Data & codes - Effects of landscape heterogeneity on bird communities in temperate and boreal forests – a review

Literature search We systematically reviewed studies examining the effects of landscape heterogeneity on bird communities found in forests ecosystems, within boreal, temperate, and montane biomes. Articles were retrieved from the Web of Science Core collection (October 2023) and supplemented with additional entries from Google Scholar (June 2024). We searched the selected terms within titles, abstracts, and keywords in Web of Science while Google Scholar applies the search string across the entire text. The review was not restricted by publication date. To refine our selection, we used specific terms (e.g. “species diversity” instead of “diversity”, “landscape composition” instead of “landscape”), to exclude, for instance, studies that referenced landscape context without explicitly analysing landscape heterogeneity effects. This approach ensured a more relevant dataset aligned with our inclusion criteria. The Web of Science search was made through University of Jyväskylä institutional access. The search string used in Web of Science and Google Scholar was the same: ("bird*" OR "avian" OR "avifauna") AND ("biodiv*" OR "species diversity" OR "species richness") AND (forest* OR "woodland*" OR "wood land*") AND ("boreal*" OR "temperate*" OR "montane" OR "mountain") AND ("landscape configuration*" OR "landscape composition*" OR "landscape structur*" OR "landscape heterogen*" OR "landscape homogen*" OR "landscape scale*" OR "landscape complex*") NOT (Urban OR tropical) The Web of Science search yielded 160 articles, to which we added 21 articles from Google Scholar, leading to a total of 181 articles considered in the review. Inclusion criteria We scrutinized all the 181 articles and excluded those conducted outside the target biomes (i.e. temperate, montane and boreal) or not studying birds. Only studies that statistically assessed the effects of landscape heterogeneity on bird communities were included. We removed articles where landscapes were not described quantitatively, e.g. described as background context. Additionally, we retained only studies using direct biodiversity metrics (e.g. taxonomic, phylogenetic and functional diversity, abundance, or species composition change), excluding those focused on single species or ecological processes such as seed dispersion. Only field-based empirical studies were considered, leading to exclusion of simulations, reviews, and meta-analyses. Finally, we included studies where observations were conducted at least partially in forested areas. After this selection process, data were extracted from 45 articles (Fig. 1). Data extraction First, we extracted general study information, i.e. publication year, location (country, geographic coordinates, and biome), main landscape features, bird sampling season, location of observation/sampling unit (forest-only, predominantly or partially in forest), scale of biodiversity measurement (i.e. alpha, beta, or gamma), sample size, number of tested landscape size(s) (usually buffers), and the minimum and maximum size of tested landscape scales. Secondly, we extracted all tested relationships between the landscape and bird biodiversity metrics, including both significant and non-significant results. For significant relationships, we recorded the direction of the effect (positive or negative) and the shape of the relationship if available (e.g. hump-shape), or the assemblage change when applicable. In studies incorporating local variables alongside landscape metrics, we also recorded their name and effects. To facilitate comparisons, we categorized landscape metrics into two hierarchical levels (Supp. Inf. Tab. S1). First, similar metrics used under different names were merged (e.g. “Deciduous wood cover”, “Amount Deciduous Forest”, “%Broadleaf Forest” were combined as “Amount Broadleaf Forest”). Second, metrics were classified based on heterogeneity component (composition or configuration) and focal habitat. For example, general habitat composition (e.g. Shannon diversity of land cover) and forest composition (e.g. amount of broadleaf forest) were treated separately. Hereafter, we use “amount” to refer to metrics quantifying specific habitat in the landscape, typically measured as proportional area. Finally, all metrics were categorized by heterogeneity component: composition, configuration, or mixed (e.g. “amount of forest core” which incorporates both). We extracted the specific biodiversity metrics (e.g. species richness) and biodiversity type (e.g. taxonomic diversity). When the entire bird community was studied, results were categorized as “All spp.”. For studies analysing species subgroups based on biological traits, we classified results accordingly (e.g. migration status: migrant vs. resident). Given our focus on bird communities in forests, we distinguished species by habitat preference (forest vs. open areas). For example, cavity nesters were classified as forest species, while early succession bird species were classified as open areas species (see Supp. Inf. Tab. S2 for details). For each single result, we also recorded the scale of biodiversity measurement (i.e. local alpha or landscape-scale gamma diversity) and, for gamma-diversity, the spatial scale of the measurement. To assess the landscape scale of effect on bird communities in forest, we analysed a subset of studies comparing the strength of landscape metrics across multiple spatial scales around the sampling unit. The scale of effect was defined as the spatial scale yielding the strongest (magnitude), the most significant (p.value), or the best fitting (AIC or R2) relationship between a landscape and a biodiversity metric.