Bat responses to changes in forest composition and prey abundance depend on landscape matrix and stand structure

Abstract

Despite the key importance of the landscape matrix for bats, we still not fully understand how the effect of forest composition interacts at combined stand and landscape scales to shape bat communities. In addition, we lack detailed knowledge on the effects of local habitat structure on bat-prey relationships in forested landscapes. We tested the assumptions that (i) forest composition has interacting effects on bats between stand and landscape scales; and (ii) stand structure mediates prey abundance effects on bat activity. Our results indicated that in conifer-dominated landscapes (> 80% of coniferous forests) bat activity was higher in stands with a higher proportion of deciduous trees while bats were less active in stands with a higher proportion of deciduous trees in mixed forest landscapes (~ 50% of deciduous forests). Moth abundance was selected in the best models for six among nine bat species. The positive effect of moth abundance on Barbastella barbastellus was mediated by vegetation clutter, with dense understory cover likely reducing prey accessibility. Altogether, our findings deepen our understanding of the ecological processes affecting bats in forest landscapes and strengthen the need to consider both landscape context and trophic linkage when assessing the effects of stand-scale compositional and structural attributes on bats.

Figure 1

Predicted responses of (a) bats to stand composition at each level (mean, mean-SD, mean + SD) of landscape composition; and (b) B. barbastellus to moth abundance at each level (mean, mean-SD, mean + SD) of shrub cover. Stand composition corresponds to the percentage cover of deciduous trees within stand while landscape composition was inferred using the proportion of deciduous forests within the forested landscape area at 1 km radius scale. Predictions from GLMMs are represented by the solid lines with 95% confidence intervals indicated in the same colour.

Figure 2

Predicted responses of bats in relation to stand structure and composition. Predictions from GLMMs are represented by the green solid lines with 95% confidence intervals indicated in light green. The raw data are indicated with fill black circles.

Figure 3

Study design. (a) Location of the 21 paired stands in the Cévennes Biosphere Reserve and National Park. Each number represents a pair of one coniferous-dominated stand and one mixed/deciduous-dominated stand. The core zone of the study area is represented in grey. Schematic representation (not in scale) of the land cover within 1-km radius buffer of one pair is depicted (dark green: coniferous forests; olive green: deciduous forests; beige: others) with a pie chart at its centre representing the percentage of deciduous trees (light green) vs. coniferous trees (dark green) at the stand scale. (b) Distribution of the sampled stands along a gradient of deciduous-conifer mixture at both stand of and landscape levels (see material and methods for more details). Stands of the same pair have the same number and are linked with a grey line. The map was created using ArcGIS v10.2 (ESRI, Redlands, California, USA; https://desktop.arcgis.com/fr/arcmap/).