Plant species composition and local habitat conditions as primary determinants of terrestrial arthropod assemblages

Abstract

Arthropods respond to vegetation in multiple ways since plants provide habitat and food resources and indicate local abiotic conditions. However, the relative importance of these factors for arthropod assemblages is less well understood. We aimed to disentangle the effects of plant species composition and environmental drivers on arthropod taxonomic composition and to assess which aspects of vegetation contribute to the relationships between plant and arthropod assemblages. In a multi-scale field study in Southern Germany, we sampled vascular plants and terrestrial arthropods in typical habitats of temperate landscapes. We compared independent and shared effects of vegetation and abiotic predictors on arthropod composition distinguishing between four large orders (Lepidoptera, Coleoptera, Hymenoptera, Diptera), and five functional groups (herbivores, pollinators, predators, parasitoids, detritivores). Across all investigated groups, plant species composition explained the major fraction of variation in arthropod composition, while land-cover composition was another important predictor. Moreover, the local habitat conditions depicted by the indicator values of the plant communities were more important for arthropod composition than trophic relationships between certain plant and arthropod species. Among trophic groups, predators showed the strongest response to plant species composition, while responses of herbivores and pollinators were stronger than those of parasitoids and detritivores. Our results highlight the relevance of plant community composition for terrestrial arthropod assemblages across multiple taxa and trophic levels and emphasize the value of plants as a proxy for characterizing habitat conditions that are hardly accessible to direct environmental measurements.

Keywords: Community ecology; Ellenberg indicator values; Functional groups; Plant–insect interactions; Species-environment relationships.

Fig. 1

Distribution of 60 landscapes (5.8 km × 5.8 km) within the federal state Bavaria. Insets show examples of a forest- and an urban-dominated landscape containing three study sites representing different local habitat types

Fig. 2

Proportion of variance explained (adjusted R²) in the assemblage composition of five functional (a) and four taxonomic arthropod groups (b). Two bars are shown for each group: the upper bar indicates independent proportions of variance explained by plant species composition, land-cover composition, climate and space. The lower bar indicates proportions of variance explained by the combination of two (colored segments) or more predictors (grey segment). Total adjusted R²-values are shown next to the bars. R²-values of single fractions are given in Tables S6. Arthropod dissimilarities between sites were calculated using the Sørensen index. Predictive dissimilarity matrices were calculated using the Sørensen index (plant species composition) and Euclidean distances (land cover, climate and space)

Fig. 3

Proportion of variance explained (adjusted R²) in the assemblage composition of five functional (a) and four taxonomic arthropod groups (b). Bar segments indicate independent and shared effects of plant species composition and Ellenberg indicator values. Arthropod dissimilarities were calculated using the Sørensen index. Predictive dissimilarity matrices were calculated using the Sørensen index (plant species composition) and Euclidean distances (indicator values)

Fig. 4

Histograms and t-values (significance levels: ***p < 0.001, **p < 0.01, *p < 0.05) comparing independent proportions of variance explained (adjusted R²) in the assemblage composition of five functional (a) and four taxonomic arthropod groups (b). Red lines indicate variance explained by plant species composition using a subset of host plants for monophagous herbivores, compared to 1000 randomly selected subsets of equal size (grey bars)