Molecular Phylogeny of the Plant Pathogenic Genus
http://www.100md.com
《分子生物学进展》
Wageningen University, Laboratory of Phytopathology, Wageningen, The Netherlands
Correspondence: E-mail jan.vankan@wur.nl.
Abstract
The cosmopolitan genus Botrytis contains 22 recognized species and one hybrid. The current classification is largely based on morphological characters and, to a minor extent, on physiology and host range. In this study, a classification of the genus was constructed based on DNA sequence data of three nuclear protein-coding genes (RPB2, G3PDH, and HSP60) and compared with the traditional classification. Sexual reproduction and the host range, important fitness traits, were traced in the tree and used for the identification of major evolutionary events during speciation. The phylogenetic analysis corroborated the classical species delineation. In addition, the hybrid status of B. allii (B. byssoidea x B. aclada) was confirmed. Both individual gene trees and combined trees show that the genus Botrytis can be divided into two clades, radiating after the separation of Botrytis from other Sclerotiniaceae genera. Clade 1 contains four species that all colonize exclusively eudicot hosts, whereas clade 2 contains 18 species that are pathogenic on either eudicot (3) or monocot (15) hosts. A comparison of Botrytis and angiosperm phylogenies shows that cospeciation of pathogens and their hosts have not occurred during their respective evolution. Rather, we propose that host shifts have occurred during Botrytis speciation, possibly by the acquisition of novel pathogenicity factors. Loss of sexual reproduction has occurred at least three times and is supposed to be a consequence of negative selection.
Key Words: Botrytis ? necrotrophic fungus ? coevolution ? host shift ? Bayesian inference ? molecular phylogeny
Introduction
Fungi of the genus Botrytis Persoon are important pathogens of many agronomically important crops, such as grapevine, tomato, bulb flowers, and ornamental crops (Jarvis 1977). Botrytis diseases appear primarily as blossom blights and fruit rots but also as leaf spots and bulb rots in the field and in stored products. Botrytis species are necrotrophs, inducing host-cell death resulting in progressive decay of infected plant tissue. The pathogen produces abundantly sporulating gray mycelium on infected tissue. Macroconidia (mitotically produced spores) can be transported by wind over long distances. Botrytis overwinters in the soil as mycelium in decaying plant debris and as sclerotia, melanized mycelial survival structures. Some species frequently produce a sexual teleomorphic stage in which ascospores are produced in an apothecium. When collected in nature, apothecia are found under cool weather conditions, arising from sclerotia, which have developed on decayed plant parts in moist soil. Botrytis and its sexual form Botryotinia Whetzel comprise 22 species and one hybrid (Hennebert 1973; Yohalem, Nielsen, and Nicolaisen 2003) and are classified within the family Sclerotiniaceae Whetzel (Inoperculate Discomycetes). Delineation of species has traditionally been based on morphological characteristics, especially macroconidium ontogeny, and species have been named based on host association. Most species have a worldwide distribution or occur wherever their host crops are grown (Jarvis 1977).
The genus Botrytis comprises one generalist, B. cinerea, infecting over 200 eudicot hosts, especially senescing or otherwise weakened or wounded plants (MacFarlane 1968). All other species are considered specialists with a narrow host range. They infect only one or a few closely related species within the same plant genus (Mansfield 1980), with the exception of B. fabae, which can infect species of the genera Vicia, Lens, Pisum, and Phaseolus, all belonging to Fabaceae (Jarvis 1977). Specialized species occur on corolliferous monocotyledons and on members from the four eudicot families Fabaceae, Ranunculaceae, Geraniaceae, and Paeoniaceae (Jarvis 1977) (table 1). Narrow host range Botrytis species parasitize living leaves or bulbs of their hosts but may also occur as saprophytes. In many cases, host-specific Botrytis species are able to cause primary lesions on a nonhost, but these primary lesions fail to expand (Prins et al. 2000).
Table 1 Species Recognized by Hennebert (1973), Sexual Stage, Disease Symptoms, and Typical Host Plant Species
Several specialists are able to infect members of the same plant family or the same plant species. Three species of Botryotinia have been described occurring on the buttercup family Ranunculaceae; B. calthae occurs on Caltha palustris L., B. ranunculi on Ranunculus spp., and B. ficariarum on Ficaria verna. Occasionally B. calthae has been observed infecting Ficaria verna, but otherwise they are specific to their host (Hennebert and Groves 1963). As many as seven Botrytis species occur on Allium spp., which is one of the largest genera of the petaloid monocotyledons (Friesen, Fritsch, and Blattner 2004). Allium includes several economically important species, such common onion, garlic, chives, and leek, all of which are infected by B. byssoidea, B. aclada, and B. allii. The last species is an allodiploid hybrid based on mitotic chromosome counts, conidia size, restriction fragment length polymorphism (RFLPs), and sequence data (Nielsen and Yohalem 2001; Yohalem, Nielsen, and Nicolaisen 2003; Shirane, Masuko, and Hayashi 1989). B. byssoidea and B. aclada are considered to be its parental species (Nielsen and Yohalem 2001; Yohalem, Nielsen, and Nicolaisen 2003). Other Botrytis species infecting Allium spp. have a more restricted host range; they have become specialized on wild Allium spp. or infect only one or two economically important Allium crops.
To investigate whether closely related Botrytis species are pathogenic on closely related plant species, we compared phylogenies of fungi and their angiosperm hosts (APG 2003) for possible cospeciation. Parallel cladogenesis between host and pathogens would be indicative of ancient coevolution, a reciprocal process in which characteristics of one organism evolve in response to specific characteristics of another. The process of coevolution between pathogen and host is often based on gene-for-gene relationships, in which resistance gene alleles in the host are matched by virulence gene alleles in the pathogen (Flor 1955; Heath 1991; Thompson and Burdon 1992; Kniskern and Rausher 2001). Coevolution is more likely expected between obligate biotrophic pathogens or symbionts and their host plants than it is with necrotrophic pathogens. Obligate parasites do not kill host cells but get their nutrients either by penetrating living cells or by establishing close contact with them. Necrotrophic pathogens such as Botrytis need to kill host cells before the cells are invaded by the fungus (Clark and Lorbeer 1976; van Baarlen, Staats, and van Kan 2004), which might suggest absence or a low degree of coevolution. There is very limited knowledge about the phylogenetic relationships among members of the genus Botrytis. Holst-Jensen, Vaage, and Schumacher (1998) analyzed nuclear ribosomal internal transcribed spacer (ITS) DNA sequences and concluded that the Botryotinia teleomorph along with Botrytis anamorphs constitute a monophyletic lineage. However, the relationships among members of the genus Botrytis could not be resolved because of the limited phylogenetically informative ITS sequence characters.
In this study, we made use of fragments of three single-copy nuclear DNA (nDNA) genes encoding glyceraldehyde-3-phosphate dehydrogenase (G3PDH), Heat-shock Protein 60 (HSP60), and DNA-dependent RNA polymerase subunit II (RPB2). All three genes encode enzymes that are involved in basic cellular processes, and both G3PDH and RPB2 evolve at moderate evolutionary rates (Smith 1989; Berbee, Pirseyedi, and Hubbard 1999; Liu, Whelen, and Hall 1999; Liu and Hall 2004). The first objective was to clarify the evolutionary history of the genus Botrytis and investigate whether DNA sequence data support the classical species delineation. Second, the effect of the hybrid species B. allii on tree structure was examined because hybrids may cause topological changes, especially when parents are distantly related (McDade 1992). The third objective was to trace the evolution of the reproductive mode and the host-range spectrum within the genus Botrytis.
Materials and Methods
A set of 52 isolates (table 2) was chosen to represent all the recognized Botrytis species. Most strains were obtained from the Belgian Coordinated Collections of Microorganisms (BCCM, Belgium, Brussels). Additional strains were obtained from the Centraalbureau voor Schimmelcultures (CBS, Utrecht, The Netherlands) and personal collections. Single isolates of Sclerotinia sclerotiorum and Monilinia fructigena were selected as outgroups on the basis of a phylogenetic study by Holst-Jensen, Vaage, and Schumacher (1998). Strains were grown on malt agar (Oxoid) for 10 to 14 days in the dark at 18°C.
Table 2 Isolates of Botrytis, Origin, Year of Collection, and GenBank Accession Numbers for DNA Sequences Used in This Study
DNA Extraction, Amplification, and Sequencing
Mycelial tissue was harvested, lyophilized, submerged in liquid nitrogen and ground into a powder. Genomic DNA was extracted from 10 to 20 mg dry tissue using the Puregene DNA isolation kit (Gentra systems Inc./Biozym systems, Landgraaf, The Netherlands) according to the manufacturer's instructions. DNA pellets were dissolved in 100 ml of TE (10mM Tris-HCl [pH 8.0], 1mM EDTA) and stored at 4°C or –20°C.
Primer combinations (table 3) were designed or modified to amplify regions of three nuclear DNA genes (G3PDH, HSP60, and RPB2). PCR primers to amplify the RPB2 region that have been described previously by Liu, Whelen, and Hall (1999) were slightly modified. To amplify G3PDH and HSP60, forward and reverse primers were designed based on homologous gene sequences from GenBank and a Botrytis cinerea cDNA library (Genoscope, Centre National de Séquen?age, France). For amplification of ITS region, primers ITS1 and ITS4 were used (White et al. 1990). To facilitate batch sequencing of different PCR products, primers were extended with M13(-20) forward primers or M13 reverse primers.
Table 3 Primers Used for PCR Amplification and Sequencing
PCR amplification was carried out in a 50-ml reaction mixture that contained 10 to 50 ng of genomic DNA, 1X GeneAmp PCR buffer (PerkinElmer, Norwalk, Conn.), 200 mM of each deoxynucleoside triphosphate (Promega, Madison, Wis.), 0.2 pmol of each primer (Amersham Pharmacia Biotech), 2 mM MgSO4 and 1.0 U of AmpliTaq DNA polymerase (PerkinElmer). Amplifications were carried out in a Peltier Thermal Cycler-200 (Biozym, Landgraaf, The Netherlands). The following thermocycling pattern was used to amplify HSP60 and RPB2 gene fragments: 94°C for 5 min (1 cycle); 94°C for 30 s, 55°C for 30 s, and 72°C for 90 s (35 cycles), and then 72°C for 10 min (1 cycle). The same program with an annealing temperature of 64°C was used for G3PDH gene fragments. For amplification of the ITS region, an amplification protocol was used as described by White et al. (1990). The PCR products were separated on a 1% agarose-Tris-borate-EDTA (1x TBE) gel, containing ethidium bromide. The fragment size of the PCR product was verified by comparison to a 100-bp DNA Ladder (GibcoBRL). PCR products were purified using the QIAquick PCR purification kit (Qiagen) following manufacturer's instructions. Purified PCR products were sequenced (BaseClear Holding B.V., Leiden, The Netherlands) in both directions using M13(-20) forward and M13 reverse primers.
Six PCR products were cloned because the yield of amplified product was insufficient for direct sequencing: M. fructigena 9201 (locus RPB2), B. hyacinthi MUCL442 (loci RPB2, HSP60 and G3PDH), B. byssoidea MUCL94 (locus HSP60), and B. calthae MUCL1089 (locus G3PDH). Furthermore, multiple alleles of allopolyploid B. allii strains MUCL403 and MUCL1150 were cloned as well. PCR products were extracted from 1% agarose gel using the GFX PCR purification kit (Amersham Biosciences). PCR products were then ligated into a pGemT-Easy Vector (Promega) using standard protocols and introduced into electrocompetent Escherichia coli DH5 cells. Transformed cells were spread on LB-plates, each containing 100 mg/ml ampicillin, 20 mg/ml X-gal, and 0.1 mM IPTG and incubated at 37°C overnight. White colonies were screened for correct insert size using gene-specific PCR primers. To distinguish between alleles of each locus, 20 positive clones of B. allii strains MUCL1150 and MUCL403 were selected for restriction digestion. Restriction enzymes were selected based on polymorphic sites between aligned sequences of each locus of the presumed parental species B. aclada (isolate MUCL8415) and B. byssoidea (isolate MUCL94). Predicted restriction profiles were compared with digestion patterns of cloned fragments of both parental species, as well as both B. allii strains. G3PDH and HSP60 fragments were digested with the restriction enzyme EcoRV (Promega), whereas RPB2 fragments were digested with HindIII (Promega). Subsequently, individual clones, containing different alleles, were selected for sequencing based on restriction profiles. Clones were sequenced in both directions using SP6 and T7 primers (see table 3).
Phylogenetic Analysis
Automated sequence outputs were imported into Vector NTI suite 8.0 (InforMax, Inc.) for visual inspection of chromatographs and assembly of the contigs. For unresolved basepairs, the IUPAC nucleotide coding system was used. Sequences were compared between complementary strands for reading errors and aligned using ClustalX version 1.8 (Thompson et al. 1997). Alignments were visually inspected and adjusted manually when necessary. Regions of sequences with ambiguous alignment were excluded from all analyses (table 4). The full alignment containing all three loci is deposited in TREEBASE (accession number S1170). Individual sequences are deposited in GenBank under accession numbers AJ704990 to AJ705044, AJ716290 to AJ716305, AJ716046 to AJ716103, and AJ745662 to AJ745716 (see table 2).
Table 4 Summary of the RPB2, HSP60, G3PDH, and Combined Sequence Alignments
The basic data set for each locus consisted of 50 taxa: 48 Botrytis specimens and two outgroup species. Phylogenetic analyses were conducted for each of the three loci (RPB2, HSP60, and G3PDH) and for a combined data matrix of all three. To assess the impact of hybrids on tree structure, separate analyses were performed for the three DNA regions with and without hybrid taxa. Basic data sets plus five to eight hybrids were analyzed separately. All alleles of each locus of B. allii MUCL1150 and B. allii MUCL403, as well as of B. aclada MUCL3106 were treated as hybrid taxa. Because multiple alleles were found for hybrid taxa, combined analysis was performed with the basic data set only.
Parsimony analyses were carried out using the software package PAUP* version 4.0b10 (Swofford 2002) on a G4 Power Macintosh. The following settings were used: heuristic search with tree bisection-reconnection (TBR) branch swapping, with MULPARS (keeping all equally most-parsimonious trees) on, 1,000 replicates of random taxon-addition order, and all character transformations treated as equally likely (Fitch parsimony [Fitch 1971]). To minimize the time spent searching large numbers of trees, a limit of five trees was set for each replicate. After completing the replicates, all trees found were then used as starting trees for another round of swapping, with no limit to the number of trees. Gaps were treated as fifth character state, and multistate characters were treated as uncertain. A jackknife analysis was carried out to assess data structure and identify significant supported clades (63%). Search type was set to "fast" stepwise-addition (for individual loci) or full heuristic (for combined analysis). Jackknife values were generated by randomly deleting 37% of the characters for 10,000 replicates and using "Jac" resampling (Farris et al. 1996).
Bayesian analyses were performed using MrBayes version 3.0b4 (Ronquist and Heulsenbeck 2003). Each run consisted of four incrementally heated Markov chains run simultaneously, with heating values set to default (0.2). Default uniform priors were used for all model parameters (six substitution rates, four base frequencies, proportion of invariable sites, and alpha value of gamma distribution). Data partitions were set up for all coding and noncoding regions of each locus. HSP60 and G3PDH both contained two introns and two exons. For both loci individually, the two introns were pooled, and the two exons were pooled and analyzed as two partitions. Likelihood model parameters were separately optimized during MCMC searches for each data partition. Markov chains were initiated from a random tree and run for five-million generations; samples were taken every 100th generation. Convergence among chains was monitored by examining plots of log-likelihood values. Burn-in period (i.e. lack of improvement of log-likelihood values) was evaluated visually. All samples taken before burn-in were discarded in PAUP*, and the remaining samples were used to determine posterior probability (PP) distributions. Each run was performed twice. Bayesian 50% majority-rule consensus trees were generated in PAUP*. Branches with at least 63% jackknife support and with at least 0.95 Bayesian PP were considered as well supported.
Concordance of combined data sets was evaluated by comparing jackknife topologies via visual inspection and partition homogeneity tests (PHT [Farris et al. 1995]) implemented with PAUP*. Visual inspection permits the precise location of areas of strong discordance (branches conflicting with jackknife values at least 85% and Bayes PP at least 0.95) between phylogenies generated by different data sets. PHTs were carried out for all pairwise combinations of the three data sets and for all data sets combined using heuristic search (TBR) with 100 random-addition replicates and informative characters only. The null hypothesis of congruence was rejected if P < 0.001 (Darlu and Lecointre 2002; Dettman, Jacobson, and Taylor 2003).
Functional characters were mapped onto the molecular phylogeny using MacClade version 4 (Maddison and Maddison 2000): reproductive mode (two states; strictly asexual or asexual and sexual); host-plant group (two states; monocotyledons or eudicotyledons) and host-plant family (10 states; Hyacinthaceae, Liliaceae, Alliaceae, Amaryllidaceae, Iridaceae, Paeoniaceae, Geraniaceae, Ranunculaceae, Fabaceae, or Polyphagous on eudicotyledons). All characters were equally weighted under parsimony. Character states from literature (Jarvis 1977, 1980; Hennebert and Groves 1963; Hennebert 1973) are listed in table 1.
Results
In preliminary experiments, sequence data of eight protein-coding loci and the noncoding ITS region were acquired for multiple isolates of three Botrytis species. The three most-informative loci (G3PDH, HSP60, and RPB2) were used as starting point for this study. A total of 188 new sequences of 54 specimens are reported here. Sequences of protein-coding gene fragments and ITS sequences were first evaluated in a standard GenBank Blast search. Nearly all obtained sequences corresponded with listed Botrytis cinerea or Sclerotinia sclerotiorum sequences. The ITS sequence of B. anthophila Bondartsev, a species reported in Jarvis (1977), was identical to Rhizoctonia sp. ITS sequences, and this sequence was, therefore, excluded from further analyses. An alignment of ITS characters, consisting of 453 nucleotide positions of the sequences listed here (table 2) and of sequences of Botrytis species reported by Holst-Jensen, Vaage, and Schumacher (1998), contained a limited number of singletons and phylogenetic informative characters (data not shown). Therefore, ITS sequence data were excluded from further analyses.
Analysis of G3PDH Sequence Data and the Influence of Hybrid Taxa on Tree Structure
The basic sequence matrix (table 4) consists of 890 sites, of which 186 (21%) were variable characters and 126 (14%) were potentially parsimony informative. Analysis of G3PDH sequences with equal (Fitch) weights resulted in two equally most-parsimonious trees of 312 steps, with a consistency index (CI) of 0.73 and a retention index (RI) of 0.89. Relationships among the main clades were topologically identical in MP and Bayesian consensus trees (data not shown), except for some minor differences. Comparison between replicate Bayesian analyses resulted in identical majority-rule consensus trees, with only marginally lower mean log-likelihood values (–2998.75 and –3025.33). Bayesian PP and jackknife percentages of branches are included on the MP semistrict consensus tree in figure 1. The semistrict consensus tree topology of the two most-parsimonious trees (MPTs) contains 24 nodes with at least 63% jackknife and at least 95% PP support.
FIG. 1.— Semistrict consensus of two most-parsimonious trees based on G3PDH data. This tree was compatible overall in highly supported lineages to the Bayesian 50% majority-rule consensus tree. Jackknife frequencies (10,000 replicates) are shown above each node; Bayesian posterior probabilities (PP) are shown below each node. Letters indicate major clades with significant or strong support (see table 5). A dash indicates support for the branch was less than 50% jackknife and less than 0.50 PP.
Table 5 Cladistic Behavior of RPB2, HSP60, G3PDH, and Combined Data Sets with and Without Hybrid Taxa
The G3PDH data set plus hybrids consists of 53 taxa, including two alleles of both B. allii MUCL1150 and B. allii MUCL403. The heuristic search yielded 11 MPTs with higher tree length and slightly lower CI (L = 323, CI = 0.71, and RI = 0.89). MP and Bayesian analyses resulted in almost identical trees (data not shown). The semistrict consensus of those fundamental trees recovers the same major clades as in the G3PDH data sets without hybrid taxa, and branch support was only marginally lower (table 5). Within these trees, one allele of B. allii strain MUCL1150 (allele 1) and two alleles of B. allii strain MUCL403 (alleles 1 and 2) are placed in a clade together with B. aclada (strains MUCL8415, MUCLPRI006, and MUCL3106). Allele 2 of B. allii strain MUCL403 is placed together with B. byssoidea.
Analysis of HSP60 Sequence Data and the Influence of Hybrid Taxa on Tree Structure
The basic sequence matrix consists of 983 characters, from which 20 positions were removed because of ambiguity in the alignment (table 4). Of the remaining 963 characters, 225 (23.4%) were variable, and 158 (16.4%) were potentially phylogenetically informative. The Fitch parsimony recovered 320 shortest trees, each 373 steps long (CI = 0.74, RI = 0.90). The Bayesian tree was consistent with the consensus of these trees, and separate Bayesian analyses resulted in identical trees (mean log-likelihood values –3357.89 and –3390.86 [data not shown]). Bayesian PP and jackknife percentages of branches are included on the MP semistrict consensus tree in figure 2. The semistrict consensus tree topology of the 320 MTPs contains 23 nodes with at least 63% jackknife and at least 95% PP support.
FIG. 2.— Semistrict consensus of 320 most-parsimonious trees based on HSP60 data. This tree was compatible overall in highly supported lineages to the Bayesian 50% majority-rule consensus tree. Jackknife frequencies (10,000 replicates) are shown above each node; Bayesian posterior probabilities (PP) are shown below each node. Letters indicate major clades with significant or strong support (see table 5). A dash indicates support for the branch was less than 50% jackknife and less than 0.50 PP.
The HSP60 data set plus hybrids consist of 58 taxa, including three alleles of B. allii strain MUCL1150 and four alleles of B. allii strain MUCL403. Fitch analysis gave 320 trees of 404 steps, with CI = 0.69 and RI = 0.89. Hybrid taxa had strong influence on tree structure in both MP and Bayesian analyses, resulting in the collapse of (sub)clades e, and k (table 5). MP and Bayesian analyses resulted in almost identical topologies (data not shown); however, there are also important differences. Bayesian analysis strongly supported (100% PP) a large clade containing members of clade e, B. convoluta, B. paeoniae, and a subclade (98% PP) placing hybrids as basal members of their derived parents B. byssoidea and B. aclada. This clade is highly supported by Bayesian posterior probabilities but is not present at all in the jackknife tree.
Analysis of RPB2 Sequence Data and the Influence of Hybrid Taxa on Tree Structure
The basic sequence matrix consists of 1096 positions, of which 274 (25%) were variable and 203 (18.5%) were potentially parsimony informative (table 4). Analysis of RPB2 sequences with equal (Fitch) weights resulted in seven equally most-parsimonious trees of 439 steps, with a CI of 0.73 and an RI of 0.91. Relationships among the main clades were topologically identical in MP and Bayesian consensus trees (data not shown). Comparison between replicate Bayesian analyses resulted in identical majority-rule consensus trees, with only marginally lower mean log-likelihood values (–3955.80 and –3970.45; not shown). Bayesian PP and jackknife percentages of branches are included on the MP semistrict consensus tree in figure 3. Of the three individual regions, RPB2 provided the highest number of significantly supported nodes (26 nodes with at least 63% jackknife and at least 95% PP support), which are mainly located at the terminal branches. The RPB2 topology contains the three main (sub)clades (R, O, and E) that also occur in the combined analysis (table 5), although relationships among these clades remained unresolved.
FIG. 3.— Semistrict consensus of seven most-parsimonious trees based on RPB2 data. This tree was compatible overall in highly supported lineages to the Bayesian 50% majority-rule consensus tree. Jackknife frequencies (10,000 replicates) are shown above each node; Bayesian posterior probabilities (PP) are shown below each node. Letters indicate major clades with significant or strong support (see table 5). A dash indicates support for the branch was less than 50% jackknife and less than 0.50 PP.
The RPB2 data set plus hybrids consisted of 55 taxa, including two alleles of both B. allii strain MUCL1150 and B. allii strain MUCL403. The heuristic search yielded six MPTs with higher tree length and slightly lower CI (L = 459, CI = 0.70, and RI = 0.91). MP and Bayesian analyses resulted in topologically identical trees (data not shown). The semistrict consensus of those fundamental trees recovers the same major clades as the basic data set (table 5). However, one extra clade was recovered, although with low support (<50% jackknife and 94% PP), consisting of clade O, B. aclada (MUCL8415, MUCLPRI006, and MUCL3106), and B. allii (MUCL1150 allele 1 and MUCL403 alleles 1 and 2). Allele 2 of B. allii strain MUCL1150 was placed together with B. byssoidea. Furthermore, as a result of hybrid taxa, jackknife branch support for nodes E and G decreased from strongly supported (86% and 87%, respectively) to well supported (73% and 78%, respectively).
Assessment of Between–Data Sets Congruence
The G3PDH tree was largely congruent with the RPB2 tree; in both trees, there is good support for three main (sub)clades (R, O, and G [figs. 1 and 3]), but arrangements within each clade could not be clearly resolved. At the base of the tree, some resolution is provided; however, branches indicating relationships between clades are not supported (fig. 3). In comparison to the RPB2 topology, more resolution is provided by HSP60 at the base of the tree (figs. 2 and 3). Contrary to the RPB2 tree, the positions of B. galanthina and B. hyacinthi/B. croci have been exchanged, as well as of B. polyblastis and B. tulipae. As with RPB2, HSP60 placed B. porri as part of a basal trichotomy. Jackknife topologies were used to test whether the three data sets have different underlying phylogenetic histories. Strong incongruence (>85% jackknife and 0.95 PP) between RPB2 and HSP60 phylogenies was evident in the placement of B. cinerea strain B05.10 and B. polyblastis. None of the strongly supported clades was incongruent between the other data sets (RPB2 versus G3PDH and G3PDH versus HSP60). The PHT was used as an alternative method to determine whether each possible combination of loci differs significantly in phylogenetic structure. No significant conflict was detected for all pairwise comparisons of each locus (RPB2-HSP60, RPB2-G3PDH, and HSP60-G3PDH; 0.007 P 0.001) and the three combined (P = 0.001).
Molecular Phylogeny Distinguishes Two Major Groups
Because no significant between–data set incongruence was detected and because the individual jackknife topologies produced nearly identical results, the three regions were combined directly. Analysis of this combined matrix resulted in 80 trees of 1,151 steps (CI = 0.72 and RI = 0.89). The trees were 27 steps longer than the combined number of steps from the RPB2, HSP60, and G3PDH trees (table 5). The Bayesian tree was consistent with the MP tree, but as with the RPB2 MP tree, B. polyblastis was strongly placed (Bayesian PP = 0.99) as a sister taxa to B. narcissicola, B. gladiolorum, and B. byssoidea (see figures 3, 4A, and 4B). Furthermore, B. tulipae was placed basal to the node tying B. globosa and B. sphaerosperma. The relative branching order of ingroup species and mean log-likelihood values were comparable among two replicate Bayesian analyses (–10389.27 and –10402.63 [data not shown]). In the combined molecular phylogeny, the overall number of supported nodes was greater (26) than for the trees of each separate analysis. Also, most nodes had greater support, especially at the spine, and most were supported strongly. The exception involved the positions of the B. cinerea B05.10 and B. polyblastis, which were poorly resolved as a result of discordant positions of both species in separate analyses (see figures 2 and 3). In contrast to other relationships in the genus, placement of B. polyblastis and B. tulipae was unresolved.
FIG. 4.— Molecular phylogeny for 48 Botrytis taxa and two outgroup species based on combined G3PDH, HSP60, and RPB2 data. (A) Semistrict consensus of 80 most-parsimonious trees. Jackknife frequencies (10,000 replicates) are shown above each node; Bayesian posterior probabilities (PP) are shown below each node. Letters indicate major clades with significant or strong support (see table 5). A dash indicates the support for the branch was less than 50% jackknife and less than 0.50 PP. Bars indicate the (sub)clades. (B) Bayesian inference showing a 50% majority-rule consensus tree (mean log-likelihood = –10389.27) from a five-million generation MCMC analysis.
Multiple isolates of the same species usually clustered together, and their gene sequences were identical or varied at fewer sites within species in the aligned sequence positions than between species. Molecular phylogenetic analysis of the DNA sequences supports the traditional morphological species classification. In addition, the hybrid status of B. allii (B. byssoidea xB. aclada [Yohalem, Nielsen, and Nicolaisen 2003]) was confirmed (fig. 5). The two isolates of B. hyacinthi did not appear to be monophyletic, because they lacked shared characters that distinguished them as a group (fig. 4B).
FIG. 5.— Schematic representation of sequence alignment of multiple HSP60 alleles for allopolyploid B. allii (strains MUCL403 and MUCL1150) and its presumed parental species B. aclada (isolate MUCL8415) and B. byssoidea (isolate MUCL94). Black and gray lines indicate stretches of sequence polymorphisms of B. aclada MUCL8415 and B. byssoidea MUCL94, respectively. Black and gray arrows indicate EcoRV restriction endonuclease cleavage sites.
Based upon the semistrict consensus tree, two main clades were identified in the genus Botrytis, and clade 2 was divided into five subclades, as within this clade strongly supported clusters of species could be identified. The base of the tree is characterized by a deep split leading to clade 1 and clade 2. Clade 1 consists of the species B. calthae, B. fabae, B. cinerea, and B. pelargonii and is separated from clade 2 by a relatively long branch. The internode separating B. pelargonii from B. cinerea was very short, reflecting the high amount of interspecific sequence similarity between both species. There was only one unique polymorphism on a total of 2,955 bp between both species.
B. aclada and B. paeoniae are less closely related to other taxa grouping at a basal position of clade 2 (81% jackknife and 1.0 PP), and they were, therefore, assigned to subclade 2a. Individual gene trees did not support this clade, although both species always grouped closely together. B. convoluta could not unambiguously be placed in either subclade within clade 2 and has, therefore, been assigned to subclade 2d. Both subclade 2b and subclade 2c were defined based on high support values (100/1.0 and 99/1.0, respectively) in combined analysis and the consistent support in separate analyses. The relationships between B. ranunculi and B. ficariarum in subclade 2b and between B. hyacinthi and B. croci in subclade 2c were unresolved. Although B. porri is closely related, there was no support for its placement in either subclade within clade 2 in combined and separate nDNA analyses. B. porri was, therefore, placed in a separate clade, designated as subclade 2e.
The Allodiploid Hybrid B. allii May Contain Up to Four Different Alleles
To confirm earlier reports about the hybrid status of B. allii (Nielsen and Yohalem 2001; Yohalem, Nielsen, and Nicolaisen 2003) restriction profiles and sequence data of cloned fragments were compared with those of its parental species B. aclada and B. byssoidea. Clones of B. aclada strain MUCL8415 and B. byssoidea strain MUCL94 could be distinguished based on the unique digestion patterns among the two species. Among cloned fragments of B. allii (strains MUCL1150 and MUCL403), digestion patterns of either the B. aclada-type or the B. byssoidea-type (data not shown) were detected for all three loci, indicating the presence of the two parental alleles. However, in case of HSP60, divergent patterns were also detected. Fragment sizes could be explained by the presence of three restriction sites (fig. 5), of which two sites are present in B. byssoidea and one is present in B. aclada. Alleles were sequenced to verify this interpretation. For HSP60, four alleles were identified for B. allii strain MUCL403 and three alleles were identified for B. allii strain MUCL1150. Hybrid alleles contained stretches of sequences originating from both parental alleles (fig. 5). For the G3PDH and RPB2 genes, hybrid alleles have also been found despite identical restriction patterns in both parental alleles.
The HSP60 locus of B. aclada strain MUCL3106 was identical to B. byssoidea except for two polymorphic sites, whereas both RPB2 and G3PDH were identical to B. aclada strain MUCL8415. Because fragments of B. aclada strain MUCL3106 were directly sequenced, allelic variants have not been detected for each locus.
Loss of Sexual Reproduction Has Evolved Three Times
Two important fitness traits, reproductive mode and host-range spectrum, were expressed as multistate "characters" and optimized on the cladogram (fig. 4A) using parsimony. Sexual reproduction has not been found in seven species, five of which parasitize monocot hosts and two parasitize eudicot hosts (table 1). Based on reproductive mode, sexual reproduction was reconstructed as the ancestral condition (fig. 6A). Strict asexual reproduction was inferred to be a homoplasious trait, and loss of sex appears to have occurred three times. Two groups, containing two and four strictly asexual reproducing Botrytis species are present in the tree (B. aclada/B. paeoniae and B. hyacinthi/B. croci/B. galanthina/B. tulipae). Within clade 1, the asexual species B. fabae separates from a clade predominated by sexual reproduction. There is indication for a reversal from strictly asexual reproduction to sexual reproduction of six taxa in clade 2c.
FIG. 6.— Reconstructions of character-state evolution mapped onto the combined molecular phylogeny (the same topology as in figure 4A) for Botrytis under parsimony: (A) reproductive mode, (B) host-plant group, and (C) host-plant family.
Host and Pathogen Phylogenies Are Not Congruent
To assess the possible occurrence of cospeciation patterns between Botrytis species and their hosts, the distribution of host-plant family or angiosperm group was mapped on the cladogram (fig. 6B and C). The order Asparagales is parasitized by 14 of the previously recognized Botrytis species, the order Ranunculales by three species, and the Liliales by two species. Three eudicot orders contain one specialist Botrytis species, and these orders can be parasitized by the generalist B. cinerea as well. Botrytis species parasitizing either eudicot or monocot hosts do not tend to cluster together (fig. 6B). Within clade 2, transitions from pathogens infecting monocot hosts to eudicot hosts seem to have occurred twice. Furthermore, Botrytis species parasitizing hosts from the same plant family appear to be unrelated (fig. 6C). The best example of incongruency occurs in the host-plant family Ranunculaceae, containing three Botrytis species, of which B. calthae clusters in clade 1, and the species B. ficariarum/B. ranunculi cluster in clade 2b.
Discussion
DNA Sequence Data Support the Morphological Species Delineation
In the present study, we conducted molecular phylogenetic analyses of 22 Botrytis species and one hybrid. Our analysis provides the first comprehensive phylogenetic study of the entire genus. Our analysis further supports monophyly of the genus Botrytis, as has previously been proposed by Holst-Jensen, Vaage, and Schumacher 1997. The phylogenetic analyses of DNA sequence data show that the sequences from multiple isolates of the same species group together and, therefore, support the classical Botrytis species status (Hennebert 1973; Jarvis 1977). Morphological characters, together with DNA sequence information, can be used to identify Botrytis species; for instance, in a diagnostic key of Botrytis species. Such a key may be a potentially powerful tool for diagnostics of this important group of plant pathogens.
The combined analysis of all three loci supported two main clades (fig. 4) and provides more resolution than individual gene phylogenies (fig. 1–3). This result shows the value of conducting phylogenetic analysis on multiple genes, which is an established tool to address complex phylogenetic questions (Rokas et al. 2003; Soltis, Soltis, and Chase 1999; O'Donnell et al. 1998).
The taxonomic status of B. croci and B. hyacinthi remains uncertain; both species exhibit high sequence similarity, which makes the separation into distinct species questionable. In this study, data of B. croci were based on a single isolate, so additional sequencing is needed to resolve its status as a separate species.
Reports about the hybrid status of B. allii (Nielsen and Yohalem 2001; Yohalem, Nielsen, and Nicolaisen 2003) were confirmed. Interspecific hybrid species were identified by the presence of multiple HSP60, RPB2, and G3PDH gene copies that consistently grouped with B. aclada and B. byssoidea within the Botrytis phylogeny. The occurrence of up to four different alleles per locus (fig. 5) is indicative of a diploid, dikaryotic genome. However, higher ploidy levels or aneuploidy in these isolates cannot be excluded. The selection of alleles for sequence analysis was based on polymorphic digestion patterns in individual clones of amplified loci; additional alleles that were nonpolymorphic in restriction sites may have escaped our attention. The mixture of parental and recombinant alleles indicates that reciprocal recombination and gene conversion have occurred several times during diversification within the hybrid B. allii. These hybrids were placed as basal members of the clade that includes their parents (table 5). Major topological change was caused by hybrid sequences in the HSP60 phylogeny, resulting in the collapse of the portion of the cladogram that included the hybrid's parents. In case of RPB2 and G3PDH, the cladistic relationships remained largely unchanged, causing only a slight decrease in support values in branches leading to the hybrid's parents. Except for B. alli, none of the 22 Botrytis species appeared to be hybrids, nor was there any indication from sequence data or host-plant range that they arose by hybridization.
Incongruence of Host and Pathogen Phylogenies Does Not Support Cospeciation
Comparison of the phylogenies of the angiosperm families (APG 2003) and Botrytis (fig. 4) showed that Botrytis species that parasitize hosts from one plant family do not often cluster together. A prominent example is found within the eudicot family Ranunculaceae, where B. calthae on the one hand and B. ficariarium/B. ranunculi on the other hand cluster apart in the Botrytis clades 1 and 2, respectively (fig. 6C). Also, the six Botrytis species infecting Alliaceae are widely dispersed over clade 2. Conversely, the related species B. aclada and B. paeoniae, clustering in clade 2a, are pathogens of monocot and eudicot hosts, respectively. Thus, there is no evidence for cospeciation of Botrytis species and their hosts. The observed lack of congruence between pathogen and host trees could be explained by host shifts. Host plants occupying similar habitats and growing during the same period of year offer possibilities for host shifting (Roy 2001; Nikoh and Fukatsu 2000). Knowledge of the distribution of genetic variation would have aided finding centers of origin of Botrytis species, and data on genetic/geographical associations would allow us to perform nested clade analyses to better delineate species based on sequence data. Unfortunately, detailed knowledge on the geographic distribution, centers of origin, and fossil data of Botrytis species are lacking. In addition, information on population dynamics and genetic variation is lacking for most Botrytis species. Studies on genetic variation have thus far only been performed with small population sizes (Van der Vlugt-Bergmans et al. 1993; Muoz et al. 2002; Huang et al. 2001), with one exception (Giraud et al. 1997).
The ability of Botrytis species to infect living host plants may result from a combination of at least four factors (see below): (1) possession of pathogenicity factors (e.g., toxins and cell-wall degrading enzymes) that confer the ability to kill and invade plant tissue (reviewed by Prins et al. [2000]), (2) the ability to avoid or counteract plant resistance mechanisms, (3) the ability to survive outside host-plant tissue under less favorable (e.g., low humidity, UV irradiation) conditions, and (4) the ability to reproduce and disperse.
Pathogenicity May Have Evolved from Saprotrophy via Acquisition of Pathogenicity Factors
In the genera Cochliobolus (Turgeon 2000) and Alternaria (Thomma 2003), the production of host-specific toxins is an important factor that distinguishes the necrotrophic pathogenic species from their saprophytic relatives. Such toxins mediate the currently described host-parasite specificity. It was proposed that plant pathogenic Cochliobolus species evolved from saprophytic or weakly pathogenic ancestors by the acquisition of genes involved in host-specific toxin production, possibly by horizontal gene transfer (Turgeon 2000).
Could plant pathogenic Botrytis species have evolved from saprophytic ancestors? Like saprophytic fungi, Botrytis species produce oxalic acid, extracellular cell-wall degrading enzymes (ten Have et al. 2002; Wubben et al. 1999), and oxidases (e.g., Edlich et al. 1989; Schouten et al. 2002; Rolke et al. 2004). Oxalic acid acidifies plant tissue to facilitate cell-wall hydrolysis (Dutton and Evans, 1996). However, in the case of Botrytis, cell-wall degradation is preceded by host-cell killing. Earlier studies of Botrytis diseases (reviewed by Brown [1965]) had already demonstrated that killing of plant cells and tissue maceration were distinct processes, likely with different causal agents. Brown (1965) proposed the involvement of a colloidal toxin, possibly an enzyme with protease or lipase activity. This hypothesis was further supported by reports that B. cinerea secretes a phospholipase able to induce cytoplasmic leakage (Shepard and Pitt 1976). The production of toxins conferring host specificity was reported for B. fabae infecting Vicia faba (Harrison 1980) and B. elliptica infecting lily (van Baarlen, Staats, and van Kan 2004).
Additional factors mediating host-parasite specificity are those that help to overcome plant resistance. The tulip secondary metabolite, tulipalin A, is toxic to all Botrytis species, except B. tulipae (Sch?nbeck and Schl?sser 1976). B. tulipae is the only Botrytis species able to infect tulip and it neutralizes tulipalin A enzymatically (our unpublished data), suggesting a causal relation between host specificity and defense compound detoxification. We propose that Botrytis speciation was driven by host shifts, resulting from the acquisition of pathogenicity genes that confer novel or wider host specificity.
Loss of Sexual Reproduction in Botrytis Species May Result from Negative Selection
Parasitic sclerotiniaceous fungi differ from their saprophytic conspecifics in their ability to survive in the absence of hosts by means of sclerotia, whereas saprophytes can only survive relatively mild adverse periods as mycelium in decayed host tissues (discussed in Willetts [1997]). Furthermore, the sclerotia serve as maternal parent in the production of apothecia, containing the sexual ascospores.
Sexual reproduction has at least three times been lost in Botrytis (fig. 6A). It is most parsimonious to assume that sexual reproduction was the ancestral form of reproduction of early Botrytis species. Botryotinia teleomorphs are generally found on decaying plant material in shaded locations with high humidity. The warm, humid climates that prevailed during the origin of present-day angiosperm Botrytis hosts were favorable for this form of reproduction (Willetts 1997). In drier climates with more intense solar radiaton, the thin-walled, easily dehydrated ascospores are possibly not an efficient manner of dispersal. The thick-walled, dehydration-tolerant and UV-tolerant macroconidia likely evolved as an adaptation to low humidity and higher UV irradiation, conferring the capacity to Botrytis species to infect the aerial parts of plants. We propose that loss of sexual reproduction in Botrytis may reflect habitat adaptation. In present-day drier habitats and under conditions of low humidity, ascospores are not produced, and only the asexual macroconidia are formed. This condition may result in negative selection against sexual reproduction. Under conditions adverse for sexual spore production and dispersal, investment in the production of the energy-costly teleomorph (Willetts 1997) will be wasted and selected against, leading to loss of sexual reproduction. It should be noted that microconidia, the "male" spermatial gametes, and sclerotia are still formed in exclusively asexually reproducing Botrytis species. Although the capacity to produce macroconidia confers a selective advantage in less humid habitats, the majority of Sclerotiniaceae spp. do not produce macroconidia.
Natural Interspecific Hybrids May Accelerate the Origin of Novel Species
The occurrence of natural hybrids in the group of Botrytis species causing onion neck rot and the resulting gene flow poses a difficulty for the maintenance of species identity. Nielsen and Yohalem (2001) suggested that polyploidization may have been the result of parasexual hybridization of B. aclada and B. byssoidea coinfecting onions. Alternatively, hybrid offspring may have been the result of interspecific sexual crosses (review by Schardl and Craven [2003]). Of the two parental species, only B. byssoidea has a sexual stage (table 1). During the hybridization events leading to the hybrid B. allii, the sclerotia of B. byssoidea have likely functioned as recipient for the spermatia (microconidia) produced by the presumed asexual B. aclada. At present, it is unknown whether the resulting hybrid offspring may only propagate vegetatively or whether backcrosses and crosses between hybrid offspring are possible. Repeated backcrossing to the parents may not be possible because of unequal chromosome number (Futuyma 1986). The hybridization event could be of recent origin because few mutations have accumulated in hybrid sequences, as was suggested previously by Nielsen and Yohalem (2001). Moreover, the observed differences in nucleotide sequence may originate from intraspecific variation present in both parental populations. There is no evidence that the hybridization events have led to host-range expansion within the B. byssoidea-aclada-allii complex, and hybrids do not exhibit increased pathogenicity to onion (Yohalem, Nielsen, and Nicolaisen 2003). The combination of interspecific sex, gene flow, and ploidy cycling gives ample opportunity for novel genotypes to arise, eventually differentiating into novel species (Kondrashov 1997).
Acknowledgements
The authors are grateful to Wilbert Flier for assistance in phylogenetic analyses and stimulating discussion, to Freek Bakker and Pedro Crous for critical review of the manuscript. This research was supported by the Dutch Technology Foundation STW, applied science division of NWO and the technology program of the Ministry of Economic Affairs (project WEB.5564).
References
APG. 2003. An update of the angiosperm phylogeny group classification for the orders and families of flowering plants: APG II. Bot. J. Linn. Soc. 141:399–436.
Berbee, M. L., M. Pirseyedi, and S. Hubbard. 1999. Cochliobolus phylogenetics and the origin of known, highly virulent pathogens, inferred from ITS and glyceraldehyde-3-phosphate dehydrogenase gene sequences. Mycologia 91:964–977.
Brown, W. 1965. Toxins and cell-wall dissolving enzymes in relation to plant disease. Annu. Rev. Phytopath. 3:1–21.
Büttner, P., F. Koch, K. Voigt, T. Quidde, S. Risch, R. Blaich, B. Bruckner, and P. Tudzynski. 1994. Variations in ploidy among isolates of Botrytis cinerea: implications for genetic and molecular analyses. Curr. Genet. 25:445–450.
Clark, C. A., and J. W. Lorbeer. 1976. Comparative histopathology of Botrytis squamosa and B. cinerea on onion leaves. Phytopathology 66:1279–1289.
Darlu, P., and G. Lecointre. 2002. When does the incongruence length difference test fail?. Mol. Biol. Evol. 19:432–437.
Dettman, J. R., D. J. Jacobson, and J. W. Taylor. 2003. A multilocus genealogical approach to phylogenetic species recognition in the model eukaryote Neurospora. Evolution 57:2703–2720.
Dutton, M. V., and C. S. Evans. 1996. Oxalate production by fungi: its role in pathogenicity and ecology in the soil environment. Can. J. Microbiol. 42:881–895.
Edlich, W., G. Lorenz, H. Lyr, E. Nega, and E.-H. Pommer. 1989. New aspects on the infection mechanism of Botrytis cinerea. Pers. Neth. J. Plant Pathol. 95:53–62.
Faretra, F., E. Antonacci, and S. Pollastro. 1988. Sexual behaviour and mating system of Botryotinia fuckeliana, teleomorph of Botrytis cinerea. J. Gen. Microbiol. 134:2543–2550.
Farris, J. S., V. A. Albert, M. Kallersjo, D. Lipscomb, and A. G. Kluge. 1996. Parsimony jackknifing outperforms neighbor-joining. Cladistics 12:99–124.
Farris, J. S., M. K?llersj?, A. G. Kluge, and C. Bult. 1995. Testing significance of incongruence. Cladistics 10:315–319.
Fitch, W. M. 1971. Toward defining the course of evolution: minimum change for a specific tree topology. Syst. Zool. 20:406–416.
Flor, H. H. 1955. Host-parasite interaction in flax rust: its genetics and other implications. Phytopathology 45:680–685.
Friesen, N., R. M. Fritsch, and F. R. Blattner. 2004. Phylogeny and new intrageneric classification of Allium L. (Alliaceae) based on nuclear ribosomal DNA ITS sequences. (in press)
Futuyma, D. J. 1986. Evolutionary biology. Sinauer Associates, Sunderland, Mass.
Giraud, T., D. Fortini, C. Levis, P. Leroux, and Y. Brygoo. 1997. RFLP markers show genetic recombination in Botryotinia fuckeliana (Botrytis cinerea) and transposable elements reveal two sympatric species. Mol. Biol. Evol. 14:1177–1185.
Harrison, J. G. 1980. The production of toxins by Botrytis fabae in relation to growth of lesions on field bean leaves at different humidities. Ann. Appl. Biol. 95:63–72.
Heath, M. C. 1991. The role of gene-for-gene interactions in determination of host species diversity. Phytopathology 81:127–130.
Hennebert, G. L. 1973. Botrytis and Botrytis-like genera. Persoonia 7:183–204.
Hennebert, G. L., and J. W. Groves. 1963. Three new species of Botryotinia on Ranunculaceae. Can. J. Bot. 41:341–370.
Holst-Jensen, A., M. Vaage, and T. Schumacher. 1998. An approximation to the phylogeny of Sclerotinia and related genera. Nordic J. Bot. 18:705–719.
Huang, J., T. Hsieh, G. A. Chastagner, and T. Hsiang. 2001. Clonal and sexual propagation in Botrytis elliptica. Mycol. Res. 105:833–842.
Jarvis, W. R. 1977. Botryotinia and Botrytis species; taxonomy, physiology and pathogenicity. Monograph No. 15, Canadian Department of Agriculture, Ottawa.
———. 1980. Taxonomy. Pp. 1–18 in J. R. Coley-smith, K. Verhoeff, W. R. Jarvis, eds. The biology of Botrytis. Academic Press, London.
Kniskern, J., and M. D. Rausher. 2001. Two modes of host-enemy coevolution. Popul. Ecol. 43:3–14.
Kondrashov, A. S. 1997. Evolutionary genetics of life cycles. Annu. Rev. Ecol. Syst. 28:391–435.
Liu, Y., and B. D. Hall. 2004. Body plan evolution of ascomycetes, as inferred from an RNA polymerase II phylogeny. Proc. Natl. Acad. Sci. USA 101:4507–4512.
Liu, Y., S. Whelen, and B. D. Hall. 1999. Phylogenetic relationships among ascomycetes: evidence from an RNA polymerase II subunit. Mol. Biol. Evol. 16:1799–1808.
MacFarlane, H. H. 1968. Review of Applied Mycology. Pp. 820 in plant host-pathogen index to volumes 1–40 (1922–1968).
Maddison, W. P., and D. R. Maddison. 2000. MacClade. Sinauer Associates, Sunderland, Mass.
Mansfield, J. W. 1980. The biology of Botrytis. Pp 181–218 in J. R. Coley- Smith, K. Verhoeff, and W. R. Jarvis, eds. The biology of Botrytis. Academic Press, London.
McDade, L. A. 1992. Hybrids and phylogenetic systematics II. The impact of hybrids on cladistic analysis. Evolution 46:1329–1346.
Muoz G., P. Hinrichsen, Y. Brygoo, and T. Giraud. 2002. Genetic characterisation of Botrytis cinerea populations in Chile. Mycol. Res. 106:594–601.
Nielsen, K., and D. S. Yohalem. 2001. Origin of a polyploid Botrytis pathogen through interspecific hybridization between Botrytis aclada and B. byssoidea. Mycologia 93:1064–1071.
Nikoh, N., and T. Fukatsu. 2000. Interkingdom host jumping underground: phylogenetic analysis of entomoparasitic fungi of the genus Cordyceps. Mol. Biol. Evol. 17:629–638.
O'Donnell, K., H. C. Kistler, E. Cigelnik, and R. C. Ploetz. 1998. Multiple evolutionary origins of the fungus causing Panama disease of banana: concordant evidence from nuclear and mitochondrial gene genealogies. Proc. Natl. Acad. Sci. USA 95:2044–2049.
Prins, T. W., P. Tudzynski, A. von Tiedemann, B. Tudzynski, A. ten Have, M. E. Hansen, K. Tenberge, and J. A. L. van Kan. 2000. Infections strategies of Botrytis cinerea and related necrotrophic pathogens. Pp. 33–65 in J. W. Kronstad, ed. Fungal pathology. Kluwer Academic Publisher, Dordrecht, The Netherlands.
Rokas, A., B. L. Williams, N. King, and S. B. Carroll. 2003. Genome-scale approaches to resolving incongruence in molecular phylogenies. Nature 425:798–804.
Rolke, Y., S. Liu, T. Quidde, B. Williamson, A. Schouten, K. Weltring, V. Siewers, K. B. Tenberge, B. Tudzynski, and P. Tudzynski. 2004. Functional analysis of H2O2-generating systems in Botrytis cinerea: the major Cu-Zn-superoxide dismutase (BCSOD1) contributes to virulence on French bean, whereas a glucose oxidase (BCGOD1) is dispensable. Mol. Plant Pathol. 5:17–27.
Ronquist, F., and J. P. Heulsenbeck. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19:1572–1574.
Roy, B. A. 2001. Patterns of association between crucifers and their flower-mimic pathogens: Host jumps are more common than coevolution or cospeciation. Evolution 55:41–53.
Schardl, C. L., and K. D. Craven. 2003. Interspecific hybridization in plant-associated fungi and oomycetes: a review. Mol. Ecol. 12:2861–2873.
Sch?nbeck, F., and E. Schl?sser. 1976. Preformed substances as potential protectants. Pp. 653–678 in R. Heitefuss and P. H. Williams, eds. Physiological plant pathology. Springer-Verlag, Berlin.
Schouten, A., C. A. M. Wagemakers, F. Stefanato, R. M. van der Kaaij, and J. A. L van Kan. 2002. Resveratrol acts as a natural profungicide and induces self-intoxication by a specific laccase. Mol. Microbiol. 43:883–894.
Shepard, D. V., and D. Pitt. 1976. Purification of a phospholipase from Botrytis cinerea and its effect on plant tissues. Phytochemistry 15:1465–1470.
Shirane, N. M. Masuko, and Y. Hayashi. 1989. Light microscopy observation of nuclei and mitotic chromosomes of Botrytis species. Phytopathology 79:728–730.
Soltis, P. S., D. E. Soltis, and M. W. Chase. 1999. Angiosperm phylogeny inferred from multiple genes as a tool for comparative biology. Nature. 402:402–404.
Smith, T. L. 1989. Disparate evolution of yeasts and filamentous fungi indicated by phylogenetic analysis of glyceraldehydes-3-phosphate dehydrogenase genes. Proc. Natl. Acad. Sci. USA 86:7063–7066.
Swofford, D. L. 2002. PAUP*: phylogenetic analysis using parsimony (*and other methods), version 4.0b10. Sinauer Associates, Sunderland, Mass.
ten Have, A., K. B. Tenberge, J. A. E. Benen, P. Tudzynski, J. Visser, and J. A. L. van Kan. 2002. The contribution of cell wall degrading enzymes to pathogenesis of fungal plant pathogens. Pp. 341–358 in F. Kempken, ed. The Mycota, Vol. 11. Springer-Verlag, Berlin, Heidelberg.
Thomma, B. P. H. J. 2003. Alternaria spp.: from general saprophyte to specific parasite. Mol. Plant Pathol. 4:225–236.
Thompson, J. D., T. J. Gibson, F. Plewniak, F. Jeanmougin, and D. G. Higgins. 1997. The CLUSTALX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25:4876–488.
Thompson, J. N., and J. J. Burdon. 1992. Gene-for-gene coevolution between plants and parasites. Nature 360:121–125.
Turgeon, B. G. 2000. Evolution of host specific virulence in Cochliobolus heterostrophus. Pp. 93–126 in J. W. Kronstad, ed. Fungal pathology. Kluwer Academic Publisher, Dordrecht, The Netherlands.
van Baarlen, P., M. Staats, and J. A. L. van Kan. 2004. Induction of programmed cell death in lily by the fungal pathogen Botrytis elliptica. Mol. Plant Pathol. (in press).
Van der Vlugt-Bergmans, C. J. B., B. F. Brandwagt, J. W. van't Klooster, C. A. M. Wagemakers, and J. A. L. van Kan. 1993. Genetic variation and segregation of DNA polymorphisms in Botrytis cinerea. Mycol. Res. 97:1193–1200.
White, T. J., T. Bruns, S. Lee, and J. Taylor. 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. Pp. 315–322 in M. A. Innis, D. H. Gelfand, J. J. Sninsky, and T. J. White, eds. PCR protocols: a guide to methods and applications. Academic Press, San Diego.
Willetts, H. J. 1997. Morphology, development and evolution of stromata/sclerotia and macroconidia of the Sclerotiniaceae. Mycol. Res. 101:939–952.
Wubben, J. P., W. Mulder, A. ten Have, J. A. L. van Kan, and J. Visser. 1999. Cloning and partial characterization of endopolygalacturonase genes from Botrytis cinerea. Appl. Environ. Microbiol. 65:1596–1602.
Yohalem, D. S., K. Nielsen, and M. Nicolaisen. 2003. Taxonomic and nomenclatural clarification of the onion neck rotting Botrytis species. Mycotaxon 85:175–182.(Martijn Staats, Peter van)
Correspondence: E-mail jan.vankan@wur.nl.
Abstract
The cosmopolitan genus Botrytis contains 22 recognized species and one hybrid. The current classification is largely based on morphological characters and, to a minor extent, on physiology and host range. In this study, a classification of the genus was constructed based on DNA sequence data of three nuclear protein-coding genes (RPB2, G3PDH, and HSP60) and compared with the traditional classification. Sexual reproduction and the host range, important fitness traits, were traced in the tree and used for the identification of major evolutionary events during speciation. The phylogenetic analysis corroborated the classical species delineation. In addition, the hybrid status of B. allii (B. byssoidea x B. aclada) was confirmed. Both individual gene trees and combined trees show that the genus Botrytis can be divided into two clades, radiating after the separation of Botrytis from other Sclerotiniaceae genera. Clade 1 contains four species that all colonize exclusively eudicot hosts, whereas clade 2 contains 18 species that are pathogenic on either eudicot (3) or monocot (15) hosts. A comparison of Botrytis and angiosperm phylogenies shows that cospeciation of pathogens and their hosts have not occurred during their respective evolution. Rather, we propose that host shifts have occurred during Botrytis speciation, possibly by the acquisition of novel pathogenicity factors. Loss of sexual reproduction has occurred at least three times and is supposed to be a consequence of negative selection.
Key Words: Botrytis ? necrotrophic fungus ? coevolution ? host shift ? Bayesian inference ? molecular phylogeny
Introduction
Fungi of the genus Botrytis Persoon are important pathogens of many agronomically important crops, such as grapevine, tomato, bulb flowers, and ornamental crops (Jarvis 1977). Botrytis diseases appear primarily as blossom blights and fruit rots but also as leaf spots and bulb rots in the field and in stored products. Botrytis species are necrotrophs, inducing host-cell death resulting in progressive decay of infected plant tissue. The pathogen produces abundantly sporulating gray mycelium on infected tissue. Macroconidia (mitotically produced spores) can be transported by wind over long distances. Botrytis overwinters in the soil as mycelium in decaying plant debris and as sclerotia, melanized mycelial survival structures. Some species frequently produce a sexual teleomorphic stage in which ascospores are produced in an apothecium. When collected in nature, apothecia are found under cool weather conditions, arising from sclerotia, which have developed on decayed plant parts in moist soil. Botrytis and its sexual form Botryotinia Whetzel comprise 22 species and one hybrid (Hennebert 1973; Yohalem, Nielsen, and Nicolaisen 2003) and are classified within the family Sclerotiniaceae Whetzel (Inoperculate Discomycetes). Delineation of species has traditionally been based on morphological characteristics, especially macroconidium ontogeny, and species have been named based on host association. Most species have a worldwide distribution or occur wherever their host crops are grown (Jarvis 1977).
The genus Botrytis comprises one generalist, B. cinerea, infecting over 200 eudicot hosts, especially senescing or otherwise weakened or wounded plants (MacFarlane 1968). All other species are considered specialists with a narrow host range. They infect only one or a few closely related species within the same plant genus (Mansfield 1980), with the exception of B. fabae, which can infect species of the genera Vicia, Lens, Pisum, and Phaseolus, all belonging to Fabaceae (Jarvis 1977). Specialized species occur on corolliferous monocotyledons and on members from the four eudicot families Fabaceae, Ranunculaceae, Geraniaceae, and Paeoniaceae (Jarvis 1977) (table 1). Narrow host range Botrytis species parasitize living leaves or bulbs of their hosts but may also occur as saprophytes. In many cases, host-specific Botrytis species are able to cause primary lesions on a nonhost, but these primary lesions fail to expand (Prins et al. 2000).
Table 1 Species Recognized by Hennebert (1973), Sexual Stage, Disease Symptoms, and Typical Host Plant Species
Several specialists are able to infect members of the same plant family or the same plant species. Three species of Botryotinia have been described occurring on the buttercup family Ranunculaceae; B. calthae occurs on Caltha palustris L., B. ranunculi on Ranunculus spp., and B. ficariarum on Ficaria verna. Occasionally B. calthae has been observed infecting Ficaria verna, but otherwise they are specific to their host (Hennebert and Groves 1963). As many as seven Botrytis species occur on Allium spp., which is one of the largest genera of the petaloid monocotyledons (Friesen, Fritsch, and Blattner 2004). Allium includes several economically important species, such common onion, garlic, chives, and leek, all of which are infected by B. byssoidea, B. aclada, and B. allii. The last species is an allodiploid hybrid based on mitotic chromosome counts, conidia size, restriction fragment length polymorphism (RFLPs), and sequence data (Nielsen and Yohalem 2001; Yohalem, Nielsen, and Nicolaisen 2003; Shirane, Masuko, and Hayashi 1989). B. byssoidea and B. aclada are considered to be its parental species (Nielsen and Yohalem 2001; Yohalem, Nielsen, and Nicolaisen 2003). Other Botrytis species infecting Allium spp. have a more restricted host range; they have become specialized on wild Allium spp. or infect only one or two economically important Allium crops.
To investigate whether closely related Botrytis species are pathogenic on closely related plant species, we compared phylogenies of fungi and their angiosperm hosts (APG 2003) for possible cospeciation. Parallel cladogenesis between host and pathogens would be indicative of ancient coevolution, a reciprocal process in which characteristics of one organism evolve in response to specific characteristics of another. The process of coevolution between pathogen and host is often based on gene-for-gene relationships, in which resistance gene alleles in the host are matched by virulence gene alleles in the pathogen (Flor 1955; Heath 1991; Thompson and Burdon 1992; Kniskern and Rausher 2001). Coevolution is more likely expected between obligate biotrophic pathogens or symbionts and their host plants than it is with necrotrophic pathogens. Obligate parasites do not kill host cells but get their nutrients either by penetrating living cells or by establishing close contact with them. Necrotrophic pathogens such as Botrytis need to kill host cells before the cells are invaded by the fungus (Clark and Lorbeer 1976; van Baarlen, Staats, and van Kan 2004), which might suggest absence or a low degree of coevolution. There is very limited knowledge about the phylogenetic relationships among members of the genus Botrytis. Holst-Jensen, Vaage, and Schumacher (1998) analyzed nuclear ribosomal internal transcribed spacer (ITS) DNA sequences and concluded that the Botryotinia teleomorph along with Botrytis anamorphs constitute a monophyletic lineage. However, the relationships among members of the genus Botrytis could not be resolved because of the limited phylogenetically informative ITS sequence characters.
In this study, we made use of fragments of three single-copy nuclear DNA (nDNA) genes encoding glyceraldehyde-3-phosphate dehydrogenase (G3PDH), Heat-shock Protein 60 (HSP60), and DNA-dependent RNA polymerase subunit II (RPB2). All three genes encode enzymes that are involved in basic cellular processes, and both G3PDH and RPB2 evolve at moderate evolutionary rates (Smith 1989; Berbee, Pirseyedi, and Hubbard 1999; Liu, Whelen, and Hall 1999; Liu and Hall 2004). The first objective was to clarify the evolutionary history of the genus Botrytis and investigate whether DNA sequence data support the classical species delineation. Second, the effect of the hybrid species B. allii on tree structure was examined because hybrids may cause topological changes, especially when parents are distantly related (McDade 1992). The third objective was to trace the evolution of the reproductive mode and the host-range spectrum within the genus Botrytis.
Materials and Methods
A set of 52 isolates (table 2) was chosen to represent all the recognized Botrytis species. Most strains were obtained from the Belgian Coordinated Collections of Microorganisms (BCCM, Belgium, Brussels). Additional strains were obtained from the Centraalbureau voor Schimmelcultures (CBS, Utrecht, The Netherlands) and personal collections. Single isolates of Sclerotinia sclerotiorum and Monilinia fructigena were selected as outgroups on the basis of a phylogenetic study by Holst-Jensen, Vaage, and Schumacher (1998). Strains were grown on malt agar (Oxoid) for 10 to 14 days in the dark at 18°C.
Table 2 Isolates of Botrytis, Origin, Year of Collection, and GenBank Accession Numbers for DNA Sequences Used in This Study
DNA Extraction, Amplification, and Sequencing
Mycelial tissue was harvested, lyophilized, submerged in liquid nitrogen and ground into a powder. Genomic DNA was extracted from 10 to 20 mg dry tissue using the Puregene DNA isolation kit (Gentra systems Inc./Biozym systems, Landgraaf, The Netherlands) according to the manufacturer's instructions. DNA pellets were dissolved in 100 ml of TE (10mM Tris-HCl [pH 8.0], 1mM EDTA) and stored at 4°C or –20°C.
Primer combinations (table 3) were designed or modified to amplify regions of three nuclear DNA genes (G3PDH, HSP60, and RPB2). PCR primers to amplify the RPB2 region that have been described previously by Liu, Whelen, and Hall (1999) were slightly modified. To amplify G3PDH and HSP60, forward and reverse primers were designed based on homologous gene sequences from GenBank and a Botrytis cinerea cDNA library (Genoscope, Centre National de Séquen?age, France). For amplification of ITS region, primers ITS1 and ITS4 were used (White et al. 1990). To facilitate batch sequencing of different PCR products, primers were extended with M13(-20) forward primers or M13 reverse primers.
Table 3 Primers Used for PCR Amplification and Sequencing
PCR amplification was carried out in a 50-ml reaction mixture that contained 10 to 50 ng of genomic DNA, 1X GeneAmp PCR buffer (PerkinElmer, Norwalk, Conn.), 200 mM of each deoxynucleoside triphosphate (Promega, Madison, Wis.), 0.2 pmol of each primer (Amersham Pharmacia Biotech), 2 mM MgSO4 and 1.0 U of AmpliTaq DNA polymerase (PerkinElmer). Amplifications were carried out in a Peltier Thermal Cycler-200 (Biozym, Landgraaf, The Netherlands). The following thermocycling pattern was used to amplify HSP60 and RPB2 gene fragments: 94°C for 5 min (1 cycle); 94°C for 30 s, 55°C for 30 s, and 72°C for 90 s (35 cycles), and then 72°C for 10 min (1 cycle). The same program with an annealing temperature of 64°C was used for G3PDH gene fragments. For amplification of the ITS region, an amplification protocol was used as described by White et al. (1990). The PCR products were separated on a 1% agarose-Tris-borate-EDTA (1x TBE) gel, containing ethidium bromide. The fragment size of the PCR product was verified by comparison to a 100-bp DNA Ladder (GibcoBRL). PCR products were purified using the QIAquick PCR purification kit (Qiagen) following manufacturer's instructions. Purified PCR products were sequenced (BaseClear Holding B.V., Leiden, The Netherlands) in both directions using M13(-20) forward and M13 reverse primers.
Six PCR products were cloned because the yield of amplified product was insufficient for direct sequencing: M. fructigena 9201 (locus RPB2), B. hyacinthi MUCL442 (loci RPB2, HSP60 and G3PDH), B. byssoidea MUCL94 (locus HSP60), and B. calthae MUCL1089 (locus G3PDH). Furthermore, multiple alleles of allopolyploid B. allii strains MUCL403 and MUCL1150 were cloned as well. PCR products were extracted from 1% agarose gel using the GFX PCR purification kit (Amersham Biosciences). PCR products were then ligated into a pGemT-Easy Vector (Promega) using standard protocols and introduced into electrocompetent Escherichia coli DH5 cells. Transformed cells were spread on LB-plates, each containing 100 mg/ml ampicillin, 20 mg/ml X-gal, and 0.1 mM IPTG and incubated at 37°C overnight. White colonies were screened for correct insert size using gene-specific PCR primers. To distinguish between alleles of each locus, 20 positive clones of B. allii strains MUCL1150 and MUCL403 were selected for restriction digestion. Restriction enzymes were selected based on polymorphic sites between aligned sequences of each locus of the presumed parental species B. aclada (isolate MUCL8415) and B. byssoidea (isolate MUCL94). Predicted restriction profiles were compared with digestion patterns of cloned fragments of both parental species, as well as both B. allii strains. G3PDH and HSP60 fragments were digested with the restriction enzyme EcoRV (Promega), whereas RPB2 fragments were digested with HindIII (Promega). Subsequently, individual clones, containing different alleles, were selected for sequencing based on restriction profiles. Clones were sequenced in both directions using SP6 and T7 primers (see table 3).
Phylogenetic Analysis
Automated sequence outputs were imported into Vector NTI suite 8.0 (InforMax, Inc.) for visual inspection of chromatographs and assembly of the contigs. For unresolved basepairs, the IUPAC nucleotide coding system was used. Sequences were compared between complementary strands for reading errors and aligned using ClustalX version 1.8 (Thompson et al. 1997). Alignments were visually inspected and adjusted manually when necessary. Regions of sequences with ambiguous alignment were excluded from all analyses (table 4). The full alignment containing all three loci is deposited in TREEBASE (accession number S1170). Individual sequences are deposited in GenBank under accession numbers AJ704990 to AJ705044, AJ716290 to AJ716305, AJ716046 to AJ716103, and AJ745662 to AJ745716 (see table 2).
Table 4 Summary of the RPB2, HSP60, G3PDH, and Combined Sequence Alignments
The basic data set for each locus consisted of 50 taxa: 48 Botrytis specimens and two outgroup species. Phylogenetic analyses were conducted for each of the three loci (RPB2, HSP60, and G3PDH) and for a combined data matrix of all three. To assess the impact of hybrids on tree structure, separate analyses were performed for the three DNA regions with and without hybrid taxa. Basic data sets plus five to eight hybrids were analyzed separately. All alleles of each locus of B. allii MUCL1150 and B. allii MUCL403, as well as of B. aclada MUCL3106 were treated as hybrid taxa. Because multiple alleles were found for hybrid taxa, combined analysis was performed with the basic data set only.
Parsimony analyses were carried out using the software package PAUP* version 4.0b10 (Swofford 2002) on a G4 Power Macintosh. The following settings were used: heuristic search with tree bisection-reconnection (TBR) branch swapping, with MULPARS (keeping all equally most-parsimonious trees) on, 1,000 replicates of random taxon-addition order, and all character transformations treated as equally likely (Fitch parsimony [Fitch 1971]). To minimize the time spent searching large numbers of trees, a limit of five trees was set for each replicate. After completing the replicates, all trees found were then used as starting trees for another round of swapping, with no limit to the number of trees. Gaps were treated as fifth character state, and multistate characters were treated as uncertain. A jackknife analysis was carried out to assess data structure and identify significant supported clades (63%). Search type was set to "fast" stepwise-addition (for individual loci) or full heuristic (for combined analysis). Jackknife values were generated by randomly deleting 37% of the characters for 10,000 replicates and using "Jac" resampling (Farris et al. 1996).
Bayesian analyses were performed using MrBayes version 3.0b4 (Ronquist and Heulsenbeck 2003). Each run consisted of four incrementally heated Markov chains run simultaneously, with heating values set to default (0.2). Default uniform priors were used for all model parameters (six substitution rates, four base frequencies, proportion of invariable sites, and alpha value of gamma distribution). Data partitions were set up for all coding and noncoding regions of each locus. HSP60 and G3PDH both contained two introns and two exons. For both loci individually, the two introns were pooled, and the two exons were pooled and analyzed as two partitions. Likelihood model parameters were separately optimized during MCMC searches for each data partition. Markov chains were initiated from a random tree and run for five-million generations; samples were taken every 100th generation. Convergence among chains was monitored by examining plots of log-likelihood values. Burn-in period (i.e. lack of improvement of log-likelihood values) was evaluated visually. All samples taken before burn-in were discarded in PAUP*, and the remaining samples were used to determine posterior probability (PP) distributions. Each run was performed twice. Bayesian 50% majority-rule consensus trees were generated in PAUP*. Branches with at least 63% jackknife support and with at least 0.95 Bayesian PP were considered as well supported.
Concordance of combined data sets was evaluated by comparing jackknife topologies via visual inspection and partition homogeneity tests (PHT [Farris et al. 1995]) implemented with PAUP*. Visual inspection permits the precise location of areas of strong discordance (branches conflicting with jackknife values at least 85% and Bayes PP at least 0.95) between phylogenies generated by different data sets. PHTs were carried out for all pairwise combinations of the three data sets and for all data sets combined using heuristic search (TBR) with 100 random-addition replicates and informative characters only. The null hypothesis of congruence was rejected if P < 0.001 (Darlu and Lecointre 2002; Dettman, Jacobson, and Taylor 2003).
Functional characters were mapped onto the molecular phylogeny using MacClade version 4 (Maddison and Maddison 2000): reproductive mode (two states; strictly asexual or asexual and sexual); host-plant group (two states; monocotyledons or eudicotyledons) and host-plant family (10 states; Hyacinthaceae, Liliaceae, Alliaceae, Amaryllidaceae, Iridaceae, Paeoniaceae, Geraniaceae, Ranunculaceae, Fabaceae, or Polyphagous on eudicotyledons). All characters were equally weighted under parsimony. Character states from literature (Jarvis 1977, 1980; Hennebert and Groves 1963; Hennebert 1973) are listed in table 1.
Results
In preliminary experiments, sequence data of eight protein-coding loci and the noncoding ITS region were acquired for multiple isolates of three Botrytis species. The three most-informative loci (G3PDH, HSP60, and RPB2) were used as starting point for this study. A total of 188 new sequences of 54 specimens are reported here. Sequences of protein-coding gene fragments and ITS sequences were first evaluated in a standard GenBank Blast search. Nearly all obtained sequences corresponded with listed Botrytis cinerea or Sclerotinia sclerotiorum sequences. The ITS sequence of B. anthophila Bondartsev, a species reported in Jarvis (1977), was identical to Rhizoctonia sp. ITS sequences, and this sequence was, therefore, excluded from further analyses. An alignment of ITS characters, consisting of 453 nucleotide positions of the sequences listed here (table 2) and of sequences of Botrytis species reported by Holst-Jensen, Vaage, and Schumacher (1998), contained a limited number of singletons and phylogenetic informative characters (data not shown). Therefore, ITS sequence data were excluded from further analyses.
Analysis of G3PDH Sequence Data and the Influence of Hybrid Taxa on Tree Structure
The basic sequence matrix (table 4) consists of 890 sites, of which 186 (21%) were variable characters and 126 (14%) were potentially parsimony informative. Analysis of G3PDH sequences with equal (Fitch) weights resulted in two equally most-parsimonious trees of 312 steps, with a consistency index (CI) of 0.73 and a retention index (RI) of 0.89. Relationships among the main clades were topologically identical in MP and Bayesian consensus trees (data not shown), except for some minor differences. Comparison between replicate Bayesian analyses resulted in identical majority-rule consensus trees, with only marginally lower mean log-likelihood values (–2998.75 and –3025.33). Bayesian PP and jackknife percentages of branches are included on the MP semistrict consensus tree in figure 1. The semistrict consensus tree topology of the two most-parsimonious trees (MPTs) contains 24 nodes with at least 63% jackknife and at least 95% PP support.
FIG. 1.— Semistrict consensus of two most-parsimonious trees based on G3PDH data. This tree was compatible overall in highly supported lineages to the Bayesian 50% majority-rule consensus tree. Jackknife frequencies (10,000 replicates) are shown above each node; Bayesian posterior probabilities (PP) are shown below each node. Letters indicate major clades with significant or strong support (see table 5). A dash indicates support for the branch was less than 50% jackknife and less than 0.50 PP.
Table 5 Cladistic Behavior of RPB2, HSP60, G3PDH, and Combined Data Sets with and Without Hybrid Taxa
The G3PDH data set plus hybrids consists of 53 taxa, including two alleles of both B. allii MUCL1150 and B. allii MUCL403. The heuristic search yielded 11 MPTs with higher tree length and slightly lower CI (L = 323, CI = 0.71, and RI = 0.89). MP and Bayesian analyses resulted in almost identical trees (data not shown). The semistrict consensus of those fundamental trees recovers the same major clades as in the G3PDH data sets without hybrid taxa, and branch support was only marginally lower (table 5). Within these trees, one allele of B. allii strain MUCL1150 (allele 1) and two alleles of B. allii strain MUCL403 (alleles 1 and 2) are placed in a clade together with B. aclada (strains MUCL8415, MUCLPRI006, and MUCL3106). Allele 2 of B. allii strain MUCL403 is placed together with B. byssoidea.
Analysis of HSP60 Sequence Data and the Influence of Hybrid Taxa on Tree Structure
The basic sequence matrix consists of 983 characters, from which 20 positions were removed because of ambiguity in the alignment (table 4). Of the remaining 963 characters, 225 (23.4%) were variable, and 158 (16.4%) were potentially phylogenetically informative. The Fitch parsimony recovered 320 shortest trees, each 373 steps long (CI = 0.74, RI = 0.90). The Bayesian tree was consistent with the consensus of these trees, and separate Bayesian analyses resulted in identical trees (mean log-likelihood values –3357.89 and –3390.86 [data not shown]). Bayesian PP and jackknife percentages of branches are included on the MP semistrict consensus tree in figure 2. The semistrict consensus tree topology of the 320 MTPs contains 23 nodes with at least 63% jackknife and at least 95% PP support.
FIG. 2.— Semistrict consensus of 320 most-parsimonious trees based on HSP60 data. This tree was compatible overall in highly supported lineages to the Bayesian 50% majority-rule consensus tree. Jackknife frequencies (10,000 replicates) are shown above each node; Bayesian posterior probabilities (PP) are shown below each node. Letters indicate major clades with significant or strong support (see table 5). A dash indicates support for the branch was less than 50% jackknife and less than 0.50 PP.
The HSP60 data set plus hybrids consist of 58 taxa, including three alleles of B. allii strain MUCL1150 and four alleles of B. allii strain MUCL403. Fitch analysis gave 320 trees of 404 steps, with CI = 0.69 and RI = 0.89. Hybrid taxa had strong influence on tree structure in both MP and Bayesian analyses, resulting in the collapse of (sub)clades e, and k (table 5). MP and Bayesian analyses resulted in almost identical topologies (data not shown); however, there are also important differences. Bayesian analysis strongly supported (100% PP) a large clade containing members of clade e, B. convoluta, B. paeoniae, and a subclade (98% PP) placing hybrids as basal members of their derived parents B. byssoidea and B. aclada. This clade is highly supported by Bayesian posterior probabilities but is not present at all in the jackknife tree.
Analysis of RPB2 Sequence Data and the Influence of Hybrid Taxa on Tree Structure
The basic sequence matrix consists of 1096 positions, of which 274 (25%) were variable and 203 (18.5%) were potentially parsimony informative (table 4). Analysis of RPB2 sequences with equal (Fitch) weights resulted in seven equally most-parsimonious trees of 439 steps, with a CI of 0.73 and an RI of 0.91. Relationships among the main clades were topologically identical in MP and Bayesian consensus trees (data not shown). Comparison between replicate Bayesian analyses resulted in identical majority-rule consensus trees, with only marginally lower mean log-likelihood values (–3955.80 and –3970.45; not shown). Bayesian PP and jackknife percentages of branches are included on the MP semistrict consensus tree in figure 3. Of the three individual regions, RPB2 provided the highest number of significantly supported nodes (26 nodes with at least 63% jackknife and at least 95% PP support), which are mainly located at the terminal branches. The RPB2 topology contains the three main (sub)clades (R, O, and E) that also occur in the combined analysis (table 5), although relationships among these clades remained unresolved.
FIG. 3.— Semistrict consensus of seven most-parsimonious trees based on RPB2 data. This tree was compatible overall in highly supported lineages to the Bayesian 50% majority-rule consensus tree. Jackknife frequencies (10,000 replicates) are shown above each node; Bayesian posterior probabilities (PP) are shown below each node. Letters indicate major clades with significant or strong support (see table 5). A dash indicates support for the branch was less than 50% jackknife and less than 0.50 PP.
The RPB2 data set plus hybrids consisted of 55 taxa, including two alleles of both B. allii strain MUCL1150 and B. allii strain MUCL403. The heuristic search yielded six MPTs with higher tree length and slightly lower CI (L = 459, CI = 0.70, and RI = 0.91). MP and Bayesian analyses resulted in topologically identical trees (data not shown). The semistrict consensus of those fundamental trees recovers the same major clades as the basic data set (table 5). However, one extra clade was recovered, although with low support (<50% jackknife and 94% PP), consisting of clade O, B. aclada (MUCL8415, MUCLPRI006, and MUCL3106), and B. allii (MUCL1150 allele 1 and MUCL403 alleles 1 and 2). Allele 2 of B. allii strain MUCL1150 was placed together with B. byssoidea. Furthermore, as a result of hybrid taxa, jackknife branch support for nodes E and G decreased from strongly supported (86% and 87%, respectively) to well supported (73% and 78%, respectively).
Assessment of Between–Data Sets Congruence
The G3PDH tree was largely congruent with the RPB2 tree; in both trees, there is good support for three main (sub)clades (R, O, and G [figs. 1 and 3]), but arrangements within each clade could not be clearly resolved. At the base of the tree, some resolution is provided; however, branches indicating relationships between clades are not supported (fig. 3). In comparison to the RPB2 topology, more resolution is provided by HSP60 at the base of the tree (figs. 2 and 3). Contrary to the RPB2 tree, the positions of B. galanthina and B. hyacinthi/B. croci have been exchanged, as well as of B. polyblastis and B. tulipae. As with RPB2, HSP60 placed B. porri as part of a basal trichotomy. Jackknife topologies were used to test whether the three data sets have different underlying phylogenetic histories. Strong incongruence (>85% jackknife and 0.95 PP) between RPB2 and HSP60 phylogenies was evident in the placement of B. cinerea strain B05.10 and B. polyblastis. None of the strongly supported clades was incongruent between the other data sets (RPB2 versus G3PDH and G3PDH versus HSP60). The PHT was used as an alternative method to determine whether each possible combination of loci differs significantly in phylogenetic structure. No significant conflict was detected for all pairwise comparisons of each locus (RPB2-HSP60, RPB2-G3PDH, and HSP60-G3PDH; 0.007 P 0.001) and the three combined (P = 0.001).
Molecular Phylogeny Distinguishes Two Major Groups
Because no significant between–data set incongruence was detected and because the individual jackknife topologies produced nearly identical results, the three regions were combined directly. Analysis of this combined matrix resulted in 80 trees of 1,151 steps (CI = 0.72 and RI = 0.89). The trees were 27 steps longer than the combined number of steps from the RPB2, HSP60, and G3PDH trees (table 5). The Bayesian tree was consistent with the MP tree, but as with the RPB2 MP tree, B. polyblastis was strongly placed (Bayesian PP = 0.99) as a sister taxa to B. narcissicola, B. gladiolorum, and B. byssoidea (see figures 3, 4A, and 4B). Furthermore, B. tulipae was placed basal to the node tying B. globosa and B. sphaerosperma. The relative branching order of ingroup species and mean log-likelihood values were comparable among two replicate Bayesian analyses (–10389.27 and –10402.63 [data not shown]). In the combined molecular phylogeny, the overall number of supported nodes was greater (26) than for the trees of each separate analysis. Also, most nodes had greater support, especially at the spine, and most were supported strongly. The exception involved the positions of the B. cinerea B05.10 and B. polyblastis, which were poorly resolved as a result of discordant positions of both species in separate analyses (see figures 2 and 3). In contrast to other relationships in the genus, placement of B. polyblastis and B. tulipae was unresolved.
FIG. 4.— Molecular phylogeny for 48 Botrytis taxa and two outgroup species based on combined G3PDH, HSP60, and RPB2 data. (A) Semistrict consensus of 80 most-parsimonious trees. Jackknife frequencies (10,000 replicates) are shown above each node; Bayesian posterior probabilities (PP) are shown below each node. Letters indicate major clades with significant or strong support (see table 5). A dash indicates the support for the branch was less than 50% jackknife and less than 0.50 PP. Bars indicate the (sub)clades. (B) Bayesian inference showing a 50% majority-rule consensus tree (mean log-likelihood = –10389.27) from a five-million generation MCMC analysis.
Multiple isolates of the same species usually clustered together, and their gene sequences were identical or varied at fewer sites within species in the aligned sequence positions than between species. Molecular phylogenetic analysis of the DNA sequences supports the traditional morphological species classification. In addition, the hybrid status of B. allii (B. byssoidea xB. aclada [Yohalem, Nielsen, and Nicolaisen 2003]) was confirmed (fig. 5). The two isolates of B. hyacinthi did not appear to be monophyletic, because they lacked shared characters that distinguished them as a group (fig. 4B).
FIG. 5.— Schematic representation of sequence alignment of multiple HSP60 alleles for allopolyploid B. allii (strains MUCL403 and MUCL1150) and its presumed parental species B. aclada (isolate MUCL8415) and B. byssoidea (isolate MUCL94). Black and gray lines indicate stretches of sequence polymorphisms of B. aclada MUCL8415 and B. byssoidea MUCL94, respectively. Black and gray arrows indicate EcoRV restriction endonuclease cleavage sites.
Based upon the semistrict consensus tree, two main clades were identified in the genus Botrytis, and clade 2 was divided into five subclades, as within this clade strongly supported clusters of species could be identified. The base of the tree is characterized by a deep split leading to clade 1 and clade 2. Clade 1 consists of the species B. calthae, B. fabae, B. cinerea, and B. pelargonii and is separated from clade 2 by a relatively long branch. The internode separating B. pelargonii from B. cinerea was very short, reflecting the high amount of interspecific sequence similarity between both species. There was only one unique polymorphism on a total of 2,955 bp between both species.
B. aclada and B. paeoniae are less closely related to other taxa grouping at a basal position of clade 2 (81% jackknife and 1.0 PP), and they were, therefore, assigned to subclade 2a. Individual gene trees did not support this clade, although both species always grouped closely together. B. convoluta could not unambiguously be placed in either subclade within clade 2 and has, therefore, been assigned to subclade 2d. Both subclade 2b and subclade 2c were defined based on high support values (100/1.0 and 99/1.0, respectively) in combined analysis and the consistent support in separate analyses. The relationships between B. ranunculi and B. ficariarum in subclade 2b and between B. hyacinthi and B. croci in subclade 2c were unresolved. Although B. porri is closely related, there was no support for its placement in either subclade within clade 2 in combined and separate nDNA analyses. B. porri was, therefore, placed in a separate clade, designated as subclade 2e.
The Allodiploid Hybrid B. allii May Contain Up to Four Different Alleles
To confirm earlier reports about the hybrid status of B. allii (Nielsen and Yohalem 2001; Yohalem, Nielsen, and Nicolaisen 2003) restriction profiles and sequence data of cloned fragments were compared with those of its parental species B. aclada and B. byssoidea. Clones of B. aclada strain MUCL8415 and B. byssoidea strain MUCL94 could be distinguished based on the unique digestion patterns among the two species. Among cloned fragments of B. allii (strains MUCL1150 and MUCL403), digestion patterns of either the B. aclada-type or the B. byssoidea-type (data not shown) were detected for all three loci, indicating the presence of the two parental alleles. However, in case of HSP60, divergent patterns were also detected. Fragment sizes could be explained by the presence of three restriction sites (fig. 5), of which two sites are present in B. byssoidea and one is present in B. aclada. Alleles were sequenced to verify this interpretation. For HSP60, four alleles were identified for B. allii strain MUCL403 and three alleles were identified for B. allii strain MUCL1150. Hybrid alleles contained stretches of sequences originating from both parental alleles (fig. 5). For the G3PDH and RPB2 genes, hybrid alleles have also been found despite identical restriction patterns in both parental alleles.
The HSP60 locus of B. aclada strain MUCL3106 was identical to B. byssoidea except for two polymorphic sites, whereas both RPB2 and G3PDH were identical to B. aclada strain MUCL8415. Because fragments of B. aclada strain MUCL3106 were directly sequenced, allelic variants have not been detected for each locus.
Loss of Sexual Reproduction Has Evolved Three Times
Two important fitness traits, reproductive mode and host-range spectrum, were expressed as multistate "characters" and optimized on the cladogram (fig. 4A) using parsimony. Sexual reproduction has not been found in seven species, five of which parasitize monocot hosts and two parasitize eudicot hosts (table 1). Based on reproductive mode, sexual reproduction was reconstructed as the ancestral condition (fig. 6A). Strict asexual reproduction was inferred to be a homoplasious trait, and loss of sex appears to have occurred three times. Two groups, containing two and four strictly asexual reproducing Botrytis species are present in the tree (B. aclada/B. paeoniae and B. hyacinthi/B. croci/B. galanthina/B. tulipae). Within clade 1, the asexual species B. fabae separates from a clade predominated by sexual reproduction. There is indication for a reversal from strictly asexual reproduction to sexual reproduction of six taxa in clade 2c.
FIG. 6.— Reconstructions of character-state evolution mapped onto the combined molecular phylogeny (the same topology as in figure 4A) for Botrytis under parsimony: (A) reproductive mode, (B) host-plant group, and (C) host-plant family.
Host and Pathogen Phylogenies Are Not Congruent
To assess the possible occurrence of cospeciation patterns between Botrytis species and their hosts, the distribution of host-plant family or angiosperm group was mapped on the cladogram (fig. 6B and C). The order Asparagales is parasitized by 14 of the previously recognized Botrytis species, the order Ranunculales by three species, and the Liliales by two species. Three eudicot orders contain one specialist Botrytis species, and these orders can be parasitized by the generalist B. cinerea as well. Botrytis species parasitizing either eudicot or monocot hosts do not tend to cluster together (fig. 6B). Within clade 2, transitions from pathogens infecting monocot hosts to eudicot hosts seem to have occurred twice. Furthermore, Botrytis species parasitizing hosts from the same plant family appear to be unrelated (fig. 6C). The best example of incongruency occurs in the host-plant family Ranunculaceae, containing three Botrytis species, of which B. calthae clusters in clade 1, and the species B. ficariarum/B. ranunculi cluster in clade 2b.
Discussion
DNA Sequence Data Support the Morphological Species Delineation
In the present study, we conducted molecular phylogenetic analyses of 22 Botrytis species and one hybrid. Our analysis provides the first comprehensive phylogenetic study of the entire genus. Our analysis further supports monophyly of the genus Botrytis, as has previously been proposed by Holst-Jensen, Vaage, and Schumacher 1997. The phylogenetic analyses of DNA sequence data show that the sequences from multiple isolates of the same species group together and, therefore, support the classical Botrytis species status (Hennebert 1973; Jarvis 1977). Morphological characters, together with DNA sequence information, can be used to identify Botrytis species; for instance, in a diagnostic key of Botrytis species. Such a key may be a potentially powerful tool for diagnostics of this important group of plant pathogens.
The combined analysis of all three loci supported two main clades (fig. 4) and provides more resolution than individual gene phylogenies (fig. 1–3). This result shows the value of conducting phylogenetic analysis on multiple genes, which is an established tool to address complex phylogenetic questions (Rokas et al. 2003; Soltis, Soltis, and Chase 1999; O'Donnell et al. 1998).
The taxonomic status of B. croci and B. hyacinthi remains uncertain; both species exhibit high sequence similarity, which makes the separation into distinct species questionable. In this study, data of B. croci were based on a single isolate, so additional sequencing is needed to resolve its status as a separate species.
Reports about the hybrid status of B. allii (Nielsen and Yohalem 2001; Yohalem, Nielsen, and Nicolaisen 2003) were confirmed. Interspecific hybrid species were identified by the presence of multiple HSP60, RPB2, and G3PDH gene copies that consistently grouped with B. aclada and B. byssoidea within the Botrytis phylogeny. The occurrence of up to four different alleles per locus (fig. 5) is indicative of a diploid, dikaryotic genome. However, higher ploidy levels or aneuploidy in these isolates cannot be excluded. The selection of alleles for sequence analysis was based on polymorphic digestion patterns in individual clones of amplified loci; additional alleles that were nonpolymorphic in restriction sites may have escaped our attention. The mixture of parental and recombinant alleles indicates that reciprocal recombination and gene conversion have occurred several times during diversification within the hybrid B. allii. These hybrids were placed as basal members of the clade that includes their parents (table 5). Major topological change was caused by hybrid sequences in the HSP60 phylogeny, resulting in the collapse of the portion of the cladogram that included the hybrid's parents. In case of RPB2 and G3PDH, the cladistic relationships remained largely unchanged, causing only a slight decrease in support values in branches leading to the hybrid's parents. Except for B. alli, none of the 22 Botrytis species appeared to be hybrids, nor was there any indication from sequence data or host-plant range that they arose by hybridization.
Incongruence of Host and Pathogen Phylogenies Does Not Support Cospeciation
Comparison of the phylogenies of the angiosperm families (APG 2003) and Botrytis (fig. 4) showed that Botrytis species that parasitize hosts from one plant family do not often cluster together. A prominent example is found within the eudicot family Ranunculaceae, where B. calthae on the one hand and B. ficariarium/B. ranunculi on the other hand cluster apart in the Botrytis clades 1 and 2, respectively (fig. 6C). Also, the six Botrytis species infecting Alliaceae are widely dispersed over clade 2. Conversely, the related species B. aclada and B. paeoniae, clustering in clade 2a, are pathogens of monocot and eudicot hosts, respectively. Thus, there is no evidence for cospeciation of Botrytis species and their hosts. The observed lack of congruence between pathogen and host trees could be explained by host shifts. Host plants occupying similar habitats and growing during the same period of year offer possibilities for host shifting (Roy 2001; Nikoh and Fukatsu 2000). Knowledge of the distribution of genetic variation would have aided finding centers of origin of Botrytis species, and data on genetic/geographical associations would allow us to perform nested clade analyses to better delineate species based on sequence data. Unfortunately, detailed knowledge on the geographic distribution, centers of origin, and fossil data of Botrytis species are lacking. In addition, information on population dynamics and genetic variation is lacking for most Botrytis species. Studies on genetic variation have thus far only been performed with small population sizes (Van der Vlugt-Bergmans et al. 1993; Muoz et al. 2002; Huang et al. 2001), with one exception (Giraud et al. 1997).
The ability of Botrytis species to infect living host plants may result from a combination of at least four factors (see below): (1) possession of pathogenicity factors (e.g., toxins and cell-wall degrading enzymes) that confer the ability to kill and invade plant tissue (reviewed by Prins et al. [2000]), (2) the ability to avoid or counteract plant resistance mechanisms, (3) the ability to survive outside host-plant tissue under less favorable (e.g., low humidity, UV irradiation) conditions, and (4) the ability to reproduce and disperse.
Pathogenicity May Have Evolved from Saprotrophy via Acquisition of Pathogenicity Factors
In the genera Cochliobolus (Turgeon 2000) and Alternaria (Thomma 2003), the production of host-specific toxins is an important factor that distinguishes the necrotrophic pathogenic species from their saprophytic relatives. Such toxins mediate the currently described host-parasite specificity. It was proposed that plant pathogenic Cochliobolus species evolved from saprophytic or weakly pathogenic ancestors by the acquisition of genes involved in host-specific toxin production, possibly by horizontal gene transfer (Turgeon 2000).
Could plant pathogenic Botrytis species have evolved from saprophytic ancestors? Like saprophytic fungi, Botrytis species produce oxalic acid, extracellular cell-wall degrading enzymes (ten Have et al. 2002; Wubben et al. 1999), and oxidases (e.g., Edlich et al. 1989; Schouten et al. 2002; Rolke et al. 2004). Oxalic acid acidifies plant tissue to facilitate cell-wall hydrolysis (Dutton and Evans, 1996). However, in the case of Botrytis, cell-wall degradation is preceded by host-cell killing. Earlier studies of Botrytis diseases (reviewed by Brown [1965]) had already demonstrated that killing of plant cells and tissue maceration were distinct processes, likely with different causal agents. Brown (1965) proposed the involvement of a colloidal toxin, possibly an enzyme with protease or lipase activity. This hypothesis was further supported by reports that B. cinerea secretes a phospholipase able to induce cytoplasmic leakage (Shepard and Pitt 1976). The production of toxins conferring host specificity was reported for B. fabae infecting Vicia faba (Harrison 1980) and B. elliptica infecting lily (van Baarlen, Staats, and van Kan 2004).
Additional factors mediating host-parasite specificity are those that help to overcome plant resistance. The tulip secondary metabolite, tulipalin A, is toxic to all Botrytis species, except B. tulipae (Sch?nbeck and Schl?sser 1976). B. tulipae is the only Botrytis species able to infect tulip and it neutralizes tulipalin A enzymatically (our unpublished data), suggesting a causal relation between host specificity and defense compound detoxification. We propose that Botrytis speciation was driven by host shifts, resulting from the acquisition of pathogenicity genes that confer novel or wider host specificity.
Loss of Sexual Reproduction in Botrytis Species May Result from Negative Selection
Parasitic sclerotiniaceous fungi differ from their saprophytic conspecifics in their ability to survive in the absence of hosts by means of sclerotia, whereas saprophytes can only survive relatively mild adverse periods as mycelium in decayed host tissues (discussed in Willetts [1997]). Furthermore, the sclerotia serve as maternal parent in the production of apothecia, containing the sexual ascospores.
Sexual reproduction has at least three times been lost in Botrytis (fig. 6A). It is most parsimonious to assume that sexual reproduction was the ancestral form of reproduction of early Botrytis species. Botryotinia teleomorphs are generally found on decaying plant material in shaded locations with high humidity. The warm, humid climates that prevailed during the origin of present-day angiosperm Botrytis hosts were favorable for this form of reproduction (Willetts 1997). In drier climates with more intense solar radiaton, the thin-walled, easily dehydrated ascospores are possibly not an efficient manner of dispersal. The thick-walled, dehydration-tolerant and UV-tolerant macroconidia likely evolved as an adaptation to low humidity and higher UV irradiation, conferring the capacity to Botrytis species to infect the aerial parts of plants. We propose that loss of sexual reproduction in Botrytis may reflect habitat adaptation. In present-day drier habitats and under conditions of low humidity, ascospores are not produced, and only the asexual macroconidia are formed. This condition may result in negative selection against sexual reproduction. Under conditions adverse for sexual spore production and dispersal, investment in the production of the energy-costly teleomorph (Willetts 1997) will be wasted and selected against, leading to loss of sexual reproduction. It should be noted that microconidia, the "male" spermatial gametes, and sclerotia are still formed in exclusively asexually reproducing Botrytis species. Although the capacity to produce macroconidia confers a selective advantage in less humid habitats, the majority of Sclerotiniaceae spp. do not produce macroconidia.
Natural Interspecific Hybrids May Accelerate the Origin of Novel Species
The occurrence of natural hybrids in the group of Botrytis species causing onion neck rot and the resulting gene flow poses a difficulty for the maintenance of species identity. Nielsen and Yohalem (2001) suggested that polyploidization may have been the result of parasexual hybridization of B. aclada and B. byssoidea coinfecting onions. Alternatively, hybrid offspring may have been the result of interspecific sexual crosses (review by Schardl and Craven [2003]). Of the two parental species, only B. byssoidea has a sexual stage (table 1). During the hybridization events leading to the hybrid B. allii, the sclerotia of B. byssoidea have likely functioned as recipient for the spermatia (microconidia) produced by the presumed asexual B. aclada. At present, it is unknown whether the resulting hybrid offspring may only propagate vegetatively or whether backcrosses and crosses between hybrid offspring are possible. Repeated backcrossing to the parents may not be possible because of unequal chromosome number (Futuyma 1986). The hybridization event could be of recent origin because few mutations have accumulated in hybrid sequences, as was suggested previously by Nielsen and Yohalem (2001). Moreover, the observed differences in nucleotide sequence may originate from intraspecific variation present in both parental populations. There is no evidence that the hybridization events have led to host-range expansion within the B. byssoidea-aclada-allii complex, and hybrids do not exhibit increased pathogenicity to onion (Yohalem, Nielsen, and Nicolaisen 2003). The combination of interspecific sex, gene flow, and ploidy cycling gives ample opportunity for novel genotypes to arise, eventually differentiating into novel species (Kondrashov 1997).
Acknowledgements
The authors are grateful to Wilbert Flier for assistance in phylogenetic analyses and stimulating discussion, to Freek Bakker and Pedro Crous for critical review of the manuscript. This research was supported by the Dutch Technology Foundation STW, applied science division of NWO and the technology program of the Ministry of Economic Affairs (project WEB.5564).
References
APG. 2003. An update of the angiosperm phylogeny group classification for the orders and families of flowering plants: APG II. Bot. J. Linn. Soc. 141:399–436.
Berbee, M. L., M. Pirseyedi, and S. Hubbard. 1999. Cochliobolus phylogenetics and the origin of known, highly virulent pathogens, inferred from ITS and glyceraldehyde-3-phosphate dehydrogenase gene sequences. Mycologia 91:964–977.
Brown, W. 1965. Toxins and cell-wall dissolving enzymes in relation to plant disease. Annu. Rev. Phytopath. 3:1–21.
Büttner, P., F. Koch, K. Voigt, T. Quidde, S. Risch, R. Blaich, B. Bruckner, and P. Tudzynski. 1994. Variations in ploidy among isolates of Botrytis cinerea: implications for genetic and molecular analyses. Curr. Genet. 25:445–450.
Clark, C. A., and J. W. Lorbeer. 1976. Comparative histopathology of Botrytis squamosa and B. cinerea on onion leaves. Phytopathology 66:1279–1289.
Darlu, P., and G. Lecointre. 2002. When does the incongruence length difference test fail?. Mol. Biol. Evol. 19:432–437.
Dettman, J. R., D. J. Jacobson, and J. W. Taylor. 2003. A multilocus genealogical approach to phylogenetic species recognition in the model eukaryote Neurospora. Evolution 57:2703–2720.
Dutton, M. V., and C. S. Evans. 1996. Oxalate production by fungi: its role in pathogenicity and ecology in the soil environment. Can. J. Microbiol. 42:881–895.
Edlich, W., G. Lorenz, H. Lyr, E. Nega, and E.-H. Pommer. 1989. New aspects on the infection mechanism of Botrytis cinerea. Pers. Neth. J. Plant Pathol. 95:53–62.
Faretra, F., E. Antonacci, and S. Pollastro. 1988. Sexual behaviour and mating system of Botryotinia fuckeliana, teleomorph of Botrytis cinerea. J. Gen. Microbiol. 134:2543–2550.
Farris, J. S., V. A. Albert, M. Kallersjo, D. Lipscomb, and A. G. Kluge. 1996. Parsimony jackknifing outperforms neighbor-joining. Cladistics 12:99–124.
Farris, J. S., M. K?llersj?, A. G. Kluge, and C. Bult. 1995. Testing significance of incongruence. Cladistics 10:315–319.
Fitch, W. M. 1971. Toward defining the course of evolution: minimum change for a specific tree topology. Syst. Zool. 20:406–416.
Flor, H. H. 1955. Host-parasite interaction in flax rust: its genetics and other implications. Phytopathology 45:680–685.
Friesen, N., R. M. Fritsch, and F. R. Blattner. 2004. Phylogeny and new intrageneric classification of Allium L. (Alliaceae) based on nuclear ribosomal DNA ITS sequences. (in press)
Futuyma, D. J. 1986. Evolutionary biology. Sinauer Associates, Sunderland, Mass.
Giraud, T., D. Fortini, C. Levis, P. Leroux, and Y. Brygoo. 1997. RFLP markers show genetic recombination in Botryotinia fuckeliana (Botrytis cinerea) and transposable elements reveal two sympatric species. Mol. Biol. Evol. 14:1177–1185.
Harrison, J. G. 1980. The production of toxins by Botrytis fabae in relation to growth of lesions on field bean leaves at different humidities. Ann. Appl. Biol. 95:63–72.
Heath, M. C. 1991. The role of gene-for-gene interactions in determination of host species diversity. Phytopathology 81:127–130.
Hennebert, G. L. 1973. Botrytis and Botrytis-like genera. Persoonia 7:183–204.
Hennebert, G. L., and J. W. Groves. 1963. Three new species of Botryotinia on Ranunculaceae. Can. J. Bot. 41:341–370.
Holst-Jensen, A., M. Vaage, and T. Schumacher. 1998. An approximation to the phylogeny of Sclerotinia and related genera. Nordic J. Bot. 18:705–719.
Huang, J., T. Hsieh, G. A. Chastagner, and T. Hsiang. 2001. Clonal and sexual propagation in Botrytis elliptica. Mycol. Res. 105:833–842.
Jarvis, W. R. 1977. Botryotinia and Botrytis species; taxonomy, physiology and pathogenicity. Monograph No. 15, Canadian Department of Agriculture, Ottawa.
———. 1980. Taxonomy. Pp. 1–18 in J. R. Coley-smith, K. Verhoeff, W. R. Jarvis, eds. The biology of Botrytis. Academic Press, London.
Kniskern, J., and M. D. Rausher. 2001. Two modes of host-enemy coevolution. Popul. Ecol. 43:3–14.
Kondrashov, A. S. 1997. Evolutionary genetics of life cycles. Annu. Rev. Ecol. Syst. 28:391–435.
Liu, Y., and B. D. Hall. 2004. Body plan evolution of ascomycetes, as inferred from an RNA polymerase II phylogeny. Proc. Natl. Acad. Sci. USA 101:4507–4512.
Liu, Y., S. Whelen, and B. D. Hall. 1999. Phylogenetic relationships among ascomycetes: evidence from an RNA polymerase II subunit. Mol. Biol. Evol. 16:1799–1808.
MacFarlane, H. H. 1968. Review of Applied Mycology. Pp. 820 in plant host-pathogen index to volumes 1–40 (1922–1968).
Maddison, W. P., and D. R. Maddison. 2000. MacClade. Sinauer Associates, Sunderland, Mass.
Mansfield, J. W. 1980. The biology of Botrytis. Pp 181–218 in J. R. Coley- Smith, K. Verhoeff, and W. R. Jarvis, eds. The biology of Botrytis. Academic Press, London.
McDade, L. A. 1992. Hybrids and phylogenetic systematics II. The impact of hybrids on cladistic analysis. Evolution 46:1329–1346.
Muoz G., P. Hinrichsen, Y. Brygoo, and T. Giraud. 2002. Genetic characterisation of Botrytis cinerea populations in Chile. Mycol. Res. 106:594–601.
Nielsen, K., and D. S. Yohalem. 2001. Origin of a polyploid Botrytis pathogen through interspecific hybridization between Botrytis aclada and B. byssoidea. Mycologia 93:1064–1071.
Nikoh, N., and T. Fukatsu. 2000. Interkingdom host jumping underground: phylogenetic analysis of entomoparasitic fungi of the genus Cordyceps. Mol. Biol. Evol. 17:629–638.
O'Donnell, K., H. C. Kistler, E. Cigelnik, and R. C. Ploetz. 1998. Multiple evolutionary origins of the fungus causing Panama disease of banana: concordant evidence from nuclear and mitochondrial gene genealogies. Proc. Natl. Acad. Sci. USA 95:2044–2049.
Prins, T. W., P. Tudzynski, A. von Tiedemann, B. Tudzynski, A. ten Have, M. E. Hansen, K. Tenberge, and J. A. L. van Kan. 2000. Infections strategies of Botrytis cinerea and related necrotrophic pathogens. Pp. 33–65 in J. W. Kronstad, ed. Fungal pathology. Kluwer Academic Publisher, Dordrecht, The Netherlands.
Rokas, A., B. L. Williams, N. King, and S. B. Carroll. 2003. Genome-scale approaches to resolving incongruence in molecular phylogenies. Nature 425:798–804.
Rolke, Y., S. Liu, T. Quidde, B. Williamson, A. Schouten, K. Weltring, V. Siewers, K. B. Tenberge, B. Tudzynski, and P. Tudzynski. 2004. Functional analysis of H2O2-generating systems in Botrytis cinerea: the major Cu-Zn-superoxide dismutase (BCSOD1) contributes to virulence on French bean, whereas a glucose oxidase (BCGOD1) is dispensable. Mol. Plant Pathol. 5:17–27.
Ronquist, F., and J. P. Heulsenbeck. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19:1572–1574.
Roy, B. A. 2001. Patterns of association between crucifers and their flower-mimic pathogens: Host jumps are more common than coevolution or cospeciation. Evolution 55:41–53.
Schardl, C. L., and K. D. Craven. 2003. Interspecific hybridization in plant-associated fungi and oomycetes: a review. Mol. Ecol. 12:2861–2873.
Sch?nbeck, F., and E. Schl?sser. 1976. Preformed substances as potential protectants. Pp. 653–678 in R. Heitefuss and P. H. Williams, eds. Physiological plant pathology. Springer-Verlag, Berlin.
Schouten, A., C. A. M. Wagemakers, F. Stefanato, R. M. van der Kaaij, and J. A. L van Kan. 2002. Resveratrol acts as a natural profungicide and induces self-intoxication by a specific laccase. Mol. Microbiol. 43:883–894.
Shepard, D. V., and D. Pitt. 1976. Purification of a phospholipase from Botrytis cinerea and its effect on plant tissues. Phytochemistry 15:1465–1470.
Shirane, N. M. Masuko, and Y. Hayashi. 1989. Light microscopy observation of nuclei and mitotic chromosomes of Botrytis species. Phytopathology 79:728–730.
Soltis, P. S., D. E. Soltis, and M. W. Chase. 1999. Angiosperm phylogeny inferred from multiple genes as a tool for comparative biology. Nature. 402:402–404.
Smith, T. L. 1989. Disparate evolution of yeasts and filamentous fungi indicated by phylogenetic analysis of glyceraldehydes-3-phosphate dehydrogenase genes. Proc. Natl. Acad. Sci. USA 86:7063–7066.
Swofford, D. L. 2002. PAUP*: phylogenetic analysis using parsimony (*and other methods), version 4.0b10. Sinauer Associates, Sunderland, Mass.
ten Have, A., K. B. Tenberge, J. A. E. Benen, P. Tudzynski, J. Visser, and J. A. L. van Kan. 2002. The contribution of cell wall degrading enzymes to pathogenesis of fungal plant pathogens. Pp. 341–358 in F. Kempken, ed. The Mycota, Vol. 11. Springer-Verlag, Berlin, Heidelberg.
Thomma, B. P. H. J. 2003. Alternaria spp.: from general saprophyte to specific parasite. Mol. Plant Pathol. 4:225–236.
Thompson, J. D., T. J. Gibson, F. Plewniak, F. Jeanmougin, and D. G. Higgins. 1997. The CLUSTALX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25:4876–488.
Thompson, J. N., and J. J. Burdon. 1992. Gene-for-gene coevolution between plants and parasites. Nature 360:121–125.
Turgeon, B. G. 2000. Evolution of host specific virulence in Cochliobolus heterostrophus. Pp. 93–126 in J. W. Kronstad, ed. Fungal pathology. Kluwer Academic Publisher, Dordrecht, The Netherlands.
van Baarlen, P., M. Staats, and J. A. L. van Kan. 2004. Induction of programmed cell death in lily by the fungal pathogen Botrytis elliptica. Mol. Plant Pathol. (in press).
Van der Vlugt-Bergmans, C. J. B., B. F. Brandwagt, J. W. van't Klooster, C. A. M. Wagemakers, and J. A. L. van Kan. 1993. Genetic variation and segregation of DNA polymorphisms in Botrytis cinerea. Mycol. Res. 97:1193–1200.
White, T. J., T. Bruns, S. Lee, and J. Taylor. 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. Pp. 315–322 in M. A. Innis, D. H. Gelfand, J. J. Sninsky, and T. J. White, eds. PCR protocols: a guide to methods and applications. Academic Press, San Diego.
Willetts, H. J. 1997. Morphology, development and evolution of stromata/sclerotia and macroconidia of the Sclerotiniaceae. Mycol. Res. 101:939–952.
Wubben, J. P., W. Mulder, A. ten Have, J. A. L. van Kan, and J. Visser. 1999. Cloning and partial characterization of endopolygalacturonase genes from Botrytis cinerea. Appl. Environ. Microbiol. 65:1596–1602.
Yohalem, D. S., K. Nielsen, and M. Nicolaisen. 2003. Taxonomic and nomenclatural clarification of the onion neck rotting Botrytis species. Mycotaxon 85:175–182.(Martijn Staats, Peter van)