The diversity of endophytic fungi in the above-ground tissue of two Lycopodium species in Poland

Endophytes are a large and diverse group of fungi that colonize healthy plant tissues without causing any symptoms. The majority of studies have focused on angiosperm and conifer hosts and few have examined the endophytes of lycophytes. In the present study, we characterized culturable endophytic fungi in two closely related Lycopodium species (L. annotinum and L. clavatum) from pine, beech, oak and spruce forests across Poland. More than 400 strains were isolated but only 18 Ascomycete species were identified. Members of the Dothideomycetes dominated the fungal endophyte communities in Lycopodium. The most abundant taxa cultured were Phoma brasiliensis (from L. clavatum) and Paraconiothyrium lycopodinum (from L. annotinum). Five taxa were isolated exclusively from L. annotinum, but only two of them (Paraconiothyrium lycopodinum and Mycosphaerella sp.) were relatively abundant. Two taxa were only found in L. clavatum, namely: Stagonospora pseudovitensis and an unidentified Dothideomycete. The taxon assigned as Ascomycota 2 (SH219457.06FU) was isolated only from strobili of both host species. Direct PCR and cloning from L. annotinum shoots revealed a substantially greater endophyte richness compared with the results from culturing.


Introduction
Lycopodium (Lycopodiaceae, Lycopodiophyta, Plantae) is a genus of flowerless, vascular, terrestrial plants that reproduce sexually via homospores, and vegetatively from rhizomes. The Lycopod lineage diversified early and was the dominant vegetation in Carboniferous period (Ranker and Haufler 2008). Reports of fossils found in the Rhynie chert suggest that arbuscular mycorrhizal fungi and Lycopods co-evolved from the beginning (Remy et al. 1994;Taylor et al. 1999Taylor et al. , 2004. Endophytes form a large and phylogenetically diverse group of fungi that colonize healthy plant tissues without causing symptoms (Wilson 1995). Endophytes sensu stricto differ from mycorrhizal fungi by the fact that they reside entirely within plant tissues (Stone et al. 2004). Four classes of endophytes have been distinguished: 1) the clavicipitaceous endophytes, 2) the nonclavicipitaceous endophytes that colonize the whole plant, 3) the nonclavicipitaceous endophytes that colonize shoots, and 4) the nonclavicipitaceous endophytes that colonize roots (Rodriguez et al. 2009). All land plants studied to date, including lycophytes, are colonized by class three endophytes which are horizontally transmitted (Davis et al. 2003;Saikkonen et al. 1998). The representatives of this class occur in above-ground tissues, form localized infections and are remarkable for their high diversity even within individual hosts (e.g. Arnold 2007;Higgins et al. 2007). Many studies have tried to determine whether endophytes exhibit tissue-and host-specificity but the results are contradictory (Arnold et al. 2001;Higgins et al. 2007Higgins et al. , 2011Joshee et al. 2009;Moricca et al. 2012;Sun et al. 2012;Wearn et al. 2012), with indications of both host-specificity and hostgeneralism (see review of older reports by Zhou and Hyde (2001).
Only few studies have been devoted to examining the endophytic community of Lycopodiaceae (e.g. Budziszewska and Szypuła 2010;Chen et al. 2011;Wang et al. 2011), and two have been driven by an interest in identifying possible novel bioactive chemical compounds (e.g. Zhu et al. 2010;Xiang et al. 2013). Indeed, some endophytic fungi from lycophyte Huperzia serrata that is used in chinese medicine are known to produce bioactive metabolites such as huaspenones, (Xiang et al. 2013) or huperzine (see Zhu et al. 2010;Zhang et al. 2011, and literature therein). Only two studies, by Holm and Holm (1981) and by Engelhardt (1987), have been dedicated studying the fungal communities of above-ground organs of Lycopodiaceae.
Representatives of the Lycopodiaceae have been included in broader comparative studies of fungal endophytic communities in different plant groups (e.g. Higgins et al. 2007;U'Ren et al. 2012). Although the Lycopodiaceae have been studied for over 100 years, most researches have focused on the symbiotic relationships between the prothalli and mycorrhizal fungi (e.g. Bruchmann 1908;Freeberg 1962;Whittier 1977;Berch and Kendrick 1982;Schmid and Oberwinkler 1993;Read et al. 2000;Lee et al. 2001;Wang and Qiu 2006;Leake et al. 2008;Kessler et al. 2010a, b;Muthukumar and Prabha 2013). Even in this respect, only about 17 species of Lycopodiaceae have been studied (Wang and Qiu 2006). Non-mycorrhizal fungi (e.g. pathogens or saprotrophs) from Lycopodiaceae have also been examined in surveys (e.g. Hagen 1950;Holm and Holm 1984;Jaklitsch and Voglmayr 2012) or when preparing monographs of particular genera (e.g. Massarina by Aptroot 1998 or Phaeosphaeria by Leuchtmann 1984).
The aim of the present study was to characterize the fungal endophytic communities of two closely related Lycopodium species. We addressed the following questions: 1) How diverse are endophytic fungi from Lycopodium in temperate forest? 2) Is species community composition influenced by host plant, plant organ, geographic location of the host, or host habitat characteristics? 3) What is the taxonomic identity of hypothetical species isolated as endophytes?
2 Materials and methods 2.1 Host plant collection and localization of sampling sites Shoots (8 individuals), strobili (15 individuals) and whole plants (shoots and strobili from the same individual; 13 individuals) of L. annotinum from 36 sites, and 10 shoots, 15 strobili, 2 whole plants (shoots and strobili from the same individual) of L. clavatum from 27 locations from temperate forests in Poland were collected (a total of 63 locations, 63 individuals and 78 samples; Fig. 1; see supplementary table S1). Sampling sites differed in terms of vegetation, with majority located in pine forests (27 sites); others in mixed forests (12 sites), acidophilic beech forests (1 site), acidophilic oak forests (4 sites), bog pine forests (11 sites), or mountain spruce forests (8 sites). Samples were further characterized by site elevation: lowlands (6 sites, 10 samples), highlands (49 sites, 58 samples), and mountains (8 sites, 10 samples; see supplementary table S1). Samples were collected in summer and autumn 2011 (see supplementary table S1). Each specimen was wrapped in a paper towel moistened with sterile water and placed in sterile plastic tube for transportation to the laboratory.

Endophytes isolation and identification
Healthy shoots and strobili of L. annotinum and L. clavatum were washed in tap water and surface-sterilized by subsequent submersion of the plant tissue for 2 min in a 0.5 % sodium hypochlorite and for 2 min in 70 % ethanol. Finally, samples were rinse in sterile distilled water, as described by Davis et al. (2003). This method eliminates epiphytic bacteria, yeasts, and fast-growing Zygomycetes (Arnold et al. 2000). The absence of epiphytic fungi was then verified by imprinting sterilized plant fragments on control PDA plates (if no fungal growth was observed, the plant fragment was considered to have been effectively surface-sterilized). Subsequently, the samples of lycophytes were cut in 2 mm pieces. Eight segments per organ (shoot or strobili) per site were plated (624 segments in total) on 4 % potato dextrose agar (PDA; without antibiotics). All plates (control and samples) were incubated at room temperature (ca. 18 o C) for 10 weeks or until fungal growth was observed. Using aseptic technique, emergent hyphae were transferred and purified on sterile PDA plates. The endophytes were grouped into morphotypes as described by Wang et al. (2011). Identification of the isolated strains was done using Domsch et al. (1993), Samson et al. (2004), and Watanabe (2002). Ninety six morphotypes were distinguished.
Genomic DNA was extracted from pure cultures of a representative isolate of each fungal morphotype using the GeneMATRIX Plant & Fungi DNA Purification Kit (EURx Ltd., Gdańsk, Poland) following the manufacturer's instructions. The complete ITS region was amplified using the primer pair ITS1f and ITS4 (White et al. 1990). PCR products were analyzed on a 1 % agarose gel stained with EtBr, and positive amplicons were purified using the GeneMATRIX Agarose-Out DNA Purification Kit (EURx Ltd). Purified DNA fragments were sequenced in both directions using the BigDye® Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Carlsbad, CA, USA). Sequencing was performed at the Institute of Biochemistry and Biophysics, the Polish Academy of Sciences. Forward and reversed sequences were aligned into contigs and manually edited for errors using the BioEdit Sequence Alignment Editor v. 7.0.0 (Hall 1999).
ITS sequences of all (96) morphotypes were grouped into operational taxonomic units (OTUs) according to 98.5 % similarity using BioEdit Sequence Alignment Editor v. 7.0.0 (Hall 1999). This 98.5 % cutoff was selected and used for the hypothetical species assignment in the UNITE database (Kõljalg et al. 2013).
In order to estimate the fraction of unculturable endophytes, total genomic DNA was extracted (as described above) from sterilized L. annotinum shoot from site no. 30 (coordinates in supplement S1). The ITS and 5.8S rDNA regions were amplified using fungal specific primer pair ITS1-f and ITS4 (as described above). PCR products were ligated into pGEM-T Easy Vector (Promega, Leiden, The Netherlands) and cloned in E. coli JM109 competent cells (Promega) following the manufacturer's instructions. Colony PCRs were performed using universal primer pair M13f and M13r.
Representative vouchers specimens for each OTU were deposited at the General Herbarium, University of Warsaw. Their numbers as well as the GenBank accession numbers of their sequences are presented in Table 1.

Data analysis
ITS sequences representing each morphotype were queried against the UNITE database using the massBLAST algorithm (http://unite.ut.ee; Abarenkov et al. 2010;Kõjalg et al. 2005). Species Hypothesis (SH) at 98.5 % of sequence similarity were used to identify isolated morphotypes (Kõljalg et al. 2013). If SH at 98.5 % of similarity was not proposed, the results of BLASTn (Altschul et al. 1997) searches were used to estimate taxonomic position of isolate and then detailed explanation concerning each case were given in Table 1.
The colonization factor (CF%; called also isolation frequency or colonization frequency) was calculated as the total number of plant segments colonized by fungi divided by the total number of all incubated segments, expressed as its percentage (Hata and Futai 1995). Relative abundance was calculated as the number of all isolates of a given taxon divided by the total number of isolates of all taxa. Frequency was calculated as the number of host individuals of fungal taxon isolated divided by the total number of all host individuals (Sun et al. 2012). Species diversity was evaluated using Shannon's Diversity Index (Shannon 1948) and Fisher's alpha (Fisher et al. 1943). The species evenness was estimated with Pielou's evenness index (Pielou 1966). The similarity of endophytic communities between different sampling sites (63) was evaluated using the Jaccard similarity coefficient (Jaccard 1912). Sample-based species accumulation curves for Fig. 1 Distribution of sample sites different hosts and organs were calculated and estimations of total richness were compared using the Jackknife 1 extrapolation algorithm. To correct for unequal sample sizes, the data was reduced to the smallest, consistent sample size (N=12, for shoots of Lycopodium clavatum). All indices were calculated using EstimateS v.9.1.0. (Colwell 2006).
Statistical analyses were then performed in four different variants in order to compare the influence of host plant, type of plant organ, host geographical location, and host habitat characteristics: (1) comparison between Lycopodium calavatum and Lycopodium annotinum; (2) comparison between shoots and strobili; (3) comparison between the samples coming from lowlands, highlands and mountains; (4) comparison between the samples collected from mixed pine, fresh pine, acidophilic oak, mountain spruce, pine bog, and acidophilic beech forests.
To correct for inconsistent sample sizes in diversity and richness comparisons between samples, the data was reduced to the smallest sample size (randomly selected host individuals) as proposed by Davis and Shaw (2008). In all the described variants, after the performance of the Shapiro-Wilk tests for normality for each variant separately, Kruskal-Wallis one-way analysis of variance was chosen as an appropriate method of comparison (non-parametric test for groups of unequal size) (Kruskal and Wallis 1952).
Variation in community composition was examined as a function of host, organ, location and vegetation type using three different ecological community similarity measures: Jaccard's similarity index, the Bray-Curtis coefficient and Euclidean distance. Resulting matrices were represented using non-metric multidimensional scale (NMDS) plots. To avoid pseudoreplication, data from different tissues types of the same host individual were pooled for NMDS analyses that examine the effect of host, location, and vegetation type. The Kruskal's stress was used to decide which grouping of data is the most accurate (commonly acceptable when lower than 0.2) (Ek-Ramos et al. 2013).
Additionally, the correlation between Jaccard similarity coefficient and geographic distance was tested for significance using simple Mantel test. The geographical distances (in km) based on coordinates of collection sites (WGS84 system) were determined using own program written in C (assuming that the earth is a perfect sphere and its radius is 6,378 km).
All statistical analyses were calculated using STATISTICA v. 10 and PAST v. 2.16 software (Hammer et al. 2001).

Results
From 624 Lycopodium segments (representing a total of 63 locations, 63 individuals and 78 samples), a total of 458 isolates, representing 96 morphotypes, 18 OTUs (hypothetical species defined in UNITE database) (Table 1), which belong only to the Ascomycota, were isolated during this study. Representatives of Dothideomycetes were isolated the most frequently. However, only in the case of L. annotinum shoots did the accumulation curve reach an asymptote. The Jacknife richness estimator did not reach an asymptote which indicates that richness would continue to increase with further sampling (Fig. 2). The most abundant taxa were Phoma brasiliense in L. clavatum and Paraconiothyrium lycopodinum in L. annotinum (Table 2). The highest frequency was observed also for these two taxa (Fig. 3).
None of the isolated taxa was present in all samples. There were 5 taxa isolated exclusively from L. annotinum, but only two of them (Paraconiothyrium lycopodinum and Mycosphaerella sp.) were relatively abundant (more than 5 isolates). There were also two taxa that were exclusive for L. clavatum, namely: Stagonospora pseudovitensis and unidentified Dothideomycetes (SH196053.06FU). The taxon assigned as Ascomycota 2 (SH219457.06FU) was isolated only from strobili of both host species.
Although we analyzed many more segments of L. annotinum than L. clavatum, the number of isolated species as well as colonization factor and values of species diversity indices for random 12 samples (the smallest sample size) were very similar (Table 3).
Only the Shannon diversity index (p=0.0294) was significantly higher in shoots than in strobili for both Lycopodium species (see also Table 3). This pattern is also visible in species accumulation and in Jacknife 1 richness estimator curves (Fig. 2).
Cluster analysis of endophytic community similarity measures presented in two-dimensional NMDS plots revealed that these communities are neither lycophyte-host-related , nor organ-related (even though differences is diversity between organs are significant). Variation in endophytic community structure could neither be explained by site vegetation type nor by geographic region of host plant origin (Fig. 4). The use of different ecological similarity measures did not significantly affect the observed patterns (data not shown). In most cases Kruskal's stress values were too high (>0.2) to confidently discern any pattern. However, the two samples of L. annotinum from the lowland that had a very different endophytic communy from samples collected from other elevations.
The shoots of L. annotinum from site 30 (as described in supplement S1) on PDA medium yielded only two species Leiosphaerella lycopodina and Mycosphaerella sp., while in the cloning experiment with the plant tissue, seven different sequences were obtained (Table 4). The sequences of both species isolated in culture were also present in cloned sequences and they were the most frequent ones. However, they still constituted less than a half of all obtained sequences.  (Table 4). Another possibility is that species diversity may vary with environmental factors at sample sites but more study is required. Among the isolated taxa, five were found exclusively in L. annotinum, of which only two (Paraconiothyrium lycopodinum and Mycosphaerella sp.) were abundant (more than 5 isolates). The isolate Ascomycota 2 (SH219457.06FU) was restricted to the strobili, regardless of the host species which suggests some level of tissue-preference, for this isolate. However, this might be a result of isolate rarity. Cluster analysis of endophytic community similarity measures presented in two-dimensional NMDS plots revealed that only two samples from L. annotinum from lowland have very different endophytic community than all others. This pattern could be explained by the relatively high abundance of Mycosphaerella sp. and by the fact that both sites are relatively close together (240 m apart). As lycophytes can reproduce asexually, it is possible that the plants represent the same host individual. The abundant presence of Mycosphaerella sp. at these sites should be verified in further research.

Discussion
Some of taxa isolated in our study are common and widespread saprotrophic fungi, e.g. Hypocrea rufa and its anamorph Trichoderma viride but they have been isolated as endophytes previously (Jaklitsch et al. 2006), but not from lycophytes. Among other taxa we isolated are well known plant pathogens, e.g. Leiosphaerella lycopodina that has been recorded from Lycopodium annotinum several times (e.g. Jaklitsch and Voglmayr 2012) but never described to date as an endophyte.
One of the most abundant taxon in our study was Paraconiothyrium lycopodinum that could be identified as Coniothyrium lycopodinum Sacc. & Paol. that was previously isolated as a possible pathogen from L. annotinum (Saccardo 1889). These two taxa were recently synonimized by Pawłowska et al. (in Crous et al. 2013) as Paraconiothyrium lycopodinum (Sacc. & Paol.) J. Pawłowska, Wilk, Śliwińska-Wyrzychowska, Mętrak & Wrzosek, comb. nov. In spite of the fact that several Paraconiothyrium species have been isolated as endophytes from asymptomatic photosynthetic tissues of plant species (e.g. Abreu et al. 2010), including lycophytes (Budziszewska et al. 2011;Wang et al. 2011), in general they are still regarded as plant pathogens (Damm et al. 2008).
It is interesting that some fungi isolated in our study are known to be, or bear close affinities, to plant pathogens. This is a common finding of studies on endophytic fungal communities (e.g. Moricca et al. 2012;Sun et al. 2012;Wearn et al. 2012 and references therein). This has led to the hypothesis that endophytism is a common phase in the life cycle of many fungal pathogens (e.g. Carroll 1988;Joshee et al. 2009;Rodriguez and Redman 1997;Schulz and Boyle 2005).
Recently, Delaye et al. (2013) performed an analysis of the evolutionary stability of biotropic, necrotrophic, and endophytic lifestyles of 163 fungal strains. They suggest that while biotrophy appears to be a stable trait, asymptomatic endophytes can easily switch to necrotrophy, even at an ecological timescale. Not only is this of paramount importance for understanding the mechanisms and factors underlying alterations in fungal lifestyle, but it is also significant for studies on endophytic fungal communities. A large number of fungal species described in 19th century that often have no holotypes and are in desperate need for revision using molecular approaches. These fungi are still found on lists of fungal taxa recorded from a particular plant hosts (e.g. Farr and Rossman 2012). They were regarded as pathogens or saprotrophs, partially because molecular identification of endophytes was beyond the technical abilities at the time. Current investigations of endophytic fungal communities often exclusively apply molecular methods (Rajala et al. 2013). However a combination of these methods with culturing is best (e.g. Arnold and Lutzoni 2007). A study of the endophytic community based on molecular identification and culturing with direct observation of fungal communities on the plant material, has been successfully applied by Chaverri and Gazis (2011) and can shed more light on the question of the 'endophytic continuum'. In this way it may also be possible to link of fungal taxa described in inventories like those by Engelhardt (1987) and by Holm and Holm (1981) with sequences of unidentified endophytes generated in numerous molecular studies. Finally, the possible roles of these endophytic fungi in interactions with lycophytes are unknown. To date, no research has been performed on whether such endophytes may in any way enhance lycophyte fitness, make them more competitive within their habitats, deter potential animal herbivores or protect from pathogenic fungi or bacteria. There is clearly much more research to be done.