{Landscape metrics as indicators of avian diversity and community measures}

The identification of landscape metrics that could be used as surrogate of biodiversity is still of broad and current interest in ecology. Several landscape metrics used to provide a quantitative description of environmental structure and related to the landscape heterogeneity were proposed in ecological modelling procedures as surrogates or indicators of biodiversity metrics. Most frequently, the species occupancy in a given area is related to the niche availability, which is correlated with the spatial heterogeneity of the landscape (e.g. number of different land use patches). However, the effectiveness of biodiversity indicators (biotic indicators: e.g. species, group of species; abiotic indicators: e.g. environmental characteristics, landscape metrics) is still discussed and more efficient surrogates are needed. In this study, we explored the associations among the most common landscape metrics and several diversity and community metrics calculated for bird assemblages in the Czech Republic. Using Generalized Linear Models, we compared the strength and direction of these associations as well as their performance in three different environments. Overall, taxonomic diversity was explained by landscape metrics most accurately, even across different types of environments. The most effective landscape metric for bird species richness was the mean patch size, which was negatively correlated. In mixed environments, the functional evenness was positively correlated with the Simpson evenness, the reason probably lying in the fact that both are measures of the regularity of the distribution of relative abundances. Finally, the surrogacy of landscape metrics was weak in forest environments, where even the most effective predictor, the Simpson evenness was only poorly associated to diversity metrics. In this regard, we hypothesize that for modelling more accurately diversity metrics in forest environments, vertical data (e.g. vegetation structure, LiDAR) could be required. Our findings are useful for ecological modelling, enabling the selection of the most appropriate landscape metrics to predict each diversity metric.