1. Natural grassland ecosystems are increasingly threatened by excessive loadings of nutrients and by the presence of species bred for high productivity. By manipulating grazing regimes and nutrient availability, agricultural practices facilitate the establishment and spread of certain forage plant species outside managed landscapes, challenging local biodiversity. The ecological success of some species in the invaded range sometimes seems to be associated with the symbiosis with foliar fungal endophytes. Symbiotic fungi may increase the competitiveness of host species, but also the resistance to herbivory through the production of toxic secondary compounds such as alkaloids. While progress has been made in understanding how soil nutrients modulate other benefits offered by fungal endophytes to plants (for example, stress tolerance, competitive ability, etc), the consequences for a higher trophic level (i.e. herbivores) and the potential feedbacks on plant invasion have not been explored yet. 2. We explored the relative and interactive importance of soil nitrogen (N) and phosphorus (P) in modulating the interaction of the invasive grass tall fescue -associated with fungal endophytes- and native herbivores in a natural grassland. We hypothesized that N and P nutrients modulate differentially leaf quality traits, namely nutritional value and fungal alkaloid contents, determining the level of damage by native insect herbivores on the exotic tall fescue. 3. We found that only P addition significantly increased native caterpillar density in the field, which corresponded to a concomitant increase in leaf damage. Contrary to expectations, the concentration of the alkaloid ergovaline in leaves was not strongly related to N. It was the level of soil P which dictated the concentration of the element (P) in the leaves and reduced the level of defence against herbivores in this endophyte-symbiotic species. Then, herbivore performance increased, and plants were more prone to be attacked. 4. Synthesis: Our study indicates a strong control of soil P fertility on the tritrophic interaction among plants, fungal endophytes and native herbivores. This highlights the potential role of increased soil nutrients on the invasion spread of endophyte-symbiotic forage plants in natural grasslands.

Experimental design and data collection

Three 50 x 25 m exclosures were established in the field site as part of a long-term experiment to study the effects of cattle exclusion on vegetation dynamics (Longo, Seidler, Garibaldi, Tognetti, & Chaneton, 2013). Despite the fact that the site was never sown with exotic species nor the soil fertilized, tall fescue and other exotics were present at low cover at the time the exclosures were established. From 2004 until now, tall fescue became dominant in the exclosures representing nearly 80% of the absolute cover. This site is part of the Nutrient Network Global Research (http://nutnet.org). In December 2012, we established a long-term experiment within three exclosures (blocks). In each block, four plots of 5 x 5 m were established and randomly assigned to control or fertilized treatments (Nitrogen, Phosphorous, and both). The nutrients were applied in commercially available granules of urea (N) [(NH₂)2CO] and triple superphosphate (P) [Ca(H2PO4)2] (Borer et al., 2014). Each nutrient was added 3 times per year since 2013 (early spring, early summer and early autumn) for a total of 10g m² year¹ to ensure a substantial increase in nutrient availability and removing potential limitations. At the third year of fertilizer addition, P+ and NP+ plots had three times more [P] in soil (31.6 and 27.3 ppm, respectively) than N+ and Control plots (12.3 and 13.66ppm, respectively) (p=0.032). Addition of urea by fertilization did not increase total soil [N] in treated plots (see Fig. S1 and analysis in Supporting Information).

a) Tall fescue leaf nutrient content: In October of the fourth year of the nutrient addition experiment, twenty fully expanded and healthy leaves were randomly selected and harvested per plot, and ground for routine chemical analysis. Leaf N was determined using the Kjeldahl digestion method. Total P was determined by digestion in a mixture of perchloric and nitric acid (Sommers and Nelson, 1972). Phosphorus concentration was determined by molybdenum blue phosphorus method (Murphy & Riley, 1962) in conjugation with UV-visible spectrophotometer (Spectronic 601).

b) Alkaloid content in tall fescue. At the same dates, five fully expanded and healthy leaves were randomly selected and harvested per plot. We made sure not to sample leaves from a same plant by separation each sample by at least 0.5 m one from each other. The plant samples for alkaloid concentration analysis were kept cold in the field and then lyophilized. The concentration of the ergot alkaloid, ergovaline, was evaluated by the following steps. A volume of 2mL of extraction solvent (methanol/water (70:30)) was added to 0.5 g of ground freeze-dried fescue. The samples were homogenized with Bio-Gen PRO200 for 3 min and sonicated for 60 min, then centrifuged for 5 min at 3000 rpm. A volume of 1.5 ml of extract were transferred into glass vials and evaporated dryness at 45 °C under a stream of N2. Samples were resuspended in 4 ml ammonium hydroxide solution (2 M pH 8.5) and extracted with liquid-liquid extraction, chloroform 3 x 3 ml. Then the organic phase were glass vials and evaporated dryness at 30 °C under a stream of N2. Samples were resuspended in methanol/water/ammonium hydroxide (90:9:1) and filtered through a 0.22-mm nylon filter before analysis.

The analyses were conducted in a Thermo scientific ™ system consisting of a degasser, quaternary pump, column oven and an LTQ XL™ ion trap mass spectrometer. Chromatographic separations were performed with a C18 100  2.1 mm Hypersil™ ODS (5μm particle size) column. A solution of ammonium formate (10 mM) and acetonitrile, at the following gradient; 0-2 min, (95:5)-(95:5); 2-5 min, (95:5)-(70:30); 5-7 min, (70:30)-(50:50); 7-10 min, (50:50)-(0:100); 10-12 min, (0:100)-( 0:100); 12-13 min, (0:100)-( 95:5); 20 min (Stop) as the mobile phase. Samples (10 uL) were analyzed at a flow rate of 0.2 mL/min at 45ºC. The selected ion monitoring mode was used in quantification analysis. The mass-spectrometer acquisition settings were: ESI positive, retention time and abundance of the confirmation ion (Ion C) relative to that of quantification ion (Ion Q) were used as identification criteria.

 

c) Endophyte infection: We collected mature seeds from 10 individual plants trying to cover the whole plot and avoiding sampling the same plant twice. We opted for checking the endophyte in seeds instead of vegetative plants, because the distribution of fungal mycelium in leaves is uneven (Christensen, Easton, Simpson, & Tapper, 1998) and it could increase the probability of having false negatives. Endophyte infection was evaluated through the ‘seed squash’ technique (Card, Rolston, Park, Cox, & Hume, 2011). Briefly, the seeds were incubated for at least 8 hours in NaOH (2.5%), then stained with Rose Bengal dye (0.5% in 95% ethanol) and finally, the fungal endophytic hyphae were searched for under a light microscope (Bacon, White, & White Jr., 1994). A previous survey in tall fescue populations that occur in different vegetation communities of the same area, indicated that 100% of established plant were endophyte-infected (Gundel et al., 2009). Here, we started by inspecting five seeds per panicle (i.e. an individual plant). If any of the five seeds were negative or yielded not reliable result, we followed until ten.

d) Herbivore field density and performance. Caterpillar individuals from the native species Paracles vulpina were counted in three 1 x 1m² quadrat within each plot in two consecutive years during October (the moment of peak abundance in the field). In the second year, all the caterpillars within a 1m² per plot were collected and brought to the laboratory where they were immediately weighted. After that, 5 larval individuals of the final instar before pupation, of similar weight and length per plot were selected and placed individually in transparent plastic containers closed with a lid. They were placed in a growth chamber with constant temperature (23°C [±1], photoperiod L16:D8 h) and the caterpillar were fed with fresh leaves of Bromus catharticus, a palatable common native grass. These containers were daily visited to record the number of days from pupation to moth emergence, then, we estimated the pupal developmental rate as days^-1.

e) Tall fescue leaf damage: Measurements were conducted one to two months after caterpillar peak abundance (November-December) in the same two consecutive years. In each plot, 10 well-spaced individual plants of tall fescue were evaluated for damage. The second newest fully expanded leaf of a randomly selected tiller per plant was removed and brought to lab for easier inspection. We differentiated between the different damage types (chewing, mining, vertebrate, etc.) and categorized the level of chewing damage using 5 categories (0%, 1-5%, 6-25%, 26-50%, greater than 50%).

f) Tall fescue cover: In each plot, one 1 x 1 m² quadrat was settled permanently from the beginning of the experiment to estimate aerial cover of tall fescue and other species. Aerial cover was visually estimated to the nearest 5% using a modified Daubenmire method (Tognetti et al. 2010, Borer et al. 2014).

g) Final biomass: In permanently designated subplots within each plot, final biomass was estimated by clipping of all plants and litter rooted within two 0.1-m² strips (10 cm x 100 cm) for a total of 0.2 m². These samples were sorted by functional group (forb, grass, legumes, standing dead biomass) and within grasses, were separated in tall fescue and other grasses. Samples were dried at 60°C to constant mass and weighed to the nearest 0.01 g.

This data package includes the files needed to reproduce the analyses of the following paper: Graff et al. 2020. Protection offered by leaf fungal endophytes to an invasive species against native herbivores depends on soil nutrients. Journal of Ecology