Interspecific hybrid endophytes of the genus Epichloë (Ascomycota, Clavicipitaceae) are prevalent in wild grass populations, possibly because of their larger gene variation, resulting in increased fitness benefits for host plants; however, the reasons are not yet known. We tested hypotheses regarding niche expansion mediated by hybrid endophytes, population-dependent interactions and local co-adaptation in the woodland grass Hordelymus europaeus, which naturally hosts both hybrid and non-hybrid endophyte taxa.
Seedlings derived from seeds of four grass populations made endophyte free were re-inoculated with hybrid or non-hybrid endophyte strains, or left endophyte free. Plants were grown in the glasshouse with or without drought treatment.
Endophyte infection increased plant biomass and tiller production by 10–15% in both treatments. Endophyte types had similar effects on growth, but opposite effects on reproduction: non-hybrid endophytes increased seed production, whereas hybrid endophytes reduced or prevented it completely.
The results are consistent with the observation that non-hybrid endophytes in H. europaeus prevail at dry sites, but cannot explain the prevalence of hybrid endophytes. Thus, our results do not support the hypothesis of niche expansion of hybrid-infected plants. Moreover, plants inoculated with native relative to foreign endophytes yielded higher infections, but both showed similar growth and survival, suggesting weak co-adaptation.
Symbioses of plants and animals with microorganisms have led to major transitions in the evolution of life (Margulis, 1996). The combination of diverse physiological functions from different species in a symbiosis may aid in dealing with dynamic temporal or spatial fluctuations of the environment (Boyle et al., 2011) and thus allow symbiota to expand into novel habitats, such as the colonization of land by ancient plants associated with mycorrhizas (Humphreys et al., 2010). The present study aims to contribute to a better understanding of the ecological and evolutionary processes necessary for host–symbiont adaptation, an important precondition for symbiota to be successful.
All modern-day plants in natural ecosystems appear to be symbiotic with fungal endophytes, a diverse group of fungi that can have profound effects on plant ecology and evolution (Rodriguez et al., 2009). Among these fungal groups, one of the best known is the clavicipitaceous endophytes, here referred to as the Epichloae (Epichloë and Neotyphodium spp., Ascomycota), which form hereditary symbioses with grasses from the subfamily Pooideae in temperate areas (Schardl et al., 2004). Endophyte infection may increase plant fitness by conferring stress tolerance to drought and poor soil, or resistance to herbivory and fungal diseases (Kuldau & Bacon, 2008). Endophyte effects on host grasses have been studied extensively in grass–endophyte systems of agricultural importance, which are characterized by artificial or anthropogenic selection on the host grass genotypes (Clay et al., 2005; Monnet et al., 2005). In these systems, endophytes mostly provide beneficial effects to their hosts, and symbioses have been characterized as mutualistic (Assuero et al., 2000; Hesse et al., 2003). However, selection acting in a natural environment may be different from that in agricultural settings, and conclusions reported from agronomic grasses may be biased (Saikkonen et al., 1998). Documented effects of endophytes on wild host grass species in non-agricultural settings are less homogeneous (e.g. Brem & Leuchtmann, 2001; Gonthier et al., 2008; Kannadan & Rudgers, 2008; Craig et al., 2011; Iannone et al., 2012) and have been found to vary from positive to negative (Morse et al., 2002; Faeth et al., 2004; Rudgers & Swafford, 2009; Kane, 2011). Infection frequencies observed in wild grasses are often very high and have been interpreted to reflect a mutualistic interaction between the host grasses and their endophytes (Clay & Schardl, 2002). However, the infection rate may also be influenced by the potential of the endophyte for contagious spread and may be modulated by the endophyte transmission rate (Gundel et al., 2008, 2011; Gibert et al., 2012). Moreover, the outcome of a symbiosis appears to depend on the particular combination of host genotype, endophyte genotype and environmental factors (Meijer & Leuchtmann, 2000; Faeth & Fagan, 2002; Cheplick & Faeth, 2009). Understanding the interactions among genotypes and environment requires experiments that combine variations in all three factors, but, to our knowledge, such a comprehensive study in wild grass species is still lacking.
Asexual Epichloae (genus Neotyphodium) include hybrid and non-hybrid taxa. Non-hybrid endophytes are haploid and genetically resemble sexual Epichloë species, from which they are derived (Moon et al., 2004). Hybrid endophytes are heteroploid (incompletely polyploid) and presumably result from somatic fusion of distinct Epichloë and Neotyphodium species infecting the same host individual (Schardl & Craven, 2003). Hybridization may remedy the accumulation of deleterious mutations of asexual strains by adding genes derived from sexual recombination in different species (Muller, 1964; Schardl et al., 1994; Tsai et al., 1994). The fusion of genomes may also allow hybrids to occupy a broader ecological niche by expressing traits from both ancestral species (Clay & Schardl, 2002). Thus, hybrid endophytes may have a selective advantage relative to non-hybrid endophytes, and are indeed often prevalent among infected host species and populations (Moon et al., 2004; Iannone et al., 2009; Oberhofer & Leuchtmann, 2012). However, the mechanisms driving hybrid predominance in natural host grasses remain unclear.
Selection favoring hybrid over non-hybrid endophytes may vary under different environmental conditions. Several studies on the grass species Festuca arizonica, which hosts one hybrid and one non-hybrid taxon, have indicated that the hybrid endophyte mainly occurs in stressful environments (Morse et al., 2007; Sullivan & Faeth, 2008; Hamilton et al., 2009). An experiment comparing naturally infected grass individuals with their disinfected counterparts revealed that hybrids particularly benefit their hosts under dry/stressful conditions (Saari & Faeth, 2012). Although the authors concluded that hybrids may extend the niche of F. arizonica, hybrid-infected grasses appear to be limited to marginal habitats. These results seem inconsistent with the presumed broader ecological niche of symbiota with hybrid endophytes relative to symbiota with non-hybrid endophytes. Festuca arizonica represents an exception among wild grass species by hosting predominantly non-hybrid endophytes, so that the restricted niche of plants hosting hybrids might be unique to this host species. It is also conceivable that F. arizonica populations hosting hybrids differ genetically from those hosting non-hybrids, so that any stress-specific effects attributed to hybrid endophytes in previous studies are actually related to the host's genotype. To distinguish between host and endophyte genotype effects, the same host accessions need to be inoculated with both endophyte types, and grown under different environmental conditions (Cheplick & Faeth, 2009).
In a survey of 28 natural populations of Hordelymus europaeus across Europe, hybrid endophytes were found in 20 populations, non-hybrids in five populations and one was endophyte free (two populations hosted both types) (Oberhofer & Leuchtmann, 2012). In Switzerland, populations containing a non-hybrid endophyte were only found at a single location on dry, shallow calcareous soil, whereas hybrid endophytes were found at all other locations, with varying water availability (A. Leuchtmann, unpublished).
These observations suggest that, in contrast with F. arizonica, H. europaeus plants hosting the non-hybrid may be more drought tolerant than hybrid-hosting plants. Furthermore, the predominance of hybrid-hosting populations may be a consequence of greater fitness, greater transmission rate or a broader ecological niche of hybrids, as proposed previously (Schardl & Craven, 2003; Moon et al., 2004; Selosse & Schardl, 2007).
To test the niche expansion hypothesis, we exposed H. europaeus plants that had been artificially infected by hybrid or non-hybrid endophytes to drought stress, and examined plant growth performance. We expected plants with hybrid endophytes to thrive in both treatments and plants with non-hybrids to perform well in the drought treatment. Source material originated from four distant European populations, including two populations with hybrid and two with non-hybrid endophytes. Each host accession was recombined with its resident endophyte, the same endophyte type (different genotype), a different endophyte type or mock inoculated with sterile water, which allowed us to differentiate between host and endophyte effects on plant growth and drought tolerance (Cheplick & Faeth, 2009). We specifically asked the following questions: (1) do endophyte-infected plants perform better than disinfected plants?; (2) do plants infected by hybrid and non-hybrid endophytes perform differently?; (3) does plant performance depend on the endophyte–host type combination, that is, are grass populations naturally hosting hybrid or non-hybrid endophytes influenced differently by either endophyte type?; and (4) is there evidence for local co-adaptation, that is, does a particular endophyte genotype provide the greatest benefit to plants from the population in which it naturally occurs? With the results of this experiment, we expect to gain more detailed insight into the nature of the symbiosis between H. europaeus and its different endophytes and to explain the distribution and abundance of hybrid and non-hybrid endophytes. Moreover, we describe the method and results of re-inoculations in detail as these may have relevance for future research.
Materials and Methods
Plant species and endophyte taxa
Hordelymus europaeus (L.) Harz is a European woodland grass of tribe Triticeae and monotypic in its genus. This caespitose and perennial species is distributed throughout Europe, occurring in various forest plant communities, mostly on nutrient-rich soils (Tutin et al., 1980) with pH values ranging from 4.5 to 7.3 (M. Oberhofer, unpublished). As the selfing rate of H. europaeus is high, genetic diversity of individuals within populations is limited. However, plant haplotype divergence is considerable between populations at border areas of the distribution because of geographical isolation (Dvořáková et al., 2010). Hordelymus europaeus represents the grass species with the highest epichloid endophyte diversity known to date, including six different hybrid and non-hybrid endophyte taxa (Oberhofer & Leuchtmann, 2012). Therefore, it is a suitable study system for the investigation of the symbioses between the host grass and different endophytes.
Source populations (accessions) for this experiment were chosen from those included in a previous survey (Oberhofer & Leuchtmann, 2012) based on their geographical distance (to represent the entire distributional range), type of endophyte (hybrid or non-hybrid) and relative ease of seed germination, which was low or absent in several populations under sterile conditions (germination rates of endophyte-free seeds from all transect populations were evaluated in a pre-experiment and only accessions with a germination rate ≥ 0.2 were considered for the experiment). Accessions included a South Italian population of H. europaeus hosting the sexual non-hybrid endophyte E. sylvatica ssp. pollinensis (A2), a Swiss accession hosting an asexual non-hybrid endophyte matching E. bromicola (B; includes subpopulations B7, B8 and B9), a German accession (C1) and a Swedish accession (E2), hosting two different genotypes of the hybrid endophyte E. hordelymi (Leuchtmann & Oberhofer, 2013). Drought regularly occurs at the two sites with non-hybrid accessions (Mediterranean summer climate at site A2, shallow calcareous soil at site B), but not at the two other sites (C1 and E2).
Endophyte removal and re-inoculation
Seeds were collected from 30 individuals per population and pooled (representing a host genotype) before heat treatment for endophyte removal (Nott & Latch, 1993). Seeds were surface sterilized (Leuchtmann & Clay, 1990) and kept in 100% relative humidity at 37°C for 21 d. Seeds were then placed on water agar plates (1%) at 4°C for 21 d in darkness for stratification, and then at 18 : 10°C with a day : night cycle of 8 : 16 h for germination. On germination, seedlings were examined under the microscope (×100) for possibly emerging mycelia. In the four-leaf stage, all seedlings were again examined by leaf-sheath scratches to confirm that seedlings were endophyte free (Bacon & White, 1994). All resulting endophyte-free plants were grown in pots on an open terrace for one season to obtain second-generation seeds. To prevent cross-pollination, inflorescences of each emerging spike were covered with a commercial tea bag. Seeds were harvested successively at ripeness to prevent excess saprophytic fungal contamination. Endophyte-free seeds were stored at 4°C over silica gel (Rubin 85815; Sigma Aldrich, St Louis, MO, USA) until use for the main experiment.
Deglumed seeds were surface sterilized and stratified on water agar plates as described above. Germination was induced in a climate chamber (Sanyo Versatile Environmental Test Chamber, Sanyo Commercial Solutions, Bensenville, IL, USA) set to an 8-h day at 20°C and 16-h night at 12°C. Illumination devices were Osram L 40W 20SA and Osram L 40W 25SA (Munich, Germany). The germination rate of seeds collected from the selected endophyte-free accessions was, on average, 46% from a total of 6450 seeds (Table 1). On germination, seedlings were transferred to fresh water agar dishes (seven per dish) to reduce the risk of contamination.
Table 1. Germination rate of seeds of Hordelymus europaeus plant accessions and, for each combination of plant accession and fungal strain, rate of seedling infection after re-inoculation, rate of seedling survival and number of plants used in the drought stress experiment
Natural population infection rate
Infection rates for natural source populations are given for reference (Oberhofer & Leuchtmann, 2012). Infection rates for resident combinations of experimental plants are indicated in bold and not attempted combinations are marked with a dot.
For the inoculation treatment, fungal isolates obtained from the selected accessions (Oberhofer & Leuchtmann, 2012) were grown on an agar medium containing 1% malt extract, 1% glucose, 0.25% bacto-peptone and 0.25% yeast extract (Leuchtmann, 1994) for a maximum of 2 wk to ensure the presence of young and vital mycelia to be used as inoculum. Inoculations of seedlings involving all four plant accessions and corresponding fungal isolates were performed as described by Leuchtmann & Clay (1993), 3–6 d after germination, when the first leaf was 3–12 mm long and still enclosed by the coleoptile. Each inoculation was performed with a fresh sterile hypodermic needle in a laminar flow hood under a binocular dissecting scope at ×6–12 magnification. Seedlings were punctured at < 1 mm height to place the inoculum next to the basally located meristematic tissue. Mock-inoculated control seedlings (endophyte free, E−) were treated in the same manner, except for the use of sterile water instead of the inoculum. Seedlings of one Petri dish were always treated with the same fungal strain. Inoculated seedlings were stored for 72 h in the dark at room temperature. Afterwards, all seedlings remained for another 14 d in the climate chamber before being transplanted into soil of pH 6.6 (85% commercial potting substrate supplemented with 10% Perlith and 5% Jura limestone).
Plants were checked for infection at the three- to four-leaf stage using one leaf sheath per plant 2–4 wk after transplanting into soil. Samples were stained with aniline blue and examined for the presence of fungal hyphae as described above. Rates of germination, survival and successful infection are given in Table 1.
Drought stress experiment
The drought stress experiment was performed in 3 l pots equipped with access tubes for the T3/22/152 access tube probe (Imko Multimodultechnik Gmbh, Ettlingen, Germany) of a Trime® FM3 moisture meter (time domain reflectometer, TDR). Access tubes consisted of tecanat (Vink AG, Dietikon, Switzerland) closed by rubber plugs at both sides and placed vertically in the center of each pot. Pots were lined with polyester cloth and filled with nutrient-poor soil characterized by a high proportion of porous structure elements, including fine-grained limestone (Dach- und Trogerde, Ökohum, Herbertingen, Germany).
The experiment included a total of 260 plant individuals from four accessions re-inoculated with the four fungal isolates (Table 1). However, the numbers of replicate plants per combination varied depending on the availability of successfully inoculated plant individuals (Figs 1-3). Attempts to clone successfully inoculated plants to increase the number of available replicates failed, because H. europaeus plants have a hierarchical tiller development with roots interlacing tiller bases.
Experimental plants were transferred to 3-l pots on 18 December 2009 and replicate plants were equally distributed among three glasshouse chambers. Within the chambers, the plants were positioned at random and rearranged every 2 wk. All plants were allowed to establish with sufficient water supply for a period of 8 wk under ambient light, supplemented with light from mercury vapor lamps (Hugentobler Spezialleuchten AG, Weinfelden, Switzerland) whenever ambient light conditions were < 15 klx. The day : night rhythm was set at 10 : 14 h with temperatures of 21 : 18°C.
After plant establishment to a size between 7 and 15 tillers (no differences in tiller number were detected between endophyte-infected (E+) and endophyte-free (E−), hybrid endophyte (H) and non-hybrid endophyte (NH) or resident and non-resident endophyte-hosting plants), drought stress was applied to half of the experimental plants for 9 wk. Water application was ceased and soil water content was monitored daily during the establishment of the drought until plants started to wilt, and afterwards every 2–3 d using the TDR. Plants were re-watered at their individual wilting point (below 4% relative soil water content) with 100 ml of tap water individually to account for the different morphologies and sizes of the plants from different accessions (300–800 ml total over 9 wk). For the other plants, which received the well-watered treatment, soil water content was kept at saturation (40–50%). No fertilizer was applied during this first phase of the experiment, hereafter called the ‘drought phase’, to avoid salinity stress in plants exposed to drought, and accounting for a pre-experiment, in which H. europaeus plants had not developed reproductive tissues with high nutrients under glasshouse conditions.
On completion of the drought phase, in April 2010, plant shoots were clipped 1.5 cm above the ground (first harvest, H1). Plants were then left to re-sprout and grow until September 2010. During this second phase of the experiment (called the ‘recovery phase’), the soil water content of all plants was kept at saturation, and plants were fertilized twice with 100 mg N, 44 mg P and 62 mg K including micronutrients (Wuxal, Syngenta Agro AG, Basel, Switzerland).
The sequence of treatments in this experiment – a drought phase followed by a harvest of shoots and a recovery phase –simulates field conditions at dry sites, where plants are most likely to experience drought and herbivory (clipping) in late summer. Shoots then die back during the winter, and new tillers and spikes are formed in spring.
At the first harvest (H1), we measured shoot fresh weight (FW) immediately after harvest, counted the tiller number and measured shoot dry weight (DW) after drying for 72 h at 80°C. We calculated the dry matter content (DMC) as the ratio between the shoot DW and the shoot FW. As the DMC is negatively related to the water content of tissues (DMC = 1 – water content), it indicates the degree of wilting (Vile et al., 2005).
At the second harvest (H2), shoots were cut off at the root transition, and roots were washed free of substrate. We determined the FW and DW of shoots (including leaves and flowering tillers), and DW of roots and seeds. The numbers of tillers, spikes and seeds were recorded. Plant DW was calculated as the sum of the shoot DW and the root DW. We estimated the average DW per seed in plants that had formed spikes, and calculated the root : shoot ratio.
Statistical data analysis was performed with the statistical package R, version 2.15.2 (R Core Team 2012). Initial seedling survival and re-inoculation success were analyzed with binomial models testing the effects of plant accession, endophyte strain, and whether the plant–endophyte genotype combination corresponds to the original (to see whether this is more compatible than new combinations).
Variables describing plant growth and reproduction were logarithmically or square root transformed if necessary to meet the requirements of linear models. As many plants did not form any spikes or seeds, the number of spikes was converted into presence–absence data (whether or not any spikes were formed) and the number and average weight of seeds were only analyzed for plants with spikes. We tested the effects of endophytes and water treatments on plant growth and reproduction using linear models; a binomial model was used for the presence or absence of spikes. Analyses of variance were based on Type II sums of squares to account for unbalanced data (Langsrud, 2003). Plant accession E2 was excluded from the statistical analyses, because only one plant of this accession could be successfully inoculated with a hybrid endophyte (Table 1), so that one treatment combination (hybrid/drought) was missing.
The effects of endophytes were tested in two ways, reflecting our research questions. First, we classified plants according to endophyte infection treatment (endophyte free (E−), hybrid endophyte (H) and non-hybrid endophyte (NH)), as well as into plant accessions originally hosting non-hybrid endophytes (A2, B) and originally hosting a hybrid endophyte (C1). The statistical model included the effects of endophyte type, plant accession type, water treatment and their interactions, as well as individual plant accessions (nested in accession type) and growth chambers as blocking factors. Our first two research questions refer to the effect of endophyte type. If this effect was significant, contrasts (function glht in R package multcomp) were used to test whether there was a difference between endophyte-free (E−) and endophyte-hosting (E+) plants (question 1) or a difference between hybrid-hosting (H) and non-hybrid-hosting (NH) plants (question 2). For test results shown in graphs, contrasts were performed separately by water treatment. The third research question refers to possible interaction effects between endophyte type and plant accession type.
For our fourth research question, we considered only re-inoculated plants and classified them according to the degree of matching between the re-inoculation and the original symbiosis: resident (endophyte strain inoculated into its original plant accession); different genotype (endophyte strain inoculated into the non-resident plant accession of corresponding type, i.e. hybrid or non-hybrid); and alternate type (hybrid endophytes inoculated into non-hybrid plant accessions and vice versa). The statistical model included the effects of re-inoculation type, plant accession, water treatment and their interactions, as well as growth chambers as blocking factors. This model could not be run for reproductive traits, because very few hybrid-hosting plants reproduced at all.
The experiment started with 6450 seeds from endophyte-free mother plants of four accessions. The germination rate was, on average, 46%, so that 2983 seedlings could be re-inoculated. Between 33% and 98% of the seedlings survived up to the start of the actual experiment, and 2–50% of these (260 plants) proved to be successfully infected by the endophyte (Table 1). Seedling survival and infection success differed among plant accessions and fungal strains (Table 1); in addition, infection success was, on average, three times higher for seedlings re-inoculated with their resident endophyte strain than for seedlings re-inoculated with non-resident strains (binomial generalized linear model (GLM), X2 = 44.8, P <0.001). Only plant accession E2 had a very low re-inoculation success even with its resident endophyte.
Drought stress strongly reduced the growth and reproduction of H. europaeus plants (Table 2; Fig. 1). At harvest 1 (H1, drought harvest), shoot DW of plants subjected to drought was decreased, on average, by 66.1% (Fig. 1a). At harvest 2 (H2, recovery harvest), the growth of previously drought-stressed plants was still reduced significantly compared with well-watered plants: tiller number was reduced, on average, by 34% (Fig. 1c), and total DW by 41.4% (P <0.0001, Fig. 1d). Drought increased significantly the DMC (DW : FW) of shoot tissues at H1 (Fig. 1b), but decreased it at H2 (Fig. 1e), and had no influence on the root : shoot ratio (Fig. 1f). A small proportion of plants that had been subjected to drought formed spikes (15.8% vs 49.2%). Seed number per plant was not affected by previous drought, but the average weight per seed was reduced (Table 3).
Table 2. Analysis of variance (ANOVA) results for the effects of Epichloë endophyte infection (uninfected, hybrid or non-hybrid endophyte), drought stress and plant accession type (naturally hybrid or non-hybrid hosting) on vegetative traits of Hordelymus europaeus
DW : FW
DW : FW
Root : shoot
For traits with significant main effects of endophytes, a priori contrasts were used to test the effects of endophyte presence (uninfected vs infected) and endophyte type (hybrid vs non-hybrid). ANOVA results are F values and significance levels for each effect (*, P <0.05; **, P <0.01; ***, P <0.001). Models also included plant accessions (nested in plant type) and growth chambers as blocking factors (not shown). For contrasts, t-values and significance levels are given. DW : FW, dry matter content of shoot tissues (dry weight : fresh weight). All plant tissue weights were measured in grams.
Table 3. Analysis of variance results for the effects of re-inoculation type (resident vs same type or different type), drought stress and plant accession on vegetative traits of Hordelymus europaeus
DW : FW
DW : FW
Root : shoot
Data are F values and significance levels for each effect (*, P <0.05; **, P <0.01; ***, P <0.001). Models also included growth chambers as blocking factors. DW : FW, dry matter content of shoot tissues (dry weight : fresh weight). All plant tissue weights were measured in grams.
Endophyte infection had no significant effect on plant growth during the first phase of the experiment (Table 2, Fig. 1a,b), but influenced significantly tiller production and biomass production during the second phase of the experiment (Table 2, Fig. 1c,d). The effect was related to endophyte presence (Table 2): at H2, the tiller number was 18.2% higher and the total DW was 6.6% higher in plants with endophytes than in endophyte-free plants (Fig. 2). Conversely, there was no significant difference between hybrid- and non-hybrid-hosting plants (Table 2). The effect of endophytes was similar in both water treatments, that is, endophyte–water interactions were not significant (Table 2, Fig. 2). The DMC of shoots and the root : shoot ratio were not influenced significantly by endophyte treatments (Table 2, Fig. 1e,f).
Plant accession types (naturally hosting hybrid or non-hybrid endophytes) differed in DMC at H1 (especially after drought), and in plant weight and root : shoot ratio at H2 (Table 2, Fig. 1b,d,f). However, the effect of endophyte treatments on plant growth did not differ between the two plant accession types: there was no significant endophyte–plant type interaction (Table 2).
When comparing plants re-inoculated with their resident endophyte with those re-inoculated with endophytes from other populations (different or alternate), no significant differences in growth were apparent, that is, plants did not generally produce more tillers or biomass when hosting an endophyte with which they had previously co-existed (Fig. 3). Plants did, on average, have a higher DMC at H2 when hosting their resident endophyte (Table 3, Fig. 3e). However, this effect was dependent on plant accession and was mainly found for accession A2 (Table 3, Supporting Information Fig. S1). Plants from accession A2 also had a 29% greater plant weight at H2 after drought stress when hosting their resident endophyte than when hosting other endophytes (Fig. S1).
Both the proportion of plants forming spikes and the average weight per seed were reduced by drought, whereas seed number was not affected (Tables 4, 5). Spike formation was influenced significantly by endophyte treatments: the proportion of plants producing reproductive tillers was higher among plants hosting non-hybrid endophytes than among plants hosting hybrid endophytes (Tables 4, 5). This difference was slightly more pronounced after drought stress (3.7-fold) than in the control treatment (2.8-fold), although the interaction was not significant. Spike formation was also dependent on plant type, with a significant endophyte–plant type interaction (Table 5): the positive effect of non-hybrid endophytes was stronger in plant accessions A2 and B, which naturally host non-hybrid endophytes (45% of plants produced spikes), than in accession C1, which naturally hosts a hybrid endophyte (only 13.6% of the plants produced spikes). Seed number per plant and average seed weight did not differ significantly among endophyte treatments (Table 5). However, it is worth noting that none of the hybrid-hosting plants produced any ripe seeds after drought stress (Table 4).
Table 4. Reproductive traits of Hordelymus europaeus plants grown with or without drought stress and with different Epichloë endophyte infection treatments
Plants with spikes (%)
No. of seeds per plant with spikes
Mean weight per seed if spikes present (mg)
Data for seed number and seed weight are means and standard errors (calculated from log-transformed data for seed number). na, data not available.
15.2 (+ 5.06)
7.0 (± 0.44)
17.7 (+ 15.2)
7.1 (± 0.76)
13.1 (+ 2.9)
6.8 (± 0.21)
8.7 (+ 3.4)
6.3 (± 0.77)
0.0 (n =1)
14.2 (+ 5.9)
5.4 (± 0.51)
Table 5. Analysis of deviance or analysis of variance results for the effect of Epichloë endophyte infection (uninfected, hybrid or non-hybrid endophyte), drought stress and plant accession type (hybrid or non-hybrid hosting) on reproductive traits of Hordelymus europaeus
Formation of spikes
No. of seeds if spikes present
Average seed weight if seed present
Data are chi-squared values (spikes) or F values (seed number, seed weight) and significance levels for each effect (*, P <0.05; **, P <0.01; ***, P <0.001). Models also included plant accessions (nested in plant type) and growth chambers as blocking factors. Some interactions could not be tested because of missing treatment combinations. Seed weight was measured in milligrams. na, data not available.
Hordelymus europaeus represents an ideal host to test the hypothesis of a niche expansion mediated by hybrid endophytes, because natural populations can be infected by several hybrid or non-hybrid endophyte taxa (Oberhofer & Leuchtmann, 2012). This is the first study of a wild grass using artificial inoculations to directly compare the effects of the two endophyte types on their host. The success rate of inoculations (germination rate × infection rate × survival rate) was generally low (1–24%), comparable with results reported from previous studies on other wild host–endophyte systems, such as Bromus spp. (Brem & Leuchtmann, 2003). Although infection rates were significantly higher in resident combinations consistent with co-adaptation, the survival rate of inoculated seedlings did not differ between resident or other combinations, suggesting equal compatibility between the different endophyte strains and host accessions. Low infection success is therefore not related to host–endophyte incompatibility, but may be explained by the inefficiency of the inoculation technique used, resulting in a limited number of plants that were available for the experiment (Table 1). This clearly reduced the statistical power for testing specific endophyte effects. Ideally, further experiments should start with even larger seed numbers, or should clone successfully inoculated seedlings to increase plant numbers. It is not yet known whether infections in non-native host–endophyte genotype combinations will be vertically transmitted to the next generation. If infections persist, the propagation of infected plants could also be performed using seeds.
By growing plants in different water treatments, we could test whether the effects of hybrid vs non-hybrid endophytes were dependent on the environment (Cheplick & Faeth, 2009). We identified drought as a potential selective pressure, because H. europaeus populations with different endophyte types occur in habitats that differ in soil moisture. Furthermore, several studies have reported previously endophyte mediated drought amelioration in non-agronomic host–endophyte systems (Brem & Leuchtmann, 2001; Gonthier et al., 2008; Kannadan & Rudgers, 2008; Craig et al., 2011). However, the generality of this effect and the underlying mechanisms are not yet established.
Regarding the first question (see the 'Introduction' section), endophyte presence had no effect on plant growth after the drought phase, but stimulated significantly re-growth during recovery (Table 2; Fig. 2). For drought-exposed plants, these results confirm the findings from studies on agricultural grasses, and support the view that endophyte benefits are greater during recovery from drought than during actual water deficit (West, 1994; Malinowski & Belesky, 2000; Hesse et al., 2003). It has been suggested that endophyte presence enhances the osmotic adjustment of plant tissues when exposed to drought, a mechanism which may protect the apical meristems and can improve the survival of tillers (Elmi & West, 1995). However, faster re-growth of infected plants was observed independent of drought treatment in this study, suggesting that the interaction between H. europaeus and its endophytes is generally mutualistic, as found for other natural endophyte–host grass systems (e.g. Iannone et al., 2012). Growth enhancement of plants has been attributed to auxins produced by endophytes (De Battista et al., 1990). These phytohormones may influence cell division, cell elongation and resource allocation within plants. In our study, the increased DW of infected plants could be entirely attributed to greater tiller numbers, that is, the average weight per tiller was not increased by endophytes (data not shown). A recent study has related higher tiller numbers in infected plants to an endophyte-induced change in the balance between auxin-controlled apical dominance in shoots towards cytokinin-promoted outgrowth of axillary buds (Eaton et al., 2011). Our specific kind of growth increase differs from that found by Iannone et al. (2012), where endophyte infection of Bromus auleticus primarily increased the size of plants. The type and amount of auxins produced vary among endophyte taxa (De Battista et al., 1990) and may alter the ratio to cytokinins differently, which may explain the contrasting results in native grasses.
Our hypothesis addressing the second research question was that hybrid endophytes might provide a selective advantage to their hosts over hosts infected by non-hybrids for two main reasons. First, hybrid endophytes were consistently observed to occur with higher frequency in natural grass populations, for which the proposed selective advantage could provide an explanation (Moon et al., 2004; Iannone et al., 2009; Oberhofer & Leuchtmann, 2012), because they carry more diverse genes derived from two different ancestors (Clay & Schardl, 2002; Schardl & Craven, 2003). In our experiment, this should become apparent as a greater ability of hybrid-hosting plants to maintain their performance in both treatments. Possible mechanisms would include a greater plasticity in biomass allocation, tissue structure or osmotic potential, allowing plant performance to be optimized under a broader range of moisture conditions. Second, reproductive isolation and a lack of recombination in asexual endophytes may have resulted in genome degradation (Brem & Leuchtmann, 2003; Moon et al., 2004), an effect that might be reversed in hybrid endophytes by the uptake of a second genome from sexually recombining species (Moon et al., 2004).
These expectations were not confirmed by the present study. Plants re-inoculated with hybrid and non-hybrid endophytes did not differ significantly in plant growth and biomass allocation. Vegetative traits were similarly reduced under drought, and none of the studied traits was more plastic with hybrid endophytes. Presuming that growth hormones produced by the endophyte induce growth differences, they may be inherent to all epichloid endophytes and therefore cause comparable effects in hybrid- and non-hybrid-infected plants. Likewise, a previous study has reported identical plant growth response to different auxin concentrations produced by endophytes in planta (De Battista et al., 1990).
The effects of genome degradation are either negligible in asexual non-hybrid endophytes or may not be effectively reversed by hybridization, as no difference between endophyte types was found. The stability and extent of hybrid zones depend on the fitness and the dispersal rate of hybrids compared with their ancestors (Arnold, 1997). In the setting of our experiment, no fitness advantage became evident, although it cannot be excluded that other biotic or abiotic factors may favor hybrid infections in grasses, which should be examined in future studies. Higher dispersal rates of hybrid endophytes may result from high vertical transmission rates, as each individual of the original populations was infected (Table 1). However, a 100% infection rate was also observed in population B containing non-hybrid endophytes. Only population A2 had a low infection frequency, which is known from other endophytes with a balanced transmission mode (Leuchtmann & Clay, 1997). Alternatively, the high transmission mode of hybrid endophytes may result from horizontal transmission of propagules, such as conidia formed on epiphyllous mycelia. These have been found by several researchers in wild grass species (Moy et al., 2000; Moon et al., 2002) and have also been observed in H. europaeus (Oberhofer, 2012).
Plant accession types
Referring to the third research question, interactions between plant accession type and endophytes were lacking in vegetative traits (Table 2), suggesting that plant accession types are not genetically differentiated depending on whether they originally hosted a hybrid or non-hybrid endophyte. This finding conforms to the overall compatibility of endophyte strains and plant accessions. However, spike formation was higher in non-hybrid plant accessions, especially when inoculated with a non-hybrid strain, which confirms our results on co-adaptation predominating in non-hybrid accessions and their native endophyte strains. Significant interactions between plant accessions and re-inoculation type may result from better growth performance of resident combinations in non-hybrid accessions, but a general trend in plant response to the different re-inoculation types is missing (Fig. S1).
Concerning the fourth research question, successful inoculations of all attempted combinations among plant accessions and endophyte strains of H. europaeus suggest that different endophytes are not specialized to particular host populations. This may be unexpected, as genetic and morphological variations among populations from central Europe are known to be considerable (Mizianty et al., 2007), while our populations originated from a wide geographical range across Europe. However, resident combinations showed the highest infection rates of all re-inoculations, suggesting the presence of a weak co-adaptation between host accessions and their original endophyte strains. In our experiment, originally non-hybrid-hosting accessions (A2, B), which naturally occurred at dry sites, had a larger root : shoot ratio. This is a typical plant adaptation to drought, which illustrates the potential for local adaptation in H. europaeus. Therefore, the weak host-endophyte co-adaptation suggests that the observed distribution of endophytes within the geographical range of H. europaeus may reflect a relatively recent colonization history of the host and its endophytes (Oberhofer & Leuchtmann, 2012). Not only were all host–endophyte genotype combinations compatible, they also performed similarly in terms of growth and seed production, that is, hosts inoculated with their resident endophyte did not perform better, generally, than hosts inoculated with other endophytes. The only exception was a higher DMC specifically found in plants from accessions A2 and B when infected by their resident non-hybrid endophytes, reflecting a tendency for these resident combinations to produce more reproductive tillers. Seed-producing tillers dry after seed ripening in grasses (Parsons & Robson, 1981) explaining the higher DMCs in these accessions.
When considering single plant accessions, the naturally non-hybrid-hosting accession A2 produced more biomass when combined with its resident endophyte, which would be consistent with a pattern of co-adaptation between host and endophyte. Interestingly, this non-hybrid endophyte is the only endophyte of H. europaeus that is capable of sexual reproduction, at least in cultivated plants (Oberhofer & Leuchtmann, 2012). Despite its beneficial effects on its host, the original population had a low infection rate, confirming recent findings that infection rates of wild grass populations depend not only on the relative fitness of the host, but also on the transmission rate (Gundel et al., 2008, 2011; Gibert et al., 2012). Co-evolution in other host–endophyte systems has been mainly detected in endophytes expressing a balanced sexual reproduction (such as the endophyte from accession A2) and has been related to common descent of hosts and their endophytes (Schardl et al., 1997). This would confirm our view originating from biogeographical considerations that the symbiosis between host and endophyte of accession A2 is particularly ancient (Oberhofer & Leuchtmann, 2012).
With regard to reproduction, non-hybrid endophytes increased seed production of their hosts, whereas hybrid endophytes reduced it, suggesting a fitness advantage for plants hosting non-hybrid endophytes. The loss of fitness was even more pronounced in relative terms for plants hosting hybrid endophytes, because they completely failed to produce ripe seeds after drought exposure. Although we cannot exclude that seed set was just delayed in hybrid-hosting plants, our experiment does suggest that seed formation was strongly reduced, as seeds formed after September would probably not ripen any further in the Central European climate. It thus seems likely that hybrid-infected plants and their hybrid endophytes would fail to reproduce in their first year of life when exposed to drought during the growing season. By contrast, the ability of plants hosting non-hybrids to form spikes and seeds after drought may explain why such plants were observed to occur in drier habitats. As plants with non-hybrid endophytes produced more reproductive tillers also in well-watered conditions, hybrid endophytes do not appear to provide a fitness advantage over non-hybrids. These findings are opposite, but still consistent, with those found in the F. arizonica system, where hybrid-infected plants are naturally restricted to dry sites and are more competitive under low nutrient and drought conditions (Saari & Faeth, 2012).
Overall, endophyte infection was found to be beneficial for all accessions of H. europaeus plants based on the growth parameters measured in the recovery phase. Our findings do not support the proposed niche expansion of hybrid-infected plants, nor can they explain the prevalence of hybrid endophytes in natural grass populations. On the contrary, non-hybrid endophytes of H. europaeus appear to convey more benefits to their hosts in terms of reproduction, particularly after drought. The absence of incompatibility of any host–endophyte genotype combination, together with only weak indications of co-adaptation, suggest a recent evolution of diversity in endophyte taxa found in H. europaeus. However, the relatively small sample sizes used (because of the low inoculation success) imply that our conclusions are provisional and need to be confirmed by further experiments, ideally including a larger number of host and endophyte accessions.
We thank the administration of the Parco Nazionale del Pollino (Italy) and the Länsstyrelsen Kalmar Län (Sweden) for permission to collect seed material in their territories, which was propagated to create the plant material for this study. We acknowledge suggestions by S. Karrenberg on the experimental set-up and the plant maintenance provided by M. Frei. Further, we would like to thank the team of gardeners and the administrative personnel for use of the glasshouse facilities of the ETH research station in Lindau-Eschikon, Switzerland. C. L. Schardl and S. H. Faeth provided helpful suggestions on earlier versions of the paper. Finally, we would like to acknowledge precious advice provided by three anonymous referees. This research was supported by a grant from the Swiss National Science Foundation (31003A-117729).