Molecular taxonomy of the Alternaria and Ulocladium ... - CBS - KNAW

270 G. S. de Hoog &R.Horréfound a substantial distance from all these groups. InFig. 1, two main clusters of catenate Alternariaspecies are found, centered around A. alternata andA. infectoria, respectively. The noncatenate plantpathogens constitute two further clusters, while agroup containing nearly all Ulocladium strains isparaphyletic to one of the plant pathogenic clustersand to A. chlamydospora. Strains of Ulocladium as wellas A. chlamydospora-like strains are morphologicallycharacterized by more or less ellipsoidal rather thanthe obclavate conidia observed in the remainder ofAlternaria. It should be noted that the Alternariaspecies with ellipsoidal conidia are heterogeneous: astrain from Carduus identified as A. mouchaccae, CBS453.74, is probably an undescribed species, whilethe two strains of A. chlamydospora studied are clearlydifferent from each other.Alternaria alternata is the oldest and most commonspecies of the genus; it is one of the short-conidialsaprotrophs. The species was redefined by Simmons[20, 22] on the basis of careful morphologicalobservations. In the absence of usable type material,he indicated strains CBS 916.96 and CBS 603.78 asrepresentative for the species (CBS List of Cultures;E.G. Simmons, personal communication). Simmons’concept of (morphologically) representativeisolates is convenient where no living or interpretablematerial is available (dried material ispresent at L: Leiden, the Netherlands). We favorthe application of a more formal approach, similarto that used by Gräser et al. [31] in the dermatophytes.Therefore the status of epitype is proposedfor CBS 916.96.In a series of publications [15, 20, 22, 32–34],E. G. Simmons maintained a narrow speciesconcept for the short-chained taxa around A.alternata, based on the three-dimensional branchingpatterns of conidial chains. Yu [35] noted that it isdifficult for nonexperts to detect morphologicaldifferences between the units; this is certainly truefor strains from humans which are often degenerateor even sterile. Most of the entities recognized bySimmons had different Random amplified polymorphicDNA (RAPD) profiles [36] and had specifichost plants [36]. RAPD-typing found a large set ofstrains which originated from closely similarsubstrates but nevertheless showed pronouncedinfraspecific variability. Similar variability wasfound by Morris et al. [37] with strains from tomatoin California: only 34 out of 137 RAPD markerswithin this ecologically unified set of strains provedto be monomorphic for all isolates. It is thereforelikely that RAPD is poorly suited for diagnosis ofnonhost-specific taxa world-wide. On the otherhand, some highly specialized species do exist thatshow remarkably consistent RAPD patterns even inwidely geographically distributed populations [38].The ITS tree is compared with ecological data inFig. 2. This tree is strongly biased towards strainsfrom economically important host plants. Forexample, Daucus carota is over-represented, beingthe host plant of six species treated (Table 1).Nevertheless a distinction can be made betweenspecies having (nearly) the same ITS sequence andcomprising isolates originating from the same hostplant, and groups ⁄ taxa with an equally low degreeof ITS diversity but showing no host specificity. Thelatter frequently contain strains from dead substrates,such as soil, wood, straw and plastic. This isthe case in Alternaria alternata (s.l.) and A. infectoria, aswell as in the saprotrophic genus Ulocladium. Themixed origin of strains clustering in these Alternariaspecies suggests that as far as saprotrophs areconcerned, their colonization of living host plants isrelatively coincidental.Occurrence on a particular host plant does notnecessarily identify the species, since several taxamay occur on the same host. This was particularlydemonstrated by Simmons [39], who found manydistinct taxa parasitizing Citrus. Most of the hostspecifictaxa may differ in toxin production [33],the toxins playing a role as pathogenicity factors[40]. Among the host-specific Alternaria specieswith ITS sequences almost identical to that ofA. alternata (Table 3) are A. longipes (on tobacco),A. lini Dey (on flax), A. mali Roberts (on apple) andA. arborescens Simmons (on tomato). The last namewas introduced by Simmons [39] for the formerA. alternata f.sp. lycopersici Gronan et al. Serdani et al.[41] listed A. arborescens as a main cause of core rotof apples in South Africa, along with A. tenuissima;the two species appeared to be very close in allcharacters analyzed. Evolutionary adaptation andspeciation on particular host plants within thebasically saprotrophic A. alternata s.l. is possible inprinciple. Indications for host specialization werefound in differential host resistance to a singlepathovar [42] and in low intraspecific variability ofanonymous markers of strains among a single host[43]. Andersen et al. [44] found A. longipes todeviate consistently from A. alternata in its metabolitespectrum. Thus the entities recognized byE. G. Simmons and R. G. Roberts are probablycorrect, but to conceive these entities as separatespecies is debatable. The groups of Nishimura[45, 46] and Adachi [47] referred to similar unitsas pathotypes within a single umbrella-species,A. alternata. This view was supported by Kuninagaand Yokosawa [48] who found high DNAreassociation values between A. alternata and relatedspecies. Our ITS sequencing data indicate thatseveral host-specific plant pathogens belong to theA. alternata species aggregate. Population geneticstudies are needed to demonstrate whether fullmycoses 45, 259–276 (2002)