A shortcut to sample coverage standardization in metabarcoding data provides new insights into land-use effects on insect diversity

Abstract

Identifying key drivers of insect diversity decline in the Anthropocene remains a major challenge in biodiversity research. Metabarcoding has rapidly gained popularity for species identification, yet the lack of abundance data complicates accurate diversity metrics like sample coverage-standardized species richness. Additionally, the vast number of taxa lacks a unified phylogeny or trait database. We introduce a new workflow for metabarcoding insect data that constructs a phylogenetic tree for most insect families, standardizes sample coverage and assesses both taxonomic and phylogenetic diversity along the Hill series. Applying this workflow to Central Europe, we analysed insect diversity from 400 families across a land-use gradient. Our results show that land-use intensity significantly affects sample coverage, highlighting the necessity of biodiversity standardization. Taxonomic diversity declined by 27-44% and phylogenetic diversity by 13-29% across 39 000 operational taxonomic units, with diversity decreasing from forests to agricultural areas. When focusing on rare species communities exhibited greater phylogenetic diversity loss than taxonomic diversity, whereas dominant species experienced smaller phylogenetic losses but more pronounced declines in taxonomic diversity. Our findings underscore the detrimental effects of agriculture on insect taxa and reveal a dramatic loss of phylogenetic diversity among rare species with potential consequences for ecosystem stability.

Keywords: Hill numbers; biodiversity; climate change; phylogenetic diversity; phylogenetic tree; taxonomic diversity.

Figure 1.

Phylogenetic tree of 39k OTUs from 400 insect families and their diversities. The main insect orders are depicted in colour (Hymenoptera in orange, Hemiptera in blue, Orthoptera in lilac, Diptera in aquamarine, Lepidoptera in yellow and Coleoptera in red). The remaining smaller orders are depicted in grey. The dots at the end of the branches represent families, with the dot size indicating the number of OTUs within each family. The tree was calibrated with fossil calibration points from Rainford et al. [29].

Figure 2.

Overview of the new workflow. Our workflow includes a collection of previously published methods (blue), linked by new and updated code (orange). Newly developed code has also been used to combine the different methods. The workflow now allows us, for the first time, to analyse metabarcoding data with a phylogenetic tree in iNEXT (orange arrow). The numbers in the blue and orange boxes refer to the R scripts in the Zenodo repository [38].

Figure 3.

Model coefficients of multiple regressions on distance matrices, calculated with focus on rare (q = 0), common (q = 1) and dominant (q = 2) species. Models included geographic distances (based on geographic coordinates in m), distances of regional and local land-use categories, day of the year and climate and weather. Models were calculated for coverage standardized dissimilarity matrices along the Hill numbers (q = 0, q = 1, q = 2) to control for relative frequency distributions of reads generated during metabarcoding. Significant results (p ≤ 0.001) are marked with three stars.

Figure 4.

Multiplicative partial effects of land use on taxonomic and phylogenetic diversity based on generalized additive mixed models. Models included local and regional land-use categories and controlled for season, climate, weather and space. Taxonomic diversity values were calculated for a standardized sample coverage of 99.6%. The values are in comparison to (a) the local land-use type forest and (b) regional land-use type semi-natural. Models were calculated along the Hill numbers (q = 0, q = 1, q = 2) to control for relative read distributions. We display all model parameters in electronic supplementary material, table S1. Letters indicate significant differences between categories (p < 0.05).