Phylogeny and morphology of Lasiodiplodia species associated with Magnolia forest plants

Two new species of Lasiodiplodia (Lasiodiplodia endophytica and Lasiodiplodia magnoliae) are described and illustrated from Magnolia forests in Yunnan, China. Endophytic and saprobic Lasiodiplodia pseudotheobromae and endophytic L. thailandica are new records from this host. The internal transcribed spacers (ITS), part of the translation elongation factor-1α (tef1) and partial β-tubulin (tub2) sequence data were analyzed to investigate the phylogenetic relationships of the new species with other Lasiodiplodia species. Lasiodiplodia magnoliae is phylogenetically sister to L. mahajangana and L. pandanicola but morphologically distinct from L. mahajangana in having larger conidia. Lasiodiplodia endophytica is most closely related to L. iraniensis and L. thailandica and the three species can be distinguished from one another by 2 base pair differences in ITS and three or four base pair differences in tef1. The new collections suggest that Magnolia forest plants are good hosts for Lasiodiplodia species with endophytic and saprobic life-styles.

stage is a unique character for Barriopsis 7,9 . Another example is the genus Neodeightonia that shows close affinity to Lasiodiplodia in having striations pigmented mature conidia and can be differentiate from Lasiodiplodia by the absence of conidiomatal paraphyses 7,9 . However, on account of morphological variability within species, morphology alone is not reliable for distinguishing different Lasiodiplodia species. Phillips et al. 4 suggest that combined LSU and ITS provide reliable resolution for phylogeny of Botryosphaeriales. However, protein coding genes such as tef1 and tub2 in addition to LSU and ITS provide greater support for species and genera level delimitation in order Botryosphaeriales 4 . In previous studies, phylogenetic analyses were solely based on ITS nucleotide sequences 3 to identify Lasiodiplodia species. Inclusion of tef1 sequences gives better resolution of phylogenetic relationships among species 3,6 . The recent multi locus phylogenetic approaches with ITS, tef1 and tub2 nucleotide sequence data has advanced the recognition of numerous Lasiodiplodia species with high phylogenetic support 3,4,10,11 . In that respect, sequence data of the internal transcribed spacers (ITS), partial translation elongation factor-1α (tef1) and partial β-tubulin (tub2) are now relied on for resolution of species in Lasiodiplodia 11 . There are 55 epithets of Lasiodiplodia recorded in Index Fungorum (March 2019) and 43 species names in MycoBank (March 2019). However, cultures and DNA sequence data are available for only 35 species (March 2019) [10][11][12] .
Species of Lasiodiplodia are cosmopolitan in tropical and subtropical regions and occur on a wide range of monocotyledonous, dicotyledonous and gymnosperm hosts 2,3,6,8,13 . They exhibit diverse life-styles as endophytes, inhabiting different asymptomatic plant tissues 8,14,15 , pathogens that cause diseases in various plant hosts 3,16 and saprobes that are commonly found on dead woody plant tissues 3,17 .
To the best of our knowledge there have been no studies on the Lasiodiplodia species associated with Magnolia species in Yunnan Province, China. The aim of this study was to characterize Lasiodiplodia isolates in terms of morphology and phylogeny based on ITS, tef1 and tub2 sequence data.

Results
phylogenetic analyses. The combined dataset of ITS, tef1 and tub2 consisted of 54 taxa of Lasiodiplodia, with Diplodia mutila (CMW 7060) as the out group taxon and comprised 1267 characters including gaps after alignment. Of these, 1011 were conserved and 123 variable characters were parsimony uninformative. Maximum parsimony analysis of the remaining 133 parsimony informative characters resulted in 1000 equally parsimonious trees of 535 steps with CI = 0.632, RI = 0.798, RC = 0.504 and HI = 0.368. The maximum likelihood analysis resulted in a tree with largely the same topology as the maximum parsimony tree. The RAxML analysis yielded a best scoring tree with the final ML optimization likelihood value of -4851.693940 (ln). Estimated base frequencies were as follows; A = 0.209292, C = 0.303982, G = 0.256083, T = 0.230643; substitution rates AC = 1.189236, AG = 3.165454, AT = 1.301265, CG = 1.047358, CT = 4.430504, GT = 1.000000; gamma distribution shape parameter α = 0.612671 (Fig. 1).
Four of the isolates from Magnolia clustered with known species, three with Lasiodiplodia pseudotheobromae and one with L. thailandica. The remaining two isolates formed distinct lineages representing two new species. Lasiodiplodia magnoliae was clustered separately and sister to L. mahajangala and L. pandanicola and L. endophytica formed a separate lineage and sister to L. iraniensis and L. thailandica. Notes. The combined ITS, tef1 and tub2 phylogeny showed that Lasiodiplodia magnoliae (MFLUCC 18-0948) clades sister to L. mahajangana and L. pandanicola with low support (53% ML, 60% MP) (Fig. 1). DNA sequence comparisons of ITS and tef1 among L. magnoliae, L. mahajangana and L. pandanicola are given in Table 1. Comparison of total length of 445 bases of ITS sequences revealed one base pair difference among L. magnoliae, L. mahajangana and L. pandanicola. Comparison of total length of 450 bases of tef1 sequences revealed an insertion of eight bases in Lasiodiplodia magnoliae when compared to L. mahajangana and L. pandanicola (Table 1). Lasiodiplodia magnoliae has longer conidia (24-30 μm) than L. mahajangana (14-24 μm) 18 . Lasiodiplodia pandanicola has overlapping range of conidial dimensions (14-38 μm) 10 with L. magnoliae. Lasiodiplodia magnoliae has longer paraphyses (60-70 μm) than L. mahajangana (27- Notes. According to the combined ITS, tef1 and tub2 phylogeny, two isolates NI173 and NI173A from M. candolii twigs clustered with Lasiodiplodia pseudotheobromae with low support (51% ML, 52% MP) (Fig. 1). Conidia of these two isolates were hyaline, (20-26 × 10-14 μm) and brown (19-25 × 12-15 μm) and thus are smaller than in the ex-type isolate (27.5-28.5 × 15.5-16.5 μm) 6 . One endophytic strain (C13) from the same M. candolii plant www.nature.com/scientificreports www.nature.com/scientificreports/ was phylogenetically closely related to L. pseudotheobromae and clustered with two saprobic strains. Conidial dimensions of the endophytic isolate (26-31 × 10-12 μm) overlap with those of the ex-type isolate. The type of L. pseudotheobromae was isolated from Gmelina arborea in Costa Rica and has also been recorded from Citrus sp., Coffea sp. 3 , Pteridium aquilinum 19 , and Plukenetia volubilis 20  Notes. Phylogenetically, the new isolate clustered with the ex-type isolate of Lasiodiplodia thailandica (CBS 138760) based on combined ITS, tef1 and tub2 sequence data. However, the new isolate has larger conidia (28-29 × 11-13 μm) than the ex-type of L. thailandica (20-26 × 12-16 μm) 21 . Lasiodiplodia thailandica was first described from symptomless twigs of Mangifera indica in Chiang Mai province, Thailand 21 and also has been recorded from a petiole of Phyllanthus acidus in Thailand 20 , from cankered branch of Podocarpus macrophyllus in China 19 and from cankered branch of Albizia chinensis in China 19 .   Notes. The combined ITS, tef1 and tub2 phylogeny showed that Lasiodiplodia endophytica (S8) (MFLUCC 18-1121) clusters sister to Lasiodiplodia iraniensis. DNA sequence comparisons of ITS and tef1 among L. endophytica, L. iraniensis and L. thailandica are given in Table 2. Comparison of total length of 477 bases of ITS sequences revealed one base pair difference among three strains and one base deletion in L. endophytica. Comparison of total length of 290 bases of tef1 sequences revealed seven base pair differences among three strains as given in Table 2. This isolate did not sporulate in culture and no conidiomata were seen on the host. Therefore it was not possible to observe conidial characters. The type of L. iraniensis was isolated from twigs of Salvadora persica in Iran 3 . Additionally, L. iraniensis was recorded from twigs of Juglans sp. Citrus sp. and Mangifera indica in Iran 3 .

Discussion
In this study two new species of Lasiodiplodia were identified and described from Magnolia candolii in the southern part of Yunnan Province, China. One species (Lasiodiplodia magnoliae) was considered to be saprobic, while the other (L. endophytica) was thought to be endophytic. In addition, two saprobic isolates of L. pseudotheobromae from dead twigs and an endophytic isolate of the same species from fresh leaves of Magnolia candolii were recorded for the first time from China. An endophytic isolate of Lasiodiplodia thailandica was also isolated for the first time from fresh leaves of Magnolia candolii in China. When Promputtha et al. 22  Phylogenetic approaches based on DNA sequence data have played a significant role in distinguishing species in Lasiodiplodia 3,4,24,25 . Previous studies have used combined ITS and tef1 regions to clarify the taxonomy and phylogenetic relationships of species in Lasiodiplodia 3,6,26 while others have used combined ITS, tef1, tub2 and rpb2 19 . The current phylogenetic analyses with combined ITS, tef1 and tub2 sequence data gave good resolution of phylogenetic separations among Lasiodiplodia species and provide insights in to taxonomic novelties. Unfortunately, www.nature.com/scientificreports www.nature.com/scientificreports/ amplification of tef1 of MFLUCC 18-0951 -C13 and MFLUCC 18-0952 -C17, and tub2 of MFLUCC 18-0951 -C13 was not successful in this study. Phylogenetic analyses were conducted using DNA sequence data available in GenBank, but unfortunately sequences of tef1 and tub2 are not available for some species (see Table S1) and some of the sequences are shorter than expected. These issues of tef1 and tub2 might compromise the number of characters in the final alignment and ultimately might affect the final phylogenetic tree construction. We provide phylogenetic analyses for single molecular markers as Supplementary Materials. Phylogenetic trees from ITS and tub2 did not provided good resolution among Lasiodiplodia species. The phylogenetic analysis of ITS gene showed that three newly isolated strains of Lasiodiplodia pseudotheobromae and L. thailandica MFLUCC 18-095 clustered in one group and L. endophytica MFLUCC 18-1121 clustered separately from that group. Lasiodiplodia magnoliae MFLUCC 18-0948 formed a separate clade with L. citricola IRAN1522C. The phylogenetic analysis of tub2 did not provide clear separation of newly isolated strains and previously described species. Two Lasiodiplodia pseudotheobromae strains, L. thailandica and L. endophytica clustered together and L. magnoliae formed a separate, distantly related lineage. Analysis of tef1 resulted in a better resolution of many taxa than single ITS and tub2 gene trees and showed similar phylogenetic relationships as combined ITS, tef1 and tub2 analyses. In both tef1 gene and combined ITS, tef1 and tub2 gene phylogenetic analyses, new strains of Lasiodiplodia pseudotheobromae formed a clade with other L. pseudotheobromae, L. endophytica clustered with L. iraniensis, L. thailandica and L. hyalina. In both tef1 gene and combined phylogenetic analyses, Lasiodiplodia magnoliae reflected similar phylogenetic affiliation with L. chonburiensis, L. caatinguensis, L. exigua, L. pandanicola and L. mahajangana. It is also worth noting that the phylogenetic relationships of species within Lasiodiplodia recovered herein from combined ITS, tef1 and tub2 gene analyses are similar to previously established ones in Dissanayake et al. 12 , Dou et al. 11 and Tibpromma et al. 10 . These three phylogenetic studies were based on different combinations of molecular markers such as Dissanayake et al. 12 who used combined ITS and tef1, Dou et al. 11 used combined ITS, tef1, tub2 and rpb2 and Tibpromma et al. 10 used ITS, tef1 and tub2. In earlier studies, Lasiodiplodia were species distinguished solely on their ITS sequences 10 . In recent studies, taxonomists frequently use highly variable protein coding genes such as tef1, tub2 together with ITS to construct phylogenies especially at species levels 4 . It can be assumed that these combination of molecular markers strengthen the support for them and to separate the existing ones 3,4 . www.nature.com/scientificreports www.nature.com/scientificreports/ Conidia of Lasiodiplodia species are initially hyaline, aseptate, ellipsoid to ovoid and become pigmented, 1-septate with longitudinal striations 3 . We observed hyaline, aseptate conidia and brown, 1-septate conidia with longitudinal striations in the saprobic isolates of Lasiodiplodia pseudotheobromae, but only hyaline conidia were seen in Lasiodiplodia magnoliae, the endophytic isolate of L. pseudotheobromae and L. thailandica. Previous studies have recorded both hyaline and pigmented conidia in L. pseudotheobromae 6 and L. thailandica 21 . Other Lasiodiplodia species have been observed with only hyaline conidia such as L. chonburiensis 10 , L. sterculiae 27 and L. thailandica in which most conidia were hyaline and only 10% were brown 19 . This character does not seem to be restricted to any particular phylogenetic groups but appears in different Lasiodiplodia species. Thus, L. magnoliae and L. chonburiensis are closely related and found in one clade. On the other hand, L. sterculiae and L. thailandica are distantly related to both L. magnoliae and L. chonburiensis and formed widely separate lineages in the phylogenetic tree. Lasiodiplodia magnoliae differs from its sister taxa by phylogeny, morphology, host species and locality as described in the notes section that support for the introduction of new saprobic taxa. We were unable to observe conidia of L. endophytica (S8) in culture even after many attempts on different media. It was considered here that phylogeny based on combined ITS, tef1 and tub2 sequence data provides sufficient evidence for the designation of L. endophytica (S8) as a novel taxon.
Lasiodiplodia species exhibit diverse life-styles as endophytes 8,15 , pathogens 3,16 and saprobes 3,17 . Lasiodiplodia species with endophytic life-styles are associated with different asymptomatic plant tissues such as L. avicenniae from asymptomatic branches of Avicennia marina in South Africa, L. bruguierae from asymptomatic branches of Bruguiera gymnorrhiza in South Africa 28 and L. mahajangana from healthy branches of Terminalia catappa in Madagascar 18 . We isolated three endophytic species; Lasiodiplodia endophytica, L. pseudotheobromae and L. thailandica from asymptomatic leaves of Magnolia candolii. Interestingly, we isolated one endophytic and 2 saprobic www.nature.com/scientificreports www.nature.com/scientificreports/ isolates of Lasiodiplodia pseudotheobromae. This might be possible because endophytes switch their nutritional mode to saprobic when environmental conditions become unfavorable to the host or during host senescence 29 . Thus, de Errasti et al. 30 stated that diatrypaceous endophytic fungi switch to a saprotrophic life-style during host senescence. They explained that this might be ecologically important as they can decay the plant part when it dies 30 . In another scenario, Osorio et al. 28 showed that endophytic Lasiodiplodia avicenniae became pathogenic and caused lesions on the branches of Avicennia marina after inoculating. It is assumed that some fungi exhibit a continuum of life-styles ranging from biotrophy (or endophytic), through to necrotrophy and ultimately to saprotrophy 29 . Endophytes are a hidden bioresource of fungal diversity that have the potential to produce important bioactive agents 15 . Chen et al. 15 chemically investigated a strain of Lasiodiplodia sp. isolated from asymptomatic leaves of the medicinal plant Acanthus ilicifolius. They studied β-resorcylic acid derivatives and showed that these compounds showed more potent inhibitory effects against α-glucosidase activity than the clinical α-glucosidase inhibitor acarbose 15 . Lasiodiplodia species with pathogenic life-styles are associated with shoot blights, stem cankers, fruit rots, dieback, grapevine trunk diseases and gummosis 3,16,31 Lasiodiplodia exigua from a branch canker of Retama raetam 32 , L. mediterranea from branch canker of Quercus ilex 32 , L. plurivora from V-shaped necrotic lesion of Prunus salicina, in Africa 33 and L. pseudotheobromae from grapevine trunk disease 16 are some examples that cause different plant diseases. We did not observe any pathogenic Lasiodiplodia species in our study. Saprobic Lasiodiplodia species have been recorded such as Lasiodiplodia iraniensis on dead twigs of Salvadora persica, L. hormozganensis on Olea sp. 3 and L. theobromae on dead twigs of Eucalyptus sp. 17 . Similarly, we introduced a new saprobic species of Lasiodiplodia magnoliae and two isolates of L. pseudotheobromae.
This study identified Lasiodiplodia species in forest plants of Magnolia candolii in Yunnan, China. The study has expanded the knowledge of Lasiodiplodia species providing two novel species and two new host records. It might be possible to identify new distribution and host associations of Lasiodiplodia species from other forest plants in the world. It is important to study endophytic Lasiodiplodia species as well as pathogenic and saprobic life-styles as novel endophytes are also yet to be explored. Lasiodiplodia endophytica

Materials and Methods
The study area was a sub-tropical rain forest inside the Xishuangbanna tropical botanical garden in Xishuangbanna at 21°55′N, 101°15′E, Yunnan province, China. Elevation ranges from 709-869 m and mean temperature and precipitation are 21.0 °C and 1532 mm respectively. The wet season is from May to October while the dry season is from November to April 34,35 .
Isolation of fungal endophytes. Fresh leaves of Magnolia candolii were collected randomly on 26 th of April 2017. The leaves were kept at 4 °C in sterile polyethylene bags until they were processed in the laboratory. Isolation of endophytes was done according to the methods described by Promputtha et al. 36 with modifications. First leaves were washed using tap water and cut in to small pieces of leaves (5 × 5 mm 2 ) and soaked in distilled water for 1 minute and then surfaced sterilized by dipping in 70% alcohol followed by 2% NaOCl for 30 s and finally washed with sterile distilled water for 30 s, dried and plated on Potato Dextrose Agar (PDA). Fungi were isolated into pure culture and grouped according to their culture morphology.
Isolation of fungal saprobes. Fungi  DNA extraction, PCR amplification and phylogenetic analysis. Mycelium was grown on PDA for one week at 25 °C in normal light in the laboratory. Genomic DNA was extracted from the mycelium using a Biospin fungus genomic DNA kit (BioFlux ® , P.R. China) following the manufacturer's protocol. DNA was kept at 4 °C for DNA amplification and maintained at −20 °C for long term storage.
The internal transcribed spacer (ITS) was amplified with primer pair ITS4 and ITS5 40 as described in Alves et al. 41 . Part of the translation elongation factor (tef1) was amplified with primer pair EF1-728F and EF1-986 Carbone and Kohn 42 and EF1-688F and EF1-1251R Alves et al. 6 . The partial β-tubulin (tub2) was amplified with primer pair Bt2a and Bt2b 43 . The expected sequence lengths are approximately 500 bp, 300-400 bp, 400 bp for ITS, tef1 and tub2 respectively. Quality of PCR products was checked on 1% agarose electrophoresis gels stained with ethidium bromide. The amplified PCR fragments were sequenced by Sangon Biotech (Shanghai) Co., Ltd, P.R. China.
Newly generated nucleotide sequences were deposited in GenBank (Table S1 in Supplementary material). Sequences of the individual loci of ITS, tef1 and tub2 were aligned with MAFFT v. 7 online version 44 using default settings. BioEdit v. 7.0.5.2 45 was used to refine the alignments manually where necessary and to exclude incomplete portions at the ends of the sequences before the analyses.
Maximum likelihood analysis was performed with RAxML GUI v. 1.3 46 and maximum parsimony analysis was done with PAUP (Phylogenetic Analysis Using Parsimony) v. 4.0b10 47 . Evolutionary models for phylogenetic analyses were selected independently for each locus using MrModeltest v. 3.7 48 under the Akaike Information Criterion (AIC). GTR + I + G model of nucleotide substitution was selected for the maximum likelihood (ML) analysis. Parameters for maximum likelihood were set to rapid bootstrapping and the analysis carried out using 1000 replicates. Maximum parsimony was run with the heuristic search option, random taxon addition, tree bisection-reconnection (TBR) for the branch swapping algorithm and 1000 random sequence additions, with maxtrees set at 1000. Gaps were treated as missing data.