Festuca thurberi plants were grown in a weatherport on site at the Rocky Mountain Biological Laboratory (N 38.95807°, W 106.98853°; 2,889 m). The experiment tested the effects of leachate from an introduced member of the Brassicaceae (Thlaspi arvense), a native member of the Brassicaceae that shares habitat with our study species (Noccaea fendleri), a positive control with leachate from additional Festuca thurberi, and a negative control treatment [deionized (DI) water] on plant growth and mortality of F. thurberi over time. We included the co-occurring native mustard N. fendleri in this experiment to ensure that any treatment effect observed by T. arvense was due to the novel interaction with an introduced mustard and not simply the effect of phytochemical interactions with any member of the Brassicaceae.

Experimental plants were collected as individual ramets from eight distinct clumps of F. thurberi and an arbitrary selection of dispersed individuals at the beginning of the growing season on 6/24/2018 from a sub-alpine meadow (∼3,826 m; Supplementary Table 1). This collection site was selected due to the absence of the introduced mustard T. arvense and the presence of N. fendleri. For the purposes of this study, we assumed that ramets from the same clump were replicates of the same genotype, given the clonal growth form of the species. We assigned a separate category (D) to replicates of dispersed ramets growing individually in the same meadow. Following soaking, each ramet was manually sorted and separated from the larger root mass. Ramets, composed of an individual shoot and root mass, were harvested from each individual clump (genotype) and yielded between 18 and 44 clones. Ramets were then cleaned, air-dried, weighed and planted into 38-cell propagation trays (Greenhouse Megastore; item CN-PRT 38-10) within 24-h. Ramets of each clone were planted in peat moss mixture, augmented with field soil collected from beneath mature F. thurberi individuals at a 6:1 ratio, in order to expose clones to microbial communities associated with F. thurberi. One ramet of F. thurberi (N = 276; Supplementary Table 1) was grown in each cell of 10 38-cell propagation trays in a randomized experimental array. Experimental plants were watered with tap water for 7 days and allowed to settle in propagation sheets to minimize transplant stress. Each plant was then randomly assigned a treatment (leacheate of T. arvense, N. fendleri, F. thurberi, or a water control) with approximately 70 Festuca plants in each treatment (Supplementary Table 2).

Treatment plants (T. arvense, N. fendleri, and F. thurberi) were also collected at the beginning of the experiment, with three replicates of three plants per pot transplanted into 5 L pots for each treatment. Treatment plants were collected from a similar elevation to the field station (∼2,889 m), with the exception of F. thurberi for which treatment plants were collected from the same site as experimental plants (∼3,826 m). Leachate was collected separately for each replicate of each species, following design by Hagan et al. (2013) by running 1.3 L of DI water through a pot until roughly 1 L of leachate had been collected. In order to determine if root exudates are responsible for any treatment effect, as has been established in A. petiolata (Stinson et al., 2006), we used leachate rather than entire plants. Each of the three, 5 L treatment plant pots, was used to extract leachate successively to water the experimental plants.

Experimental plants were watered with treatment leachate or control DI until liquid started to pool at the surface of individual cells every other day or every third day depending on how long plants took to dry out. Plants were harvested over the course of 5 days, 51 days after transplant into the experimental array. Root sub-samples (5 fine roots) were collected from a subset of experimental plants (n = 106) and stored in 70% ethanol and whole individual plants were stored in brown paper bags until processing for biomass at the University of Massachusetts, Amherst.

New growth was assessed on every experimental plant in the array every 10 days starting 10 days after the first leachate treatment was applied (7/10/2018, 7/20/2018, and 7/30/2018) and marked as either new growth (i.e., new sprouts from the base of the plant) present (1) or absent (0). Similarly, mortality was evaluated twice, 15 and 30 days after transplant (7/15/2018 and 7/30/2018) and marked as either above ground green biomass absent (1) or present (0). Relative measure of greenness linked to chlorophyll content was measured using Spad technology (Spad 502; Minolta) every 2 days from 7/10/2018–7/20/2018. Only plants that produced new growth since the start of treatment application as assessed on 7/10/2018 were evaluated for relative greenness. Once harvested, plants were separated into above ground biomass (shoots and blades) and belowground biomass (roots).

Colonization by arbuscules, vesicles, and dark septate endophytes (DSE) were quantified in fine roots using staining followed protocols established in Ranelli et al. (2015). Samples were rinsed in tap water to remove ethanol and then partitioned into tissue cassettes for clearing and staining. Cassettes were cleared in a 10% KOH solution for 5–7 days, with variations of up to 2 days in clearing time due to individual thickness of roots. Potassium hydroxide was rinsed from cassettes and transferred to acidified water. Roots were then stained with a 5% Schaffer ink/white vinegar solution for 15–20 min and then de-stained in 10% KOH for several days. Quantification was performed using the magnified intersection method at 450 × magnification, spanning the upper and lower limit of each root piece in order to satisfy the requirements of the method (McGonigle et al., 1990).

All analyses were conducted in R version 4.0.2. (R Core Team, 2020) with the significant p-value set to below or equal to 0.05.

To confirm that plants were effectively randomized across the treatments we ran an ANOVA with treatment as the main effect and fresh biomass measured prior to transplant into the experimental array as the response variable.

Because we were interested in the role of intraspecific variation in plant responses to invasion, we adapted a standard quantitative genetics modeling approach (Falconer, 1960) to assess the relative contributions of the leachate treatments (environment) and intraspecific variation among F. thurberi clones (genotype). Our models included main effects of leachate, genotype, and the leachate × genotype interaction. We use “genotype” in this study in the broadest sense, in that ramets within individual clumps of the clonal study species, F. thurberi are assumed to be genetic replicates of one another and that clonal clumps separated by several meters may be genetically distinct. For the purposes of this study, we interpret a leachate effect as evidence of allelopathic effects of native and non-native mustard plants on F. thurberi phenotypic traits (i.e., differences in the mean response of F. thurberi plants to the leachate treatments), a genotype effect as a test for evidence of intraspecific variation in F. thurberi phenotypic responses to the leachate treatment (visualized as the mean response of the genotypes in the figures), and a significant GXE interaction as evidence for variance among genotypes in the direction of the phenotypic response to the treatments (visualized as lines that intersect in the figures).

To test our hypotheses that T. arvense leachate would affect growth through time, we constructed binomial regression models with treatment, genotype, and a treatment × genotype interaction term as the main effects and observations of new growth [0 (absent) or 1 (present)] as the response variable. We included experiment days (i.e., days since experiment began) as a covariate to account for changes in growth over time. We also constructed linear models using treatment, genotype, and a treatment × genotype interaction as the main effects with mean growth as the response variable. For mortality, we used a binomial regression with the same predictive variables and mortality (live or dead) as the response variable.

To determine if T. arvense leachate reduced general greenness linked to chlorophyll content we constructed linear models using experiment days, treatment, genotype, and a treatment × genotype interaction term as predictive variables and Spad reading as the response variable. We also constructed linear models using treatment, genotype, and a treatment × genotype interaction as the main effects with Spad reading as the response variable.

To determine if T. arvense leachate affected F. thurberi growth we constructed linear models using treatment, genotype, and a treatment × genotype interaction for the predictive variable related to final size (response variable): whole plant dry biomass, above ground biomass, belowground biomass, and root:shoot ratio. Dry biomass was normalized through log transformation. A constant was added to all values prior to log transformation to correct for negative values resulting from the transformation.

To determine the effects of T. arvense leachate on fungal root colonization we constructed a linear model with genotype, treatment, and a treatment × genotype interaction as the predictive variable and colonization of fine roots by arbuscules, vesicles, and DSE as the response variable. These response variables were normalized through an arcsin transformation. Dark septate endophytes were analyzed with the same effects in the model but with a Poisson distribution.

To visualize effects of the environmental treatments and the potential for underlying genetic variation in each response variable, we plotted reaction norms (Falconer, 1960; Via and Lande, 1985) of the mean trait value for each genotype.

Effects of biological invasion on the recruitment and mortality of native plant species are likely to affect the future composition of vegetation communities around the world (Alvarez and Cushman, 2002). Recruitment in clonal plants such as F. thurberi is linked to new growth of ramets (Shumway, 1995). Here we found that new growth of individual ramets varied across genotypes regardless of exposure to leachates from native and non-native mustards or to the control treatments. Several genotypes lagged in new growth early after experiment establishment but had new growth present 20 or 30 days into the experiment. We also noted an interaction between genotype and time, as indicated by the intersecting trendlines in Figure 1. Such plastic responses are likely to play an important role in the persistence of plant populations as global change introduces novel pressures, including the incursion of non-native plants (Frei et al., 2014). This variation in recruitment may be amplified in natural settings due to the natural fragmentation of high elevation populations, similar to island habitats (Burkey, 1995). Mortality in F. thurberi was similarly affected by individual genotypes with some genotypes having little to no mortality and others experiencing higher levels of mortality across the different leachate treatments (1.3–30.7%; Figure 2). The interaction effect of genotype and leachate treatment on mortality further suggests genotypes within a population of F. thurberi may vary widely in their ability to survive in the presence of leachate from mustard species.

Contrary to our hypotheses, simulated invasion by T. arvense with leachate application resulted in no significant effects on mean phenotypic values for new growth, biomass, or percent colonization with arbuscular mycorrhizal fungi of F. thurberi, and the treatment effect on relative greenness was attributable to the native mustard, N. fendleri as discussed below. The lack of an effect of treatment on these traits was surprising, given that leachates from T. arvense’s close relative, the invasive plant Alliaria petiolata, were shown to prevent germination of AMF, inhibit AMF symbioses, and reduce root lengths in native North American tree seedlings (Roberts and Anderson, 2001). Moreover, lower levels of AMF colonization in three native tree species were observed in soils invaded by A. petiolata (Stinson et al., 2006). Growth of AMF in pure culture was also completely inhibited by application of an isothiocyanate associated with sinigrin, one of the main glucosinolates found in both T. arvense and A. petiolata (Wolfe et al., 2008). Despite the lack of effects of T. arvense on F. thurberi in this experiment, further study may be warranted to determine if future interactions between T. arvense and endemic plants and soil microorganisms will result in more significant disruption to normal ecosystem function. The fact that we found evidence for genotype effects in many F. thurberi traits and responses to our experimental treatments underscores the importance of understanding intraspecific variation in plant species’ responses to biological invasion and other global change factors (Gibson et al., 2014; Johnson et al., 2015). In addition to recruitment, mortality, and relative greenness, dry-weight biomass allocation to roots and shoots had a high degree of intraspecific variability (Figures 4A,B). Although heterogeneity of the environment has been shown to be an important component of invasive plant dynamics (Pyšek et al., 2012), genetic diversity of endemic populations, as well as species richness, is also an important variable in evaluating the effects associated with novel interactions in a changing climate.

AMF colonization did not vary significantly between genotypes or respond to synthetic invasion by T. arvense. Dark septate endophytes were found less frequently than either arbuscules or vesicles, but percent of fine roots with DSE present varied in distribution across genotypes and showed a genotype × treatment interaction (Figure 4C). Dark septate endophytes are an enigmatic group of fungi that may play ecological roles similar to mycorrhizal fungi. Additionally, DSE have been shown to be more abundant in stressful environments (Mandyam and Jumpponen, 2005). Mycorrhizal association is generally understood to be a mutualistic symbiosis where plant C is exchanged soil P, N, and water (Smith and Read, 2008), although direct mineral cycling has recently been questioned and further and tight coupling of AMF and decomposers may play a more important role than historically understood (Bunn et al., 2019). Abundance in our experiment, as indicated by the percentage of fine roots with arbuscules present, also varied significant across treatment when controlling for genotype (Figure 4D). Arbuscules are specialized interfaces across which nutrients are exchanged between a plant and fungal partner (Smith and Read, 2008), and therefore critical to understanding how F. thurberi may respond to novel interactions. Future research on high elevation populations of F. thurberi should take into account questions regarding the consistency of mycorrhizal response to non-native plants across the spectrum of life histories as well as whether these responses will be consistent under future climate change scenarios.

Interestingly, we found a significant effect of N. fendleri leachate, but not T. arvense leachate on relative greenness (Figure 3). Treatment with leachate from N. fendleri resulted in elevated relative greenness that was not significantly different from T. arvense but was significantly different from control and F. thurberi treatments. While N. fendleri is a native mustard present at the elevation of the source F. thurberi, T. arvense is a non-native mustard that has become cosmopolitan at lower elevations. We chose to include N. fendleri in this study in order to compare the effects of a locally native Brassicaceae species to that of the encroaching non-native, the latter of which contains phytochemicals known to relax constraints on arbuscular mycorrhizal fungal phylogenetic diversity (Trautwig et al., 2021).

Mycorrhizal fungi have been shown to interact with plant hosts along a mutualism/parasitism continuum (Johnson et al., 1997). No significant effects were found in our study between final dry-weight biomass, root:shoot ratio, or percent of fine root colonization with arbuscules by treatment. However, a significant interaction between genotype and treatment on mortality, general greenness linked to chlorophyll content, and percent of fine root colonized by DSE indicated that evidence for variance among genotypes in the direction of the phenotypic response to the treatments is present (Figures 2, 3, 4D and Table 1). Further research could explore whether both members of the Brassicaceae filter microorganisms (including carbon hungry AMF whose abundance inversely correlates with plant biomass; Lynn et al., 2019) out of a community. In addition, novel microbes associated with T. arvense at lower elevations may disrupt previously existing dynamics in high elevation meadows (Trautwig et al., 2021). A co-culturing experiment that examines prolonged plant-plant and plant-microbe interactions may add additional context to our experiment.

One reason for the interactions we observed may have been that novel microorganism communities associated with T. arvense may be more adapted to disturbance than microorganisms associated with N. fendleri and therefore recovered more quickly following transplanting. Arbuscular mycorrhizal communities have been shown to be highly influenced by mechanical disturbance (Schnoor et al., 2010) and the presence of T. arvense may have selected for more disturbance-tolerant species.