Genetic Evidence for Colletotrichum gloeosporioides Transmission Between the Invasive Plant Ageratina adenophora and Co-occurring Neighbor Plants

To understand the disease-mediated invasion of exotic plants and the potential risk of disease transmission in local ecosystems, it is necessary to characterize population genetic structure and spatio-temporal dynamics of fungal community associated with both invasive and co-occurring plants. In this study, multiple genes were used to characterize the genetic diversity of 165 strains of Colletotrichum gloeosporioides species complex (CGSC) isolated from healthy leaves and symptomatic leaves of invasive plant Ageratina adenophora, as well as symptomatic leaves of its neighbor plants from eleven geographic sites in China. The data showed that these CGSC strains had a high genetic diversity in each geographic site (all Hd > 0.67 and Pi > 0.01). Haplotype diversity and nucleotide diversity varied greatly in individual gene locus: gs had the highest haplotype diversity (Hd = 0.8972), gapdh had the highest nucleotide diversity (Pi = 0.0705), and ITS had the lowest nucleotide diversity (Pi = 0.0074). Haplotypes were not clustered by geographic site, invasive age, or isolation source. AMOVA revealed that the genetic variation was mainly from within-populations, regardless of geographic or isolation origin. Both AMOVA and neutrality tests indicated these CGSC strains occurred gene exchange among geographic populations but did not experience population expansion along with A. adenophora invasion progress. Our data indicated that A. adenophora primarily accumulated these CGSC fungi in the introduced range, suggesting a high frequency of CGSC transmission between A. adenophora and co-occurring neighbor plants. This study is valuable for understanding the disease-mediated plant invasion and the potential risk of disease transmission driven by exotic plants in local ecosystems.


Introduction
Invasive plants may initially benefit from enemy (such as pathogens) release from native range and hence obtain competitive advantage over local species in a new domain [1,2]. However, some invasive plants can accumulate pathogens over the longer-term invasion progress according to the accumulation of local pathogens hypothesis [3,4]. Several reports recently indicated that richness and diversity of pathogens increased with the increase in diversity and range of invaded habitats [5,6], and the delayed invasion time [7].
Pathogen accumulation may be a potentially important factor for shaping the dynamics of invaded plant communities through two ecological consequences [4,8]. On the one hand, pathogen accumulation has direct adverse effect on the invader, known as the Pathogen Accumulation and Invasive Decline (PAID) hypothesis. For example, Hilker et al. performed a model analysis and found that pathogens can slow down, stop, and even reverse the host invasion [9]. Stricker et al. indicated that the accumulated fungal pathogens suppressed the invasive plant Microstegium expansion [10]. Pile Knapp et al. inoculated a native pathogen Verticillium nonalfalfae to effectively control the invasive plant Ailanthus ZhaoYing Zeng and ZhiPing Yang contributed equally to this work and share first authorship.
altissima [11]. On the other hand, pathogen accumulation can exacerbate the invasion, because pathogens may cause the increasing diseases on susceptible co-occurring species and indirectly reduce the competitive inhibition of these neighbors to invasive species. Such dynamics are termed as "spillover" when the pathogens are introduced with the invader and as "spillback" when an invasive species host native pathogens [4]. For example, invasive cheat grass (Bromus tectorum) can accumulate pathogen Pyrenophora semeniperda from the native seed and cause higher death rate of native seed in invaded range (spillover) [12]; Spartina alterniflora serves as host and repository of non-native pathogen Fusarium palustre to reduce the advantage of native plants Phragmites australis (spillback) [13].
Regardless of spillover or spillback, it is expected a common fungal transmission driven by invasive plants. Understanding this process depends on the comparison of genetic diversity of fungal community associated with invasive and neighbor plant. Till now, only limited experiment has been performed to quantify the genetic diversity of fungal community associated with invader. For example, an investigation of endophytic fungi of Centaurea stoebe in both native (Europe) and invaded ranges (North American) showed that some fungal endophytes appeared to have been introduced with C. stoebe, but the invader also acquired new endophytes after introduction [14]. In contrast, low genotypic diversity of the pathogen Teratosphaeria destructans on the nonnative Eucalyptus's leaves were found in all investigated populations from China, Indonesia, South Africa, Thailand and Vietnam, indicating that T. destructans is introduced to these regions with exotic host [15].
Ageratina adenophora (Sprengel) R.M.King & H.Robinson (Asteraceae), known as Crofton weed or Mexican Devil weed, has invaded over 30 countries and areas, including South Africa, Australia, New Zealand, Hawaii, India and China [16,17]. The species is one of the most invasive weeds in China. Following the outbreak in Yunnan in the 1960s, A. adenophora has been spreading north-and eastward through Guizhou, Sichuan, and Guangxi provinces in Southwest China at approximately 20 km/year [18]. Our previous research found that A. adenophora contained diverse fungal endophytes; in particular, Collectotrichum sp. are 30% of total isolated strains [19].
Colletotrichum species are regarded as the most commonly occurring pathogens and foliar endophytes in terrestrial plant [20,21]. Current understanding of Colletotrichum species is largely restricted to populations infecting crop [22] or ornamental plants [23,24]. Recently, it has concerned that non-agricultural environments are reservoirs of plant pathogens with broad host ranges and may impact the disease epidemiology and evolutionary dynamics of crop pathogens [25,26]. For example, non-cultivated hosts adjacent to strawberry fields can serve as potential inoculum sources for Colletotrichum crown rot of strawberry [27]. Interestingly, our research further found that numerically dominant strains belonging to Colletotrichum gloeosporioides species complex (CGSC) were isolated from healthy and symptomatic leaves of A. adenophora, as well as symptomatic leaves of native plants in invasive region [28,29]. If the CGSC shared by A. adenophora and co-occurring plants showed high genetic similarity, it is expected a high potential of phytopathogenic interactions and disease epidemics driven by invasive plant. Thus, it is necessary to evaluate the genetic diversity of these CGSC fungi by using multiple gene analysis.
In this study, we sequenced multiple gene loci, including internal transcribed spacer (ITS), calmodulin (cal), glutamine synthetase (gs), glyceraldehyde-3-phosphate dehydrogenase (gapdh), chitin synthetase (chs-1), and β-tubulin 2 (tub2), to characterize the genetic diversity of CGSC strains associated with invasive plant A. adenophora and local co-occurring neighbor plants. We aim to (1) quantify the haplotype and nucleotide diversity of CGSC isolated from A. adenophora from eleven invasive regions divided into four invasion ages; (2) identify potential associations of the genotypic variation among different isolation sources (invasive plant vs neighbor plants, healthy leaves vs symptomatic leaves); and (3) assess if these CGSC transmit or expand with the invasion of A. adenophora.

Samples Collection
From August 2012 to September 2017, we collected healthy mature leaves and symptomatic leaves of A. adenophora, as well as symptomatic leaves of co-occurring neighbor plants, including Alnus nepalensis, Rosa xanthine, Camellia japonica, Fallopia multiflora, Dioscorea hemsleyi, Ageratina heterophyllum, Gonostegia hirta, Dioscorea hispida, Nicotiana tabacum, Mangifera indica, Spermacoce alata, and Psidium guajava. These samples are from eleven A. adenophora invasion locations in Yunnan Province and cover four invasion ages based on the previous research [18] (see Table 1). Briefly, both the healthy mature leaves and any symptomatic leaves collected from A. adenophora population were selected stochastically in a location. Each pair of healthy mature leaves was the fourth pair leaves coming down from top in an individual, because of highest fungal diversity in the fourth pair leaves of A. adenophora [30]. Symptomatic leaves found in co-occurring neighbor plants with A. adenophora individuals were also collected. These collected leaves were placed respectively in sterile polypropylene bags and brought back to the laboratory by boxes with ice packs.

Fungal Isolation
Isolation of pure cultures followed methods described in previous study of Chen el at. [28]. Each sampled leaf was first washed with tap-water. Healthy leaves were cut into 2 cm × 2 cm square, symptomatic leaves were cut along the infected area edge with the sterile knife. All the pieces were respectively surface-sterilized with 0.5% (active ingredient) sodium hypochlorite solution for 30 s and then 75% alcohol for 2 min following by three times washes with sterilized distilled water. They were further subdivided into 6 mm 2 sections after drying sections surface with sterile filter paper. Six subdivided sections were picked out randomly and then placed in PDA medium in a Petri-dish and incubated at ambient temperature (20-25℃) for 6-8 days or until mycelia growing from the leaf fragments were observed. Hyphal tip cultures were then transferred onto new PDA plates and incubated until pure colony appeared.
The obtained sequences were edited using EDITSEQ and SEQMAN software in the DNASTAR package (DnaStar Inc., Madison, WI, USA). ITS consensus sequences were clustered into OTUs based on 100% sequence similarities using DISTSEQS in MOTHUR v.1.34.4 [38]. A local blast search was used to classify the OTUs phylogenetically using CLASSIFY.SEQS in MOTHUR and the UNITE database of reference fungal ITS sequences [38]. Isolations belonging to the top two dominant OTU were further amplified the other five loci (tub2, gapdh, chs-1, cal, and gs) and then used BLASTN analyses at each individual locus against GenBank database. CGSC may be simpler to consider as a species complex rather than using intraspecific designations for practical reasons [39,40]. Moreover, because genetic diversity in a species is generally limited, researches on population genetic diversity have been conducted on a species complex, such as CGSC [23,41,42]. In this study, therefore, all DNA sequences that were identified as the CGSC were included in subsequent genetic analyses.

Population Genetic Analysis
Multiple sequence alignment was performed with ClustalW in BioEdit [43]. Aligned sequences were concatenated by BioEdit [44]. DnaSP6 was used for calculating the number of haplotypes, haplotype diversity (Hd), and nucleotide diversity (Pi) of each population for concatenated sequences (ITS-gapdh-chs-1-gs-cal-tub2) and each locus [45]. Minimum spanning haplotype network was constructed to visualize the genetic relationship between haplotype and geographic populations, invasion ages as well as isolations sources in PopART version 1.7 [46]. Analyses of molecular variance (AMOVA) within each population and among different populations and matrices of pairwise genetic distance (Φ PT ) among populations as well as isolation sources were calculated by using GenAlEx6.51. Mantel test was used to determine the potential correlation between genetic and geographical distances among populations in GenAlEx6.51 [47].

Genetic Diversity
In total, 165 isolations belong to Colletotrichum gloeosporioides species complex (CGSC) were obtained from eleven geographical sites. All 990 sequences of six loci including ITS, tub2, gapdh, chs-1, cal, and gs, had been deposited in GenBank at www. ncbi. nlm. nih. gov (accession numbers see Table S2). Eleven geographic populations varied in haplotype diversity (Hd: 0.7 ~ 1); five populations (i.e., SL, ZT, GM, ES, KM) shared the highest haplotype diversity (Hd = 1), whereas NJ had the lowest (Hd = 0.7). Nucleotide diversity of all eleven geographic populations was more than 0.01 ( Table 1). The six sequenced gene fragments of the 165 strains varied in number and percentages of polymorphic nucleotide sites (see Table 2). The locus gapdh had the highest percentages of polymorphic nucleotide sites (35.56%), followed by gs (20.49%), and the lowest was ITS (5.08%). Nucleotide diversity (Pi) in 5 out of 6 loci (except ITS) was more than 0.01, and the highest was gapdh (Pi = 0.0705), followed by gs (Pi = 0.0341), and the lowest was ITS (Pi = 0.0074). In six loci, gs locus had the highest haplotype diversity (Hd = 0.8972), with 48 haplotypes, followed by ITS (Hd = 0.8264, 40 haplotypes), and the lowest was tub2 (Hd = 0.6702, 17 haplotypes). The 165 concatenated sequences by six loci (ITS-gapdh-chs-1-gs-cal-tub2) totaling 2976 bp were divided into 105 haplotypes, with a haplotype diversity of 0.9710; in total, these concatenated sequences showed a medium percentages of polymorphic nucleotide sites (15.29%) and nucleotide diversity (Pi = 0.0294) comparing with such six single loci.

Haplotype Network Analysis
Haplotype network analysis indicated that 165 strains were not clustered according to geographic site, isolation source or invasion age whenever based on six individual loci or the concatenated sequences by six loci (Fig. 1 & S1 & S2 & S3). For individual locus, only several haplotypes from the Ah isolation source were clustered together based on ITS locus (Fig. S2 C). Five loci (tub2, gapdh, chs-1, cal, and ITS) had two dominant haplotypes with 40 ~ 60 sequences while the gs locus have only one (H_8, containing 45 sequences) (Fig. S2). For the concatenated sequences, the major haplotype is H_10 (25 sequences, detected in 9 of 11 geographic sites, and in all three isolation sources), followed by H_5 (10 sequences, detected in 6 of 11 geographic sites, and in all three isolation sources) (Fig. 1). Sixteen haplotypes with more than one sequence stochastically distributed in different geographic sites, isolation sources and invasion ages (Fig. 1C, D & S1 B). Other 99 haplotypes had only one sequence (Fig. 1).

Population Structure
Analysis of molecular variance (AMOVA) revealed that 92.5% of the total genetic variation was from infra-individual geographical populations (p = 0.011), with a low (7.5%) but statistically significant genetic differentiation among the geographical populations (p = 0.011) ( Table 3). The total genetic variation within individual isolation source populations was 99.5% (p = 0.250), with a low (0.5%) and statistically insignificant genetic differentiation among isolation source populations (p = 0.250) ( Table 3). Further analysis of population genetic distance among geographic populations found that only 9 out of 55 pairs showed high Fst value and significantly genetic differentiation (P < 0.05, Table S3). The biggest Fst value was between Cangyuan and Silan (Fst = 0.268, P < 0.001), follow by that between Cangyuan and Eshan (Fst = 0.255, P < 0.001). Mantel test indicated that there was no significant correlation between geographic distance and genetic distance by PhiPT Values (P = 0.250, Fig. 2).  (Table 4), suggesting all populations did not experience population expansion.

N e u t r a l i t y t e s t s s h ow e d t h a t Ta j i m a's D
The abbreviations of eleven geographical populations and three isolation source are same as Fig. 1, and the abbreviation of four invasion ages see Table 1 Discussion Colletotrichum gloeosporioides species complex (CGSC) is one of the most widely distributed and common plant fungi [48]. They can host among 470 different plants as a pathogen or an endophyte [49]. Most research of CGSC has focused on economic crops [50] or fruits [42]. In this study, we obtained 165 strains belonging to CGSC from three isolation sources (the healthy and symptomatic leaves of invasive plant A. adenophora and the healthy leaves of the native plants) in eleven geographic sites (covering four invasion ages of A. adenophora). We quantified the genetic diversity of these CGSC strains by multiple gene loci sequencing and found distinct haplotype and nucleotide diversity for different gene locus; moreover, all eleven geographic populations showed a high level of haplotype and nucleotide diversity based on the concatenated sequences (aim 1). The haplotype of these CGSC did not cluster according to geographic site, isolation source or invasion age; AMOVA revealed that the genetic variation was mainly from within-populations, regardless of the geographic or isolation origin (aim 2). Finally, both AMOVA and neutrality tests indicated these CGSC strains occurred gene exchange among geographic populations but did not experience population expansion along with A. adenophora invasion progress (aim 3).
The results indicated that 5 out of 6 loci showed high nucleotide diversity (Pi > 0.02), only ITS locus had a low nucleotide diversity (Pi = 0.0074). Specifically, as previously reported [50], gapdh locus had the highest nucleotide diversity (Pi = 0.0705), followed by gs (Pi = 0.0341). The result also supported that gapdh and gs can be used to effectively distinguish most taxa in CGSC [51]. On the contrary, low level nucleotide diversity of ITS explained why single ITS is poor to define CGSC [51,52]. Thus, multilocus analysis could provide a more precise distinguish in differential genetic diversity from CGSC [23,53].
Unexpectedly, we found that the genetic diversity of our 165 CGSC isolates was very high, with a haplotype Fig. 1 Haplotype networks for the distribution and frequency (circle sizes) of 105 haplotypes based on the concatenated sequences of six loci (ITS-gapdh-chs-1-gs-cal-tub2) across eleven geographic populations (a) and three isolation sources (b) as well as distributions and frequency of haplotyps with more than one sequences in geographic populations (c) and solation sources (d). Note: Sample location of each isolate was deemed as a proxy to represent differences in geographic population and isolations sources (A. adenophora healthy leaves (Ah), A. adenophora symptomatic leaves (As) and other native plants symptomatic leaves (Os) were regarded as another proxy for different isolation source. The size of the circle is proportional to the number of sequences in each haplotype, one transverse line represents simple mutations. CY, Cangyuan population; ES, Eshan; YX, Yunxian; YJ, Yuanjiang; SL, Silan; XM, Ximeng; KM, Kunming; NR, Ninger; GM, Gengma; NH, Nanhua; NJ, Nanjian; LC, Lancang; ZT, Zhutang diversity (Hd) ranging from 0.67 to 1 and all nucleotide diversity (Pi) > 0.01 for all eleven geographic populations ( Table 1). Since population variability is generally considered to be large when the polymorphism index Pi is more than 0.01 [54], it is concluded that there is a significant genetic variation within these CGSC populations. Previously, Moges el at. found a low level genetic diversity of isolated 163 CGSC strains in two host plants belonging to Citrus from four cultivated orchards in Ethiopia [55]. The high nucleotide and haplotype diversity of CGSC in our study may be the result of diverse host species and wide geographic range (including invasive plant A. adenophora plus twelve neighbor plants from eleven geographic sites, see Table S2), as well as long introduction time of A. adenophora (the longest invasion time more than 80 years, see Table 1), since introduction time and alternative host species diversity may play a role in generating genetic variability within populations [56]. Moreover, we performed genetic analysis of these isolates on a species complex as reported previously [23,41,42]. One possible reason for high genetic diversity of these isolates may also be related to diverse species including in these CGSC in our study.
Therefore, it is necessary to clarify the species identity of these strains based on the morphological feature.
Alternatively, high diversity of our CGSC may suggest a weak selection for infection of CGSC strains by A. adenophora in the invaded range, which is favorable for frequent cross-over transmission among A. adenophora populations in different geographic sites, and between invasive and neighbor plant species. Accordingly, we found that CGSC haplotypes did not clustered by geographic sites, invasion ages or three isolation sources. AMOVA results showed that the major genetic differentiation was within populations (92.7 ~ 99.5%), whenever based on geographical populations or isolation source populations. Moreover, analysis of population genetic distance among geographical populations found that only 9 out of 55 pairs showed significantly genetic differentiation (P < 0.05), and Mantel test also indicated that there was no a significant correlation between geographical distance and genetic distance among geographic populations. These results suggested that the genotypic variation of CGSC in A. adenophora invasion system did not associated with geographic distance, invasion ages or isolation sources. Fungal lifestyle switching between endophyte and pathogen in different host plants even in the same host have been frequently reported [57,58]. For example, the same haplotype of Fusarium circinatum can be an endophyte colonizing in herbaceous plants but act as a reservoir of a pitch canker disease inoculum in Pinus radiate [59]. Tian et al. also indicated that Sclerotinia sclerotiorum severed as a widespread pathogen of dicotyledons, while it can protect wheat, rice, barley, maize, and oat against Fusarium head blight, stripe rust, and rice blast by acting as an endophyte [60]. In particular, for C. gloeosporioides, it is common a lifestyle switching between endophyte and necrotrophy [57]. C. gloeosporioides could transform endophytic strategies into saprotrophic lifestyle in the same host Magnolia liliifera [61]. Previously, Colletotrichum strains were isolated from asymptomatic leaves of A. adenophora and symptomatic leaves of neighbor plants, and these strains were nonvirulent to A. adenophora but virulent to some neighbor plants by Koch's postulates [28]. In this case, our multilocus analysis of CGSC further found that, the dominant haplotype H_10 shared by all eleven geographic populations and occurred in all three isolation sources; another dominant haplotype H_5 were detected in 6 of 11 geographic populations and in all three isolation sources (Fig. 1). The occurrence of these generalist haplotypes of CGSC verified that it was common for fungal lifestyle switches between endophyte and pathogen within host A. adenophora, as well as between A. adenophora and neighbor plants in invaded ecosystem. In this study, we ignored the reservoir potential of target CGSC associated with healthy leaves of the co-occurring native plants of A. adenophora. Generally, hosts with short-lived, poorly defended, nutrient rich and high metabolism tissue have great values for pathogen reservoir potential [62]. Moreover, the high richness of a susceptible host species may increase its capacity to serve as a pathogen reservoir [63]. Invasive exotics species often grow monoculture and high-density, and are often expected to be short-lived, fecund and poorly defended relative to natives [64,65], and are expected as ideal pathogen reservoir [62]. It thus may be greatly underappreciated of the potential disease risk driven by asymptomatic, endophytic CGSC associated with A. adenophora.
Fungal pathogens shared by the invader and native plants, either through spillover or spillback, may facilitate invasion [3,4], referred as disease-mediated invasion (DMI) [66]. In reviewing DMIs, Strauss et al. found that most diseasemediated animal invasions benefit from spillover rather than spillback; in contrast, most disease-mediated plant invasions benefit from spillback because most non-indigenous plants can be introduced by seeds and may not be accompanied by the same avirulent parasites, especially soil pathogens infected in the native ranges [66].To date, it is difficult to quantify the degree variation that spillover or spillback occurs in plant invasions, because DMIs have been reported only in a few invasive plants so far [13,67]. A. adenophora mainly disperses into a new scope by its seeds with tiny size and light weight [68]; moreover, there was no molecular evidence that Colletotrichum sp. is seedborne for A. adenophora [69]. In this case, neutrality tests indicated that CGSC gene exchanged among populations but population did not expand with invasion of A. adenophora (Table 4, all P > 0.05). Therefore, if CGSC-mediated invasion can occur in A. adenophora, this process is high possibility through spillback.
In conclusion, our multilocus analysis verifies that Colletotrichum gloeosporioides species complex (CGSC) associated with A. adenophora and other neighbor plants in the introduced range are high in haplotype and nucleotide diversity. Furthermore, several dominant haplotypes shared by most geographic populations, invasion ages, and by healthy and symptomatic leaves of A. adenophora, as well as symptomatic leaves of neighbor plants. The major genetic differentiation of CGSC is within populations rather than among populations, and geographical distance had no significant relation with population genetic distance among. The data suggested that genotypic variation of CGSC did not relate to geographic distance, invasion ages or isolation sources. It is concluded that there is high possibility that A. adenophora obtain local foliar CGSC from co-occurring neighbor plants