Prunus trees in Germany—a hideout of unknown fungi?

Prunus belongs to the economically most important genera of fruit crops in Germany. Although wood pathogens possess the capability to damage the host substantially, the knowledge of the fungal pathogenic community and the mycobiome of Prunus wood in general is low. During a survey in important fruit production areas in Germany, branches with symptoms of fungal infection were sampled in Prunus avium, P. cerasus and P. domestica orchards, and 1018 fungal isolates were obtained primarily from the transition zone of symptomatic to non-symptomatic wood. By a combination of blastn searches and phylogenetic analyses based on ITS and LSU sequences with a strong focus on reliable reference data, a diversity of 172 fungal taxa belonging to Ascomycota, Basidiomycota and Mucoromycota were differentiated. The majority of the strains belonged to three classes of Ascomycota, namely Sordariomycetes, Leotiomycetes and Dothideomycetes. The dominant species were Aposphaeria corallinolutea (Dothideomycetes) and Pallidophorina paarla (Leotiomycetes) that were isolated more than a hundred times each, while all other taxa were isolated ≤ 30 times. Only part of them could be identified to species level. Because of the high plasticity of species boundaries, the identification certainty was divided into categories based on nucleotide differences to reference sequences. In total, 82 species were identified with high and 20 species with low (cf.) certainty. Moreover, about 70 species could not be assigned to a known species, which reveals Prunus wood to represent a habitat harbouring high numbers of potentially new species, even in a well-explored region like Germany.


Introduction
Fungal pathogens inhabiting the woody plant body can plug vessels and necrotise tissue, which causes wilting, inhibition of blossoming and dieback of branches and whole trees. The resulting decrease in fruit or timber yield can ruin the productivity of orchards, vineyards and forests and can even require replanting. Additionally, some of the pathogens can reduce the quality of fruits, which causes further yield losses. Moreover, trees in forests and orchards are usually grown in monocultures and are therefore especially threatened by fungal plant pathogens, both due to the increasing global plant trade (Roy et al. 2014, Ghelardini et al. 2017) and effects of climate change (Anderson et al. 2004, Gange et al. 2011, Luck et al. 2011, Fisher et al. 2012, Altizer et al. 2013). An example for the threat an exotic pathogen can pose to native trees is Hymenoscyphus fraxineus, the causal agent of ash dieback that moved from eastern Asia to Europe, encountering ash tree species being more susceptible (McMullan et al. 2018). Due to extreme conditions like drought, trees become also more susceptible to fungi that are already living as endophytes inside their wood, so-called weak parasites. They include species of Botryosphaeriales that have frequently been isolated from Prunus trees in South Africa (Damm et al. 2007a, b). In Germany, one of these species, Diplodia pinea, has been reported to cause serious damage to pine trees that suffered from drought stress and had been attacked by bark beetles (Heydeck andDahms 2012, Petercord 2017). Furthermore, trees can become more susceptible to pathogens or encounter new potential pathogens if they are planted outside their typical growing region, for example by the northward expansion Section Editor: Marc Stadler Electronic supplementary material The online version of this article (https://doi.org/10.1007/s11557-020-01586-4) contains supplementary material, which is available to authorized users.
of European crop production due to global warming (Maracchi et al. 2005, Santos et al. 2017. In order to allow an early detection and control of known and new threats for the fruit industry, knowledge of the wood mycobiome of fruit trees is crucial.
Fungal communities inside wood have frequently been studied using culture-independent high-throughput sequencing (HTS) (e.g. Kubartová et al. 2012, Hoppe et al. 2016) and isolation techniques (e.g. Kowalski 1983, Butin and Kowalski 1986, Lygis et al. 2005, Santamaría and Diez 2005, Simeto et al. 2005, Cloete et al. 2011, Markakis et al. 2017, Fischer et al. 2016. However, many studies focused on endophytic fungi (e.g. Barengo et al. 2000, Fröhlich et al. 2000, Gonthier et al. 2006 or were restricted to grapevine wood (e.g. Hofstetter et al. 2012, Pancher et al. 2012, Bruez et al. 2014. Sweet cherry (Prunus avium), sour cherry (P. cerasus) and plum (P. domestica) are the most important stone fruit crops in German fruit industry (Garming et al. 2018). In 2018, more than 350,000 t of sweet cherry, sour cherry and plum fruit were produced on an area of around 12,000 ha (FAO 2020). In spite of this economic importance, there are only a few studies on the fungal diversity of aboveground woody parts of Prunus trees (e.g. Bernadovičová and Ivanová 2011, Haddadderafshi et al. 2011, Hortová and Novotný 2011, Gramaje et al. 2012, Abdollahi Aghdam and Fotouhifar 2016, 2017. Most of these studies are limited by a small sample size, by a narrow sampling area or by relying solely on morphological features for species identification. The most extensive work so far has been conducted in a survey on the fungal diversity of Prunus species in South Africa (Damm et al. 2007a, b, 2008a, b, c, 2010, Moyo et al. 2018. More than 40 taxa were reported, predominantly within Botryosphaeriales (nine species) and Phaeoacremonium (14 species). During this survey, 24 species of Botryosphaeria-ceae, Calosphaeriaceae, Togniniaceae, Montagnulaceae, Coniochaetaceae, Celotheliaceae, Tympanidaceae and Ploettnerulaceae were recognised as new to science. However, these publications aimed only on selected, very abundant or specifically interesting taxa of wood-inhabiting fungi from Prunus wood in South Africa; the complete diversity collected was not evaluated. Moreover, no comprehensive study has been done on the mycobiome of Prunus trees in Germany. In a study on several tree species in the vicinity of a vineyard in Germany, only a selection of eight fungal species (belonging to Botryosphaeriaceae, Stereaceae, Tympanidaceae and Valsaceae) isolated from wood of six Prunus species (including P. cerasus and P. domestica) was reported (Gierl and Fischer 2017).
With an extensive study such as the evaluation of a mycobiome, time is the most limiting factor. For the selection of an appropriate approach for identification, quantity and quality have to be balanced against each other. Uncertainties in identifications of fungi can arise due to deficiencies of both morphological and molecular approaches. Morphological identification of fungal cultures is hindered or impossible, if strains do not develop identification-relevant features (fruiting structures) or show phenotypic plasticity (Slepecky and Starmer 2009), belong to a complex of cryptic species that cannot be differentiated by morphological features (e.g. Damm et al. 2012) or species had been described based on one morph only, usually the sexual morph, that does not develop in culture (e.g. . Even if morphological identification is possible, each genus requires a certain amount of expertise (Hofstetter et al. 2019), as well as time to obtain necessary literature and reference/type material. If many taxa extending over the entire fungal kingdom need to be identified in a reasonable time frame, an overall morphology-based approach is not appropriate; identification based on sequence data is the method of choice.
Fungal identification solely based on blastn searches with ITS sequences is common practice (Hughes et al. 2009, Hofstetter et al. 2019); however, it has a lot of shortcomings as well. Although the ITS region is considered as the universal barcode region for fungi and the most commonly sequenced locus in mycology, it is not suitable for species delimitation in each genus (Schoch et al. 2012). Species identification in surveys using HTS is even less certain, because the sequences generated are very short, and the high number of sequences generated puts even more time pressure on identification, allowing only unquestioned/unvalued blastn searches. Moreover, identification results cannot be verified by morphology as no cultures are available. Therefore, species can often only be identified up to genus level (LoBuglio and Pfister 2010, Johnston et al. 2014, Ekanayaka et al. 2017, Pärtel et al. 2017 or result in doubtful identifications like those of Collophorina species that are discussed in .
The purpose of this study was to reveal the mycobiome of Prunus trees in a temperate climate focusing on potential pathogens associated with wood necroses of P. avium, P. cerasus and P. domestica in three important fruit production areas in Germany. Some of the genera isolated within this study, belonging to the Leotiomycetes and Eurotiomycetes, have previously been analysed in depth and several new taxa were revealed . The aim of this study was to give an overview of the complete fungal diversity based on LSU and ITS sequences, to highlight the possible depth of identification based on these loci as part of a mycobiome study and to detect potential new taxa. A culture-dependent approach allowed verifying results by morphology, if necessary, and facilitates further taxonomic studies.

Sampling and fungal isolation
Branches with wood symptoms (e.g. canker, necroses, wood streaking, gummosis) were collected from Prunus domestica (61 branches), P. cerasus (64) and P. avium (43) orchards in Saxony; from P. domestica (30) and P. avium (60) orchards in Lower Saxony; and from P. domestica (38) and P. avium (48) orchards in Baden-Württemberg, Germany, in 2015 and 2016. Additionally, a symptomatic wood sample from a P. cerasus tree located in a private garden in Bavaria was included. From each of these 345 branches, ten wood pieces (5 × 5 × 5 mm) from the transition zone of symptomatic to nonsymptomatic wood tissue as well as each three pieces of the same size from non-symptomatic wood of the same branch were surface sterilised 30 s in 70% ethanol, 1 min in 3.5% NaOCl and 30 s in 70% ethanol and washed for 1 min in sterilised water. Five pieces from symptomatic tissue were placed on synthetic nutrient-poor agar (SNA, Nirenberg 1976) medium, and the remaining five pieces from symptomatic tissue as well as the three pieces from non-symptomatic tissue on oatmeal agar (OA; Crous et al. 2019) medium both supplemented with 100 mg/L penicillin, 50 mg/L streptomycin sulphate and 1 mg/L chloramphenicol. After incubation for several days at 25°C, hyphal tips of developing fungi were transferred to SNA medium with a sterilised pine needle. Single-spore or single-hyphae isolates were obtained from the fungi for further study.
The resulting strains are preserved in cryotubes containing sterile distilled water with 10% glycerol at − 80°C and in sterile distilled water at + 4°C in the culture collection of the Senckenberg Museum of Natural History Görlitz, Germany (GLMC). Specimens (dried cultures) were deposited in the fungarium of the Senckenberg Museum of Natural History Görlitz (GLM).

Phylogenetic analysis
Genomic DNA of the isolates was extracted using the method of Damm et al. (2008b). A partial sequence of the 28S nrDNA (LSU) and the 5.8S nuclear ribosomal gene with the two flanking internal transcribed spacers ITS-1 and ITS-2 (ITS) were amplified and sequenced using the primer pairs LROR (Rehner and Samuels 1994) + LR5 (Vilgalys and Hester 1990) and ITS-1F (Gardes and Bruns 1993) + ITS-4 (White et al. 1990), respectively.
The PCR mixture contained 1 μL of 1:10 DNA template, 2.5 μL 10X buffer (Peqlab, Erlangen, Germany), 1 μL of each primer (10 mM), 2.5 μL MgCl 2 (25 mM), 0.1 μL Taq polymerase (0.5 U, Peqlab, Erlangen, Germany) and 2.5 μL of 2 mM dNTPs. Each reaction was made up to a final volume of 20 μL with sterile water. DNA amplifications were carried out in a Mastercycler® pro S (Eppendorf, Hamburg, Germany). Conditions for the amplification of LSU and ITS were set according to Paulin and Harrington (2000) and , respectively. The PCR products were visualised on a 1% agarose gel and sequenced by the Senckenberg Biodiversity and Climate Research Centre (BiK-F) laboratory (Frankfurt, Germany). The forward and reverse sequences were assembled by using BioEdit Sequence Alignment Editor (v. 7.2.5;Hall 1999).
All strains were grouped based on comparison of their ITS sequences. One strain of each group with an identical ITS sequence was selected for blastn searches and phylogenetic analysis. For generic determination of the isolates and selection of reference strains, blastn searches were performed on the NCBI GenBank (www.ncbi.nlm.nih.gov) and EPPO-Q-Bank (qbank.eppo.int) databases. For each genus, sequences of strains identified to species level, preferably of ex-type strains and strains of the type species, with at least 97% identity were included as reference strains in the phylogenetic analyses. If no type strains were available, strains with a CBS (culture collection of the Westerdijk Fungal Biodiversity Institute, Utrecht, the Netherlands) number were favoured. Strains without species determination were only used, if blastn searches did not result in any close match with a strain identified to species level.
For the phylogenetic analyses, the sequences downloaded were added to the sequences generated in this study and those of the appropriate outgroup sequences in five LSU-ITS datasets depending on phylum and class. Four datasets were assembled for species of the Ascomycota classes Sordario-mycetes, Dothideomycetes, Leotiomycetes and Eurotiomycetes, respectively. A fifth d a t a s e t e n c o m p a s s e s s p e c i e s o f t h e c l a s s e s A g a r i c o m y c e t e s , T r e m e l l o m y c e t e s a n d Cystobasidiomycetes (Basidiomycota); Lecanoromycetes, Pezizomycetes and Saccharomycetes (Ascomycota); as well as the subdivision Mucoromycotina of the Mucoromycota. The datasets of each locus were aligned automatically using MAFFT v. 7.308 (Katoh et al. 2002, Katoh andStandley 2013), manually adjusted where necessary and subsequently concatenated using Geneious v. 10.2.2 (Kearse et al. 2012).
The phylogenetic analyses were conducted using Bayesian inference (BI) and maximum likelihood (ML) as described in . The DNA sequences generated in this study were deposited in GenBank (Table 1) and the alignments in TreeBASE (treebase.org/treebase-web/ home.html; TB2:S25316). The complete list of strains included in the phylogenetic analyses is provided in the supplementary material table (suppl. material tab.).

Identification
The strains were identified to species, genus or higher level, depending on the affinity to the available reference sequences. These identifications were assigned to a level of identification certainty based on an evaluation of the respective clades in the phylogenetic trees and nucleotide differences in the respective ITS alignments. A species was assigned to "identified with high certainty", if the strain showed ≤ 4 nucleotide differences in the ITS sequence to a named reference sequence. Letters at the species name indicate a sequence variation within strains that were identified as the same species. A low certainty was indicated with "cf.", if the ITS sequence of a strain differed in 5-10 nucleotides from the closest named reference sequence. The strain was assigned to a genus, but not to a species, if the ITS sequence differed in > 10 nucleotides from the closest named reference sequence or matched with more than one named reference sequence and marked with "sp." or "cf. spp.", respectively. If the strain belonged to a clade, for which no named reference sequence was available or with reference sequences belonging to more than one genus, the name of family, order or class was applied. Identifications of part of the taxa to genus level were verified based on microscopic examination of morphological features formed on the used standard media.

Results
In total, 1018 fungal strains were isolated from Prunus wood, which belonged to 172 species. The numbers of species isolated per host species were as follows: 113 species from Prunus domestica, 70 from P. avium and 61 from P. cerasus Nov., novelties and potential first reports during this survey; sy., from symptomatic wood tissue; n-sy., from non-symptomatic wood tissue; P.d., from Prunus domestica; P.c., from P. cerasus; P.a., from P. avium; Sa, from Saxony; LSa, from Lower Saxony; BW, from Baden-Württemberg; Ba, from Bavaria; N, newly described in  or in Bien and Damm 2020; G, P, a, c, d, potential first report from Germany, Prunus, P. avium, P. cerasus or P. domestica, respectively; rep. strain, representative strain for the taxon 1 LSU, 28S nrDNA; ITS, internal transcribed spacers and intervening 5.8S nrDNA Saxony (n=168)  . 1a). While 66, 31 and 20 species, respectively, were exclusively isolated from one of these hosts, 17 species occurred in all of them. Regarding the main sampling regions, 122, 75 and 43 species were isolated from Prunus wood collected in Saxony, Baden-Württemberg and Lower Saxony, respectively. While 73, 34 and 10 species, respectively, were exclusively isolated from wood collected from one of these regions, 13 species occurred in all of them (Fig. 1b). Five species were isolated from all three Prunus species and in all collection areas, namely Alternaria destruens, Aposphaeria corallinolutea, Aureobasidium pullulans, Pallidophorina paarla and Cladosporium cf. spp. 1 (Table 1). Aposphaeria corallinolutea and Pa. were isolated 138 and 112 times, respectively, all other taxa ≤ 30 times. Most of the taxa with 15-30 strains were isolated from at least two host species and in at least two collection regions, except for Collophorina africana and Lepteutypa sp. 1 that were collected only from P. domestica, and Coniothyrium ferrarisianum that was collected only from Saxony. The majority of the species (166 species) was isolated from the transition zone between symptomatic and non-symptomatic tissue, 138 species exclusively from this tissue, while 34 species were isolated from asymptomatic tissue, six species (each one isolate) exclusively from asymptomatic tissue.

Baden-Württemberg
Of the 172 species, 152 species belonged to the Ascomycota (965 strains), 16 to the Basidiomycota (45 strains) and four to the Mucoromycota (eight strains). Within the Ascomycota, 75 species belonged to the Sordariomycetes (356 strains), 30 to the Leotiomycetes (290 strains), 30 to the Dothideomycetes (287 strains) and 13 to the Eurotiomycetes (27 strains), representing 43.6%, 17.4%, 17.4% and 7.6%, respectively, of the total diversity and 35%, 28.5%, 28.2% and 2.7%, respectively, of the abundance of the complete mycobiome of Prunus wood isolated in this study ( Fig. 2a, b). The sequences of the four most abundant classes of Ascomycota were analysed in separate alignments, while the remaining classes of the Ascomycota were analysed together with Basidiomycota and Mucoromycota.

Phylogenetic analyses
The combined sequence dataset 1 of the Sordariomycetes consisted of 246 strains including the reference strains and the outgroup Cadophora luteo-olivacea strain CBS 141.41 (Leotiomycetes) and comprised 1884 characters (gene boundaries: LSU: 1-902, ITS: 903-1884, including gaps). The final ML optimisation likelihood of ML analysis was lnL = -39,461.875859. In total, 356 isolates from Prunus wood belonged to 75 taxa (Fig. 3). Thirty-one species (136 isolates) were placed in the order Xylariales, of which 15 taxa were determined to species, 13 to genus and three to family level. Six species (77 isolates) were placed in the Diaporthales; the generic determination of one of them was unclear. Three taxa (35 isolates) were placed in the Calosphaeriales, four (14 isolates) in the genus Phaeoacremonium, Togniniales, and two (six isolates) in the Ophiostomatales. Seventeen species (57 isolates) were placed in the order Hypocreales; the generic placement of three of them was unclear. Two taxa (nine isolates) were placed in the order Glomerellales and determined to species level. Four species (13 isolates) were placed in the order Sordariales, one of which not determined to genus level. One isolate was placed in the genus Chaetosphaeria, Chaetosphaeriales. Four species (six isolates) were placed in the genus Coniochaeta, Coniochaetales; one of them was identified to species level. One species (two isolates) was placed in a clade formed by strains of Phialemonium sp., sister to the single-strain clade of the ex-type strain of Ph. dimorphosporum (incertae sedis  Calosphaeria pulchella (Calosphaeriales) was the most frequently isolated species in the Sordariomycetes.
The combined sequence dataset 2 of the Dothideomycetes consisted of 113 strains including the outgroup Penicillium resticulosum strain CBS 609.94 (Eurotiomycetes) and comprised 1585 characters (gene boundaries: LSU: 1-870, ITS: 871-1585, including gaps). The final ML optimisation likelihood of ML analysis was lnL = -17,660.376634. In total, 287 isolates belonged to 30 taxa (Fig. 4). Twenty-three taxa (258 isolates) were placed in the Pleosporales, of which 17 were determined to species, four to genus and each one to family and order level. Four taxa (14 isolates) were placed in the Capnodiales and determined to species or genus level. One taxon (15 isolates) of the Dothideales was identified as Aureobasidium pullulans. Each one isolate was identified as Diplodia mutila and D. seriata (Botryosphaeriales). With 138 strains, Aposphaeria corallinolutea (Pleosporales) was the most frequently isolated species in the Dothideomycetes.
The combined sequence dataset 3 of the Leotiomycetes consisted of 84 strains including the outgroup Colletotrichum godetiae strain CBS 133.44 (Sordariomycetes) and comprised 1557 characters (gene boundaries: LSU: 1-912, ITS: 913-1557, including gaps). The final ML optimisation likelihood of ML analysis was lnL = -11,950.782384. In total, 290 isolates belonged to 30 taxa (Fig. 5). Twenty-four taxa (137 isolates) were placed in the Helotiales, of which 15 were determined to species and nine to genus level. Five taxa (152 isolates) were placed in Phacidiales and determined to species level. One strain remained in an uncertain taxonomic position on order level. With 112 strains, Pallidophorina paarla (Phacidiales) was the most frequently isolated species in the Leotiomycetes.
The combined sequence dataset 4 of the Eurotiomycetes consisted of 38 strains including the outgroup Diplodia intermedia strain CBS 124462 (Dothideomycetes) and comprised 1573 characters (gene boundaries: LSU: 1-908, ITS: 909-1573, including gaps). The final ML optimisation likelihood of ML analysis was lnL = -9837.561743. In total, 27 isolates belonged to 13 taxa (Fig. 6). Seven taxa (15 isolates) were placed within Eurotiales, of which five were determined to species and two to genus level. Four taxa (eight isolates) were placed in Chaetothyriales; one was determined to species, two to genus and one to family level. Two taxa (four isolates) were placed in Phaeomoniellales, of which one was determined to species and one to genus level. All species of the Eurotiomycetes were isolated with low frequencies (≤ 6 strains).
The combined sequence dataset 5 of the remaining classes of the Ascomycota, as well as all Basidiomycota and Mucoromycota consisted of 105 strains including the outgroup Entomophthora sphaerosperma strain CBS 530.75 (Entomophthoromycotina, Zoopagomycota) and comprised 2058 characters (gene boundaries: LSU: 1-1115, ITS: 1116-2058, including gaps). The final ML optimisation likelihood of ML analysis was lnL = -33,834.310764. Within the 16 taxa (45 strains) of Basidiomycota, 14 taxa (41 strains) belonged to the in Agaricomycetes, of which 11 taxa were identified to species and three to genus level (Fig. 7). One isolate of the Tremellomycetes and one taxon (three isolates) of the Cystobasidiomycetes were determined to genus and species level, respectively. With 13 strains, Peniophora cinerea was the most frequently isolated species in the Basidiomycota. Species of the other phyla were isolated with low frequencies (< 10 strains). One strain of the class Pezizomycetes ( Ascomycota) was determined as T r i c h o p h a e o p s i s b i c u s p i s . O n e t a x o n o f t h e Lecanoromycetes (two strains) could not be further determined. Two strains of the Saccharomycetes were determined as Nakazawaea cf. holstii and Wickerhamomyces silvicola, respectively. Within the 8 strains of Mucoromycota, five strains were identified as Umbelopsis isabellina, two strains as Mucor circinelloides and M. hiemalis, respectively, while one further Mucor strain could not be assigned to a species.

Identification certainty
In total, 102 taxa were assigned to a particular species with high (82 taxa) or low (20 taxa) certainty. A further 57 species were determined to genus level. Thirteen species could not be assigned to any genus and were identified to family (six), order (five) or class (two), level, respectively. Almost all of the 70 taxa that were not identified to species level belonged to the Ascomycota, with the largest number of taxa belonging to the Sordariomycetes (39), followed by Leotiomycetes (ten), Dothideomycetes (nine) and Eurotiomycetes (five) (Fig. 8).
Only few undetermined species belonged to Basidiomycota, Mucoromycota and to the remaining classes of Ascomycota.

Fungal diversity of necrotic Prunus wood in Germany
In total, 172 fungal species were detected in the wood samples of Prunus trees studied. The diversity detected in this study far exceeds the number of taxa usually reported from isolation studies of woody plants. In many cases, not more than 30 taxa were reported (e.g. Barengo et al. 2000, Gonthier et al. 2006, Hortová and Novotný 2011, Markakis et al. 2017. Only in few studies up to or more than a hundred taxa were isolated   Fig. 3 (continued) from wood (Lygis et al. 2005, Simeto et al. 2005, Hofstetter et al. 2012. The high number of detected taxa in our study presumably results from the high sample number of three different target host species over a wider geographical area, in contrast to most of the studies that display less diversity. However, the isolated taxa only encompass those fungi present at the time of sampling and accessible by isolation; a multitude of fungi cannot be cultured in general or on standard media (Allen et al. 2003, O'Brien et al. 2005, Tsui et al. 2011, Muggia et al. 2017. Therefore, studies using cultureindependent high-throughput sequencing (HTS) techniques usually report much higher species numbers from the fungal diversity inside living or dead plant parts (up to 2000 operational taxonomic units, OTU) than studies using isolation techniques (e.g. Kubartová et al. 2012, Hoppe et al. 2016, Dissanayake et al. 2018, Jayawardena et al. 2018. As most of the taxa were isolated in this study only once or few times, we would expect the number of taxa to increase tremendously, if the number of wood samples would be increased. The mycobiome of the wood of the three Prunus species in Germany is far from being complete. The two most abundant species, Aposphaeria corallinolutea (Dothideomycetes, Pleosporales) and Pallidophorina paarla (Leotiomycetes, Phacidiales), were isolated > 100 times each and from all three host species and in all three collection areas. Aposphaeria corallinolutea was revealed as the most dominant inhabitant of Prunus wood in Germany in our study, while there are only five reports from previous studies: from Fraxinus excelsior and Kerria japonica in the Netherlands (de Gruyter et al. 2013), from decaying wood in Thailand (Li et al. 2016), from dead branches of Prunus padus in Russia (Tibpromma et al. 2017) and from needles of Pseudotsuga menziesii in the USA (Daniels 2017). The only ITS sequence of this species in GenBank originates from the study in Thailand. Thus, A. corallinolutea is known from several hosts, including Prunus, and from different countries, however, has not previously been reported from Germany or from any of the Prunus species studied here. The low number of reports could be explained by the lack of studies on its main host plants/substrates that, based on this study, includes necrotic wood of Prunus in Germany, but also by the facts that A. corallinolutea was described only 2013 (de Gruyter et al. 2013) and that the first and so far only ITS sequence of a strain identified as this species was submitted to GenBank only 2017 (Li et al. 2016). A blastn search with the ITS sequence of strain GLMC 1355 revealed a 100% match with an unidentified Ascomycota strain from leaves of Fagus sylvatica in Germany (Unterseher and Schnittler 2009) indicating the occurrence of this fungus on a further host as well as in Germany. In contrast, the second most abundant species, Pa. paarla, has previously frequently been reported from a number of Prunus species in several countries including Germany (Gierl andFischer 2017, Bien et al. 2020).
Part of the taxa isolated in this study probably represent first reports for the genus Prunus, for specific Prunus species or for Germany. We conducted a search of the 82 taxa identified to species level with high certainty on the USDA database (Farr and Rossman 2019). For 41 of these taxa, no previous report from Germany and for 40 taxa, no previous report from the host genus Prunus was listed (Table 1). Of further 25 taxa, there was no previous report from one or more of the Prunus hosts, on which they were collected from in our study. However, as some of the latest publications are missing, this database is apparently not complete. Therefore, and due to the unreliable identification results of many species, we consider these reports as potential first reports. They need to be confirmed by in-depth studies of the respective species, which was beyond the scope of this study.
The aim of this study was to reveal the mycobiome associated with necroses of Prunus wood in Germany as complete as possible in a reasonable time frame using a cultivation approach. As the study was based on commercial orchards, it was not possible to collect the exact amount of samples from each host species with the same age, same cultivars etc. at the same collection area. For some of the orchards, data like tree age and cultivar were not even available. Therefore, a direct comparison of the three collection areas and host species regarding strain or species numbers cannot be made as it is most probably biased by other factors.

Comparison with other studies from Prunus
The results obtained in this study could only be compared to a few other studies that used similar methods (culturing, sequence-based identification). However, most of them were conducted on different Prunus species and in different climates. The extensive survey of fungi in Prunus wood (P. armeniaca, P. dulcis, P. persica, P. persica var. nucipersica, P. salicina) in South Africa resulted in reports of 47 species in several publications by Damm et al. (2007aDamm et al. ( ,b, 2008aDamm et al. ( ,b,c, 2010, Moyo et al. (2018) and  focusing on specific genera. Gramaje et al. (2012) isolated nine fungal species from Prunus dulcis in Spain (Island of Mallorca) including five species belonging to the Botryosphaeriales as well as Collophorina hispanica, Diaporthe amygdali, Eutypa lata and Phaeoacremonium amygdalinum. The study of Inderbitzin et al. (2010) was restricted to Botryosphaeriaceae from Prunus dulcis in CA, USA, and that of Tian et al. (2018) to Diaporthe amygdali and Botryosphaeria dothidea of P. persica in Yangshan, China. The only study from Germany was that by Gierl and Fischer (2017), who reported only eight fungal species from symptomatic wood of six Prunus species, two of which were also sampled by us, namely P. cerasus and P. domestica.
Botryosphaeriales are known as pathogens and endophytes of various woody hosts (Slippers et al. 2007, Cloete et (Damm et al. 2007a, b, Inderbitzin et al. 2010, Gramaje et al. 2012, Tian et al. 2018. The dominating species in the studies from South Africa and Spain were D. seriata and Neofusicoccum parvum, respectively, while only Botryosphaeria dothidea was reported in that from China. Moreover, D. pinea, a pathogen of several Pinus species in many countries (Farr and Rossman 2019), that also cause serious damage to pine trees suffering from drought stress and bark beetle attacks in Germany (Heydeck andDahms 2012, Petercord 2017), had frequently been isolated from P. persica in South Africa and tested positive for its pathogenicity on this host (Damm et al. 2007a). Therefore, host jumps from infected Pinus plantations to Prunus orchards in close vicinity are possible. However, Botryosphaeriales were surprisingly rare in this study. Only one strain each of D. seriata and D. mutila was detected in wood of P. domestica in the most southern sampling region in Germany; D. pinea was not isolated at all. Brodde et al. (2019) documented an outbreak of Diplodia tip blight on Pinus sylvestris stands in Sweden in 2016 caused by D. pinea and attributed it to the increased summer temperatures. An influence of different climatic conditions on distribution patterns of Botryosphaeriales species has also been observed in the USA and Australia (Taylor et al. 2005, Úrbez-Torres et al. 2006, Pitt et al. 2010). However, a climatical or geographical explanation in general can be ruled out, since species of this order have been detected from fruit trees and grapevine in Central Europe before, even in different parts of Germany, including a report of the same two species from P. armeniaca (Trapman et al. 2008, Quaglia et al. 2014, Fischer et al. 2016, Gierl and Fischer 2017. Based on the results in this study, species of Botryosphaeriaceae are currently not regarded as a threat for German Prunus orchards. With 14 species, Phaeoacremonium was the genus with the highest diversity in the study on Prunus wood in South Africa (Damm et al. 2008b), while only four Phaeoacremonium species were isolated in Germany (this study). Three of them were isolated in both studies, namely Pm. iranianum, Pm. scolyti and Pm. viticola, provided the identification of the latter, which was with low certainty (cf.), is correct. Although the genera were found in Prunus wood in both countries, completely different species of Coniochaeta (Coniochaetales, Sordariomycetes), Calosphaeria, Jattaea (Calosphaeriales, Sordariomycetes), Paraconiothyrium/ Paraphaeosphaeria (Pleosporales, Dothideomycetes) and Phaeomoniellales (Eurotiomycetes) were collected in Germany and in South Africa (Damm et al. 2008a, c, 2010. The latter order was much more diverse and frequent in Prunus wood in South Africa; in Germany, only two Minutiella species were collected. In contrast, Cadophora species were more frequently detected in wood of different Prunus species in Germany, but only rarely detected in South Africa; only Ca. prunicola was collected in Prunus wood in both countries . Collophorina (syn. Collophora) and Pallidophorina species were isolated frequently in Prunus wood both in South Africa and in Germany . The dominating Collophorina species isolated from several Prunus species in South Africa was C. rubra, a species not reported from Germany, while the dominating one in Germany was C. africana (syn. Collophora capensis). The latter was originally found exclusively on wood of P. salicina in South Africa, while in our study, it was exclusively present on P. domestica. In the study by Damm et al. (2010), Pa. paarla (syn. C. paarla, Collophora pallida) was mostly isolated from P. salicina in South Africa, while this species was one of the two dominating species in this survey occurring in all Prunus species studied (Bien et al. 2020, this study). The Collophorina species isolated from P. dulcis wood in Spain (Gramaje et al. 2012), C. hispanica, was not found in our study. Gierl and Fischer (2017) isolated Pa. paarla from symptomatic wood of P. cerasus and P. persica, as well as C. hispanica and C. africana from P. armeniaca and P. dulcis, respectively.
Although five species of Diatrypaceae were collected in the surveys in Germany and South Africa, Eutypa lata was the only species found in both of them, in wood of P. cerasus and P. domestica in Germany, as well as in P. armeniaca, P. avium, P. dulcis and P. salicina in South Africa (Moyo et al. 2018, this study). It was also found in wood of P. dulcis in Mallorca (Gramaje et al. 2012). Furthermore, Diaporthe species have been isolated in all three studies as well. Based on preliminary studies, none of the species is overlapping with those found in this study (Gramaje et al. 2012, U. Damm, unpubl. data). The remaining taxa cannot be compared as no data were published from the survey in South Africa.

Function of the fungal species inside wood
Only for part of the species/genera isolated in this study information on lifestyle, like pathogenicity on Prunus species, is available. In the survey on Prunus wood in South Africa, preliminary pathogenicity tests on detached shoots revealed t h e m a j o r i t y o f t e s t e d s p e c i e s b e l o n g i n g t o Botryosphaeriaceae, Celotheliaceae, Coniochaetaceae, Togniniaceae and Tympanidaceae to be potentially Fig. 4 Phylogeny of dataset 2 obtained by Bayesian inference analysis of the combined LSU and ITS sequence alignment of Dothideomycetes. Penicillium resticulosum strain CBS 609.94 is used as outgroup. BI posterior probability support values above 0.9 (bold) and ML bootstrap support values above 70% are shown at the nodes. The strains isolated in this study are emphasised in bold. Numbers in parentheses indicate the number of isolated strains per taxon. Branches that are crossed by diagonal lines are shortened by 50%. T, ex-type strain; #, type species pathogenic to P. persica var. nucipersica and/or P. salicina (Damm et al. 2007a(Damm et al. , 2008b. Species of all these families have been isolated in this study as well. However, apart from the fact that these pathogenicity tests were preliminary and not followed up by field tests, these results cannot be transferred to this study, because most of the fungal species isolated were different, and even the few species isolated in both studies, for example Pa. paarla and C. africana, were not isolated from the same Prunus species. Therefore, the pathogenicity of each fungal species isolated in this study would need to be tested on its host species in Germany. As we aimed at isolating pathogens causing necroses inside Prunus wood, the majority of wood pieces we isolated from were from the transition zone of symptomatic to non-  Fig. 5 Phylogeny of dataset 3 obtained by Bayesian inference analysis of the combined LSU and ITS sequence alignment of Leotiomycetes. Colletotrichum godetiae strain CBS 133.44 is used as outgroup. BI posterior probability support values above 0.9 (bold) and ML bootstrap support values above 70% are shown at the nodes. The strains isolated in this study are emphasised in bold. Numbers in parentheses indicate the number of isolated strains per taxon. Branches that are crossed by diagonal lines are shortened by 50%. T, ex-type strain; #, type species symptomatic wood tissue. From most of the wood samples, we isolated several fungi. Wood diseases are caused by a complex of fungal pathogens, which is known from grapevine trunk diseases like esca and Botryosphaeria dieback Dubos 1997, Bertsch et al. 2013). Therefore, more than one of the isolated fungi could be responsible for the symptoms on the respective branch. Moreover, transitions between different lifestyles have been shown in a high number of fungi (Promputtha et al. 2010, Álvarez-Loayza et al. 2011, Eaton et al. 2011, O'Connell et al. 2012, Kuo et al. 2014. As an example, many of the wood-inhabiting fungi, including Botryosphaeriaceae, are known as weak pathogens: they do not cause symptoms and live inside their host endophytically and become pathogenic, if the host plant is exposed to stress, e.g. drought (Desprez-Loustau et al. 2006, Slippers andWingfield 2007). However, not only the presence of one or more pathogens decides, if a disease develops, but also the absence of other fungi or other organisms that prevent the disease and keep the tree healthy. Thus, in a study of Gennaro et al. (2003), the endophytic communities on declining oaks were less diverse than those on healthy trees, and endophyte communities of needles of Norway spruce have been proposed as indicators of tree health (Rajala et al. 2014). It is therefore hardly possible to draw conclusions concerning the particular role of the individual species within the temporal-spatial succession of fungal communities associated with wood necroses of Prunus trees in Germany.

Helotiales
We isolated fungi both from the transition zone of symptomatic to non-symptomatic tissue and from non-symptomatic tissue of the same branch providing that the sole isolation of a certain species from one of the two zones would indicate a certain life style, e.g. the sole isolation from non-symptomatic tissue would indicate an endophytic life style. However, the resulting data are not directly comparable, because the number of wood pieces of the non-symptomatic tissue of a branch with wood symptoms studied was lower than the number of wood pieces from the transition zone of symptomatic to nonsymptomatic wood tissue. Moreover, in some branches, little non-symptomatic tissue was available due to the large expansion of the necroses and the "symptomless tissue" placed on OA for isolation was very closely located to the necrotic tissue. Biggs et al. (1983) detected hyphae of Cytospora chrysosperma up to 2 cm away from xylem tissue of Populus with visible necroses caused by this fungus. Therefore, isolation of a pathogenic fungus from nearby symptomless tissue cannot be excluded and a similar fungal diversity and abundance was expected. However, the number of fungi isolated from non-symptomatic tissue was exceptionally low compared to that from the transition zone of symptomatic to non-symptomatic tissue. The fungi isolated solely from symptomless tissue were isolated only once. And none of the few species isolated more often from symptomless than from symptomatic tissue was found more than five times in total. This cannot be explained by the lower subsample number of non-symptomatic wood pieces. We attribute this to the larger number of ecological niches of the wood pieces from the transition zone resulting in a temporal-spatial succession of fungal communities including endophytes, pathogens and saprobionts.

Uncertainties in identifications
Of the 172 species isolated from Prunus wood in this study, 102 could be assigned to a particular species with different levels of certainty. The ITS region of many species is highly variable, which decreases the similarity values and results in unjustified uncertainty (Nilsson et al. 2008, Simon and Weiß 2008, Hughes et al. 2009). In contrast, ITS sequences of closely related species can be identical or nearly so, which results in similarity values of up to 100% and therefore unjustified certainty (e.g. Houbraken et al. 2011, Damm et al. 2019). This means, on the one hand, some of the 70 taxa not assigned to a species could possibly be identified to species level by including the whole variability of the ITS sequences of the respective genus. On the other hand, some of the 102 species that were assigned to a particular species with high or low (cf.) certainty, even of those with identical ITS sequences, could represent new species or species with no sequence data in GenBank. This demonstrates how imprecise an identification based on solely ITS data is, even by availability of full-length sequences, careful selecting the reference data and inclusion of the nucleotide differences/identities. The different inter-and intraspecific variability of ITS sequences is a dilemma of species identification in fungal diversity studies dealing with big and diverse sampling datasets. It is simply not possible to study the variability of each taxon and consider DNA variability of each species while defining a uniform threshold for species differentiation. On the one hand, a rigorous application of strict criteria for species delimitation ignores the variability of different fungal taxa. On the other hand, if no clear criteria are applied, species delimitation is to a certain degree subjective and the different reasons for a specific decision hard to compare.  Moreover, blastn searches often not only result in uncertain but more importantly in wrong identifications due to the sequence data in NCBI GenBank, of which many are incomplete, include artefacts, are mixed up or derived from wrongly identified samples and therefore not suitable as reference data (Vilgalys 2003, Nilsson et al. 2006, Bidartondo 2008, Ko et al. 2011. Therefore, only sequences of extype strains can be reliable references. However, even sequences of ex-type strains can be unreliable, if they are based on sequences with low quality or mixed up with other species, as revealed for example in Colletotrichum hymenocallidicola (Damm et al. 2019). Nonetheless, the main drawback of identification based on sequence data (in GenBank) is the limited part of the overall known fungal diversity with available sequence data, especially those from type material. It is possible that the number of species that were not identified to species level and regarded as new species is lower. That means, some of these 70 potentially new species as well as some of those with uncertainty identified species could represent species that had previously been described based on morphology, however, lacking sequence data in GenBank.

Species identified with high certainty
In total, 82 taxa (630 strains) were identified to species level with certainty. All species belonging to the genera Arboricolonus, Cadophora, Collophorina, Pallidophorina and Proliferodiscus in the Leotiomycetes and Minutiella in the Eurotiomycetes have previously been studied in detail morphologically as well as by multi-locus phylogenetic analyses . Therefore, their identifications are reliable. Three and two species of Cadophora and Collophorina, respectively, and one species each of Proliferodiscus and Minutiella, as well as the genus Arboricolonus have been described in these two previous studies based on strains isolated from Prunus wood in Germany collected within our survey.
Strain GLMC 380 belonging to the Dothideomycetes shows that also strains assigned to a species with certainty can actually be of uncertain systematic position. The sequence of this strain (representing further 23 strains) was identical with that of a strain referred to as Coniothyrium ferrarisianum (CBS 285.74). Both strains form a clade sister to a clade formed by two ex-type strains of Sclerostagonospora. Other strains of Coniothyrium, Co. dolichi, Co. glycines and Co. telephi, formed a distant clade also within the Pleosporales. There is no DNA sequence of the type species Co. palmarum available; the genus Coniothyrium is currently regarded as polyphyletic (Verkley et al. 2004(Verkley et al. , 2014. Therefore, the systematic placement of the genus Coniothyrium as well as of the individual species, including Co. ferrarisianum, still needs to be clarified.

Species identified with low certainty
In total, 20 taxa (98 strains) were assigned to a species with low certainty, because the ITS sequences differed in 5-10 nucleotides from the closest named reference sequences. These are taxa that need to be studied in depth; species boundaries need to be evaluated, etc. It is possible that part of these taxa represent new species. Even the affiliation of some of the taxa to genus level still needs to be clarified, for example Anthostomella cf. pinea strain GLMC 451 (see below).

Taxa identified to genus level
In total, 57 taxa (255 strains) were assigned to a genus, but not to a species. Of these, 16 taxa (86 strains) matched with more than one named reference sequence (cf. spp.); these fungi are unlikely to represent new species and can probably be identified to species level based on secondary barcodes. It is also possible that one of the species in the respective clade represents a synonym that has previously not been revealed yet.
The ITS sequences of the other 41 taxa (169 strains) differed in > 10 nucleotides from the closest named reference sequence. Most of these taxa represent new species, unless the species was described only based on morphology and no ITS sequence is available.

Species not identified to genus level
Thirteen species could not be identified to genus level (35 strains), because they did not match with named reference sequences in blastn searches and were placed isolated within the phylogenies (e.g. Leotiomycetes sp. GLMC 792, Lecanoromycetes sp. GLMC 1733) or because the respective genus is polyphyletic and sequences of the type species are either not available or belong to a different clade within the phylogeny. The 13 taxa were therefore identified to family (six taxa), order (five taxa) or class (two taxa) level only; most of them belong to the Sordariomycetes.
Although some of the closest matches in blastn searches with the ITS sequences of strains GLMC 1660 (Xylariaceae sp. 1) and GLMC 1594 (Xylariaceae sp. 2) were strains previously identified as Rosellinia sp., we doubt these taxa belong to this genus, because sequences of ex-type strains of two species and of a strain of the type species, R. aquila (Wendt et al. 2018), belong to different clades. Affiliation of the strains isolated in this study to the genus Rosellinia cannot be clarified with the data at hand.
Strain GLMC 848 (Xylariaceae sp. 3) is placed together with two strains from Juniperus deppeana in the USA referred to as Sordariomycetes sp. (Hoffman and Arnold 2010). The clade formed by these strains is sister to a clade formed by strain GLMC 451 (Anthostomella cf. pinea, this study) and the ex-type strain of Anthostomella pinea (CBS 128205). However, the genus Anthostomella is polyphyletic (Daranagama et al. 2015), which is confirmed here as the ex-type strain of another species, An. proteae (CBS 110127), belongs to a different clade. None of these clades was confirmed to represent the genus Anthostomella, because there is no sequence of the type species of the genus, An. limitata, available. Therefore, the affinity of both strains, GLMC 848 and GLMC 451, to Anthostomella is unclear.
Strain GLMC 1232 (Sordariales sp.) groups with a strain referred to as Cercophora sp. (CIM1_17, Mapperson and Dearnaley, unpubl. data), an uncultured Ascomycota (dfmo0690_036) from soil in the USA (O'Brien et al. 2005) and the ex-type strain of Zopfiella tardifaciens (CBS 670.82). A strain of the type species of Zopfiella, Z. tabulata (CBS 230.78), is placed in a singlestrain clade sister to this group. The intergeneric relationships of Lasiosphaeriaceae genera including Zopfiella and Cercophora were described as inconclusive due to the uncertainty about the phylogenetic value of different morphological characters (Cai et al. 2005).
Strain GLMC 1316 (Pleosporales sp.) clustered with two strains referred to as Leptosphaeria sp. (LCC1-2, Li et al. unpubl.;LQ122417, Qiong et al., unpubl. data) that are distant from a clade formed by strains of three further Leptosphaeria species, none of which are ex-type strains. The affiliation of the isolated strain to this genus is therefore doubtful.
The ITS sequence of strain GLMC 1563 (Lentitheciaceae sp.) is identical with that of a strain previously identified as Sclerostagonospora cycadis (CBS 291.76). Both strains form a clade sister to a clade formed by two ex-type strains of Murilentithecium species, including the type species of the genus. As the ex-type strain of S. cycadis (CBS 123538) belongs to a different clade within the Pleosporales, sister to the ex-type strain of S. ericae, strain CBS 291.76 must have been wrongly identified. Both strains are likely to be a Murilentithecium species, which needs to be confirmed.
Strain GLMC 792 (Leotiomycetes sp.), belonging to the Leotiomycetes, grouped with strain 30404-E that had been isolated from wood in Greenland and identified as Pseudeurotium sp. (Pedersen et al., unpubl. data). However, the placement in this genus is doubtful, because this clade is distant from the Pseudeurotium clade formed by three ex-type strains including the type species of the genus.
This study highlights that a common substrate like wood of fruit trees in Germany actually represents an underexplored habitat and houses a widely unknown mycobiome with widely unknown host spectrum/specificity, distribution, conservation status, life cycle and function and probably large potentials for applications. We expect most of the taxa not assigned to a species and part of the species identified with more or less certainty to represent new species or even new genera. In order to clarify their identity, these species should be treated in depth in further follow-up studies by a polyphasic approach consisting of multilocus sequence analyses and sound morphological examinations.
Availability of data and material The DNA sequences generated in this study were deposited in GenBank (Table 1, suppl. material tab.). The datasets generated and analysed during the current study are available from the TreeBASE website, http://purl.org/phylo/treebase/phylows/ study/TB2: S25316.
Authors' contributions Both authors have contributed equally. Both authors read and approved the final manuscript.
Funding information Open Access funding provided by Projekt DEAL. This study contributes to the German Barcode of Life project, funded by the Federal Ministry of Education and Research of Germany (www. bolgermany.de).

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflicts of interest.

Consent for publication Not applicable
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.