High nitrogen supply and carbohydrate content reduce fungal endophyte and alkaloid concentration in Lolium perenne


Author for correspondence: Susanne Rasmussen Tel: +64 6351 8182 Fax: +64 6351 8042 Email: susanne.rasmussen@agresearch.co.nz


  • • The relationship between cool-season grasses and fungal endophytes is widely regarded as mutualistic, but there is growing uncertainty about whether changes in resource supply and environment benefit both organisms to a similar extent.
  • • Here, we infected two perennial ryegrass (Lolium perenne) cultivars (AberDove, Fennema) that differ in carbohydrate content with three strains of Neotyphodium lolii (AR1, AR37, common strain) that differ intrinsically in alkaloid profile. We grew endophyte-free and infected plants under high and low nitrogen (N) supply and used quantitative PCR (qPCR) to estimate endophyte concentrations in harvested leaf tissues.
  • • Endophyte concentration was reduced by 40% under high N supply, and by 50% in the higher sugar cultivar. These two effects were additive (together resulting in 75% reduction). Alkaloid production was also reduced under both increased N supply and high sugar cultivar, and for three of the four alkaloids quantified, concentrations were linearly related to endophyte concentration.
  • • The results stress the need for wider quantification of fungal endophytes in the grassland–foliar endophyte context, and have implications for how introducing new cultivars, novel endophytes or increasing N inputs affect the role of endophytes in grassland ecosystems.


An estimated 20–30% of all grass species host systemic endophytes (Leuchtmann, 1992), which makes these associations of wide interest to studies of plant–fungal interactions. The largest plant family hosting these endophytes is the Poaceae and this contains a number of significant agricultural species which are often hosts of systemic endophytic fungi of the Clavicipitaceae family (Clay & Schardl, 2002). Lolium perenne–Neotyphodium lolii associations are the most common in temperate grasslands in Europe and Australasia, and L. arundinaceum (=Festuca arundinacea)–N. coenophialum associations in North America (Christensen et al., 1993). Hyphae grow in the intercellular spaces of above-ground organs and rely on nutrients produced by the plant (Clay, 1990; Clay & Schardl, 2002). Endophytic fungi have been demonstrated to confer benefits to their grass hosts, partly through altered physiology but notably through insect resistance/repellence because of the production of alkaloids (see review by Malinowski & Belesky, 2000; Schardl et al., 2004). These associations have widely been regarded as mutualistic, but there is growing evidence of negative interactions between host and endophyte under certain circumstances (Cheplick, 2004; Müller & Krauss, 2005; Malinowski & Belesky, 2006) and a recently published meta-analysis of grass/endophyte literature suggests that the degree of mutual benefit is conditional on environmental factors such as nutrient availability (Saikkonen et al., 2006). Changes in the balance of grass and endophyte can alter the fitness of both species and can have implications for wider ecosystem function (Clay & Holah, 1999; Müller, 2003; Rudgers et al., 2004; Clay et al., 2005; Malinowski & Belesky, 2006).

The common strain (CS) of the endophyte N. lolii in ryegrass produces three known alkaloids: lolitrem B, peramine, and ergovaline. Lolitrem B is the putative cause of ryegrass staggers in sheep (Gallagher et al., 1984), while ergovaline (the cause of fescue toxicosis; Lyons et al., 1986) and peramine provide the host plant with protection against herbivorous insects such as the Argentine stem weevil (Prestidge & Gallagher, 1988). For use in pasture grazing systems, novel associations between naturally occurring endophyte strains and high-yielding grass cultivars have been developed to retain insect deterrent properties but to have much reduced detrimental impact on grazing herbivores (Fletcher & Easton, 1997; but see Hunt & Newman, 2005); for example, the N. lolii strain ‘AR1’ has peramine only and ‘AR37’ lacks all three major alkaloids but produces janthitrems.

The production and concentration of endophyte alkaloids in plant tissues vary considerably in the field with season, weather and management (for L. perenne see Keogh, 1983; Ball et al., 1995; Salminen & Grewal, 2002; Hume & Barker, 2005; for L. arundinaceum see Lyons et al., 1986; Belesky et al., 1988; Arechevaleta et al., 1992; Malinowski et al., 1998). But there have been surprisingly few controlled studies of the effects of individual environmental components on alkaloid accumulation, sufficient to predict the consequences of environments on alkaloids. Few reports have been published on interactions between nitrogen (N) availability and endophytic alkaloid production, with sometimes contrasting results. In tall fescue, Lyons & Bacon (1984) and Belesky et al. (1988) showed increased alkaloids at elevated N availability, especially under moderate water stress (Arechevaleta et al., 1992), but Faeth et al. (2002) found no such increase in Arizona fescue. In L. perenne, ‘no consistent pattern has been observed’ (Lane et al., 1997), although Hunt et al. (2005) found N decreased alkaloid concentrations. We are not aware of any studies of the impact of new grass cultivars, bred for altered sugar metabolism, on alkaloid production, although sugar content has been proposed to alter function in a root endophyte (Peucedanum spec.) (Hadacek & Kraus, 2002).

One of the major difficulties encountered in previous studies has been in assessing the concentration of the fungus itself, as is necessary to address whether an environmental factor alters alkaloids by altering metabolism (e.g. alkaloid production per unit fungus) or by altering the concentration of the fungus. Although real-time PCR (qPCR) has been used previously to quantify, for example, arbuscular mycorrhizal fungi (Alkan et al., 2004), this method has only rarely been used in grass–Neotyphodium associations (Young et al., 2005). We present here an extension of this technique, for this purpose, which also has implications for understanding the scale of the contribution of the fungal genome to the expression and regulation of the metabolism of the association. Previous techniques based on hyphal counts, genetically modified reporter genes (Spiering et al., 2005) or ELISA (Stewart, 1986) have not led to widespread quantification of endophyte content in experimentation or practice in the productive pasture context.

The N. lolii/ryegrass association prevails in some of the most fertile, productive temperate grassland regions typified by New Zealand. It is important to confirm whether recent trends for greatly increased use of N fertilizer, and the introduction of novel traits such as ‘high sugar’ grass cultivars, will lead to sufficiently retained toxicity to insects, and yet remain safe for grazing mammals. However, the insights gained from relatively controlled manipulations of nutrient and energy supply to the host and fungus can have wider significance to understanding the nature of grass/endophyte associations, in general. Notably, more controlled manipulations are required to test the extent to which the association is truly mutualistic, or whether changes in resource supply in the host plant, to its direct advantage, may have adverse impacts on fungal presence, as would be implied by more recent hypotheses of the nature of plant/endophyte associations in natural as well as cultivated grassland ecosystems (Faeth & Fagan, 2002; Saikkonen et al., 2006).

Here, we report an environmental study of how N fertilization interacts with a selected ‘high sugar’ trait in ryegrass and how both features interact with abundance and alkaloid production of three N. lolii strains.

Materials and Methods

The fundamental design of the experiment was a three-way anova comprising two grass cultivars, four endophyte treatments (three fungi and ‘nil’), and two concentrations of N supply, as detailed below.

Plant material

We used two Lolium perenne L. cultivars (CV), ‘AberDove’ (IGER, UK) and ‘Fennema’. ‘AberDove’ was bred to produce high concentrations of water-soluble carbohydrates (so we denote ‘AberDove-HSG’) and these have been shown to be greater than in ‘Fennema’, which is used here as a ‘normal sugar’ comparison (Parsons et al., 2004).

Seedlings of the two cultivars were inoculated with three N. lolii strains differing in their alkaloid profiles: Common Strain Lp19 (CS), ‘AR1’ and ‘AR37’ (Latch & Christensen, 1985). One set of plants remained uninfected (‘nil’). In seed stocks in the UK, neither cultivar contains endophyte. This removes the uncertainty of having to use chemical treatments to obtain endophyte-free controls, and ensures all associations (including that with CS fungi) are de novo infections. Individual plants were grown for several months outdoors and subsequently tillers (shoot and root) of approximately the same size were separated and transferred into pots (two genotypes of the same CV infected with the same endophyte per pot) filled with a 1 : 1 mix of vermiculite and perlite and grown in controlled climate chambers (14 h light, 20°C; 10 h dark, 10°C; light intensity 620 µmols m−1 s−1). The endophyte status was initially confirmed by immunoblotting (Gwinn et al., 1991). A total of 160 pots (10 replicates × two cultivars × four endophytes × two N concentrations) representing 320 genotypes were kept under these conditions in two chambers in a random setup for 18 wk. During that time plants received daily water and a modified Hoagland nutrient solution containing 9 mm N supplied as nitrate (Brooking, 1976). Plants were cut back every 2 wk to a height of approx. 6 cm above ground (leaf blades, harvestable component).

Differences in N treatment were imposed after 18 wk. Half the pots remained on 9 mm N (high N), while the other half were supplied with 2.25 mm N (low N). Plants were cut back every 2 wk for a further 8 wk under the same conditions, as already described. The concentrations of N in nutrient solution were chosen to produce a total N content in harvested grass parts of 3%, typical of nonmineral N fertilized L. perenne seen in grass/legume-based pastures, where N is supplied only by the legume (low N), and of 4%, typical of L. perenne in well fertilized pastures receiving, for example, 200+ kg N ha−1 (Parsons et al., 1991).

Sample preparation

For the analysis of sugar, alkaloid and endophyte content, leaf blades (6 cm above-ground material) from all plants were harvested at the end of week 8, within 1 h of each other, on the same day, 7 h after the start of the daylight period. Blades were cut off with scissors and immediately frozen and ground in liquid N and subsequently freeze-dried. The material was stored at −20°C until further analysis.

Carbohydrate, amino acid and protein concentration

Water-soluble carbohydrates (WSCs) were extracted and quantified essentially as described previously (Hunt et al., 2005). In short, 25 mg of powdered plant material were extracted with 2 × 1 ml of 80% ethanol (low-molecular-weight (LMW) sugars) and subsequently with 2 × 1 ml water (high-molecular-weight (HMW) sugars) for 30 min at 65°C. Extracts were centrifuged, and supernatants of each fraction were combined and analysed separately using anthrone as a colorimetric reagent (Jermyn, 1956).

Total free amino acids were determined colorimetrically with ninhydrin, as described previously (Yemm, 1955), using l-leucine as standard. Soluble proteins were determined according to Bradford (1976) with absorbance measured at 595 nm.

Alkaloid concentration

For the analysis of peramine and ergovaline, dried and ground material (50 mg) was extracted with 1 ml of 50% aqueous isopropanol containing 1% lactic acid (w/v) and internal standards of ergotamine and homoperamine and analysed by HPLC as described by Spiering et al. (2002). Lolitrem B was analysed using an adaptation of the method of Gallagher et al. (1985) as described by Hunt et al. (2005) and estimated by comparison of peak areas with those obtained with authentic external standard lolitrem B (obtained from Chris Miles, AgResearch, New Zealand). Janthitrems were extracted and analysed by HPLC as described by Tapper & Lane (2004). Amounts are represented as ‘relative peak areas’, because no standards are available for this compound class.

Endophyte concentration

Fungal endophyte amounts were estimated based on quantitative PCR (qPCR, real-time PCR) of genomic DNA (gDNA) isolated from infected plants using a MyiQ™ cycler (Bio-Rad Laboratories Pty. Ltd., Auckland, New Zealand). DNA was extracted from 100 mg freeze-dried powdered material using DNeasy® Plant Mini kit (Qiagen, Biolab Ltd., Auckland, New Zealand) following the manufacturer's handbook. Extracted DNA was treated with RNAse H (Sigma New Zealand Ltd., Auckland, New Zealand) to remove RNA from the samples. Primers suitable for qPCR were designed for fragments of two individual endophyte-specific genes, encoding a chitinase and a nonribosomal peptide synthetase (NRPS-1) (see Supplementary Material). The gel-purified PCR products were cloned into TOPO vectors (Invitrogen NZ Ltd., Auckland, New Zealand) and transformed into One Shot® E. coli cells by chemical transformation. Single colonies were grown in liquid culture and plasmids extracted and purified (Qiagen Plasmid Kit). A dilution range of the plasmids from 2 × 101 to 2 × 106 copies was used for calibration of the PCR reaction. Four nanograms of gDNA isolated from infected and uninfected plant material was mixed with iQ SYBR Green Supermix (Bio-Rad), primer pairs and water in a total volume of 20 µl per qPCR assay. The PCR protocol was as follows: cycle 1 (1×), 95°C for 3 min; cycle 2 (45×), 95°C for 20 s, 62°C for 30 s, 72°C for 1 min (data collection and real-time analysis enabled); cycle 3 (1×), 95°C for 1 min; cycle 4 (85×), 95°C for 10 s (decrease setpoint temperature after cycle 2 by 0.5°C; melt curve data collection and analysis enabled); cycle 5 (1×), 12°C hold. Data points were collected at the annealing temperature (62°C) and measured as fluorescence (excitation maximum 494 nm, emission maximum 521 nm). The formation of single PCR products in the assays was confirmed by subsequent gel electrophoresis and ethidium bromide staining of amplified DNA (data not shown). A verification test of this procedure is available in the Supplementary Material.

Statistical analysis

The experiment was analysed as a full three-way anova (Edwards, 1985) with two cultivars (‘Fennema’ and ‘AberDove-HSG’), four levels of endophyte (nil, CS, AR1, AR37), and two concentrations of N (9 and 2.25 mm). Each of the 4 × 2 × 2 = 16 treatment combinations was replicated 10 times with ‘pots’ (two plants per pot) as the unit of replication. All analyses were conducted using JMP statistical software version 5.1. We used Box-Cox transformation to homogenize the error variances. We report, uniformly, the untransformed means, and standard error of the means as a measure of data dispersion.

In the case of fungal concentration, we used a principal-component analysis (McGarigal et al., 2000) to reduce the two highly correlated gene copy numbers to a single variable (the first principal component, PC1).


In all cases, the full model (i.e. all main effects and two- and three–way interactions) was analysed. Full anova tables are available in the Supplementary Material, if required. For ease of comprehension, the text and figures below focus on the statistically significant effects.

Endophyte concentration in plant tissue

The concentration of endophyte in the harvested plant material was assessed using qPCR and is expressed as the number of copies of each of two fungal-specific genes per total (plant + fungal) genomic DNA. The endophyte strains used in this experiment are haploids with one genome per cell (compartment) and both genes studied are present as single copies in the endophyte genome (S. Bassett & L. Johnson, pers. comm.).

The main effects of sugar cultivar, N availability and endophyte strain on the concentration of endophyte are shown in Fig. 1a–c. Copy numbers of both endophyte genes, chitinase A and NRPS-1, were substantially reduced in ‘AberDove-HSG’ compared with ‘Fennema’ (Fig. 1a; chitinase A: F1,107 = 48.09, P < 0.0001; NRPS-1: F1,106 = 68.88, P < 0.0001). Copy numbers of these two genes were also substantially reduced in plants supplied with high N concentration (Fig. 1b; chitinase A: F1,107 = 26.12, P < 0.0001; NRPS-1: F1,106 = 25.23, P < 0.0001). Comparing the different endophytes, copy numbers were significantly reduced in plants infected with ‘AR37’ compared with the common strain or ‘AR1’ (Fig. 1c; chitinase A: F2,107 = 15.33, P < 0.0001; NRPS-1: F2,106 = 15.37, P < 0.0001). There were no significant two- or three–way interactions.

Figure 1.

Main effects of Lolium perenne sugar cultivar (a), nitrogen (N) availability (b) and endophyte strain (c; CS, common strain) on fungal concentrations expressed as gene copies ng−1 total genomic DNA. Closed columns, NRPS-1; open columns, chitinase. Bars, untransformed means ± SE. There were no significant interactions.

Copy numbers of the two genes were highly correlated (Pearson correlation coefficient 0.905, P < 0.001, R2 = 0.7752) and so separate analyses for the two genes are not independent. However, we note that exactly the same conclusions are reached if we subject these two variables to a principal component analysis first, and then conduct the anova on PC1, which accounts for 95% of the total variance. Both genes load positively and strongly on to PC1. For brevity, results not shown.

Alkaloid concentration in plant tissue

The main effects of sugar cultivar and N on the concentration in dry matter (DM) of alkaloids peramine and lolitrem are shown in Fig. 2a–d. Recall that the endophyte strains have known differences in alkaloid profiles.

Figure 2.

Main effects of sugar cultivar (a, c) and nitrogen (N) availability (b, d) on peramine (a, b) and lolitrem B (c, d) concentrations. Closed columns, alkaloid concentrations in common strain (CS); open columns, alkaloid concentrations in AR1-infected Lolium perenne plants. Lolitrem B is only produced in CS-infected plants. The untransformed means ± SE are plotted. All four pairwise comparisons on the Box-Cox-transformed least-squared means are significantly different; differences in peramine concentrations between the strains CS and AR1 in each treatment were not significant. There were no significant interactions.

The alkaloid peramine is produced only in plants infected with CS and AR1 endophytes. Peramine concentrations were significantly reduced in ‘AberDove-HSG’ compared with ‘Fennema’ (Fig. 2a; F1,72 = 37.04, P < 0.0001) and under high N compared with low N (Fig. 2b; F1,72 = 30.82, P < 0.0001). There were no significant interactions and no significant effect of endophyte strain (CS, AR1) on peramine.

Lolitrem B and ergovaline are produced only in plants infected with CS endophyte. Both cultivar (Fig. 2c; F1,36 = 5.79, P < 0.05) and N (Fig. 2d; F1,36 = 12.53, P < 0.005) reduced the concentrations of lolitrem B. There was no significant interaction. There were no significant effects of cultivar, N or their interaction on the concentrations of ergovaline (not shown).

Janthitrems are produced only in plants infected with AR37. They followed the same general pattern as peramine and lolitrem B, being significantly lower in ‘AberDove-HSG’ than in ‘Fennema’ (0.54 ± 0.1 vs 1.02 ± 0.1 µg g−1 DM; F1,36 = 9.44, P < 0.005) and in high N than in low N (0.48 ± 0.1 vs 1.08 ± 0.1 µg g−1 DM; F1,36 = 14.86, P < 0.0005) (not shown). There was no significant interaction.

Factors such as cultivar (high sugar trait) or N supply could alter the production of alkaloids either by altering the concentration of fungus per se or by altering the rate of alkaloid production per unit endophyte. In Figs 1 and 2 there appears to be a similar trend in the way cultivar and N affect both endophyte and alkaloid concentration. To examine this further, the two highly correlated estimates of endophyte concentration (copy numbers of the two genes) were condensed to a single principal component, as described earlier, and alkaloid concentrations were regressed against PC1 using Box-Cox transformations to homogenize the error variance. Increasing values of PC1 can be readily interpreted as increasing fungal quantity. In all cases except ergovaline, the regression was highly significant. A post-hoc power analysis on the slope of the line for ergovaline regression yielded an estimate of power = 0.354 for the observed standardized effect size of 0.264 (JMP version 5.1). The untransformed data are shown in Fig. 3.

Figure 3.

The relationship between untransformed alkaloid concentration (µg g−1 DM) and the fungal concentration represented by a first principal component which is a positive linear combination of copy numbers ng−1 total gDNA of each of the two fungal-specific genes, chitinase-A and NRPS-1. Below we give the overall F-ratio for each regression on the transformed concentrations and the equation for the untransformed relationship. Peramine = 3.05 + 8.57x, R2 = 0.56, F1,76 = 92.75, P < 0.0001; lolitrem B = 0.79 + 0.39x, R2 = 0.53, F1,36 = 52.19, P < 0.0001; ergovaline = 0.28 + 0.03x, R2 = 0.47, F1,36 = 2.64, P > 0.05; janthitrems = 0.78 + 0.37x, R2 = 0.69, F1,38 = 92.99, P < 0.0001.

Water-soluble carbohydrate content

Our measurements confirm that HMW sugars were greater in ‘AberDove-HSG’ than in ‘Fennema’, but there was a significant interaction between the cultivar and the endophyte treatments (Fig. 4; F3,144 = 3.28, P < 0.05) such that ‘AberDove-HSG’ always had higher concentrations of HMW sugars than ‘Fennema’, but the difference between the cultivars was reduced when both were infected with AR1 or AR37.

Figure 4.

High-molecular-weight (HMW) carbohydrates. Closed columns, ‘AberDove-HSG’; open columns, ‘Fennema’. The untransformed means ± SE are plotted. We used Tukey's Honestly Significant Difference test on the Box-Cox-transformed least-squared means to separate the treatment combinations (different letters denote means that are significantly different).

High-molecular-weight sugar concentrations were reduced, independently of endophyte and cultivar, by high N compared with low N (mean of N effect, 17.32 ± 1.4 vs 34.34 ± 2.9 mg g−1 DM; F1,144 = 45.64, P < 0.0001) (not shown).

There was less difference between ‘AberDove-HSG’ and ‘Fennema’ in LMW carbohydrates, which is consistent with previous observations in endophyte-free material (Parsons et al., 2004). Again, in the present study there was a strong interaction between the cultivar and the endophyte treatments (Fig. 5; F3,144 = 12.67, P < 0.0001). ‘Fennema’ without an endophyte had lower LMW carbohydrates, and ‘Fennema’ with AR37 had higher LMW carbohydrates, than all other combinations.

Figure 5.

Low-molecular-weight (LMW) carbohydrates. Closed columns, ‘AberDove-HSG’; open columns, ‘Fennema’. The untransformed means ± SE are plotted. We used Tukey's Honestly Significant Difference test on the Box-Cox-transformed least-squared means to separate the treatment combinations (different letters denote means that are significantly different).

As for HMW carbohydrates, LMW carbohydrate concentrations were reduced, independently of endophyte and cultivar, under high N compared with low N conditions (76.68 ± 1.9 vs 93.68 ± 2.0 mg g−1 DM; F1,144 = 61.47, P < 0.0001) (not shown as graph).

Soluble protein and total amino acid content

Soluble protein concentrations tended to be lower in ‘AberDove-HSG’ than in ‘Fennema’ (Fig. 6), but there was a significant interaction between cultivar and endophyte (F3,144 = 15.90, P < 0.0001). ‘Fennema’ infected with the common strain of the endophyte resulted in the greatest concentration of soluble protein, while ‘Fennema’ infected with AR37 and the uninfected ‘AberDove-HSG’ had the lowest concentrations of soluble protein.

Figure 6.

Soluble protein concentrations. Closed columns, ‘AberDove-HSG’; open columns, ‘Fennema’. The untransformed means ± SE are plotted. We used Tukey's Honestly Significant Difference test on the Box-Cox-transformed least-squared means to separate the treatment combinations (different letters denote means that are significantly different).

Soluble protein concentrations were, as expected, higher under high N than under low N (81.05 ± 1.0 vs 70.37 ± 1.0 mg g−1 DM; F1,144 = 56.22, P < 0.0001) independent of endophyte and cultivar (not shown).

Total amino acid concentrations (not presented as graph) were lower in ‘AberDove-HSG’ than in ‘Fennema’ (14.47 ± 0.5 vs 16.11 ± 0.5 mg g−1 DM; F1,144 = 10.71, P < 0.005) and substantially higher under high N than under low N (20.91 ± 0.5 vs 9.67 ± 0.5 mg g−1 DM; F1,144 = 343.65, P < 0.0001). There was a marginally significant effect of the endophyte on total amino acid content (F3,144 = 2.65, P = 0.051).

Note that the effect of N treatment on total amino acids was proportionally far greater (×2) than its effect on soluble proteins (×1.2).


Endophyte and alkaloid concentration in host plant tissue

The use of a high sugar ryegrass cultivar (AberDove) substantially reduced the concentrations (by c. 50%) of both the endophytic fungus and alkaloids in plant leaves – the component predominantly harvested by mammalian herbivores. Likewise, a high N environment substantially reduced the concentration (by c. 40%) of both the endophyte and alkaloids. The two effects were additive (i.e. no significant interaction) so that the high sugar cultivar AberDove, at high N, had only c. 25% of the endophyte and alkaloid content seen in the control grass, ‘Fennema’, at low N. The effects were seen for all three endophyte strains (no interaction) regardless of the overall endophyte concentration, which was lower in AR37 than in AR1 and CS.

There are few previous studies in which sugar concentration has been manipulated to investigate its impacts on shoot endophyte function. High sugar concentrations have been implicated as reducing endophytic growth in roots in some Peucedanum associations (Hadacek & Kraus, 2002), but the Neotyphodium species used in our study are restricted to above-ground plant material

With regard to N inputs, our results are inconsistent with reports from tall fescue showing increased alkaloid production in grazed tall fescue plants under increased N supply (Belesky et al., 1988). But they are consistent with field trials using similar grass and endophytes to ours. Hunt et al. (2005) found decreased alkaloid accumulation in perennial ryegrass when N supply was increased. Lane et al. (1997) reported that alkaloid concentrations may both increase and decrease with high N inputs, depending on other environmental factors, such as drought. But in none of the studies above was it possible to confirm whether the N treatments altered fungal concentration per se, and in trials with multiple changes of environmental factors, results may be expected to be variable. In a trial using a similar L. perenne and endophyte association as in our present study, and where endophyte concentration was assessed, Stewart (1986) showed N application at spikelet initiation reduced endophyte concentration not only in the collected seed crop, but throughout the seedling, maturation and next generation of the crop grown from that seed.

In the present study, alkaloid concentrations varied linearly with endophyte concentration, in all cases except for ergovaline (which is the least abundant alkaloid and so effects are necessarily small). One study (Spiering et al., 2005) reported only a weak relationship between alkaloid and endophyte concentration (assessed using the GUS reporter gene) in a comparison between clonal populations of three plant genotypes selected for different endophyte content and distribution. In that study, much of the variation in endophyte content was between different tissues and tissue ages. In our present study, the variation in alkaloid and endophyte content, used to explore statistically the relationship between the two, comes from differences between genotypes within and between cultivars, and represents alkaloid and fungal concentrations in tissues of overall similar age. This may provide a more general representation of endophyte/grass population responses to external environmental factors.

Not only does qPCR enable us to quantify endophyte content more widely, it also allows us to reconsider the scale of the contribution of the endophyte to metabolism and gene expression in the association. Based on the comparison of total amounts of DNA, it is often argued that the endophyte represents only 0.5–2% of the association (Young et al., 2005). However, the haploid Neotyphodium lolii genome is very small (38 Mb) compared with the diploid ryegrass genome (6.18 × 103 Mb) and ratios based on total DNA are heavily biased towards plant DNA. Our results based on fungal genome copies per total gDNA show that, in leaf blades, the ratios of fungal compartments (one copy = one compartment) to plant cells may be as high as 1 : 2 to 1 : 6. The majority of plant DNA is noncoding, repetitive (approx. 50%) and of largely unknown function. The plant genome codes for 10 000–60 000 genes (Messing et al., 2004), but likewise the genome of the fungus Neurospora crassa (40 Mb, similar to Neotyphodium) also codes for approx. 11 000 genes (Mannhaupt et al., 2003). The relatively high ratio of potentially expressed fungal genes suggests a more substantial contribution for the fungus in the metabolism of the association than has previously been perceived, and re-emphasizes the importance of a quantitative estimation of fungal concentrations if metabolic interactions between both organisms are to be interpreted.

Possible explanations for changes in endophyte concentration in host grass

Lane et al. (1997) have proposed that changes in endophyte and alkaloid concentration could arise by ‘dilution’, if an environmental factor (such as N) stimulated the growth of the grass plant more than it stimulated the growth of the fungus. In the present study, because of the need to flash-freeze material for analysis, growth rates were not recorded. The 50% overall reduction in fungal concentration could be consistent with a ‘dilution effect’, however (presuming a doubling of grass growth and no change in fungal growth), given the large but realistic difference in N supply. But ‘dilution’ is an unlikely explanation for the effects of the high sugar cultivar on fungal and alkaloid concentration. Endophyte and alkaloid concentrations in ‘AberDove’ were reduced 50% relative to ‘Fennema’, but differences in growth between ‘AberDove’ and ‘Fennema’ are known to be small (Smith et al., 2002). The accumulation of sugars themselves is also an unlikely explanation for a physical dilution effect (the difference in sugars being < 2% of total biomass), and a measure of endophyte concentration expressed as fungal DNA per unit total genomic DNA is a measure that cannot be ‘diluted’ in any physical sense by sugar or other metabolites. However, in the case of both N and sugar, changes in endophyte and alkaloid concentration could arise at an organ scale, if the environmental factor altered the proportion of new leaves and whole tillers (branches) that were successfully colonized. Further controlled studies are needed to distinguish whether changes in endophyte and alkaloid concentrations arise from an altered success of the endophyte in entering new leaves and tillers at all, or from changes in the amount of infection within all organs. This is essential to identify the mechanism for the substantial variation and uncertainty in alkaloid concentrations seen with the multiply changing environmental conditions in the field.

We cannot preclude the possibility that the difference in endophyte concentration between the sugar cultivars is a cultivar-specific effect per se. It is unlikely to be an endophyte-specific effect because sugar cultivar (like N) reduced endophyte concentration in all three endophyte strains. Differences in infection rates and alkaloid concentrations between cultivars (indeed, between individual genotypes) are known to exist (Easton et al., 2002; Faeth et al., 2006), suggesting strong host × endophyte specificities (Cheplick & Cho, 2003; Saikkonen et al., 2004). To explore this we could have used a single plant genotype (isogenic lines) but these would not differ in sugar content without imposing different environmental treatments that might themselves affect endophyte/grass balance and so confound interpretation. To our knowledge, isogenic lines of ryegrass that differ only in the quantities of sugar do not exist currently, but these lines would be very valuable for studies on the effects of WSCs on endophyte growth and alkaloid production. Further research is needed with other cultivars that differ in WSC production to confirm if the reductions in fungal concentration are related to host specificity or sugar content per se. But host specificity cannot explain the major impact of N.

Metabolism: interactions between grass/endophyte

Our results provide evidence that the presence of an endophyte may elevate the sugar content in the grass host, notably elevating sugar content in the control grass ‘Fennema’ closer to that of the high sugar grass, ‘AberDove’, although the effect depended on which endophyte was involved (Figs 4, 5). The characteristic difference in HMW sugars between cultivars, seen in the ‘nil’ treatment, was still evident when both were infected with the ‘common strain’ but was much smaller and no longer significantly different when the two cultivars were infected with either AR1 or AR37. Likewise, the characteristic difference in LMW sugars was only significant in the absence of any endophyte (see ‘nil’ column, Fig. 5) and LMW sugars were no longer significantly different between cultivars with CS and AR1. With AR37, LMW sugars were greater in ‘Fennema’ than in ‘AberDove’. The possibility that endophytes may elevate sugar content in the host grass would be intuitive, given the value of sugar supply to the endophyte, but creates a dilemma in understanding why the effects of endophyte in elevating sugars are dwarfed by the large effect of sugar cultivar in reducing endophyte concentration.

Our results show no consistent effect of endophyte in elevating (or reducing) protein (Newman et al., 2003; Hunt et al., 2005). In our present study, AR37 elevated plant-soluble protein concentration (relative to uninfected plants) in the case of ‘AberDove’, but lowered protein concentration in ‘Fennema’. The CS endophyte elevated protein in ‘Fennema’. Hunt et al. (2005) found CS lowered soluble protein in the Samson cultivar of perennial ryegrass. Stimulating nitrogen availability in the host plant would intuitively be of benefit to the fungus and to its alkaloid production, given that this requires a supply of N, but this is not evident from the large impact of increased mineral N availability in decreasing the concentrations of fungus. Establishing the direction of cause and effect in the grass/endophyte association and a comprehensive account of the biotic interaction between the grass and endophyte requires a more detailed analysis of metabolome and transcriptome-wide changes (Johnson et al., 2006).

The major negative impacts of increased N and energy supply on fungal (and alkaloid) concentration seen in this study appear counterintuitive. But these may be in keeping with recent new perspectives of the nature of plant/fungal mutualism. Recent studies propose that there may be metabolic costs to the host plant of sustaining endophyte presence, and that this may be tolerated by the host (i.e. of net benefit) when outweighed by the benefits to the plant of improved nutrient uptake and/or insect defences (Faeth & Fagan, 2002; Lehtonen et al., 2005; Saikkonen et al., 2006). The current results would be in keeping with the premise that the prevalence of endophyte within each plant might be reduced (or the prevalence of infected vs noninfected plants might decline in the population) when conditions for the host plant are substantially improved. It has been proposed that grass/endophyte associations are only part of a continuum from mutualism to parasitism/pathogenicity and that the growth of the endophyte is under continual and dynamic control by the host (Kogel et al., 2006; Saikkonen et al., 2006).

Conclusions/practical implications

We have seen evidence of a relatively direct relationship between alkaloid production and the abundance of fungal endophyte in harvested tissues, and have demonstrated that simple changes, such as the application of grass cultivars with novel traits, or the simple use of increased N supply, can substantially alter both alkaloid and fungal concentrations. These changes would affect the balance of risk (toxicity) and benefit (insect deterrence) of having endophyte in the grassland ecosystem. It is highly likely that ‘high sugar’ cultivars and high N inputs will be used together. The HSG trait is associated with reducing N release from the rumen (ultimately in urine) and so mitigating some of the environmental impacts of high N inputs. Our work suggests this combination would not lead to increased toxicity of grass to livestock, as elevated sugar supply and N supply to the fungus might have been anticipated to do. Any increase in alkaloid production would have been more serious, of course, in a common strain endophyte (full toxic alkaloid complement) than in the novel endophyte associations. But there is a possibility, if our results prove general in field conditions, that these two factors in combination may reduce alkaloid concentrations (e.g. to 25%) at times below a critical threshold for insect deterrence.

Endophyte–grass interactions are complex and play a pivotal role in wider multitrophic assemblages (Omacini et al., 2001; Malinowski & Belesky, 2006). Their signals and metabolites can influence mycorrhizal infection (Müller, 2003), and their toxins can pass through insect herbivores to potentially antagonize desirable biocontrol agents (e.g. parasitoids; Bultman et al., 2003). Resolving the dynamics of the plant endophyte equilibrium will require substantially more study to analyse the fitness consequences and measure costs and benefits in physiological as well as population currencies, but in all cases will benefit from a wider use of techniques to quantify endophyte abundance.


This work has been partially funded by the NZ Foundation of Research, Science and Technology (contracts C10X0203 and PROJ 10333-ECOS-AGR). We acknowledge Dr Gregory Bryan (AgResearch) for his suggestion to use the endophyte chitinase gene for qPCR and for reviewing the manuscript. We thank E. Davis and Dr B. Tapper (AgResearch) for alkaloid analysis. We also thank Michael Hickey (AgResearch) for plant maintenance. Plasmid pTEFEGFP was kindly provided by Dr M. Kodama (Tottoti University, Japan) with permission from Dr J. Andrews (University of Wisconsin, USA). Plasmid pCYhph was kindly provided by Prof Barry Scott (Massey University, New Zealand).