• cereal aphids;
  • endophytic fungi;
  • Neotyphodium lolii;
  • parasitoids;
  • trophic cascades


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    Endophytic fungi are associates of most species of plants and may modify insect community structures through the production of toxic alkaloids. Fertilization is known to increase food plant quality for herbivores, but it is also conceivable that additional nitrogen could increase the production of the insect toxic alkaloid, peramine, in endophyte-infected plants.
  • 2
    The relative importance of soil fertility and endophyte infection on herbivores and their natural enemies is unknown. As performance of the host plant is often affected by an interaction between endophyte infection and genetic background, four different plant cultivars were tested. The main questions addressed in this study were whether plant cultivar and fertilizer addition to endophyte-infected and endophyte-free Lolium perenne affect alkaloid concentrations, plant life-history traits and the abundances of aphid species and their parasitoids.
  • 3
    In a full factorial outdoor experiment we found a strong positive effect of fertilizer on plant biomass and on the abundance of aphids and parasitoids. While plant traits differed between cultivars, there was little effect of cultivar on either aphid or parasitoid abundance. Only endophyte-infected plants contained alkaloids, and the concentration of peramine was enhanced in fertilized plants. However, endophyte infection had no negative effect on aphid or parasitoid abundances. Plant traits were only weakly influenced by endophyte infection in the field, which contrasts with plant growth room studies, where both germination rate and plant height were influenced by endophyte–cultivar interactions.
  • 4
    The generally weak effects of endophytes in the outdoor experiment could be explained by various additional constraints under field conditions and the relatively low peramine concentration that we observed.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Most plants have microbial associates (Clay 2004) and such associations may alter processes of plant succession (Clay & Holah 1999), general relationships between plant species diversity and productivity (Rudgers, Koslow & Clay 2004), and insect food web interactions (Omacini et al. 2001). Endophytic fungi of cool season grasses are often seen as mutualistic symbionts, with the fungi receiving shelter, nutrients and transmission to the next generation via grass seeds, and host grasses having higher stress tolerance and herbivore resistance (Schardl, Leuchtmann & Spiering 2004; Müller & Krauss 2005). Cereal aphids, which are common grass herbivores, often show strong negative responses when feeding on endophyte-infected agronomic grass species (Breen 1994; Hunt & Newman 2005; Meister et al. 2006). The grass–fungus association produces a cocktail of alkaloids and, in the grass Lolium perenne, the main insect toxic substance is peramine. The alkaloids ergovaline and lolitrem B are also found in L. perenne (Spiering et al. 2002; Schardl et al. 2004), the latter being responsible for ryegrass staggers in sheep (Schardl et al. 2004). All alkaloids vary in concentration and distribution within a single host plant (Fannin, Bush & Siegel 1990; Ball, Prestidge & Sprosen 1995; Keogh, Tapper & Fletcher 1996; Ball et al. 1997; Spiering et al. 2002) and toxic effects depend on environmental conditions (Faeth, Bush & Sullivan 2002) and the genetic backgrounds of fungus and grass host (Roylance, Hill & Agee 1994; Faeth et al. 2002). As nitrogen is a key component of alkaloids, it could be expected that nitrogen addition will increase the alkaloid concentration in infected grasses (Lyons, Plattner & Bacon 1986; Marks, Clay & Cheplick 1991; Latch 1993; Faeth & Fagan 2002). Indeed, concentrations of lolitrem B and peramine have been shown to be higher in well fertilized ryegrass compared with poorly fertilized plants (Latch 1993). However, even though plant nitrogen concentrations typically increase in response to fertilization (Davidson & Potter 1995), Faeth et al. (2002) found that the peramine concentration of Arizona Fescue was not altered by fertilizer treatment.

Generally, aphid densities are enhanced when plants are grown with additional fertilizer (Honek 1991; Davidson & Potter 1995); this could result in a conflicting situation for aphids on endophyte-infected plants where insect growth rates are enhanced by fertilization, but reduced through higher concentrations of toxic alkaloids. The aphid Rhopalosiphum padi benefits from fertilizer addition, showing higher growth rates on fertilized plants of Lolium (formerly Festuca) arundinacea. However, when the grass is infected with the endophyte Neotyphodium coenophialum, the positive effect of fertilizer is counteracted and aphid population densities decrease (Davidson & Potter 1995). In this latter study, effects on the population densities of natural enemies of aphids were not considered. It is, however, conceivable that not only herbivores, but also their natural enemies are affected by both endophyte presence and fertilizer addition, with further feedbacks on herbivore densities. Flying predators (Müller & Godfray 1999) and particularly parasitoids (Schmidt et al. 2003) can have strong negative effects on aphid colony growth. Several laboratory studies on endophytes have found that predators (de Sassi, Müller & Krauss 2006) and parasitoids (Barker & Addison 1996; Bultman et al. 1997; Bultman, McNeil & Goldson 2003) are negatively affected by the presence of endophytic fungi. However, these studies were conducted under laboratory conditions, with insects being fed on endophyte-infected food. Providing natural enemies with a choice, under field conditions, might result in less distinct fitness losses.

As with most studies on the effects of endophytes on herbivores and predators, effects on plant life-history traits are often measured only in greenhouse experiments and only during the first few months of the lifespan of grasses (e.g. Cheplick 1998, 2004; Cheplick & Cho 2003). Field conditions may alter these results, because more species, at different trophic levels, will interact in the field, potentially resulting in higher order interactions (Wootton 1994; van Veen, Morris & Godfray 2006). In addition, all endophyte-mediated effects on plant life-history, alkaloid concentration, density of herbivores and natural enemies may be influenced by the plant's genotype or cultivar (Cheplick 1998, 2004; Faeth et al. 2002; Cheplick & Cho 2003; Meister et al. 2006). Here we present data from four agronomically important cultivars of Lolium perenne L., with the asexually transmitted endophyte, Neotyphodium lolii Glenn, Bacon and Hanlin, which relies entirely on seed production of the host plant to pass to the next generation. It would be expected that such an endosymbiont would manipulate its host plant to allocate more resources to reproduction, compared with uninfected plants.

The main aim of this study was to understand the relationships between fertilizer treatment, endophyte infection and plant cultivar on plant life-history of L. perenne and the associated insect population densities in the field. This was achieved by a full factorial outdoor experiment, in which insects were left to colonize the plants naturally. The main predictions addressed were that: (1) endophyte infection alters plant performance, especially the allocation of resources to reproduction; (2) fertilizer addition and grass cultivar affect plant life-history traits and these may interact with endophyte infection; (3) peramine and nitrogen concentrations are enhanced after fertilizer addition; and (4) endophyte infection decreases aphid and parasitoid abundances, but grass cultivar and fertilizer treatment modify this effect.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

study species

Seeds of English ryegrass, L. perenne (Poaceae), of four different agronomically important plant cultivars, with and without endophyte infection by the common endophyte strain N. lolii, were used. The plant cultivars were Imp (Lolium × boucheanum, Grassland Impact), Nui (L. perenne Grassland Nui), Pac (L. perenne Grassland Pacific) and Sam (L. perenne, Grassland Samson). Each cultivar was either uninfected (E–) or infected (E+) with the fungus N. lolii. All uninfected grass cultivars showed no infection after diagnostic staining of seeds, whereas 67–97% of the stained seeds of infected plants had fungal hyphae, depending on cultivar (Meister et al. 2006).

experimental design

Establishment of plant material

In December 2003, infected and endophyte-free L. perenne of all four cultivars were grown on commercially available garden compost in a climate controlled plant growth room, under artificial light. Each seed was sown separately in a cell of 48 propagation plug trays containing 150 seedling cells per tray ( part number 65150). This represents 7296 single plants in total. The cells for each plant had a diameter of 2·8 cm with a volume of 17 cm3. Each tray contained only seeds from one infection status and one cultivar, resulting in a replication of n = 6 trays per treatment. Thirteen days after the seeds were sown, the height of 20 randomly chosen germinated plants per tray was measured. One day later, the germination success was also assessed by counting the proportion of germinated plants per tray.

For the main experiment, plants that had not germinated after 14 days were replaced by reserve plants of the same age and the 48 propagation plug trays were moved to a greenhouse for 4 months. At the end of the 4 months all trays were split in two equally sized half-trays containing 75 plant cells (Fig. 1). Sprawling roots at the bottom were removed before planting the half-trays into the soil of three experimental blocks in a field site at the end of April 2004. The base of each experimental block was fenced to prevent mouse entry, and filled with 20 cm of normal soil from nearby agricultural lands with an estimated nitrate content of 10 mg kg−1 (pers. comm. Theres Zwimpfer, University of Zürich). The three blocks were placed in snail proof enclosures in an experimental field site at the University of Zürich. The experimental field site was surrounded by grassland with naturally occurring grass aphids, a nearby forest and the university buildings. The 96 half-trays (henceforth referred to as 96 plots) are considered the experimental unit, and were equally distributed between the three experimental blocks, with two nested sub-blocks (Fig. 1). Randomization of the plots took place within the sub-blocks; half of the 96 plots were fertilized with a balanced fertilizer (Wuxal–MaagAgro, N 100 g l−1, P2O5 100 g l−1, K2O 75 g l−1, B 120 mg l−1, Cu 81 mg l−1, Fe 190 mg l−1, Mn 162 mg l−1, Mo 10 mg l−1, Zn 61 mg l−1), at a rate equivalent to 200 kg N ha−1 provided in seven doses at 2-week intervals between April and July. The amount of nitrogen added was representative of typical agricultural application rates in western Europe (Carsten Thies, pers. comm., University of Göttingen, Germany). Each of the three treatments [plant cultivar (Imp, Nui, Pac, Sam), endophyte infection (E–, E+) and fertilizer (F–, F+)] were present in each sub-block once, resulting in a replication of n = 6 per treatment (Fig. 1). Plants were watered as required and any weeds occurring between plots were regularly removed. A minimum of approximately 20 cm between all 96 plots reduced direct competition between the plants of the different treatments. Competition within the treatment in a single plot (half-tray containing 75 plants) was reduced due to the separated cells in the half-tray. However, root competition in deeper soil may have occurred.


Figure 1. The experimental design, showing the three experimental blocks with two nested sub-blocks and the within sub-blocks randomized plots (half-trays), with a total n = 96. Each plot contains 75 single plants in a half propagation plug tray. Imp, Nui, Pac, Sam = abbreviations for the four different grass cultivars. E–, E+ = endophyte-free and endophyte-infected plants. F–, F+ = not fertilized and fertilized plants.

Download figure to PowerPoint

Plant traits and chemical analyses

In August 2004, 8 months after L. perenne was planted, the total number of ears per plot was determined to estimate reproductive allocation. Thereafter, the above-ground biomass of all plots was harvested at ground level, and divided into ear and shoot biomass to provide a further estimate of the allocation to reproductive vs. vegetative growth. The oven-dried biomass was weighed and the number of spikelets and length of ears were measured for 10 randomly chosen ears per plot.

Prior to harvest, five randomly chosen shoots with ears were selected from each treatment plot. These were immediately frozen in liquid nitrogen, ground in a mill and dried by lyophilization for chemical analyses. Analyses of carbon and nitrogen concentrations were carried out using a CHNS-932 determinator (LECO Corporation, St Joseph, MI, USA). For peramine analysis, the powdered samples were extracted with a methanol : water (4 : 1 v/v) mixture and the extract washed with hexane five times. The hexane fractions were discarded. Using an HPLC, separation of peramine was performed on a C18/cation exchange column (150 mm × 4·6 mm with 5 µm beads, Alltech Associates,, part number 72574). The elution programme was performed with 5% solvent A and 95% solvent B for the first 9·5 min, then a linear change to 35% A and 65% B over the next 22 min and then back to 5% A and 95% B over 0·5 min and held for 2 min. Solvent A was acetonitrile : 0·1 m ammonium acetate (4 : 1, v/v) and solvent B was acetonitrile : water (9 : 1, v/v). The flow was 1·8 mL min−1. Peramine retention was approximately 25 min and was detected at 280 nm. The calibration was done with an authentic chemical standard.

Lolitrem B analyses were conducted to test the viability of the N. lolii used in the experiment. To avoid destructive sampling of the experimental plants, samples were collected from reserve greenhouse plants in April, 4 months after planting. The sheaths and blades from 20 different plants for each cultivar and infection status were pooled. Fifty milligrams of freeze dried and lyophilized material were extracted with dichloromethane : methanol (1 : 1 v/v, 7 mL, 30 min) and purified by solid phase extraction (Varian Bond Elut Si, 100 mg/1 mL, elution with dichloromethane : acetonitrile 4 : 1). The solvent was evaporated and the sample dissolved in acetonitrile : dichloromethane (2 mL/100 µL) for HPLC-MS analyses. This was performed on an Agilent 1100 HPLC system (Agilent Technologies, Palo Alto, CA, USA) connected to a Bruker ESQUIRE-LC quadrupole ion trap instrument (Bruker Daltonik GmbH, Bremen, Germany) in the (+)-APCI ionization mode. Chromatographic conditions: Waters Symmetry C18 column (150 × 2 mm) and a flow rate of 0·3 mL min−1. Mobile phase: gradient within 8 min from 50 to 90% of solvent B, then 4 min at 90% of B (solvent A: 0·05% formic acid solution in water, solvent B: 0·05% formic acid solution in acetonitrile). MS acquisitions were performed in the ‘single reaction monitoring’ mode (m/z 686·3–628·3).

Aphids and parasitoids

Aphids were counted and identified to species level on four occasions, at 2-wkly intervals (June–July 2004). Sampling times were standardized at 5 min per plot (75 plants), to ensure uniform sampling effort, and complete aphid counts. Aphids were not removed from the plants during these surveys and counts were pooled for statistical analyses. Aphid species identification followed Blackman & Eastop (2000). The three most common cereal aphids on L. perenne in the region are: Sitobion avenae Fabricius, Rhopalosiphum padi L. and Metopolophium festucae Theobald, which are all easy to identify in the field. These species are native to Europe and occur on numerous species of Gramineae (Blackman & Eastop 2000).

To detect the abundances of primary and secondary parasitoids of aphids, all aphid mummies were collected from all plants on two of the survey dates in July 2004. Each mummy was placed into individual gelatine capsules and left to emerge in the laboratory. The identification of primary parasitoids was based on Starý (1966, 1973), that of Alloxystinae on van Veen (1999) and van Veen, Belshaw & Godfray (2003), and that of the other secondary parasitoids on Graham (1969) and Fergusson (1980). All parasitoid identifications were confirmed by Dr Frank van Veen (Imperial College London, UK).

Further grass herbivores (a total of 26 beetles and bugs) and aphid predators (a total of 109 ladybirds, syrphids, lacewings, beetles, bugs and spiders) were detected during the four aphid surveys. The overall low species abundances did not allow meaningful population density analyses for these groups.

statistical analyses

All statistical analyses were conducted using R (version 2.1.1). Linear mixed effects models with a maximum likelihood method were calculated for the main experiment with fixed factors (1) plant cultivar, (2) endophyte infection, and (3) fertilization, and all interactions between these. Block and nested sub-blocks were treated as random factors (Pinheiro & Bates 2000). Statistical analyses of peramine concentration were restricted to endophyte-infected plants only, as uninfected plants did not contain this alkaloid. Response variables were transformed when necessary to meet the assumptions of normality and homoscedasticity. Biomass measurements and number of spikelets were log10 transformed, number of ears and length of ears were square-root transformed. The count data for aphids and parasitoids were pooled over the four sampling dates for the aphids and over the two sampling dates for the parasitoids. The pooled counts were square-root transformed. Pearson correlations were calculated to identify correlations between the transformed response variables at the different trophic levels. Data collected in the plant growth room had plant cultivar and endophyte infection as fixed factors and block as a random factor. The response variables germination rates and height of seedlings were not transformed. Arithmetic means and standard errors of back-transformed data are given throughout the text and shown in all figures.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

plant life-history traits and alkaloids

There was a significant interaction between endophyte infection and plant cultivar, in terms of germination success and plant height for L. perenne on days 13 and 14 (Fig. 2). Eight months later, at the end of the experiment, there were no longer clear differences in plant traits between endophyte-infected and endophyte-free plants. There was a significant three-way treatment interaction (cultivar × endophyte × fertilizer) for total biomass and shoot biomass, as well as a significant interaction between endophyte and fertilizer for the number of ears produced (Table 1). Fertilizer addition strongly increased total above-ground biomass, shoot biomass (F– = 43·7 ± 1·5, F+ = 101·0 ± 3·7 g), ear biomass (F– = 9·7 ± 0·7, F+ = 15·4 ± 1·1 g) and number of ears (F– = 70·1 ± 5·2, F+ = 98·6 ± 6·5). These plant traits were also clearly affected by plant cultivar. For number of spikelets and ear length, plant cultivar was the only significant predictor; endophyte infection and fertilizer addition had no significant effects (Table 1).


Figure 2. Effects of endophyte infection on (a) germination rate and (b) plant height on four different cultivars of L. perenne. (a) Mean (± SE) germination rate in percent of the four different plant cultivars of L. perenne infected (E+, grey bars) and uninfected (E, white bars) by N. lolii showed a significant interaction between cultivar and endophyte infection after 14 days (endophyte × plant cultivar: F3,35 = 36·72; P < 0·0001). Endophyte infection (F1,35 = 11·59; P = 0·002) and plant cultivar (F3,35 = 20·43; P < 0·0001) were both significant. (b) Similarly, the mean (± SE) plant height in centimetres of the four cultivars after 13 days interacted with infection status (endophyte × plant cultivar: F3,35 = 4·09; P = 0·014). Neither endophyte infection (F1,35 = 1·91; P = 0·176) nor plant cultivar (F3,35 = 0·98; P = 0·414) were significant.

Download figure to PowerPoint

Table 1.  Mixed effects models showing the relationship between predictor variables (plant cultivar, endophyte infection, fertilizer treatment) and plant response variables. The three biomass variables plus number of spikelets were log10 transformed, No. of ears per plot and ear length were square-root transformed
Predictor variablesBiomassShoot biomassEar biomassNumber of ears per plot
Cultivar (C)F3,75 = 6·26F3,75 = 4·15F3,75 = 8·70F3,75 = 35·82
P = 0·0008P = 0·009P < 0·0001P < 0·0001
Endophyte infection (E)F1,75 = 0·19F1,75 = 0·13F1,75 = 0·02F1,75 = 0·08
P = 0·663P = 0·724P = 0·883P = 0·781
Fertilization (F)F1,75 = 450·71F1,75 = 571·73F1,75 = 21·89F1,75 = 38·00
P < 0·0001P < 0·0001P < 0·0001P < 0·0001
C × EF3,75 = 0·30F3,75 = 0·19F3,75 = 0·78F3,75 = 0·65
P = 0·827P = 0·910P = 0·509P = 0·584
C × FF3,75 = 0·26F3,75 = 0·32F3,75 = 0·43F3,75 = 0·74
P = 0·857P = 0·809P = 0·730P = 0·532
E × FF1,75 = 1·43F1,75 = 0·87F1,75 = 2·05F1,75 = 5·21
P = 0·235P = 0·354P = 0·156P = 0·025
C × E × FF3,75 = 3·12F3,75 = 3·09F3,75 = 1·90F3,75 = 2·34
P = 0·031P = 0·032P = 0·137P = 0·081
Predictor variablesNo. of spikeletsEar lengthNitrogenPeramine
  1. Significant P-values are presented in bold.

Cultivar (C)F3,75 = 7·48F3,75 = 23·77F3,75 = 1·45F3,32 = 7·30
P = 0·0002P < 0·0001P = 0·235P = 0·0007
Endophyte infection (E)F1,75 = 0·00F1,75 = 0·25F1,75 = 0·33 
P = 0·954P = 0·622P = 0·569 
Fertilization (F)F1,75 = 1·85F1,75 = 0·51F1,75 = 0·00F1,32 = 15·76
P = 0·178P = 0·478P = 0·985P = 0·0004
C × EF3,75 = 0·38F3,75 = 2·21F3,75 = 0·88 
P = 0·769P = 0·093P = 0·454 
C × FF3,75 = 1·85F3,75 = 1·76F3,75 = 0·40F3,32 = 0·96
P = 0·146P = 0·163P = 0·757P = 0·426
E × FF1,75 = 1·24F1,75 = 1·03F1,75 = 0·09 
P = 0·269P = 0·313P = 0·769 
C × E × FF3,75 = 0·26F3,75 = 2·31F3,75 = 0·59 
P = 0·855P = 0·083P = 0·625 

The nitrogen concentration (Table 1, Fig. 3a), as well as carbon concentration and C : N ratios (result not shown), were not significantly influenced by any treatment. The concentration of the two alkaloids lolitrem B and peramine, which are produced by the endophyte, could only be detected in infected L. perenne plants. Lolitrem B concentrations were 0·41 µg g−1 for Imp, 0·17 µg g−1 for Nui, 0·16 µg g−1 for Pac, and 0·36 µg g−1 for Sam (these data could not be analysed statistically because material had to be pooled, see Materials and methods). In unfertilized plants, peramine concentrations were 5·07 ± 0·88 µg g−1 for Imp, 3·43 ± 0·64 µg g−1 for Nui, 6·41 ± 1·41 µg g−1 for Pac and 7·46 ± 0·67 µg g−1 for Sam, indicating a strong cultivar effect on peramine production (Table 1). Fertilizer addition significantly increased peramine concentrations in all cultivars except Pac (Table 1, Fig. 3b).


Figure 3. (a) Effects of endophyte infection on nitrogen concentration on fertilized and not fertilized L. perenne. (b) Effects of fertilization on peramine concentration on four different cultivars of L. perenne. (a) Mean (± SE) nitrogen concentration in Lolium perenne above-ground tissue was not significantly affected by fertilizer addition (F+) compared with no fertilizer addition (F–), endophyte infection (E+, grey bars) compared with no infection by endophytes (E–, white bars) or the four different plant cultivars (pooled in the figure). (b) The mean (± SE) concentration of insect toxic peramine was significantly higher in fertilized L. perenne plants (F+, grey bars) than in not fertilized plants (F–, white bars) and significantly differed for the four plant cultivars (Imp, Nui, Pac, Sam). Note that endophyte-free plants did not contain peramine (for statistics see Table 1).

Download figure to PowerPoint

aphids and parasitoids

The total number of aphids summed over the four survey dates was 13 182 individuals, with Sitobion avenae (9487 individuals), Rhopalosiphum padi (2530) and Metopolophium festucae (1131) the three most abundant species. The abundances of S. avenae (F– = 61·3 ± 7·4, F+ = 136·3 ± 11·1) and M. festucae (F– = 5·5 ± 0·5, F+ = 18·1 ± 2·9) were increased by fertilizer treatment and S. avenae was also affected by plant cultivar (Table 2). There was a significant interaction between fertilizer × endophyte treatments on the abundance of R. padi; numbers were greater on endophyte-infected, unfertilized plants (F–/E– = 19·2 ± 2·6, F–/E+ = 37·4 ± 6·2, F+/E– = 25·1 ± 3·9, F+/E+ = 23·7 ± 3·2), but effects were only just statistically significant (Table 2).

Table 2.  Mixed effects models showing the relationship between predictor variables (plant cultivar, endophyte infection, fertilizer treatment) and insect response variables. Aphids, primary and secondary parasitoids each include all species of their trophic level and are pooled over all sampling dates. All response variables were square-root transformed
Predictor variablesAphidsSitobion avenaeRhopalosiphum padiMetopolophium festucae
Cultivar (C)F3,75 = 2·67F3,75 = 2·85F3,75 = 0·78F3,75 = 0·58
P = 0·054P = 0·043P = 0·511P = 0·626
Endophyte infection (E)F1,75 = 0·88F1,75 = 0·32F1,75 = 3·99F1,75 = 0·44
P = 0·352P = 0·575P = 0·049P = 0·508
Fertilization (F)F1,75 = 31·45F1,75 = 35·78F1,75 = 0·59F1,75 = 27·02
P < 0·0001P < 0·0001P = 0·447P < 0·0001
C × EF3,75 = 0·11F3,75 = 0·19F3,75 = 0·08F3,75 = 0·69
P = 0·954P = 0·905P = 0·970P = 0·563
C × FF3,75 = 0·49F3,75 = 0·40F3,75 = 1·04F3,75 = 0·30
P = 0·747P = 0·750P = 0·381P = 0·827
E × FF1,75 = 1·65F1,75 = 0·39F1,75 = 4·20F1,75 = 0·02
P = 0·203P = 0·534P = 0·044P = 0·892
C × E × FF3,75 = 0·61F3,75 = 0·96F3,75 = 0·25F3,75 = 0·67
P = 0·612P = 0·418P = 0·861P = 0·573
Predictor variablesPrimary parasitoidsAphidius rhopalosiphiSecondary parasitoidsDendrocerus aphidium
  1. Significant P-values are presented in bold.

Cultivar (C)F3,75 = 0·93F3,75 = 0·46F3,75 = 2·16F3,75 = 0·47
P = 0·433P = 0·711P = 0·099P = 0·702
Endophyte infection (E)F1,75 = 0·37F1,75 = 0·61F1,75 = 0·00F1,75 = 0·73
P = 0·548P = 0·436P = 0·963P = 0·396
Fertilization (F)F1,75 = 14·34F1,75 = 14·36F1,75 = 14·86F1,75 = 1·66
P = 0·0003P = 0·0003P = 0·0002P = 0·202
C × EF3,75 = 1·85F3,75 = 1·92F3,75 = 0·11F3,75 = 0·55
P = 0·145P = 0·133P = 0·952P = 0·648
C × FF3,75 = 0·67F3,75 = 1·02F3,75 = 1·58F3,75 = 0·39
P = 0·574P = 0·387P = 0·202P = 0·758
E × FF1,75 = 0·01F1,75 = 0·01F1,75 = 0·13F1,75 = 3·42
P = 0·916P = 0·931P = 0·723P = 0·068
C × E × FF3,75 = 0·40F3,75 = 0·41F3,75 = 0·45F3,75 = 0·78
P = 0·757P = 0·748P = 0·715P = 0·507

The number of parasitoids emerging from the collected mummies was 212 for primary parasitoids and 227 for secondary parasitoids, six species were identified as primary parasitoids and nine species as secondary parasitoids. The most abundant primary parasitoids were Aphidius rhopalosiphi (182 individuals), followed by A. picipes (13), A. ervi (12), Ephedrus plagiator (three), Praon volucre (one) and Aphelinus sp. (one). The most abundant secondary parasitoids were Dendrocerus aphidium (62), followed by Asaphes suspensus (41), D. carpenteri (39), Asaphes vulgaris (37), Phaenoglyphis villosa (20), Alloxysta victrix (15), Coruna clavata (nine), Syrphophagus aphidivorus (two) and Alloxysta tscheki (two). Individual numbers of A. rhopalosiphi were increased by fertilizer treatment (individuals per plot: F– = 1·1 ± 0·2, F+ = 2·6 ± 0·3), whereas there was no significant treatment effect on the number of D. aphidium (Table 2). All other species occurred at densities which were too low for population density analyses.

The total number of aphids (F– = 95·6 ± 8·7, F+ = 179·0 ± 12·8) and both primary (F– = 1·4 ± 0·2, F+ = 3·0 ± 0·3) and secondary parasitoids (F– = 1·6 ± 0·2, F+ = 3·1 ± 0·3) were enhanced by fertilizer addition to L. perenne (Fig. 4, Table 2). The effect of plant cultivar on total number of aphids was slightly above the significance level, whereas endophyte infection had no negative effect on the abundances of species at the three trophic levels, even though the insect-toxic peramine occurred only in plants with the fungal endophyte (Table 2; Fig. 3b). Plant biomass and the numbers of aphids, primary and secondary parasitoids were all positively correlated (Fig. 4), indicating that the effects of increased resource availability through fertilizer addition moves up the food chain. Therefore, it is not surprising that fertilizer addition had strong positive effects on all trophic levels in this insect food web. To remove the direct (plant biomass) effect of fertilizer, aphid and parasitoid numbers were divided by plant biomass. These biomass-corrected densities showed a significant plant cultivar effect on aphids (F3,75 = 6·06, P = 0·0009); all other predictors for aphid and parasitoid numbers were not significant (all P > 0·1).


Figure 4. Summary of the mixed effects models for the main treatment effects and Pearson correlations between the transformed response variables (log10: biomass and square root: aphids, primary (Prim.) and secondary (Sec.) parasitoids). Fertilization increased plant biomass, and aphid and parasitoid abundance significantly. ***P < 0·001; **P < 0·01; (*)P = 0·05–0·06; NS, P  0·1 (for statistics see Table 2 and Material and methods). Footnote: between predictor variables and response variables solid lines show significant relations, dotted lines show not significant relations..

Download figure to PowerPoint


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

In our fully factorial field experiment, fertilizer addition strongly enhanced the abundances of naturally colonizing aphids and parasitoids on agricultural grasses. Plant cultivar had a small effect on insect species abundance, while endophyte infection of the resource plant had no negative effect on insect abundances in this study. The absence of an effect of endophyte infection is in contrast to short-term laboratory trials. For example, clear negative effects of the endophyte N. lolii have been shown for herbivores and predators associated with L. perenne (Meister et al. 2006; de Sassi et al. 2006).

The aphid R. padi is known to be negatively affected by the presence of N. coenophialum in Tall Fescue (L. arundinacea), with the associated insect-toxic loline group of compounds (Davidson & Potter 1995; Hunt & Newman 2005), and by N. lolii and associated peramine production in L. perenne (Meister et al. 2006). It is surprising, therefore, that there was a trend towards higher densities of this aphid species on infected unfertilized plants, compared with uninfected and fertilized plants in our experiment. In another field experiment conducted in 2005, R. padi was also more abundant on infected L. perenne (Jochen Krauss, unpublished data). We currently have no explanation for why this aphid species shows such contrasting results. Endophyte effects on the aphids M. festucae and S. avenae could not be detected in our study; this supports data from laboratory studies that show that M. festucae has no clear negative response to endophyte infection (Simone Härri, unpublished data). Sitobion avenae colonizes ears of the grass and therefore depends on ear rather than leaf quality (Honek 1991). The concentration of peramine in ears is, however, unknown and was not measured separately in our study. Ear biomass, number of ears, ear length and number of spikelets all differed between plant cultivars; cultivar also significantly affected the abundance of S. avenae. In the absence of a fertilizer-related increase in foliar nitrogen concentration, the increase in abundance of S. avenae and M. festucae is likely to be linked to the overall increase in above-ground plant biomass following fertilizer treatment. Growth dilution of foliar N concentrations is a common phenomenon (Johnson, Ball & Walker 1997) and is likely to explain the lack of concentration increase observed in this study.

The increase in parasitoid numbers associated with fertilizer treatment appears to be a direct result of increased aphid availability resulting from the treatment-related increase in plant biomass. Correlations between the four trophic levels – plants, aphids, primary and secondary parasitoids – make this interpretation plausible. Such cascading trophic interactions are common in food webs and have frequently been described for terrestrial webs (e.g. Schmitz 1993; Dyer & Stireman 2003).

Endophyte infection did not provide any significant defence against the aphid herbivores in our study. Similarly, neither the treatment-related increase in peramine production nor the effect of plant cultivar affected the level of herbivore protection offered by the endophyte. Elsewhere, a further peramine producing Neotyphodium species has also been shown to provide no protection against a grasshopper species feeding on infected Arizona Fescue (Saikkonen et al. 1999). The relatively small effects of endophytes on aphids and parasitoids in our study might be explained by the relatively low concentrations of peramine found in our L. perenne plants (unfertilized: 5·5 µg g−1, fertilized 8·0 µg g−1). Other studies have reported concentrations in excess of 10·0 µg g−1 (e.g. Ball et al. 1995; Spiering et al. 2002), which is also the threshold level for feeding deterrence for the Argentine Stem Weevil (Keogh et al. 1996). Peramine concentrations below 3·0 µg g−1 are generally considered nontoxic for invertebrate herbivores (Siegel & Bush 1996).

Another reason for the relatively small effect of endophytes on insect herbivores and their parasitoids might be as a result of the experiment being conducted under field conditions, with numerous indirect interactions between species and insects having a wide choice of plants on which to feed and oviposit, in contrast to more controlled laboratory conditions. Furthermore, the clear effect of endophytes on plant performance in our 2-week growth room experiment disappeared 8 months later under field conditions. The main driver for plant performance in the field was fertilization and, to a lesser degree, plant cultivar. In the growth room study, endophyte infection and plant cultivar showed significant interactions in terms of germination success and plant height. These findings are consistent with other laboratory studies where plant performance is often affected by interactions between endophyte infection and host-plant genotype (Cheplick 1998, 2004; Cheplick & Cho 2003).

In conclusion, our study showed that under field conditions endophyte effects on plant performance, herbivores and natural enemies are less consistent than laboratory studies suggest. For aphid populations and their parasitoids, fertilizer addition at agricultural rates has much stronger effects on abundance than endophyte and alkaloid presence. The increase in peramine concentrations associated with fertilizer addition was not sufficient to decrease aphid population sizes. Overall, in this study system with four endophyte-infected agronomic grass cultivars and trophic interactions based on aphids and their parasitoids, we found that the effect of fertilizer on aphid and parasitoid abundance was greater than the effect of plant cultivar on L. perenne, and the effect of endophyte infection by N. lolii was minimal.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We thank Thomas Bultman, Stan Faeth and Dan Hare for helpful comments and Adrian Leuchtmann and Brian Tapper for their grass and endophyte information. Frank van Veen kindly helped with the identification of the aphid parasitoids. We thank Urs Stalder, Claudio de Sassi, Susanne Müller, Julia Nüscheler and Tobias Züst for laboratory and field assistance and Barbara Meister for pilot studies. The project was funded by a grant of the Swiss National Science Foundation to Christine Müller (Grant number 631-065950).


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Ball, O.J.P., Prestidge, R.A. & Sprosen, J.M. (1995) Interrelationships between Acremonium lolii, peramine, and lolitrem B in perennial ryegrass. Applied and Environmental Microbiology 61, 15271533.
  • Ball, O.J.P., Barker, G.M., Prestidge, R.A. & Sprosen, J.M. (1997) Distribution and accumulation of the mycotoxin lolitrem B in Neotyphodium lolii-infected perennial ryegrass. Journal of Chemical Ecology 23, 14351449.
  • Barker, G.M. & Addison, P.J. (1996) Influence of clavicipitaceous endophyte infection in ryegrass on development of the parasitoid Microctonus hyperodae Loan (Hymenoptera: Braconidae) in Listronotus bonariensis (Kuschel) (Coleoptera: Curculionidae). Biological Control 7, 281287.
  • Blackman, R.L. & Eastop, V.F. (2000) Aphids on the World's Crops. An Identification and Information Guide. Wiley, New York.
  • Breen, J.P. (1994) Acremonium endophyte interactions with enhanced plant resistance to insects. Annual Review of Entomology 39, 401423.
  • Bultman, T.L., Borowicz, K.L., Schneble, R.M., Coudron, T.A. & Bush, L.P. (1997) Effect of a fungal endophyte on the growth and survival of two Euplectrus parasitoids. Oikos 78, 170176.
  • Bultman, T.L., McNeil, M.R. & Goldson, S.L. (2003) Isolate-dependent impacts of fungal endophytes in a multitrophic interaction. Oikos 102, 491496.
  • Cheplick, G.P. (1998) Genotypic variation in the regrowth of Lolium perenne following clipping: effects of nutrients and endophytic fungi. Functional Ecology 12, 176184.
  • Cheplick, G.P. (2004) Recovery from drought stress in Lolium perenne (Poaceae): are fungal endophytes detrimental? American Journal of Botany 91, 19601968.
  • Cheplick, G.P. & Cho, R. (2003) Interactive effects of fungal endophyte infection and host genotype on growth and storage in Lolium perenne. New Phytologist 158, 183191.
  • Clay, K. (2004) Fungi and the food of the gods. Nature 427, 401402.
  • Clay, K. & Holah, J. (1999) Fungal endophyte symbiosis and plant diversity in successional fields. Science 285, 17421744.
  • Davidson, A.W. & Potter, D.A. (1995) Response of plant-feeding, predatory, and soil-inhabiting invertebrates to Acremonium endophyte and nitrogen fertilization in tall fescue turf. Journal of Economic Entomology 88, 367379.
  • Dyer, L.A. & Stireman, J.O. (2003) Community-wide trophic cascades and other indirect interactions in an agricultural community. Basic and Applied Ecology 4, 423432.
  • Faeth, S.H. & Fagan, W.F. (2002) Fungal endophytes: common host plant symbionts but uncommon mutualists. Integrative and Comparative Biology 42, 360368.
  • Faeth, S.H., Bush, L.P. & Sullivan, T.J. (2002) Peramine alkaloid variation in Neotyphodium-infected Arizona fescue: effects of endophyte and host genotype and environment. Journal of Chemical Ecology 28, 15111526.
  • Fannin, F.F., Bush, L.P. & Siegel, M.R. (1990) Analysis of peramine in fungal endophyte-infected grasses by reversed-phase thin-layer chromatography. Journal of Chromatography 503, 288292.
  • Fergusson, N.D.M. (1980) A revision of the British species of Dendrocerus Ratzeburg (Hymenoptera: Ceraphronoidea) with a review of their biology as aphid hyperparasites. Bulletin of the British Museum of Natural History (Entomology) 41, 255314.
  • Graham, M.W.R.D.V. (1969) The Pteromalidae of north-western Europe. Bulletin of the British Museum, Natural History (Entomology) Supplement 16, 1908.
  • Honek, A. (1991) Nitrogen fertilization and abundance of the cereal aphids Metopolophium dirhodum and Sitobion avenae (Homoptera, Aphididae). Journal of Plant Diseases and Protection 98, 655660.
  • Hunt, M.G. & Newman, J.A. (2005) Reduced herbivore resistance from novel grass–endophyte association. Journal of Applied Ecology 42, 762769.
  • Johnson, D.W., Ball, J.T. & Walker, R.F. (1997) Effects of CO2 and nitrogen fertilization on vegetation and soil nutrient content in juvenile ponderosa pine. Plant and Soil 190, 2940.
  • Keogh, R.G., Tapper, B.A. & Fletcher, R.H. (1996) Distributions of the fungal endophyte Acremonium lolii, and of the alkaloids lolitrem B and peramine, within perennial ryegrass. New Zealand Journal of Agricultural Research 39, 121127.
  • Latch, G.C.M. (1993) Physiological interactions of endophytic fungi and their hosts. Biotic stress tolerance imparted to grasses by endophytes. Agriculture, Ecosystems and Environment 44, 143156.
  • Lyons, P.C., Plattner, R.D. & Bacon, C.W. (1986) Occurrence of peptide and clavine ergot alkaloids in tall fescue grass. Science 232, 487489.
  • Marks, S., Clay, K. & Cheplick, G.P. (1991) Effects of fungal endophytes on interspecific and intraspecific competition in the grasses Festuca arundinacea and Lolium perenne. Journal of Applied Ecology 28, 194204.
  • Meister, B., Krauss, J., Härri, S.A., Schneider, M.V. & Müller, C.B. (2006) Fungal endosymbionts affect aphid population size by reduction of adult life span and fecundity. Basic and Applied Ecology 7, 244252.
  • Müller, C.B. & Godfray, H.C.J. (1999) Predators and mutualists influence the exclusion of aphid species from natural communities. Oecologia 119, 120125.
  • Müller, C.B. & Krauss, J. (2005) Symbiosis between grasses and asexual fungal endophytes. Current Opinion in Plant Biology 8, 450456.
  • Omacini, M., Chaneton, E.J., Ghersa, C.M. & Müller, C.B. (2001) Symbiotic fungal endophytes control insect host–parasite interaction webs. Nature 409, 7881.
  • Pinheiro, J.C. & Bates, D.M. (2000) Mixed-Effects Models in S and S-plus. Springer, New York.
  • Roylance, J.T., Hill, N.S. & Agee, C.S. (1994) Ergovaline and peramine production in endophyte-infected tall fescue: independent regulation and effects of plant and endophyte genotype. Journal of Chemical Ecology 20, 21712183.
  • Rudgers, J.A., Koslow, J.M. & Clay, K. (2004) Endophytic fungi alter relationships between diversity and ecosystem properties. Ecology Letters 7, 4251.
  • Saikkonen, K., Helander, M., Faeth, S.H., Schulthess, F. & Wilson, D. (1999) Endophyte–grass–herbivore interactions: the case of Neotyphodium endophytes in Arizona fescue populations. Oecologia 121, 411420.
  • De Sassi, C., Müller, C.B. & Krauss, J. (2006) Fungal plant endosymbionts alter life history and reproductive success of aphid predators. Proceedings of the Royal Society of London B 273, 13011306.
  • Schardl, C.L., Leuchtmann, A. & Spiering, M.J. (2004) Symbiosis of grasses with seedborne fungal endophytes. Annual Review of Plant Biology 55, 315340.
  • Schmidt, M.H., Lauer, A., Purtauf, T., Thies, C., Schaefer, M. & Tscharntke, T. (2003) Relative importance of predators and parasitoids for cereal aphid control. Proceedings of the Royal Society of London B 270, 19051909.
  • Schmitz, O.J. (1993) Trophic exploitation in grassland food-chains – simple models and a field experiment. Oecologia 93, 327335.
  • Siegel, M.R. & Bush, L.P. (1996) Defensive chemicals in grass–fungal endophyte associations. Recent Advances in Phytochemistry 30, 81118.
  • Spiering, M.J., Davies, E., Tapper, B.A., Schmid, J. & Lane, G.A. (2002) Simplified extraction of ergovaline and peramine for analysis of tissue distribution in endophyte-infected grass tillers. Journal of Agricultural and Food Chemistry 50, 58565862.
  • Starý, P. (1966) Aphid parasitoids of Czechoslovakia. A Review of the Czechoslovak Aphidiidae (Hymenoptera). Dr W. Junk, The Hague.
  • Starý, P. (1973) A review of the Aphidius-species (Hymenoptera, Aphidiidae) in Europe. Annotationes Zoologicae et Botanicae 84, 185.
  • Van Veen, F.J.F. (1999) Aphid hyperparasitoids: taxonomy, ecology and evolution. PhD Thesis, University of London.
  • Van Veen, F.J.F., Belshaw, R.D. & Godfray, H.C.J. (2003) The value of the ITS2 region for the identification of species boundaries between Alloxysta hyperparasitoids of aphids. European Journal of Entomology 100, 449453.
  • Van Veen, F.J.F., Morris, R.J. & Godfray, C.J. (2006) Apparent competition, quantitative food webs, and the structure of phytophagous insect communities. Annual Reviews of Entomology 51, 187208.
  • Wootton, J.T. (1994) The nature and consequences of indirect effects in ecological communities. Annual Reviews of Ecology and Systematics 25, 443466.