Geographical generality of bird‐habitat relationships depends on species traits

Abstract

Aim:The environmental filtering process is often considered as static in ecological studies. However, growing evidence shows that species-environment relationships vary in space and time. In this study, we assessed to what extent bird responses to landscape components can be geographically generalised and whether differences in response generality can be explained by traits. Location: France.Methods: We collected a large bird data set (1968 point counts over two years) with a standardised protocol in three agricultural regions with different levels of intensification in France. We modelled the relationships between the distribution of 26 bird species and three landscape components (percentage of woodland, hedgerow density and landscape heterogeneity) and assessed whether differences between regions in bird responses to landscape components (i.e., landscape-region interactions) can be explained by three species traits (habitat specialisation, diet and migration strategy).We also examined the response of total species richness. Results:We found that 16 species showed regional differences in their response at least for one of the three landscape variables. Importance of landscape-region interactions was significantly correlated with two species traits. Responses of specialist species to landscape components were geographically more constant than those of generalists. The geographical variability of responses was higher for migrants than for sedentary species. There were no significant relationships for the diet trait. Species richness responded positively to the three landscape metrics in a similar way in the three regions. Main conclusions:The results underline the need to take into account the spatial differences between species responses to habitats according to their traits when modelling species-habitat relationships at large scales. From a conservation point of view, we suggest that conservation measures could be generalised at a large scale for specialist species which are declining in agricultural landscapes. K E Y W O R D Sbirds, land cover, model generality, species-habitat relationship, traits

Improved prediction of Canada lynx distribution through regional model transferability and data efficiency

“…SDMs are often constructed with opportunistic data collected across large spatial extents or with intensive data collection across smaller extents (Aubry et al, 2017; Thuiller et al, 2006). Few studies have the resources required for extensive data collection at multiple locations across a large area (Bonthoux et al, 2017). However, we combined GPS data from multiple collaborators to directly assess regional differences in habitat selection across populations within a large spatial area.…”

“…Generalizability of SDMs is also predicted to be related to specificity in diet or habitat selection (Bonthoux et al, 2017; Yates et al, 2018), although this pattern appears to be born out in some species and not others. A similar lack of transferability in habitat selection was observed in koalas (McAlpine et al, 2008), a specialist on eucalyptus leaves, while the opposite pattern was found in several species of European birds living in mixed agricultural land, which demonstrated increased model transferability with habitat specialization (Bonthoux et al, 2017). Specialists are generally predicted to select a narrower range of environmental conditions (Kassen, 2002; Peers et al., 2012), and thus are predicted to favor homogenous environments with resource use similar and transferable across populations.…”

“…Lynx rely almost entirely on snowshoe hares ( Lepus americanus ) as a food source (Aubry et al, 2000; Squires & Ruggiero, 2007), and thus are closely tied to boreal forests with high horizontal vegetation cover (Holbrook et al, 2017; Squires et al, 2010). Lynx are an excellent species to assess geographic generalizability of SDMs across populations, because we expect habitat specificity and selection for a narrow range of environmental conditions to result in less intraspecific variation and more habitat generalizability compared to generalist species (Bonthoux et al, 2017; Yates et al, 2018). We used data from three geographically distinct populations at the species' southern range periphery in Washington, Montana, and Wyoming, USA.…”

Spatial and Temporal Dependency of NDVI Satellite Imagery in Predicting Bird Diversity over France

“…The lowest values of NDVI correspond to the most open landscapes (Appendix E) that facilitate farmland ground-nesting species avoiding high vegetation structures (e.g., Alauda arvensis, Alectoris rufa, Coturnix coturnix). Intermediate values of NDVI represent heterogeneous landscapes with grasslands and hedgerows that favor farmland species nesting in hedgerows and requiring complementation of habitats (e.g., Carduelis cannabina, Emberiza cirlus, Emberiza citrinella) [6]. Landscape heterogeneity, through standard deviation NDVI, also explained a part of the species richness of the two bird groups but to a lesser extent.…”

“…At finer spatial scales, the components of landscape mosaics play a key role in explaining species habitat selection [4][5][6]. Remote sensing-based predictors of habitat characteristics may contribute to improve the performances of species distribution models (SDM) [7].…”