Climate-induced forest dieback drives compositional changes in insect communities that are more pronounced for rare species

Species richness, abundance and biomass of insects have recently undergone marked declines in Europe. We metabarcoded 211 Malaise-trap samples to investigate whether drought-induced forest dieback and subsequent salvage logging had an impact on ca. 3000 species of flying insects in silver fir Pyrenean forests. While forest dieback had no measurable impact on species richness, there were significant changes in community composition that were consistent with those observed during natural forest succession. Importantly, most observed changes were driven by rare species. Variation was explained primarily by canopy openness at the local scale, and the tree-related microhabitat diversity and deadwood amount at landscape scales. The levels of salvage logging in our study did not explain compositional changes. We conclude that forest dieback drives changes in species assemblages that mimic natural forest succession, and markedly increases the risk of catastrophic loss of rare species through homogenization of environmental conditions.


Supplementary Figures
Supplementary Figure 1: Observed and estimated insect MOTUs richness.
MOTUs richness observed and estimated, based on Chao2 and Jack1 (first order jackknife) analyses for the 56 plots. While the recovered insect diversity of the present study is of 2972 MOTUs, both estimation methods give around 4000 MOTUs trappable using Malaise trap on the four-months sampling period considered. Dark green bars show the total number of MOTUs recovered for each of the five most represented insect orders (Diptera, Hymenoptera, Lepidoptera, Coleoptera and Hemiptera).
Grey-blue, blue and and light-blue bars represent the number of MOTUs for each order that has been associated with no ambiguity to family, genus and species level, respectively. Taxonomic assignment was performed using BOLD System DNA reference libraries from April 2019. Zeta-diversity analyses per Zeta order (i.e. ζi) here referring to plots (from ζ2 to ζ56) for two computing schemes. Representations consider non-geographically structured scheme that computes all combinations and assemblages (ALL) with parameter sample set to 5000 and Monte-Carlo (mc) sampling. (a, b) Zeta-diversity decline and 0 to 1 scaled ratio of Zeta diversity decline, representing species shared and species retention rate (i.e. the retention probability of common species in the community) across ζi, respectively. (c, d) Zeta-diversity model fitting to exponential and power-law regressions, respectively. Representation of the species retention rate (i.e. zeta ratio) per plot (i.e. Zeta order) following all plot combinations scheme (ALL) with parameter sample set to 5000 and Monte-Carlo (mc) sampling for low, medium and high dieback levels, respectively (a, b, c) for the five main insect Orders (Coleoptera, Diptera, Hemiptera, Hymenoptera and Lepidoptera) and (d, e, f) for the four main ecological functions recovered from taxonomic assignment (floricolous / non floricolous and parasitoid / non-parasitoid species). Green line with plain dots represents mean species retention rate of the total dataset in each respective dieback category. Increasing curves express that common MOTUs are more likely to be retained in additional samples than rare ones (with presence of common species over all plots if zeta ratio = 1) and decreasing curves indicates species turnover. Representation of the species retention rate (i.e. zeta ratio) per plot (i.e. Zeta order) following all plot combinations scheme (ALL) with parameter sample set to 5000 and Monte-Carlo (mc) sampling for healthy, disturbed and salvaged stands, respectively (a, b, c) for the five main insect Orders (Coleoptera, Diptera, Hemiptera, Hymenoptera and Lepidoptera) and (d, e, f) for the four main ecological functions recovered from taxonomic assignment (floricolous / non floricolous and parasitoid / non-parasitoid species). Green line with plain dots represents mean species retention rate of the total dataset in each respective dieback category. Increasing curves express that common MOTUs are more likely to be retained in additional samples than rare ones (with presence of common species over all plots if zeta ratio = 1) and decreasing curves indicates species turnover.

Supplementary Tables
Supplementary Table 1: Impact of forest dieback and salvage logging on insect communities.
Generalized linear models from mvabund object for community composition changes of different study groups (i.e. total insects, five most represented insect orders, and four functional group) compared across respective environmental conditions of diebacks (i.e. low, medium and high forest dieback levels) and stand types (i.e. healthy, disturbed and salvaged logged. Functional groups (i.e. parasitoid/non-parasitoid, floricolous/non-floricolous insects) were assigned using each MOTU's taxonomic family. Reported values are the two extremes given from 10 consecutive analyses. Post-Hoc Holm correction was applied on the complete 10 trials range of each group. Significance is given by "*" with relative values in bold, while "N.S." stands for non-significant. We highlight that all studied groups but Coleoptera had community composition changes driven by forest dieback. Significant compositional changes also arose in each case between study districts.   14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 N. 14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 N.