Sampling was performed in November 2009 in three spruce forests of the biosphere area of Schwäbische Alb (Swabian Jura, Baden-Württemberg) of the German Biodiversity Exploratories framework [32]. Additional details on these biotopes are given by Fischer et al. [33]. Soils were collected underneath logs and from adjacent areas, approximately one meter apart, placed in sterile plastic bags and transferred to the laboratory for analysis. We analyzed soils underneath logs larger than 20 cm in diameter. Samples were taken at 0–10 cm soil depth. Leaf litter was removed before sampling in areas adjacent to wood logs. Characteristics of studied logs and basic abiotic soil parameters (5 cm depth) are provided in Table 1. Soil acidity (pH) was measured in 0.01 M CaCl₂ suspensions, according to the guidelines for soil description [34]. For each sampled wood log, stem length, stem diameter, remaining stem mass, as well as the type of rot (brown or white), were estimated (Table 1). Remaining mass of stem was estimated as proportion of decaying stem mass (volume * density) to the average wood weight of beech and spruce. In order to estimate stem mass, we measured the deadwood density by taking samples from the logs. The type of rot was determined based on fungal fruiting bodies growing on stems and by examining wood samples. Detailed description of methods to measure other abiotic soil properties and the respective results are given in Alvarez [35].

From each soil sample, 7 g was placed in 50 mL plastic tube, suspended (w/v) 1:5 in 35 mL of sterile 0.01% peptone-water solution and shaken on an orbital shaker at 200 rpm for 1 hour. Aliquots were taken from 1:5 suspension to produce two other step-wise dilutions, 1:10 and 1:20. All soil samples were analyzed in duplicates (two sub-samples) and each of the replicates was used to produce three dilutions (1:5, 1:10 and 1:20), each of which was plated in a duplicate again. An aliquot of 0.15 mL was plated on the surface of acidified glucose-yeast extract-peptone agar (GPYA) [8,26]. Plates were incubated at room temperature for 2–3 days and kept at lower temperatures (6–10 °C) to prevent fast development of molds. Plates were examined after 7, 14 and 21 d of incubation. Colonies were differentiated into macro-morphological types using dissection microscopy, counted and 1–2 representatives of each morphological type per plate were retained as a pure culture.

PCR-fingerprinting with microsatellite-specific oligonucleotides were used to group pure cultures. Strains showing identical electrophoretic profiles were considered as conspecific clones and only 1–2 representatives of them were chosen for further identification by sequencing of rDNA regions. Details on oligonucleotide primer sequences and PCR conditions are given by Yurkov et al. [8].

Yeast cultures were identified using nucleotide sequences of the D1/D2 domains of the large subunit rDNA (LSU) and the internal transcribed spacer region (or rDNA ITS). A combination of these two genetic markers provides the highest probability of correct identification of asco- and basidiomycetous fungi [31]. Protocols describing DNA extraction, amplification, purification and sequencing are given by Yurkov et al. [8]. For species identification, the nucleotide sequences were compared with sequences deposited in the NCBI [36] and MycoID [37] databases, respectively. Nucleotide sequences were deposited in GenBank under the accession numbers given in Table 2. Pairwise sequence comparisons were derived from alignments obtained with the MAFFT algorithm [38].

For each sub-sample, yeast quantity and community structure were determined. Yeast quantity was calculated as colony forming units (CFU) per gram of soil at natural humidity. Frequency of occurrence (incidence) was calculated as the number of samples, where a species was observed, as a proportion of the total number of samples. Relative abundance was calculated as proportion of a particular species in the sample and is based on colony counts (CFU).

A total of 72 sub-samples were included in the analysis, 36 sub-samples collected underneath wood logs and 36 sub-samples from adjacent areas. Statistical evaluations were performed with Statistica 8–9 (StatSoft Inc., United States). Quantity values were Log₁₀ transformed for the analyses. Normality of distribution was tested for absolute abundance values. Significant effects were confirmed with T-test. Differences in soil abiotic factors underneath logs and apart from them were assessed using the Wilcoxon signed-rank test (also referred to as Wilcoxon T-test) and the paired T-test. Effects were considered to be statistically significant at the level p < 0.05.

Beta-diversity was estimated using incidence- and abundance-based Jaccard similarity indices [39]. Similarity matrices were obtained using EstimateS 8.0 software [40]. Similarity matrices were analyzed with Multidimensional Scaling technique implemented in the Statistica software. The number of dimensions (axes) properly describing the observed similarity between samples was estimated on the basis of a scree plot, containing G-Star stress values.

Transient species were defined based on known ecological preferences (see Section 3) and from the analysis of Species Abundance Distribution (SAD) in the form of Rank-Abundance Diagram, or RAD [41]. Species evenness and productivity-evenness relationship in analyzed yeast communities were assessed on the basis of the distribution of species ranks in a community, RAD plots [41].

A total of 25 species were found, 13 underneath logs and 18 species adjacent to logs (Table 2). Three yeasts, TSN-67, TSN-36, and TSN-5, showed low similarity to any known species and may thus represent novel taxa (Table 2). Analysis of the large subunit (D1/D2 domains) suggested the placement of the isolate TSN-67 in the Cystofilobasidiales (Tremellomycetes, Agaricomycotina). The closest match among currently recognized species was obtained with Phaffia rhodozyma (Xanthophyllomyces dendrorhous), showing more than 50 substitutions in the D1/D2 domains of LSU rRNA. Isolate TSN-36 (=CBS 11767) showed no difference in LSU to the strain CBS 7170, previously misidentified as Candida atmosphaerica (AY574389). Phylogenetic analysis suggested the placement of this novel species near the Candida tanzawaensis clade [42]. The other novel ascomycete, represented by the isolate TSN-5 (=CBS 11766) was closely related to the Candida chilensis, showing 10 substitutions in LSU. Additionally, several isolates showed a few nucleotide substitutions to the type strains of the currently described species (Table 2).

The following six yeasts were common in soils collected underneath logs and in adjacent areas: Kazachstania piceae, Schizoblastosporion starkeyi-henricii (Saccharomycetetales, Saccharomycetes), Cryptococcus terricola (Filobasidiales, Tremellomycetes), Rhodotorula colostri (Sporidiobolales, Microbotryomycetes), Trichosporon dulcitum and Trichosporon porosum (Trichosporonales, Tremellomycetes). Cryptococcus podzolicus (Tremellales, Tremellomycetes) was frequent in soils underneath logs, but was absent in soils of adjacent areas. Additionally, several single isolates of pigmented yeasts, Aureobasidium pullulans (Dothideales, Dothideomycetes), Cr. tephrensis (Tremellales, Tremellomycetes), R. colostri and Sporobolomyces ruberrimus (Sporidiobolales, Microbotryomycetes) were found in soils collected underneath logs. In contrast, soils collected adjacent to logs were characterized by regular occurrence of yeasts, which were reported as inhabitants of plant surface [43,44,45,46,47]: Cr. filicatus, Cr. stepposus, Cr. victoriae (Tremellales, Tremellomycetes), Cystofilobasidium capitatum, Cy. macerans (Cystofilobasidiales, Tremellomycetes), R. colostri, R. fujisanensis (Microbotryomycetes) and Sp. ruberrimus. However, none of these species was observed in each of three studied plots.

Total species richness estimated per plot varied from 5 to 9 yeasts in soils underneath logs and from 6 to 11 in soils in adjacent areas (Figure 1). Average species richness values per sample were slightly higher in soils collected in adjacent areas (2.5–3.5 species) than underneath logs (2.0–2.5 species). Also, Shannon index values were significantly higher in soils adjacent to logs (ANOVA: F = 22.72; p < 0.001). Communities’ composition was strongly influenced by spatial factors, which resulted in observed dissimilarities between experimental plots. Three to four species were common for soil communities underneath logs, but the similarity between soil samples taken apart from the logs was always lower (Table 3). Specifically, five species were shared between AEW2 and AEW3 plots and one species only (Cr. terricola) was common for all three plots. Overall, Cr. terricola was the only species found in all soil samples underneath logs and in adjacent areas.

Up to 130 species have been reported from soils worldwide but evidence for strong association with soil-related substrata is still lacking for many of these species [7,28,48]. Unlike a few true soil-borne species (autochthonous soil yeasts) exclusively found in soils, transient yeast species are not restricted to this habitat but enter soil profile passively (e.g. with decaying leaves, fruits or mushrooms) or are transmitted to soils by animals (see, for example, [47,49]). Soil yeast communities underneath logs consisted to great extent of soil-borne yeasts and a few transient (phylloplane-related) species. Specifically, yeasts A. pullulans, Cr. tephrensis, R. colostri and Sp. ruberrimus were found in low numbers in a single replicate (Table 3). In total, relative abundance of these latter species did not exceed 4% (Table 4 and Figure 2) underneath logs. In contrast, communities of adjacent areas contained up to 30% (relative abundance) of phylloplane-related yeasts, which, therefore, contributed substantially to the total yeast quantity.

Analysis of communities’ structure also revealed striking differences between soils influenced by decaying logs and ones of adjacent areas (Figure 3). For example, Cr. podzolicus was the most frequent and abundant species underneath logs, but it was absent in soils collected apart from logs (Table 3). In contrast, the relative abundance of Cr. terricola was slightly higher in soils apart from logs, where it was clearly dominant (ANOVA: F = 5.48; p = 0.02). In contrast, K. piceae and T. porosum were significantly more abundant in soil underneath logs (ANOVA: F = 5.17; p = 0.02 and F = 5.58; p = 0.02, respectively). In contrast to species clearly reacting on the deposition of logs in all plots, yeasts K. piceae and T. dulcitum showed a plot-specific pattern being more abundant at the AEW3 plot than at AEW1 and AEW2 (Table 3 and Figure 3). Relative abundance Cr. podzolicus showed a contrasting trend declining at the AEW3 plot (Table 3 and Figure 3). In soils from adjacent areas, Cr. musci and Schizoblastosporion starkeyi-henricii were more abundant at the AEW1 plot (Table 3), whereas relative abundance of K. piceae and Cr. terricola were significantly higher at the AEW3 plot (Table 3 and Figure 3).

Analyses of communities’ similarity using incidence- and abundance-based Jaccard similarity coefficient showed a low similarity level between all yeast communities (Figure 4). In all pair-wise comparisons, incidence-based Jaccard coefficient values did not exceed 0.30 (mean values 0.28 and 0.24 for soils under logs and from adjacent areas, respectively). Interestingly, similarity within a single plot, between soils collected underneath and apart from wood logs ranged from 0.20 to 0.29. Application of the abundance-based Jaccard coefficient showed that yeast communities in soils collected underneath logs were more similar in average than the ones of adjacent areas, 0.63 and 0.42, respectively (Figure 4). Additionally, the ordination of these results using the multidimensional scaling technique clearly displayed a separation between soil yeast communities underneath logs and of adjacent areas (Figure 5).

Total yeast quantity varied largely from 2.4 × 10³ to 9.7 × 10⁵ CFU/g or 3.38 to 5.99 Log₁₀ (CFU/g), respectively (Table 4 and Figure S1). Average quantity in soils collected in adjacent areas was slightly higher than in soils underneath logs, 5.24 and 5.10 Log₁₀ (CFU/g), respectively (ANOVA: F = 17.67; p < 0.001). However, when colony counts corresponding to phylloplane-related yeasts were not considered (Figure 4), quantity of soil-borne species appeared to be in the same range in both soils, 5.15 and 5.16 Log₁₀ (CFU/g), respectively (Table 4).

Temperature, annual precipitation, pH and carbon content are the most frequent soil parameters, traditionally provided to explain distribution of soil yeasts (see, for example, [7,21,28]). Two studies that analyzed yeast distribution along a large latitudinal gradient revealed a positive correlation between high rainfall values, low soil pH, and occurrence of Cr. podzolicus [27,28]. However, none of these parameters explain the association, in our study, of Cr. podzolicus with soils underneath decaying logs (Table 1 and Figure S2). This disagreement might be explained by the fact that both pH and rainfall do not impact directly environmental factors determining the ecological optimum of this yeast. According to the original description, Cryptococcus podzolicus (basionym Candida podzolica) is associated with soddy-podzolic soils or podzols (according to FAO classification) [53]. Formation of this type of soil relies both on high raw organic matter income and on washing water regime [54]. However, rainfall values do not necessarily correlate with the type of soil water regime, as the latter strongly depends on physical field properties such as soil structure, clay content, ground water table and parent rock. Similarly, soil acidity (often referred to as pH) is the result of two distinct processes: decomposition of organic matter, and parent rock weathering [54]. In podzols, decomposing plant material produces a large amount of organic acids and aromatic compounds that are readily utilized by soil microbes, including yeasts. In contrast, acid acrisols and ferralsols contain less labile carbon, and their highly weathered parent rock releases toxic metal ions such as aluminum and iron [54,55]. Even though all these acid soils may have pH in the same range, environmental conditions for development of soil microorganisms in such habitats are quite different [54,55,56].

Our results showed positive effects of decaying logs on abundance of Cr. podzolicus, thereby suggesting the importance of dissolved (leaching) carbon compounds rather than soil pH for development of this soil yeast. Therefore, the distribution range of Cr. podzolicus might be significantly larger and include different acid well-drained soils worldwide, like podzols, andosols, umbrisols, acrisols and ferrasols. Even though the soil seems to be the primary habitat for this species, its occurrence in other substrates with similar environmental properties is feasible. For example, Cr. podzolicus was isolated from Sphagnum moss and peat [57,58], habitats that are also characterized by high moisture, acidity, and raw organic matter content. Although in some cases the distribution range of Cr. podzolicus may correlate with high water content in a substrate (washing water regime in case of soils), temperature, pH, or rainfall data alone are unlikely to be able to explain the distribution of this soil yeast.

Many different ascomycetous and basidiomycetous yeasts were reported from soils but just a few of them were found exclusively in this substrate [7,8,28,51]. Other species, which could be also detected in soils, are entering soil profile with animal activity or annual leaf fall [47]. In particular, leaf fall may contribute substantially to the composition of soil yeast communities because phylloplane contains an abundant yeast population reaching millions of cells per gram [43]. In the present study, typical phylloplane-related yeasts were numerous and abundant in soils collected from adjacent areas that are exposed to the annual litter fall (Figure 2). These transient (or allochthonous) species substantially affected estimations of yeast diversity values, such as species richness and Shannon diversity index (Figure 1). In contrast, yeast communities underneath logs contained only three so far phylloplane-associated species, viz. Aureobasidium pullulans, Cr. tephrensis and Rhodotorula colostri. Three more ascomycetous yeast species (retrieved as single isolates) could represent transient species, as well. They are closely related to yeasts that were previously isolated from decaying material. Specifically, strain TSN-95 was identified as Candida cretensis, which was originally isolated from a rotten polypore Inonotus tamaricis [59]. Another yeast (strain TSN-36) is prospectively conspecific to the strain Candida sp. CBS 7170, which was isolated from a fungal fruiting body of Tyromyces ptychogaster growing on a fallen spruce trunk (CBS database). Lastly, strain TSN-5 was closely related to C. chilensis, a species described from rotting Nothofagus wood in Chile [60]. All these yeasts were observed as single colonies during plating experiments that utilized 1:5 and 1:10 soil-to-water dilutions. Because these species were not detected from soil suspension with a lower soil-to-water ratio, we assume that these three ascomycetous yeasts are present in soils in very low quantities. They might originate from aboveground sources such as decaying logs and fungal fruiting bodies growing on fallen stems or trunks [8,13,42,59,60]. Overall, our results suggest that unlike phylloplane-related yeasts, yeasts inhabiting decaying logs provide minor contribution to soil communities.

While influence of transient species on communities’ parameters in animal ecology has been considered [41], the division of microbes into resident and transient species is rare and has little application in microbial ecology. In the 1970s, extensive analysis of species abundance distribution patterns (SAD) along gradients by animal and plant ecologists resulted in revealing an empirical pattern of increasing community evenness with ecosystem productivity [41]. We found that the extrapolation of results based on species distribution patterns (abundance and incidence) in soils might be biased by the substantial presence of transient species (comp. Figure 2). This may also lead to incorrect conclusions regarding ecosystem properties. Both species richness and the Shannon diversity index suggested higher diversity in soils of adjacent areas in our study. Furthermore, we found that distribution of taxa (ranks) in Rank-Abundance diagram (RAD) was more even (flat) in soils collected apart from tree logs (Figure 6 A). This leads to a contradictory conclusion that without any regard to the soluble organic carbon leaching from decaying wood (Table 1), soils underneath logs are less productive and, therefore, harbor less diverse yeast communities. It is worth mentioning that yeasts entering soil with litter develop their biomass by consuming plant exudates and mainly retain their activity on fresh fallen leaves during early stages of succession [47,48,61]. These yeasts do not necessarily rely on the nutrients available in soil and, therefore, do not explain differences in productivity between soils collected underneath logs and in adjacent areas. Remarkably, these yeast species were rarely detected at the same plots in soils collected in spring [8]. After removal of transient (phylloplane-related) species from the analysis, the structure of soil yeast community underneath logs displayed higher evenness than communities of adjacent areas (Figure 6 B). Therefore, our results show that external carbon originating from decaying logs supports higher productivity of soil-borne yeast species.

It has been repeatedly pointed out that proper identification of fungi using molecular markers is important for microbial biodiversity assessments. Our results additionally suggest that correct attribution of resident and transient species is crucial for microbial diversity assays, both culture-dependent and culture-independent ones. Importantly, detection of fungal taxa with culture-independent techniques does not necessarily require the presence of living cells and is often possible from resting stages, such as spores [62]. This may also bias estimations of sampling depth. It has been shown that completely surveyed communities usually follow lognormal distribution, whereas undersampled communities are better described with power law functions [41]. Recent analysis of fungal communities in the rhizosphere and on plant surfaces showed the presence of core and satellite species (molecular operational taxonomic units, MOTUs) groups [63]. While species from the core group followed lognormal distribution, species from the satellite group were log-series. At the same time, a few studies have shown that the resident species of fish and beetles were lognormal, but the transient species were log-series [41]. We would like to emphasize that additional studies (also using culture-dependent approaches) are required to resolve the ecology of rare (or satellite) fungal species (and MOTUs) in order to find out whether or not they represent transient community members.

Yeast populations in soils collected from adjacent areas was strongly dominated by a single species, Cr. terricola, which contributed with 40%–60% to the total yeast abundance. In soils collected underneath logs, relative abundance of several soil-related yeasts, Cr. podzolicus, T. dulcitum and T. porosum, was higher than in soils from adjacent areas (Table 3 and Figure 3). These species are known to be able to sustain low pH and their occurrence in acid soils has been reported previously [10,21,28,48]. Besides the tolerance to soil acidification with soluble organic carbon substances, good development of these yeasts in soils underneath logs could be additionally explained by their assimilation profiles, i.e., abilities to assimilate intermediates of wood decomposition such as complex polysaccharides, organic acids and aromatic compounds [9,11,13]. For example, strains of Cr. podzolicus and T. porosum isolated from decomposing wood were found to be the most active in degradation of plant-related carbohydrates [13]. In contrast, the members of the Cystofilobasidiales and Tremellales (Tremellomycetes), frequently detected in phylloplane and litter [43,47,48,61] either exhibited a reduced capacity to metabolize aromatic compounds, or did not utilize them at all [14].

Even though the concentration of dissolved organic carbon in topsoil under logs was higher than in adjacent areas (Table 1), we did not observe any pronounced correlation of this factor with yeast quantity, species richness, abundance of soil-related yeasts, and relative abundance of Cr. podzolicus (data not shown). Indeed, log deposition had little effect on total yeast quantity and species richness values (Table 4 and Figure 1, Figure S1). However, it caused substantial alteration of yeast communities’ structure and composition (Table 3 and Figure 2, Figure 3). Although detailed analysis of environmental factors associated with wood decay was not in the scope of this study, our observations imply that selection towards well-adapted soil-borne species is driven by the same mechanisms in all analyzed plots. It has been repeatedly pointed out that yeasts in soils are distributed unevenly [7,8]. However, analysis of spatial variability using abundance-based Jaccard similarity coefficient showed that yeast communities formed underneath logs are more similar than the ones in adjacent areas (Figure 4). Additionally, ordination of this data displayed larger between-plot heterogeneity (distance) of yeast communities in soils collected apart from logs (Figure 5). This, in our opinion, suggests that aboveground deadwood deposition establishes selective (but stable) conditions in soils that promote growth of a few highly specialized soil-borne yeasts. Nevertheless, additional studies are required to find out which substances influenced soil yeasts and, therefore, are responsible for the observed effects. Besides altering belowground microbial communities with soluble organic carbon, large wood logs cover soil surface and thereby hamper the proliferation of plant-related yeasts in the soil profile.