Hemiparasitic plants increase alpine plant richness and evenness but reduce arbuscular mycorrhizal fungal colonization in dominant plant species

Study site

We utilized pre-existing sites along an elevational gradient that spanned from 2,480 to 3,460 m (∼1,000 m) near Gothic, CO, USA (Read, Henning & Sanders, 2017, Table 1). The sites are located on USDA Forest Service land and are covered under Forest Service Special Use Permit #GUN1120. The gradient site receives about 439–668 mm/year of precipitation and has a mean annual temperature of −1.6 to 1.5 °C (Hijmans et al., 2005; Read, Henning & Sanders, 2017; Table 1). Additionally, nutrient availability shifts across the gradient, with high elevation sites having more available phosphorus but lower available nitrogen relative to low elevation sites (Read, Henning & Sanders, 2017). Plant diversity is highest at middle elevation sites, while plant community composition transitions from a sagebrush steppe at low elevation to montane meadows at high elevation (Read, Henning & Sanders, 2017).

Plant community sampling

At each elevation, we identified all Castilleja present within a 20 × 20 m area. Next, we haphazardly-selected 10 Castilleja individuals of similar size and identified an adjacent “Castilleja free” area that was located within two m from our focal Castilleja, but contained no Castilleja within a 1.5 m diameter around the centroid. Thus, we sampled 100 total plots (10 plots ×2 Castilleja treatments × 5 elevations). From our sampling design, we were unable to determine if plots were truly absent of Castilleja influence, as shallow-rooted Castilleja can infect across a semi-broad spatial area (Ducharme & Ehleringer, 1996). Logistical constraints of the pre-existing elevational gradient and the observational focus of our study prevented us from removing Castilleja individuals, which would have provided a more direct measurement of Castilleja effects on plant community composition and ecosystem function (Reed, 2012).

To measure plant community composition, we identified all plant species present within a 0.5 × 0.5 m quadrat in each plot. Plant community composition was measured in the Castilleja present plots, but placing the focal Castilleja in the center of our quadrat and Castilleja-free surveys were conducted by placing the quadrat in the center of the Castilleja-free area to maximize the distance from all surrounding Castilleja plants. Next, we visually estimated the cover of each species present in the quadrat. We estimated percent coverage to the nearest 1% for species <20% and the nearest 5% for species >20% coverage. We also estimated the amount of bare ground and rock within each plot.

Root sampling

From our cover data, we identified the most abundant plant species that occurred in all 20 plots at each elevation to sample for fungal colonization. At one of the five sites (2,480 m), our sampled plant species (Balsamorhiza sagittata) was also the dominant taxa, however, at four of the five sites, we sampled either the second or third most abundant taxa, as the dominant taxa did not occur in all 20 plots. We collected a single 2.5 × 15 cm core from our focal individual to measure the colonization of AMF and DSE. A single focal plant (Table 1) was haphazardly selected near the center of each plot (10 plots × 2 Castilleja treatments × 5 elevations = 100 total soil cores). We transported the soil cores back to the Rocky Mountain Biological Laboratory on ice and stored them at 4 °C in lab until being processed within 24 h. Next, we extracted live roots from the core using a 0.5 mm mesh sieve and placed roots in Fisher brand Histosette II tissue cassettes (Catalog #1000965, Thermo Fisher Scientific, Waltham, MA, USA). Tissue cassettes were then placed in deionized water to remove any remaining soil. Next, we cleared roots in a 10% potassium hydroxide solution, acidified root samples in a 2% hydrochloric acid solution, and then stained roots in a 0.01% trypan blue solution (Koske & Gemma, 1989). We then mounted roots on microscope slides using polyvinyl lactic acid glycerol glue (International culture collection of (Vesicular) Arbuscular Mycorrhizal fungi (INVAM), 2017) and slides were oven dried at 50 °C for 48 h. We removed two root samples (Castilleja present plant five and eight) from our 3,200 m elevation site because of low root biomass. Next, we quantified the presence of AMF hyphae, DSE hyphae, as well as AMF vesicles, arbuscules, coils, and spores, DSE microsclerotia, and any potential pathogens, of at least 50 root intersections per slide, using the magnified grid-line intercept method (McGonigle et al., 1990). Finally, we calculated percent colonization of AMF, DSE, and any structures and potential pathogens by dividing the number of positive observations by the total number of root intersections observed.

Statistical analysis

All analyses were conducted in R (R Development Core Team, 2008) and RStudio (RStudio Team, 2015), with packages cited within. To test for differences in plant richness, evenness, and community composition independent of the contribution of Castilleja we removed the Castilleja cover from all analyses. However, Castilleja cover is located in Table 1. We calculated species richness using the specnumber function in the vegan package (Oksanen et al., 2017) and species evenness as the probability of interspecific encounter (PIE, Simpson’s Evenness) (Hurlbert, 1971) as: $\text{PIE}=N/\left(N-1\right)\times \left(1-\sum p_i{}^2\right)$, where N = total number of individuals, and p_i = is the relative abundance of species i. Within each plot using the calcPIE function in the mobr package (McGlinn et al., 2018). We chose to calculate PIE as our evenness measurement because of its independence from sample size when comparing across elevation sites (Chase & Knight, 2013). We tested for differences in plant richness and evenness between plots associated with the presence of Castilleja across sites by constructing linear mixed-effect models with elevational site, Castilleja presence (absent or present), and their interaction as fixed factors using the nlme package (Pinheiro et al., 2014). Within each mixed model, we allowed the intercept to vary by plot pairings (random effect). Next, we constructed mixed effect models with and without fixed factors (elevation and Castilleja presence) and compared AIC scores to determine if adding each fixed factor improved model fit. Next, we calculated the deviance and compared the inclusion of each factor with a likelihood ratio test using the Anova function (car package, Fox & Weisberg, 2011).

To test whether the presence of Castilleja was associated with changes in plant community composition, we performed a PERMANOVA using the adonis function in the vegan package (Oksanen et al., 2017) with Castilleja presence, elevation, and plot pairing as predictors. To separate the effects of Castilleja presence on abundance shifts among dominant taxa vs changes in taxa gains and losses, we performed PERMANOVAs using abundance-weighted Bray–Curtis distances as well as presence-absence data using Jaccard’s distance. During our initial PERMANOVA fitting across all sites (Table S2), we observed significant interaction terms within both elevation × Castilleja presence (F_(1,99) = 2.532, p = 0.01) and elevation × plot pairing (F_(1,99) = 13.48, p = 0.01), which suggested that effect of Castilleja presence on plant community composition differed by site. To explore when and where Castilleja was associated with plant community composition, we conducted separate PERMANOVAs for each elevation. Within our site level PERMOANOVAs, we included Castilleja presence, plot pairing, and their interaction as predictor variables.

To visualize the results of our PERMANOVAs, we performed non-metric multidimensional scaling based on Bray–Curtis distances using the metaMDS function in the vegan package (Oksanen et al., 2017) for each elevational site.

Finally, we compared fungal colonization patterns within the dominant plant species with mixed effect models as described above, with Castilleja presence/absence, elevational site and the interaction between Castilleja presence and elevation as predictor variables. For fungal mixed models, we constructed mycorrhizal (AMF + ERM) and DSE models separately. Structures of AMF and DSE (vesicles, arbuscules, coils, spores, microsclerotia) were extremely rare within our samples, therefore we present only AMF and DSE hyphal colonization data.

Castilleja association with plant richness and evenness

Overall, we found that the presence of Castilleja and elevation were retained within our best models to predict plant richness and evenness (Table S1). Taken as a whole, we found that the presence of Castilleja was associated with an increase in plant richness of 11% (χ² = 52.360, p < 0.0001, Table 2), and an increase in plant evenness of 5% (χ² = 6.805, p = 0.009, Table 2). As expected, elevation significantly impacted plant richness and evenness, with the highest plant species richness and evenness values observed at the middle elevation sites (3,200 m: richness μ = 11.8, evenness μ = 0.835; 3,392 m: richness μ = 11.2, evenness μ = 8.46) (Table 2; Fig. 1). Surprisingly, we observed no significant interaction terms between Castilleja presence and plant richness or evenness (Table 2).

Castilleja association with plant community composition changes

Overall, we found that elevation was the best predictor of plant community composition change, accounting for 30% of the data variation in community composition (Table S2). Using abundance-weighted measures, Castilleja presence was associated with a significant (p = 0.01), but weak effect (R² = 0.021) on overall plant community composition (Table S2). Site-level heterogeneity (plot pairings) explained an additional 9% of data variation across plant community composition.

At the site level, we found that the presence of Castilleja was associated with plant composition shifts in abundance-weighted Bray–Curtis distance at two (2,740 and 3,200 m) of the five elevation sites (Table 3; Fig. 2). At 2,740 m, we observed higher abundance of Adenolinum lewisii, Phacelia sericea, Antennaria rosea and reduced abundance of shrub species Artemisia tridentate and Symphoricarpos rotundifolia, and forb Delphinium nuttallianum when Castilleja was present (Fig. 2). At 3,200 m, we observed reduced cover of invasive grass Bromopsis inermis, when Castilleja was present (Fig. 2). At three of the five sites (2,480, 3,392, and 3,460 m), we observed a significant relationship of plot pairing, suggesting that spatial heterogeneity in plant community composition may cloud our ability to observe the effect of Castilleja on community composition. Unlike our abundance-weighted results, we found no effect of Castilleja and only weak effects of plot pairings on plant community composition when sites were compared using presence-absence based Jaccard’s index (Table S3).

Castilleja association with mycorrhizal and dark-septate endophyte colonization

Overall, we found that Castilleja presence was associated with reduced mycorrhizal colonization at each site by 20% (Table 2; Fig. 3) and was the only factor retained within our best fit model (Table S1). At our four lowest elevation sites Castilleja presence was correlated with a reduction in AMF colonization in Balsamorhiza sagittata (2,480 m), Chrysothamnus viscidiflorus (2,740 m), Viola adunca (3,200 m), Ligusticum porteri (3,392 m), and at our highest (3,460 m) elevation site, Castilleja presence was correlated to a reduction in ericoid mycorrhizae colonization within Arctostaphylos uva-ursi. We found that climate contexts had no impact on Castilleja effects, as elevation was not retained within our best-fit model for mycorrhizal colonization (Table S1). Surprisingly, we found that Castilleja presence was not related to DSE colonization and only elevation was retained within the best fit model predicting DSE colonization. Overall, we found that DSE colonization was highest (μ = 51.9%) at our high elevation site (3,460 m), and colonization levels were similar at the four other elevations, ranging from 27% to 33% (Table 2; Fig 3).

Castilleja increases plant richness and evenness

Overall, we found the presence of Castilleja was associated with an 11% increase in plant richness and a 5% increase in plant evenness. Our results are in line with several studies that have found increased plant richness and evenness with the presence hemiparasites (Davies et al., 1997; Press & Phoenix, 2005; Bardgett et al., 2006; Westbury et al., 2006; Reed, 2012), although this result can be contingent on hemiparasite identity and the identity of the surrounding plant community (Press & Phoenix, 2005; Westbury & Dunnett, 2007). Interestingly, we found the effect of Castilleja associated with plant richness and evenness did not differ by elevation, suggesting in this system Castilleja effects may be consistent despite the shifting biotic and abiotic contexts along the gradient. Plant richness and evenness shifted across the elevational gradient, displaying the characteristic unimodal pattern with highest richness and evenness at middle elevations. Additionally, the species of Castilleja present differed by elevation, with C. angustifolia present at 2,480 and 2,740 m, C. miniata at 3,392 m, and C. sulphurea at 3,200, 3,392, and 3,460 m.

Castilleja has inconsistent effects on plant community composition

The presence of Castilleja was associated with a significant, but weak, effect on overall plant community composition across the elevational gradient. However, site-level analyses revealed that the effect of Castilleja on abundance-weighted plant community composition differed by site. For example, at two of our middle elevation sites (2,740 and 3,200 m), the presence of Castilleja was the strongest predictor of plant community composition. The presence of Castilleja was correlated with a reduction in aboveground cover of shrub species Artemisia tridentata and S. rotundifolia at 2,740 m and invasive grass Bromopsis inermis at 3,200 m. The reduction of shrub and grass taxa was matched with an increase in low abundant forb taxa: Adenolinum lewisii, P. sericea, and Antennaria rosea. However, it is important to note the nature of our study did not allow us determine which hosts Castilleja were parasitizing, how much carbon was being parasitized from hosts, or the long-term dynamics of Castilleja in an ecosystem comprised mostly of long-lived perennial plant taxa. In a previous study by Ducharme & Ehleringer (1996), authors found Castilleja received ∼40% of its carbon from critical host Artemisia tridentata. However, Castilleja had no effect on the photosynthetic rates or water potentials of Artemisia over a 2-year period. Thus, it is unclear how detrimental Castilleja is for host production and survival (Ducharme & Ehleringer, 1996; but see Reid, Yan & Fittler, 1994; Bowie & Ward, 2004).

The association between Castilleja presence on plant community composition was only apparent when we compared plant community composition using abundance-weighted Bray–Curtis distance metrics. When we compared plant community composition using presence/absence data we found no association of Castilleja presence. This suggests that the association of Castilleja is primarily driven by abundance changes among dominant plant taxa and less by taxa gains and losses. Taken together, our results align with previous studies demonstrating that few hemiparasites reduce host resources enough to cause mortality, but reduce host growth rates which allow the establishment of subordinate taxa (Reid, Yan & Fittler, 1994; Watson, 2009).

Although our study is not a direct test of the effect of Castilleja on plant community composition, our results align with previous direct, manipulative hemiparasite removal and addition experiments that report a strong negative effect of hemiparasites on dominant taxa abundance with an overall increase in plant diversity, especially with forb taxa (Pennings & Callaway, 1996; Davies et al., 1997; Press, 1998; Press & Phoenix, 2005; Bardgett et al., 2006; Těšitel et al., 2015, 2018). For instance, sowing of hemiparasite Rhinanthus into a Calamagrostis invaded grassland reduced Calamagrostis cover from ∼45% to ∼2% within two growing seasons (Těšitel et al., 2018). Overall, plant diversity increased to compensate for the reduction in Calamagrostis biomass and cover, although authors found that Rhinanthus had inconsistent effects on overall plant community composition. Taken together, this suggests that hemiparasite effects on plant community composition are likely driven by the identity of the surrounding plant species, competitive ability of surrounding plant taxa, and abundance levels of surrounding plant taxa (Press & Phoenix, 2005; Watson, 2009; Těšitel et al., 2018). To confirm previous findings, a direct test manipulating both plant community composition and the presence of hemiparasites needs to be performed.

Site-level spatial heterogeneity was a strong driver of plant community composition

At three of the five sites (2,480, 3,392, 3,460 m), plot pairing was the strongest predictor of plant community composition explaining ∼13%, 18%, and 13%, respectively, of the data variation in community composition. At each of these three sites, the presence of Castilleja had no effect on plant community composition, suggesting that site-level spatial heterogeneity obscures our ability to detect changes in community composition mediated by hemiparasites. This was surprising because we observed a similar increase in plant richness and plant evenness between plot pairings at all five sites. The overall richness and evenness effects, but limited compositional effects likely reflects the generalist hemiparasite life history of Castilleja (Ducharme & Ehleringer, 1996; Adler, 2003; Spasojevic & Suding, 2012). Additionally, the effect of plot pairing was stronger using abundance-mediated measures of community composition (Bray–Curtis distance), compared to presence-absence weighted-metrics (Jaccard’s distance). With the exception of our low elevation site, we observed no effect of plot pairing on plant community composition using Jaccard’s distance metrics. This suggests that at each site, differences in community composition were driven by abundance differences in taxa plot to plot, and less by taxa turnover among plots.

Hemiparasitic plants shape abundant plant interactions with mycorrhizal fungi

Overall, we found that the presence of Castilleja was associated with a reduction in AMF colonization by 18% and ERM by 17%. However, we found no effect of Castilleja presence on DSE colonization within focal taxa roots. The presence of Castilleja reduced mycorrhizal colonization similarly across elevation sites even though the overall rate of colonization differed by elevation, suggesting a ubiquitous pattern independent of biotic and abiotic contexts, counter to our expectations. However, our results provide an indirect test of the effect of Castilleja on mycorrhizal colonization because we did not confirm if sampled plants were actively parasitized by Castilleja. Additionally, we were unable to determine the origin and identity of every single root we quantified for colonization. However, even with the indirect measure and the uncertainty of host identity, we still observed a consistent reduction in mycorrhizal colonization when Castilleja was present.

Our results align with previous studies that observed a reduction in mycorrhizal fungal colonization with hemiparasite colonization (Gehring & Whitham, 1992; Davies & Graves, 1998; reviewed in Press & Phoenix, 2005). One potential explanation for the reduction in mycorrhizal colonization is that hemiparasites outcompete mycorrhizal fungi for host carbon resources (Gehring & Whitham, 1992; Davies & Graves, 1998; reviewed in Press & Phoenix, 2005). Because AMF have a higher reliance on host-derived carbon resources relative to DSE and ERM, competition with hemiparasites for host carbon may explain why we observed reduced AMF colonization, but no difference in DSE colonization (Caldwell, Jumpponen & Trappe, 2000; Jumpponen, 2001; Usuki & Narisawa, 2007; Knapp et al., 2018). Although at our high elevation site, we observed reduced ERM colonization in Arctostaphylos uva-ursi even though ERM produce a suite of carbon-degrading enzymes making them less reliant on plant-derived carbon (Read, Leake & Perez-Moreno, 2004; Smith & Read, 2008; Averill, Turner & Finzi, 2014). Future studies should directly test whether hemiparasites reduce carbon allocation to mycorrhizal fungi and whether the differences in the reliance on host carbon resources will determine the response of fungal colonization to hemiparasite presence.

Hemiparasites also directly associate with mycorrhizal fungi and DSE (Bouwmeester et al., 2007; Li & Guan, 2008; Stein et al., 2009; Li et al., 2012). Associations with mycorrhizal fungi and DSE may provide an additional mechanism for hemiparasites to access soil and plant-derived resources (Li & Guan, 2008), so it is unclear whether a reduction fungal colonization in dominant plant species would be detrimental to hemiparasites. In a previous field census, we found both AMF and DSE actively colonizing the roots of C. angustifolia, C. miniate, and C. sulphurea across this elevational gradient (J. Henning, 2013, 2015, unpublished data). However, we did not measure fungal colonization rates of Castilleja in this study. Our results highlight an interesting pattern, however further study is required to determine if interactions among hemiparasites, mycorrhizal fungi, and plant hosts could cascade to influence plant community composition and ecosystem function.

Hemiparasite effects on plant diversity, plant community composition, and mycorrhizal colonization are independent of climate contexts

We sought to explore how climate contexts shape the impact of hemiparasites on plant richness, evenness, community composition, and the relationship of dominant plant species with root symbionts. Although we found that the presence of Castilleja or elevation were always retained in the best fit models for plant richness, evenness, mycorrhizal colonization, and DSE, we did not observe any significant interactions between the presence of Castilleja and elevation for any response variable. This suggests that hemiparasites increase plant richness, increase plant evenness and reduce mycorrhizal fungal colonization within dominant species independent of biotic and abiotic contexts. Our results are counter to several studies, which have found hemiparasite effects are contingent on-site properties like: plant composition; nutrient availability, soil moisture, mycorrhizal fungi present, plant productivity (Callaway & Pennings, 1998; Matthies & Egli, 1999; Stein et al., 2009; Těšitel et al., 2015, 2018). Thus, across a wide array of ecosystems, hemiparasites may be critical for the maintenance of plant diversity and the regulation of competitively dominant plant taxa.