Influence of landscape heterogeneity on vertebrate assemblages of fragmented woodland

Abstract

Environmental fragmentation leads to the transformation of continuous habitats in several patches, separated from each other by a habitat ifferent from the original that is generally efined matrix. To study its effect on the iodiversity of the residual patches, this process has initially being assimilated to the theories developed on insular fauna dynamics (e.g. acArthur & Wilson). However, the reductionist approach of describing patches as islands, in a sea of non-habitat, does not allow the inclusion of the complexity of real landscapes. The aim of the present study is therefore the investigation of the relationship existing between animal species distribution and landscape heterogeneity, to examine the influence of matrix composition and configuration, and patch structure on the vertebrate assemblages of residual woods. The field part of the study took place in the Marche Region, in an area of 100,000 ha approximately, which encompasses the Chienti and Potenza River catchments. This area was selected since it offered the opportunity of comparing woodland fragments included in landscapes that have suffered a diverse human impact. In the area, 24 sample sites were selected, characterised by the presence of Turkey oak (Quercus cerris) and downy oak (Q. rubescens). In each site, the bird communities composition was described using 66 point counts. The presence and absence of two arboreal mammals was assessed, with 115 hair tubes for the European red squirrel (Sciurus vulgaris), and 132 nestboxes for what concerns the hazel dormouse (Muscardinus avellanarius). Each site spatial structure was also studied, through circular sample plots in which the vertical organization and composition of the herbaceous, shub and arboreal layers was described. Aerial photographs were used to describe and quantify landscape patterns. Four classes of land cover were identified (woodland, cropland, hedges, shrubs, grassland and anthropic areas) and mapped with GIS. Two software were employed to describe the landscape configuration according to the developed map, Patch Analyst and Image Analyzer. The extracted metrics were both structural and functional, and for every land cover class, percentage of occurrence, nearest neighbor, edge density, spatial aggregation and several more parameters were calculated. To avoid multicollinearity a principal components analysis (PCA) was performed on the variables, divided in three subsets describing respectively patch characteristics, landscape composition and landscape configuration. Six factors were extracted from the PCA and used as explanatory variables for the analysis of the target species distributions. The most abundant species present in the bird communities were also the most widespread. The consideration of the existing relationship between single species and environmental variables stressed how the whole community is influenced only by factors describing matrix composition and configuration, and in particular the lack of isolation of the patches, their connectivity and the presence of woods, grasslands and anthropic areas in the landscape. The factors associated to the patch features did not influence the observed pattern. The only feature of the patches that had an effect on the observed abundances was the size of the wood (R2 =0.435, p<0.001). Considering species diversity, the smaller and poorer sites tended to contain subsets of the species present in bigger and richer woods, showing a significant nested pattern that mostly involved generalist species. Using the species’ life history traits, the communities were subdivided in several different guilds, considering their breeding period diet, feeding technique, nest location, average laying date, clutch size and body mass. For each subgroup, the percentage of species and individuals present in each site was calculated, and included in multiple regression models. The fact that patch related factors, as the extent of the canopy cover, the canopy height and the average tree diameter, could not be disregarded in any of the models indicates the importance of the available internal components for species persistence. The contemporaneous consideration of landscape and patch analysis scales led to the emergence of different patterns, highlighting the presence of different driving forces, belonging to the two scales, contemporaneously shaping community structure. The effect of landscape heterogeneity on the two considered mammal species was studied using logistic regressions. The resulting models were ranked using second order Aikaike Information Criterion (AICc and AICc) and the importance of the included parameters was evaluated with Akaike weights (wi). The use of three different sets of variables confirmed the importance of the amount of hedgerows in the landscapes surrounding the patches for the occurrence of both species. The convergence of the two arboreal mammals towards landscapes that offered a higher amount of hedges proved the critical role of this element to ensure species persistence through landscape structural connectivity. The relevance of other elements, as the amount of shrublands and grasslands in the matrix, suggests that since the local forest management impoverished the woods available resources, species occurrence is may be linked to the contemporaneous exploitation of sub-optimal habitats, through a process defined “habitat compensation”. The inclusion of species belonging to different taxa emphasised the matrix essential role, both in terms of composition and configuration, in determining the actual use of residual patches. In a heterogeneous landscape, structure heterogeneity matters in terms of connectivity and lack of isolation, and the presence and position of hedgerows and woods in the landscape were key determinants of species distribution. The contemporaneous consideration of parameters describing landscape and patch features showed how these two elements and their associated scales affect species in a different degree, and their lack of simultaneous consideration may lead to misleading conclusions. The matrix therefore holds the capability to at least mitigate the effects of isolation and habitat loss, if its management, and hence its permeability, allows animal movement.