Long-term effects of traditional and conservation-oriented forest management on the distribution of vertebrates in Mediterranean forests: a hierarchical hybrid modelling approach

Abstract

Aim Recently, increasing attention has been devoted to the development of sustainable forestry practices aimed at finding a balance between the maintenance and enhancement of different forest resources. However, the long‐term, large‐scale effects of conservation‐oriented forest management on vertebrates have been poorly studied. We tested the hypothesis that conservation‐oriented forest management, being conceived to mimic the dynamics of a natural forest succession more closely than does traditional forestry, causes a less severe long‐term impact on the distribution of forest vertebrates. Location Molise region, Central Italy. Methods We proposed a hybrid modelling framework based on the integration of a forest dynamic model (‘LANDIS‐II’) with species distribution models (SDMs). The framework was applied on four forest specialist vertebrates (tiny salamander, slow worm, Eurasian nuthatch and Leisler's bat) and included three components: forest dynamic model (FDM), SDMs and spatial pattern analysis (SPA). FDM was used to simulate spatially explicit patterns of forest succession for the current time and for 2050, imposing three alternative forestry scenarios. The simulated forest succession patterns were analysed through SPA to calculate spatialized landscape metrics that were adopted as environmental predictors for SDMs. Landscape trajectories were calculated on current and future species distributions predicted with SDMs to evaluate the effect of alternative forestry practices on their extent and fragmentation. Results Forest management mainly affected the spatial configuration rather than the extent of the species potential distributions. Conservation‐oriented forest management was more favourable than traditional forestry in increasing the extent and reducing the fragmentation of the studied species’ distributions. Main conclusions Conservation‐oriented forest management, by mimicking the dynamics of a natural forest succession more closely than did traditional practices, favoured elements, such as forest unevenness, species richness, aggregation of patches and variability in their distances, which emerged as fundamental characteristics for preserving the long‐term persistence of forest vertebrates.

Integrating climate and land‐use change scenarios in modelling the future spread of invasive squirrels in Italy

“…We calculated the relative importance of variables from the ensemble model using the functionality provided in the biomod2 package (Jiguet, Barbet‐Massin, & Henry, ). Lastly, we calculated the spatial autocorrelation in regional SDMs residuals through Moran's I correlograms (Pottier et al, ; Di Febbraro et al, ; Supporting Information Appendix , Figure .1).…”

“…We calculated the relative importance of variables from the ensemble model using the functionality provided in the biomod2 package (Jiguet, Barbet-Massin, & Henry, 2010). Lastly, we calculated the spatial autocorrelation in regional SDMs residuals through Moran's I correlograms (Pottier et al, 2013;Di Febbraro et al, 2015; Supporting Information Appendix S3, Figure S3.1). We projected regional SDMs over two climate change scenarios derived by the fourth assessment of IPCC (IPCC, 2007), as well as two land-use change scenarios developed by Verburg et al (2006).…”

The decline of the lanner falcon in Mediterranean landscapes: competition displacement or habitat loss?

“…Current and future potential distributions of the three entities were built in order to evaluate how land‐use change could alter the current competition pattern between the two species by affecting the future potential distributions of their exclusive and shared breeding sites. SDMs were calibrated using a hierarchical structure, from global to regional scale (Gallien et al ., ; Di Febbraro et al ., ). Accordingly, models were first fitted considering the global species range (global SDM, Supporting Information, Appendix ).…”

“…The predictive performance of each model was assessed by measuring the area under the receiver operating characteristic curve (AUC; Hanley & McNeil, ) and the true skill statistic (TSS; Allouche, Tsoar & Kadmon, ). To avoid using poorly calibrated models, only projections from models with AUC ≥ 0.75 were considered in all subsequent analyses (Di Febbraro et al ., ). Model averaging was performed by weighting the individual model projections by their AUC scores and averaging the result (Marmion et al ., ).…”

Climate change, woodpeckers, and forests: Current trends and future modeling needs

“…Dynamic community process‐based forest landscape models (Scheller & Mladenoff, ) such as the LANDIS models (LANDIS‐II and LANDIS PRO; Figure a) that incorporate finer scale climate–vegetation–disturbance interactions compared to bioclimatic DGVMs are ideally suited for this integration (Di Febbraro et al, ; Iverson, Prasad, Matthews, & Peters, ; LeBrun et al, ; Tremblay, Boulanger, Cyr, Taylor, & Price, ). These models could improve woodpecker distribution modeling, especially within the context of multi‐objective management scenarios (Martin, Hurteau, Hungate, Koch, & North, ).…”