1. In forest communities, the Janzen–Connell (J–C) model proposes that species diversity is maintained by noncompetitive distance- or density-dependent seedling mortality caused by host-specific natural enemies. Host specificity, however, has not been fully elucidated.
2. We conducted a cross-inoculation experiment to evaluate the host specificity of a pathogenic fungus, Colletotrichum anthrisci. The fungus was isolated from seedlings of four tree species (Prunus grayana, Fraxinus lanuginosa, Cornus controversa and Magnolia obovata), all of which were killed by damping-off disease beneath conspecific adults. Each fungal isolate was then inoculated into seedlings of P. grayana and F. lanuginosa. Molecular identification [internal transcribed spacer (ITS) sequences] also confirmed that all isolates (strains) showed 99–100% similarity with C. anthrisci, irrespective of their origin.
3. In both P. grayana and F. lanuginosa seedlings, all isolates of the pathogen caused damage, irrespective of origin, but the damage was more severe with the isolate from conspecifics rather than from any of the three heterospecifics.
4. In response to infection, callose papillae were deposited on the inner side of the leaf cell wall of seedlings in P. grayana; then, circular abscission layers formed between two layers of leaf cells surrounding the locus of infection. The central area of the infection was completely cut off from the rest of the leaf. In F. lanuginosa, infected leaves of seedlings were shed immediately after inoculation. This defensive behaviour, which may prevent further pathogen invasion, was more frequent in seedlings inoculated with isolates from conspecifics than from heterospecifics in both species.
5.Synthesis. Although the pathogenic fungus C. anthrisci is ubiquitous and attacks a wide range of host species, virulence was much stronger for strains derived from conspecifics rather than from heterospecifics, suggesting local adaptation and development of host specificity. If host specificity is common for several pathogens within a microbial community in a given area occupied by an adult, and the trait is also common for multiple tree species co-occurring within a forest community, the J–C model would be applicable to explain tree species diversity.
In forest communities, Janzen (1970) and Connell (1971) hypothesized that species diversity can be maintained by host-specific parasites (e.g. pathogens and herbivores) if they are more likely to damage and kill seeds and seedlings near the parent tree than far away because they respond to a high density of the juveniles. This high mortality might in turn liberate areas for colonization and recruitment by heterospecific tree species, thus promoting species coexistence [Janzen–Connell (J–C) hypothesis].
In inoculation experiments with microbes (e.g. Pythium, rhizosphere bacteria), Mills & Bever (1998) and Westover & Bever (2001) revealed that each microbe species caused a strong reduction in plant biomass, particularly in species from which the strains had been isolated in old-field perennial plants. This suggests that infection history (i.e. origin of the strain from a local host species) affects the virulence of multihost pathogens, generating host specificity. In forest communities, however, the extent to which infection history affects the virulence of multihost pathogens remains unknown in the context of local adaptation.
Several studies in forest communities have also explored whether high mortality of juveniles near conspecific adults can be attributed to the accumulation of host-specific pathogens. Soil biota associated with soil beneath conspecific adults had a more negative influence on seedlings than that associated with soil beneath heterospecifics (i.e. far from conspecifics; Packer & Clay 2003; Reinhart et al. 2005; Reinhart & Clay 2009). Although these findings suggest that the pathogens attack the seedlings in a host-specific manner, the studies bulked soil communities from multiple distant trees of different species into one composite sample and thus included considerable natural variation in soil pathogen communities (McCarthy-Neumann & Kobe 2010a,b). Packer & Clay (2000) found that in soil with Pythium (an oomycete) collected under Prunus serotina, all seedlings of P. serotina were killed, but mortality of two heterospecific tree seedlings was low. Even though the soils are sampled from under one parent, they contain multiple pathogens, each with some degree of host specificity; the extent of host specificity is different among individual pathogen species even within a genus. Therefore, overall effects of the pathogens would be different among areas occupied by different species of adult. To more precisely clarify the role of host specificity of pathogens in the context of the J–C mechanism, one must perform inoculation experiments for multiple host species using identified pathogens from different origins that were isolated from different host tree species.
Colletotrichum species (Coelomycetes) can live either as saprophytes on dead plants or as parasites or endophytes on plants in agricultural and forest communities (Prusky, Freeman & Dickman 2000; Agrios 2005; Osono & Mori 2005). In temperate forests of Japan, Colletotrichum spp. are pathogenic fungi that cause seedling death in several tree species (Sahashi, Kubono & Shoji 1994, 1995; Ichihara & Yamaji 2009). Colletotrichum spp. were isolated more frequently for seedlings killed by pathogens beneath conspecific compared with heterospecific adults in six of eight hardwood species, in all of which seedling mortality caused by disease was higher beneath conspecific compared with heterospecific adults (Yamazaki, Iwamoto & Seiwa 2009), suggesting greater abundance or stronger virulence beneath conspecifics. Colletotrichum spp. usually overwinter within latently infected leaves, and these leaves can be a source of primary inoculation in the following year (Sahashi, Kubono & Shoji 1995; Buchwaldt et al. 1996; Yoshida & Shirata 1999), suggesting lower mobility of conidia. These traits also suggest that the virulence of the strains of Colletotrichum differs depending on the origin of the strain (Thrall & Burdon 1997; Burdon & Thrall 2009), although Colletotrichum is ubiquitous in the forest community.
In this study, we evaluated the extent of host specificity (i.e. local adaptation) of the multihost pathogen Colletotrichum anthrisci in a cross-inoculation experiment. Strains of the fungus were isolated from the seedlings of four tree species that were killed by damping-off disease beneath conspecific adults. We also conducted molecular identification (ITS sequences) to evaluate similarities within C. anthrisci among strains of different origin. Each isolate was then inoculated to seedlings of two of the four tree species. We investigated whether virulence (degree of damage) was much stronger for strains from conspecific seedlings compared with heterospecific ones. We also investigated whether the frequency of structural defences is higher against strains that originate from heterospecific compared with conspecific seedlings. Finally, we discuss these results in the light of the observations on host specificity of a pathogen to provide insight into the potential impacts of a soil pathogen on the forest community structure in the context of the J–C hypothesis.
Materials and methods
The focal tree species in this study, Prunus grayana (Rosaceae), Fraxinus lanuginosa (Oleaceae), Cornus controversa (Cornaceae) and Magnolia obovata (Magnoliaceae), were late successional and shade tolerant (Kikuzawa 1983; Seiwa & Kikuzawa 1991; Seiwa et al. 2006). Shade tolerance strongly influences the susceptibility to pathogens, or conversely, the virulence of pathogens (O’Hanlon-Manners & Kotanen 2004; McCarthy-Neumann & Kobe 2008). As these species belong to different genera, we hereafter refer to them by their genus names. Prunus, Cornus and Magnolia are monoecious trees, whereas Fraxinus is a dioecious species. In all species, flowering occurs from May to June and fruits ripen from September to October. Seeds of the three monoecious species and of Fraxinus are dispersed by birds and wind, respectively, although most are disseminated beneath the canopy (Seiwa et al. 2008; M. Yamazaki and K. Seiwa unpubl. data). For all four species, seeds germinate in the following spring, although half of the seeds germinate 2 years later in Cornus. Dense seedling mats often formed in both Prunus and Cornus (Masaki & Nakashizuka 2002; Seiwa et al. 2008). In the four species, large fruit crops occur every 2–5 years.
A damping-off fungus, Colletotrichum anthrisci, was collected from a deciduous broadleaf forest in the reserve area (c. 168 ha) of the experimental forest of the Field Science Centre of Tohoku University, north-east Japan (38°48′ N, 140°44′ E, altitude 500–610 m). Mean monthly temperatures ranged from 1.0 °C (January) to 22.5 °C (August) in 2004. Mean annual temperature and rainfall were 11.0 °C and 1563 mm, respectively. Trees in the reserve area have been re-established after clear-cutting 60 years ago, and the area has been protected from human activity for at least 50 years as a forest preserve. The forest community in this area is diverse, with a total species number of 60 per 6 ha (Terabaru et al. 2004). These species are, in order of importance value (basal area, %) in the area, Quercus mongolica var. grosseserrata (30.2%), Fagus crenata (18.0%), Castanea crenata (15.3%), Aesculus turbinata (8.1%), Carpinus laxifolia (4.5%), Acer mono (3.2%), Quercus serrata (2.0%) and Acer palmatum var. matsumurae (1.7%). The relative basal areas (%) of the focal species are 1.7% (Magnolia), 1.0% (Cornus), 0.3% (Prunus) and 0.04% (Fraxinus). The three dominant species, which are distributed as patches of a mosaic, occupied 63.5% of the total basal area (Terabaru et al. 2004), whereas the other species were distributed randomly (K. Seiwa et al. unpubl. data).
To obtain the isolates of the fungus from dead seedlings of the four hardwood species (i.e. Prunus, Fraxinus, Cornus and Magnolia), seeds of each species were sown in four replicated quadrats (34 × 45 cm), which were randomly established within 0–3 m of the boles of the three replicate conspecific adults in 2003 (Yamazaki, Iwamoto & Seiwa 2009). In Fraxinus, the seeds were sown beneath female trees. In each species, the adults beneath which the seeds were sown were isolated from each other and were at least 25 m from the nearest conspecific and heterospecific adults of the focal species. In each species, the seeds were collected from more than three adults in the reserve. The details of the experiment (e.g. numbers of seeds sown, number of seedlings emerged, seedling mortality, mortality agents and genera of pathogens that caused seedling death) are provided by Yamazaki, Iwamoto & Seiwa (2009).
Isolation of the fungus was conducted from a lesion on the leaves of the seedlings that died from damping-off diseases, based on macroscopic (red pigment production) and microscopic characteristics (curved conidia formation) on cornmeal agar plates (Nissui Pharmaceutical, Tokyo, Japan). One or two of the dead seedlings were randomly selected beneath two of three conspecific adults from May to July 2004. Three strains of Colletotrichum species were obtained for each of the four focal tree species. For all strains, single spore isolates were established and incubated on cornmeal agar plates at 20 °C for molecular identification and inoculation experiments.
For the 12 Colletotrichum strains, we conducted PCR amplification and sequencing of the ribosomal DNA (rDNA) ITS, including 5.8S rDNA, ITS1 and ITS2 regions. Fungal DNA was extracted from the mycelium on cornmeal agar plates using the FastDNA kit (MP Biomedicals, Solon, OH, USA). Primers ITS5 and ITS4 (White et al. 1991) were used to amplify the ITS region. PCR amplification was performed using Takara EX Taq (Takara Bio Inc., Otsu, Japan) with an ABI 9700 thermal cycler (Applied Biosystems, Foster City, CA, USA) and PCR product was purified using the MagExtractor kit (Toyobo, Tokyo, Japan). Sequencing was performed using an ABI PRISM 3700 DNA autosequencer with BigDye Terminator chemistry (Applied Biosystems).
To investigate the phylogenetic position of our isolates, sequence data from nine Colletotrichum species selected on the basis of our BLAST results and recent taxonomic reports of the genus Colletotrichum (Damm et al. 2009; Hyde et al. 2009) were obtained from GenBank, in addition to two isolates sequenced in this study. Culture collection and GenBank accession numbers follow taxon names on the tree (Fig. S1 in Supporting information). Sequence alignments and phylogenetic analysis were performed using mega 4.1 (Tamura et al. 2007). A phylogenetic tree was generated using a neighbour-joining method with the Kimura two-parameter model. Support for individual nodes was tested using bootstrap analysis based on 1000 bootstrap replicates.
To obtain the seedlings of Prunus and Fraxinus for inoculation, seeds were collected from more than 10 trees in the experimental forest in 2004. Seeds were floated in water to eliminate nonviable ones, surface-sterilized in a 50% solution of commercial bleach (6% sodium hypochlorite) for 15 min and then rinsed in tap water for additional 15 min (Packer & Clay 2003). After stratification at 5 °C for 5 months, seeds were sown in planters filled with sterilized soil. In May 2005, seedlings were transplanted individually in each pot (6 cm diameter, 10 cm depth) filled with forest soil. The soils were autoclaved (121 °C, 110 kPa) for 1 h. To avoid contamination with other strains and to maintain high humidity conditions in the inoculation experiment, 15 vinyl-covered cells (40 cm wide, 80 cm deep, 80 cm high) were established for inoculation (12 cells) and for control (three cells) in a glasshouse. In each of the 12 cells, 36 pots (seedlings of Prunus and Fraxinus were planted in each of 12 and 24 pots, respectively) were set on a planter (30 cm wide, 60 cm deep, 3 cm high). A single strain of C. anthrisci, isolated from each of the four origins (one conspecific and three heterospecific seedlings), was inoculated to each of the two target species, Prunus and Fraxinus. In each of three control cells, 12 pots (seedlings of Prunus and Fraxinus were planted in each of eight and four pots, respectively) were set on the same planter as the inoculation. In total, 12 strains (three strains from each of four origins) and three controls (no inoculation) were randomly set. The total number of seedlings investigated was 156 for Prunus [(12 seedlings × 4 origins + 4 seedling in control) × 3 replicates] and 312 for Fraxinus [(24 seedlings × 4 origins + 8 seedling in control) × 3 replicates]. In this study, seeds of two host species (i.e. Cornus, Magnolia) were not available for the experiment because of the lack of seeds in a nonmast year.
Small wounds were made at the centre of a leaf by touching the surface of the leaf with a bundle of five sterile insect pins; this resulted in 25 pin pricks in an area of 0.3–0.5 cm2 (Gilbert & Webb 2007). For each of the wounded seedlings, a mycelial disc (5 mm2) or a nonmycelial (control) disc was transferred to the centre of the wounded leaf. After inoculation, the pots were randomly placed in a plastic case.
Measurements of Virulence and Defences
In this study, we defined virulence as the degree of damage caused to a host by pathogen infection, and this was assumed to be negatively correlated with host fitness (Sacristan & Garcia-Arenal 2008). Damage to individual seedlings from pathogen infection was observed at 4- or 5-day intervals in the 20 days following inoculation. The proportion of the area damaged per seedling was assessed visually at 1% intervals. In all seedlings of both Prunus and Fraxinus, we confirmed whether the symptoms observed were because of the inoculation of C. anthrisci by observing the morphological characteristics of the fungus (e.g. acervuli, cetae and conidia) under a microscope. The observations were made whenever seedlings died or leaves were shed during the study period (20 days). At the end of this period, the remaining seedlings were also assessed. Furthermore, following the inoculation test, we conducted a re-isolation of C. anthrisci from a total of 34 Prunus and 43 Fraxinus seedlings. For each of these two species, 10, 12, 13 and 8 seedlings and 9, 8, 9 and 8 seedlings were inoculated with isolates originating from dead seedlings of Prunus, Fraxinus, Cornus and Magnolia, respectively.
Plants usually respond to pathogen attack by forming one or more types of defence structure (e.g. cytoplasmic, cell wall, historical and hypersensitive defences) that is more or less successful in defending the plant from further pathogen invasion (Simms & Vision 1995; Agrios 2005), even after pathogens have penetrated the defence structure (e.g. waxes, cuticles, thick and tough outer walls of epidermal cells). In this study, seedlings of Prunus and Fraxinus showed cell wall defences and early defoliation of infected leaves, respectively, both of which may defend the plant from further pathogen invasion. We calculated the proportion of seedlings showing defensive behaviour to the total number of seedlings inoculated for each species.
To test for differences in virulence among strains of different origin, we conducted a repeated-measures analysis of variance (rm-anova), testing the effect of observation dates (time period) and strains of different origins (between-subject factor) on each of two variables indicating pathogen damage (virulence): proportion of pathogen damage (PP) and index of pathogen damage (IP). PP showed the proportion of the area damaged per seedling. In the analysis of PP, the data were arcsine-transformed to meet the assumption of anova, and the isolates were nested within origin. The IP was obtained according to the formula,
where k is the category of damage (six categories based on the proportion of area damaged in each individual: 0, no damage; 1, 1–25%; 2, 25–50%; 3, 50–75%; 4, 75–99%; 5, 100% of total leaf area damaged or dead), Nk is the number of seedlings in each category of damage and N is the total number of seedlings per species (Kremer & Unterstenhofer 1967; Garcıa-Guzman & Dirzo 2001).
To compare the defence ability of host seedlings against the strains of the pathogen originating from different species, the proportion of seedlings showing structural defences was compared among strains using a one-way anova. These analyses were conducted for Prunus and Fraxinus separately. Tukey–Kramer HSD tests were used to compare differences among means when the anova gave significant values. The computer software jmp version 4.05 (SAS Institute Inc., Cary, NC, USA) was used for the analysis.
Internal transcribed spacer sequences obtained from 12 strains (three strains from four origins) were classified into two genotypes (Fig. S1). In the sequence alignment, one nucleotide indel was observed between two genotypes. Sequences of WCK 0887 (representative of genotype A) and WCK 0894 (representative of genotype B) were deposited in DDBJ (Accession Nos. AB476424-476425). BLAST searches against two representative sequences revealed that ITS sequences of WCK 0887 and WCK 0894 showed 100% and 99% similarity, respectively, with Colletotrichum anthrisci CBS 125334 (Accession No. GU227845). While C. anthrisci Damm, P. F. Cannon and Crous were recently described by Damm et al. (2009) based on a fungus from dead stems of Anthriscus sylvestris, the molecular analyses suggest that our Colletotrichum species is C. anthrisci. The phylogenetic position of our isolates is also presented (Fig. S1).
Infection and Re-isolation
After inoculation, morphological characteristics (i.e. acervuli, cetae and conidia) specific to C. anthrisci were observed in 30.6%, 16.7%, 27.8% and 11.1% of Prunus seedlings and 9.7%, 36.1%, 29.2% and 12.5% of Fraxinus seedlings, for which the isolates inoculated originated from Prunus, Fraxinus, Cornus and Magnolia, respectively. Following the inoculation experiments, C. anthrisci was re-isolated for 40%, 100%, 100% and 25% of the Fraxinus seedlings and 50%, 44%, 56% and 75% of the Prunus seedlings, for which inoculated isolates originated from Prunus, Fraxinus, Cornus and Magnolia, respectively. In controls, the wound caused necrosis, from which no C. anthrisci was isolated for Prunus and Fraxinus seedlings. These traits strongly suggest that the symptoms observed after inoculation were caused by infection with the isolates of C. anthrisci.
In the seedlings of both Prunus and Fraxinus, the PP damage increased with time after inoculation for all strains, irrespective of the origin (rm-anovas, time period; Prunus: F3,118 = 162, P <0.0001; Fraxinus: F4,273 = 26.7, P <0.0001; Fig. 1). PP was usually higher for seedlings inoculated with strains from conspecifics rather than from heterospecifics (between-subject factor; Prunus: F3,120 = 52.0, P <0.0001; Fraxinus: F3,276 = 8.57, P <0.0001; Fig. 1). Interaction between time and origin was observed only in Prunus (F9,287 = 16.1, P <0.0001), in which large differences were only observed during the later days after inoculation (Fig. 1).
The IP showed similar patterns to PP (Fig. S2). In the seedlings of both Prunus and Fraxinus, IP increased with time after inoculation for all strains, irrespective of origin (rm-anovas, time period; Prunus: F4,5 = 725, P <0.0001; Fraxinus: F5,4 = 36.1, P <0.002; Fig. 1). IP was usually higher for seedlings inoculated with strains from conspecifics rather than from heterospecifics (between-subject factor; Prunus: F3,8 = 54.0, P <0.028; Fraxinus: F3,8 = 5.20, P = 0.028; Fig. S2). Interaction between time and origin was observed only in Prunus (F12,14 = 6.80, P = 0.0007), showing large differences only during the later days after inoculation (Fig. 1). In the seedlings of both Prunus and Fraxinus, both PP and IP were constantly lowest in the control (Fig. 1), in which both PP and IP included the discolouration of leaves caused by the pin pricks, but not by any fungal attack.
In Prunus seedlings, callose papillae were deposited on the inner side of leaf cell walls after infection with C. anthrisci; then, circular abscission layers formed between the two layers of leaf cells surrounding the locus of infection. Thus, the central area of the infection was completely cut off from the rest of the leaf (Fig. S3). The proportion of individual seedlings showing cell wall defences was lower for seedlings inoculated with strains from conspecifics than those inoculated with strains from heterospecifics, although this was marginally nonsignificant (Tukey–Kramer HSD tests, after one-way anova, F3,2 = 3.09, P = 0.090; Fig. 2a). In Fraxinus, the seedlings shed the infected leaves soon after infection, whereas noninfected leaves were retained. The proportion of individuals showing this defence (early leaf shedding) was lower for seedlings inoculated with strains from conspecifics than those inoculated with strains from heterospecifics (Tukey–Kramer HSD tests, after one-way anova, F3,2 = 4.41, P = 0.041; Fig. 2b). In contrast, no control individuals showed structural defences in either species.
The specialization of multihost pathogens has been demonstrated in herbaceous plant communities and agricultural fields (Mills & Bever 1998; Westover & Bever 2001; Thrall et al. 2005; Malpica et al. 2006; Spitzer 2006; Sicard et al. 2007; Kniskern, Barrett & Bergelson 2011). In forest communities, a cross-inoculation experiment was the first study to clarify the evolutional consequences of microbial pathogenesis and host resistance of a multihost pathogen. The virulence of the pathogenic fungus C. anthrisci was greater in strains originating from conspecific compared with heterospecific hosts, although the pathogen is ubiquitous and attacks at least four hardwood species in the temperate forest of northern Japan. Colletotrichum spp. (e.g. Colletotrichum dematium and Colletotrichum truncatum) do not inhabit soil, but usually live in and overwinter in debris or the litter layer (Sahashi, Kubono & Shoji 1995; Yoshida & Shirata 1999); infectivity in buried debris remains high for at least 4 years (Buchwaldt et al. 1996). Thus, the fungal inoculum could have been effectively amplified beneath adult trees where more conspecific litter including dead seedlings accumulated, even though the nonmast year provided few seeds. Furthermore, in forest communities, the generation time of most pathogens will be much faster than that of the host tree species, and pathogens may pass through many generations under strong selection for increased virulence before resistant hosts reach reproductive maturity. The scenario is well supported beneath other large and more fecund trees of P. serotina, which experience greater mortality of conspecific seedlings in North America (Packer & Clay 2004) and also in Prunus grayana in northern Japan (Seiwa et al. 2008). The process of rapidly evolving pathogens adapting to the most common host encountered beneath adults leads to the local adaptation of the pathogens, enhancing their performance in conspecific host populations (Mills & Bever 1998; Westover & Bever 2001; Packer & Clay 2004; Sicard et al. 2007; Seiwa et al. 2008). In contrast, a generalist pathogen may be maladapted to particular species (i.e. heterospecific seedlings), probably because hosts are rarely encountered (Spitzer 2006; Kniskern, Barrett & Bergelson 2011). Moreover, recent theoretical and experimental studies suggest that when gene flow of parasites is greater than that of the host, local adaptation is more pronounced owing to an inherent increase in the genetic variation of parasite populations and thus increased efficacy of selection (Gandon et al. 1996; Greischar & Koskella 2007). Colletotrichum spp. can be dispersed among debris not only by wind-splashed rain but also by wind to distances of at least 240 m in commercial crop fields (Buchwaldt et al. 1996). If C. anthrisci also has higher migration ability than the host tree species, local adaptation of the pathogenic fungi would be enhanced, resulting in greater host specificity. In forest communities, however, the virulence should evolve towards optimum levels to maximize fitness because so many alternative hosts are available. Further studies comparing the dispersal ability of genes between fungi and host trees (pollen, seeds) will be useful to evaluate the evolution of host specificity.
In infected seedlings, cell wall defences and early shedding of infected leaves were observed for Prunus and Fraxinus, respectively. These structural defences induced by C. anthrisci would effectively protect other leaf tissues or organs from being invaded by the pathogen (Agrios 2005). In turn, these host responses are likely to reduce fungal reproduction by decreasing the proportion of leaf area occupied by lesions. These traits suggest that the two hardwood species have evolved adaptive responses to lower the detrimental impact of pathogens on plant fitness (Simms & Vision 1995). Our study also revealed that the defences were less frequent in seedlings inoculated with strains from conspecifics compared with those from heterospecifics. The lower defence capacity could also result from a build-up of detrimental pathogens over time through negative biotic feedback in the local area beneath conspecific adults.
In this experiment, however, the original trap plants used to isolate different strains were the same as the resident host in each of the two cases (e.g. Prunus seedlings were used to trap fungal isolates from beneath Prunus adults). Therefore, the design of the experiment suggests to some extent that selection is occurring for isolates better able to infect conspecifics. To evaluate the local adaptation in more detail, multiple species of trap plants (i.e. four species in this study) should be used for each focal adult host species.
Host Specificity and Species Diversity
The laboratory experiment clearly revealed that the pathogenic fungus C. anthrisci attacked hardwood seedlings in a host-specific manner. This strongly suggests that the development of negative feedback beneath the parent canopy gave less susceptible seedlings of locally rare species a competitive advantage over the parental much more common seedlings, promoting the replacement of conspecific by heterospecific seedlings. Therefore, further experiments examining the effects of the different isolate in mediating competition among competing tree species are required.
In the forest where we collected the fungus of C. anthrisci, we observed 22 genera of pathogenic fungi (e.g. Phoma, Fusarium, Cylindrocarpon, Cladosporium, Alternaria and Clyndrocladium), most of which attacked hardwood seedlings of many species (Yamazaki, Iwamoto & Seiwa 2009). In turn, individual seedlings were simultaneously infected by multiple pathogens. If each of the fungal pathogens has host-specific virulence like C. anthrisci, synergistic effects of the multiple pathogens may enhance the damage to conspecific seedlings and consequently local replacement in the forest community. However, recent studies showed that the degree of susceptibility to each pathogen varied among host species (Augspurger & Wilkinson 2007) and that the high host specificity of one pathogen alone was not enough to favour the establishment of heterospecifics under the tree canopy (Seiwa et al. 2008). Thus, to reveal the role of pathogens (i.e. host specificity and virulence) in maintaining local tree diversity, further cross-inoculation experiments are needed for each of the substantial pathogens.
Furthermore, recent studies showed that the interaction between multiple pathogens often has opposite effects; pathogens activate systemic resistance, which offers protection against a wide variety of microbes (Van Wees, Van der Ent & Pieterse 2008; Barrett et al. 2009). In a natural forest, most individual seedlings could be infected not only by pathogenic micro-organisms (e.g. soilborne pathogens and foliar diseases) but also by symbiotic micro-organisms (e.g. bacteria, arbuscular mycorrhiza and ectomycorriza) in a spatially and temporally variable manner (Hood, Swaine & Mason 2004; Gallery, Dalling & Arnold 2007; Gallery, Moore & Dalling 2010). Further study is needed to clarify the complexity of synergetic effects regarding the multiple micro-organisms on seedling performance beyond the context of the J–C hypothesis.
In conclusion, our cross-inoculation experiments clearly revealed that the virulence of a pathogen (C. anthrisci), which was ubiquitous and attacked a wide range of host species, was stronger in strains derived from conspecifics compared with those from several heterospecifics. The traits strongly suggest an important role of local adaptation in developing host specificity of a pathogen that is assumed to be a host generalist. If host specificity is common for a variety of soil pathogens within a microbial community in a given area occupied by an adult, the J–C model would be applicable in explaining the species diversity of the forest community.
We thank Yu, Ichihara and members of the Laboratory of Forest Ecology, Tohoku University, for help with the experiments. This research was funded by the Ministry of Education, Culture, Sports, Science and Technology, Japan (No. 20380084 to KS).