Journal list menu

Volume 56, Issue 6 p. 1367-1378
Free Access

The significance of retention trees for survival of ectomycorrhizal fungi in clear-cut Scots pine forests

Erica Sterkenburg

Erica Sterkenburg

Department of Forest Mycology and Plant Pathology, Swedish University of Agricultural Sciences, Uppsala, Sweden

Search for more papers by this author
Karina E. Clemmensen

Karina E. Clemmensen

Department of Forest Mycology and Plant Pathology, Swedish University of Agricultural Sciences, Uppsala, Sweden

Search for more papers by this author
Björn D. Lindahl

Björn D. Lindahl

Department of Soil and Environment, Swedish University of Agricultural Sciences, Uppsala, Sweden

Search for more papers by this author
Anders Dahlberg

Corresponding Author

Anders Dahlberg

Department of Forest Mycology and Plant Pathology, Swedish University of Agricultural Sciences, Uppsala, Sweden


Anders Dahlberg

Email: [email protected]

Search for more papers by this author
First published: 15 February 2019
Citations: 37


  1. Forestry with short stand generations and simplified forest structures has markedly affected forest biodiversity. One group of organisms adversely affected by clear-cutting is ectomycorrhizal (ECM) fungi, as they are associated with the roots of living trees. Retention forestry is a way of reducing logging impacts and enhancing biodiversity conservation. Increasing the proportion of trees retained at harvest may improve ECM fungal diversity.
  2. We investigated the potential for lifeboating of ECM fungi through the harvesting phase in an experimental field study in a 190-year-old Scots pine forest in northern Sweden. The experiment comprised four levels of tree retention—unlogged forest, plots with 60% or 30% of evenly distributed trees retained and clear-cuts without retained trees. We sampled soils and determined identities, frequencies and relative abundances of ECM fungal species during 3 years following logging through the use of high-throughput sequencing of amplified ITS2 markers.
  3. We identified 149 ECM fungal species, with the five most abundant species accounting for 50% of the total ECM fungal amplicons. Three years after harvesting, the proportion of ECM sequences in the total amplicon pool had decreased proportionally to the extent of tree removal. In clear-cuts, ECM fungal relative abundance had decreased by 95%, while ECM fungal species richness had declined by 75%, compared to unlogged plots.
  4. Tree retention enabled the maintenance of the most frequent ECM species, while more lowly abundant species were progressively lost at random with increasing level of tree removal. Five of the most frequent ECM fungal species remained present after clear-cutting, probably associated with pine seedlings.
  5. Synthesis and applications. Tree retention can moderate short-term and potentially also long-term logging impacts on ECM fungi. Local ECM fungal diversity is preserved in proportion to the amount of retained trees. Abundant species may be largely maintained, even by low levels of tree retention and on naturally established seedlings. However, conservation of more infrequent species requires higher levels of tree retention, and our results suggest that around 75% of the ECM species are lost with the forest certification standard of 5% retention trees left at logging.


Forests constitute a major part of terrestrial ecosystems in the world and contain a large part of the global biodiversity (Perry, Oren, & Hart, 2008). Industrialized forestry has resulted in simplified forest structures and even-aged stands with short rotation times, having adverse impacts on biodiversity and ecosystem functions (Butchart et al., 2010). In Fennoscandia, where the conservation status of animals, fungi and plants has been assessed since 1980s, clear-cutting is the main reason why forest species are included in national Red-Lists (Tingstad et al., 2018).

Aiming to mitigate negative impacts of forestry on biodiversity, retention forestry was initiated in the early 1990s (Lindenmayer & Franklin, 2002). At this point, conservation actions were integrated into silvicultural methods, as it was obvious that protection of old forests in reserves would not be sufficient to maintain forest biodiversity at the national level. Retention forestry moderates negative harvesting impacts on biodiversity, for example, by leaving single trees, tree groups, buffering tree zones bordering lakes and wetlands as well as by leaving and creating dead wood (Fedrowitz et al., 2014). These actions are primarily associated with clear-cutting with the objective of ‘lifeboating’ species through the regeneration phase, increasing habitat diversity and enhancing connectivity in the forest landscape. Retention forestry is increasingly being practiced in temperate and boreal forests in North America, Australia and northern Europe. The proportion of trees retained varies greatly among regions and forest ownership and may range from a few per cent to about 30%. In Fennoscandia, commonly 5%–10% of the trees are retained (Gustafsson et al., 2012). There is also an increasing interest to retain much larger proportions of trees, using continuous cover forestry methods to conserve biodiversity and promote a broader variety of ecosystem services in boreal production forests (Peura, Burgas, Eyvindson, Repo, & Mönkkönen, 2018).

The groups of organisms that benefit most from tree retention are ectomycorrhizal (ECM) fungi, birds and lichens (Rosenvald & Lohmus, 2008). ECM fungi represent a large part of the biodiversity in boreal forests. They depend on carbohydrates from their host trees and are vital for forest production, as uptake of nutrients and water by the trees is mediated by the ECM symbiosis. ECM fungal mycelium forms a basis for soil food webs. The largely cryptic life of ECM fungi has hampered understanding of their biology and their importance for ecosystem processes, impeding adaptation of forestry to sustain ECM fungal diversity.

Field experiments have shown that ECM community dynamics may be under strong influence of dispersal limitation and slow recruitment (Smith, Steidinger, Bruns, & Peay, 2018). The longevity of ECM fungal genotypes may range from less than a year to potentially exceeding that of the host trees, conditional on a continuity of living trees at the site (Douhan, Vincenot, Gryta, & Selosse, 2011). Retention trees may be a conservation measure that enables ECM fungal mycelia to survive harvesting. Clear-cutting has profound and long-lasting effects on the abundance and composition of ECM fungal communities (Jones, Durall, & Cairney, 2003; Kyaschenko, Clemmensen, Hagenbo, Karltun, & Lindahl, 2017; Varenius, Kårén, Lindahl, & Dahlberg, 2016; Varenius, Lindahl, & Dahlberg, 2017; Wallander, Johansson, Sterkenburg, Durling, & Lindahl, 2010). Clear-cutting may cause a complete disappearance of energy supply to ECM fungi while thinning reduces energy input (Jones et al., 2003; Luoma, Eberhart, Molina, & Amaranthus, 2004). However, ECM fungi may survive logging if associated with retained trees or roots extending into the area from forest edges (Cline, Ammirati, & Edmonds, 2005; Varenius et al., 2017). A limited set of ECM fungi, for example, Suillus, Rhizopogon and Cenococcum species, may also survive without their hosts by forming a bank of spores or sclerotia in the soil (Glassman et al., 2015). Spore banks can be important in providing compatible ECM fungi to seedlings in temperate forests (e.g. Wen et al., 2018), but their roles in boreal forests are poorly known.

ECM fungal communities are typically species rich and spatially heterogeneous, with a few widespread species and the large majority being rare (Wolfe et al., 2009). Thus, β-diversity (i.e. the species turnover among sites or habitats) is high, particularly on small spatial scales. Tree harvesting may, thus, be considered as habitat loss for mycorrhizal fungi and may be expected to result in a random loss of ECM fungal mycelia in relation to the level of tree cutting. As a consequence of their low frequencies, rare species should have a higher risk of local extinction than common species. Habitat loss due to harvesting is also expected to increase stochasticity in communities, further increasing β-diversity (Chase, 2003).

To approach the question of how the amount of retention trees relates to the maintenance of ECM fungal biomass and diversity during 3 years following forest harvest, we established an experiment, in which two levels of tree retention, 60% and 30% of the trees retained, were compared with unlogged forest (100% of trees retained) and clear-cuts (0% retained). The results may be related to the standards of the Swedish Forest Stewardship Council (FSC), of retaining 5% of the trees at harvest. The experiment was conducted in a 190-year-old Scots pine forest close to the northern extension limit of the boreal biome, as we wanted to study a diverse ECM fungal community unaffected by modern forestry. Scots pine forest is the most abundant type of forest ecosystem in northern Fennoscandia and commonly monodominant with respect to tree species (Stokland et al., 2003).

We hypothesized that (a) abundance of ECM fungi would decline in relation to the total fungal community (i.e. including saprotrophic and pathogenic fungi as well as fungi associated with roots of understorey plants) in the soil with decreasing amount of trees and that (b) ECM fungal species richness would decrease proportionally to the amount of trees logged with (c) most adverse effects on infrequent species. We also expected (d) β-diversity of ECM fungal communities to increase with increasing levels of tree removal.


2.1 Site description

The study site Ätnarova is located within a 19 ha stand of an old unmanaged Pinus sylvestris forest in northern Sweden (66°98N′, 20°46E′; 400 m.a.s.l.; Figure 1). It is located in the northern boreal zone on nutrient-poor moraine soil with understorey vegetation characterized by the ericaceous shrubs Empetrum hermaphroditum, Vaccinium myrtillus and V. vitis-idaea (Cladino-Pinetum sensu Kielland-Lund, 1973). Forest wildfire has been the major type of forest disturbance in this area (Engelmark, Kullman, & Bergeron, 1994). Mean annual precipitation is 700 mm and mean annual temperature is 0°C, with monthly mean temperatures ranging from −13°C in January to +14°C in July (Swedish Metrological and Hydrological Institute, records 1961–1990). The site fertility index is T13, that is, low fertility with pine as the prevalent tree and an expected tree height of 13 m at an age of 100 years (Hägglund & Lundmark, 1977). In 2007, the average tree age was 192 years, and the standing wood volume 100 m3/ha with 208 trees ha−1 (data from the forest owner Sveaskog). There was no wind-thrown trees or signs of recent forest fires and cut stumps, although cutting of individual trees may have occurred until the early 20th century.

Details are in the caption following the image
Location of the study area, experimental design and layout of soil core sampling. The study contained four retention levels: 100% (unlogged), 60%, 30% and 0% (clear-cut) retained trees, in five replicated blocks. Nine soil cores were collected according to the figure 1 year before harvest (2009) and 1 (2011) and 3 (2013) years after logging. In 2013, an additional 10 soil cores were collected from the O-horizon and kept separate; nine at the same location as for the pooled samples. The circle indicates where the extra 10th core was sampled

2.2 Study design

The study consisted of four treatments with different proportions of trees retained: (a) unharvested control (100% of the trees retained); (b) 60% of the trees retained; (c) 30% of the trees retained and (d) all trees cut (0% of the trees retained). The plots were not planted after harvest, but naturally established seedlings were left at the site. Treatments were replicated five times across five randomized blocks (Figure 1). Each treatment plot was 50 × 50 m, and the retained trees were evenly distributed throughout each plot. Logging was performed in November 2010 and the slash was left on site. The number of seedlings was counted in four 1 m2 subplots within the inner 100 m2 of each plot in September 2013 and ranged between 0 and 4 per m2 with an average of 1.4 per m2. A majority (83%) of the seedlings were less than 15 cm in height and the remaining less than 1 m (Appendix Table S1). A vast majority of the seedlings were established before the experiment was initiated.

2.3 Soil sampling

In each plot, nine soil cores (diameter of 2.5 cm, 10 cm deep) were collected in a centrally located 10 × 10 m grid pattern (Figure 1). The litter and moss layer were discarded, and the cores were divided into O-, E- and B-horizons, which were separately pooled into three composite samples per plot. This separation was done in order to account for vertical niche separation of ECM fungi in the soil profile. The samples were stored in a cooler and frozen at −20°C within 8 hr. Pretreatment samples were collected in September 2009 and additional samples were collected 1 year (September 2011) and 3 years (September 2013) after harvest. Moreover, 10 additional O-horizon samples per plot were taken in 2013 and processed separately, to allow higher spatial resolution in the community analysis. In total, we analysed 380 soil samples; 180 pooled samples (5 replicates × 4 treatments × 3 horizons × 3 sampling times) and 200 individual O-horizon samples collected 3 years after harvest (5 replicates × 4 treatments × 10 individual soil cores).

2.4 DNA analysis

Samples were freeze-dried, cleared from roots larger than 2 mm in diameter and carefully homogenized to powder in a mortar. Approximately 50 mg of soil (including fine roots) was used for Deoxyribonucleic acid (DNA) extraction with 1 ml extraction buffer (3% cetyl trimethylammonium bromide (CTAB), 2 mM EDTA, 150 mM tris-HCl and 2.5 M NaCl, pH8) at 65°C for 1 h. After extraction with chloroform, DNA was precipitated from the aqueous phase with 1.5 volumes of isopropanol. Following centrifugation, the pellet was rinsed with ethanol and resuspended in 50 μl of water. DNA extracts were further purified with the Wizard DNA clean-up kit (Promega, Madison, WI) and subjected to Polymerase chain reaction (PCR), using the gITS7-ITS4 primer combination (Ihrmark et al., 2012). The ITS4 primer was elongated by a unique sample tag of eight bases, designed using the BARCRAWL software (Frank, 2009). PCR cycle numbers were optimized for each sample and ranged between 25 and 35. Three PCR reactions from each sample were pooled and purified with the AMPure kit (Beckman Coulter Inc., Brea, CA). The concentration of the purified PCR products was established with a Qubit fluorometer (Life Technologies, Carlsbad, CA) and equal amounts of DNA from each sample were pooled and subjected to sequencing. Adaptor ligation and sequencing were performed by LGC Genomics, Berlin, Germany, on a 454-Genomic Sequencer (Roche, Basel, Switzerland), using Titanium chemistry.

2.5 Sequence analysis

Sequences were filtered and clustered using the SCATA pipeline ( Sequences were filtered for quality, removing data with an average quality score below 20 or with a score below 10 at any position, screened for primers and sample identifying tag and complementary reversed when necessary. Sequences were compared for similarity, using USEARCH (Edgar, 2010) as the search engine, requiring a minimum match length of 85%. Pairwise alignments were scored using a scoring function with 1 in penalty for mismatch, 0 for gap opening and 1 for gap extension. Homopolymers were collapsed to 2 bp before alignment. Sequences were clustered into species hypotheses (SHs; Koljalg et al., 2013) by single linkage clustering, with a 1.5% maximum distance allowed for sequences to enter clusters.

2.6 Taxonomic identification

For each sample, SHs accounting for less than 2‰ of the total reads (amplicons) were removed, to eliminate any potential impact of tag switching (Carlsen et al., 2012). The most abundant sequence in each SH was used as a representative (after restoration of homopolymers). For each SH, the most closely related reference sequences were retrieved from the UNITE database and INSD (Koljalg et al., 2013), using the BLASTn massblaster in PlutoF (Abarenkov et al., 2010), and aligned together with SH representatives in MegAlign (DNAStar Inc., Madison, WI). Taxonomic identities were assigned based on neighbour joining trees, and SHs assigned to ECM fungi were subjected to detailed phylogenetic analysis. A list of all identified ECM fungal taxa, their frequency of occurrence and relative sequence abundance within each treatment is supplemented (Appendix Tables S2 and S5). The nomenclature follows the Swedish Taxonomic Database (

2.7 Statistics

The relative abundance of ECM fungi was analysed as a proportion of the total number of fungal sequences (including all fungal life-forms), and effects of sampling year, soil horizon (categorical factors) and tree removal treatment (continuous factor, given as the proportion of trees removed) were tested by a linear mixed model (nlme) in r (R Core Team, 2013). Differences in sequencing depth were accounted for by including the square root of the total number of obtained sequences per sample as an explaining variable in the model. Plot was included as a random effect variable to account for dependency of responses among samples collected in the same plot at different time points or in different horizons. Pre-logging data (2009) were set as the reference point in the model, and interactions between time (2011 or 2013) and tree removal and between time and soil horizon were included in the model if significant.

The total number of species detected (observed species richness) within each plot was integrated across O-, E- and B-horizons. The whole plot was excluded from this analysis if data from any horizon were missing (eight plot samples of 60 were excluded). Effects on species richness of sampling year, tree removal and sequencing depth were tested by a linear mixed model as described above. Interactions between time and tree removal were included in the model if significant.

Detrended correspondence analysis (DCA) was performed in Canoco 5 (Microcomputer Power, Ithaca, NY, USA) in order to obtain a graphical representation of ECM fungal community similarity between samples collected 3 years after tree removal. The DCA was based on separately processed O-horizon samples (10 per plot), where the presence of an SH in a sample was defined as one observation, and species abundances expressed as arc sine transformed frequencies out of the total number of observations. Only species present in three or more plots were included in the analysis.

Correlations between tree removal and ECM fungal community composition were established by CCA and evaluated for statistical significance by Monte Carlo permutation tests. Tree removal was included as a continuous variable and analyses were performed both with this variable log transformed and without log transformation.

Mantel tests were performed using the package ade4 (Dray & Dufour, 2007) in R in order to analyse whether more closely located samples had more similar ECM fungal communities. The test was conducted on preharvest samples (pooled O-horizon samples; n = 20) and on samples collected 3 years after harvest (O-horizon samples kept separate; n = 200). The latter test was also conducted within each treatment (n = 50 per treatment). The Mantel tests showed that the ECM fungal community composition of the plots did not correlate significantly with spatial distance for any of the years, treatments or horizons (P = 0.22–0.58).

Beta diversity (calculated as γ-diversity divided by α-diversity) of ECM fungal communities was calculated for samples collected 3 years after harvest and with α and γ diversities considered at different spatial scales; (a) species richness of the entire site (γ-diversity) in relation to the average richness of 10 × 10 m plots (α-diversity), based on 20 pooled O-horizon samples (n = 5 per treatment), (b) species richness of the entire site (γ-diversity) in relation to the average richness of individual soil cores (α-diversity), based on 200 separate O-horizon samples (n = 50 per treatment), and (c) species richness of plots (γ-diversity) in relation to the average richness of individual soil cores (α-diversity), based on 200 separate O-horizon samples (n = 10 per plot).


3.1 Sequence output

A total of 739,345 reads passing quality control (37% of the total number of reads) were clustered into 2,650 SHs. Of these, 1,529 SHs (94% of the reads passing quality control) were assessed for phylogenetic and functional affiliation. Of these, 149 SHs (8% of assessed reads) were identified as ECM fungi. On average, the total fungal communities were represented by 2,580 (415–15,205) reads per pooled sample and by 1,020 (112–4,237) reads per sample in the individually processed samples from 2013. Ninety-nine (66%) of the ECM SHs were identified to species, whereas the remaining 50 species were identified only to the genus level (Appendix Table S2).

3.2 Abundance of ECM fungi relative to the total fungal soil community

The proportion of cores with ECM fungi present 3 years after logging (2013) remained high (85%–95%) in plots with trees retained but decreased significantly (P = 0.02) to an average of 51% after complete clear-cutting (Figure 2). The relative abundance of sequences ascribed to ECM fungi in the O-horizon of the unlogged plots varied between years: 10% (±0.03), 26% (±0.09) and 17% (±0.03) in 2009, 2011 and 2013 respectively. Accounting for these inter-annual differences in the unlogged controls, the relative abundance of ECM fungal amplicons in the O-horizon declined post-harvest proportionally to the harvest intensity (Figure 3; Appendix Table S3). After 3 years, the relative abundance of ECM fungal amplicons in the clear-cuts had declined to 5% of that in the unlogged plots. The trend was the same in the E and B horizons, although variation was larger. The square root of total read number was negatively correlated with ECM relative abundance, and there were significant interactions between read number and O-horizon and between year 2011 and E-horizon (Appendix Table S3).

Details are in the caption following the image
Proportion of soil cores with ectomycorrhizal (ECM) fungi present 3 years after harvest a Scots pine forest with 100%, 60%, 30% or 0% of trees retained. The estimate is an average of five plots, each with 10 individually analysed soil cores. Mean values with standard error, n = 50 for each treatment
Details are in the caption following the image
The relative abundance of all ectomycorrhizal (ECM) fungi combined (a) O-, (b) E- and (c) B-horizons in a Scots pine forest with 60%, 30% and 0% of trees retained at harvest. To account for variation between years, abundances are expressed in relation to ectomycorrhizal (ECM) abundance at the unlogged plots. Open circles represent samples collected 1 year after logging (2011) and closed circles represent samples collected 3 years after logging (2013) n = 5

3.3 ECM fungal species richness

In the unlogged treatment, ECM species richness did not vary significantly between years (p = 0.17) with 8.7 (±0.06), 9 (±1.7) and 5.6 (±1.0) ECM species per pooled sample in 2009, 2011 and 2013 respectively. ECM species richness declined linearly with decreasing level of tree retention (Figure 4; Appendix Table S4). With 60%, 30% and 0% of the trees retained, the average number of detected species per plot had declined to 70%, 50% and 25% of that in the unlogged plots 3 years after logging.

Details are in the caption following the image
Ectomycorrhizal (ECM) fungal species richness in Scots pine forest plots with 60%, 30% and 0% of trees retained at harvest. To account for variation between years, species richness is expressed in relation to unlogged plots (n = 3–5). Open circles represent samples collected 1 year after logging (2011) and closed circles represent samples collected 3 years after logging (2013)

3.4 ECM fungal community composition

The most species-rich and abundant genera were Cortinarius (51 species, 37% of reads), Russula (11, 4%) Inocybe (10, 2%), Tomentella (10, 0.3%), Piloderma (9, 32%), Lactarius (3, 4%) and Suillus (3, 10%) (Appendix Tables S2, S4 and S5). Over the 4 years, three species were recorded in more than 15 of the 20 plots; Piloderma sphaerosporum (20), Suillus variegatus (19) and Cortinarius obtusus (18). Eight additional species occurred in more than 10 of the 20 plots (Figure 5). Sixty-eight of the 149 ECM species were recorded in one plot and one time only (Appendix Table S2). In total, the four most frequent ECM species accounted for 43% of the total ECM reads and the 15 most frequently registered ECM species accounted for 76% (Appendix Table S5). Six species nationally red-listed in Sweden were recorded in 1–5 plots each (Appendix Table S2).

Details are in the caption following the image
The relative number of records of ectomycorrhizal fungal species detected by 454-sequencing of ITS2 amplicons in a harvest experiment in Scots pine forest plots with four levels of tree retention. Logging was conducted in 2010. Treatments were replicated five times. Only species recorded in > 6 of the 20 plots are shown

Clear-cut plots differed systematically in ECM fungal community composition from plots with retained trees (Figure 6a). No systematic differences could be observed between plots with retention trees and unlogged plots. ECM fungal community composition was significantly (p = 0.006) related to the log transformed proportion of retained trees, 3 years after logging, but without log transformation significance power was lost (p = 0.08), indicating that the clear-cut treatment was the primary driver of the relationship, in concordance with the DCA. A CCA species plot (Figure 6b) indicated that the sparse ECM fungal communities on the clear-cuts were characterized by the overall most abundant species, whereas less abundant species generally were found preferably on plots with retained trees. Species of Cortinarius almost disappeared after clear-cutting and their relative abundance decreased from 50% of the total ECM fungal community to 4% (Figure 5; Appendix Table S5).

Details are in the caption following the image
(a) Sample plot of a detrended correspondence analysis (DCA) of ectomycorrhizal (SCM) fungal communities in the humus layers of Scots pine forest plots with 100%, 60%, 30% and 0% of trees retained at harvest, conducted 3 years after logging. The dotted circle delimits the plots with retained trees. Community composition was assessed by 454-pyrosequencing of amplified ITS2 markers. (b) Species plot of a CCA of ECM fungal communities in relation to tree retention. Note that only the first axis is canonical. The area of symbols reflects frequency of occurrence. Analyses are based on samples collected 3 years after logging

3.5 Spatial variation in community composition (β-diversity)

Three years after harvest, local variation in community composition (β-diversity) was markedly higher after clear-cutting than when trees were retained (Table 1). This is also shown by the large variation between clear-cut plots in the DCA figure (Figure 6a). The reason was a progressive decrease in α-diversity with tree removal, whereas larger scale γ-diversity was maintained or less dramatically decreased.

Table 1. Alfa, beta and gamma diversity of ECM fungi in the O-horizon 3 years after logging with different levels of tree retention and at different spatial scales. In (a) and (b), the analysis is based on 10 individual soil cores per 100 m2 plot; in (c), the analysis is based on 10 pooled soil cores from each of the 5,100 m2 plots per treatment. In (a), standard error of the mean is given for β-diversity (n = 5)
Retention treatment
Spatial scale of α and γ diversity Partitioning of diversity 100% 60% 30% 0%
α = plot, γ = site Average α (= 5 plots per treatment) 5.6 3.4 3.4 1.2
γ 17 9 10 6
β 2 1.6 1.9 4
α = core, γ = site Average α (n = 50 cores per treatment) 2.8 2.2 1.9 0.9
γ 40 36 33 23
β 13.2 15.3 16.9 25.8
α = core, γ = plot Average α (= 10 cores per plot) 2.7 2.4 2.1 0.9
Average γ (n = 5 plots per treatment) 11.8 10.8 9.6 11
Average β 5.1 ± 0.8 4.8 ± 0.5 4.7 ± 0.3 6.6 ± 0.4


Understanding how different amounts of retained trees at forest harvest effect the survival of ECM fungi is essential for the development of management methods that sustain high diversity of ECM fungi in the boreal forest. We investigated effects of retention trees on ECM fungi in an old natural Scots pine forest in northern Sweden by setting up an experiment representing a gradient from no to all trees retained with effects on the abundance, richness and composition of ECM fungal communities followed for 3 years.

Our results demonstrated that ECM fungi may disappear completely after clear-cutting, but that their abundance and species richness are closely dependent on the amount of trees retained. The presence and composition of frequent ECM fungal species were largely maintained even with low retention levels, although at lower relative abundance out of the total fungal community. Infrequent species, however, were gradually eliminated with increasing level of tree removal.

The results corroborate previous studies that have highlighted ECM fungi as highly sensitive to tree removal (e.g. Högberg et al., 2001; Jones et al., 2003; Kyaschenko et al., 2017; Varenius et al., 2016, 2017; Wallander et al., 2010) and confirm earlier studies showing increased survival of ECM fungi close to retention trees and forest edges (e.g. Kranabetter & Kroeger, 2001; Luoma et al., 2004; Varenius et al., 2017). The positive correlation between retained tree densities and ECM fungal species richness supports a species–area relationship, where habitat area corresponds to the density of alive roots or trees (Arrhenius, 1921; Peay, Bruns, Kennedy, Bergemann, & Garbelotto, 2007).

4.1 Rare species fade but may benefit from retention

As expected, we found that the community composition showed higher local variation (β-diversity) after logging, related to increasing stochasticity in community composition as the root habitat radically decreased and α-diversity declined. Just by chance, locally infrequent species with few mycelia will be more likely to disappear after cutting than frequent species, although individual mycelia may face the same local extinction risk. Clear-cutting is the main reason that about 80 species of ECM fungi associated to Scots pine currently are red-listed in Fennoscandian (Tingstad et al., 2018). It is, therefore, a pressing issue to develop and implement methods to ameliorate the conditions for these ECM fungi. One approach is to use retention trees to lifeboat the potentially long-lived ECM fungal genotypes through the clear-cut phase so that they can benefit by priority effects, as forest re-establish and grow older. Assemblies of ECM fungal communities are affected by priority effects, and early colonizers have a competitive advantage (Kennedy, Peay, & Bruns, 2009). We did not specifically study the practice to leave 5% trees corresponding to the Swedish FSC standard (Gustafsson, Kouki, & Sverdrup-Thygeson, 2010). However, our results imply that the most frequent species and about 25% of the ECM fungal species may be maintained at this retention level. Retention of more trees enables preservation of higher richness, as more of the less frequent species are maintained. However, our results imply even modest harvesting operations (60% retention), corresponding to continuous cover forestry or thinning, lead to significant losses, about 30%, of ECM fungal species richness.

4.2 ECM fungi surviving clear-cutting

Contrary to our expectations, several ECM species were still detected in the soil 3 years after clear-cutting with no retained trees, although at very low abundance. This was probably mainly due to the high number of Scots pine seedlings that were naturally established prior to the logging. Such seedlings largely host the same ECM fungi as nearby old trees, implying the importance of mycelium for local spread and persistence (e.g. Teste, Simard, & Durall, 2009). This possibility for ECM fungi to survive logging may be important as old managed Scots pine forests can have significant numbers of seedlings surviving harvesting into the next rotation, although their numbers are normally much reduced by the clear-cut operation and soil scarification (Karlsson et al., 2017). The presence of a fungal soil spore bank, typically consisting of only a small subset of the fungal community in a forest, may also be important in providing compatible ECM fungi to new seedlings, as documented after stand-replacing fires in temperate coniferous forests (e.g. Glassman et al., 2015). Interestingly, Piloderma and Suillus species contributing to the spore banks in temperate forests were among the frequent species detected in this study after clear-cutting. However, the significance of an ECM spore bank is probably smaller in boreal forests, as natural boreal forest dynamics are characterized by continual presence of trees and driven by low-severity surface fires and gap dynamics rather than by stand-replacing disturbances (Kuuvalainen & Aakala 2001). This may probably have selected for ECM fungi with persistent and long-lived mycelia in boreal forest (e.g. Jonsson 1999).

4.3 Forestry may select for different ECM fungi

The clear-cutting management in Fennoscandia with rotation times of 60–100 years and even aged stands may select for different ECM fungal traits than natural boreal forest dynamics. Re-establishment of a rich ECM fungal community after clear-cutting may take several decades and depends on successively increasing photosynthetic rates, changing edaphic factors and the diversity of the ECM fungi in the surrounding landscape (cf. Kyaschenko et al., 2017; Varenius et al., 2016, 2017; Visser, 1995; Wallander et al., 2010). A meta-analysis of ECM fungi reports that species richness takes, on average, 90 years to recover to old-growth forest levels (Spake et al. 2015). Another analysis across broad taxonomic groups in secondary forests similarly implies that species richness converges to old-growth reference values within a century, that similarity in species occurrence may take about twice as long and that community composition may take up to an order of magnitude longer to return to its original status (Curran, Hellweg, & Beck, 2014). Varenius et al. (2016, 2017) reported the most frequent ECM fungi to be present in similar frequencies in 50-year-old Scots pine forests established after clear-cutting as in nearby old natural forests. These findings imply that, although clear-cutting almost eradicate ECM mycelial biomass, the most frequent ECM fungal species in pine forests, such as P. sphaerosporum and S. variegatus, survive or re-establish at forest management conditions. However, it is unclear whether these frequent ECM fungal species are also the most important species, for example, in terms of mediation of tree nutrition and growth, although these particular species, together with Cortinarius species, have been documented to correlate with efficient nitrogen mobilization and low C sequestration in boreal forests (Clemmensen et al., 2015). On the contrary, we found the abundant and species-rich ECM fungal genus Cortinarius, to contain many species sensitive to the clear-felling, as found earlier by Wallander et al. (2010) and Kyaschenko et al. (2017).

4.4 Constrains with DNA-based fungal soil studies

There are several methodological considerations to be made related to the use of high-throughput sequencing to study fungal communities in order to avoid artificial results and misleading conclusions. The selection of molecular marker may bias the ability to detect different species of interest and the relative abundance of markers representing different species may not represent their relative biomass contribution (see discussion in, e.g. Baldrian, 2017). However, as this study is a relative comparison of the same set of species in the same forest context, we judge the results to be robust.

Another consideration is that DNA analyses of soil samples have a limited capacity to capture the presence and abundance of rare species. Despite our extensive study including a total of 380 soil samples, only a small fraction of the forest soil was analysed, that is, less than 0.2 m2 of the 20 ha forest. Long-term sporocarp surveys have shown that rare ECM fungal species typically are present as single or a few mycelia within forest sites (Dahlberg & Mueller, 2011). In practice, this means that searching and monitoring rare ECM fungal species, for example, red-listed species of conservational interest, by DNA barcoding of soil cores are impractical. An exception may be when soil sampling is used to monitor-specific locations with known occurrences of rare ECM fungi (Gordon & van Norman, 2014). Otherwise, extensive sporocarp search during years with good fruiting is a better approach, as large areas can be completely and efficiently searched. However, soil studies may also detect rarely fruiting species to be more common and widespread than their sporocarps indicate.

4.5 Implications for management

Our results confirm the value of retaining trees in forest management as a measure to maintain ECM fungal biodiversity. There was a clear and positive relationship between the amount of retention trees and ECM fungal species richness as well as the relative abundance of ECM fungi in the total fungal community. Frequent ECM fungi are likely to withstand logging with at least 30% of the trees retained, but at reduced mycelial abundance in the soil. Although clear-cutting cause ECM fungal communities to be strongly impoverished even with FSC requirements of tree retention met, the most common species survive harvest. Higher levels of tree retention, that is, in continuous cover forestry, may counteract local extinctions also of less frequent species and thus support efforts to manage for sustained high ECM fungal diversity. Several rare species, and species predominantly confined to old natural forests, appear to rarely re-establish after clear-cutting and are hence red-listed. For the survival of these species, protection of forests with high conservation values and forest management directed towards conservation needs are unequivocally needed.


This research was funded by Formas (grant id 2010-180), Sveaskog and Skogssällskapets Research Foundation. We acknowledge Hans Winsa at Sveaskog for support and setting up the field experiment, the Swedish Forest Agency Board to initiate the study and Joachim Hjältén for commenting the manuscript.


    E.S., A.D. and B.D.L. designed the study and collected the data. E.S., A.D. and B.D.L. analysed the data. E.S. and A.D. led the manuscript writing. All authors contributed critically to the drafts and gave approval for publication.


    Appendix data available via the Dryad Digital Repository. (Sterkenburg, Clemmensen, Lindahl, & Dahlberg, 2019a). Data available via the NCBI SRA as Bioproject PRJNA506533 ( (Sterkenburg, Clemmensen, Lindahl, & Dahlberg, 2019b).