Phylogenetic lineages in the Botryosphaeriales: a systematic and evolutionary framework

PCR and sequencing

A total of six partial gene portions were used in this study: the nuclear ribosomal small subunit (SSU), the nuclear ribosomal large subunit (LSU), the intergenic spacer (ITS), the translation elongation factor 1-alpha (EF1), the β-tubulin gene (BT) and the mitochondrial ribosomal small subunit (mtSSU).

The primers used were NS1 and NS4 (White et al. 1990) for SSU, LROR and LR5 (Vilgalys Laboratory, Duke university, www.biology.duke.edu/fungi/mycolab/primers.htm) for LSU, ITS-1 and ITS-4 (White et al. 1990) for ITS, EF-AF and EF-BR (Sakalidis et al. 2011) for EF1, BT2A and BT2B (Glass & Donaldson 1995) for BT and mrSSU1 and mrSSU3R (Zoller et al. 1999) for mtSSU. All PCR reactions were conducted in 15 μL containing 1.5 mM of MgCl₂, 0.5 mM of dNTP, 1 × final concentration of buffer, 1 μM of each primer, 0.25 U of FastStart Taq Polymerase (Roche), 1.5 μL of DNA template and Sabax sterilised water (Adcock Ingram) to complete up to 15 μL. The cycling parameters were as follows: a first step of denaturation at 95 °C for 5 min followed by 35 cycles of (i) denaturation at 95 °C for 60 s, (ii) annealing at optimal temperature (55 °C for ITS, EF1, LSU and 45 °C for SSU, mtSSU, BT) for 80 s, (iii) elongation at 72 °C for 90 s, and a final elongation step of 5 min was applied.

Sephadex columns (Sigma-Aldrich) were used to clean the samples both before and after the sequencing reactions. The sequencing PCRs were performed in 10 μL containing 1 μL of PCR product, 0.7 μL of Big Dye Terminator v. 3.1 (Applied Biosystems), 2.5 μL of sequencing buffer (provided with Big Dye), 1 μL of primer (10 μM) and 4.8 μL of Sabax sterilised water. Cycling parameters consisted of 25 cycles with three steps each: 15 s at 95 °C, 15 s at 55 °C (for ITS, EF1, LSU) or 45 °C (for SSU, mtSSU, BT) and 4 min at 60 °C. The sequencing PCR products were sent to a partner laboratory for sequencing of both strands (Sequencing Facility, University of Pretoria).

Data analyses

Sequences were aligned using MAFFT v. 6 online (Katoh & Toh 2008) and refined visually. In order to retrieve the genetic information from indels, GapCoder (Young & Healy 2003) was used to code the gaps contained in the EF1 and the mSSU alignments. Analyses were run both with gaps coded and not coded.

The datasets were combined because this is believed to increase phylogenetic accuracy (Bull et al. 1993, Cunningham 1997). The six phylogenies resulting from the six data sets were first inspected visually to check whether there were conflicts between the histories of the genes that would preclude the combination of data. Additionally, the Incongruence Length Difference (ILD) test (Farris et al. 1995a, b) also known as the partition homogeneity test was run using PAUP v. 4.0b10 (Swofford 2003).

MrAIC (Nylander 2004) was used to determine the best fit model of nucleotide substitutions for each gene. Phylogenetic relationships of the samples were investigated using both Maximum Likelihood (ML) and Bayesian Inference (BI). The ML analysis was conducted using PhyML online (Guindon et al. 2005). The reliability of each node was assessed using the bootstrap (Felsenstein 1985) resampling procedure (100 replicates).

The BI was conducted using the software Beast v. 1.7.4 (Drummond et al. 2012). The phylogenetic relationships were estimated by running 10 000 000 generations and sampling every 100^th generation. Bayes Factors were computed to choose between the different options available in Beast (four clock models: strict clock, exponential or lognormal uncorrelated relaxed clocks, and random local clock and two tree priors: Yule process or Birth-Death speciation model). The Birth-Death speciation model and a relaxed uncorrelated exponential clock were selected as best fitting our data. Six independent runs were performed and outputs were combined using LogCombiner (in the Beast package). The programme Tracer v. 1.5 (available on the Beast website) was used to check that the effective sampling sizes (ESS) were above 200 (as advised by the programmers to ensure an accurate estimation of phylogeny and parameters of interest). The programme Tree Annotator (available in the Beast package) was used to summarize the resulting trees using the maximum clade credibility option. The final tree was visualised in FigTree v. 1.4.

Molecular clock dating

Using a mean molecular rate of 1-1.25 % per lineage per 100 million years, commonly accepted in fungi for the SSU gene (Berbee & Taylor 2010), a rough estimation of the times to the most recent common ancestors of groups of interest was assessed using Beast. A normal distribution was used as prior and this was centred on a 95 % interval spanning 1.0-1.25 % (mean = 0.000113; standard deviation = 0.000006). The dating of the distinct groups of interest and their 95 % highest posterior density (HPD) were retrieved using Tracer v. 1.5.

Molecular dating

Based on the models used, families split between 57 (28-100) - 103 (45-188) mya. Divergence within some families was also very ancient, such as the split between Pseudofusicoccum [65 (28-112) mya] and Endomelanconiopsis [52 (27-78) mya] and the rest of the Botryosphaeriaceae. The split between sub-clades 1 to 4 within the Botryosphaeriaceae was estimated to be between 33 (15-55) and 44 (25-64) mya.

Analysis of phylogeny-trait association

None of the five morphological traits that were investigated had a significant phylogenetic signal (Table 2; Fig. 2A-E).

Open in a separate window

Fig. 2.

Ancestral state reconstruction using parsimony and mapping the five traits onto the phylogenetic tree. Traits are coded as presence (in black) or absence (in white).

Botryosphaeriaceae

The Botryosphaeriaceae as it has been defined in this study includes the type genus Botryosphaeria (asexual morph Fusicoccum), as well as 16 other genera. This group corresponds to a group traditionally referred to as botryosphaeria-like. This term is, however, now understood to be much more restricted, including B. dothidea and a few closely related species (as defined in Slippers et al. 2004, Crous et al. 2006, Phillips et al. 2013, this volume). Using Botryosphaeria to refer to the assemblage of genera including Diplodia, Lasiodiplodia, Neofusicoccum and others, of which the sexual morphs were formerly described in Botryosphaeria, is thus taxonomically incorrect. These groups are now referred to using a single genus name, which is typically the asexual morph, irrespective of whether a sexual morph is known or not. This convention has been applied subsequent to the taxonomic changes introduced by Crous et al. (2006), and is also consistent with the recent decisions to abolish the dual nomenclatural system for fungal taxonomy (Hawksworth et al. 2011, Wingfield et al. 2012).

The data presented in this study, together with those emerging from more focused earlier studies, as well as the recent changes resulting in the abandonment of a dual nomenclature, necessitates reducing a number of genera in the Botryosphaeriaceae to synonymy with others. Most importantly, there is no longer just cause to maintain Botryosphaeria and Fusicoccum as distinct genera. In the interests of maintaining taxonomic stability, and the fact that B. dothidea is the type species of the order and family, it is recommended that Botryosphaeria be retained (Slippers et al. 2004, Schoch et al. 2006, Phillips et al. 2013, this volume). Botryosphaeria must thus be redefined to include species where only the asexual (Fusicoccum and dichomera-like) morphs are known. Species such as F. ramosum and Dichomera saubinetti must then be redefined in Botryosphaeria. For F. ramosum, an ex-type isolate was available and it has thus been redescribed in Botryosphaeria in a companion paper (Phillips et al. 2013, this volume).

Dichomera is polyphyletic and most likely also includes synasexual morphs of other genera. Two of the species included in this analysis, D. eucalypti and D. versiformis, clearly group in Neofusicoccum and we consider them as synonyms of species in this genus. Dichomera saubinetti grouped with Botryosphaeria and should be redescribed in this genus. Unfortunately no ex-type isolates of these species are presently available.

A number of genera grouped in Lasiodiplodia s. lat. in our analyses and are possibly synonyms of this genus. Macrovalsaria (see Sivanesan 1975) clearly grouped amongst species of Lasiodiplodia. This was also pointed out by Liu et al. (2012), but they did not find the available LSU and SSU data sufficiently convincing to make taxonomic changes. Isolates of Macrovalsaria however, grouped extremely closely with L. theobromae and we view them as representing a synonym of Lasiodiplodia, rather than Lasiodiplodia being polyphyletic. Lasiodiplodia and Macrovalsaria are both tropical fungi. Our analyses differ from those of Liu et al. (2012), indicating that Auerswaldia is a synonym of Lasiodiplodia. This was also confirmed in Phillips et al. (2013, this volume), who redescribed A. lignicola as L. lignicola. These findings suggest that the taxonomy of Lasiodiplodia needs to be re-evaluated, but for the present we do not recognize Auerswaldia as a genus in the Botryosphaeriaceae.

Diplodia juglandis and D. corylii both grouped in Dothiorella, which is consistent with the results of a previous study (Phillips et al. 2008). Type specimens were not available for these species and they were, therefore, not re-described. These species require epitypification. It is likely that a number of other Diplodia species will similarly reside in Dothiorella or Spencermartinsia, or vice versa, given the confusion of these names in the past (Phillips et al. 2005, 2008). The conidia of these genera remain difficult to distinguish, which can create problems when interpreting older descriptions or poorly preserved herbarium specimens.

There was no statistically significant pattern that could be discerned in the Botryosphaeriaceae with respect to the evolution of hyaline or pigmented conidia or ascospores. These characters appear to be more or less randomly spread amongst the clades of this section of the phylogenetic tree for the family. This would suggest that these characters predate the divergence of the genera in this family, and that they have been independently lost or suppressed (character not expressed under all conditions) in different groups. This would also explain the “appearance” of darker or even dark muriform conidia in genera such as Botryosphaeria and Neofusicoccum that were traditionally considered not to have such synasexual morphs (Barber et al. 2005, Phillips et al. 2005b, Crous et al. 2006). Clearly these characters, which have traditionally been commonly used for phylogenetic and taxonomic purposes, have very little phylogenetic and taxonomic value above the genus level.

The distinction and more narrow definition of the Botryosphaeriaceae is important when considering the economic and ecological importance of this group. Many of the Botryosphaeriaceae species share a common ecology in being endophytic and latent pathogens in virtually all parts of woody plants (Slippers & Wingfield 2007). While not all species have been isolated as endophytes, or from all plant parts, most of those that have been carefully studied have conformed to this pattern, and it is thus expected for the group as a whole. Many of the genera in the family are also very widespread, with wide host ranges and broad levels of environmental tolerance (e.g. N. parvum, N. australe, B. dothidea, L. theobromae, L. pseudotheobromae). Sakalidis et al. (2013) for example reported N. parvum from 90 hosts in 29 countries on six continents. This broad ecological range, together with their cryptic nature as endophytes, makes these fungi important to consider as a group prone to being spread with living plant material. Ample evidence exists that these fungi can infect both native and non-native trees, once they have been introduced into a region. The observed (Dakin et al. 2010, Piskur et al. 2011) and expected (Desprez-Loustau et al. 2006) increase of the importance of this group due to pressure on plant communities as a result of climate change provides another reason to focus future efforts on characterising the diversity, distribution and pathogenicity of this group of fungi.

Aplosporellaceae

An unexpected outcome of this study was the consistent connection between Aplosporella and Bagnisiella and their distinction from other members of the Botryosphaeriales. While it has been suggested that some Aplosporella spp. might be asexual morphs of Bagnisiella, this connection has never been proven. Neither of these genera were treated in molecular phylogenetic re-evaluations of the Botryosphaeriaceae until very recently. The first analyses to include DNA sequence data for Aplosporella (Damm et al. 2007, Liu et al. 2012) and Bagnisiella (Schoch et al. 2009) hinted at a distant relationship with other Botryosphaeriales. However, none of these studies included both genera. The phylogenetic relationship between these genera revealed in this study is further supported by their remarkably similar multiloculate sporocarps, expressed in both the asexual morphs and sexual morphs, which is thus not the product of parallel evolution in two distinct groups. There are, however, undoubtedly many species of Aplosporella and Bagnisiella that would not be connected to the phylogenetic clade identified here, because both genera are heterogeneous and likely contain unrelated species.

The data presented in this study suggest that Aplosporella and Bagnisiella are not only related, but that they should be synonymized. Both genera were described in 1880, and historical precedence can thus not be used to choose an appropriate genus. Aplosporella includes many more species (352) than Bagnisiella (65) (www.MycoBank.org, accessed August 2013). More species have also recently been described in the former genus, possibly because the asexual structures are more common than the sexual structures (as is found in other Botryosphaeriales). An argument based on taxonomic stability, relevance and frequency of occurrence is thus favoured and has led us to decide that Bagnisiella should be reduced to synonymy with Aplosporella.

Examination of the distribution of species of Aplosporella and Bagnisiella is complicated by the fact that the literature is old (which means the taxonomic accuracy is difficult to judge) and commonly lacking relevant information. However, most of the well-known and recently characterised species, and all species included here, are from the Southern Hemisphere (Damn et al. 2007, Taylor et al. 2009). This result suggests the possibility of a Southern Hemisphere and Gondwanan origin and divergence pattern, which should be considered in future studies based on more robust sampling.

Melanopsaceae

Melanops tulasnei and an undescribed Melanops sp. grouped most basal in the Botryosphaeriales, together with the Aplosporellaceae, Planistromellaceae and Saccharataceae. This group is unique amongst these families in having persistent mucous sheath around its ascospores and conidia (Phillips & Alves 2009). Conidia in M. tulasnei are typically hyaline and fusoid and it resembles the Aplosporellaceae in having multiloculate ascomata and conidiomata, often with locules at different levels. Little is known regarding the ecology and distribution of this group, given the paucity of recent reports that could be verified using DNA sequence data. It appears similar, however, to other Botryosphaeriales that infect woody tissue of plants, and sporulates on the dead tissue. Whether it is pathogenic or endophytic is not known.

Saccharataceae

Saccharata (the only genus in the Saccharataceae) grouped separately from all other families that were basal in the phylogenetic tree, suggesting a long, separate evolutionary history. The genus was first described by Crous et al. (2004) from Proteaceae in the South Western Cape region of South Africa. Subsequently, three additional species were added to the genus, two also from Proteaceae and one from Encephalartos (Marincowitz et al. 2008, Crous et al. 2008, 2009). All known species are thus from the same region on indigenous flora. These species are typically associated with leaf spots and stem cankers and they appear to be pathogens. Separate studies have also shown that they are endophytes (Swart et al. 2000, Taylor et al. 2001), similar to members of the Botryosphaeriaceae.

Apart from its restricted distribution and host range, Saccharata is also unique in its asexual morphology, which includes a hyaline, fusicoccum-like and a pigmented diplodia-like asexual morph. These characters are shared with its related families; fusicoccum-like conidia in Melanopsaceae and pigmented diplodia-like conidia in Aplosporellaceae. It is clear that these variations in conidial morphology are very old (tens of millions of years) ancestral characters that must have existed prior to the divergence of this group from other Botryosphaeriales. It is thus remarkable how similar, especially in the fusicoccum-like conidia and ascospore morphology, the spores of these fungi have remained over time. The diplodia-like state is somewhat different from other Botryosphaeriales in that the conidia are typically almost half the size of other Diplodia conidia. We do not currently have enough data to address the selective pressure that could have played a role in the development these interesting morphological changes.

Phyllostictaceae and Planistromellaceae

Phyllosticta clearly warrants a separate family to accommodate this morphologically and ecologically unique, widespread and economically important genus in the Phyllostictaceae. These fungi typically infect leaves and fruit, rather than woody tissue, and they can cause serious damage (Glienke et al. 2011, Wong et al. 2012). Species of Phyllosticta are also known to have an endophytic phase (Wikee et al. 2013a), as is true for most other Botryosphaeriales. The Phyllostictaceae is also morphologically unique in terms of the ascospores and conidia (Van der Aa & Vanev 2002). The species included in this study are only representative of a small extent of the diversity in this group and a more complete analysis of the Phyllostictaceae is presented in Wikee et al. (2013b, this volume).

The Planistromellaceae has previously been recognised as distinct within the Botryosphaeriales (Minnis et al. 2012) and this is supported by the analyses in the present study. The family is currently considered to include only species of Kellermania. Species in the Planistromellaceae have unique conidia with fairly long appendages that are quite distinct from other genera in the Botryosphaeriales. Kellermania spp. are mostly leaf infecting, and one species has been associated with Yucca Leaf Blight in California and Florida in the USA (Horst 2008). Species are commonly collected sporulating on dead leaves of Agavaceae, and appear to be endophytic, as they have been isolated from healthy leaves in many countries where Yucca spp. are grown as ornamentals (P.W. Crous, unpubl data). Minnis et al. (2012) have also shown apparent patterns of host specificity and co-evolution with major plant lineages. As is true for many other families in the Botryosphaeriales, additional sampling is needed for a better understanding of species diversity, host range and geographic distribution.

We thank members of the Tree Protection Co-operative Programme (TPCP), the DST/NRF Centre of Excellence in Tree Health Biotechnology (CTHB), THRIP (project no. TP2009061100011) and the University of Pretoria, South Africa for financial support. We also thank Ms Katrin Fitza and Ms Fahimeh Jami for assistance with submissions of sequences.