Near-to-nature logging influences fungal community assembly processes in a temperate forest

Authors


Summary

  1. As the Earth's biota enters the sixth great mass extinction event recorded in the history of the planet, it is predicted that the erosion of biodiversity will result in the reduction of the goods and services that ecosystems provide. To mediate the loss of biodiversity and ecosystem function associated with wood production in temperate forests in Europe, a near-to-nature strategy has been developed. Whether this strategy enables natural assembly mechanisms of fungi responsible for major ecosystem processes is unknown.

  2. We analysed variation in species richness and both the functional and phylogenetic structure of fungal assemblages of different trophic life histories (soil saprotrophic, wood saprotrophic, and ectomycorrhizal fungi) in 69 beech forest plots along a steep gradient of management intensity. We focused on reproductive traits to test the hypothesis that management intensity shifts community assembly mechanisms from limitations in niche overlap that promote the coexistence of dissimilar species to environmental filtering that selects for similar species. Specifically, we hypothesized that unpredictable resources in production forests filter the assemblages, promoting species with small fruit bodies and with small and elongated spores.

  3. As management intensity increased, functional diversity decreased from a random pattern to a clustered pattern, which indicated that management intensity increased the strength of environmental filtering. However, phylogenetic diversity increased from a random pattern to an overdispersed pattern. Combining phylogenetic diversity with functional diversity did not provide additional insight into the traits but revealed a contrasting pattern. Reproduction traits of the assemblages shifted, with a decrease in mean fruit body size and an increase in spore elongation.

  4. Synthesis and applications. Near-to-nature logging concepts are not able to mimic the major processes that shape fungal community assembly in protected forests. This could have severe consequences for important ecosystem functions provided by fungi. The phylogenetic overdispersion indicated that analyses of other traits in addition to reproductive traits are required to disentangle the factors affecting fungal community structure.

Introduction

Ecologists have debated for almost a century whether the assembly of communities is driven by environmental filters (Gleason 1926), biotic interactions (Clements 1916), or dispersal processes (Hubbell 2001). Leibold et al. (2004) unified these contrasting concepts and showed that these paradigms trade off along ecological gradients. One of the strongest ecological gradients in modern landscapes is anthropogenic disturbance (Millennium Ecosystem Assessment 2005). An understanding of the impact of anthropogenic disturbance on processes involved in community assembly is crucial for improving conservation strategies (Cadotte, Carscadden & Mirotchnick 2011). For example, functional approaches have provided new insights for coral fish conservation strategies beyond strategies based on species richness alone (Stuart-Smith et al. 2013).

European forests display a gradient of naturalness as human activities alter natural diversity (Hannah, Carr & Lankerani 1995; Grove 2002a). To protect and promote forest ecosystem functions and biodiversity in the face of commercial timber harvesting, forest authorities in Europe developed a ‘near-to-nature’ strategy beginning in the mid-20th century. The concept has primarily focused on forests dominated by beech Fagus sylvatica L., the most common natural ecosystem in Europe (Brunet, Fritz & Richnau 2010). This strategy is based on selective cutting that promotes regeneration of native tree species (Boncina 2011), but overmature trees and dead wood are regularly removed.

Despite the impetus for near-to-nature strategies, evidence shows that wood harvesting decreases species richness and shifts community compositions (e.g. Brunet, Fritz & Richnau 2010). The broader ecological consequences of near-to-nature activities are poorly understood (but see Gossner et al. 2013). By understanding how community traits respond to management activities, we can better link diversity change to potential ecological mechanisms. Moreover, combining both functional and phylogenetic approaches allows an assessment of well-measured traits that explain an ecological pattern in comparison with unmeasured conserved traits represented by a phylogeny (Cadotte, Albert & Walker 2013).

According to theory, the total trait space occupied by a local assemblage should non-randomly decrease in extent as a result of human disturbance (deterministic model sensu, Lin et al. 2011; Mouillot et al. 2012). Therefore, with increasing anthropogenic disturbance, we expect that environmental trait filtering should become more important and possibly override biotic processes that determine species assembly, for example, competitive exclusion. To test these hypotheses for beech forests, we used fungi of different trophic lifestyles as a model system since this diverse group drives major ecosystem functions in these forests (e.g. regulating nutrient cycling, Carlile, Watkinson & Gooday 2001). We considered fungal reproduction traits strongly related to dispersal and establishment in the exploitation of new resources (fruit body size and spore characteristics; see Norden et al. 2013). We expected management intensity to favour species that follow an r strategy, characterized by small fruit bodies enabling them to react flexibly to ephemeral and limited resources and by small spores, which have been shown to travel further (Norros 2013). Although overall evidence of the relevance of spore shape in dispersal capability is still lacking, we expected a shift along management intensity towards species characterized by elongated spores, which are assumed to float better through air (Ingold 1965; Norros 2013). Specifically, along a steep gradient of management intensity, we tested the hypothesis that despite near-to-nature management, logging shifts the processes that influence fungal species assembly from limiting similarity to habitat clustering, thereby filtering the assemblages and promoting species with traits that allow them to respond to patchily distributed resources (small fruit bodies) and to disperse efficiently (small and elongated spores).

Material and methods

Study area and study design

The Steigerwald (49°50′N 10°29′E) is a temperate (mean annual temperature 7–8 °C; mean annual rainfall 850 mm), colline to sub-montane, forested, low mountain range and covers about 1000 km2 in Southern Germany. The natural forests are dominated by beech Fagus sylvatica L. The study plots are situated at 325–520 m a.s.l. Beech is the dominant tree species in the study area (43% cover), followed by oak (Quercus robur L. and Quercus petrea Liebl., 20% cover). Deciduous trees cover more than 70% of the area. For our study, only beech-dominated stands with low proportions (<10%) of Q. robur, Q. petrea, Fraxinus excelsior L., Acer sp., or coniferous trees were selected. In a 15 × 15 km2 block, we selected 69 inventory plots in mature beech stands of commercial forests and, in protected reserves, two of which have not been managed for the past 40 years, and management of the third ceased about 25 years ago. The size of each circular plot was 0·1 ha. The distance between two plots was always ≥100 m (for details, see Appendix S1, Supporting Information).

Management intensity

Deriving indices of forest management intensity is challenging and still under discussion for Central European forests (Luyssaert et al. 2011; Luyssaert, Hessenmöller & Schulze 2013; Schall & Ammer 2013a,b). However, available indices are in principle based on the age of the trees, amount of dead wood and presence of coniferous trees and stumps (Schall & Ammer 2013b). We decided to use the recently published axis of a canonical correspondence analysis (CCA) combining most of the above-mentioned forest structural features (e.g. amount of dead wood) and a species group known to be highly sensitive to forest management (saproxylic beetles; 9303 individuals – 284 species, see also Grove 2002b) as a measure of management intensity (Gossner et al. 2013). Such a combination of available structural resources with community data creates a gradient with a high resolution because beetles may represent variation in microhabitats beyond coarse measures of forest structure (see also Winter, Fischer & Fischer 2010). Community ecology provides evidence that species assemblages often respond more sensitively to environment than other measures (Schaffers et al. 2008). For example, highly managed plots (indicated by large score values in our gradient) are physically characterized by mature but not overmature trees with lower amounts and a highly patchy distribution of dead wood colonized only by generalist saproxylic beetles (Müller, Bussler & Kneib 2008). The gradient used in our study proved to be highly robust; correlations with various other measures considering, for example, only forest structural variables or availability of stumps, or from a CCA excluding beetles related to fungi were considerable (Figs S1 and S2, Supporting information).

Fungal inventory and traits

Fungi were sampled in 2004 in three campaigns (spring, summer, and fall). In each campaign, all 69 plots were sampled by an expert mycologist (Heinz Engel) within two weeks with similar moisture conditions. We focused mainly on species with conspicuous fruit bodies (macromycetes) that could be identified macroscopically, but we considered all trophic lifestyles [soil saprotrophic, wood saprotrophic, parasitic and ectomycorrhizal fungi; Table S1 (Supporting information), complete species list; Fig. 1]. Critical species were later examined with a microscope. Fungal fruit bodies of corticoid-like fungi were considered when macroscopic identification was possible. For wood saprotrophic fungi, all pieces of dead wood with a diameter >12 cm within each plot were inspected irrespective of the length. Because sampling was restricted to fruit bodies, it covers only a subset of the actual fungal assemblage. However, in larger field studies, fruit body inventories provide a reliable and the only feasible method to mirror fungal assemblages (Tóth & Barta 2010). All soil-related and wood-object-related samples collected during the three campaigns were compiled to yield a presence/absence matrix. In the final data set, we excluded taxonomically critical species, myxomycetes, anamorph forms, and parasites.

Figure 1.

Phylogenetic tree of 274 fungal species. Bar colour indicates the trophic lifestyle: black, soil saprotrophic fungi (a, Clitocybe vibecina); green, wood saprotrophic fungi (b, Oudemansiella mucida); red, ectomycorrhizal fungi (c, Amanita muscaria).

To account for important functional processes, effect traits were derived from literature covering the main trophic lifestyles of each species (soil saprotrophic, wood saprotrophic and ectomycorrhizal fungi; e.g. Rinaldi, Comandini & Kuyper 2008). As response traits, we used characteristics related to sexual reproduction and dispersal, namely fruit body size, spore volume and spore shape (e.g. Kauserud et al. 2010; Norden et al. 2013). These traits have been also used for the classification of lifestyle strategies, such as the r/K strategy (e.g. Moore et al. 2008). To calculate the fruit body size, we used a measure suggested by Tóth & Feest (2007). Because the cap area (mm2) has been shown to be a reliable measure for fruit body size (relative biomass), we extracted mean cap sizes from the literature (e.g. Knudsen & Vesterhold 2008). For resupinate fungi, we considered the fruit body thickness as the basis for calculating size. This approach proved to be very robust (Fig. S3, Supporting information). Spore volume was calculated from spore length and spore diameter, assuming an ellipsoid; spore shape was calculated as the ratio between spore length and spore diameter (the same measures as Kauserud, Colman & Ryvarden 2008; Kauserud et al. 2010; Norden et al. 2013). For the complete species list, related traits and associations to the sampling plots, see Table S1 (Supporting information).

Phylogeny

The fungal phylogeny was estimated using a data assembly pipeline in r language (R Development Core Team 2013; detailed description in Appendix S2, Supporting information). The tree is based on available nuclear and mitochondrial sequences from the nucleotide repository in GenBank. Using a literature-based guide tree (Hibbett et al. 2007; and references therein), we successively aligned the sequences of each genetic marker. Tree topology and branch lengths were modelled in a maximum-likelihood framework (raxml v7.6.6; Stamatakis 2006). Sequence evolution was modelled for each marker separately (Stamatakis 2006) on a common topology. The confidence in the tree topology was assessed via 1000 nonparametric bootstrap replicates (Stamatakis, Hoover & Rougemont 2008). Branch lengths were converted from substitution per site to relative time, that is, making the tree ultrametric, using the r package ape (Paradis 2013) with a penalized likelihood approach using a relaxed clock model of rate evolution and the smoothing parameter lambda = 1·0 (Sanderson 2002; Paradis 2013). Since not all of the 303 sampled species were represented in GenBank, the final tree for the analysis consisted of 274 species (90% of all species sampled; Table S2, Supporting information). It is unclear how differences in resolution and branch lengths among trees would change ecological inferences (Gonzalez et al. 2010). We therefore tested whether the phylogenetic diversity measures yield comparable results based upon ultrametric and non-ultrametric trees; the results were very robust (Fig. S4, Supporting information). To overcome the uncertainty of covering only 90% of all species when applying the above-mentioned method, we also constructed a tree with branch lengths from relative node ages for all 303 species (Fig. S4, Supporting information) and ran the analysis again. Since the results were robust, irrespective of the method applied, we present only the results obtained with the ultrametric tree based on the sequence data (for the other results, see Fig. S4, Supporting information).

Analysis

To investigate the independence of traits in the data set, we analysed the co-variance structure by principal component analysis (PCA) and phylogenetic principal component analysis (PPCA), using the R package phytools. According to both ordinations, the trophic lifestyles showed no distinct pattern, whereas soil saprotrophic and ectomycorrhizal fungi nested within wood saprotrophic fungi (Fig. S5, Supporting information). According to PCA, fruit body size negatively correlated with spore shape; according to PPCA, spore volume negatively correlated with spore shape. However, pairwise correlation coefficients (r) were <0·45; therefore, we used the raw values as independent measures in our analysis. For fruit body size, spore volume and spore shape, we calculated the phylogenetic signal based on Pagel's λ (Pagel 1999). Significance was estimated using 999 randomizations.

We tested the effect of management intensity on the overall number of species (log10 transformed) with a linear model. To test for differences between the number of species among the lifestyles, we fitted a linear mixed-effect model (r package lme4) and used management intensity as fixed factor and study plot as random factor. To test the influence of management intensity on the community composition, we applied redundancy analysis (RDA; DCA length of gradient of 2·96, see Leps & Smilauer 2003).

To test the effects of management intensity on the processes affecting the assembly of species (limiting similarity vs. habitat filtering), we calculated the functional distance (Gower distance) of all traits, the phylogenetic distance (patristic distance) and a distance matrix that integrated both of these measures in one (mean pairwise functional phylogenetic distance; MFPD) recently developed by Cadotte, Albert and Walker (2013). The key to MFPD is a weighting parameter, a, which weights the contributions of phylogenetic and functional distances to MFPD. When = 0, MFPD only includes functional distances, and when = 1, MFPD only includes phylogenetic distances. At intermediate values of a, both functional and phylogenetic distances contribute to MFPD. Based on these distance matrices, we calculated the standardized effect size of the mean distance between co-occurring species within each assemblage using the R package picante. The effect size was calculated based on null models with 999 randomizations by reshuffling the tip labels to achieve independence from species numbers (Webb et al. 2002). This null model retains the structure of the assemblage matrix of fungi and tests whether the functional composition of species within assemblages is random. Values above zero indicate overdispersion; values below zero indicate clustering (Pausas & Verdu 2010). To test for robustness of the results related to the functional diversity, we additionally calculated the functional dispersion (Laliberté & Legendre 2010). Finally, we calculated the mean (log10 transformed) and diversity (ses.mpd) for each continuous trait across all lifestyles and tested the relationship with management intensity and for interaction among the lifestyles again by applying a linear mixed-effect model. To check the residuals of our models for spatial independence, we used cross-correlograms provided by the r package ncf (Bjornstad & Falck 2001; see also Fig. S6, Supporting information).

Results

Our final data set included 303 sampled fungal species, with a mean of 29 species per plot; 67 species were assigned to soil saprotrophic fungi, 76 to ectomycorrhizal fungi and 160 to wood saprotrophic fungi. Along the management intensity gradient, we found no effect on the number of species (math formula = −0·01). When we grouped species according to their lifestyles, the number of soil saprotrophic fungi species decreased significantly with increasing management (Table 1, Fig. 2). In contrast, the number of ectomycorrhizal fungi species increased, but the effect was not significant. Only the number of wood saprotrophic fungi showed no trend along the gradient (Fig. 2). Overall fungal community composition showed a turnover with increasing management intensity (Fig. 2).

Table 1. Phylogenetic signals for the traits using Pagel's λ
TraitPagel's λ
  1. Significance was estimated by using 999 randomizations. Significant values are in bold (*< 0·05, **< 0·01, ***< 0·001).

Fruit body size 0·09
Spore volume (μm3) 0·23 *
Spore shape 1·00 ***
Figure 2.

Number of fungal species (log10 transformed) of three different lifestyles along a forest management intensity gradient. (a) Soil saprotrophic fungi (SSF); (b) ectomycorrhizal fungi (EMF); (c) wood saprotrophic fungi (WSF). (d) Biplot based on redundancy analysis of 303 sampled fungal species using management intensity as the constraint variable.

Spore volume and spore shape but not fruit body size showed a significant phylogenetic signal (Table 1). Some lifestyles were strongly associated with certain clades (Figs 1 and S7, Supporting information). Functional trait diversity decreased with management intensity (β = −0·34, math formula = 0·05, < 0·05; Fig. 3). Within forest stands with low management intensity, the traits of assemblages were randomly distributed, whereas in stands with high management intensity, traits of assemblages clustered (Figs 3 and S8, Supporting information). In contrast, phylogenetic diversity increased significantly with increasing management intensity, from a more random pattern towards an overdispersion pattern (β = 0·19, math formula = 0·05, < 0·05, Fig. 3).

Figure 3.

Fungal functional and phylogenetic relationships with management intensity. (a) Standardized effect sizes (SES) of the mean pairwise distance for all traits based on 303 sampled species. (b) Histogram of the contribution of the individual traits to the pattern in functional diversity. 1−R2 is the correlation of overall functional diversity with functional diversity in which one trait was removed. Correlation was based on the SES of the mean pairwise distance (Fb, fruit body; Sp, spore). (c) SES of the mean pairwise distance of the species based on their phylogenetic relationship. (d) Adjusted R-squared values and (e) slopes of the regression between the SES of the mean pairwise distance based on functional–phylogenetic distances (MFPD) and management intensity across the range of a-values (phylogenetic weighting parameter). (c–e) are based on 274 fungal species (90% of all sampled species) for which sequence data in GenBank were available.

Our approach of integrating both functional and phylogenetic distances into one distance (MFPD) corroborated the above findings. However, with increasing weighting of phylogeny (> 0), R2 from the linear model of the effect of management intensity on standardized effect sizes of the mean pairwise distance was U-shaped, which indicated that phylogeny provides no substantial improvement in explanatory power over functional traits. However, in a correlation of standardized effect sizes of the mean pairwise distance with management intensity (phylogenetic overdispersion in highly managed stands), contrasting signs of slopes indicated that there were unmeasured traits relevant for species assembly represented by phylogenetic distances (Fig. 3).

An examination of within-lifestyle patterns showed an almost contrasting pattern (Fig. S9, Supporting information). While the functional diversity of soil saprotrophic and wood saprotrophic fungi decreased with increasing management intensity, the phylogenetic diversity of soil saprotrophic fungi decreased but was not significant and that of wood saprotrophic fungi increased significantly (Fig. S9, Supporting information). In contrast, although the scatter was considerable, the functional diversity of ectomycorrhizal fungi increased and the phylogenetic diversity decreased with increasing management intensity.

The mean fruit body size of fungal assemblages significantly decreased, and the mean elongation of the spores significantly increased as management intensity increased (Fig. 4). When we focused on the mean trait values within each of the lifestyles, the mean fruit body size of wood saprotrophic fungi significantly decreased and that of ectomycorrhizal fungi significantly increased with increasing management intensity (Table 2). The effect of management intensity on mean assemblage spore shapes was strongest for wood saprotrophic fungi, whose spores significantly elongated (Table 2). The diversity of the fruit body size of soil saprotrophic and wood saprotrophic fungi significantly decreased with increasing management intensity (Table 2). Neither the mean spore volume nor the diversity model of spore volume was significantly related to management intensity. The residuals of our models showed no or only a weak spatial autocorrelation (Fig. S6, Supporting information).

Table 2. Standardized effect sizes (SES) from linear mixed-effect models for number of fungal species, mean values of reproductive traits and diversity of traits along a management gradient
 SSFWSFEMFAdjusted R2
  1. SSF, soil saprotrophic fungi; WSF, wood saprotrophic fungi; and EMF, ectomycorrhizal fungi.

  2. Significant effect sizes are in bold (*< 0·05, **< 0·01, ***< 0·001). Significant differences among the lifestyles are indicated by different small letters.

Number of species2·96**a−1·01a1·89b 0·60 ***
Mean fruit body size (mm2)−1·71a3·14**a2·58**b 0·10 ***
Mean spore volume (μm³)−1·31a0·36a1·08a 0·12 ***
Mean spore shape−0·52a3·91***b0·28a 0·68 ***
SES fruit body size2·85** ab6·32***a1·77b 0·33 ***
SES spore volume−1·14a1·77a−0·54a 0·03 *
SES spore shape0·37a1·44a1·38a 0·14 ***
Figure 4.

Proportion and mean trait values of fungi and fungal assemblages along a forest management intensity gradient. (a) Proportion of soil saprotrophic fungi (SSF, light grey, dotted line), wood saprotrophic fungi (WSF, dark grey, solid line) and ectomycorrhizal fungi (EMF, medium grey, dashed line) along the gradient. (b) Mean spore volume, (c) mean fruit body size and (d) mean spore shape of fungal assemblages along the gradient.

Discussion

Our results showed that the mere number of species is a poor indicator of the ecological impacts of near-to-nature management on biodiversity. The species composition and the functional composition of assemblages on the other hand provided crucial information on human impacts on species assemblages and their ecosystem functions. Moreover, our results supported the view that logging acts as a habitat filter in promoting species able to respond more flexibly to the patchy resources in managed forests. Although our results were consistent and robust (Figs S4 and S8, Supporting information), most of the R2 values of effect size vs. management intensity were rather small (Fig. 2). However, it has been stressed that biological relevance of an effect must not necessarily be related to a large value or to statistical significance (Martinez-Abrain 2008). Knowing a priori the magnitude of a biologically significant effect would be optimal (Nakagawa & Cuthill 2007), but such knowledge is not available for our study system. Thus, we evaluated our hypotheses of the relevancy of the effects. One explanation of the weak R2 values in our study might be that the fungi are characterized by a rather cryptic and ephemeral lifestyle, thereby producing considerable scatter.

Recent studies have suggested that phylogenetic diversity can be used as a proxy for measures of functional diversity; this relationship is premised on the reasonable assumption that patterns of relatedness reflect ecological similarities so long as diversification rates are not lower than rates of niche evolution (Cadotte, Cardinale & Oakley 2008). However, the phenotypic and phylogenetic structure of a community is a consequence of the evolution of a number of different traits, which may show patterns of convergence and rapid divergence, and the dominant assembly process (habitat filtering vs. limiting similarity; see Webb et al. 2002; Pausas & Verdu 2010 for an conceptual overview). In our study, phylogeny provided no additional information on the functional distance. However, patterns of phylogenetic relatedness contrasted with functional diversity patterns despite strong phylogenetic signals in the individual reproductive traits. This indicated that our results for the measured traits are robust and informative, and independent from patterns of relatedness in the phylogeny, but it also showed that phylogeny is not simply a surrogate for measured functional traits and may provide complementary information on the assembly process (Cadotte, Albert & Walker 2013). That is, the phylogeny likely reflects conserved unmeasured traits that interact with the assembly mechanisms.

To draw conclusions about the true assembly process (habitat filtering vs. limiting similarity), one would need information on the evolution of the missing assembly-relevant traits. If the missing traits hidden in our phylogeny are conserved, this should lead to overdispersion similar to that shown with our observed phylogenetic distances, which would indicate that limiting similarity is the predominant assembly rule. If, however, the traits evolved convergently, this would necessarily lead to phenotypic clustering, similar to that of functional diversity, which would indicate that habitat filtering is the predominant assembly rule along the management intensity gradient. The variety of possibilities has been referred to as ‘the jungle for evaluating phenotypic and phylogenetic structure of communities’ (Pausas & Verdu 2010). Not surprisingly, we were only able to speculate about assembly-relevant fungal traits since no study provides comprehensive information on fungal life-history strategies.

In our study, mean fruit body size decreased and mean spore elongation increased as management intensity increased. This effect was most pronounced for fungi decomposing organic material and might be a response to a limited and patchily distributed resource pronounced by forest management. Commercial forests have negligible coarse woody debris (Bauhus, Puettman & Messier 2009). On the one hand, branches, stumps and other logging residuals are typically patchily distributed; on the other hand, most of the residual branches are available only for a short time because of their small diameter and the sudden death due to logging. Fungi therefore might have to reach and colonize these resources quickly and to reproduce immediately after establishment. Irrespective of an assumed trade-off between the size and number of fruit bodies produced, fungi with small fruit bodies flexibly react by adjusting the number of fruit bodies to the resources available.

Norros (2013) has shown that spore size is a relevant dispersal trait in terms of determining spore deposition rate from the air to a surface and that small spores disperse considerably further than large spores (but see Kuparinen et al. 2007). However, at the scale of our study (15 × 15 km), characterized by less-fragmented homogenous forested landscape, spore size seems not to be a critical trait. Conversely, a study in Finland covering a much larger scale has demonstrated that generalists with larger spores bridge large distances in highly fragmented landscapes (Norden et al. 2013).

Although spore shape has been used in many studies, no evidence has been presented that it responds to environmental gradients (e.g. Norden et al. 2013). Our study used a broad phylogenetic range of assemblage data (e.g. spore length in Norden et al. 2013: 3–18 μm; in our study: 3·35–67·5 μm). The often-neglected ascomycete species frequently have very long and narrow spores. Elongate spores probably influence aerodynamic properties, which in turn lead to better floating through air and thereby improve wind dispersal, as suggested by Ingold (1965). Our results on fungal fruit body size and dispersal ability may be interpreted in terms of the concept of r/K strategy; organisms following r strategy cope better with disturbed environments through their ability to rapidly colonize habitats and by being less competitive (e.g. Grime 1988).

Ecologists have traditionally explored the disturbance–diversity relationship by examining species richness, evenness or population abundances (Mackey & Currie 2001). These metrics have been criticized as weak quantitative tools because different processes may affect species in different ways, potentially providing no signal of disturbance or even a false signal of ecosystem recovery (Cadotte et al. 2010; Mouillot et al. 2012). In our study, management had a strong influence on the proportion of fungi with a specific trophic lifestyle, while at the same time the overall number of species remained stable (see also Ernst, Linsenmair & Rödel 2006 for anurans). In our study, soil saprotrophic fungi and ectomycorrhizal fungi showed contrasting responses in species richness to the management intensity gradient. Both groups of fungi are affiliated with the forest soil, although they generally use different carbon sources. Logging of beech forests results in a reduced basal area and litter fall and at the same time increases decomposition as an effect of canopy interruption (microclimate effect; Merino, Real & Rodríguez-Guitán 2008). Although this might lead to a short-term availability of nutrients, there is evidence that logging leads to a decreased uptake of nitrogen and phosphorous along with denitrification, respiration, leaching and runoff (e.g. Rosen 1984; Tiedeman, Quinley & Anderson 1988). In contrast, litter input in undisturbed old-growth forests is higher with a long-term supply of nutrients (humus accumulation; Finer et al. 2003). Therefore, the decrease in soil saprotrophic fungi might be explained by the limited, probably more patchily distributed resources (soil organic matter) owing to reduced litter fall.

The resource for wood saprotrophic fungi is dead wood. Despite 90% less dead wood in managed beech forests compared with that in natural forests (Christensen et al. 2005), the number of wood saprotrophic fungal species on the highly managed plots in our study did not decrease. Our results contrast the results of other studies dealing with wood-inhabiting fungi (e.g. Bader, Jansson & Jonsson 1995). One possible explanation is that a specific amount of small dead wood objects harbour more species than comparable amounts of large dead wood objects (Heilmann-Clausen & Christensen 2004; Bässler et al. 2010) owing to reduced competition on the variety of dead wood sizes typical after logging compared with competition on one large dead tree in natural forests. However, inconsistent results of the effect of forest management on species diversity might be masked by the set of species groups considered in a certain study. Most of the studies in this field deal with polypores, which are very sensitive to forest management (Junninen & Komonen 2011). As discussed previously, our range of taxa spans a broad gradient across the fungal phyla.

Conclusions

Our results suggested that near-to-nature logging schemes alter natural fungal community assembly processes. The clear shift in the functional composition of fungi should be a signal for forest managers that ecosystem functions and services are likely to be impacted and perhaps reduced in near-to-nature managed forests. Experiments testing the removal of litter fall or dead wood are required to determine how different fungal strategies or traits influence overall ecosystem function. For mycologists, the contrasting patterns of functional diversity and phylogenetic diversity indicate that other assembly-relevant traits of fungi need to be identified and considered in future studies. We urge community ecologists to use combined phylogenetic and functional approaches to understand the complex mechanisms responsible for species assembly.

Acknowledgements

This research was supported by the Bavarian State Forestry. We are grateful to Heinz Engel, a brilliant mycologist, for sampling and identifying fungi.

Ancillary