A systematic survey of regional multi-taxon biodiversity: evaluating strategies and coverage

Abstract

Background: In light of the biodiversity crisis and our limited ability to explain variation in biodiversity, tools to quantify spatial and temporal variation in biodiversity and its underlying drivers are critically needed. Inspired by the recently published ecospace framework, we developed and tested a sampling design for environmental and biotic mapping. We selected 130 study sites (40 × 40 m) across Denmark using stratified random sampling along the major environmental gradients underlying biotic variation. Using standardized methods, we collected site species data on vascular plants, bryophytes, macrofungi, lichens, gastropods and arthropods. To evaluate sampling efficiency, we calculated regional coverage (relative to the known species number per taxonomic group), and site scale coverage (i.e., sample completeness per taxonomic group at each site). To extend taxonomic coverage to organisms that are difficult to sample by classical inventories (e.g., nematodes and non-fruiting fungi), we collected soil for metabarcoding. Finally, to assess site conditions, we mapped abiotic conditions, biotic resources and habitat continuity.

Results: Despite the 130 study sites only covering a minute fraction (0.0005%) of the total Danish terrestrial area, we found 1774 species of macrofungi (54% of the Danish fungal species pool), 663 vascular plant species (42%), 254 bryophyte species (41%) and 200 lichen species (19%). For arthropods, we observed 330 spider species (58%), 123 carabid beetle species (37%) and 99 hoverfly species (33%). Overall, sample coverage was remarkably high across taxonomic groups and sufficient to capture substantial spatial variation in biodiversity across Denmark. This inventory is nationally unprecedented in detail and resulted in the discovery of 143 species with no previous record for Denmark. Comparison between plant OTUs detected in soil DNA and observed plant species confirmed the usefulness of carefully curated environmental DNA-data. Correlations among species richness for taxonomic groups were predominantly positive, but did not correlate well among all taxa suggesting differential and complex biotic responses to environmental variation.

Conclusions: We successfully and adequately sampled a wide range of diverse taxa along key environmental gradients across Denmark using an approach that includes multi-taxon biodiversity assessment and ecospace mapping. Our approach is applicable to assessments of biodiversity in other regions and biomes where species are structured along environmental gradient.

Keywords: Abiotic gradients; Biotic factors; Continuity; Denmark; Disturbance; Moisture; Productivity; eDNA.

Fig. 1

a Map of Denmark showing the location of the 130 sites grouped into 15 clusters within five regions (Njut: Northern Jutland, Wjut: Western Jutland, Ejut: Eastern Jutland, FLM: Funen, Lolland, Møn, Zeal: Zealand). b Site layout with four 20 × 20 m quadrants each containing a 5 m radius circle (plot) (Reprinted and modified from Ejrnæs et al. [50], Copyright (2018), with permission from Elsevier)

Fig. 2

Validation of the stratification scheme used in site selection. Boxplots of measured values of nutrient levels (soil N g/m²), moisture levels (trimmed site mean % Volumetric Water Content (VWC)), and vegetation height (mean LIDAR canopy height (m)) for the a 90 natural sites of different fertility levels (infertile, fertile), moisture levels (dry, moist, wet), and successional stages (early (open), mid (herb/scrub), late (forest)) and b the 90 natural sites, 15 plantations, 15 fields and 10 perceived areas of high species richness (HighSpcRich)

Fig. 3

95 percentile convex hull plots of Ellenberg F, L and N values from a reference data set (http://www.naturdata.dk) of open and forest habitat types (blue, n = 59,227) as well as the data set used in this study, Biowide (red, n = 130). Black dots represent Ellenberg values of the 130 Biowide sites

Fig. 4

Habitat mean values for various carbon resources in the 130 40 × 40 m sites. Volume of dead wood (m³/ha), density of dung (cow, sheep, deer, horse, hare) (number/m²), summed flower density in April, June and August (number/m²), litter mass (g/m²) and plant species richness per site are depicted for natural habitat types (early, mid and late successional stage), arable sites and plantations

Fig. 5

Three dimensional NMDS plots for plants with a showing axis 2 against axis 1 and b showing axis 3 against axis 2 and fungi with c showing axis 2 against axis 1 and d showing axis 3 against axis 1. The three main gradients used for selecting the 130 sites (fertility, moisture, successional stage) are overlaid as arrows (from an envfit analyses in the R package Vegan). The ordinations are based on plant species lists from the 130 sites (a, b) or macrofungi species lists from the 124 sites with more than five species (c, d) and the arrows reflect soil moisture measured using a soil moisture meter, fertility measured as soil N and light measured as light intensity using HOBO loggers. The ordination plots illustrate that the community composition of vascular plants and macrofungi actually reflect the main gradients the sites were selected to cover. The scatter of dots shows the variation in abiotic conditions across the 130 sites. Correlations and p-values can be seen in Additional file 8: Appendix H

Fig. 6

Correlation between a observed site plant species richness and plant OTU site ‘richness’ for the 130 sites (Spearman Rho: R² = 0.652, S = 70,457, p-value < 0.001), b–d observed site plant community composition and plant OTU community composition for the 130 sites, b NMDS axes 1 (Spearman Rho: R² = 0.576, S = 644,210, p-value < 0.001), c NMDS axes 2 (Spearman Rho: R² = 0.594, S = 648,480, p-value < 0.001), and d NMDS axes 3 (Spearman Rho: R² = 0.697, S = 671,850, p-value < 0.001)

Fig. 7

Cross correlation among the main taxonomic groups included in the study. The colour and shape of the symbol is scaled according to spearman rank correlation coefficients and non-significant (p > 0.05) correlations are indicated by a cross