• Phialocephala fortinii sensu lato was isolated from Picea abies roots that had been collected from the same 3 × 3-m forest plot in 2001 and 2004, to examine the spatial and temporal dynamics of this fungal community.
• RFLP analysis was used to define the multilocus haplotype (MLH) of each isolate. Pielou's measure of association and χ2 tests of independence were employed to examine the randomness of patterns of spatial distribution of MLH observed in 2001 and 2004. Population differentiation between the two samplings was tested using the GST statistic.
• In 2001, 144 strains of 28 MLH were isolated; in 2004, 139 strains of 29 MLH. Abundant MLH in 2001 also prevailed in 2004, and the same two cryptic species were dominant in both collections. The probability of being isolated in both years increased with increasing frequency of an MLH. The patterns of spatial distribution of most MLH did not differ between years. The GST values indicated identity of the two collections.
• Communities of P. fortinii sensu lato remain spatially and genetically stable for at least 3 yr.
Dark septate endophytes (DSE) are among the most widely distributed fungal endophytes of plant roots, and have been isolated from >600 plant species to date (Jumpponen & Trappe, 1998). Dark septate endophytes are characterized by their darkly pigmented and septate mycelia which allow them to be distinguished from members of the Glomeromycota known to form endomycorrhizae, and from other endophytes with hyaline hyphae. Phialocephala fortinii sensu lato is the dominant dark septate endophyte in roots of species belonging to the Pinaceae, but is also known to colonize roots of broad-leaved trees, shrubs and herbaceous plants (Stoyke et al., 1992; Ahlich & Sieber, 1996; Ahlich-Schlegel, 1997; Sieber, 2002). Recently, P. fortinii sensu lato was shown to form ectomycorrhiza with a Populus tremula × Populus tremuloides clone (Kaldorf et al., 2004). New reports suggest that P. fortinii sensu lato is not restricted to fine roots of higher plants. Menkis et al. (2004) were able to isolate P. fortinii from (living) stem bases and stumps of Pinus sylvestris, Betula pendula and Picea abies. Furthermore, Jumpponen et al. (2003) reported P. fortinii sensu lato from the rhizoid environment of the liverwort Cephaloziella varians in Antarctica, showing the broad distribution of these ubiquitous species. Recent population genetic studies in Europe suggest that P. fortinii sensu lato is composed of several cryptic species (CSP) that occur sympatrically in the same forest site and even in the same root fragment (Grünig et al., 2004; Sieber & Grünig, 2005). Thus the species name carries the addendum ‘sensu lato’ (s.l.). Sieber & Grünig (2005) estimated that at least eight CSP of P. fortinii s.l. may be found in Europe, based on single-copy restriction fragment length polymorphism (RFLP) data. One was recently described as Acephala applanata (Grünig & Sieber, 2005). The existence of several CSP in P. fortinii s.l. was also suggested for North America, based on amplified fragment length polymorphism (AFLP) fingerprints (Piercey et al., 2004). However, the low number of strains included in that study did not allow a proper analysis of species boundaries. Whether the CSP found in Europe and Canada are conspecific has not been investigated to date. As P. fortinii s.l. is a species complex, we use the terms ‘population’ for genets belonging to the same cryptic species; and ‘community’ if more than one CSP is concerned.
Study site, sampling design and isolation of dark septate endophytes
The study was performed near Zürich in a P. abies stand that had been planted in 1969 after a windthrow. The plants used for this artificial regeneration originated from a local nursery. The former stand was composed of P. abies and Abies alba planted in c. 1840 and replaced a coppice-with-standards, a typical and widespread historical forest structure in Europe (Etter, 1892). In this former stand P. abies, A. alba and Fagus sylvatica plants were of unknown origin, and were regularly planted under canopy cover.
A hierarchical sampling design was chosen. A 3 × 3-m square was subdivided with a 1 × 1-m grid, and four of the 1 × 1-m squares were further subdivided with a 0.5 × 0.5-m grid, as described previously (Grünig et al., 2002). A total of 32 grid points were sampled in 2001 and 2004, respectively. Five root segments of P. abies were chosen arbitrarily per grid point and year, and DSE were isolated as described previously (Grünig et al., 2002). To assess the efficacy of the sampling, 10 instead of five root segments were sampled in each of three grid points in both years. In total, 144 and 139 strains were collected in 2001 and 2004, respectively (Table 1).
Table 1. Comparison of number of isolates and multilocus haplotypes (MLH) found in the first and second samplings
Number of strains with complete RFLP data sets indicated in brackets.
When a weak demarcation line formed between mycelia that emerged from the same 5-mm-long root segment in the 2004 collection, single hyphal tip cultures were prepared from both mycelia to test whether such mycelial heterogeneities rely on the interaction of different genotypes or CSP.
DNA extraction, Southern blotting and single-copy RFLP analysis
Strains were grown in Erlenmeyer flasks containing 50 ml 2% (w/v) malt extract broth for 14 d at 20°C. Mycelia were harvested by filtration and lyophilized. Approximately 50–80 mg lyophilized mycelium was ground and DNA extracted as described previously (Grünig et al., 2003). DNA quality and concentration were estimated by gel electrophoresis (0.8% agarose gels in 0.5 × TBE at 2.5 V cm−1 for 1 h, stained in a 10 µg ml−1 ethidium bromide bath, and visualized by UV fluorescence). Strains from the 2004 collection were analysed using 11 single-copy RFLP probes which are available from our laboratory (Grünig et al., 2003), and were compared with the data set from the first collection (Grünig et al., 2004). Definitions of single-copy RFLP alleles are given by Grünig (2004). In brief, fungal DNA (5 µg) was digested with 50 U of the restriction enzyme HindIII and digests were separated on 0.8% agarose 1 × TBE gels for 19 h (2.4 V cm−1). After partial hydrolysis in 0.25 M HCl for 15 min, DNA was blotted onto Hybond-N+ membranes (Amersham, Dübendorf, Switzerland) using the capillary transfer method under alkaline conditions (Sambrook & Russell, 2001).
Probes were labelled with [α-32P] dCTP using a nick translation kit (Invitrogen, Basel, Switzerland) following the manufacturer's instructions. Blots were prehybridized for 5–8 h at 65°C and hybridized overnight at 65°C, before being subjected to washes twice with 1 × SSC, 0.1% SDS; twice with 0.2 × SSC, 0.1% SDS; and twice with 0.1 × SSC, 0.1% SDS. Membranes were exposed to X-ray film (Kodak BioMax MS) for 24–72 h at −80°C.
RFLP data were scored based on the presence of RFLP fragments with similar sequence but different size, and each fragment of a particular size was assumed to represent a specific allele at a single genetic locus.
Isolates having the same MLH were assumed to be individual members of the same clone, and cluster analysis was therefore performed using clone-corrected data sets (Grünig et al., 2003).
Reisolation frequency of identical multilocus haplotypes
Multilocus haplotypes were assigned to one of four categories according to the number of isolates per MLH (1, 2–5, 6–10 or >10). If an MLH was found in both collections, the collection in which this MLH was more abundant was responsible for assignment to one of the categories. Reisolation frequency was calculated for each category as the number of reisolated MLH divided by the number of observed MLH in that category, multiplied by 100.
Assuming uniform distribution, the proportion of root segments colonized by a defined MLH in the 2001 sampling was used as an estimate of the probability π of detecting this MLH in a root segment; π was then used to calculate the probability P(X = x) of isolating this MLH from x = 0, 1, 2, 3, 4 or 5 segments of the n = 5 segments examined per grid point. P(X = x) follows a binomial distribution and can be calculated according to:
(eqn 1 )
The probability P(X ≥ 1) of isolating the MLH at least once at a grid point can be determined by setting x = 0 in equation 1 and deducing the probability of not finding the MLH at a grid point:
( eqn 2 )
In addition, the expected value (expected number of segments colonized by the MLH at each grid point) and its variance were determined:
(eqn 3 )
(eqn 4 )
The expected value was compared with the observed value of the 2004 sampling and tested for significant differences using the z test (Stahel, 2002).
Spatial distribution of reisolated MLH
A map was prepared to display the spatial distribution of isolates according to their MLH. Then a contingency table test for independence of the two samplings was performed for each MLH found in both collections (Pielou, 1977). In addition, a measure of association was calculated as Q = (ad −bc)/(ad + bc) where a = frequency of points where the MLH was collected at both samplings; b = frequency of points where the MLH was collected only at the first sampling; c = frequency of points where the MLH was collected only at the second sampling; and d = frequency of points where the MLH was not present (Pielou, 1977).
Population genetic analysis
To test the genetic stability of the two collections GST values for population differentiation, as estimated by θ (Weir, 1996), were calculated between the two collections, and between CSP of the two collections which were represented by more than eight MLH. Significance of GST values was tested by comparing the observed GST with 1000 randomized data sets, as implemented in the software multilocus 1.3 (Agapow & Burt, 2001).
Associations among loci within CSP were analysed for both collections with multilocus 1.3 using the index of association (IA). The significance of IA was tested by comparing the observed IA with the distribution of IA expected under the null hypothesis of random mating from 1000 randomized data sets.
In total, 144 strains and 139 strains of P. fortinii s.l. were isolated from the 3 × 3-m plot in 2001 and 2004, respectively. RFLP analysis was successful for 135 strains in 2001 and 136 strains in 2004 (Table 1). These represented 43 MLH, most of which belonged to one of three CSP described by Grünig et al. (2004) (Fig. 1). The full data sets for both collections are given in Appendix S1, available online as supplementary material.
The number of strains collected per CSP differed slightly between the two samplings. Cryptic species 1 (CSP1) was most abundant, but its frequency of isolation decreased from 77 to 65% between the two samplings (Table 1). Correspondingly, other CSP were isolated slightly more frequently in the 2004 sampling. Fourteen of the 43 MLH were detected in both samplings (Fig. 1). There was a clear tendency for MLH represented by many isolates to have a higher chance of being reisolated in the second sampling (Fig. 2). Fifteen MLH had been isolated from just one root segment in the 2001 sampling, and only three (20%) were reisolated in 2004. Similarly, only two (17%) of the 12 MLH isolated just once in 2004 had already been found in 2001. The two most abundant MLH from the first collection (UeB-05 and UeB-01) were also the most abundant MLH in the second collection. Whereas the number of isolates for UeB-05 was slightly higher for the second sampling, with 32 compared with 28 isolates, the number of isolates for UeB-01 was reduced from 42 in the first collection to 29 isolates in the second collection. However, the observed mean number of isolates per grid point in 2004 did not deviate significantly from the number of isolates expected based on the 2001 data for either of the two MLH (Table 2).
Table 2. Comparison of estimated and observed number and proportion of isolations of the two most abundant multilocus haplotypes (MLH)
Probability of isolation from ≥1 segment per grid point*
Observed proportion of grid points where MLH isolated from ≥1 segment
The spatial distribution and number of isolates per sampling of the 14 resampled MLH are presented in Fig. 3. Most resampled MLH were sampled more often at the same grid point(s) than expected, using the contingency table test of Pielou (1977) (Table 3). Only four MLH showed a negative association (Q): these MLH were isolated more often at different than at the same grid points in the two samplings (UeB-08, UeB-07, UeB-24, UeB-01) (Table 3). However, association Q of Pielou (1977) does not account for the spatial pattern of presence/absence of an MLH, thus Q values for some MLH, especially rare ones, must be interpreted with caution. For example, UeB-07 appears to be restricted to the left side of the study plot (Fig. 3): the two samplings showed a rather positive spatial arrangement for this MLH. However, although UeB-07 was isolated at spatially close, but not identical, grid points in 2001 and 2004, association Q of the two samplings was negative.
Table 3. χ2 test for independence and measure of association (Q) of the two collections regarding the occurrence of each of 14 resampled multilocus haplotypes (MLH) in the plot (Pielou, 1977)
Measure of association Q
, Significant at P ≤ 0.05; ns, not significant; 1 ≥Q ≥−1, 1 = MLH was isolated at exactly the same grid points in both years; −1 = MLH was never isolated at the same grid point in the two samplings; 0 = two samplings were independent (distribution of MLH was random in both years).
Population subdivision and multilocus disequilibrium
Neither the GST value for the complete data set, nor the GST values for two cryptic species (CSP1 and 3) with more than eight MLH, differed significantly between the two samplings (Table 4). GST values were very small, ranging from 0 to 0.01. Nevertheless, some novel alleles and consequently MLH were found for CSP1 and 3 in the second sampling.
Table 4. Population differentiation (GST) between communities of Phialocephala fortinii s.l. collected in 2001 and 2004, and between populations of cryptic species (CSP) represented by at least eight multilocus haplotypes (MLH) in each collection
Number of MLH
The index of association (IA) did not deviate significantly from zero for the two CSP when analysing either the data sets from each collection separately, or the combined and clone-corrected data sets (Table 5).
Table 5. Index of association (IA) for cryptic species (CSP) represented by more than eight multilocus haplotypes (MLH)
Number of MLH
Sampling in detail and heterogeneities within single roots
The sampling of 10 instead of five root segments resulted in the detection of new MLH (Table 6). Up to six MLH were found per grid point in both the 2001 and 2004 samplings. If MLH of both samplings were combined per grid point, up to nine MLH were detected. Three MLH were found, on average, if only five segments were examined per grid point; but 5.3 MLH were found if 10 segments were studied. From each of six root segments, two mycelia emerged that were separated by a demarcation zone. Single hyphal tip cultures were prepared from each of these mycelia to test whether such ‘incompatibilities’ rely on the interaction of different genotypes or even CSP. In four of the six cases, two different CSP were involved, and these must have occurred in the same 5-mm-long root segment. In one case the mycelial heterogeneities were the result of two genetically different strains of the same CSP, and in another case no genetic differences were obvious, based on RFLP and inter-simple-sequence-repeat-anchored polymerase chain reaction (ISSR–PCR) data, respectively.
Table 6. Analysis of number of multilocus haplotypes (MLH) found in two collections from grid points where 10 instead of five root segments were collected
Number of isolates
Number of MLH
In both samplings
Two single hyphal tip cultures were prepared from one 5-mm-long root segment.
The spatial and temporal dynamics of a community of the tree-root endophyte P. fortinii s.l. was studied in a small forest plot. The community was stable with regard to the total number of isolates and MLH found in both collections; the genetic composition; and the spatial distribution of most MLH.
Phialocephala fortinii s.l. was the dominant root endophyte in both collections. Only 16 strains in the first collection and four strains in the second belonged to other species, including Cylindrocarpon sp., Geniculosporium spp., Mortierella sp., Oidiodendron griseum, Chalara sp., and various sterile mycelia. The number of MLH of P. fortinii s.l. detected per unit area was high compared with that of mycorrhizae that inhabit a similar ecological niche (Dahlberg & Stenlid, 1995; Gryta et al., 1997; Gherbi et al., 1999). The relationship between number of isolates per MLH and number of MLH followed a hyperbolic distribution in both samplings, as has been observed for other fungal populations including Sclerotinia sclerotiorum (Kohn, 1994; Kohli et al., 1995), Cryphonectria parasitica (Milgroom & Cortesi, 1999) and Phytophthora infestans (Grünwald et al., 2001). The actual frequency distribution of strains per MLH could be explained by several hypotheses, including recombination (Hartl & Clark, 1989); selection for the more abundant MLH (Kohli et al., 1995); or the stand history of the study site including the introduction of less abundant genotypes by planting (Grünig et al., 2002).
More abundant MLH had a higher chance of being reisolated in the second sampling (Fig. 2). This was expected for stable populations not subjected to seasonal bottlenecks, such as populations in agricultural ecosystems (Wolfe & McDermott, 1994). In agriculture, populations of most pathogens start from sexually produced spores each year, mostly because of the biology of these pathogens. Other reasons are crop rotation and harvest of the host (McDermott & McDonald, 1993; Chen et al., 1994; Lamour & Hausbeck, 2001a, 2001b). Therefore resampling of identical genotypes has not been reported for such pathosystems. However, in S. sclerotiorum asexually produced propagules are responsible for the establishment of new populations, and resampling of clones, even in fields over 450 km apart, has been reported (Kohn, 1994; Kohli et al., 1995). In contrast, continuous vegetative growth is possible, and persistence of mycorrhizal mycelia has been reported several times in forest ecosystems (Dahlberg & Stenlid, 1995; Gherbi et al., 1999; Gryta et al., 2000).
The probabilities of isolating the two dominant MLH in at least one of the five segments examined per grid point was high (Table 2). Of course, the probability of finding an MLH can be increased by increasing the sample size (Table 6). For example, the probability of finding MLH UeB-01 (UeB-Res-22) at least once at each grid point could be increased to 95% by the examination of 12 instead of only five segments per grid point. For abundant MLH (π ≈ 0.2, 20%), examination of five root segments per grid point is good enough to detect the spatial distribution and dynamics reliably. The story is different for rarely isolated MLH. The estimated π was 0.005 (0.5%) for MLH isolated from only one root segment in the whole plot. To detect a uniformly distributed MLH with π = 0.005 with a 95% probability in at least one segment, 530 root segments per grid point would have to be examined. This is not feasible. Even if a 50% probability of finding the MLH were considered good enough, 125 segments per grid point would still be needed. However, if we assume that rare MLH are not uniformly distributed in the study site, but are spatially restricted to one or only a few grid points, the probability of reisolating them at such a grid point is much higher. Supporting this view, we have reisolated rare MLH at the same or adjacent grid points, for example UeB-30 (UeB-Res-05), UeB-27 (UeB-Res-01) and UeB-17 (UeB-Res-16). In addition, the medium used to isolate the fungi might have been selective to the advantage of the abundant MLH: rarely isolated MLH were possibly more abundant in reality, but were not competitive enough to emerge from the segments if another MLH was present.
Most resampled MLH showed a positive measurement of association (Q), being detected in the same part of the plot in both years, indicating temporal and spatial stability. The two most abundant MLH, UeB-01 (UeB-Res-22) and UeB-05 (UeB-Res-20), occurred everywhere in the plot and were stable for at least 3 yr. In order to examine the spatial and temporal dynamics of these MLH, the limits of these clones needs to be known, thus the size of the study plot needs to be enlarged. In addition to the measure of association, Q, approaches similar to those described by Milgroom et al. (1991) and Kohli et al. (1995) were tested to quantify the spatial aggregation. Average Euclidian distances between points where the MLH was isolated, or the average Euclidian distance to the nearest point where the MLH was isolated, were calculated. These average distances were then compared with the average Euclidian distances within 500 artificially generated data sets (a program based on visual basic is available from the authors). We also tried to calculate the index of dispersion according to Morisita (1959). Unfortunately, the above tests were not suitable for our data set (data not shown). Nevertheless, results from our analysis indicate that a spatial correlation exists for most of the resampled MLH. These results must be interpreted cautiously because the exact spatial distribution of MLH is difficult to assess, for various reasons, including: (i) possible competition among MLH; (ii) selectivity of media used for isolation; and (iii) limited resources to allow statistically appropriate sampling sizes (Grünig et al., 2002). Spatial correlation of genets was also reported by Gryta et al. (2000), who analysed the fine-scale population structure of two Hebeloma cylindrosporum genets over 4 yr based on sporocarp samples. The two genets occupied nonoverlapping territories, and extended their territories by a rate of c. 0.45–0.6 m yr−1. Genotypes of Laccaria amethystina were resampled in the study of Gherbi et al. (1999). However, only eight genotypes were detected again after 3 yr. They concluded that populations of L.. amethystina maintain a genetic structure more consistent with a high frequency of sexual reproduction. In a resampling of a P. fortinii s.l. population at a glacier forefront after 1 yr, none of the RAPD phenotypes was detected in both years (Jumpponen, 1999). Our results clearly deviate from those of Jumpponen (1999): frequently occurring genets appeared to be quite stable in space and time in the present study.
According to the GST statistic, populations of both samplings were nearly identical, and populations in the second sampling represent well the populations in the first sampling. Temporal stability of populations has been reported regularly from other fungal species, especially plant pathogens (Chen et al., 1994; Lamour & Hausbeck, 2001b). However, in these cases the new populations were established from sexually or asexually produced propagules. The index of association (IA) did not deviate from zero, confirming the findings of Grünig et al. (2004) that recombination occurs, or had occurred, in the CSP of P. fortinii.
Mycelial heterogeneities indicate the presence of more than one genotype per root segment. In five out of six cases, different MLH were isolated that belonged to different CSP in four cases. Isolation of different genotypes from the same root was also observed by Piercey et al. (2004). This is further supported by the fine-scale analysis of P. fortinii s.l. in a 20-cm-long root using ISSR–PCR and single-copy RFLP. Six ISSR phenotypes belonging to several CSP were found among 19 strains of this root (Sieber & Grünig, 2005). However, other roots similar in length and collected in the same forest stand were colonized by a single ISSR phenotype (N. Nuessli, unpublished data). To avoid a mixture of genotypes, it is therefore essential to prepare single hyphal tip cultures carefully. This may be crucial if dominantly inherited markers are to be used for population analysis.
We have shown that abundant and possibly also some rare genotypes of P. fortinii s.l. are stable in space and time, at least on a small plot. However, this study opens up many new questions: What is the clone size of abundant genotypes? What happens at the edge of clones? Do clones compete, and/or do they exchange genetic material? Does recombination occur in the interaction zone between sexually compatible clones – do newly combined genotypes appear in this zone? Is the absence of some genotypes in one year only a stochastic effect, or the result of mortality and colonization by propagules dispersed by vectors (including man) or wind? It will be fascinating to find answers to these questions to broaden our knowledge of this interesting and ecologically important fungal species complex.
We would like to thank Angelo Duò, Swiss Federal Institute of Technology, Zürich, Switzerland for excellent technical assistance.