Birds as indicators of landscape changes: a lesson from southern Mediterranean semi-arid streams

Benchaiba, Lamir , Elafri, Ali , Telailia, Salah

2026-01-01 LANDSCAPE AND ECOLOGICAL ENGINEERING 2026   22(卷), 1(期), (49-62页)

查看原文

We investigated the impacts of landscape changes on bird communities in semiarid/arid southern Mediterranean streams. Bird surveys were conducted at 62 sampling riparian units during the breeding season. Patterns in species composition were explored with advanced statistical tools including Multivariate generalized linear models (ManyGLM), non-metric ordinations, and generalized linear mixed models (GLMMs). A total of 4158 individuals belonging to 107 species across 33 families were recorded. The most diverse family was Muscicapidae, with 19 species. The most parsimonious model explaining changes in breeding bird assemblages, as indicated by the lowest AIC, retained tree cover, shrubland, grassland, and aridity intensity (IDM) as significant predictors (P < 0.05). In contrast, stream-related variables such as slope, flood-prone areas, and proximity to surrounding forests, along with human-related factors including fruit orchards and built-up surfaces, were not retained in the final model (P > 0.05). The ANOVA.manyglm test identified ten indicator species (from the 27 resident breeders analyzed; P < 0.05), including three North African endemics, Cyanistes teneriffae, Phoenicurus moussieri, and Emberiza sahari, which are of particular relevance for stream management and conservation efforts. Among the environmental gradients, tree cover emerged as the most influential land cover variable, showing the strongest positive association with bird richness and abundance (as confirmed by higher GLMM estimates). This variable may also buffer or mitigate the negative effects of increasing aridity. Finally, no evidence was found to suggest that human-related land-use activities significantly shaped the composition of breeding bird assemblages, likely due to their overall low intensity in the study area, where such land uses did not exceed an average of 25% of riparian land cover.