Overview

This analysis examines fungal community composition across an experimental carrion decomposition system, comparing mycobiomes from:

  • Carcass swabs — Surface fungi on decomposing rabbit carcasses
  • Gravesoil — Soil directly beneath carcasses
  • Flies (Cochliomyia macellaria) — Blow fly mycobiomes
  • Beetles (Necrodes surinamensis) — Carrion beetle mycobiomes (abdomen and head-thorax fractions)

The study tracks fungal community dynamics across three decomposition stages (Fresh, Active decay, Advanced decay) under manipulated insect diversity treatments in outdoor mesocosms.

Key Findings:

  • Habitat specialization: Strong differentiation between substrate types (PERMANOVA R² = 0.22, p < 0.001), with soil harboring highest diversity (94.4 ASVs), followed by carcass (59.5), flies (47.7), and beetles (44.6)
  • Contrasting temporal patterns: Carcass diversity peaks at Active decay (71.6 ASVs), while soil diversity declines from Fresh (133.5) to Advanced stages (81.5)
  • Network dynamics: Carcass networks show dramatic restructuring during decomposition—0 edges at Fresh stage, peak connectivity at Active decay (274 edges), then simplification during Advanced decay (46 edges)
  • Insect-fungal associations: Beetles harbor stinkhorn fungi (Phallus, Clathrus) with distinct mycobiomes between head-thorax and abdomen regions
  • Minimal treatment effects: Insect diversity treatments showed no significant impact on fungal community composition
  • High classification accuracy: Random Forest models achieved 87.6% accuracy for sample type prediction and 72.8% for decomposition stage

Table of Contents

Sequencing & Processing Library Sizes & Rarefaction Community Composition Alpha Diversity Beta Diversity Differential Abundance Differential Prevalence Correlation Testing Network Analysis Machine Learning Yarrowia Analysis Summary References

Sequencing and Processing

Sample Collection

Experimental Design:

  • 14 rabbit carcasses in outdoor mesocosms (May 2025, Huntsville, TX)
  • Four insect diversity treatments (Tr0-Tr3): natural colonization, single species, mixed community, high diversity
  • Sampling at three decomposition stages:
    • Fr (Fresh): Day 0-1
    • Ac (Active decay): Day 3
    • Adv (Advanced decay): Day 6
  • Sample types collected:
    • Carcass surface swabs
    • Gravesoil beneath carcass
    • Necrophagous fly specimens (C. macellaria)
    • Necrophagous beetle specimens (N. surinamensis, abdomen and head-thorax separated)

Library Sizes and Rarefaction

Sequencing Depth Distribution

Figure 1: Library size distribution across sample types. Boxplots show read counts per sample after DADA2 denoising. Red dashed line indicates rarefaction threshold (500 reads). Most samples exceed 2,000 reads.

Community Composition

Phylum-Level Relative Abundance

Figure 2: Phylum-level fungal community composition across sample types. Each bar represents one sample. Ascomycota and Basidiomycota dominate across all sample types, with Mucoromycota present in some carcass and soil samples.

Genus-Level Composition by Decomposition Stage

Carcass Samples

Figure 3: Genus-level relative abundance for carcass samples across decomposition stages (Fresh, Active, Advanced). Top 20 most abundant genera shown. Distinct community shifts visible across decay progression.

Soil Samples

Figure 4: Genus-level relative abundance for gravesoil samples across decomposition stages (Fresh, Active, Advanced). Community composition differs from carcass samples, reflecting soil-specific fungal populations.

Insect Mycobiomes by InsectType and Body Fraction

Figure 5: Genus-level relative abundance for insect samples. Flies are dominated by Cladosporium. Beetles show diverse mycobiomes with stinkhorn fungi (Phallus, Clathrus) and yeasts (Rhodotorula, Hasegawaea). Beetle samples are further separated by body fraction (Abdomen vs Head-Thorax) via the Spectrum variable.

Alpha Diversity

Alpha diversity measures the richness and evenness of fungal communities within samples. Four complementary metrics were calculated to capture different aspects of diversity: Observed ASV richness (simple taxon count), Shannon diversity (accounts for both richness and evenness with logarithmic weighting), Simpson’s index (probability that two randomly selected reads belong to the same taxon), and Inverse Simpson (effective number of dominant taxa). These metrics provide increasingly nuanced views of community structure, from raw richness to dominance patterns. Statistical comparisons employed non-parametric tests due to non-normal distributions typical of diversity data: Kruskal-Wallis tests for omnibus comparisons across groups, followed by pairwise Wilcoxon rank-sum tests with Benjamini-Hochberg false discovery rate correction to control for multiple testing. This approach makes no assumptions about data distribution and is robust to outliers and heteroscedasticity common in ecological datasets.

Alpha Diversity by SampleType

Figure 8: Alpha diversity metrics across sample types. Four metrics shown: Observed ASV richness, Shannon diversity, Simpson's index, and Inverse Simpson. Beetles show highest diversity, followed by soil and carcass. Flies have lowest diversity (dominated by Cladosporium).

Alpha Diversity by Decomposition Stage

Figure 9: Alpha diversity across decomposition stages for carcass and soil samples. Separate boxplots for each sample type within each facet. Diversity generally increases from Fresh to Advanced decay stages.

Alpha Diversity by Treatment (Carcass)

Figure 10: Alpha diversity by insect diversity treatment (Tr0-Tr3) for carcass samples. Each metric (Observed, Shannon, Simpson, InvSimpson) shown in separate panel. Insect diversity treatments show variable effects on carcass fungal diversity.

Alpha Diversity by Treatment (Soil)

Figure 11: Alpha diversity by insect diversity treatment (Tr0-Tr3) for soil samples. Each metric (Observed, Shannon, Simpson, InvSimpson) shown in separate panel. Insect diversity treatments show variable effects on gravesoil fungal diversity.

Alpha Diversity by InsectType

Figure 12: Alpha diversity for insect samples, comparing Fly, Beetle Head/Thorax, and Beetle Abdomen. Each metric (Observed, Shannon, Simpson, InvSimpson) shown in separate panel. Beetles show significantly higher diversity than flies across all metrics, with variation between head/thorax and abdomen regions.

Alpha Diversity Statistical Tests

Statistical test results for alpha diversity by sample type. Kruskal-Wallis omnibus tests followed by pairwise Wilcoxon comparisons with Benjamini-Hochberg correction.

Beta Diversity

Beta diversity measures between-sample differences in fungal community composition. This analysis employs multiple complementary approaches to visualize and test community structure. Principal Coordinates Analysis (PCoA) provides linear ordination of samples based on dissimilarity matrices, preserving pairwise distances between samples. Two distance metrics are used: Bray-Curtis dissimilarity, which accounts for both taxon presence and relative abundance, and Jaccard distance, a presence/absence metric that emphasizes rare taxa without weighting by abundance. Non-metric Multidimensional Scaling (NMDS) offers an alternative ordination approach that maximizes rank-order correlation with the original dissimilarity matrix, providing a complementary visualization less sensitive to PCoA’s assumption of linear relationships. PERMANOVA (permutational multivariate analysis of variance) tests whether community composition differs significantly between categorical groups using permutation-based hypothesis testing, which makes no distributional assumptions and is robust to the compositional nature of microbiome data.

PCoA Ordination (Bray-Curtis Distance)

Figure 13: Principal Coordinates Analysis (PCoA) of Bray-Curtis dissimilarity, colored by sample type. Clear separation between carcass, soil, fly, and beetle mycobiomes. Ellipses show 95% confidence intervals for each group.

PCoA by Stage (Carcass + Soil)

Figure 14: PCoA (Bray-Curtis) colored by decomposition stage, faceted by sample type. Shows temporal dynamics of fungal communities during decomposition. Fresh, Active, and Advanced stages form distinct clusters.

PCoA (Jaccard Distance)

Figure 15: PCoA of Jaccard dissimilarity (presence/absence), colored by sample type. Similar separation pattern to Bray-Curtis, but emphasizes rare taxa presence over abundance.

NMDS Ordination (Bray-Curtis)

Figure 16: Non-metric Multidimensional Scaling (NMDS) of Bray-Curtis dissimilarity. Stress value indicates goodness of fit (<0.2 is acceptable). Confirms separation between sample types observed in PCoA.

PCoA for Insects Only

Figure 17: PCoA (Bray-Curtis) for insect samples only, colored by insect type, with shapes indicating body fraction. Beetles and flies form distinct clusters, with beetle head-thorax and abdomen mycobiomes also distinguishable.

PERMANOVA Results

PERMANOVA (permutational multivariate analysis of variance) tests whether fungal community composition differs significantly between groups.

PERMANOVA results testing effects of sample type, decomposition stage, and treatment on fungal community composition. All models use Bray-Curtis dissimilarity with 999 permutations.

Differential Abundance

MaAsLin2 (Microbiome Multivariable Associations with Linear Models) identifies genera that are differentially abundant between groups. Unlike compositional methods like ANCOM-BC that focus on pairwise ratios or count-based models like DESeq2 that assume negative binomial distributions, MaAsLin2 uses a flexible multivariable framework that can simultaneously model multiple covariates and control for potential confounders in a single analysis. This approach is particularly useful for complex experimental designs with multiple factors (sample type, decomposition stage, treatment), allowing for the detection of taxon-specific associations while accounting for other variables. MaAsLin2 supports various normalization methods and can handle both categorical and continuous predictors, making it well-suited for ecological studies with heterogeneous metadata.

Figure 18: Volcano plot for SampleType effect (Soil vs Carcass). Hover over points for taxon details. Positive coefficients indicate enrichment in Soil.

Figure 19: Volcano plots for Stage effect. Left: Active vs Fresh. Right: Advanced vs Fresh. Positive coefficients indicate enrichment in later stages.

Figure 20: Volcano plots for Treatment effect (Tr1, Tr2, Tr3 vs Tr0). Positive coefficients indicate enrichment in treatment relative to control.

Insect Samples: InsectType Effect

Figure 21: Volcano plot of differential abundance for insect samples. Hover over points for taxon details. Reference levels: SampleType (Fly), Spectrum (Abdomen). Positive coefficients indicate enrichment in Beetle (vs Fly) or Head-Thorax (vs Abdomen).

All Samples: SampleType Effect

Figure 22: Volcano plot of differential abundance across all sample types. Hover over points for taxon details. Reference level: Carcass. Positive coefficients indicate enrichment in Soil, Fly, or Beetle relative to Carcass.

Differential Prevalence

MaAsLin3 identifies genera that differ in prevalence (presence/absence) in addition to abundance. While differential abundance tests detect changes in the quantity of a taxon when present, differential prevalence tests detect whether a taxon is consistently present or absent across conditions. This complementary approach is particularly useful for identifying environmental indicator taxa that may occur at low abundance but are reliably present in specific habitats, or for detecting taxa that are completely excluded from certain environments regardless of their abundance elsewhere. Differential prevalence is less sensitive to extreme abundance outliers and can reveal ecological patterns that abundance-based methods might miss.

Decomposition Samples: Stage Effect on Prevalence

Figure 23: MaAsLin3 prevalence effect plot for decomposition samples showing genera with differential prevalence (presence/absence) independent of abundance. Top: Soil vs Carcass (reference). Bottom: Active decay (left) and Advanced decay (right) vs Fresh stage (reference). Positive coefficients (red) indicate higher prevalence in test level; negative coefficients (blue) indicate lower prevalence.

Insect Samples: InsectType Prevalence

Figure 24: MaAsLin3 prevalence effect for insects. Shows genera with differential prevalence (presence/absence) independent of abundance. Reference levels: SampleType (Fly), Spectrum (Abdomen). Positive coefficients indicate higher prevalence in Beetle or Head-Thorax vs reference.

Correlation Testing

Alpha Diversity vs Continuous Metadata

Tests whether fungal diversity correlates with placement days and temperature.

Figure 25: Scatter plots showing Spearman correlations between alpha diversity metrics and continuous variables (PlacementDays, Temperature_C). Regression lines with 95% confidence intervals shown.

Table 5: Spearman correlation coefficients and p-values for alpha diversity vs continuous metadata variables.

Network Analysis

Fungal co-occurrence networks were inferred using SpiecEasi (Sparse InversE Covariance Estimation for Ecological Association Inference), which estimates conditional dependence relationships between taxa using sparse inverse covariance estimation. Networks were constructed at the genus level using the Meinshausen-Bühlmann neighborhood selection method. To focus on ecologically relevant taxa, genera were filtered to retain only those present in at least 20% of samples with a minimum total abundance of 10 reads. Model selection was performed using the StARS stability criterion with 50 subsamples across 20 lambda values (lambda.min.ratio = 0.01, seed = 42). Edge weights were assigned based on the estimated partial correlation coefficients, with positive and negative values indicating co-occurrence and mutual exclusion relationships, respectively. Networks were constructed for each sample type (Carcass, Soil, Fly, Beetle) and additionally for decomposition samples (Carcass and Soil) stratified by stage (Fresh, Active, Advanced) to capture temporal dynamics in fungal associations. Network topology metrics (nodes, edges, density, transitivity, mean degree, components, modularity) were calculated for all networks, and edge overlap analyses quantified shared associations between decomposition stages.

Overall Networks by SampleType

Carcass

Soil

Fly

Beetle

Figure 25a: Overall fungal co-occurrence networks for each sample type. Nodes = genera (size proportional to mean abundance), edges = significant associations (blue = positive, red = negative). Networks learned via SpiecEasi neighborhood selection. Fly networks show minimal connectivity compared to other sample types.

Stage-Specific Networks: Carcass

Carcass - Active Decay

Carcass - Advanced Decay

Figure 25b: Stage-specific fungal co-occurrence networks for carcass samples. Left: Active decay stage (Ac) — dense network with 274 edges among 161 genera showing peak complexity. Right: Advanced decay stage (Adv) — contracted network with 46 edges among 86 genera as dominant taxa emerge. Fresh stage omitted (0 edges detected, no associations formed).

Stage-Specific Networks: Soil

Soil - Active Decay

Soil - Advanced Decay

Figure 25c: Stage-specific fungal co-occurrence networks for gravesoil samples. Left: Active decay stage (Ac) — 317 edges among 189 genera. Right: Advanced decay stage (Adv) — 150 edges among 148 genera. Both stages show substantial network connectivity. Fresh stage soil not included due to insufficient sample size.

Network Dynamics Across Decomposition:

  • Carcass networks: No fungal associations detected in fresh stage. Network complexity explodes during active decay (274 edges), then contracts during advanced decay (46 edges) as dominant taxa emerge
  • Soil networks: Maintain high connectivity across both active (317 edges) and advanced (150 edges) stages, reflecting more stable fungal communities
  • Edge overlap: Only 10 edges shared between carcass active and advanced stages. Soil shows 21 shared edges, suggesting greater stability in fungal associations
  • Fly networks: Minimal connectivity (16 edges, 78 nodes) reflects dominance by Cladosporium with few co-occurring taxa
  • Beetle networks: Moderate connectivity (37 edges, 61 nodes) with higher transitivity, indicating clustered associations around stinkhorn fungi

Network Topology Comparison

Table 7: Network topology metrics comparing fungal networks across sample types and decomposition stages. Includes metrics for all networks, stage-specific carcass and soil networks, and edge overlap between stages showing shared associations.

Machine Learning

Random Forest classifiers were used to predict sample type and decomposition stage from genus-level fungal composition. Two models were trained using relative abundance data from rarefied samples. The first model classified all 279 samples into four sample types (Carcass, Soil, Fly, Beetle), while the second model predicted decomposition stage (Fresh, Active decay, Advanced decay) using only carcass and soil samples (n = 224). Both models were built with 1,000 trees using mean decrease in accuracy as the variable importance metric. Model performance was assessed using out-of-bag (OOB) error estimates from bootstrap samples during training, as well as 10-fold cross-validation repeated 3 times (30 total iterations per model). Cross-validation models used 500 trees for computational efficiency. Performance metrics include overall accuracy, per-class accuracy, and confusion matrices.

Predicting SampleType

Figure 26: Random Forest classification of sample type. Left: Out-of-bag (OOB) confusion matrix heatmap showing classification accuracy (rows = actual, columns = predicted). Right: Top 20 most important genera for prediction, measured by mean decrease in classification accuracy. Model achieves high classification accuracy across all sample types.

Cross-validation performance for SampleType classification. Includes overall model accuracy and per-class accuracy metrics from 10-fold cross-validation (3 repeats).

Predicting Decomposition Stage

Figure 27: Random Forest classification of decomposition stage. Left: Out-of-bag (OOB) confusion matrix heatmap showing classification accuracy (rows = actual, columns = predicted). Right: Top 20 most important genera for predicting decomposition stage (Fresh/Active/Advanced decay). Trained on carcass and soil samples only.

Cross-validation performance for Stage classification. Includes overall model accuracy and per-class accuracy metrics from 10-fold cross-validation (3 repeats).

Yarrowia as a Focal Taxon

Yarrowia is a genus of ascomycetous yeasts within the family Dipodascaceae (Saccharomycotina) that has been documented as a characteristic associate of carrion beetles, particularly burying beetles (Nicrophorus spp.). Yarrowia lipolytica has been isolated from the hindgut and anal secretions of multiple silphid species, where it is hypothesized to play roles in resource conditioning, antimicrobial competition, and lipid metabolism during carrion utilization. Given the documented associations between Yarrowia and carrion beetles, and the presence of Necrodes surinamensis in this study, we conducted a targeted analysis of Yarrowia distribution and abundance across all sample types.

Yarrowia ASV Diversity

Table 9: Summary of Yarrowia ASVs detected across all samples. Five distinct ASVs were recovered, including four assigned to Y. lipolytica and one to Y. deformans. The dominant ASV (0a434385...) accounted for 877 total reads across 49 samples.

Five distinct Yarrowia ASVs were recovered across the dataset, comprising a total of 1,002 reads distributed across 55 samples. Four ASVs were assigned to Yarrowia lipolytica, while one was identified as Yarrowia deformans. A single dominant ASV (0a434385763e63efd7fff30d6c0dfa1e) accounted for 87.5% of all Yarrowia reads (877 reads) and was detected in 49 samples, representing the most widespread Yarrowia genotype in the study. The presence of multiple co-occurring ASVs suggests either strain-level diversity within Y. lipolytica populations or sequential colonization by distinct genotypes over the course of decomposition.

Prevalence and Abundance Patterns

Figure 28: Yarrowia prevalence (left) and mean relative abundance (right) across sample types. Beetles show the highest prevalence (43.6%) and abundance (0.152%), followed by flies, carcass, and soil.

Yarrowia prevalence varied significantly across sample types, with the highest detection rate in beetle samples (43.6% of samples, 17/39) followed by flies (22.2%, 4/18), carcass surfaces (17.6%, 19/108), and soil (12.7%, 15/118). The elevated prevalence in beetles aligns with documented Yarrowia-silphid associations in the literature and suggests that carrion beetles serve as primary vectors or reservoirs for this yeast in decomposition communities. Notably, Yarrowia was detected in 22% of fly samples, indicating that necrophagous Diptera may also carry or acquire these yeasts, though at lower frequencies than beetles.

Mean relative abundance followed a similar pattern to prevalence, with beetles harboring the highest Yarrowia loads (mean 0.152% of reads, max 2.04% in a single sample), followed by flies (0.048%), soil (0.016%), and carcass surfaces (0.002%). When considering only Yarrowia-positive samples, mean abundance among positive detections was 0.348% for beetles, 0.218% for flies, 0.127% for soil, and 0.103% for carcass, indicating that when Yarrowia is present, it achieves higher relative abundance in insect-associated samples compared to environmental substrates. The low abundance on carcass surfaces despite moderate prevalence suggests that Yarrowia may be present as transient surface contamination from insect activity rather than as an actively growing carcass colonizer.

Temporal Dynamics

Figure 29: Yarrowia prevalence (left) and abundance (right) across decomposition stages in carcass and soil samples. Yarrowia was absent from Fresh stage samples in both substrates, appearing only during Active and Advanced decay stages.

Yarrowia showed striking temporal dynamics, with complete absence from Fresh stage samples (0/11 carcass, 0/4 soil). Prevalence increased during Active decay (18% for carcass, 3.6% for soil) and Advanced decay (21.3% carcass, 22.4% soil), indicating that Yarrowia colonization occurs after the onset of decomposition rather than at initial carcass placement. This pattern could reflect: (1) delayed arrival of beetle vectors that carry Yarrowia, (2) environmental conditions during Fresh stage (e.g., low pH, high antimicrobial activity from early bacterial colonizers) that inhibit Yarrowia establishment, or (3) competitive exclusion by early-arriving fungi that are later displaced. Soil samples showed lower prevalence than carcass during Active decay but converged during Advanced decay, suggesting that soil accumulation of Yarrowia may result from downward migration of yeast cells via carcass fluids or fecal deposition by insects as decomposition progresses.

Insect Body Region Differences

Figure 30: Yarrowia abundance across insect body regions. Beetle abdomens harbor significantly higher Yarrowia abundance (0.291%) compared to beetle head-thorax regions (0.019%) and whole fly specimens (0.048%).

Analysis of Yarrowia distribution across insect body regions revealed pronounced spatial heterogeneity in beetles. Beetle abdomens contained mean Yarrowia abundance of 0.291% (median 0.092%, indicating high abundance in a subset of individuals), approximately 15-fold higher than beetle head-thorax regions (0.019%, median 0%). This abdomen-biased distribution is consistent with hindgut or anal gland localization, supporting previous microbiological studies that have cultured Yarrowia from silphid beetle hindguts and exocrine secretions. Whole fly specimens showed intermediate abundance (0.048%), though the lack of body region dissection for flies precludes direct comparison with beetle gut-associated loads. The strong abdomen signal in beetles, combined with temporal patterns showing Yarrowia appearance coinciding with beetle arrival during decomposition, suggests that N. surinamensis serves as the primary inoculum source for Yarrowia in this carrion system.

Summary

Community Structure and Diversity: Fungal communities exhibit strong habitat specialization, with PERMANOVA analyses confirming significant separation between sample types (p < 0.001, R² = 0.22). Gravesoil harbors the highest alpha diversity (mean 94.4 observed ASVs, Shannon 3.52), reflecting the heterogeneous fungal reservoir beneath decomposing carcasses. Carcass surfaces show intermediate diversity (59.5 ASVs, Shannon 2.60), while necrophagous insects maintain lower, more specialized mycobiomes: flies (47.7 ASVs, Shannon 2.35) and beetles (44.6 ASVs, Shannon 2.44). The dramatically reduced diversity on fly specimens, dominated by Cladosporium, suggests highly selective filtering or competitive exclusion, whereas soil’s elevated diversity likely reflects diverse saprotrophic and environmental taxa accumulating in this stable substrate.

Temporal Dynamics: Decomposition stage drives significant shifts in fungal community composition (PERMANOVA p < 0.001), with contrasting diversity trajectories between substrates. Carcass fungal diversity increases from Fresh (41.6 ASVs) to Active decay (71.6 ASVs), then declines during Advanced decay (50.4 ASVs), peaking at mid-decomposition when resource availability and moisture are optimal. In contrast, soil diversity shows a steady decline from Fresh (133.5 ASVs) to Active (104.7) to Advanced stages (81.5), possibly reflecting initial contamination from carcass fluids followed by competitive exclusion or resource depletion. Network analyses reveal dramatic temporal restructuring in carcass samples: Fresh stage showed no detectable fungal associations (0 edges), Active decay exhibited peak connectivity (274 edges), followed by network simplification during Advanced decay (46 edges, with only 10 edges shared with Active stage). This suggests that mid-decomposition represents a window of maximum fungal diversity and interaction complexity on carcass surfaces, while soil communities become progressively simplified.

Insect-Fungal Associations: Carrion beetle samples showed detectable presence of stinkhorn fungi (Phallus, Clathrus), which are known to attract insects for spore dispersal in other systems. Differential prevalence analyses identified taxa that consistently occur in beetle samples across individuals, though the ecological significance of these associations remains unclear. Beetle head-thorax and abdomen fractions harbor distinguishable fungal communities based on alpha and beta diversity patterns, indicating spatial heterogeneity in fungal carriage across body regions. Network analyses for beetle samples showed moderate connectivity (37 edges, 61 nodes) with higher transitivity than fly networks, reflecting more clustered co-occurrence patterns among beetle-associated fungi.

Treatment Effects: Insect diversity treatments (Tr0-Tr3) showed minimal effects on fungal community composition in both carcass and soil samples, with no significant differences detected by PERMANOVA or alpha diversity comparisons. This suggests that broad-scale fungal community structure is more strongly driven by substrate type and decomposition stage than by the diversity of necrophagous insect communities present, though subtle taxon-specific effects may exist.

Predictive Modeling: Random Forest classifiers achieved high accuracy for sample type prediction (87.6% cross-validated), with key predictive genera including substrate-specific indicators. Stage classification proved more challenging (72.8% accuracy), likely reflecting gradual successional transitions and within-stage heterogeneity. Per-class accuracy revealed that Fresh stage samples were most difficult to classify, consistent with the low fungal diversity and network connectivity observed at this stage. These models demonstrate that fungal community composition contains strong predictive signals for both habitat and decomposition state.

Network Architecture: Co-occurrence network topology varied dramatically across sample types. Soil networks showed the highest connectivity (281 edges across 171 nodes), reflecting complex multi-taxon interactions in this heterogeneous habitat. Carcass networks exhibited intermediate complexity (129 edges, 113 nodes), while insect networks were sparse, particularly for flies (16 edges across 78 nodes). Edge overlap analyses between decomposition stages revealed minimal conservation of fungal associations over time (only 10 of 274 Active stage edges persisted into Advanced stage for carcass samples), indicating rapid turnover in co-occurrence patterns as decomposition progresses.

Yarrowia-Beetle Associations: Targeted analysis of Yarrowia yeasts revealed a clear pattern of insect-mediated dispersal and maintenance in carrion communities. Five Yarrowia ASVs (primarily Y. lipolytica) were detected across 55 samples, with the highest prevalence in beetles (43.6%) compared to flies (22.2%), carcass (17.6%), and soil (12.7%). Beetle samples also harbored the highest Yarrowia abundance (mean 0.152%, max 2.04%), with pronounced spatial heterogeneity: beetle abdomens showed 15-fold higher abundance than head-thorax regions, consistent with hindgut or anal gland localization documented in other silphid species. Temporal patterns revealed complete absence of Yarrowia from Fresh stage samples, with colonization occurring only during Active and Advanced decay, suggesting that N. surinamensis vectors this yeast to carcasses after initial placement. The low abundance on carcass surfaces despite moderate prevalence, combined with abdomen-localized distribution in beetles, supports a model where Yarrowia is primarily an insect-associated yeast rather than a saprophytic carcass colonizer. Detection in fly and soil samples likely reflects secondary dispersal from beetle sources through direct contact or environmental contamination. The documented lipase activity of Y. lipolytica may provide functional benefits for beetles processing lipid-rich tissues, while the yeast’s antimicrobial properties could contribute to beetle-mediated suppression of competitors.

Note on Clathrus/Phallus in N. surinamensis

The observed enrichment of stinkhorn fungi (Phallus, Clathrus) in carrion beetle mycobiomes may reflect several ecological mechanisms. Necrodes surinamensis adults and larvae spend extended periods moving through carcass surfaces and soil-litter matrices where stinkhorn fruiting bodies may be present, providing opportunities for repeated spore acquisition and retention on beetle cuticles. Beetles are known to engineer their carrion habitat through antimicrobial secretions and thermogenic biofilms, creating selective conditions that may favor macrofungal spores tolerant of elevated temperatures, variable oxygen, and antimicrobial compounds while reducing generalist mold abundance. Additionally, the robust spores characteristic of macroscopic basidiomycetes can strongly adhere to beetle integument and potentially survive gut passage, leading to higher retention times compared to airborne molds. This combination of prolonged substrate contact, habitat engineering, and differential spore retention could explain the overrepresentation of stinkhorn genera in beetle-associated fungal communities, though experimental validation of these mechanisms remains needed.

Note on Cladosporium in C. macellaria

The dominance of Cladosporium in blow fly mycobiomes likely reflects fundamentally different fungal acquisition and filtering processes compared to beetles. Adult necrophagous flies repeatedly land on exposed surfaces and are constantly exposed to airborne spores, with Cladosporium being among the most ubiquitous fungi in aerial spore rain and on plant surfaces. Their contact with carcasses may be briefer and less intimate than beetles walking through biofilms, resulting in fungal loads that more closely reflect general atmospheric fungal composition. Blow flies maintain host-specific bacterial microbiomes despite variable environmental exposures, suggesting strong host filtering that likely extends to fungi and favors taxa capable of tolerating gut passage and the fly’s immune environment. Cladosporium species are known gut-passage survivors and common surface contaminants in insects, making them likely to dominate a “background” mycobiome even when other spores are encountered. Furthermore, the rapid larval development and brief exposure window in fly life cycles favor fast-growing opportunists rather than stable associations with specialized macrofungal taxa. This combination of atmospheric acquisition, host filtering, and rapid development may explain the simpler, Cladosporium-dominated mycobiomes observed in C. macellaria compared to the more diverse, macrofungi-enriched communities in carrion beetles.

Note on Claviceps in C. macellaria

The detection of Claviceps (ergot fungi) in blow fly samples may reflect the known role of Diptera as vectors for ergot spores in agricultural and grassland systems. During grass infections, Claviceps produces a honeydew stage—a sugary exudate rich in asexual conidia that attracts a wide range of insects including flies, beetles, and wasps. More than 40 insect species have been documented visiting ergot honeydew on grasses, where they acquire conidia both externally on their bodies and internally through feeding. Targeted field studies using PCR detection found Claviceps spores in 36-39% of dipteran insects sampled across consecutive years, demonstrating that flies frequently carry ergot conidia after honeydew visitation. While early work noted uncertainty about the relative importance of insect versus wind dispersal for ergot spread, recent evidence indicates that Diptera and Lepidoptera can be effective dispersers of asexual Claviceps spores at field scale, complementing wind dispersal. The presence of Claviceps in carrion-associated blow flies likely reflects incidental acquisition from the broader environment during adult foraging or oviposition flights, though the quantitative contribution of necrophagous flies to ergot epidemiology and the ecological significance of this association remain areas for further investigation.

Note on Rhodotorula in N. surinamensis

The presence of Rhodotorula in carrion beetle samples aligns with the ecology of these ubiquitous basidiomycetous yeasts as stress-tolerant environmental colonizers. Rhodotorula species are cultured from diverse substrates including soil, water, air, plant material, and various foods, and are notable for carotenoid production, tolerance of low-nutrient conditions, and ability to persist under environmental stress. These yeasts readily colonize animal guts and surfaces when encountered through feeding or environmental contact; in vertebrates, intestinal colonization is considered common and typically commensal, indicating their capacity to persist in animal digestive tracts. Their physiological traits—including stress tolerance, utilization of diverse carbon sources, and biofilm formation—make them well-suited to patchy, fluctuating substrates like decomposing organic matter. Carrion beetles inhabit carcasses and associated soils where Rhodotorula is likely present as airborne fallout and on plant/soil particles, with frequent grooming and feeding on contaminated tissues providing ample acquisition opportunities. Silphid beetles maintain distinct microbiomes linked to their carrion lifestyle, with documented associations with specific yeasts such as Yarrowia in other burying beetle species, suggesting that yeasts tolerant of beetle secretions and carcass conditions can establish as stable associates. Necrodes surinamensis produces defensive antimicrobial secretions including necrodane monoterpenes that repel competitors and likely shape its microbiota; yeasts like Rhodotorula capable of tolerating chemical stress may be disproportionately retained on cuticles and in guts compared to more sensitive fungi. Without species-level resolution and functional experiments, Rhodotorula in N. surinamensis is most parsimoniously interpreted as a common environmental and enteric yeast that beetles can repeatedly acquire and maintain, rather than as a specialized obligate symbiont, though its consistent presence suggests ecological compatibility with the beetle-carrion system.

References

  1. Kurtz, Z. D., Müller, C. L., Miraldi, E. R., Littman, D. R., Blaser, M. J., & Bonneau, R. A. (2015). Sparse and Compositionally Robust Inference of Microbial Ecological Networks. PLOS Computational Biology, 11(5), e1004226. https://doi.org/10.1371/journal.pcbi.1004226

  2. Liaw, A., & Wiener, M. (2002). Classification and Regression by randomForest. R News, 2(3), 18–22. https://journal.r-project.org/articles/RN-2002-022/RN-2002-022.pdf

  3. Mallick, H., Rahnavard, A., McIver, L. J., Ma, S., Zhang, Y., Nguyen, L. H., Tickle, T. L., Weingart, G., Ren, B., Schwager, E. H., Chatterjee, S., Thompson, K. N., Wilkinson, J. E., Subramanian, A., Lu, Y., Waldron, L., Paulson, J. N., Franzosa, E. A., Bravo, H. C., & Huttenhower, C. (2021). Multivariable association discovery in population-scale meta-omics studies. PLOS Computational Biology, 17(11), e1009442. https://doi.org/10.1371/journal.pcbi.1009442

  4. Bangay, R., Matsuoka, S., & Tuno, N. (2026). Comparative Analysis of Fungal Spore Flora Among Birds, Insects and Air in a Temperate Japanese Forest. Ecology and Evolution, 16(1). https://doi.org/10.1002/ece3.72929

  5. Brinkrolf, K., Shukla, S. P., Griep, S., Rupp, O., Heise, P., Goesmann, A., Heckel, D. G., Vogel, H., & Vilcinskas, A. (2021). Genomic analysis of novel Yarrowia-like yeast symbionts associated with the carrion-feeding burying beetle Nicrophorus vespilloides. BMC Genomics, 22(1). https://doi.org/10.1186/s12864-021-07597-z

  6. Eisner, T., Deyrup, M., Jacobs, R., & Meinwald, J. (1986). Necrodols: Anti-insectan terpenes from defensive secretion of carrion beetle (Necrodes surinamensis). Journal of Chemical Ecology, 12(6), 1407–1415. https://doi.org/10.1007/bf01012360

  7. Fu, C., Shah, A. A., Alissa, M., Alsuwat, M. A., Ullah, S., Khan, M. S., & Zaman, A. (2024). Exploring the potential of insect gut microbes for advancing renewable energy production. Journal of Applied Entomology. https://doi.org/10.1111/jen.13311

  8. Hof, H. (2019). Rhodotorula spp. in the gut – foe or friend? GMS Infectious Diseases, 7. https://doi.org/10.3205/id000042

  9. Kaltenpoth, M., & Steiger, S. (2013). Unearthing carrion beetles’ microbiome: characterization of bacterial and fungal hindgut communities across the Silphidae. Molecular Ecology, 23(6), 1251–1267. https://doi.org/10.1111/mec.12469

  10. Kaur, N., Cating, R. A., Rondon, S. I., Scott, J. C., Alderman, S. C., Walenta, D. L., Frost, K. E., Hamm, P. B., & Dung, J. K. S. (2019). Dispersal Potential of Ergot Spores by Insects Foraging in the Perennial Ryegrass Fields in the Columbia Basin of Oregon and Washington. Crop, Forage & Turfgrass Management, 5(1), 190020. https://doi.org/10.2134/cftm2019.04.0020

  11. Lunde, L. F., Boddy, L., Sverdrup-Thygeson, A., Jacobsen, R. M., Kauserud, H., & Birkemoe, T. (2023). Beetles provide directed dispersal of viable spores of a keystone wood decay fungus. Fungal Ecology, 63, 101232. https://doi.org/10.1016/j.funeco.2023.101232

  12. Matuszewski, S., & Mądra-Bielewicz, A. (2021). Heat production in a feeding matrix formed on carrion by communally breeding beetles. Frontiers in Zoology, 18(1). https://doi.org/10.1186/s12983-020-00385-7

  13. Pratt, C. J., Meili, C. H., Youssef, N. H., & W Wyatt Hoback. (2023). Culture-independent analyses of carrion beetle (Coleoptera: Silphidae) secretion bacterial communities. Microbiology Spectrum, 11(6). https://doi.org/10.1128/spectrum.01694-23

  14. Shukla, S. P., Plata, C., Reichelt, M., Steiger, S., Heckel, D. G., Kaltenpoth, M., Vilcinskas, A., & Vogel, H. (2018). Microbiome-assisted carrion preservation aids larval development in a burying beetle. Proceedings of the National Academy of Sciences, 115(44), 11274–11279. https://doi.org/10.1073/pnas.1812808115

  15. Wirth, F., & Goldani, L. Z. (2012). Epidemiology of Rhodotorula: An Emerging Pathogen. Interdisciplinary Perspectives on Infectious Diseases, 2012, 1–7. https://doi.org/10.1155/2012/465717