Responses of small mammals to habitat characteristics in Southern Carpathian forests
We surveyed small mammal communities in a montane area along the elevational gradient in relation to habitat characteristics and human impact, this study being the first to assess habitat use by small mammals in the Southern Carpathians.Compared to a similar study conducted in the Eastern Tatra Mountains31, the species richness (12 species captured) was lower in our survey; part of the reason could be that the North Carpathian endemic Microtus tatricus and the boreal species Sicista betulina are absent in our study area, which is beyond the limits of their geographical distribution. Species composition of small mammals was overall comparable to those reported for forested areas of Northern Carpathians17,32,33, although a high variability, both spatial and temporal, in the number and abundance of species characterized all surveyed communities. Although A. flavicollis was seldom captured in 2003 and 2005 and only at low elevations23, overall it, together with M. glareolus, dominated the small mammal community, representing over 75% of the captured individuals (Table 1). This is the common pattern of small mammal communities in temperate zones, i.e., to be dominated by two species, usually rodents34,35,36. M. glareolus and A. flavicollis are the dominant species in most forests of central and eastern Europe32,33,37,38, with one or the other being more numerous depending on habitat conditions and geographic position35. M. glareolus and A. flavicollis were also found to remain dominant in small-sized clearings39.Box-trapping results for shrews are often considered underestimates because of their small size40 and because seed baits are not attractive to them41. However, during our survey S. araneus had wider distribution than A. flavicollis; we captured it in low numbers in a large number of trapping sites, having the highest ratio between occurrence (45.2%) and relative abundance (16%) of all small mammal species (Table 1). S. araneus was higher in abundance in our research area in comparison to both natural and planted montane forests in Northern Carpathians17,32,33, possibly as an effect of the long-term conservation practices in the national park.Besides the three dominant species and S. minutus, all the other captured species are of regional conservation interest, being included in the Red Book of Vertebrates from Romania42, which highlights the conservational value of this landscape.Small mammals showed significant responses to habitat characteristics at population and community levels, regardless of the metrics considered. Tree cover was an important predictor for small mammal communities (Table 2, Table 3). Increased tree cover limits light available for understory plants, reducing habitat structure43, hence the usually negative correlation between canopy cover and both shrub and herbaceous cover. The reduced vegetation complexity of closed-canopy forests may limit resources important to small mammals. Most studies show that forests with a greater percentage of tree cover harbour less abundant small mammal communities44. In the Sierra Nevada mountains in North America, small mammals showed a limited response to canopy thinning, reflecting the generalist habits of the common species in those forests, which may be a legacy of more than a century of human impacts generating a process of biotic homogenization via differential success of some native species over the others45. In Europe, there is a legacy of much longer human impacts, thus common forest species should have even more generalist habits. However, in our research area tree cover was positively correlated with all parameters, except for the abundance of A. flavicollis, which did not significantly respond to it (Table 2). The small mammal fauna in our study area is a primarily forest fauna, with dominant species responding negatively to the decrease in tree canopy cover, even when this means an increase in the understory cover and complexity. The response to tree cover was strongest in M. glareolus (Fig. 2a, Fig. 3). In boreal forests of Scandinavia tall vegetation and structural heterogeneity of trapping stations positively influenced the total abundance of this species15. This may mean that there is an important geographic variability in the ecological behavior of M. glareolus. There are differences in the habitat preferences not only along the latitudinal gradient15,35,46,47 but also on elevation. At the foothills of Southern Carpathians M. glareolus is limited mainly to forest edges and riparian forests with tall hygrophilous vegetation48. During this study we did not find a significant effect of the interaction between elevation and tree cover, probably because of the relatively short elevational gradient (of 1200 m), which did not include lowland forests outside the ecological optimum of M. glareolus. The short gradient may also explain the lack of response by M. glareolus, both as absolute and relative abundance (Fig. 3) to elevation itself, although this species is known to increase in density towards the north and at higher elevations35.Although shrub cover is an important element of vegetation structure, and one which increases its complexity, it had a significant effect only on the abundance of A. flavicollis. In opposition to our expectations, we found increased abundances of A. flavicollis in forests with little or no shrub layer (Table 2). In forests, shrubs may serve as shelter for mice against physical disturbances such as soil compaction, trampling or rooting49, although some studies failed to find evidence for this50. A positive effect of cover and height of shrub layer was also found on the abundance of A. flavicollis in the Northern Carpathians in forest clearings51. However, besides the positive effects of greater vegetation complexity and increased availability of food and shelter resources, the shrub layer also reduces visibility and hinders rapid movement, so that mobile species such as mice, which rely on running rather than hiding to escape predation, are exposed to higher predation risk in habitats with dense undergrowth.The feature related to habitat heterogeneity to which small mammals responded positively in our study area was the abundance of rocks (Fig. 2a, Table 2). Rocky outcrops and large boulders are stable elements of the landscape that enhance the availability of shelters and refuges providing hard protection for nest sites50. Some species that do not burrow are dependent on rocks for shelter, occurring only in rocky sites. Among these is C. nivalis, but the small number of captured individuals did not allow testing its habitat use.Unlike rocks, woody debris is more ephemeral, and apparently it was less valued as a shelter resource (Table 3). Many studies show the importance of coarse woody debris as a quantitative habitat feature for forest small mammals44; their value increases in the late decay stages52. Woody debris in mid-to-late decay state is often a suitable substrate for lichen and fungi, and can support a rich insect fauna53, all potential foods for omnivorous rodents and shrews. In our research area the sites with the largest amounts of coarse woody debris were those recently logged, so availability of food resources for small mammals was not optimal.Soil moisture, which has a very strong effect on the primary productivity and vegetation diversity, may also have an important role in the habitat selection, with various effects on small mammal populations. In our study area the two dominant rodents had opposite responses to soil moisture, with M. glareolus showing a strong preference for dry habitats (Fig. 3, Table 2), in contrast to its response to moisture in other parts of its distribution. At the southern limit of its geographical distribution35 or at the limit of its elevational distribution48, M. glareolus is usually confined to damp habitats, but there it does not develop abundant populations, with Apodemus species usually dominating the small mammal community. In the northern part of its distribution, where Apodemus species are absent, M. glareolus also shows a preference for moist woodlands54. We may thus infer that the response of M. glareolus to soil moisture is modulated by the interaction with mice species, in our case A. flavicollis. This conjecture is also supported by the fact that moisture did not significantly affect community abundance, only species composition (Table 3). Other studies have also reported conflicting results of the role of soil moisture for A. flavicollis. For example, it was one of the most important factors influencing population dynamics of A. flavicollis in a beech forest in northern Germany55 but it did not predict its distribution in Britain56.Sites closer to watercourses are damper, so an overlap of the effect of the two variables—moisture and distance to water—would be expected. However, the significant negative effect of distance to water on the abundance of A. flavicollis also had a component that was independent of soil moisture (Table 2), and this may have a spatial significance. The increased abundance of A. flavicollis in sites close to watercourses could be explained by a potential fence effect that these may exert on small mammal populations. River banks are linear habitats bordered on one side by a physical barrier, more or less penetrable depending on the local habitat morphology. Linear habitats with favourable conditions sometimes shelter rodent populations at densities much higher than those in wide habitats, although the underlying mechanism, involving probably territoriality and dispersal, is not yet understood57. In our research area, river banks were important for A. flavicollis especially in low abundance years, when we captured this species exclusively here and only at low elevations, suggesting that besides a source of habitat heterogeneity watercourses may be involved also in the spatial dynamics of populations, with their banks being used as routes for dispersal.Neither species richness nor species abundance changed along the elevational gradient in our research area when also considering yearly fluctuations and habitat characteristics (Table 2), and our result is in contradiction with the pattern frequently described for mountains worldwide58,59, including the Eastern Tatras31, which shows a reduced species richness with the increase in elevation. But on the other hand, we found species composition to be affected by elevation, with A. flavicollis responding negatively and S. araneus positively. The thermophilous character of A. flavicollis is more evident in the Northern Carpathians, where this species was found only up to 1328 m, well below the timberline31. But as latitude compensates for elevation, at least in part, in our research area A. flavicollis was found along the entire elevational gradient, up to above 2000 m (Table 1), beyond the timberline, in the subalpine shrubs, perhaps as a result of its lack of preference for the tree cover. S. araneus had a similarly wide elevational distribution and, unlike A. flavicollis, it was captured at high elevations also in low abundance years23. This result supports the classification of S. araneus as a habitat generalist. In contrast to these species, M. glareolus was only once captured in the shrubs beyond the timberline, suggesting that in our study area this vole avoids habitats with no tree layer. This may also be because the subalpine sites that we surveyed were heterogenous, with relatively small patches of shrubs separated by open meadows, areas avoided by M. glareolus.Logging is the main human activity causing disturbance of forests. In our study area only selective logging was recent, while older clearcuts were already reforested. The overall impact was negative and significant on species richness and total abundance, as well as on the abundance of S. araneus. The sensitivity of S. araneus to logging may be one cause of its increased abundance at higher elevations, as in the study area recent timber exploitation was concentrated at low elevations (mostly in mixed forests). Although we did not find a significant response of M. glareolus to logging, other studies revealed that this species is influenced by habitat alterations caused by logging15 but also by the inter- and intraspecific competition, which is considered by some investigators to be the main mechanism causing the decline of vole populations in harvested forests60. We learned that timber exploitation caused a drastic reduction of the small mammal populations in the disturbed area, to the point where no animal was captured during a trapping session, with the neighbouring habitats being also affected. However, since habitat changes were not substantial, timber extraction had a relatively short time impact on the small mammals, and the year following logging the community structure resembled that of undisturbed areas. This suggests that selective logging with the extraction of a relatively small amount of timber affects small mammals rather by direct disturbance than by changes in habitat characteristics. The influence of logging on species of conservation interest, such as the mostly arboreal M. avellanarius and the rare S. alpinus, still needs to be evaluated. The main effect of logging is the decrease in canopy cover or its complete removal in case of clearcuts. But there are also other effects, such as degradation of shrub and herbaceous layers, soil compaction and erosion, and also direct disturbance involving presence of humans and sometimes domestic animals (in the research area logged trees were removed by horses and watch dogs usually roamed the logged forest patches and their surroundings), noise and soil vibrations. Following reduction of canopy cover, improvement in light conditions cause development of understory and decrease of soil moisture, affecting the abundance and composition of animal communities. Most studies on the influence of forest management on small mammals in Europe have focused mainly on clearcutting, one of the most common methods of forest harvest, and have revealed a positive effect on most analyzed small mammals, which can be attributed to an increase in forb and grass cover in the harvested areas61. In managed forest in Czech Republic it was found that the practice of felling within relatively small-sized clearings may help preserve the diversity of small mammal community39. However, the observed positive effect of clearcuts may be a biased result caused by the fact that most surveyed sites were in homogenous conifer plantations, a low-quality habitat for small mammals61.We found that tourism had less impact on small mammals compared to logging, with M. glareolus showing the only significant negative response. Tourism may also represent an additional source of food for the small mammal species that tolerate the presence of humans, such as A. flavicollis, which we found on campgounds. Touristic buildings may also represent important daily or hibernation shelters for some rodents, such as Glis glis, which we observed in autumn in a chalet. In contrast to logging, the effect of tourism on small mammals has been less researched and most such studies have focused on winter sports resorts and mainly on the impact of ski-run development, which involves substantial alteration of forest habitat, sometimes with a significant change in small mammal communities62. In case of ecotourism, damage to the vegetation and soil compaction that result from trampling during tourist season is only local and temporary, thus the regeneration of soil fauna and vegetation is possible63, hence the weaker effect of ecotourism on small mammals.Habitat characteristics had a stronger influence on community abundance than on species composition (Table 3), suggesting that, being primarily forest dwellers, the small mammal species in our study area have somewhat similar responses, especially towards tree cover, but they also show some differentiation, which is reflected by the divergent responses of A. flavicollis, M. glareolus, and S. araneus in their relative abundances in the community. The differences in the relative habitat use, along with the divergent dietary niche, enables their coexistence as dominant species, exploiting the same wide range of habitat resources.In conclusion, habitat use by small mammals in the continuous forest landscape in the Southern Carpathians was overall similar to that reported from the Northern Carpathians, with some notable differences related to recent and historical forest management practices and to latitude. Variation partitioning showed that yearly fluctuations were more important than habitat selection in shaping community composition. Temporal variations eclipsed the effects of habitat selection and elevational gradient, temporal fluctuations in community abundance and species composition having higher amplitudes than spatial variations. Relative habitat use by most species also changed among years. Thus, our results suggest that ignoring the time dimension of habitat selection may lead to the inability to comprehend the forces and processes that structure small mammal communities. More
