1. Fungal mycelium is increasingly recognized as a central component of soil biogeochemical cycling, yet our current understanding of the ecological controls on fungal necromass decomposition is limited to single sites and vegetation types.

2. By deploying common fungal necromass substrates in a temperate oak savannah and hardwood forest in the midwestern USA, we assessed the generality of the rate at which high- and low-quality fungal necromass decomposes; further, we investigated how the decomposer ‘necrobiome’ varies both across and within sites under vegetation types dominated by either arbuscular (AM) or ectomycorrhizal (EM) plants.

3. The effects of necromass quality on decay rate were robust to site and vegetation type differences, with high-quality fungal necromass decomposing, on average, 2.5 times faster during the initial stages of decay. Across vegetation types, bacterial and fungal communities present on decaying necromass differed from bulk soil microbial communities and were influenced by necromass quality. Moulds, yeasts and copiotrophic bacteria consistently dominated the necrobiome of high-quality fungal substrates.

4. Synthesis: We show that regardless of differences in decay environments, high-quality fungal substrates decompose faster and support different types of decomposer microorganisms when compared with low-quality fungal tissues. These findings help to refine our theoretical understanding of the dominant factors affecting fast cycling components of soil organic matter (SOM) and the microbial communities associated with rapid decay.

Parallel necromass decomposition experiments were conducted at two sites and under two different vegetation types at each site: a temperate savannah containing EM-associated trees and AM-associated grasses and a temperate hardwood forest containing adjacent EM- and AM-dominated stands. The savannah site was located at Cedar Creek Ecosystem Science Reserve in central Minnesota, USA (N 45.42577 W 093.20852). Cedar Creek is a 2266 ha reserve affiliated with the University of Minnesota, which contains of a mix of prairie and forest ecosystems. The mean annual temperature at Cedar Creek is 6.7°C and the mean annual precipitation is 801 mm. The forest site was located at Moores Creek in south-central Indiana, USA (N  39.08333 W  086.46666). Moores Creek, which is part of the Indiana University Research and Teaching Preserve system, is comprised of 105 ha of mixed deciduous hardwood forest (~80 years in age). The mean annual temperature at Moores Creek is 11.6°C and the mean annual precipitation is 1200 mm. Within both sites, vegetation communities differed in their AM- and EM-associated plant species and edaphic characteristics. 

Plot locations at each site were chosen based on dominant vegetation type and mycorrhizal association. Three replicate plots were established at two locations in the savannah site; 10 m into EM-dominated Quercus forest and 20 m into the adjacent AM-dominated grassland. These distances were chosen based on previous work at the same site by Dickie and Reich (2005), which found little to no EM colonization of Quercus seedlings at 20 m away from the forest edge. At the forest site, 7 replicate plots were established based on known mycorrhizal associations of dominant tree species. In all plots, trees from the dominant mycorrhizal type (AM or EM) represented >85% of the basal area of the plot and AM and EM plots were paired according to geographic proximity.

Two fungal species, Mortierella elongata and Meliniomyces bicolor, were chosen to represent high- and low-quality necromass. M. elongata is a fast-growing saprotrophic fungus in the phylum Mucromycota, which is frequently found in both forest and agricultural soils (Li et al., 2018). M. bicolor is an EM and ericoid mycorrhizal (ErM) Ascomycotan fungus frequently found in temperate and boreal forest soils. These two species have contrasting melanin and nitrogen levels, with M. elongata representing a high-quality substrate and M. bicolor representing a low-quality substrate. Complete details on the methods used for the chemical characterization of both species are provided in the online Supplementary Information.

Fungal biomass for both species was produced in liquid cultures by individually inoculating 50 mL flasks containing half-strength potato dextrose broth with 3 mm diameter mycelial plugs (one plug per flask). Following inoculation, cultures were transferred to an orbital shaker and left to shake at 80 rpm for at least 30 days or until growth stopped. To produce fungal necromass, cultures were rinsed with distilled water and dried at 26°C for 24 hours. Dried fungal necromass (~25 mg) were then placed into nylon mesh litter bags constructed from 53-micron mesh (Elko, Minneapolis, MN, USA) and heat-sealed. The 53-micron mesh size excluded both tree and grass root in-growth.

Separate litter bags were constructed for replicates of each fungal species. During deployment, litter bags were buried at organic-mineral soil interface (0-5 cm depth). To determine if there was any mass loss due to transport and handling, an additional set of litter bags was carried into the field (n = 3). Necromass recovery was greater than 98% and did not differ between fungal species, so masses were not corrected for any loss during transport. At each harvest, litter bags were individually bagged, placed on ice, and taken directly to the laboratory for processing. For each sample, necromass was carefully removed from the litterbag and dried at 30°C to a constant mass to determine mass remaining.

While the preparation and processing of fungal necromass was standardized across the two sites, the specific incubation times varied slightly between studies due to logistical constraints. At the savannah site, fungal necromass was incubated for 14, 28, 42, and 56 days beginning in July 2017 (n = 3 litter bags of each fungal species for each vegetation type for each sampling date). At the forest site, fungal necromass was incubated for 14, 31, and 92 days beginning in July 2017 (n = 7 litter bags of each fungal species for each vegetation type for each sampling date). Soil moisture measurements at both sites were taken at the time litter bags were harvested. Gravimetric soil moisture data was collected from the composite of two 5 × 10 cm soil cores per plot at the savannah site and three 6.35 × 10 cm soil cores per plot at the forest site. To determine pH, a sub-sample of soil collected at the time of the first litter bag harvest was air-dried and analyzed in a 0.01 M CaCl₂ solution using a bench-top pH meter. 

Statistical analyses and data visualization were conducted in R version 3.5.1 (R Core Team, 2018). Analysis of variance (ANOVA) was used to test for differences in soil pH between sites and vegetation types within sites. To test for differences in soil moisture, ANOVAs were run with vegetation type (AM- vs. EM-associated vegetation) and sampling date as the predictor variables for each site. Prior to running the ANOVAs, soil moisture data was log-transformed to meet the assumptions of normality. Linear mixed-effect (LME) models were used to analyze the amount of fungal necromass remaining within each site. Fixed predictor factors included vegetation type (AM- vs. EM-associated vegetation), necromass type (high- vs. low-quality), incubation period, and soil moisture. Replicate sampling locations (either plots or plot pairs) were designated as a random factor. Because pH was only measured during one harvest at each site it was not included in this analysis. Mass remaining data were log logit-transformed to meet statistical assumptions.To evaluate the significance of linear mixed-effects models the Kenward-Roger approximation was used to estimate F statistics and denominator degrees of freedom (Halekoh & Højsgaard, 2014). Least square means were computed for each fixed effect and post-hoc comparisons were carried out on pairs of the least-squares means using the Tukey’s adjustment for multiple comparisons.

Decay constants were calculated separately for each necromass type at each site. To calculate decay constants, we fit the proportion of remaining necromass against incubation time (days) using single- and double-exponential decay models. The best fitting model was selected using Akaike’s Information Criteria (AIC). According to AIC values, a double-exponential decay model (Equation1) produced the best fit.

Equation 1: [mass]_t= ae^-k₁t+(1-a) e^-k₂t

The proportional mass remaining ([mass]_t) was calculated by dividing the mass remaining at time (t) by the initial mass for each litterbag. In equation 1, a refers to is the initial proportion of fast decomposing or labile material, 1-a is the initial proportion of slow decomposing or recalcitrant material. k₁ and k₂ are the degradation rate constants of the labile (fast-decomposing) and recalcitrant (slow-decomposing pool), respectively. The nonlinear least-squares Levenberg-Marquardt algorithm used to estimate model parameters, a, k₁ and k₂, using the ‘minpak.lm’ package. 

The microbial community data that accompanies the fungal necromass decomposition data is available under the following NCBI BioProject Accessions: Moores Creek Bacteria (PRJNA607032), Moores Creek Fungi (PRJNA607029), Cedar Creek Bacteria (PRJNA607034), Cedar Creek Fungi (PRJNA607030).