GoKawiilStudy sites and sampling design
We used a pan-European network of 64 mature, uneven-aged forest plots (30 × 30 m2) consisting of three-species mixture stands (34 plots) and corresponding monospecific stands (30 plots; Extended Data Table 1). These plots are part of the FunDivEUROPE (Functional Significance of Forest Biodiversity in Europe) exploratory platform59, and were established across European forests over 2011–2012 to investigate the role of the diversity and composition of regionally common and economically important tree species on ecosystem functioning. The studied plots were distributed across four locations featuring different European forest types and spanning a large biogeographic gradient: North Karelia (Finland), Białowieża (Poland), Râşca (Romania) and Colline Metallifere (Italy), corresponding to typical boreal, hemi-boreal, mountainous beech and thermophilous deciduous (Mediterranean) forests, respectively (Supplementary Table 4 and Supplementary Fig. 5a). In each location, plots were carefully selected based on tree species richness and composition while minimizing as much as possible covariation with potentially confounding environmental factors such as topography and soil conditions59 (Supplementary Fig. 2a). Plot selection was performed so as to include monospecific stands of all tree species from the local species pool and replicate the three-species mixture treatment with different tree species combinations while maximizing community evenness (Extended Data Table 1). This allowed strict avoidance of a dilution gradient, such as would occur in a design with monospecific stands of only one species combined with mixture stands including this species, along with a clear distinction between the effects of species mixing and composition. Our stratified plot selection procedure enabled us to mimic formal biodiversity experiments, given that such manipulative approaches are virtually impossible to undertake in mature forests owing to the high longevity of tree species. Tree species diversity and composition in the studied plots were predominantly the result of natural community assembly from the regional species pool, combined with local forest management practices. The investigated levels of species richness, that is, one versus three tree species, are typical for European forest ecosystems (https://forest.eea.europa.eu/topics/forest-biodiversity-and-ecosystems/forest-ecosystems), boreal forests in Asia and North-America, as well as managed forests and plantations worldwide, although clearly much less diversified than many (sub)tropical forests and some mid-latitude temperate forests47. Overall, our sampling design encompassed a total pool of 13 tree species, including 12 ectomycorrhizal and one arbuscular mycorrhizal tree species, and the local species pool ranged from three to five tree species per location (Extended Data Table 1 and Supplementary Table 4).
Soil organism sampling and analysis
In each plot, we assessed energy fluxes through the soil food web by measuring the biomass of major groups of organisms within this food web22,23,43,60, including both microbial (bacteria and fungi) and faunal (nematodes, microarthropods and macroinvertebrates) groups. Biomass data for all soil organism groups were expressed per unit surface area (g dry weight m−2) at the plot level. Further details on biomass calculation and the trophic classification of soil organisms are provided in the Supplementary Methods.
Sampling
Soil organisms were sampled in all plots during the phenological spring of 2017 (Supplementary Table 4), a period of high soil biological activity. We sampled both the litter layer (unfragmented aboveground litter, OL horizon) and the soil layer (including both fragmented/humified organic matter and mineral soil, OF/OH/A horizons). In each plot, we selected five 10 × 10-m2 subplots, with samples taken equidistantly from three trees of either the same species in monospecific stands or of different species for three-species mixture stands (Supplementary Fig. 5b). For each subplot, soil samples for microbial analyses and nematode extraction were collected by taking five soil cores (10-cm depth, 5.3-cm diameter), spaced approximately 35 cm apart around the equidistant point between the three trees, weighted by tree individual size, that is, individual diameter at breast height. The five cores were gently sieved through 6-mm mesh (to avoid damaging nematodes), homogenized and pooled at the subplot level for nematode extraction. Pooled soil was then sieved through 2-mm mesh for microbial analyses. All subsamples were stored at 4 °C until further processing. For microarthropod extraction, an intact core (10-cm depth, 10-cm diameter), including both the litter and soil layers, was collected from each of three subplots along a southwest–northeast transect and stored at 4 °C until further processing. For the hand-sorting of soil macroinvertebrates, an intact monolith (25-cm depth, 25 × 25-cm2 surface), including both the litter and soil layers, was collected from each of the same three subplots. To express all data per unit surface area, an extra core was sampled, sieved through 2-mm mesh, dried at 105 °C for 48 h and weighted to measure soil bulk density.
Microorganisms
The biomass of bacteria (gram-positive and gram-negative), arbuscular mycorrhizal fungi and non-arbuscular mycorrhizal fungi was quantified at the plot level using phospholipid fatty acid data61. Fungal community data based on metagenomic amplicon sequencing and bioinformatics62 were used to partition non-arbuscular mycorrhizal fungal biomass into five trophic guilds: ericoid mycorrhizal fungi, ectomycorrhizal fungi, general saprotrophic fungi, wood saprotrophic fungi and plant pathogenic fungi (Supplementary Methods). The biomass of each fungal trophic guild was calculated by multiplying its relative abundance, that is, number of reads divided by the total number of reads for the five trophic guilds, by the total non-arbuscular mycorrhizal fungal biomass.
Nematodes
Nematodes were extracted for each subplot within 72 h after sampling from approximately 100 g of fresh soil using a modified sugar flotation method63, before being heat-killed and fixed in 4% formaldehyde. Nematodes were then pooled and counted at the plot level, and a subsample of approximately 160 randomly selected individuals were identified to family level. The biomass of nematode families was calculated based on body mass data retrieved from the Nemaplex database (http://nemaplex.ucdavis.edu). Nematode families were assigned to five trophic guilds64: herbivores, bacterivores, fungivores, omnivores and carnivores.
... continue reading