Data from: Distinct diversity Trajectories of Boreal wood-inhabiting fungi following fire vs. clear-cutting
Data files
Feb 26, 2026 version files 831.12 KB
-
README.md
16.74 KB
-
Script_1_-_Laser_scanning_indices.R
4.34 KB
-
Script_2_-_Diversity_trajectories.R
37.90 KB
-
Script_3_-_Community_analysis.R
18.41 KB
-
Script_4_-_Random_forest_analysis.R
19.78 KB
-
Script_5_-_Threshold_analysis.R
23.57 KB
-
Stand_Level_Data_Agaricomycetes.xlsx
330.17 KB
-
Stand_Level_Data.xlsx
380.23 KB
Abstract
Here, we describe the data and methods supporting a study on how disturbances such as forest fires and clear-cutting influence wood-inhabiting fungal diversity trajectories in boreal Scots pine (Pinus sylvestris) forests in Northern Sweden. The research contrasts two chronosequences, one under rotational management (clear-cutting, soil scarification, and thinning) and one unmanaged and fire-origin.
Two contrasting disturbance regimes – wildfire and clear-cutting – are common in boreal forests, and create fundamentally different conditions for succession of wood-inhabiting fungi. We investigated (i) how species richness and community composition change after these two disturbances, and (ii) which stand-level characteristics drive diversity trajectories.
In two chronosequences – managed clear-cut (1–109yr since clear-cut; n=18) and unmanaged fire (4–375yr since fire; n=18) – we combined fruiting-body surveys with DNA metabarcoding to quantify species richness of wood-inhabiting fungi, including total number of species, Agaricomycete and red-listed species. To identify drivers, we measured deadwood attributes and forest structural complexity using terrestrial laser scanning.
Species richness, including red-listed species, was highest in unmanaged fire stands. Unmanaged fire stands had ~55 more total species than managed clear-cut stands at comparable time since disturbance (≤109yr), and ~156 more species in stands >109yr. Communities differed markedly between chronosequence types. Managed clear-cut stands harbored a subset of species found in unmanaged fire stands, and nearly all red-listed and indicator species were exclusive to unmanaged fire stands. Total and Agaricomycete species richness increased with time in both chronosequences without saturating. Red-listed species richness remained low and did not increase with time in managed clear-cut stands, but was higher and increased with time in unmanaged fire stands. Conditional random forest models identified spruce deadwood percentage, deadwood volume, and forest structural complexity as dominant diversity drivers, with deadwood quality replacing forest structure as the best predictor for red-listed species. Species richness rose steadily with deadwood volume, leveling at ~50m³ha⁻¹ for total species and >100m³ha⁻¹ for red-listed species.
Synthesis: Clear-cutting altered fungal recovery trajectories differently from fire. While fires leave standing and fallen dead trees that host fungal communities for centuries, clear-cutting removes these legacies and simplifies forest structure, resulting in a lack of recovery of red-listed species. These contrasting disturbance pathways shape boreal fungal communities through their effects on deadwood and structural continuity. Retaining high deadwood volumes and structural complexity can help maintain fungal diversity in managed forests, however maintaining old-growth stands is essential for conserving highly diverse communities and red-listed species.
Dataset DOI: 10.5061/dryad.tmpg4f5bs
Description of the data and file structure
This dataset supports the research article "Distinct Diversity Trajectories of Boreal Wood-Inhabiting Fungi Following Fire vs. Clear-Cutting" submitted to Journal of Ecology. The study examines how two contrasting disturbance regimes - natural fire and rotational clear-cutting - create fundamentally different conditions for successional development of wood-inhabiting fungal communities in boreal Scots pine (Pinus sylvestris) forests in Northern Sweden (approximately 64.8-66.5 degrees N and 17.1-20.5 degrees E).
The dataset contains comprehensive stand-level measurements from 36 forest stands representing different stages of forest development following disturbance:
18 rotational management stands (1-109 years since clear-cut) subjected to typical even-aged forestry practices including clear-cutting, soil scarification, planting, and 2-3 thinning operations
18 unmanaged fire stands (4-375 years since fire) with varying fire severity from non-stand-replacing ground fires to stand-replacing wildfire.
Files and variables
FILES INCLUDED
Stand_Level_Data.xlsx - Complete dataset with all stand-level variables and fungal species occurrences (37 rows including header x 2631 columns)
Stand_Level_Data_Agaricomycetes.xlsx - Subplot-level dataset containing Agaricomycete fungi only (180 rows including header x 550 columns). Each row represents one subplot (5 subplots per stand, 36 stands; one subplot excluded due to insufficient sequencing depth). Metadata columns differ from Stand_Level_Data.xlsx as described below.
Script_1_-_Laser_scanning_indices.R - Processes terrestrial laser scanning (.laz) point clouds to calculate forest structural complexity metrics
Script_2_-_Diversity_trajectories.R - Analyzes fungal species richness trajectories using negative binomial GAMs, compares detection methods, and performs post-hoc group comparisons
Script_3_-_Community_analysis.R - Performs NMDS ordination of fungal communities, indicator species analysis, and environmental vector fitting
Script_4_-_Random_forest_analysis.R - Identifies key habitat drivers of fungal diversity using conditional random forests with cross-validation
Script_5_-_Threshold_analysis.R - Detects ecological thresholds in fungal richness responses to habitat gradients using GAMs with bootstrap confidence intervals
====================
VARIABLE DEFINITIONS FOR Stand_Level_Data.xlsx
Empty cells indicate true absence of measurement or non-applicability (e.g. variable not relevant for a given stand), not zero values.
STAND CHARACTERISTICS (Columns A-C)
- Stand: Stand identifier (format: M[number][letter] for managed, F[number] for fire)
- Management: Management type
- M = Managed (rotational forestry with clear-cutting)
- F = Fire (unmanaged, fire-origin)
- Time: Years since disturbance event (clear-cut or fire)
FUNGAL DIVERSITY METRICS (Columns D-H)
- TotalSpecies: Total number of wood-inhabiting fungal species (DNA + fruiting-body combined)
- TotalAgaricomycetes: Total number of Agaricomycete species (DNA + fruiting-body combined)
- SpeciesDNA: Species detected through DNA metabarcoding
- SpeciesSporocarp: Species detected through fruiting-body surveys
- RedListedSpeciesDNA: Red-listed species detected through DNA metabarcoding
- RedListedSpeciesSporocarp: Red-listed species detected through fruiting-body surveys
- RedListedSpecies: Total red-listed species according to Swedish Red List (Artdatabanken, 2020)
STAND AND SAMPLING VARIABLES (Columns I-J)
- StandSize: Total stand area (hectares)
- SeqDepth: DNA sequencing depth (number of high-quality reads after filtering)
FOREST STRUCTURE METRICS (Columns K-P)
- Derived from terrestrial laser scanning using handheld MLS (ZEB Horizon, GeoSLAM Ltd, UK):
- CanopyRugosity: Surface roughness of the canopy calculated at 2m grid resolution (dimensionless, ranges ~0.4-3.7)
- TopRugosity: Surface roughness of tree tops calculated at 2m grid resolution (dimensionless, ranges ~1.1-5.4)
- DeepGapFraction: Proportion of 2m grid cells classified as deep canopy gaps, defined as cells where canopy point density falls below max(1, 5% of mean cell density) above 2m height (dimensionless, ranges ~0.03-0.70)
- CoverFraction: Proportion of 2m grid cells with canopy cover, defined as cells with >=30 return points above 2m height (dimensionless, ranges ~0.65-1.00)
- VerticalGini: Gini coefficient of vertical vegetation distribution (ranges ~0.54-0.96, higher values indicate more uneven distribution)
- BoxDimension: Fractal dimension of the three-dimensional point cloud estimated via box-counting voxelization (dimensionless, ranges ~2.10-2.18).
DEADWOOD CHARACTERISTICS (Columns Q-AA)
All measured within 17.8m radius plots (1,000 m2 area, 5 subplots per stand):
- TotalDeadwoodVolume: Total deadwood volume calculated as truncated cones for logs and standing deadwood (m3/ha). Equivalent to DeadwoodQuantity; retained as the variable name referenced in Script 4.
- ShareDeciduousDeadwood: Percentage of deadwood volume from deciduous species (Betula pendula, Populus tremula, Sorbus aucuparia, Salix spp.) (%)
- ShareSpruceDeadwood: Percentage of deadwood volume from Norway spruce (Picea abies) (%)
- MeanDecayClass: Mean decay class following Swedish National Forest Inventory protocol (1-5 scale; 1=fresh wood, 5=highly decomposed)
- MeanDiameter: Mean diameter of deadwood items measured at both ends (cm)
- LateDecay: Proportion of deadwood items in decay classes 3-5 (%)
- ProportionLargeLogs: Proportion of logs with diameter >=30cm (%)
- ProportionStanding: Proportion of standing deadwood (snags and stumps) versus lying logs (%)
- DeadwoodDiversity: Shannon diversity index calculated from deadwood categories (tree species x size class x decay stage)
- DeadwoodQuantity: Total deadwood volume calculated as truncated cones for both logs and standing deadwood (m3/ha)
- DeadwoodQuality: First principal component (PC1) from PCA of MeanDiameter, ProportionLargeLogs, and LateDecay (explains 63.4% variance; higher values indicate larger, more decayed logs)
FUNGAL SPECIES OCCURRENCE DATA (Columns AB onwards)
- Column headers: Species names following Swedish Taxonomic Database (Artdatabanken, 2025) nomenclature
- Values for DNA-detected species: Number of sequence reads after quality filtering and rarefaction
- Values for fruiting-body detected species: Presence (>0) or absence (0)
- Species naming format: Genus_species_scata[number] for DNA sequences, Genus_species for fruiting bodies
SUMMARY COLUMNS (after species occurrence data)
- SpeciesDNA: Total species detected through DNA metabarcoding
- SpeciesSporocarp: Total species detected through fruiting-body surveys
====================
VARIABLE DEFINITIONS FOR Stand_Level_Data_Agaricomycetes.xlsx
This file is organized at the subplot level. Variable definitions are the same as for Stand_Level_Data.xlsx except for the following:
- Stand: Stand identifier (same as above)
- Subplot: Subplot identifier within each stand (A-E)
- Management: Management type (same as above)
- Time: Years since disturbance event (same as above)
- TotalSpecies: Total number of wood-inhabiting fungal species per subplot
- TotalAgaricomycetes: Total number of Agaricomycete species per subplot
- RedListedSpeciesDNA: Red-listed species detected through DNA metabarcoding per subplot
- RedListedSpeciesSporocarp: Red-listed species detected through fruiting-body surveys per subplot
- RedListedSpecies: Total red-listed species per subplot
- SeqDepth_subplot: DNA sequencing depth per subplot (number of high-quality reads after filtering)
- Forest structure metrics (CanopyRugosity, TopRugosity, DeepGapFraction, CoverFraction, VerticalGini, BoxDimension) are at the stand level and repeated across subplots within a stand.
- Species occurrence columns contain Agaricomycete fungi only, with the same naming conventions as Stand_Level_Data.xlsx.
- DeadwoodDiversity, DeadwoodQuantity, DeadwoodQuality, SpeciesDNA, and SpeciesSporocarp appear after the species occurrence columns.
====================
DETAILED METHODS
FIELD SAMPLING DESIGN
Sampling period: September 2023
Stand selection: Based on established chronosequence from Buness et al. (2025)
Plot configuration: 5 subplots per stand arranged as 1 central subplot plus 4 subplots at 35m distance in intercardinal directions (NE, SE, SW, NW)
Deadwood assessment area: 17.8m radius circular plots (1,000 m2 per subplot, 5,000 m2 total per stand)
DNA sampling area: 10m radius circular plots (314 m2 per subplot) nested within deadwood plots
Deadwood inclusion criteria: Minimum 10cm diameter at larger end and minimum 1m length
Deadwood volume calculation: Truncated cone formula for all deadwood pieces: V = (pi x L)/12(D12 + D1D2 + D22)
DNA METABARCODING PROTOCOL
Sample collection: Wood cores drilled from deadwood using 30cm increment drill (8mm diameter) along 8 transects per subplot
Drilling pattern: Three samples per log (0.5m left, center, 0.5m right of transect); for snags/stumps at base, 0.5m, and 1m height
Sample processing: Pooled by subplot (180 composite samples total), freeze-dried, ball-milled
DNA extraction: NucleoSpin Soil kit with SL1 buffer + enhancer, 200mg material
PCR amplification: ITS2-LSU region using primers gITS7 and TW13
Sequencing: PacBio Revio Platform (25M SMRT cell) at SciLifeLab NGI Uppsala
Bioinformatics: SCATA pipeline with 99.25% similarity clustering threshold
Quality filtering: Mean Q >= 20, removal of chimeras and singletons using MUMU
Rarefaction: 2,000 reads per sample for community analyses
Final dataset: 1,800 fungal species hypotheses after quality control
FRUITING-BODY SURVEY PROTOCOL
Target taxa: All polypore species plus tooth fungi genera Odonticium and Asterodon
Survey effort: ~30 minutes per subplot by two experienced surveyors
Identification: Field identification when possible, microscopic verification using Melzer's reagent, 3-10% KOH, iron sulfate, and cotton blue at 400-1000x magnification
References: Ryvarden et al. (2017), Christensen & Heilmann-Clausen (2013), Laessoe & Petersen (2019), Larsson & Ryvarden (2021)
External validation: ~20 specimens verified by taxonomic experts
Total detected: 60 species by fruiting-body surveys, 19 species detected by both methods
TERRESTRIAL LASER SCANNING
Equipment: Handheld mobile laser scanner (MLS, ZEB Horizon, GeoSLAM Ltd, UK)
Scanning pattern: Circular paths at 12m, 24m, and 35m radii from stand center
Point cloud processing: 1cm voxelization, 5cm DTM resolution, height normalization
Structural metrics: Calculated using forestr R package following Atkins et al. (2018)
Original data: Raw .laz point cloud files available at https://doi.org/10.5061/dryad.2547d7x1f. Raw .laz point cloud files can be viewed using free software such as CloudCompare or LAStools (lasview).
Processing infrastructure: National Academic Infrastructure for Supercomputing in Sweden (NAISS)
====================
Code/software
STATISTICAL ANALYSES OVERVIEW
SCRIPT WORKFLOW
Script 1: Processes raw laser scanning data (.laz files) to derive structural complexity metrics. Requires substantial computational resources (NAISS infrastructure recommended)
Script 2: Models species richness trajectories using GAMs, compares management types and successional stages, analyzes detection method performance
Script 3: Analyzes community composition patterns using NMDS ordination and identifies environmental drivers through vector fitting
Script 4: Identifies key habitat predictors using conditional random forests with 50 x 10-fold cross-validation
Script 5: Determines ecological thresholds for management targets using GAM derivatives with bootstrap confidence intervals
KEY STATISTICAL APPROACHES
GAMs: Negative binomial family with sequencing depth offset for modeling richness trajectories
NMDS: Jaccard dissimilarity for community ordination (stress = 0.194)
Conditional Random Forests: 500 trees per model, variable importance assessed through permutation
Indicator Species Analysis: IndVal.g function with 999 permutations for six successional classes
Threshold Detection: First derivatives of GAM smooths with parametric bootstrap (200 iterations) for uncertainty
====================
SOFTWARE REQUIREMENTS
R: Version 4.3.0 or higher
Essential R packages:
Core: tidyverse (v2.0.0+), readxl, writexl
Statistical modeling: mgcv (GAMs), vegan (ordination), party (random forests), multcomp (post-hoc tests)
Visualization: ggplot2, patchwork, gridExtra, cowplot, ggsignif
LiDAR processing: lidR, raster, data.table, stringr
Additional: indicspecies (indicator analysis), broom (model tidying), MuMIn (model selection)
====================
USAGE INSTRUCTIONS
File paths: Update all paths marked with "/.../" in scripts to match your directory structure
Execution order: Run scripts in numerical sequence (1-5) for complete reproduction
Script 1 requirements: Access to high-performance computing infrastructure for processing .laz files
Scripts 2-5: Can be run on standard desktop computers
Output: Scripts generate figures and supplementary tables as described in the manuscript
====================
DATA COMPLEMENTARITY
The DNA metabarcoding and fruiting-body survey methods are complementary:
DNA metabarcoding detected 1,800 species (primarily capturing cryptic diversity)
Fruiting-body surveys detected 60 species (excelling at red-listed species detection)
Only 19 species were detected by both methods, highlighting the importance of dual approaches
====================
REFERENCES
Artdatabanken (2020) Rodlista 2020: Svampar (Fungi). SLU Artdatabanken. https://www.slu.se/globalassets/ew/subw/artd/6-publikationer/31.-rodlista-2020/rodlista-2020.pdf
Artdatabanken (2025) Swedish Taxonomic Database. Swedish University of Agricultural Sciences. https://www.artdatabanken.se
Atkins JW, Bohrer G, Fahey RT, Hardiman BS, Morin TH, Stovall AEL, Zimmerman N, Gough CM (2018) Quantifying vegetation and canopy structural complexity from terrestrial LiDAR data using the forestr r package. Methods in Ecology and Evolution 9:2057-2066
Buness V, Sundqvist MK, Ali ST, Annighofer P, Aragon CM, Lanzrein I, Metcalfe DB, Nilsson M-C, Gundale MJ (2025) Resource quantity and heterogeneity drive successional plant diversity in managed and unmanaged boreal forests. Ecography e07676
Christensen M, Heilmann-Clausen J (2013) The genus Tricholoma. Danish Mycological Society
Laessoe T, Petersen JH (2019) Fungi of temperate Europe. Princeton University Press
Larsson K-H, Ryvarden L (2021) Corticioid fungi of Europe. Volume 1: Acanthobasidium - Gyrodontium. FUNGIFLORA
Ryvarden L, Melo I, Niemela T (2017) Poroid fungi of Europe (Second revised edition). Fungiflora
Swedish University of Agricultural Sciences (2023) Faltinstruktion 2023: Riksinventeringen av skog. Institutionen for skoglig resurshushallning, SLU
====================
FUNDING
This study was supported by a grant from FORMAS (Swedish Research Council for Sustainable Development; Project Formas grant no. 2021-02116) awarded to MJG.
====================
CONTACT INFORMATION
Corresponding authors:
Vincent Buness (vincent.buness@slu.se)
Michael J. Gundale (michael.gundale@slu.se)
====================
LICENSE
This dataset is made available under CC0 1.0 Universal (CC0 1.0) Public Domain Dedication.
====================
ACKNOWLEDGMENTS
We thank Sveaskog AB for providing access to forest stands and their database. We extend our gratitude to Länsstyrelsen for allowing us to conduct research in protected areas. We acknowledge Morgan Karlsson, Per Schönning, and Carlos Miguel Aragon for their assistance with fieldwork, and Ilse van Duuren and Yasaman Najafi for laboratory work. The Centre for Statistics (SLU), particularly Magnus Ekström, provided valuable statistical advice. We also thank Mikael Brandström Durling for establishing and maintaining the bioinformatics pipeline. The data handling and processing was enabled by resources provided by the National Academic Infrastructure for Supercomputing in Sweden (NAISS), partially funded by the Swedish Research Council through grant agreement no. 2022-06725. The authors declare no conflicts of interest.
