Metabolite analysis of the effects of elevated CO 2 and nitrogen ferti l ization on the association between tall fescue ( Schedonorus arundinaceus ) and its fungal symbiont

Atmospheric CO2 is expected to increase to between 550 ppm and 1000 ppm in the next century. CO2-induced changes in plant physiology can have ecosystem-wide implications and may alter plant-plant, plant-herbivore and plant-symbiont interactions. We examined the effects of three concentrations of CO2 (390ppm, 800ppm and 1000ppm) and two concentrations of nitrogen fertilizer (0.004gN/week vs. 0.2gN/week) on the physiological response of Neotyphodium fungal endophyte-infected and uninfected tall fescue plants. We used quantitative PCR to estimate the concentration of endophyte under altered CO2 and N conditions. We found that elevated CO2 increased the concentration of water-soluble carbohydrates and decreased the concentration of plant total amino acids in plants. Fungal-derived alkaloids decreased in response to elevated CO2 and increased in response to nitrogen fertilization. Endophyte concentration (expressed as the number of copies of an endophyte-specific gene per total genomic DNA) increased under elevated CO2 and nitrogen fertilization. The correlation between endophyte concentration and alkaloid production observed at ambient conditions was not observed under elevated CO2. These results suggest that nutrient exchange dynamics important for maintaining the symbiotic relationship between fungal endophytes and their grass hosts may be altered by changes in environmental variables such as CO2 and nitrogen fertilization. The effects of elevated CO2 on fungal endophytes Ryan et al. 2 INTRODUCTION Atmospheric CO2 is expected to rise from a current ambient concentration of 390ppm to between 550 and 1000ppm by the year 2100 (IPCC 2007). The effects of elevated CO2 on plant growth, physiology, metabolism, morphology and competition have been well studied (Bazazz 1990; Robinson, Ryan & Newman 2012). CO2 enrichment has been shown to increase plant photosynthetic rates resulting in increased growth and biomass production and physiological changes brought about by altered resource allocation. A commonly observed CO2-induced change is an increase in carbohydrate metabolites and a decrease in nitrogen metabolites such as amino acids and soluble protein, resulting in an increased plant C:N ratio (Robinson et al. 2012). Such changes in resource allocation are also expected to alter plant defensive chemistry. The carbonnutrient balance (CNB) hypothesis has often been employed as a model of allocation to defensive compounds under elevated CO2 and predicts that carbon products in excess of those needed for primary metabolic functions will result in increased carbonbased secondary metabolites and subsequent decreased N-based secondary metabolites (Bryant, Chapin & Klein 1983). While there is some evidence that supports this outcome on average (Robinson et al. 2012) generalizations concerning these changes have been difficult to formulate and may be subject to complex regulation by phytohormones (Ryan, Rasmussen & Newman 2010; DeLucia et al., 2012). Such alterations in plant primary and secondary metabolism are expected to have bottomup trophic effects that are likely to affect broader community interactions. While these effects have been relatively well studied in the case of herbivores and their host plants, very little work exists on the broader foodweb impacts of elevated CO2. The perennial grass Schedonorus arundinaceus (Schreb; tall fescue) is an agricultural pasture grass that is commonly associated with the mutualistic clavicipitaceous fungus Neotyphodium coenophialum (Glenn, Bacon, & Hanlin). Neotyphodium endophytes are asexual, vertically transmitted fungi whose hyphae grow in the intercellular spaces of host grasses (Christensen et al. 2008). Grasses themselves are relatively poor producers of defensive compounds and the presence of mutualistic Neotyphodium fungi has been shown to directly increase host plant resistance to both vertebrate and invertebrate herbivores through the production of toxic alkaloids (Clay 1988). The presence of endophytic fungi can alter the outcome of competitive interactions between plants by providing increased herbivore resistance, drought tolerance, disease resistance, nutrient stress tolerance and reproductive vigour in host plants (Kuldau & Bacon 2008; Cheplick & Faeth 2009). As such, endophytes have a critical role in shaping community composition. Recent studies have suggested that the degree of mutual benefit for endophytic fungi and their hosts may be conditional on environmental factors such as nutrient availability (Saikkonen et al. 2006). Heterotrophic microorganisms such as fungal The effects of elevated CO2 on fungal endophytes Ryan et al. 3 endophytes and mycorrhizae depend on carbon and energy provided by their autotrophic host plants. It is not entirely clear which organism controls such interactions and to what extent abiotic mechanisms contribute to these dynamics. Though these mechanisms have yet to be elucidated, nutrient exchange between plant and fungal partners seems to play an important role (Kiers & Denison 2008; Ryan et al. 2008; Draper, Rasmussen & Zubair 2011). Thus, changes in environmental variables that can alter plant resource allocation may have implications for the way in which symbiotic partners interact. Both elevated CO2 and nitrogen fertilization can alter plant nutrient allocation and may affect nutrient exchange in ways that impact the production of Nbased mycotoxins and carbon energy sources for endophytic growth. Few studies have examined the interaction between fungal endophytes and host plants under elevated CO2 (but see: Marks & Clay 1990; Groppe et al. 1999; Newman et al. 2003; Hunt et al. 2005). Hunt et al. (2005) found that N. lolii-derived alkaloids increased in perennial ryegrass grown under elevated CO2, but only when plants were well fertilized. CO2induced increases in defensive compounds are likely to impact herbivores in a high CO2 atmosphere. Studies have also shown that endophyte infection can buffer CO2induced decreases in plant protein production in both tall fescue (Newman et al. 2003) and perennial ryegrass (Hunt et al. 2005), which may affect herbage quality for herbivores beyond changes in defensive chemistry. Changes in fungal growth in response to altered environmental variables may have implications for the strength and direction of plant-endophyte interactions, particularly in cases where nutrient dynamics are altered. The growth responses of endophytes to changes in plant physiology have been difficult to study due to the fact that endophyte and plant tissues cannot be physically separated. As such, most studies in the endophyte literature report only fungal presence/absence data in host grasses. However, quantitative PCR (qPCR) provides a promising technique for estimating endophyte concentration by quantifying the number of copies of endophyte-specific genes found in the total (plant + fungus) genomic DNA (Rasmussen et al. 2007; Liu et al. 2011). In this study we examine the effects of elevated CO2, nitrogen fertilization and endophyte status on plant physiological parameters associated with carbon and nitrogen metabolism. In endophyte-infected plants, we examine the effects of CO2, nitrogen fertilization and their interaction on endophyte-derived alkaloids. We use qPCR to examine N and CO2 effects on the concentration of N. coenophialum in plant tissues. To our knowledge this is the first time this technique has been used to examine in planta fungal growth of N. coenophialum in response to altered atmospheric CO2 concentration. The effects of elevated CO2 on fungal endophytes Ryan et al. 4 MATERIALS AND METHODS CO2-controlled chambers Nine plexiglass chambers housed in a greenhouse in a 3 x 3 square pattern were used to control CO2. Chambers were constructed and operated according to Grodzinski et al. (1999). An Argus Greenhouse Control System was used to control and monitor the concentration of CO2, relative humidity and temperature. We used three CO2 concentrations (ambient (390 ppm), 800 ppm and 1000 ppm) with three replicates per concentration, which were blocked according to a light gradient within the greenhouse (one of each CO2 concentration per block, arranged as a latin squares design). These elevated CO2 concentrations are within the range of the projected increase by the year 2100. Chambers were maintained at a relative humidity of ~40% and a temperature of 23oC. The light regime was set to an 18:6 L/D cycle. Plant material Experimental S. arundinaceus (tall fescue) plants were taken from a group of longstanding nursery plants where each plant was the product of a single seed and thus represented a single genotype. Georgia 5 seed lines were originally obtained from Donald Wood (University of Georgia, USA). Seed stock was originally infected with the common strain (CS) of N. coenophialum fungal endophyte. Following one year of seed storage at room temperature, some seeds lost the endophyte infection. Thus, all plant genotypes used were compatible with endophyte infection. Six E+ genotypes and six Egenotypes were randomly selected from plants grown from these seed stocks for the experiment. Prior to the experiment, endophyte infection status was confirmed for each genotype using the immunoblot method (Agrinostics Ltd.). Two tillers were removed from the appropriate parent plants and were cut back to 5-6 cm (aboveand below-ground tissue) and repotted in a mixture of 50% Sunshine Mix #2 (Sun Gro Horticulture) and 50% vermiculite (2A; Thermorock). Pots were constructed from 0.635 cm thick PVC piping cut to 9 cm x 38 cm (D x H). Each genotype (6 Eand 6 E+) was replicated twice per chamber (one set of replicate genotypes per level of nitrogen fertilization) for a total of 216 pots. To avoid unwanted insect colonization prior to the beginning of the experiment, plants were sprayed with insecticidal soap after placement in the CO2-controlled chambers. Plants were watered as necessary with de-ionized water. Plants were fertilized weekly with either a low N (0.004 gN/week) or high N (0.2 gN/week) treatment, delivered as 40 ml of a 0.3 g/L NH4NO3 or 14.7 g/L NH4NO3 solution respectively. All plants received 40 ml of a micronutrient solution weekly (nitrogen-free Long Ashton solution; Ponder et al. 2000). Plants were cut back to 10 cm above soil level weekly over the whole experimental period and the cuttings were dried and weighed. In order to test the resulting plant nitrogen status, we measured %N in blade tissues of aboveThe effects of elevated CO2 on fungal endophytes Ryan et al. 5 ground material at two time points, pooled across endophyte and CO2. Low N plants contained 2.69 ± 0.12 %N, while high N plants contained 4.24 ± 0.10% N; these values reasonably approximate N-limited and N-rich pastures respectively (Parsons, Harvey & Woledge 1991). Plants were grown in treatment conditions for 6 months, at which point two 4th or 5th instar Rhopalosiphum padi aphids were released onto each plant. Details of aphid population dynamics and phloem biochemistry will be published in a separate manuscript. Alkaloid extraction and analysis All alkaloid analyses were performed on above-ground plant tissues (blade and sheath tissues combined). Ergovaline extraction was performed using a method modified from Spiering et al. (2002). A sample of lyophilized grass tissue (50 mg) was extracted for one hour with 1ml of extraction solvent (50% isopropanol, 1% lactic acid with 1.16 ng/ml ergotamine tartrate as internal standard). The sample was then centrifuged (8000 g, 5 min) and a 500 μl aliquot of the supernatant transferred to an HPLC vial for analysis. Extraction and analysis of lolitrem-B was performed using a method modified from Spiering et al. (2005). Samples of lyophilized grass tissue (50 mg) were extracted for one hour with 1 ml of the extraction solvent (9:1 dichloroethane/methanol). Samples were centrifuged for 5 min at 8000 g and 500 μl aliquots of the supernatant were transferred to HPLC vials for analysis. Loline alkaloids (N-acetyl loline, N-formyl loline and N-acetyl norloline) were measured using gas chromatography (GC) from a method modified from Kennedy & Bush (1983) and Yates, Fenster & Bartelt (1989). A sample of lyophilized grass tissue (100 mg) was extracted for 1 hour with 50 μl of 40% methanol/5% ammonia and 1ml of 1,2-dichloroethane (containing 53.7 ng/ml 4-phenylmorpholine as internal standard). The tubes were then centrifuged for 5 min at 8000 g, and the supernatant transferred to a glass GC vial via a 10 μm filter for analysis. The analysis was conducted on a GCflame ionization detector (GC-FID) equipped with a ZB-5 capillary column (30m x 0.32mm x 0.25μm film; Phenomenex, Torrance, CA, USA). Peramine was extracted using a method modified from Spiering et al. (2002). A sample of lyophilized grass tissue (50 mg) was extracted for one hour with 1ml of extraction solvent (50% methanol with 2.064 ng/ml homoperamine nitrate as internal standard). The sample was then centrifuged (8000 g, 5 min) and a 500 μl aliquot of the supernatant transferred to an HPLC vial for analysis. Endophyte concentration The concentration of N. coenophialum endophyte in plant tissue was estimated based on qPCR of genomic DNA (gDNA) isolated from infected plants using the Roche The effects of elevated CO2 on fungal endophytes Ryan et al. 6 LightCycler® 480 system (Roche Diagnostics). DNA was extracted from 25 mg of freezedried powdered plant material (pseudostem + blade) using DNeasy® Plant Mini kit (Qiagen, Bio-Strategy Ltd., Auckland, New Zealand) following the manufacturer’s handbook. Primers (forward: 5’-cacgtactgactgaagcgtagc-3’; reverse: 5’caatgcagcgagtgaacatc-3’) suitable for qPCR were designed for a fragment of a N. coenophialum-specific gene, encoding the translation elongation factor 1-alpha (Takach et al. 2012; GenBank Acc. # JX028264). The gel-purified PCR products were cloned into pCR®2.1-TOPO® vectors (Invitrogen NZ Ltd., Auckland, New Zealand) and transformed into One Shot® Escherichia coli cells by chemical transformation. Single colonies were grown in liquid culture and plasmids extracted and purified using a QIAprep®Miniprep kit (Qiagen, Bio-strategy Ltd., Auckland, New Zealand). A dilution range of the plasmids from 2 ×101 to 2 × 106 copies was used to generate a standard curve. Three nanograms of gDNA isolated from infected and uninfected plant material were mixed with LightCycler 480 SYBR Green I Master Mix (Roche Diagnostics), primer pairs and water, for a total volume of 15 μl per qPCR reaction. The PCR protocol was as follows: pre-incubation: one cycle of 95°C for 5 min; amplification: 45 cycles of 95°C for 10 s, 64°C for 10 s and 72°C for 10 s; the formation of single PCR products in the assays was confirmed by melting analysis: 95°C for 5 sec, 65°C for 1 min and a continuous melting curve fluorescent acquisition at 97°C; cooling: 40°C for 30 sec. The concentration of endophyte present in the grass is expressed as the number of copies of fungal-specific gene per ng total (plant + fungal) genomic DNA. Water-soluble carbohydrates, protein and amino acids Low molecular weight (LMW) carbohydrates (mostly glucose, fructose and sucrose) and high molecular weight (HMW) carbohydrates (mainly fructans) were extracted and quantified from above-ground tissues as previously described (Hunt et al. 2005; Rasmussen et al. 2007). Briefly, 25 mg of dried tissue (blade + sheath) were extracted with 2 × 1 ml of 80% ethanol (LMW sugars) and subsequently with 2 × 1 ml water (HMW sugars) for 30 min at 60° C. Extracts were centrifuged, and supernatants of each fraction were analyzed separately using anthrone as a colorimetric reagent (Jermyn 1956). Standard calibration curves were generated using sucrose and inulin for LMW and HMW sugars respectively. Absorbance was measured at 620 nm. Total free amino acids were analyzed from the ethanol fraction used for carbohydrate extraction and were determined colorimetrically with ninhydrin, as described previously (Yemm 1955). L-glutamine was used as a standard and absorbance was measured at 570 nm. Soluble proteins were extracted using 0.1% mercaptoethanol in 100 mM potassium phosphate buffer. Extracts were centrifuged and the resulting supernatant The effects of elevated CO2 on fungal endophytes Ryan et al. 7 was analyzed according to Bradford (1976) with absorbance measured at 595 nm using Bovine Serum Albumin (BSA) as a standard. Statistics We used a blocked split-plot experimental design with CO2 as the whole-plot factor and nitrogen and endophyte as sub-plot factors. All analyzes were performed using a mixed effects ANOVA model with CO2, nitrogen and endophyte as fixed factors and plant genotype as a random factor. We also included a block term but as we had no a priori hypothesis about block x treatment interactions, we excluded these interactions from our model and collapsed these terms into the residuals (Newman, Bergelson & Grafen 1997). Table 1 shows the degrees of freedom for the analysis. For endophyte and alkaloid concentrations, we used a reduced version of the model that excluded “endophyte” as a factor, since these are only found in endophyte-infected plants. All analyses were conducted using JMP statistical software version 8. A Box-Cox transformation was used to homogenize the residual variance and an examination of the residuals following transformations suggested that the assumptions of the ANOVA were met. The untransformed means, and standard error of the means, are reported as measures of response magnitude and data dispersion. RESULTS Water-soluble carbohydrates LMW-WSC concentration increased under elevated CO2 (F2,4 = 49.76, p < 0.0001) and decreased in the high nitrogen treatment (F1,142 = 10.08, p < 0.01). However there was a significant CO2 x N interaction (F2,142 = 10.23, p < 0.0001; Fig. 1a), whereby the increase in LMW concentration under elevated CO2 was observed only in the high N treatment. The effect of CO2 on LMW concentration was also dependent on endophyte (CO2 x endophyte interaction: F2,142 = 8.15, p < 0.01; Fig. 1b) and the magnitude of the CO2-induced increase in LMWs was higher in endophyte-infected plants. The concentration of HMW-WSCs also increased under elevated CO2 (F2,4 = 43.23, p < 0.0001) and decreased under high N (F1,142 = 80.05, p < 0.0001). Again the effects of elevated CO2 were dependent on a CO2 x N interaction (F2,142 = 7.63, p < 0.001; Fig. 1c) but unlike LMW response, the CO2-induced increase in HMWs was observed only in the low N condition. There was no significant effect of endophyte infection on HMW WSCs. Nitrogen metabolites Total soluble protein increased under high N fertilization (F1,142 = 59.45, p < 0.0001; Fig. 2a). Total amino acids also increased under high N fertilization (F1,142 = 441.93, p < 0.0001). However, there was a significant interaction between CO2 and N fertilization The effects of elevated CO2 on fungal endophytes Ryan et al. 8 (F2,142 = 6.52, p < 0.01; Fig. 2b) whereby amino acids decreased under elevated CO2, but only in the low N condition. There was also a significant N x endophyte interaction (F2,142 = 12.78, p < 0.01; Fig. 2c); under high N conditions, E+ plants had a greater concentration of amino acids than E-, but this was not the case under low N conditions. Endophyte and alkaloid concentrations Total lolines (F1,70 = 78.56, p < 0.001; Fig. 3a), ergovaline (F1,70 = 188.60, p < 0.0001; Fig. 3c) and peramine (F1,70 = 200.10, p < 0.0001; Fig. 3e) all increased with high N fertilization. Total lolines decreased significantly under elevated CO2 (F2,4 = 14.67, p = 0.001; Fig. 3b), while there was a marginally significant (p < 0.1) decrease in ergovaline under elevated CO2 (F2,4 = 3.41, p < 0.1; Fig. 3d). The concentration of endophyte (expressed as the number of copies of an endophyte-specific gene per ng of total genomic DNA) increased under elevated relative to ambient CO2 (F2,4 = 5.72, p < 0.05; Fig. 4a), but this difference was seen only in the 800 ppm condition and not in the 1000ppm condition. Endophyte concentration increased in the high N treatment (F1,70 = 37.37, p < 0.01; Fig. 4b). In order to examine differences in the production of alkaloids by endophyte under our treatment conditions, we expressed alkaloid concentration (μg/gDM) per unit endophyte (copies/ng gDNA) and subjected the per unit endophyte production of all alkaloids individually to the ANOVA model above. We found that the production of alkaloids per unit endophyte significantly decreased for all three alkaloids under elevated CO2 (Lolines: F1,70 = 9.22, p < 0.01; Peramine: F1,70 = 5.23, p < 0.05; Ergovaline: F1,70 = 6.44, p < 0.05). The reduction in loline production per unit endophyte is shown in Figure 5a. This CO2-induced reduction was similar for both peramine (390 ppm: 0.21 ± 0.01; 800 ppm: 0.09 ± 0.01; 1000 ppm: 0.10 ± 0.01) and ergovaline (390 ppm: 0.027 ± 0.003; 800 ppm: 0.018 ± 0.002; 1000 ppm: 0.023 ± 0.003). In order to examine whether endophyte concentration was a good predictor of alkaloid concentration in general, we regressed endophyte content against alkaloid concentration. There was a significant positive correlation between endophyte and alkaloid concentration, but only under ambient conditions. This relationship became less significant as CO2 increased. Figure 5b, c and d shows the correlation for loline alkaloids (390 ppm: R2 = 0.38, F1,32 = 19.58, P < 0.0001; 800ppm: R2 = 0.11, F1,32 = 4.13, P < 0.1; 1000 ppm: R2 = 0.001, F1,32 = 0.032, P > 0.1). The relationship for peramine (390 ppm: R2 = 0.32, F1,32 = 14.99, P < 0.001; 800ppm: R2 = 0.11, F1,32 = 3.94, P < 0.1; 1000 ppm: R2 = 0.02, F1,32 = 0.667, P > 0.1) and ergovaline (390 ppm: R2 = 0.23, F1,32 = 9.55, P < 0.01; 800ppm: R2 = 0.07, F1,32 = 2.47, P > 0.1; 1000 ppm: R2 = 0.001, F1,32 = 0.019, P > 0.1) were similar to that of lolines shown in figure 5. The effects of elevated CO2 on fungal endophytes Ryan et al. 9 DISCUSSION Plant metabolite responses Plant carbohydrate responses in the present study were similar to those seen in other studies, in that carbohydrates (here LMW and HMW sugars) decreased in high N (Pettersson, McDonald & Stadenberg 1993; Skinner, Morgan & Hanson 1999; Curtis et al. 2000; Rasmussen et al. 2007, 2008) and increased under elevated CO2 (Pettersson et al. 1993; Skinner et al. 1999; Curtis et al. 2000; Robinson et al. 2012). For nitrogen metabolites, only total amino acids changed under elevated CO2, with the concentration decreasing under elevated CO2 in the low nitrogen treatment (nitrogen x CO2 interaction). We found no CO2 x endophyte interaction for amino acids and soluble protein concentrations. This is inconsistent with the results of Hunt et al. (2005) and Newman et al. (2003) who found that a CO2-induced reduction in protein in uninfected perennial ryegrass and tall fescue plants respectively, disappeared when plants were infected with endophyte. We found no evidence here of this buffering effect of endophyte in tall fescue. It is difficult to say why this discrepancy might occur, but we note that both previous studies where conducted under field conditions, and thus in very different environments than the present study. Additionally, the Newman et al. (2003) study used the cultivar “Kentucky-31” and analyzed cuttings from the “grazed horizon” (5cm above soil level) combined over three time points, each of which had a higher regrowth interval (4 weeks) than the plants used here (1 week). The Hunt et al. (2005) study used harvested leaf and sheath tissues of perennial ryegrass combined together, but in 50-day post-germination plants. Thus, discrepancies may be accounted for by differences in growth conditions, genetic background, sampling intervals and/or plant age. Endophyte and alkaloid responses Our observation that alkaloid concentrations more than doubled in the high N treatment suggests that the production of alkaloids in tall fescue is N-limited. The importance of plant nitrogen status on the production of N-based alkaloids may explain the slight CO2-induced decrease in alkaloids seen here for both lolines and ergovaline. A general hypothesis has emerged in the CO2 literature which predicts that increased carbohydrate accumulation under elevated CO2 results in a dilution of nitrogen which may result in a subsequent reduction in N-based metabolites (Cortrufo, Ineson & Scott 1998; Ryan et al. 2010; Robinson et al. 2012). This study shows that CO2-induced nitrogen limitation may also affect symbiont-derived metabolites, which may have implications for nutrient exchange between symbiotic partners. Decreased alkaloid concentrations under elevated CO2 may result in reduced herbivore resistance, which may decrease the competitive advantage of endophyte infection in a high CO2 atmosphere. The effects of elevated CO2 on fungal endophytes Ryan et al. 10 Endophyte concentration (copies/g gDNA) increased in the current study under elevated CO2, but only for the 800ppm treatment. This may be due to increased carbohydrate resources being shunted to endophyte growth due to increased photosynthetic output. However, the dynamics of resource exchange in grassendophyte symbioses are not well understood. Rasmussen et al. (2007) found that the concentration of the closely related N. lolii endophyte in perennial ryegrass decreased in plants that were selectively bred to produce high levels of water-soluble carbohydrates. More recent work suggests that such changes may be brought about in part by carbohydrates but are also strongly dependent on the genetic background of host plants (Ryan et al. unpublished). The grass-endophyte mutualism and elevated CO2 We observed that alkaloid production per unit endophyte decreased in elevated CO2 (Fig. 5a). Interestingly, we also found that a linear correlation between endophyte and alkaloid concentration in ambient conditions disappeared in elevated CO2, which may indicate that the symbiotic nutrient economy may be disrupted to some degree in a high CO2 atmosphere (Fig.5 b,c,d). The relationship between alkaloid and endophyte concentration in plants grown under elevated CO2 appears to be quadratic, with very low levels of alkaloids in plants with either very low or very high endophyte concentrations, and this was particularly apparent in the 1000 ppm treatment. This may suggest that plants undergoing the largest CO2-induced changes in resource allocation have a higher endophyte concentration due to excess carbohydrates, and the lowest alkaloid concentrations due to simultaneous N dilution. If this interpretation is correct then we would expect that plants where the C:N ratio are highest (i.e. strong CO2induced C/N allocation responses) would have the lowest alkaloid per unit endophyte concentrations. While there was some support for this here (Log[lolines per unit endophyte] = 3.17 – 0.38 (Log[C:N]); R2 = 0.06, F1,102 = 7.03, P < 0.01), the C:N ratio explained only a small amount (~6%) of the variation in the response. It is unknown whether host plants can exert control over the degree of endophyte growth (Ryan et al. 2008). The current study suggests that nutrient exchange may in part be a passive process: an “overflow” mechanism whereby nutrient exchange between symbiotic partners is subject to availability. Our results point to the intriguing possibility that the degree of benefit conferred by endophyte infection and associated anti-herbivore toxin production for the host plant may decrease in a high CO2 atmosphere by shunting nutrients toward endophyte growth and away from alkaloid production. The idea that the strength and/or direction of the symbiosis might be altered in high CO2 could be more appropriately tested by examining the fitness responses of endophyte-infected vs. endophyte-free plants in elevated CO2, and the degree to which they are The effects of elevated CO2 on fungal endophytes Ryan et al. 11 dependent on the presence of herbivores. In practice, lifetime fitness is difficult to measure and vegetative reproduction and biomass are often used as proxies (Cheplick & Faeth 2009). However, it is worth noting that there was no CO2 x endophyte interaction in the present study for plant growth parameters, suggesting that the relative difference between E+ and Eplants was similar across all CO2 treatments. Endophyte-infected plants consistently produced more tillers and aboveand belowground biomass across all treatments (data not shown). However, endophytes may provide a range of benefits to host plants outside of herbivore resistance (Clay 1990) and the changes in resource allocation observed here will be dependent on the response of other factors affecting this mutualism. Increased tillering and biomass production in endophyte-infected grasses could be due to e.g. increased production of growth affecting phytohormones such as gibberellins and auxin (Depuydt & Hardtke 2011). It has been shown previously that clavicipitaceous endophytes can produce auxin (indole-3-acetic acid and related compounds) in culture (Porter et al. 1985; De Battista et al. 1990; Yue et al. 2000; Tanaka et al. 2003), but it is unclear if fungal auxin production affects plant growth in vivo (De Battista et al. 1990). It is also possible that endophyte infection regulates plant hormone production and a transcriptome study of L. perenne infected with a mutated strain of E. festucae impaired in a stress-activated mitogen-activated protein kinase showed significant effects on a number of plant genes related to phytohormone, including auxin biosynthesis and translocation in vivo (Eaton et al. 2010). How generalizable are endophyte responses to CO2 and N fertilization? Research on the nutrient dynamics of grass-endophyte interactions in response to altered environmental variables suggests that results may not be generalizable across Neotyphodium species. In N. coenophialum-infected tall fescue, increased nitrogen fertilization has been shown to result in increased alkaloid production (Lyons, Plattner & Bacon. 1986; Arechavaleta et al. 1992) and the current study suggests that N fertilization also results in increased endophyte concentration. On the other hand, in N. lolii-infected perennial ryegrass, increased nitrogen input appears to cause a dilution effect resulting in decreased alkaloid (Hunt et al. 2005; Rasmussen et al. 2007) and endophyte (Rasmussen et al. 2007) concentrations. While we observed a decrease in alkaloid concentration under elevated CO2 in the present study using tall fescue, endophyte-derived alkaloids have been shown to increase under elevated CO2 in perennial ryegrass (Hunt et al. 2005). Taken together, these results suggest that treatment-induced changes in plant physiology do not induce similar responses in different Neotyphodium–grass associations. Continued research on how changes in plant physiology affect growth, physiology and nutrient dynamics in fungal endophytes may illuminate some of the mechanisms underlying this interaction. In particular, The effects of elevated CO2 on fungal endophytes Ryan et al. 12 variables which alter the balance of storage and growth in plants, such as manipulation of CO2 levels, nutrient fertilization and selective plant breeding, may be useful tools for the study of plant-fungal endophyte interactions. Additionally, comparisons in plants sampled at similar ages, growth conditions and regrowth intervals may illuminate the extent to which such differences are truly divergent rather than the product of sampling differences. ACKNOWLEDGMENTS We gratefully acknowledge Lui Qianhe (AgResearch) for advice on qPCR, Wade Mace and Heike Schwendel (AgResearch) for assistance with HPLC and GCMS (alkaloid analysis) and Kruti Shukla for assistance with plant maintenance. GR was supported by a fellowship from the Natural Science and Engineering Research Council (NSERC) Canada. Funding to support this research was provided by grants from NSERC and the Ontario Ministry of Agriculture, Food and Rural Affairs, to JN.