Emerging fungal pathogen of an invasive grass: Implications for competition with native plant species

Infectious diseases and invasive species can be strong drivers of biological systems that may interact to shift plant community composition. For example, disease can modify resource competition between invasive and native species. Invasive species tend to interact with a diversity of native species, and it is unclear how native species differ in response to disease-mediated competition with invasive species. Here, we quantified the biomass responses of three native North American grass species (Dichanthelium clandestinum, Elymus virginicus, and Eragrostis spectabilis) to disease-mediated competition with the non-native invasive grass Microstegium vimineum. The foliar fungal pathogen Bipolaris gigantea has recently emerged in Microstegium populations, causing a leaf spot disease that reduces Microstegium biomass and seed production. In a greenhouse experiment, we examined the effects of B. gigantea inoculation on two components of competitive ability for each native species: growth in the absence of competition and biomass responses to increasing densities of Microstegium. Bipolaris gigantea inoculation affected each of the three native species in unique ways, by increasing (Dichanthelium), decreasing (Elymus), or not changing (Eragrostis) their growth in the absence of competition relative to mock inoculation. Bipolaris gigantea inoculation did not, however, affect Microstegium biomass or mediate the effect of Microstegium density on native plant biomass. Thus, B. gigantea had species-specific effects on native plant competition with Microstegium through species-specific biomass responses to B. gigantea inoculation, but not through modified responses to Microstegium density. Our results suggest that disease may uniquely modify competitive interactions between invasive and native plants for different native plant species.


Introduction
Both plant invasions and infectious diseases can affect natural plant communities by reducing plant diversity and biomass production [1][2][3][4]. Invasive species and disease outbreaks can co-occur in communities because the species are co-introduced, or because invasive species amplify disease transmission [5]. Invasive plants can negatively impact native plant communities through competition [6] and diseases may increase, decrease, or have no net effect on invasive plant impacts [7,8]. The responses of both the invasive species and competing native species to infection can determine which of these outcomes occurs [9,10]. For example, infected invasive plants are predicted to have lower competitive effects than uninfected invasive plants when native species have greater disease resistance or tolerance [11]. In empirical and theoretical tests of disease-mediated competition between a single native plant species and a single invasive plant species, disease has both increased [12,13] and decreased [14,15] impacts of the invasive species. However, the relevant guild of native species in natural communities is often diverse and species vary in their susceptibility to pathogen infection [16], making it unclear whether results from studies of disease-mediated competition with one native species apply to the broader native community.
A shared pathogen can create an antagonistic interaction between two or more host species (i.e., apparent competition), even in the absence of other forms of interaction, such as resource competition [17]. Increases in the abundance of one host species can decrease the fitness of another through transmission and the negative effects of disease [17]. Invasive species may enhance pathogen transmission to or disease impacts on native species [18,19]. For example, high densities of invasive cheatgrass (Bromus tectorum) promoted infection of native plant seeds by a fungal pathogen [20]. Pathogen infection of native or invasive plants can reduce growth, reproduction, and survival [2,21], as well as induce compensatory growth or reproduction [22,23]. Because disease can influence the productivity and composition of natural plant communities [4], disease amplification by invasive plants could have potentially strong effects on native plant communities [24].
Disease can modify the effects of plants on shared resources and their responses to resource limitation [25]. For example, disease-induced reductions in total leaf area can decrease light interception [26], potentially increasing light availability to lower canopy levels and decreasing photosynthesis of infected plants. If disease disproportionately impacts invasive species, reductions in growth and resource uptake may release native species from competition, which was reported when a powdery mildew fungus infected invasive garlic mustard (Alliaria petiolata) [15]. In contrast, disease-induced fitness costs may reduce the competitive ability of native species, which is hypothesized to have promoted invasion of European grasses in California [12,13]. Plant species can vary widely in their competitive ability [27,28], susceptibility to infection [29,30], and performance losses due to disease [23,31]. Differential responses of native species to disease and competition with an invasive species could determine how the native plant community responds to disease-mediated competition.
Microstegium vimineum (Trin.) A. Camus (stilt grass, hereafter Microstegium) is an annual grass species native to Asia that was first recorded in the United States in 1919 [32]. Microstegium forms dense populations and litter layers in eastern and midwestern U.S. forest understories, suppressing herbaceous plants and tree seedlings [28,33]. Over the last two decades, Microstegium populations have acquired fungal leaf spot diseases caused by species in the genus Bipolaris that reduce its biomass and seed production [29,34]. Bipolaris gigantea (Heald & F.A. Wolf) B. Lane, Stricker, M.E. Sm., S.L. Flory & Harmon is a common pathogen of Microstegium [34] that causes zonate leaf spots with dark brown margins [35] and likely disperses via wind and splashing [36], but with restricted distances due to its large spore size [35,37].
Dichanthelium clandestinum L. Gould (syn. Panicum clandestinum L.; deer-tongue grass, hereafter Dichanthelium), Elymus virginicus L. (Virginia wild rye, hereafter Elymus), and Eragrostis spectabilis (Pursh) Steud (purple lovegrass, hereafter Eragrostis) are perennial grass species that are native to the U.S. and co-occur with Microstegium [29,34]. Bipolaris gigantea can infect, produce lesions on, and produce spores on Elymus virginicus [35]. Bipolaris gigantea infections have been observed on Dichanthelium clandestinum in the field, but it may not be a competent host for spore production [37]. At least three species in the genus Eragrostis are susceptible to B. gigantea infection [37][38][39] and closely related plant species are more likely to share pathogens than distantly related species [30], suggesting that Eragrostis spectabilis may also be susceptible to B. gigantea. Infection by B. gigantea may be more likely when these grass species co-occur with Microstegium due to high infection rates in some Microstegium populations [34,40]. Bipolaris gigantea may alter the competitive ability of Microstegium or the ability of these native species to compete with Microstegium.
Competitive ability depends on the fitness of species in the absence of competition and their responses to changes in intraspecific and interspecific competitor densities (i.e., competition coefficients) [41,42]. Here we investigated how B. gigantea inoculation affected the competitive ability of the three native perennial grass species in a greenhouse experiment by measuring their competition-free biomass and responses to Microstegium density. We were uncertain about how B. gigantea would affect native species biomass, but acknowledged that a range of outcomes were possible given interspecific variation in host-pathogen interactions [29][30][31], including decreased, increased (e.g., through compensatory growth), and no change in biomass. We hypothesized that B. gigantea infection would reduce the negative effect of Microstegium density on native plant biomass because diseased Microstegium would be smaller [29,34]. However, we also expected that disease-induced biomass reduction experienced by some native species would increase their sensitivity to Microstegium density.

Greenhouse experiment
We performed the experiment in a greenhouse in Gainesville, FL, USA, from June 19, 2019 to September 12, 2019. We used Microstegium seeds collected from Big Oaks National Wildlife Refuge (BONWR) in Madison, IN, USA (38.9365, -85.4148) in 2015, Elymus and Eragrostis seeds purchased from Prairie Moon Nursery (Winona, MN, USA) in 2018, and Dichanthelium seeds purchased from Sheffield's Seed Company (Locke, NY, USA) in 2018. All seeds were stored at 4˚C. Prior to the experiment, seeds of each species were planted in a greenhouse and seedlings developed no lesions, suggesting that lesions caused by B. gigantea inoculation were unlikely to be confused with lesions caused by potential seedborne pathogens. The potting mix used in the experiment (Jolly Gardener Pro-Line Custom Growing Mix) was autoclaved at 120-130˚C for 30 minutes and all pots and trays were sprayed with 10% bleach solution (0.6% sodium hypochlorite) and rinsed with tap water after approximately five minutes to minimize risk of contamination by non-focal pathogens.
To quantify the effect of Microstegium competition on the native species, we used an additive competition experimental design [41,43] with one individual of a native species surrounded by 0, 2, 10, 50, or 100 Microstegium plants per 1 L pot (Fig 1A and 1B). First, we sowed seeds for each native species into separate trays to germinate. Seven days later, we planted Microstegium seeds in 1 L pots according to their density treatment (50 and 100 seed numbers estimated by weight). The native species were transplanted from germination trays to the 1 L pots with Microstegium after growing in the greenhouse for 28 days. We chose native plant individuals that were similar in size (2 to 3 true leaves) to transplant into pots. The 15 plant combinations (each of the three native species with five Microstegium densities) were replicated eight times, half of which were inoculated with B. gigantea and half of which were mock inoculated with a control solution.
The pure culture of B. gigantea (BGLMS-1 in the collection of Dr. Philip Harmon, University of Florida) used in this research was originally isolated from Microstegium as part of a previous study and had been stored as previously described [35]. Bipolaris gigantea was revived from 4˚C storage by placing colonized, 3 to 5 mm diameter, filter paper pieces on half-strength V8 media agar plates. Fungal colonies grew under 12 h day and night fluorescent light at 26˚C for one week and were transferred to new half-strength V8 media agar plates. Conidia were harvested from fungal colonies by flooding plates with 10 ml of sterile deionized water with 0.1% Tween 20 (Sigma-Aldrich, St. Louis, MO, USA). The resulting conidia suspension was filtered through a layer of cheese cloth, and conidia were quantified with a Spencer Bright-Line hemocytometer (American Optical Company, Buffalo, NY, USA). The concentration of inoculum was adjusted to 15,000 conidia/ml and applied to plants with a Passche H-202S airbrush sprayer (Kenosha, WI, USA). Inoculations occurred six days after planting the native species with the Microstegium, and half of the pots were sprayed until runoff with the spore suspension while the other half were sprayed with the same volume of sterile deionized water with 0.1% Tween 20 (i.e., mock inoculation control). To encourage infection, we placed a paper towel wet with deionized water in each pot and sealed each pot with a transparent plastic bag secured with a rubber band. The plastic bags and paper towels were removed after seven days [44]. Plants were watered daily before and after they were contained in the plastic bags. Ten days

Data collection
To assess disease incidence (i.e., the percentage of leaves per pot with at least one lesion [45]) 14 days after inoculation, we recorded the number of Microstegium leaves with one or more Bipolaris-like lesions ( Fig 1C) and the total number of leaves for three Microstegium plants per pot (or two plants for the pots with only two). No other types of lesions were observed on the plants. The number of leaves per plant were averaged within pots and multiplied by the total number of plants per pot, based on seeding rate, to estimate the total number of Microstegium leaves per pot. For native plants that received the B. gigantea inoculation treatment and had lesions, we counted the number of leaves with at least one lesion and the total number of leaves per plant. None of the plants in mock-inoculated pots had lesions with one exception: in one pot that contained Dichanthelium and 100 Microstegium plants, 46 Microstegium leaves had lesions. We removed this replicate from analyses.
We used these visual assessments of Bipolaris-like lesions to approximate B. gigantea infection of experimental plants. Bipolaris gigantea association with foliar lesions can be assessed by using microscopy to identify conidiophores on leaves after incubation [35]. The absence of B. gigantea conidiophores in lesions, however, does not confirm that it is not the causal agent. In leaf samples collected from BONWR in 2018 and 2019, 67% of Microstegium samples (n = 238) and 48% of Elymus samples (n = 65) that had Bipolaris-like lesions also had B. gigantea conidiophores identifiable by microscopy. In addition, 28% of Microstegium samples (n = 29) and 1% of Elymus samples (n = 106) that did not have Bipolaris-like lesions had B. gigantea conidiophores identifiable by microscopy (S1 Table). Therefore, Bipolaris-like lesions are commonly associated with B. gigantea infection and it is less common for leaves to be infected without lesions. Because we did not test leaves for infection with B. gigantea following inoculation, we present results in the context of the inoculation treatments rather than infection status.
To assess plant performance, we harvested the aboveground biomass of all pots on September 12, 2019 (51 days after inoculation), separated the native plants from the Microstegium, dried the biomass at 60˚C to constant mass, and weighed it. Biomass production can act as a proxy for perennial plant fitness [41]. While seed production is a more meaningful measure of annual plant fitness [41], Microstegium biomass is correlated with its seed production [46].

Statistical analyses
To evaluate disease incidence on plants across the Microstegium density gradient, we fit a generalized linear regression to the estimated proportion of Microstegium leaves with lesions per pathogen-inoculated pot using Microstegium density (the number of Microstegium seeds added to each pot), native species identity, and their interaction as the explanatory variables. The model was fit with Bayesian statistical inference using the brm function in the brms package [47], an interface for Stan [48], in R version 3.5.2 [49]. The model contained three Markov chains with 6000 iterations each and a discarded burn-in period of 1000 iterations. We chose prior distributions based on whether model variables could reasonably take on negative values in addition to positive values (Gaussian or Cauchy) or not (gamma or exponential). We chose parameters for prior distributions that reflected limited a priori information about variable values. We used a binomial response distribution (logit link) and a Gaussian distribution for the intercept and coefficient priors (location = 0, scale = 10). We calculated the mean and 95% highest posterior density interval (hereafter, "credible interval") of back-transformed (from logit to percentage) model estimates using the mean_hdi function in the tidybayes package [50]. There were too few native plant leaves with lesions to statistically analyze disease incidence, nevertheless, we present these results graphically to assess qualitative patterns.
To evaluate the effects of Microstegium density and B. gigantea inoculation on Microstegium performance, we fit a linear regression to Microstegium biomass: This formulation allowed us to estimate quadratic relationships between Microstegium biomass and Microstegium density for each native species-inoculation treatment combination. We used a Gaussian response distribution, a Gaussian distribution for the intercept prior (location = 2, scale = 10) and the coefficient priors (location = 0, scale = 10), and a Cauchy distribution for the standard deviation prior (location = 0, scale = 1). Otherwise, the model was fit using the same methods described for disease incidence.
To evaluate the effects of inoculation treatment and Microstegium density on native plant biomass, we fit a Beverton-Holt function to native plant biomass: We fit this function to all of the native plant biomass data, estimating separate b 0 (biomass in the absence of competition) and α (biomass response to Microstegium density, i.e., competition coefficient) values for each native species-inoculation treatment combination. We used a Gaussian response distribution, a Gamma distribution for the b 0 prior (shape = 2, scale = 1), an exponential distribution for the α prior (rate = 0.5), and a Cauchy distribution for the standard deviation prior (location = 0, scale = 1). Otherwise, the model was fit using the same methods described for disease incidence. To evaluate differences in b 0 and α between treatments, we subtracted the estimate for one treatment from the other for each posterior sample (n = 1500) and then calculated the mean and 95% credible intervals [50]. To assess model fits, we checked that the r-hat value for each parameter was equal to one, visually examined convergence of the three chains, and compared the observed data to simulated data from the posterior predictive distributions using the pp_check function [47]. We report a model coefficient as statistically significant if its 95% credible interval ("CI", i.e., 95% probability that this interval of the posterior distribution contains the true estimate value) omits zero [51,52]. We used the tidyverse packages to clean data and create figures [53].

Results
We observed Bipolaris-like lesions on Microstegium leaves in 94% of pots in which Microstegium was planted and inoculated. The average Microstegium disease incidence in low-density pots (i.e., two Microstegium plants) was 8% (95% CI: 6%-11%). The species identity of the native plant in low density pots did not significantly affect Microstegium disease incidence (Table 1). Microstegium disease incidence was constant across the Microstegium density gradient when Elymus was present (Fig 2B). However, Microstegium disease incidence decreased by four percentage points (95% CI: -7%--2%) and five percentage points (95% CI: -7%--3%) when 100 Microstegium were grown with Dichanthelium and Eragrostis, respectively, relative to pots with two Microstegium plants and the same native species (Fig 2A and 2C). The total number of leaves per pot increased across the Microstegium density gradient while the number of leaves with lesions increased more slowly or not at all (S1 Fig).
Bipolaris gigantea inoculation resulted in lesions on all three native plant species but only in some of the Microstegium density treatments (Fig 3). Lesions formed on 7 out of 20 Elymus plants (Fig 3B), but only 3 out of 20 plants for each of the other species (Fig 3A and 3C). Of the Dichanthelium and Elymus plants with lesions, higher Microstegium density tended to increase the percentage of leaves with lesions. For example, 17% of Dichanthelium leaves had lesions when grown with 100 Microstegium compared to 10% with 10 Microstegium. Similarly, 38% of Elymus leaves had lesions when grown with 100 Microstegium compared to 23% with 10 Microstegium.
Bipolaris gigantea inoculation did not significantly affect Microstegium biomass relative to the mock inoculation control ( Table 2). In addition, Microstegium biomass was not significantly different among treatments with different native species (Table 2). Microstegium biomass increased with Microstegium density (0.16 g plant -1 , 95% CI: 0.05 g plant -1 -0.28 g plant -1 ) and the relationship between biomass and density varied in shape, although not significantly, when grown with the three species: saturating at high densities when grown with Dichanthelium (Fig 4A), increasing nearly linearly when grown with Elymus (Fig 4B), and peaking at intermediate densities when grown with Eragrostis ( Fig 4C).

Discussion
We evaluated how inoculation with the emerging fungal pathogen B. gigantea affected the biomasses of three native species in competition with the invasive plant Microstegium. Bipolaris  gigantea inoculation did not significantly affect Microstegium biomass relative to the mock inoculation control and it had contrasting effects on the native plant species by increasing Dichanthelium biomass, decreasing Elymus biomass, and having no effect on Eragrostis biomass. The negative effect of Microstegium density on biomass for each of the native species was the same whether plants were inoculated with B. gigantea or the control. These results suggest that B. gigantea may differentially affect native plant species in competition with Microstegium through species-specific responses to B. gigantea exposure.

Disease-mediated competition
Studies on disease-mediated competition between invasive and native plant species indicate that infection of invasive plants can contribute to either native plant persistence or recovery [14,15,34,54]. However, in our experiment, inoculation with Bipolaris did not modify the effect of Microstegium density on the three native species relative to the mock inoculation control, suggesting that competitive effects of Microstegium on native species are likely to be consistent in the presence or absence of low levels of disease incidence. Our experimental design captured two components of native species' competitive ability: their growth in the absence of competition and their biomass responses to changes in Microstegium density (i.e., interspecific competition coefficients) [41,42]. To better characterize the competitive ability of the native species, it is also necessary to estimate their biomass responses to changes in their own density (i.e., intraspecific competition coefficients) [41,42]. Because B. gigantea inoculation had unique effects on the growth of the three native species, it may also uniquely affect their per capita impacts on competitor growth, altering their intraspecific competition coefficients. If, however, B. gigantea does not affect the intraspecific competition coefficients of the native species, we would expect low levels of B. gigantea exposure to increase the competitive ability of Dichanthelium in interactions with Microstegium, decrease the competitive ability of Elymus, and to have no effect on the competitive ability of Eragrostis, potentially leading to shifts in the relative abundances of native species. Predicting the long-term outcome of disease impacts on native-invasive interactions requires also understanding the invasive species' competitive ability and niche overlap between the native and invasive species (e.g., overlapping resource requirements or natural enemies) [25,42,55]. Because the native species are perennial and the invasive species is an annual, studies that examine the effects of disease on aspects of plant fitness other than biomass, for example, annual seed survival and perennial adult survival, are necessary for characterizing the comprehensive impacts of disease on native-invasive plant competition [41,55].
Our experimental methods and conditions may have limited the levels of B. gigantea leaf spot disease (see Limitations section), and there were no effects of B. gigantea inoculation on Microstegium biomass, likely leading to limited changes in Microstegium resource acquisition. Therefore, it is crucial to explore disease-mediated competitive effects of Microstegium in the field or with methods that may result in disease incidence approaching levels observed in the field. For example, disease incidence on Microstegium decreased as Microstegium densities increased, which is likely because the single B. gigantea inoculation infected a relatively constant number of leaves regardless of Microstegium density, leading to lower percentages of leaves with lesions at higher densities. In contrast, higher plant densities in the field may promote higher disease incidence and greater inoculum production [56,57], in which case disease may have stronger impacts on Microstegium competition than what we observed in the experiment. Greater B. gigantea inoculum levels and multiple disease cycles may also have larger effects on native plant responses to competition, for example, through reduced ability to capture resources [58]. Nonetheless, the impacts of B. gigantea on Microstegium and native plant competition may simply be minor, as has been demonstrated for disease effects on cheatgrass competition [14] and herbivory effects on Amur honeysuckle (Lonicera maackii) competition with native species [59]. In that case, competitive effects of the invasive plant on native species are likely to overshadow the effects of disease, which may be common across plant communities [60].

Consequences of B. gigantea inoculation
It is likely that pathogen amplification by invasive plants has distinct effects on different native species [20,31]. The three native species in our experiment showed unique biomass responses to B. gigantea inoculation in the absence of Microstegium competition. The range of B. gigantea inoculation effects on biomass, from negative (on Elymus) to positive (on Dichanthelium) is consistent with the theory that plant-microbe interactions can vary from mutualism to parasitism depending on context, such as environmental conditions and host identity [61,62]. For example, infection with Cucumber mosaic virus increased the biomass and seed weight of one Arabidopsis thaliana genotype while it reduced the biomass and seed weight of another genotype relative to a mock inoculation control [63]. Our results suggest that B. gigantea could increase aboveground growth for some host species (e.g., through compensatory growth or reallocation of resources [22,63]) and suppress the aboveground growth of other species. Studies encompassing a broader range of environmental conditions and host diversity are needed to better predict when B. gigantea will have positive, negative, or neutral effects on host biomass and other traits. In studies of soil microbes, the effects of microbial inoculations on plant growth can predict plant species relative abundances in the field [64,65]. However, whether disease-induced changes in growth of plant species, as observed in our study, are sufficient to shift plant community structure is an important area of future research [25].
Interestingly, no fungal lesions were observed on the three native plant species in the absence of competition despite seven days of incubation inside plastic bags. The absence of visible symptoms, however, does not necessarily indicate a lack of infection. For example, some fungi are asymptomatic endophytes of invasive Crofton weed (Ageratina adenophora) but cause visible leaf spots on co-occurring plant species [66]. Fungal lesions were observed on some native plants grown with Microstegium, perhaps because Microstegium biomass altered the microclimate of the pots (e.g., increased humidity), which can be more suitable for lesion formation [56]. In addition, transmission from Microstegium to native plants may have maintained infections. Indeed, B. gigantea transmission from more competent host species to less competent host species has been inferred from field observations [37]. If Microstegium biomass amplifies B. gigantea incidence on native species in the field, B. gigantea could drive apparent competition between Microstegium and species negatively affected by B. gigantea [17].

Limitations
Experimentally suppressing Bipolaris infection using fungicide in the field increased Microstegium biomass by 33-39% [29,34], suggesting substantial effects of severe disease symptom development. However, despite using a pathogenic Bipolaris isolate in our experiment [35], inoculation caused low levels of disease incidence (relative to approximately 40% of leaves with lesions documented in the field, S2 Table), which had no effect on Microstegium biomass relative to the mock inoculation control. While such results could be explained by Microstegium tolerance or compensatory growth [22,23], it is more parsimonious to assume that B. gigantea exposure was below levels experienced in the field. Pathogen transmission and disease incidence depend on the favorability and duration of environmental conditions and the inoculum load [67,68]. Our experiment relied on a single inoculation and extended incubation; however, field conditions that result in cycles of leaf wetness events (e.g., dew or precipitation) can enhance fungal infection [67] and promote multiple disease cycles throughout the growing season. The concentration of B. gigantea conidia in our experimental inoculations was limited by the number of conidia we could harvest from agar plates in the lab and was relatively low (15,000 conidia/ml compared to e.g., 10 5 conidia/ml [68]). While the conditions for leaf wetness and conidia suspension concentration likely limited the possible extent of disease incidence, they may reflect initial disease dynamics in the field, which is consistent with the age of plants we used in the experiment.
Although we did not test plants for infection with B. gigantea infection following the experiment, our results indicate that observed lesions and biomass effects were likely due to the B. gigantea inoculation treatment. Only a single pot in the mock inoculation treatment exhibited disease symptoms while 94% of B. gigantea-inoculated pots with Microstegium (n = 48) displayed disease symptoms. In addition, a relatively high number of leaves in the one mockinoculated pot had lesions. These two results suggest that the pot was inadvertently inoculated when treatments were applied. An alternative explanation is that the one mock-inoculated pot was contaminated (e.g., seeds harbored pathogenic fungi or external contaminants were introduced). While either is a possibility, the latter does not explain the much higher percentage of inoculated pots with lesions relative to mock-inoculated pots. In addition, B. gigantea can cooccur with other pathogens in the field (S3 Table). Therefore, infection of plants in our experiment that is not confounded with the inoculation treatment does not negate our ability to evaluate the effects of the inoculated B. gigantea strain on competition between native plants and Microstegium. Future efforts that aim to better characterize host-pathogen interactions between Microstegium or native plant species and Bipolaris fungi could confirm infection by attempting to re-isolate the fungus after inoculation.

Conclusions
We used a greenhouse experiment to demonstrate that inoculation with a fungal leaf spot pathogen that has accumulated on a widespread invasive grass had unique effects on the growth of native species but did not modify biomass responses of native species to Microstegium density. Complementary experiments in the field could help determine whether these findings are consistent across other native species and when disease pressure is higher. Transmission of B. gigantea may depend on Microstegium densities, potentially creating feedbacks between infection and density, which we controlled for in our experiment. The competitive effects of native plant species on Microstegium in the presence and absence of disease also may be important for understanding long-term community dynamics [8]. The emergence of infectious diseases in invaded plant communities may lead to natural biological control of the invasive species [7], exacerbated effects of invasion if the pathogen negatively impacts native species [12], or there may be no effect of disease [8]. Altogether, our study suggests that low levels of disease caused by B. gigantea may have unique effects on native species but are unlikely to modify the large negative impact of invasive Microstegium density on native species.
Supporting information S1 Fig. Microstegium leaves. The estimated number of Microstegium leaves with lesions across the Microstegium density gradient following B. gigantea inoculation when grown in the presence of (A) Dichanthelium, (B) Elymus, and (C) Eragrostis (mean ± 95% confidence intervals). All leaves with lesions were counted and the total number leaves per pot were estimated by counting the number of leaves on up to three plants per pot. (TIF) S1 Table. Bipolaris gigantea identification on field-collected leaves. Raw data collected from an experiment at BONWR in which fungicide or water (control) were added to plots with ten planting treatments. Leaves of Microstegium vimineum (Mv) and Elymus virginicus (Ev) were assessed for visible eyespots and B. gigantea conidiophores using microscopy. (DOCX) S2 Table. Microstegium vimineum infection incidence in the field. Raw data collected from Microstegium in an experiment at BONWR in which the total number of leaves per stem and the number of leaves with at least two foliar lesions per stem were recorded. The plots included were sprayed monthly with water a control for fungicide (not included). (DOCX) S3 Table. Fungi identification on field-collected leaves. Raw data collected from an experiment at BONWR in which fungicide or water (control) were added to plots with ten planting treatments. Leaves of Microstegium vimineum (Mv) and Elymus virginicus (Ev) were assessed for B. gigantea conidiophores using microscopy. Bipolaris gigantea was isolated, as well as some co-occurring fungi, including Pyricularia spp., Bipolaris spp. other than B. gigantea, and Curvularia spp. Leaves were collected in late August of 2018. (DOCX)