Introduction
Land-use change is a major driver of biodiversity loss (Chapin et al., Reference Chapin, Zavaleta, Eviner, Naylor, Vitousek and Reynolds2000). As a consequence of human activities, natural habitats have been fragmented and degraded, posing a major threat to wildlife through marked reductions in habitat quantity and quality (Didham et al., Reference Didham, Kapos and Ewers2012; Tscharntke et al., Reference Tscharntke, Tylianakis, Rand, Didham, Fahrig and Batary2012; Haddad et al., Reference Haddad, Brudvig, Clobert, Davies, Gonzalez and Holt2015). Developments such as forestry plantations, croplands, cattle-raising grasslands and urban areas have encroached on natural habitats, posing a major threat to wildlife (Echeverría et al., Reference Echeverría, Coomes, Salas, Rey-Benayas, Lara and Newton2006; Newbold et al., Reference Newbold, Hudson, Hill, Contu, Lysenko and Senior2015). Although habitat disturbance is usually associated with biodiversity loss, intermediate levels of disturbance may be beneficial, as was demonstrated by Connell (Reference Connell1978), who proposed the intermediate disturbance hypothesis. The proposal that high diversity can be maintained by intermediate disturbance events that limit strong competitors and allow more species to coexist has been widely examined. In forest ecosystems, intermediate disturbance events are usually associated with forest gaps that allow shade-intolerant understorey plants to thrive (Dalling & Hubbell, Reference Dalling and Hubbell2002). These shade-intolerant plants usually have flowers and fruits that are important food resources for native animals (Fontúrbel et al., Reference Fontúrbel, Salazar and Medel2017).
Historically, Indigenous communities have obtained natural resources from native forests (Smith-Ramirez, Reference Smith-Ramirez2007), but these traditional uses are being replaced by large scale land-use changes, resulting in habitat fragmentation and degradation, such as in the temperate rainforest of southern South America (Echeverría et al., Reference Echeverría, Coomes, Salas, Rey-Benayas, Lara and Newton2006). Those forests are a biodiversity hotspot with high levels of endemism (Myers et al., Reference Myers, Mittermier, Mittermier, da Fonseca and Kent2000) and plant–animal mutualism. They are being affected by large-scale deforestation and subsequent replacement by other land uses (mainly forestry plantations; Nahuelhual et al., Reference Nahuelhual, Carmona, Lara, Echeverria and Gonzalez2012) and by small-scale disturbance associated with selective logging (Smith-Ramirez, Reference Smith-Ramirez2007). Selective logging causes less marked effects than habitat loss and fragmentation but can reduce habitat quality by removing large trees that provide resources such as nesting cavities (Lindenmayer et al., Reference Lindenmayer, MacGregor, Welsh, Donnely and Brown2008) and habitat for other species (Tejo & Fontúrbel, Reference Tejo and Fontúrbel2019), ultimately altering forest composition and ecological processes (Asner et al., Reference Asner, Knapp, Broadbent, Oliveira, Keller and Silva2005).
Forest-dependent animals are good models for examining the effects of disturbance from selective logging (Castellón & Sieving, Reference Castellón and Sieving2006). A charismatic example is the monito del monte Dromiciops gliroides, a small arboreal marsupial (Hershkovitz, Reference Hershkovitz1999) categorized as Near Threatened on the IUCN Red List (Martin et al., Reference Martin, Flores and Teta2015). Although D. gliroides depends on forest habitats (Fontúrbel & Jiménez, Reference Fontúrbel and Jiménez2011), it is tolerant of disturbance and capable of persisting in disturbed habitats if some structural features are retained (Salazar & Fontúrbel, Reference Salazar and Fontúrbel2016). Although the effects of large-scale disturbance on this species have been assessed (Fontúrbel et al., Reference Fontúrbel, Silva-Rodriguez, Cardenas and Jimenez2010, Reference Fontúrbel, Jordano and Medel2015; Uribe et al., Reference Uribe, Chiappe and Estades2017), the potential consequences of the small-scale disturbance associated with selective logging are poorly known. As fleshy fruits are an important component of the diet of D. gliroides, their abundance can influence its occurrence, abundance and behaviour (García et al., Reference García, Rodríguez-Cabal and Amico2009; Morales et al., Reference Morales, Rivarola, Amico and Carlo2012; Tiribelli et al., Reference Tiribelli, Amico, Sasal and Morales2017). Fleshy fruit abundance may increase in selectively logged forest, where light reaching the understorey results in the proliferation of shade-intolerant plants (Fontúrbel et al., Reference Fontúrbel, Salazar and Medel2017), a factor that could explain the presence of D. gliroides in disturbed habitats (Salazar & Fontúrbel, Reference Salazar and Fontúrbel2016). Forest-dependent species also coexist with Indigenous communities that traditionally extract native wood for their livelihoods (Smith-Ramirez, Reference Smith-Ramirez2007).
To examine this issue, we compared the density and occupancy of D. gliroides in three native forest stands experiencing different intensities of selective logging. Following Connell's (Reference Connell1978), we hypothesized that the density of D. gliroides would be higher at an intermediate logging intensity as a result of a higher diversity of fleshy-fruited plants, which are associated with D. gliroides occupancy.
Study area and species
We conducted this study in Pucatrihue in southern Chile, at 150 m altitude (Fig. 1). The mean annual temperature is 12 °C and mean total annual precipitation is 2,500 mm. We defined three study sites, 1, 2 and 3, separated by 1–4 km to ensure independence, as the maximum movement range of D. gliroides is c. 500 m (Fontúrbel et al., Reference Fontúrbel, Silva-Rodriguez, Cardenas and Jimenez2010). Pucatrihue is a rural locality inhabited by Indigenous communities of the Huilliche branch of the Mapuche ethnic group. The Huilliche people (meaning ‘people from the south’ in Mapudungun) traditionally exploit marine resources for food but also extract wood from the surrounding native forests. Given the low population density of the area, most of the wood extraction is for subsistence.
Location of study sites 1, 2 and 3 in the temperate rainforest region of Pucatrihue, southern Chile, indicating the trapping locations, area of influence of each trapping location, and households where we interviewed Indigenous people regarding their use of the forest.
Dromiciops gliroides is a small arboreal marsupial, endemic to the temperate rainforests of southern South America, the only extant species of the order Microbiotheria, which is closely related to the Australian marsupials (Hershkovitz, Reference Hershkovitz1999). D'Elia et al. (Reference D'Elia, Hurtado and D'Anatro2016) proposed there are three Dromiciops species, but this has been refuted based on morphological and genetic evidence (Valladares-Gomez et al., Reference Valladares-Gomez, Celis-Diez, Palma and Manriquez2017; Martin, Reference Martin2018; Suárez-Villota et al., Reference Suárez-Villota, Quercia, Núñez, Gallardo, Himes and Kenagy2018). Dromiciops gliroides is one of the few hibernating marsupials of South America (Hadj-Moussa et al., Reference Hadj-Moussa, Moggridge, Luu, Quintero-Galvis, Gaitan-Espitia, Nespolo and Storey2016), and a seed dispersal agent for at least 16 native plant species (Amico et al., Reference Amico, Rodríguez-Cabal and Aizen2009). Although formerly considered to be restricted to old-growth forests (Hershkovitz, Reference Hershkovitz1999), it also occurs in secondary forests (Fontúrbel et al., Reference Fontúrbel, Silva-Rodriguez, Cardenas and Jimenez2010) and abandoned exotic plantations (Fontúrbel et al., Reference Fontúrbel, Candia and Botto-Mahan2014; Uribe et al., Reference Uribe, Chiappe and Estades2017). Despite being tolerant of habitat disturbance, D. gliroides depends on habitat structure and heterogeneity (Salazar & Fontúrbel, Reference Salazar and Fontúrbel2016), and feeds on fleshy fruits (its primary food source) and animal protein (invertebrates and eggs; Cortés et al., Reference Cortés, Franco, Sabat, Quijano and Nespolo2011).
Methods
Habitat characterization
At each trapping location we measured per cent canopy cover, with a spherical crown densiometer (as this variable is associated with occurrence of D. gliroides; Salazar & Fontúrbel, Reference Salazar and Fontúrbel2016), plant diversity (using the Shannon entropy index, H'; Jost, Reference Jost2006) by counting individuals of all plant species within a 2.5 m radius, and quantified the number of fleshy fruits per plant and estimated their biomass from a sample of 10–20 ripe fruits of each species (Fontúrbel & Medel, Reference Fontúrbel and Medel2017). From the latter, we estimated fruit biomass density and diversity, and the diversity of fleshy-fruited plants, using H' (Goenster et al., Reference Goenster, Wiehle, Kehlenbeck, Jamnadass, Gebauer and Buerkert2011). As there are no meteorological stations near the study site, we obtained daily temperature and precipitation records from meteoblue (2016) and YR (2016), averaging the values from the two sources. We selected these sources because of their high data resolution and precision.
Wood extraction and use assessment
As selective logging is the main disturbance in the study area, we conducted an assessment of how local people extract and use native wood from each of the three survey areas. We defined a 700-m buffer of influence around each area so as to include all households involved in local wood extraction. We used structured interviews (Amare et al., Reference Amare, Mekuria, Wondie, Teketay, Eshete and Darr2017) to assess which tree species were logged, and the approximate wood volume extracted. Respondents were assured anonymity, and no personal data or information other than wood use and extraction were stored or analysed. Interviews were limited to permanent Pucatrihue residents and were conducted by HGA. We interviewed only the head of the household, asking about the family's economic activities, household characteristics, land ownership and wood use (Cinner et al., Reference Cinner, McClanahan and Wamukota2010). We used these data to estimate how often tree species were logged and the wood volume extracted.
Trapping
We conducted live trapping surveys at the three sites. As D. gliroides is an arboreal marsupial, we used custom-made wire-mesh traps (26 × 13 × 13 cm) placed 1.5–2.5 m above the ground, baited with fresh banana slices (Fontúrbel, Reference Fontúrbel2010). At each site, we set 40 traps in an 8 × 5 array with 10 m between traps. We opened traps at 19.00 and checked them at 7.30 the next day. We measured, weighed and sexed all captured individuals, and marked them using fur cuts in unique patterns, for identifying any recaptures, and released them in the capture location. Live trapping was conducted during November 2015–April 2016, the time of year during which D. gliroides is most active (Fontúrbel et al., Reference Fontúrbel, Candia and Botto-Mahan2014). We operated traps for 4–5 consecutive nights bimonthly, giving a total of 1,680 trap-nights.
Data analysis
We used a non-parametric multivariate analysis of variance (MANOVA) using the adonis function of the vegan package (Oksanen et al., Reference Oksanen, Blanchet, Kindt, Legendre, Minchin and O'Hara2013) in R 3.6 (R Development Core Team, 2019) to assess habitat differences between the three sites. We used canopy cover, plant diversity, fruit biomass, fruit biomass diversity, and diversity of fleshy-fruited plants as the response variables, and site as a factor. As we found significant differences, we conducted individual ANOVA tests for each response variable to examine differences among sites. We used factor analysis to describe the variability among the five response variables. Data variability comprises communality (variability explained by linear combinations of potential factors) and uniqueness (variability not explained by these linear combinations). We performed factor analysis using the function factanal in R, with two factors, regression scores, and a Promax rotation (Long & Teetor, Reference Long and Teetor2019). We used a principal component analysis to visualize differences. We compared plant species composition among sites using a non-parametric analysis of similarities (ANOSIM; Clarke, Reference Clarke1993) with a Bray Curtis similarity measure and 9,999 permutations to estimate significance. We visually represented differences using non-metric multidimensional scaling (nMDS) (Fontúrbel & Jiménez, Reference Fontúrbel and Jiménez2014). We used a Bray Curtis similarity measure for nMDS and optimized the result to maximize the variance explained by the two components. We estimated nMDS components using the function metaMDS in the package vegan in R (Oksanen et al., Reference Oksanen, Blanchet, Kindt, Legendre, Minchin and O'Hara2013).
We used capture–recapture to estimate D. gliroides abundance and density at the three sites (Fontúrbel et al., Reference Fontúrbel, Silva-Rodriguez, Cardenas and Jimenez2010, Reference Fontúrbel, Franco, Rodríguez-Cabal, Rivarola and Amico2012). As this marsupial has a mean home range of 1.6 ha (Fontúrbel et al., Reference Fontúrbel, Franco, Rodríguez-Cabal, Rivarola and Amico2012), we assumed a closed population model; no individuals were recaptured at a different site from where they were originally captured. As D. gliroides populations may extend beyond the area covered by the trap arrays, we estimated abundance using a reversible jump algorithm in a Monte Carlo Markov chain, which is able to simulate individual distributions over undefined surfaces (Green, Reference Green1995). We performed abundance estimations using the package multimark (McClintock, Reference McClintock2015; McClintock, Reference McClintock2019) in R. To estimate population densities, we calculated effective sampling areas using the area of the trap array plus a buffer corresponding to the mean recapture distance (Parmenter et al., Reference Parmenter, Yater, Anderson, Burnham, Dunnum and Franklin2003). We then calculated population densities by dividing the estimated abundance by the effective sampling area (following Parmenter et al., Reference Parmenter, Yater, Anderson, Burnham, Dunnum and Franklin2003).
We used occupancy models to assess habitat selection patterns (MacKenzie et al., Reference MacKenzie, Nichols, Lachman, Droege, Royle and Langtimm2002). Occupancy models are a useful approach to estimate a species’ distribution, by taking detection probability into account and reducing the probability of obtaining false negatives (Royle, Reference Royle2006). We used the R package unmarked (Fiske & Chandler, Reference Fiske and Chandler2011) to estimate occupancy models. Firstly, we assessed correlation among all climate and habitat variables, to discard any highly correlated variables (r ≥ 4). Then, we used model-based recursive partitioning (Zelleis et al., Reference Zelleis, Hothorn and Hornik2008) to perform a selection process. This method is based on multivariate recursive partitioning (Cook & Goldman, Reference Cook and Goldman1984), which is built upon a parametric regression model. The advantages of this approach are the ease with which the results can be interpreted and the identification of those variables causing model distortion (Zelleis et al., Reference Zelleis, Hothorn and Hornik2008; Strobl et al., Reference Strobl, Malley and Tutz2009). We conducted model-based recursive partitioning using the R package partykit (Hothorn et al., Reference Hothorn, Hornik and Zeileis2006; Zelleis et al., Reference Zelleis, Hothorn and Hornik2008), using a generalized linear model (GLM). We included the densities of all plant species, and retained plant diversity, biomass diversity and fruiting plant diversity in all models as we consider them fundamental for the feeding and forest structure preferences of D. gliroides, to reduce the variable subset and keep only those significant for D. gliroides habitat selection (Zelleis et al., Reference Zelleis, Hothorn and Hornik2008). We fitted 25 potential occupancy models (Supplementary Material 1). After removing the non-convergent models, the 17 candidate models were compared using the Akaike information criterion (AIC; Burnham & Anderson, Reference Burnham and Anderson2002). We retained three models within the ΔAIC ≤ 5 subset (representing a cumulative model weight of 0.95). We plotted occupancy probabilities (ψ) and the 95% confidence intervals for each variable using the R package ggplot2 (Wickham, Reference Wickham2016).
Results
We found significant differences in habitat characteristics among the three sites (non-parametric MANOVA F 2,117 = 13.67, P < 0.001; Fig. 2), which were explained by significant variations in canopy cover (Supplementary Fig. 1a), fruit biomass (Supplementary Fig. 1b), fruiting plant diversity (Supplementary Fig. 1c) and fruit biomass diversity (Supplementary Fig. 1d), but not by variations in plant diversity (Supplementary Fig. 1e). Site 1 had a more closed canopy than the other sites. Site 2 had some canopy openings and the highest fruit biomass, plant diversity, fruiting plant diversity and fruit biomass diversity (Table 1). Site 3 had a relatively open canopy, and the lowest fruit biomass, fruiting plant diversity and fruit biomass diversity (Supplementary Fig. 1). Factor analysis showed that these five variables made differential contributions to the variability between the three sites (Table 2). Plant species composition was significantly different among the three sites (ANOSIM R = 0.374, P = 0.001; Fig. 3).
Per cent canopy cover, fruit biomass, plant diversity, fruiting plant diversity and fruit biomass diversity at the three study sites (Fig. 1). All figures are mean ± SE.
Comparison of habitat characteristics of Dromiciops gliroides among the three study sites (1–3; Fig. 1) using a principal component analysis. Ellipses depict 95% confidence intervals for each habitat type, and arrows represent the five measured variables.
Plant composition differences among the three study sites illustrated using a non-metric multidimensional scaling ordination (stress = 0.125).
Factor analysis results for the five habitat variables (Table 1, see text for details; factor correlation = −0.462; model goodness of fit: χ 2 = 2.14, df = 1, P = 0.143). Communality is the variability explained by linear combinations of the five variables, and uniqueness is the variability not explained by these linear combinations.
We identified 18 households whose occupants were extracting wood: two, seven and nine households within the influence of sites 1, 2 and 3, respectively (Fig. 1). There were wood extraction activities at all three sites, but with a large variability in intensity and number of tree species used (Table 3). Wood extraction was lowest in site 1 (a mean of 0.22 m3/ha/year; three species logged), intermediate in site 2 (0.33 m3/ha/year; eight species logged) and highest in site 3 (2.55 m3/ha/year; nine species logged). The tree species logged in site 1 were also the most commonly logged species in sites 2 and 3, but extraction intensity varied between sites (Table 3).
Wood use and extraction patterns by local people at the three sites. Per cent indicates the number of households responding affirmatively to each question. Wood extraction volumes were calculated based on data provided by the respondents.
In the live trapping survey, we captured D. gliroides 36 times, corresponding to 28 individuals (recapture rate was 33%). We estimated a mean abundance of 8.26 ± SE 0.01, 35.29 ± SE 0.13 and 13.38 ± SE 0.07 at sites 1, 2 and 3, respectively, and mean population densities of 8.10 ± SE 3.83, 27.89 ± SE 11.59, and 13.57 ± SE 6.46 individuals/ha, respectively.
We captured D. gliroides at 29 of the 120 trap locations. The recursive partitioning model indicated that the trees Luma apiculata and Drimys winteri were significantly associated with detection of D. gliroides (Supplementary Fig. 2). The density of these two plant species along with the estimated diversity indices and the climatic variables produced 16 convergent models, from which we retained three models based on their AIC ranking (Table 4). Detection probability did not vary with temperature (Fig. 4a), but increased with increased precipitation (Fig. 4b) and decreased with increased fruit biomass (Fig. 4c). Occupancy increased with density of L. apiculata (Fig. 4d) and D. winteri (Fig. 4e), fruit biomass diversity (Fig. 4f) and fruiting tree diversity (Fig. 4g), but decreased with overall species diversity (Fig. 4h).
The relationship between detection of D. gliroides and (a) temperature, (b) precipitation and (c) fruit biomass, and occupancy and density of the trees Luma apiculata (d) and Drimys winteri (e), and fruit biomass (f), fruiting tree (g) and plant species (h) diversity. Mean values and their 95% confidence intervals are presented.
The three top occupancy models (for all 25 models, see Supplementary Material 1) fitted for the detection of Dromiciops gliroides, representing a 95% cumulative AIC weight (cωAIC).
1Occupancy (ψ) variables: div, plant species diversity; fpdiv, fleshy-fruited plant diversity; biom, dry fruit biomass; Lapic, Luma apiculata density; Dwint, Drimys winteri density. Detection (d) variables: pp, precipitation; temp, temperature; ffbm, fleshy fruit biomass density.
Discussion
The different intensities of selective logging at our three study sites could have been responsible for the habitat differences that influenced D. gliroides abundance and habitat selection. Differences in habitat factors between the three sites (canopy cover, fruit biomass, and fruiting plant and fruit biomass diversity) could be a result of the increasing level of small-scale wood extraction disturbance from sites 1 to 3, with site 2 having an intermediate level of disturbance. This would be consistent with the intermediate disturbance hypothesis (Connell, Reference Connell1978), as D. gliroides was most abundant at site 2. Low levels of wood extraction can increase fruiting plant diversity, mainly as a result of the proliferation of shade-intolerant plant species (Fontúrbel et al., Reference Fontúrbel, Salazar and Medel2017) and the consequent increase in fruit biomass density and diversity.
Previous studies of the response of D. gliroides to habitat disturbance have focused on habitat fragmentation (Rodríguez-Cabal et al., Reference Rodríguez-Cabal, Aizen and Novaro2007; Fontúrbel et al., Reference Fontúrbel, Silva-Rodriguez, Cardenas and Jimenez2010), degradation, and transformation by exotic plantations (Fontúrbel et al., Reference Fontúrbel, Candia and Botto-Mahan2014; Uribe et al., Reference Uribe, Chiappe and Estades2017), all of which are large-scale disturbances. As far as we are aware, this is the first study that explicitly assesses the responses of D. gliroides to different intensities of small-scale selective logging. For small-bodied arboreal animals such as D. gliroides, habitat structure plays a major role in determining occurrence, as they need a structurally complex habitat that provides movement pathways, nesting sites and shelter (Bro-Jørgensen, Reference Bro-Jørgensen2008). Unlike large-scale deforestation, selective logging does not have area or edge effects (Didham et al., Reference Didham, Kapos and Ewers2012), but non-random tree removal (larger trees are usually logged first) alters characteristics (Asner et al., Reference Asner, Knapp, Broadbent, Oliveira, Keller and Silva2005) such as availability of nesting cavities (Reem & Lõhmus, Reference Reem and Lõhmus2011). Nevertheless, low to medium levels of selective logging can create forest gaps, allowing shade-intolerant plants to thrive that would not usually grow beneath a dense canopy (Dalling & Hubbell, Reference Dalling and Hubbell2002). These shade-intolerant plants often have large flowers and fruits, an important food source for frugivores (Fontúrbel et al., Reference Fontúrbel, Salazar and Medel2017).
Our estimates of the density of D. gliroides are similar to those for other locations in southern Chile (Celis-Diez et al., Reference Celis-Diez, Hetz, Marín-Vial, Fuster, Necochea and Vásquez2012; Fontúrbel et al., Reference Fontúrbel, Franco, Rodríguez-Cabal, Rivarola and Amico2012). The density at site 1 is similar to that on Chiloé island (Celis-Diez et al., Reference Celis-Diez, Hetz, Marín-Vial, Fuster, Necochea and Vásquez2012), and the densities at sites 2 and 3 are similar to those at continental sites in Chile and Argentina (Fontúrbel et al., Reference Fontúrbel, Franco, Rodríguez-Cabal, Rivarola and Amico2012; Balazote-Oliver et al., Reference Balazote-Oliver, Amico, Rivarola and Morales2017). Differences in density between sites could be related to differences in plant species composition, which may be influencing habitat selection. For example, common shade-intolerant species such as Aristotelia chilensis and Rhaphithamnus spinosus were absent from site 1 but were abundant at sites 2 and 3. Gevuina avellana, a shade-tolerant species, was present only at site 1. Such plant species turnover is consistent with a light-incidence gradient as a result of habitat disturbance (Gianoli et al., Reference Gianoli, Saldaña, Jiménez-Castillo and Valladares2010; Fontúrbel et al., Reference Fontúrbel, Salazar and Medel2017), and D. gliroides seems to be responding to these changes. The low density at the least disturbed site could be related to the lower diversity of fleshy-fruited plants (there were few shade-tolerant plant species with fleshy fruits), with individuals needing to move longer distances to forage (Salazar & Fontúrbel, Reference Salazar and Fontúrbel2016; Fontúrbel et al., Reference Fontúrbel, Salazar and Medel2017).
As expected, given its frugivorous habit (Amico et al., Reference Amico, Rodríguez-Cabal and Aizen2009), the probability of D. gliroides occupancy increased with plant species diversity and fleshy-fruited plant species diversity. The availability of fleshy fruits increases after disturbance following establishment of fast-growing secondary vegetation (Greenberg et al., Reference Greenberg, Perry, Harper, Levey, McCord, Greenberg, Collins and Thomson2011). Despite being considered an old-growth forest species (Hershkovitz, Reference Hershkovitz1999), D. gliroides selects secondary forests with a high diversity of fleshy-fruited plants. The presence of the native bamboo Chusquea quila and hemiparasitic mistletoe Tristerix corymbosus were the best predictors of the occurrence of D. gliroides in a fragmented landscape (García et al., Reference García, Rodríguez-Cabal and Amico2009; Rodríguez-Cabal & Branch, Reference Rodríguez-Cabal and Branch2011), but small-scale disturbance such as selective logging, fruit diversity and abundance appear to influence habitat selection. The fact that the densities of L. apiculata and D. winteri, common species of secondary forests, had significant effects on the probability of occupancy, indicates that D. gliroides is able to use secondary forest, and even abandoned exotic plantations, as long as there is some landscape heterogeneity to provide nesting sites (Salazar & Fontúrbel, Reference Salazar and Fontúrbel2016).
The Indigenous community is a crucial part of this story. Their houses, tools and fishing boats are constructed mainly from the wood of native species, and wood from native trees is used for heating and cooking (Smith-Ramirez, Reference Smith-Ramirez2007). These communities have inhabited this area for centuries, using these natural resources sustainably (Herrmann, Reference Herrmann2006; Molares & Ladio, Reference Molares and Ladio2012). Our findings show that small-scale local wood extraction and biodiversity conservation can coexist, with intermediate levels of disturbance producing beneficial conditions for D. gliroides. This could also be the case for other forest-dependent species with habitat needs similar to those of D. gliroides (e.g. understorey birds; Fontúrbel & Jiménez, Reference Fontúrbel and Jiménez2011). However, any increase in wood extraction could threaten D. gliroides and other native animals. In southern Chile, mean wood extraction volumes from old-growth forests are 2.5–8.5 m3/ha/year, and in secondary forests 7.5–15.0 m3/ha/year (Nahuelhual et al., Reference Nahuelhual, Donoso, Lara, Núñez, Oyarzún and Neira2007), well above the wood volumes extracted from our study area. Extraction of wood from site 3, where we recorded the highest extraction rate, increased during 2017–2019 as economic activities related to tourism increased. Approximately 60% of the native forest in site 3 was cleared during April 2018–January 2019 (F.E. Fontúrbel, unpubl. data).
Despite being a forest-dependent species, D. gliroides appears to be able to persist in logged habitats if wood extraction volumes are low, and intermediate disturbance could result in an increase in the species’ density in response to the increase of fleshy-fruited plant diversity. Responses to small-scale disturbance are important for understanding how biodiversity is responding and adapting to a changing world (Armesto et al., Reference Armesto, Manuschevich, Mora, Smith-Ramirez, Rozzi, Abarzua and Marquet2010). Indigenous communities play a key role in conserving native forests, but increasing extraction pressure is harming this balance between people and nature. The evidence presented here could be used as a guideline to establish a wood extraction quota, to protect the extant remnants of the declining temperate rainforests of Chile and its many endemic species, and the sustainable use of these forests by Indigenous communities.
Acknowledgements
We thank Tito Álvarez, Marina, Alfonso and Neto Las Casas, and Orlando Melillanca for their support and for access to land; the people of Bahía Mansa and Pucatrihue, and the Choroy-Traiguén Indigenous community, for responding to our survey; and two anonymous reviewers for their comments. FEF was supported by FONDECYT Fondo Nacional de Desarrollo Científico y Tecnológico projects 3140528 and 11160152.
Author contributions
Study design: HG-A, MS, JMM-B, JCS; fieldwork: HG-A, FEF; social survey: HG-A, AP; data analysis: HG-A, MS, FEF; writing: HG-A, FEF.
Conflicts of interest
None.
Ethical standards
Animal trapping and handling followed the guidelines of the American Society of Mammalogists (Sikes et al., Reference Sikes and Gannon2011), captures were authorized by the Chilean Agriculture and Livestock Bureau (licence 302/2015 to FEF and HG-A), interviews with people were approved by the ethics committee of the Instituto Internacional en Conservación y Manejo de Vida Silvestre (FCTM-ICOMVIS-CGA-TA-091-2014), and this research otherwise abided by the Oryx guidelines on ethical standards.
Data availability
Data for this article are available at doi.org/10.6084/m9.figshare.11451267
Open access
