being both singletons (Table 1). In addition, more than half of the species observed in the current study were singletons (s = 33), and six species were doubletons (see in Supplementary Material).

Table 1 Abundance and species richness of beetle families collected in Atlantic forest patches in Usina Trapiche, Pernambuco, Brazil
Full size table

Dichotomius iannuzziae, Scarabaeidae sp1 and Nitidulidae sp1 were the most abundant species, accounting for 69.49% of the collected individuals (Fig. 2). The region of Usina Trapiche presented approximately 17 species per fragment (0Dα = 17.18), with nearly four abundant species per forest patch (1Dα = 3.98). In the whole region, almost eight species were considered abundant (1Dɤ = 7.98). Beta diversity based on beetle richness indicated four distinct beetle assemblages in the 12 sampled forest fragments (0Dβ = 3.78), whereas two distinct assemblages were observed based on Shannon diversity (1Dβ = 2.00).

Fig. 2
figure 2

Rank-abundance curves of beetle species sampled in forest patches in Usina Trapiche, Pernambuco, Brazil

Full size image

The degree of collinearity between the explanatory variables of both RDA and GLM were low (VIF = 1.79–4.26), indicating no problems related to data multicollinearity. In the analysis with D. iannuzziae (the most abundant beetle species), beetle abundance was explained best by litter dry weight and fragment size (Table 2). Abundance was positively correlated with litter dry weight (Fig. 3a) and fragment size (Fig. 3b). In the analysis without D. iannuzziae, beetle abundance was explained best by tree density and litter depth; however, these factors did not significantly affect beetle abundance (Table 2). Observed beetle richness was better explained by tree density, but there was no significant effect over richness (Table 2). However, estimated beetle richness was positively affected by tree density (Table 2; Fig. 3c).

Table 2 Generalized linear models to test the effect of environmental variables and forest fragment size on observed and estimated species richness and abundance of beetles in Atlantic forest fragments in Usina Trapiche, Pernambuco, Brazil. The environmental variables DBH and canopy openness were removed from this table as they did not significantly affect beetle species richness and abundance
Full size table
Fig. 3
figure 3

Relationship between environmental parameters and beetle assemblage: beetle abundance (with Dichotomius iannuzziae) and litter dry weight (a) and fragment size (b); estimated beetle richness and tree density (c)

Full size image

Beetle species richness and composition was not influenced by the spatial arrangement of the forest fragments, according to the RDA (richness: F = 2.86, p = 0.171; composition: F = 1.30, p = 0.056). In the analysis with D. iannuzziae, the environmental variables explained 84.5% of the variability of species composition, of which 95.6% was explained by axis 1 and 2.8% by axis 2 of the RDA. Only fragment area was correlated with the beetle composition, according to the RDA (Table 3). The patch area explained ca. 36% of the species composition variability (RDA: F = 9.55; p = 0.026). When data were analyzed without D. iannuzziae, the variation of species composition was explained to 74% by environmental variables, of which 42.4% was represented by axis 1 and 16.9% by axis 2 of the RDA. Forest fragment area (RDA: F = 3.06; p = 0.038) and litter depth (RDA: F = 3.03; p = 0.035) together explained 39% of the species composition variation (Table 3).

Table 3 RDA of beetle richness and species composition constrained by environmental variables in Atlantic forest fragments in Usina Trapiche, Pernambuco, Brazil
Full size table

Discussion

Beetle communities comprise species that are vulnerable to habitat disturbance, and others that are eurytopic, and can thus occupy habitats with varying environmental structures (Didham et al. 1998a; Rodríguez et al. 1998; Rainio and Niemelä 2003; Nichols et al. 2007; Fagundes et al. 2011). In the present study, both the environmental parameters and forest fragment size affected beetle assemblages in a north Atlantic rainforest. Due to the natural diversity and heterogeneity of tropical rainforests, larger forest patches are assumed to represent more complex habitats, providing high microhabitat diversity for beetle species (Bierregaard et al. 1992; Turner 1996). As a consequence, large patches provide a wider array of resources and therefore facilitate high species diversity of edaphic beetles.

Scarabaeidae and Nitidulidae were the most abundant families in the current study. This result is related to the dominance of a few species of these families. For example, the dominant species in the current study, D. iannuzziae, is widely dominant in forest fragments of north Atlantic rainforest and occurs in both disturbed and conserved patches (Filgueiras et al. 2011; Costa et al. 2013; Salomão and Iannuzzi 2015). Although this species was described only recently, it was previously referred to as D. aff. sericeus and was observed in the Atlantic forest of northeastern Brazil (see Valois et al. 2017). Similarly, there are Nitidulidae beetles that are highly abundant in some areas due to their successful association with specific food resources (e.g., plant matter, fruits, pollen, or honey) (Barbosa et al. 2002; Aguirre and Dirzo 2008; Fagundes et al. 2011). In addition, Staphylinidae and Carabidae were the most diverse families in the current study. These families are diverse in conserved and disturbed environments, both in tropical and temperate ecosystems (Didham et al. 1998b; McIntyre 2000; Lassau et al. 2005; Buddle et al. 2006; Salomão et al. 2018). Furthermore, Staphylinidae and Carabidae were already shown to occur at high diversity in Amazon and north Atlantic rainforest (Didham et al. 1998b; Salomão et al. 2018). The north Atlantic forest is a highly fragmented landscape and typically experiences strong edge effects (Ranta et al. 1998). This may be favoring high species richness of Staphylinidae and Carabidae, and its assemblage may be dominated by species that successfully occupy conserved and disturbed habitats. Based on these results, we suggest that Staphylinidae, Carabidae, Scarabaeidae, and Nitidulidae are good indicators of landscape fragmentation in the north Atlantic rainforest.

The sampling effort in this study was similar to that of other studies of insect diversity in tropical rainforests (Moeed and Meads 1985; Lassau et al. 2005; Fagundes et al. 2011), however, the relatively low sampling efficiency must be considered (see Brehm et al. 2005; Rocha-Ortega and Favila 2013; Hernández et al. 2014). This survey was conducted only during the dry season. Thus, additional sampling during the rainy season, which is the time of highest insect activity in most tropical ecosystems (Wolda 1978, 1989; Tanaka and Tanaka 1982; Vasconcellos et al. 2010, but see), could potentially improve the sampling efficiency. However, previous diversity surveys in the same ecosystem did not produce substantial differences between the rainy and dry seasons regarding beetle diversity (i.e. Scarabaeidae; see Filgueiras et al. 2009; Iannuzzi et al. 2016). Although low sampling efficiency may indicate insufficient surveying to assess biodiversity, it is important to consider the huge number of rare insect species in tropical rainforests (Novotny and Basset 2000), as confirmed by the high number of singleton and doubleton species in the present study. This factor seems to be a determinant of the low sampling efficiency of the current study, where the majority of the recorded species were rare. In addition, this may be a result of using an entire order of insects as a focal group and using a sampling method which captured beetles that randomly fell in the traps. Therefore, we suggest that increased sampling effort is necessary to achieve sufficient sampling efficiency of the edaphic beetle fauna in north Atlantic rainforest fragments.

Beetle abundance was positively correlated with litter dry weight, and fragment size affected beetle assemblage structure, abundance, and species composition. Edaphic beetles included the group Coleoptera that mainly thrives in litter and therefore depends on the condition of this habitat. Invertebrate communities that inhabit the litter layer use this habitat as shelter and food resource (Sayad et al. 2012; Luz et al. 2013). Ecosystems with thicker litter layers may therefore be more suitable for more abundant edaphic community, due to the higher habitat availability (Sayer et al. 2010; Sayad et al. 2012). However, apparently few species seem to benefit particularly from this increased habitat space, as there was no significant correlation between litter dry weight and species richness. Besides the amount of litter, the microhabitat condition and the plant composition of litter typically affect the distribution of litter-dwelling beetles (Magura et al. 2004; Vanderwall et al. 2006; Wiezik et al. 2007). Therefore, we suggest that such environmental parameters may result in synergistic effects influencing the distribution of litter-dwelling beetles in the soil. Further studies are needed to assess how litter parameters (e.g., quality, plant species, microhabitat characteristics, and suitability as a food resource) affect beetle distribution in tropical forests.

Fragment size significantly affected beetle community structure. Larger forest patches presented a more abundant beetle fauna and a distinct species composition, compared with smaller forest fragments. Previous studies attributed larger forest patch size to higher habitat availability, which thus favors more individuals and different species, compared with smaller fragments (Bennet and Saunders 2010; Haddad et al. 2015). In addition, the more representative families belonging to the Coleoptera in the current study (i.e., Staphylinidae, Carabidae, Scarabaeidae and Nitidulidae) are sensitive to habitat fragmentation with shifts in their communities according to fragment size and environmental structure (Didham et al. 1998a; Rainio and Niemelä 2003; Nichols et al. 2007; Fagundes et al. 2011). Tropical rainforest are heterogeneous landscapes, comprising myriads of different microhabitats, which favors high insect biodiversity (Wilson 1988; Larsen et al. 2006; Novotny et al. 2006). Due to the high fragmentation of tropical rainforests, recent studies are presenting alarming scenarios with a considerable loss of biodiversity (Myers et al. 2000; Santos et al. 2008; Tabarelli et al. 2012). Based on the results of the current study, we suggest that patch size is one of the main drivers of changes in beetle community structure in fragmented landscapes. It is unclear whether there is higher microhabitat diversity in larger Atlantic rainforest patches than in smaller fragments, as no correlation between environmental parameters and fragment size was observed. Thus, further studies are needed to test this hypothesis.

Species dominance affects species coexistence, which may affect metacommunity dynamics and ecosystem properties (Hooper et al. 2005; Hillebrand et al. 2008). Interestingly, the analyses conducted with and without the dominant species of the current study (D. iannuzziae) presented contrasting results regarding the response of edaphic beetle assemblage to environmental parameters and fragment size. In the current study, the weighting of each species was either considered independently (e.g. Mantel, NMDS), or the overall abundance of the edaphic beetles was used (e.g. GLM). Thus, the weighting of a superabundant species such as D. iannuzziae, which was almost fourfold more abundant that the second most abundant species, may bias results that are abundance-dependent. Our results suggest that dominant species affect responses of ecological assemblages to environmental parameters. Thus, analyses encompassing assemblages must be interpreted with care, considering the statistical bias that may be introduced by dominant species, such as D. iannuzziae. Future studies on assemblages with superabundant species should consider their responses by using analyses with and without such taxa to achieve a more precise and detailed interpretation of assemblage responses.

With this study, we demonstrate that forest fragment size and micro-environmental characteristics significantly affected edaphic beetle community. Habitat loss in north Atlantic rainforests has led to a substantial loss in species richness and abundance, and produced a turnover in species composition of edaphic beetles. Considering that the most diverse and abundant beetles of this landscape are decomposers and predators, the decrease of patch area and modification of environmental structure may induce drastic changes in the nutrient cycling process and in community dynamics.