Introduction

Most terrestrial habitats are known to rely on fungi for key parts in the nutrient recycling process. It is mainly their decomposing and mineralisation activities that create important habitats for many micro-organisms, protists, fungi, and arthropods. Their important trophic role notwithstanding, fungi remain a poorly understood kingdom of the eukaryotic tree of life with respect to species richness, species composition and ecological functionality of species (Molina et al. 2011). Their abundance in virtually all ecosystems with wooden substrata has been fully realized only in the last few decades (Lonsdale et al. 2008; Ovaskainen et al. 2010).

Apart from biological reasons for this lack of knowledge (cryptic fungal life as subterranean mycelium), further arguments are found in a comparatively low societal relevance of fungi (many fungi, such as moulds and poisonous agarics are perceived negatively by the European public) and in the politics of science, which are partly responsible for the limited number of mycologists with taxonomic expertise that are employed. Because biodiversity research is generally underfunded with significant lack of manpower, the use of volunteers is becoming increasingly important in these scientific areas (Lovell et al. 2009). Especially in Germany there exists a long tradition of high-quality data collection by hundreds of amateur mycologists for many decades, leading to the accumulation of millions of data of macrofungal (fruit body) occurrence. Such information was, for instance, used to compile distribution atlases of macrofungi in Germany (Krieglsteiner 1991, 1993), various Red Lists (e.g. Anon 1992; Schwik and Westphal 1999; Schmitt 2007), or the recent Funga of Saxony-Anhalt (Täglich 2009).

The analyses of such comprehensive datasets with new and enhanced statistical tools is of particular importance for the urgently needed updates of regional, national and global species richness (Hawksworth 1991; Hawksworth and Rossman 1997; Mueller et al. 2007). Nearly all existing extrapolations of fungal species richness are based on a number of assumptions such as ratios of fungi to their associated hosts, anamorph–teleomorph relationships, or the consideration of understudied fungal groups and habitats (Hawksworth 2001; Fröhlich and Hyde 1999). Consequently, species richness estimations differ strongly, depending on the taxonomic group, the geographical area under investigation, and the number of specialists involved in the surveys. This has led to sceptical considerations with respect to credibility of species richness estimations (May 1991). In addition, undersampling might prevent meaningful estimations of any organism's species richness (Coddington et al. 2009). Only recently, microbiologists, protistologists and mycologists became aware of new, promising methods to extrapolate total species richness (Bohannan and Hughes 2003; Ugland et al. 2003; Chao et al. 2006) from physical species (Unterseher et al. 2008), molecular operational taxonomic units (Jumpponen and Jones 2009) or from cultured isolates (Joshee et al. 2009; Unterseher and Schnittler 2010). The next generation sequencing technologies have, more than other methodologies, the potential to overcome undersampling thus launching a new era of fungal diversity research (Öpik et al. 2009; Amend et al. 2010; Unterseher et al. 2011).

It is the large number of observed fungal species, especially the many rare and rarest ones (‘rare biosphere’) that further complicates statistical analysis and communication of results (Novotny and Basset 2000; Cunningham and Lindenmayer 2005; Reeder and Knight 2009). In the 1980s, the concept of core and satellite species was introduced by Hanski (1982) to cope with the problematic analysis of too strongly varying abundances of species within a natural community.

In theory, Hanski's approach distinguishes between species with differing niche requirements: an abundant, enduring group of species (the core group) and a more diverse, variable and transient component of the community (the satellite/occasional group) (Hanski 1982; Magurran and Henderson 2003; Ulrich and Zalewski 2006). Recently, this theory was picked up by Pedrós-Alió (2006), Dolan et al. (2009) and Unterseher et al. (2011) to analyse the 'rare biosphere' of marine bacteria, planctonic ciliates and fungi, respectively.

Apart from generating new and continuously growing datasets, legacy data are of high value. Many specimens are preserved in stand-alone fungal culture collections, public exsiccaria at Universities and botanic gardens and linked to curated online databases and further web resources (Wilkinson and Foster 2004; Crous et al. 2004; Abarenkov et al. 2010). Moreover, there exist dozens of private exsiccaria and even more voluminous digitalised datasets and printed card indices owned and maintained by highly experienced amateur mycologists.

For the present study, we used existing data about the occurrence of macromycetes and concentrated on species with mycorrhizal, saprobic, pathogenetic and otherwise unknown modes and degrees of association to wooden substrata in Mecklenburg-Western Pomerania (MV). We aimed at testing the following hypotheses: (1) the present data gathered by volunteers allow analyses of fungal community ecology, i.e. of species richness, species turnover and host preferences; (2) species richness estimators allow serious predictions of total species richness of tree-associated macrofungi in MV; and (3) the separate analysis of abundant fungal ‘core’ and rare ‘satellite’ species (Magurran and Henderson 2003) helps to define indicator taxa for species, habitat and landscape conservation, irrespective of undersampling or an open community structure with a high temporal or spatial species turnover.

Materials and methods

Basic geobotanical data of the collection area Mecklenburg-Western Pomerania

The German state Mecklenburg-Western Pomerania (MV) is located in the very northeast of the country (Fig. 1). It covers an area of 23,180 km2 and has the lowest population density of all German states with 72 inhabitants per km2 (1.65 M inhabitants in total). Population is characterised by small and middle-sized cities (20,000–50,000 inhabitants) The climate is characterised by a transition from maritime to continental-temperate climate.

Fig. 1
figure 1

The German state Mecklenburg-Western Pomerania (shaded area, left). The right part displays an ordinance survey map (“Messtischblatt”) of Mecklenburg-Western Pomerania and the locations of the database entries (black dots) as used for the present paper. Size of the dots correspond to the number of database entries in that grid. (source: map of Germany from http://de.wikipedia.org/wiki/Mecklenburg-Vorpommern)

Full size image

The Pomeranian landscape was shaped by the pleistocenian glacial period and has a mean elevation below 50 m a.s.l. Three national parks were established in MV: the “Müritz”, the “Vorpommersche Boddenlandschaft” and the “Jasmund”. In MV, there exist 58 habitat types according to Appendix I of the EU Council Directive on the conservation of natural habitats and of wild fauna and flora including 15 with priority (Anon 2004). Among these, several habitats are of particular importance for the preservation of regional fungal diversity: (1) nutrient-poor Pinus- and Fagus-rich forests (spine fungi, coralloid fungi, agaricoid fungi); (2) calcareous forests of Fagus, Quercus, Pinus (Boletales, Cortinariales, Russula, Lactarius); (3) dunes and dry, sandy grasslands (Geastrales); (4) rough pastures and grasslands (Hygrocybe, Clavariaceae, Geoglossaceae, Entolomataceae); (5) fens, transitional mires and raised bogs (bryophilous and peat-inhabiting fungi); (6) wet habitats such as marshland forests with Alnus, Salix, Populus, Fraxinus (saprobic and mycorrhizal fungi); and (7) old-growth trees, especially of Quercus, Fagus, Ulmus and Pinus (poroid and corticioid fungi).

Dataset

This study relied on wood-inhabiting, mycorrhizal and saprobic macrofungi associated with abundant and both economical and ecological important deciduous and coniferous tree species. The data are entirely based on fruitbody observations and belonged to the private database of Mr. Benno Westphal. Data were gathered by numerous honorary mycologists and nature conservationists throughout decades (see Acknowledgements). Data were regularly verified by B.W. They are currently made available through the internet portal “Biological survey databases and herbaria in Mecklenburg-Vorpommern”Footnote 1 of the University of Greifswald, Germany.

Fungi associated with herbs, shrubs and further non-tree plants, parasitic rusts, smuts, mildews as well as myxomycota, asexual taxa (e.g. Coelomycetes) or taxa with tiny fruit bodies (e.g. cyphelloid Agaricales) were removed. The remaining data were then sorted for nutrition strategy (mycorrhiza, saprophyte) at the generic level according to ecological annotations, authors' knowledge and literature (e.g. Breitenbach and Kränzlin 1984–2005; Horak 2005; Kirk et al. 2008). Additionally, data were filtered with respect to eight tree genera: Abies (minus Picea abies), Acer, Fagus, Fraxinus, Picea (minus Abies alba), Pinus, Quercus and Tilia.

Analysis of community ecology

Species abundance distribution

To avoid bumbling through dozens of different species abundance distribution (SAD) models (McGill et al. 2007), a simple hypothesis was presumed according to current discussions in community ecology (Coddington et al. 2009; Ulrich et al. 2010; Unterseher et al. 2011): the community of macrofungi in MV follows a log-normal distribution.

In case the fungal data did not follow a log-normal SAD, the core-satellite analysis (Hanski 1982) was applied to test for the possibility of overlapping SADs, e.g. a log-series pattern overlapping a log-normal SAD (Magurran and Henderson 2003). For this analysis, 20 years (1988–2007) were extracted from the dataset and analysed with respect to persistence (number of years in which a species was recorded) and abundance (the number of occurrences in a particular sample or year) of that species according to Magurran and Henderson (2003).

Different methods of separating core from satellite species exist (Magurran and Henderson 2003; Ulrich and Zalewski 2006; Galand et al. 2009; Dolan et al. 2009) because it is the organism's ecology and the nature of the data that deserve particular attention.The following assumptions should therefore account for the nature of fungi in core-satellite analysis: fungal species may be well established and abundant in their habitat but grow most of the time as invisible vegetative mycelium. Fruit body formation may occur rarely, e.g. every 5–10 years. If, for example, a 50% persistence threshold would be applied separating satellite taxa with a persistence of less than 10 years from core species rarely fruiting species would then be classified as satellite species by mistake and conclusions might depart from reality. To account for such phenomena, core species were defined as those species that were recorded with ≥75% of maximal persistence (more than 15 years) but down to 25% (15 quadrants of ordinance survey maps: MTB-Q) of maximal abundance for any year in MV. The remaining species were treated as satellite group and analysed accordingly. In order to assign possible surrogate species (Favreau et al. 2006), the core group was examined in detail.

Observed and estimated species richness

Randomised (rarefied) species accumulation curves (SAC) were calculated in the ‘vegan’ package (Oksanen et al. 2010) of the R environment (R Development Core Team 2011) to display the accumulation of species when the number of sites or individuals increases (Gotelli and Colwell 2001). By the analysis of the curves' shape (e.g. initial slope, approaching an asymptote or not), it was possible to evaluate basic patterns of species richness as well as sampling efforts.

The species richness estimators Chao1, Jackknife1 and Bootstrap (e.g. Colwell and Coddington 1994) were calculated in R. By analysing the estimator curves' shape, only those values were considered as serious extrapolations, that remained stable, i.e. whose curve showed a stable asymptote for a considerable part at the right end of the diagram. A fourth estimator was used for extrapolations over the whole of Germany. It was introduced by Ugland et al. (2003) and since then has also been used for fungal data (Unterseher et al. 2008).

Community analyses

Multivariate statistics, such as ordination, and an appropriate graphical display of results are of great importance to evaluate and explain any community structure in the light of natural fluctuations (i.e. species turnover between different habitats, host plants or ecosystems). As for the issue of species richness estimations, there exist different algorithms to analyse such multivariate (multidimensional) community data. All have their strengths and weaknesses (Gauch 1982; McCune and Grace 2002) whose description, however, is beyond the scope of this communiciation. In any case, it is recommended to apply different methods in parallel. If the results then show similar patterns, one can demonstrate and discuss the findings with more confidence than if only one calculation was done.

In the present study, community structures of macrofungi in MV were traced with the three ordination techniques: canonical correspondence analysis (CCA), detrended correspondence analysis (DCA) and non-metric multidimensional scaling (NMDS or NMS). Analyses were performed for those species that were recorded at least at ten different MTB-Q samples. Deleting the rarest species prior to ordination is a common procedure to strengthen the apparent differences among habitats by reducing noise from very infrequent species (McCune and Grace 2002). The raw data were not transformed or relativised, thus allowing differences in sample totals to be expressed in the analyses. Multivariate analysis included (1) detrended correspondence analysis (DCA) with default settings of the 'decorana' command of the R-package 'vegan'; (2) non-metric multidimensional scaling (NMDS) with the 'auttransform' option of command 'metaMDS' (also R-package 'vegan') set to FALSE; and (3) canonical correspondence analysis (CCA) with default settings of the 'cca' command and the four environmental variables: mycorrhiza, saprophytism, coniferous host and deciduous host.

All data used here (Online Resources 1 and 2) as well as analyses in the R environment (Online Resource 3) are provided as electronic supplementary material.

Results

Observed and predicted species richness

The total dataset used for this study comprised 65,535 geo-referenced observations (counts) with host tree and substratum characteristics. In total, 3,307 species were recorded from 1,019 MTB-Q. Database entries were dominated by saprobic fungi associated with deciduous trees (Fig. 2a). The overall number of saprobic fungi clearly outnumbered mycorrhizal fungi (Fig. 2a). Fagus, Quercus and Pinus trees harboured the highest absolute numbers of macrofungi, Tilia, Acer and Abies the lowest (Fig. 2b). With respect to the mean number of species per MTB-Q, Fagus was most species-rich with 2.3, followed by Tilia and Pinus (Table 1). A third value of observed species richness was calculated, the number of species per count. The higher those values are (Table 1), the faster fungal species accumulated with increasing number of counts. Here, Tilia is taking the lead with one new species every two counts, followed by Acer (0.4 species per count).

Fig. 2
figure 2

Species accumulation curves of macrofungi in Mecklenburg-Western Pomerania. Line numbes: 1 Fagus; 2 Quercus; 3 Pinus; 4 Picea; 5 Fraxinus; 6 Tilia; 7 Acer; 8 Abies

Full size image
Table 1 Basic data of macrofungal species richness in Mecklenburg-Western Pomerania
Full size table

Table 1 further displays values of three different species richness estimators and an assessment of the estimators confidence (rightmost two columns). With exception of the Chao1 estimator for mycorrhizal fungi on Picea, none of the estimators' curves displayed a stable levelling-off (Online Resource 3). The resulting equation of Ugland's estimator was y = 821.8 × log (x) – 2,401.31 with y = the number of predicted species and x = the number of MTB-Q (calculations not shown). For all MTB-Q, Ugland's formula estimated 3,291 species. Extrapolation to the entire forest area in MV resulted in 3,511 predicted woodland species.

Fungal communities

Canonical correspondence analysis (CCA) displayed species scores with respect to the parameters host group (deciduous-coniferous) and ecology (saprophyte-mycorrhiza) in a two-dimensional ordination space (Fig. 3). Table 2 lists taxonomic information, number of occurrences and preferred habitat for the five most distinct species of each community: deciduous-mycorrhiza (upper right of Fig. 3), deciduous-saprophyte (lower right), coniferous-saprophyte (lower left), coniferous-mycorrhiza (upper left). The five species closest to the centroid could not be assigned clearly to one of the four groups and are also considered in Table 2.

Fig. 3
figure 3

Position of macrofungi within a two-dimensional ordination space after canonical correspondence analysis (CCA). The first two axes account for 56 and 31% of the constrained variability. The ordination shows species (circles) in sample space according to substratum requirement (coniferous-deciduous) and nutrition strategy (mycorrhiza-saprophyte). The five most distinct species for each group (those nearest to the four corners of the ordination) are highlighted as black dots. Additionally the five species nearest to the centroid (without clear ecology) are highlighted. All 25 species are described in Table 2. Abbreviations are composed of the first four letters each of genus (upper case) and epithet (normal case), e.g. BOLEreti = Boletus reticulatus

Full size image
Table 2 Full species names, abundance and habitat requirements of those fungi highlighted in Fig. 3
Full size table

Using NMDS and DCA and focusing on host trees instead of fungal species, the fungal assemblage appeared tripartite (Fig. 4a, b): saprophytes of the five deciduous trees clearly separated from saprophytes of the three coniferous hosts. A third group consisted of mycorrhizal fungi of both coniferous and deciduous host trees (right hand side of Fig. 4a, b). Separate analysis of mycorrhizal communities (Fig. 4c, d) and saprobes (Fig. 4e, f) provided further details: For mycorrhizae, the eight host trees, representing their respective fungal assemblages, were clearly separated from each other (Fig. 4c, d). Deciduous trees and their associated saprobic fungi were overlapping for DCA on both axes (Fig. 4f), and separated clearly for NMDS (Fig. 4e). Coniferous host trees and their fungal communities separated clearly from each other and from those of deciduous trees. Saprobic fungi of Abies were the most distinct assemblage for both NMDS and DCA (Fig. 4e, f).

Fig. 4
figure 4

Ordination plots of nonmetrical multidimensional scaling (NMDS; upper row) and detrended correspondence analysis (DCA; lower row) display host trees (grey dots) in fungal species (black crosses) space. a, b Three distinct groupings of host trees and their associated fungal communities are visible. c, d Patterns from the mycorrhizal dataset. The trees' coordinates in ordination space are separated clearly. This means that each tree houses a unique community of mycorrhizal fungi. Several fungal species (black crosses) are located in between thus occurring on several hosts. e, f Coniferous trees and their fungal saprophytes separate from each other and from deciduous trees. Fungi without clear host preferences are located inbetween

Full size image

Core-satellite analysis

The 20-year dataset comprised 2,697 macrofungi. The core group comprised 100 taxa (4% of total species number and 25% of total records; Online Resource 1). The remaining 2,597 species (35,055 records) were treated as occasional species, i.e. satellite group (Online Resource 1). Visual inspection and statistical testing (Chi-square, Kolmogorov–Smirnov, Shapiro–Wilk and Anderson–Darling) for log-normal SAD of all data were negative, i.e. not significant (Fig. 5a, b; statistical tests not shown). Analyses of the core group revealed a significant approach to log-normality (Fig. 5c, d; statistical tests not shown). Analysis of satellite taxa resulted in rejection of log-normality (Fig. 5e) and confirmation of a log-linear relationship (Fig. 5f).

Fig. 5
figure 5

Species abundance distributions (SADs) of all fungi (left column), core species (middle column) and satellite group (right column) of a 20-year dataset. a, c, e ‘Binned’ species abundances overlaid with a fitted truncated log-normal SAD. Inserts show Whittaker plots (rank-log-abundance plots) which are better suited for identification of SAD other than log-normal. b, d The probability plots additionally test for log-normality of the data. The better the grey line in a probability plot fits to the plotted abundances (open circles) the better is the fit to a log-normal species abundance distribution. f Satellite taxa follow a log-linear SAD: only a few species had high numbers of counts whereas most fungi were rarely recorded

Full size image

Discussion

Observed and predicted species richness

Despite the long-term voluminous dataset used here, several aspects of fungal diversity could not be resolved. On the basis of the present data, it is hard to tell which host tree supports the richest fungal community (Fig. 1). Without doubt, beech (Fagus sylvatica) and oak (Quercus spp.) accommodate very high numbers of fungi (Heilmann-Clausen 2001; Heilmann-Clausen and Christensen 2005; Binion et al. 2008). However, the observed species richness on the different trees (Fig. 2a; Table 1) seemed to be rather a function of host tree abundance (column “area” in Table 1) than of the trees' ability to support high fungal diversity. By considering the number of species per count, other host trees became more important for maintenance of fungal diversity, such as Tilia (Unterseher et al. 2005) or Acer (Hein et al. 2009).

Estimations of total species richness ranged widely from 3,710 to 4,682 fungal woodland species (Table 1) and should therefore be communicated with caution. However, somewhat more than 4,000 macrofungal species (not only woodland species) in MV may be a realistic figure when compared with the recently published Funga of Saxony-Anhalt (Täglich 2009; 3,612 asco- and basidiomycota), or the Funga of Saxony that currently comprise 3,378 Basidiomycota and 1,593 Ascomycota from about 190,000 observations (Dämmrich, personal communication).

Ugland et al. (2003) provided an extrapolation method that could also be used with fungal communities (Unterseher et al. 2008), because their species richness estimator takes into account very diverse communities from large and heterogeneous environments. Whereas the chosen host trees represent dominant genera and 76.5% of total forest area in MV, extrapolation to 100% increased species numbers to ca. 3,500.

Fungal communities and the benefit of surrogate species for biodiversity and conservation

Multivariate statistics revealed distinct mycorrhizal communities of both coniferous and deciduous trees. Saprophytes seemed to behave in a more complex manner than mycorrhizal fungi. They showed clearly overlapping species composition on deciduous wood with DCA (Fig. 4f), but well-separated groups with NMDS (Fig. 4e). In both cases, deciduous hosts clearly separated from coniferous trees with respect to their fungal species composition (Fig. 4e, f).

Throughout various studies applying the core–satellite theory, core taxa were discussed as species that are biologically related to the habitat they live in (Magurran and Henderson 2003) or having the largest impact on their ecosystem (Ulrich and Zalewski 2006). Furthermore, changes in core species composition may be translatable to changes in environmental conditions (e.g. climate change, habitat disturbance), when new core species, drawn from the pool of satellite species, will replace existing ones (Magurran and Henderson 2003). Such interpretation of core species is close to the concept of surrogate species (Favreau et al. 2006) that comprises flagship (Dietz et al. 1994), focal (Lambeck 1997), indicator (Landres et al. 1988; Noss 1999) or keystone (Power et al. 1996) species. According to Rolstad et al. (2002), the biology of indicator species should be, among others, fairly well known. This would be the case for those four species groups in Fig. 3, which are located close to the four corners of the ordination. Except for 3 species, all those 20 fungi belonged to the core group (Online Resource 1).

Among the actual core group, Amanita muscaria, A. phalloides, Boletus edulis or Fomes fomentarius can without doubt be considered flagship species, given their charismatic appearance and their high level of public awareness that might support the protection of that species' entire habitat. Focal species are, e.g., those ectomycorrhizal taxa that are especially threatened by overfertilisation of forest soils (Gryndler and Lipavsky 1995) and deforestation (Zhang et al. 2004; Tedersoo et al. 2007). In order to protect focal species, such threatening processes should be eliminated so that further species will also benefit. The indicator species approach itself is complex and applicable in a hierarchical manner (Noss 1990). Fungal indicators can be assigned to plant communities and ecotypes (e.g. fungi associated with coniferous trees), ecosystem processes (e.g. litter and wood decay by the core group species Mycena galericulata, Auriscalpium vulgare, Fomitopsis pinicola, Hypoxylon fragiforme or Oudemansiella mucida) or to health assessment of major plant groups (e.g. Armillaria spp., Heterobasidion annosum). Additional to their significant role in forest ecosystems, all core species are common in MV, fruit regularly, are well known to most mycologists, and many of them are well represented in scientific literature (Kirk et al. 2008). They further represent all host trees and all ecologies considered in the present study.

Whereas more than 75% of total forest area in Mecklenburg-Western Pomerania were considered in the present study, numerous, from a fungal perspective, important host trees were neglected, such as Populus, Salix, Corylus, Larix, Pseudotsuga or the wooden rosids Malus, Sorbus and Prunus. These substrata surely provide rich sources for fungal species, and studies incorporating those trees will unravel further, so far unknown and unpredictable, patterns of fungal diversity.