Genetically based vertical transmission drives the frequency of the symbiosis between grasses and systemic fungal endophytes

Authors


Correspondence author. E-mail: anais.gibert@gmail.com

Summary

  1. Understanding the variation in hereditary symbiont frequency among host populations is a prerequisite to predict symbiont fixation processes. However, the mechanisms driving this variation remain elusive. Exploring the mechanisms responsible for the observed variability on an ecological time scale requires simultaneous study of fitness differentials between symbiotic (S) and non-symbiotic (NS) hosts, and the symbiont transmission rates to host offspring.
  2. We studied these two key mechanisms using a grass–endophyte symbiosis in the alpine grass Festuca eskia. Plants from four native populations varying in endophyte frequencies (ranging from 13% to 100%), and environmental conditions (water availability, and grazing pressure) were transplanted in a common garden. Soil nutrient levels were manipulated to assess genetic and environmental differences within and among populations in fitness-related traits (i.e. traits linked to clonal growth, sexual reproduction and resource acquisition).
  3. A fitness differential favouring S over NS plants was detected in all studied populations: in heavily grazed populations, sexual reproduction was higher in S compared with NS plants, whereas in minimally grazed populations, clonal growth increased. Results showed a positive correlation between endophyte transmission rates and population endophyte symbiotic frequencies. The population endophyte transmission rates were not affected by soil resource level.
  4. According to selection pressures acting in each population, symbiotic plants appear to perform better in all F. eskia populations. The correlation between endophyte frequencies and transmission, and the positive effect of S on NS plants under our experimental conditions, indicated a predominant role of endophyte transmission in endophyte frequency variation in F. eskia. The endophyte transmission rate variation is genetically based at the population level and can be explained by a trade-off with a specific host trait subjected to strong selection; here, we suspected traits linked to plant resource acquisition.
  5. Synthesis: Our study provides evidence for the (i) dominant role of endophyte transmission and its responsibility for endophyte frequency variation in a native grass when compared with fitness differential process between S and NS plants at an ecological time scale, and (ii) genetically based variation in endophyte transmission rates. We also confirm the population specificity of positive endophyte effects in a native grass.

Introduction

Symbioses have been responsible for the origin of major ecological and evolutionary transitions in the history of life. One of the most striking examples is the eukaryotic cell, which gained the capacity for oxygenic photosynthesis by fixing the cyanobacterial ancestors of plastids 1–1.5 billion years ago (Rodríguez-Ezpeleta et al. 2005). Although current research provides us with a better understanding of symbioses establishment and persistence (Douglas 2010), loss or fixation over generations of a symbiont in host populations cannot yet be accurately predicted. Heritable symbioses, which include some of the most ecologically abundant symbioses on earth, for example, grass-fungal endophytes (see Leuchtmann 1992; Cheplick & Faeth 2009) and arthropod-Wolbachia bacteria (see Werren, Baldo & Clark 2008), are likely to be intermediate states towards symbiosis fixation in hosts (Herre et al. 1999; Schardl & Selosse 2007; Gundel, Rudgers & Ghersa 2011b): they are vertically transmitted to host offspring, but the frequency of the symbiont in host populations varies. A better understanding of symbiont fixation processes will be gained by the study of the mechanisms involved in hereditary symbiosis variation at population level.

Theoretically, two primary mechanisms might contribute to hereditary symbiont variation at the population level: the impact of symbioses on host fitness and symbiont transmission rate to offspring. Most field studies have focused attention to the symbiosis effects to host fitness, while symbiont transmission rate has been overlooked (Herre et al. 1999; Bright & Bulgheresi 2010; Gundel, Rudgers & Ghersa 2011b). Transmission is often considered one mechanism underlying the biological success of symbioses, favouring mutualism when it is exclusively vertical (e.g. Fine 1975; Ewald 1987). Studies considering these two variables (i.e. host fitness and vertical transmission) will shed light on the underlying mechanisms to the variation in hereditary symbiosis and ultimately the symbiont fixation.

Here, we use this framework to understand the underlying mechanism to the variation in grass–endophyte symbiosis in nature. This hereditary symbiosis between cool-season grasses (Poaceae) and Neotyphodium fungal endophytes (asexual forms of Epichloë- Clavicipitacea, Ascomycota) has been reported to influence host species ecology and evolution (Clay 1990; Schardl & Selosse 2007; Gibert et al. 2012b), and ecosystem functioning (e.g. Clay & Holah 1999). It is considered to be a classic plant–microbe mutualism example (e.g. Schardl 2001; Clay & Schardl 2002; Douglas 2010). The host may receive increased resistance to stresses, such as herbivory and drought from the endophyte (Clay & Schardl 2002; Malinowski & Belesky 2006), and in return, the fungus receives nutrition, protection and dispersion from the host plant. The endophytic fungus grows asymptomatically in aerial grass tissues and is vertically transmitted from the plant to its offspring via seeds and tillers. Variability in symbiont frequencies in grass populations has commonly been observed in wild environments (e.g. Rudgers et al. 2009; Gonzalo-Turpin et al. 2010). Several authors have recently emphasized the need for simultaneously evaluating the endophyte effect on host fitness and endophyte transmission rates to understand variations in the endophyte frequency of wild grass populations (e.g. Davitt, Chen & Rudgers 2011; Gundel, Rudgers & Ghersa 2011b; Gibert, Magda & Hazard 2012a).

The common approach used to estimate the effect of endophytes on a host grass is based on comparing performance of hosts with the same genotype (or group of genotypes) with and without the fungal endophyte; i.e. artificially removed or added (see Cheplick & Faeth 2009). This approach, although straightforward to establish endophyte host benefits, is probably not suitable to characterize the processes resulting in endophtyte frequency variability in natural systems. Indeed, in a given population, the endophyte frequency corresponds to the fitness differential outcome between naturally symbiotic (S) and non-symbiotic (NS) plants, not S plants fitness differential with and without their endophyte. This is especially the case in the absence of differential crossing between S and NS plants within a population, i.e. no neutral genetic differentiation observed between S and NS plants (see for Lolium perenne Gibert et al. 2012b; and for Festuca eskia Gonzalo-Turpin et al. 2010). Consequently, while the large majority of studies on endophyte effect on host fitness have documented a list of benefits attributed to these fungi (see Cheplick & Faeth 2009), very few have produced pertinent results to understand variation in endophyte frequencies (e.g. Gibert, Magda & Hazard 2012a).

Moreover, there is a lack of biological data to test endophyte transmission rates within natural grass–endophyte systems. Several authors revealed the importance of endophyte transmission rates using modelling approaches and found it an integral mechanism underlying the variation in symbiotic plant frequency (Ravel, Michalakis & Charmet 1997; Saikkonen, Ion & Gyllenberg 2002; Gundel et al. 2008). However, endophyte transmission rate is investigated far less often than an endophyte's effect on host fitness (Gundel et al. 2008; Gundel, Rudgers & Ghersa 2011b). The few studies performed so far suggest that endophyte transmission might be a specific host trait, rather than a consequence of environmental differences. Indeed, endophyte transmission failures reported at different phenological stages and in different endophyte/grass associations (Afkhami & Rudgers 2008; Canals, San Emeterio & Oreja 2008), appear not to be affected by resource availability (e.g. Davitt, Chen & Rudgers 2011; Gundel, Rudgers & Ghersa 2011b; Gibert, Magda & Hazard 2012a).

The objective of our study was to determine the relative importance of fitness differential between S and NS plants, and the endophyte transmission rates, in driving endophyte frequency variation in host populations. We concurrently estimated these two processes in natural populations of the alpine grass F. eskia. In this species, Gonzalo-Turpin et al. (2010) did not detect significant differences in neutral genetic diversity between S and NS plant groups across populations; therefore, S and NS plants within each population can be considered as a single population. We tested how these two processes vary based on variable levels of endophyte frequency within populations. Plants from four populations exhibiting endophyte frequencies ranging from 13% to 100% were grown in a common garden under three water and nutrient levels to assess genetic and environmental differences within and among populations in fitness-related traits between S and NS plants, and in endophyte transmission rates. The populations were chosen along a water availability gradient (Gonzalo-Turpin et al. 2010) with different grazing histories (heavy vs. light grazing pressures).

Materials and methods

Study Organisms

Festuca eskia Ram. (Poaceae) is a perennial grass endemic to the Pyrenees and Cantabric mountains, which forms the dominant habitat structure for many subalpine and alpine species-rich communities over 1500 m in the Pyrenees (Tosca 1996). Festuca eskia is primarily distributed on acidic soils, snow-covered combs and scree. The species harbours an asexual form of the endophytic fungus Epichloë festucae, with population dependent endophyte frequencies ranging from 0% to 100% (Gonzalo-Turpin et al. 2010).

Epichloë festucae Leuchtmann, Schardl and Siegel (Ascomycota: Clavicipitaceae) is a fungal endophyte that commonly infects cool-season grasses (Schardl 2001). Because there are no records of stroma in F. eskia, E. festucae appears to be an asexual form (pers. obs). Epichloë festucae has been shown to induce drought and cutting stress resistance in F. eskia (Gibert & Hazard 2011). The term ‘endophytic symbiosis’ is now used in reference to the E. festucae/F. eskia relationship.

Population Sampling

Festuca eskia populations evaluated in this study are distributed across the Pyrenean massif that forms a natural border between France and Spain. We sampled four natural F. eskia populations along 350 km of the Pyrenean massif during spring 2007. The populations are distributed along a climatic gradient from east to west (i.e. xeric to mesic, respectively), experiencing Mediterranean, mountainous to oceanic influences. The populations were selected from previously documented endophytic symbiosis frequencies (Gonzalo-Turpin et al. 2010) and distinctive environmental conditions (see Table S1).

Plants were collected from each population; nine naturally symbiotic (S) and nine NS, with the exception of the Puymorens population. At this site, only nine S plants were sampled because naturally NS plants were not detected in the population. In the Puymorens population, we experimentally generated nine NS individuals (E-) by removing the endophyte from S plants (see following subsection). We sampled at ≥ 5 m intervals to prevent the repeated collection of the same genetic individual (i.e. via clonal reproduction). Each plant was propagated into three clones (a total of 216 plants). In July 2008, individual clones were grown in pots for 3 months in a nursery located at 1500 m a.s.l. This protocol provided a reduction in the environmental origin conditioning effect, before translocation to the experimental site. After translocation, we used only neoformed tillers in the experiment to avoid maternal effects.

Plant endophyte detection was performed using the Phytoscreen tissueprint immunoblot kit (Agrinostics Ltd, Inc., Watkinsville, GA, USA). Epichloë festucae presence on plant tissues was confirmed in a sample subset by microscopic observation following Hiatt et al. (1999). In addition, PCR analyses using ®-tubulin sequences were employed to definitively establish the endophytic fungus as E. festucae (Navaud and Gryta, data not shown) and the efficiency of the immunoblot kit in endophyte detection on F. eskia. E- plants in the Puymorens population were derived from seeds collected in 2006 and treated with an incubation method subjecting the seeds to heat and humidity (Harvey, Fletcher & Emms 1982; Siegel et al. 1984). Disinfected plants were verified for endophyte elimination with Phytoscreen tissueprint immunoblot kit and maintained in the nursery.

Common Garden Experiment

The common garden was located at Guzet, one of the population study sites (see site characteristics, Table S1). The experimental site was natural scree in primary succession (not yet colonized by local vegetation). In October 2008, when clones reached five-six tillers, they were planted directly to common garden soil. The experimental design comprised three blocks, and each clone was randomly attributed to one of each block. Plants were planted 40 cm apart in a block.

Each individual clone was also randomly assigned to one of three treatments; therefore, each block comprised clones from different plants with different treatments. The three following treatments were applied in 2009 and 2010: (i) control condition (C), (ii) water addition (W), plants received 1.5 L of water every 2 weeks during the vegetative growth period, and (iii) water and nutrient addition (W+N), plants received 1.5 L of nutrient solution (N:P:K: 19:2.6:10) at the same frequency as the water addition treatment. Water and nutrient solution diffused regularly during the 2 weeks at the tussock base. The study area was fenced to protect the plants from large herbivores. All plants survived during the 2 years of our study.

Plant Trait Measurements and Endophyte Transmission Rates

Shoot biomass was chosen to represent clonal growth in our experiment. It was positively correlated with the number of vegetative tillers (r²: 0.75; Pearson test P = 2.2e−16). In October 2010, all plants were harvested, and shoot biomass was measured in the laboratory following oven drying for 24 h at 80 °C. The traits used to establish reproductive success included the proportion of flowered plants by population, spike number per plant and reproductive allocation (RA). In August 2010, the number of spikes and vegetative tillers was counted. RA was determined as spike number per vegetative tiller. Functional traits associated with plant resource acquisition, specific leaf area (SLA, mm² mg−1) and leaf dry matter content (LDMC, mg g−1) were measured in September 2010 on five leaves per tiller according to Cornelissen et al. (2003). In addition, SLA and LDMC were measured from the four studied population origin sites following the same methodology; 20 plants were field collected and subsequently measured from each population. These plants were screened for endophyte presence with Phytoscreen tissueprint immunoblot kit.

Endophyte transmission from plant to seeds includes the rates at which the endophyte colonizes tillers, young seedlings and developing seeds (see Gundel et al. 2008). Here, endophyte transmission rate from plant to seeds was established using 15 seeds per each S plant panicle. We applied a dilute methylene blue stain that detected the endophyte regardless of seed viability. For each population, we averaged the proportion of endophyte-infected seeds per plant. Moreover, endophyte transmission rate from plant to tillers was specifically assessed because clonal growth is successful in generating new individuals in F. eskia (Tosca 1996), and there is no evidence of a linear correlation between endophyte transmission from plant to tillers and from plant to seeds. Consequently, in F. eskia, it was necessary to dissociate plant to tillers endophyte transmission from plant to seeds transmission. We tested endophyte occurrence on 20 new tillers that emerged from the initial ramet in each plant, using the Phytoscreen tissueprint immunoblot kit (Agrinostics Ltd.). Tested tillers were collected following seed production in 2010. For each population, we averaged the proportion of endophyte-infected seeds or tillers per plant.

Data Analysis

All analyses were performed using S-Plus® (Version 2.4.0, R Foundation, 2006). Because some traits were recorded on individual plants during the same observation period, the traits might not be independent. Consequently, multivariate analysis of variance (manova) was applied to analyse significant differences among phenotypes and check for consistency of effects across different traits. manova can also be used for multiple dependent, correlated variables. We performed three separate manovas, on three sets of traits: (i) traits for clonal growth and sexual reproduction (shoot biomass, spike number, flowered plant percentage RA), (ii) traits for resource acquisition (SLA and LDMC measured under common garden conditions), and (iii) endophyte transmission rates (from plant to seeds and from plant to tillers). When manova detected significant differences, and also for SLA and LDMC measured under natural conditions, we tested each dependent trait using analysis of variance (anova) and analysis of deviance. The models were based on three or four fixed factors: water and nutrient treatments, population (i.e. geographical origin), endophytic status and block. The models considered all interactions between population, treatment and endophytic status. Endophyte transmission rates and the flowered plant percentage were generated as percentages, and therefore analysed by a generalized linear model assuming a Poisson error distribution, and applying a log-linear link function. Analyses met assumptions of normality of residuals and homogeneity of variances. Consequently, post hoc Tukey's tests were performed to detect any significant differences among means using SPLUS. Pearson's correlation was performed to detect correlations between average endophyte transmission rates and endophytic frequencies per population.

Correlations between RA and shoot biomass were investigated to verify whether differences in sexual reproduction according to population and/or endophytic status were not the result of differences in plant size. Analysis of covariance (ancova) was conducted on flowering plants to assess the effects of population, endophytic status and interactions on the relationship between RA and shoot biomass (Table 2a). This analysis was performed on the W+N treatment data because these conditions induced a plant shift from the vegetative to reproductive stage (Table S2), resulting in the production of spikes. Analyses met assumptions of normality of residuals and homogeneity of variances. Consequently, post hoc Tukey's tests were conducted to detect significant differences between intercept and slope (Table 2b).

Results

Phenotypic Structure of F. eskia Based on Treatments and Populations

Festuca eskia exhibited high levels of phenotypic variation. Traits for clonal growth and sexual reproduction were primarily driven by the treatment effect; in contrast to traits associated with resource acquisition (SLA and LDMC), which discriminated based on population (Table 1). W+N plants were significantly larger and exhibited an increased spike number than W and C plants, with 3-fold higher shoot biomass and 26-fold more spikes in plants during flowering. W+N addition elicited a shift from the vegetative to reproductive stage, with an increased number of flowering plants from 26.5 ± 5% under C and W treatments to 83 ± 4% under W+N addition treatment. Plants from C and W addition treatments were not significantly different, with the exception of the number of spikes produced. The control treatment plants generated 3-fold more spikes than plants under water addition treatment (Table S2).

Table 1. Results of manova, anova and analysis of deviance for Festuca eskia fitness-related traits in natural populations measured under the common garden experiment, or natural habitat (loc): (a) plant clonal growth and sexual reproduction traits; (b) plant resource acquisition traits; and (c) endophyte transmission rates. The effect of population origin, endophytic status, water and nutrient treatments (W+N treatments), all interactions and block were examined. Three manova were performed: (i) on traits for plant clonal growth and sexual reproduction (shoot biomass, log of plants flowered, no. of spikes and reproductive allocation), (ii) on traits for resource acquisition (SLA and LDMC), and (iii) on endophyte transmission rates (from plant to tillers, from plant to seeds)
(a)
Source of variation manova anova and analysis of deviance
Traits for plant clonal growth and sexual reproductionShoot biomassPercentage of flowered plantsNo. of spikeReproductive allocation
d.f.Approximate F P valued.f.MS P valueDeviance P valueMS P value MS P value
A. Population origin33.8 0.000 32698 0.000 2.910.2767097 0.009 0.0780.158
B. Endophytic status16.3 0.000 1854 0.044 5.57 0.007 11689 0.011 0.678 0.000
C. W+N treatments224.4 0.000 29088 0.000 89.8 0.000 40028 0.000 0.463 0.000
D. Block10.60.6651160.7780.150.64928330.2090.0410.334
A × B32.7 0.001 31340 0.000 19.2 0.000 5430 0.030 0.118 0.051
A × C62.4 0.000 61556 0.000 6.950.1635506 0.007 0.0580.255
B × C22.8 0.006 26120.0540.0050.9977662 0.015 0.315 0.001
A × B × C61.8 0.008 6619 0.008 7.010.1594795 0.016 0.172 0.001
Residuals189 189207 - 1787 0.044
(b)
Source of variationTraits for acquistion of ressourcesSLALDMCSLAlocLDMCloc
d.f.Approximate F P-valued.f.MS P-valueMS P-valued.f.MS P-valueMS P-value
A. Population origin313.7 0.000 32346 0.000 4392 0.000 32333 0.000 13998 0.012
B. Endophytic status10.40.69711.220.9202880.50913730.3431540.833
C. W+N treatments23.9 0.004 2404 0.038 17040.078-----
D. Block11.30.26012.080.89618810.093-----
A × B30.90.43932020.1779620.22721630 0.025 5190.861
A × C61.30.18961820.1817000.386-----
B × C20.40.8172480.6712930.642-----
A × B × C61.20.26961710.21412260.089-----
Residuals189 189121 659 48407 3461
(c)
Source of variationEndophyte transmission rate (ETR)ETR from plant to tillersETR from plant to seeds
d.f.Approximate F P-valued.f.Deviance P-valueDeviance P-value
  1. d.f., degrees of freedom; MS, mean square; Deviance, deviance residual P < 0.05 in bold; SLA, specific leaf area; LDMC, leaf dry matter content. ETR, endophyte transmission rate.

A. Population origin344.4 0.000 335.9 0.000 7.21 0.000
C. W+N treatments20.20.93720.0050.8810.0010.996
D. Block20.90.43520.0180.6330.0540.488
A × C60.30.98860.0020.9990.0060.999
Residuals--------

Traits linked to resource acquisition (SLA and LDMC) discriminated principally according to population. Guzet and Puymorens populations exhibited a 27% lower SLA and 14% higher LDMC relative to the Ansabère and Rioumajou populations. SLA and LDMC values did not differ when measured in the common garden or natural habitat (Fig. 1). W+N addition induced a significant increase in SLA and a decrease in LDMC, but the population ranking remained unchanged: Guzet and Puymorens populations exhibited lower SLA and higher LDMC compared with Ansabère and Rioumajou populations.

Figure 1.

Traits linked to resource acquisition in four populations of Festuca eskia: specific leaf area (SLA) and leaf dry matter content (LDMC) measured under common garden conditions (Mean ± SD). Different letters indicate significant differences according to Tukey's post hoc tests.

Population-specific Effect of Endophytic Symbiosis on Traits for Clonal Growth and Sexual Reproduction in F. eskia

All traits for clonal growth and sexual reproduction exhibited a significant endophytic status effect according to manova and anova (Table 1). Symbiotic plants showed increased shoot biomass (S: 32.5 ± 2 mg vs. NS: 28.3 ± 2 mg), produced more spikes (S: 36.6 ± 6 vs. NS: 27.7 ± 5), more numerous plants flowered (S: 57 ± 5% vs. NS: 41 ± 5%) and finally exhibited higher RA (S: 0.19 ± 0.03 vs. NS: 0.07 ± 0.01) than NS plants.

However, these results must be interpreted with caution based on significant three-way (population × endophytic status × treatments) or two-way interactions (population × endophytic status for the percentage of flowered plants) also recorded for clonal growth and sexual reproduction traits (Table 1). All these traits discriminated differently between S and NS plants according to population (significant population × endophytic status interaction). The population × endophytic status × treatment interactions corresponded to significant differences between S and NS within and among populations exclusively under the W+N condition (Table 1, Fig. 2). In the Ansabère population, S plants exhibited an increased number of spikes (+54%) compared with NS flowering plants, resulting in a significantly higher RA under W+N addition (+66%, P < 0.01, Fig. 2). Similarly, in the Guzet population, results showed S flowering plants produced more spikes (+33%) than NS flowering plants, resulting in significantly higher RA under W+N addition (+58%, P < 0.01, Fig. 2). However, in the Rioumajou population, S flowering plants had a 2-fold lower RA than NS flowering plants under W+N addition. A higher shoot biomass was responsible for the lower RA (+49%, P < 0.01, Fig. 2). In the Puymorens population, E- plants experienced a dramatic reduction in all fitness-related traits compared with S plants. E- plants exhibited decreased shoot biomass, spike production (−32%), percentage of flowered plants (−4%) and RA (−99%) relative to S plants (Fig. 2).

Figure 2.

Phenotypic differences in clonal growth (shoot biomass) and sexual reproduction (reproductive allocation, number of spikes and percentage of plants flowered) between symbiotic (S) and non-symbiotic (NS) or disinfected (E-) plants in four native Festuca eskia populations along the Pyrenean massif. All traits (Mean ± SD) were measured under water + nutrient treatments in a common garden experiment, with the exception of percentage of plants flowered calculated under all treatments (control, water, and water + nutrient conditions). The dark area in each circle represents endophyte frequency in the host population. RA: reproductive allocation.

Reproductive allocation summed up differences between S and NS plants in all populations. RA was negatively correlated with shoot biomass (r 2: 0.27, P = 0.002), which indicated allocation to sexual reproduction decreased with plant size. ancova showed significant (P < 0.01) effects of shoot biomass, population origin, endophytic status and population × endophytic status interactions on RA (Table 2a), but an interaction effect with shoot biomass was not detected. Therefore, origin ordinates were different among populations and between endophytic status, but not slopes. Consequently, for a given shoot biomass and a given population, S plants were provided a significantly different RA than NS plants, with the exception of the Rioumajou population (Table 2b).

Table 2. Correlation between reproductive allocation and shoot biomass in Festuca eskia a) ancova to examine shoot biomass influence, population origin, endophytic status and all interactions on reproductive allocation b) Parameter estimates (intercept and slope) of linear regressions between RA and shoot biomass, for all populations × endophytic status interactions
Reproductive allocation InterceptSlope
a) Results of ancova d.f.MSP-valueb) Parameters estimates of linear regressions for the population × endophytic status interaction
  1. S, Symbiotic plants; NS, Non-symbiotic plants; E-, Disinfected plants.

  2. Different letters indicate significant differences according to Tukey's post hoc tests. Data were collected in a common garden experiment on reproductive individuals under water + nutrient treatment. This treatment induced a shift from the vegetative to the reproductive stage in the plant individuals. In bold: intercept and slope significantly different from 0.

A. Shoot biomass10.8615 0.009
B. Population origin30.3644 0.034 Ansabère S 1.14a 0.02
C. Endophytic status10.6984 0.018 Ansabère NS0.36bc 0.005
B × C30.3286 0.048 Rioumajou S0.16c 0.0001
Rioumajou NS 0.58 b 0.013
A × B30.05370.711Guzet S 1.37 a 0.02
A × C10.01050.765Guzet NS 0.67 a 0.01
A × B × C20.09250.458Puymorens S 1.01 a 0.01
Puymorens E-NANA
Residuals480.1126

Endophyte Transmission Rates

Population influenced endophyte transmission rates from plant to tillers and from plant to seeds (Table 1), but not water and nutrient treatments (Table S2). Endophyte transmission rate averages for the two transmission stages were correlated with endophyte frequencies in natural populations (r 2: 0.89, Pearson test: P = 0.044, Fig. 3). S plants from Ansabère and Rioumajou populations, which showed the lowest infection frequencies (18% and 13%, respectively), also exhibited the lowest endophyte transmission rates (15% and 34%, respectively). Guzet S plants, with a 50% infected population, had a 63% endophyte transmission rate. The Puymorens S plant population exhibited 100% infection frequency and showed the highest transmission rate (86%).

Figure 3.

Relationship between symbiosis frequencies observed in native Festuca eskia populations and endophyte transmission rates measured in each population under a common garden experiment (Mean ± SD). Circle: endophyte transmission rate from plant to tillers; Triangle: endophyte transmission rate from plant to seeds.

Discussion

The objective of our study was to determine the relative importance of symbiosis effect on host fitness and symbiont transmission in driving symbiont frequency in host populations. We showed that in all populations, S plants outperformed NS plants, and endophyte frequencies were correlated with endophyte transmission rates. These results suggested the observed variation in endophyte frequencies among F. eskia populations was principally explained by endophyte transmission rates.

Differences in fitness-related traits between S and NS plants were detected in all F. eskia populations. At heavily grazed sites (e.g. Ansabère and Guzet), RA increased in S plants compared with NS plants, exhibited by increased spike production. Grazing pressures are probably responsible for this differentiation. Indeed, grasses grazed for long periods of time are known to increase tillering and reduce flowering (e.g. Painter, Detling & Steingraeber 1993; Hazard, Barker & Easton 2001). It was the case for NS plants in our study. Increased spike production in S plants could be explained by herbivore-deterrent alkaloids produced by E. festucae (Koh & Hik 2007): mammal herbivores tend to avoid grazing S plants and prefer NS plants. At low grazed sites (e.g. Rioumajou), S and NS plants exhibit similar reproductive output, but S plants show increased vegetative biomass. Grime (1977) reported that in areas not subjected to stress or perturbation, high shoot biomass enhanced effective above-ground competition, suggesting an adaptive differential in favour of S plants under native habitat conditions. This finding is congruent with previous results on L. perenne, demonstrating that endophytic symbioses promote shoot growth in populations from nutrient rich environments (Gibert et al. 2012b). Consequently, S plants always outperformed NS plants in our study, and our results confirmed the presence of a population-specific endophytic symbiosis effect on host species (Hesse et al. 2003; Gibert et al. 2012b).

Removing the endophyte from S plant in the Puymorens population markedly decreased E- plant fitness. The manipulation revealed a clear mutualistic endophytic symbiosis effect in this population (e.g. De Mazancourt, Loreau & Dieckmann 2005). However, extrapolation from our results to fixed benefits of S over NS plants in F. eskia populations is inaccurate. Indeed, the S advantage observed may be transient over a longer time scale; there is no evidence that advantages occur over the entire plant life history. In our experiment, the hypothesis of a specificity of endophyte effect on a plant stage is supported by a symbiosis effect visible only under W+F conditions, i.e. when plants reached a particular size. Modelling population dynamics would be an elegant means to address S plant fitness advantages maintained over plant life history in all populations.

Existing data suggest the most common circumstances in nature are those in which the endophyte induces negative or positive but tiny effects on host plants (see Müller & Krauss 2005; Faeth & Hamilton 2006; Gundel et al. 2008). In our study, we found a notable symbiosis effect in the native grass F. eskia by comparing naturally S from naturally NS plants. Considering the use of this non-traditional approach to understand endophytic symbioses (see Gibert et al. 2012b), it facilitates our interpretation of the plant dynamics resulting in population endophyte frequencies. Yet, with this approach, we cannot be able to reveal the underlying mechanisms of the differences between S and NS plants. Neutral genetic differentiation was not observed in F. eskia associated to the endophyte (i.e. meaning random crossing between S and NS plants, Gonzalo-Turpin et al. 2010); therefore, differences between S and NS in our experiment may be explained by a direct fungal endophyte effect on the host and/or adaptive differentiation between S and NS plants. Consequently, we believe that an enhanced understanding of the grass–endophyte interaction dynamic will be gained by simultaneously analysing differences between S, NS and E- plants in native grass species (see e.g. Gundel et al. 2012).

Our results provide biological evidence for the dominant role of endophyte vertical transmission responsible for endophyte frequency variation in native grass populations. Indeed, given a high symbiont frequency was reached with a tiny fitness differential between S and NS plants (Ravel, Michalakis & Charmet 1997; Gundel et al. 2008), even with a transient S plant advantage, an imperfect endophyte transmission mechanism is required to explain intermediate levels of endophyte frequencies in host populations. Furthermore, endophyte transmission rates observed under common garden conditions in F. eskia populations were positively correlated with endophyte frequencies in individual populations; high symbiont frequency in grass populations was associated with high rates of endophyte transmission, and low symbiont frequency with low rates of endophyte transmission. The observed endophyte transmission rates < 100% are consistent with former studies (Afkhami & Rudgers 2008; Canals, San Emeterio & Oreja 2008; Gundel et al. 2009).

Our study suggests the variability in population endophyte transmission rates is the result of genetic rather than environmental factors. Indeed, we found that endophyte transmission rates measured under a common garden experiment were not influenced by water and nutrient availability treatments. This is in accordance with previous studies showing that diverse environmental factors had no significant effects on endophyte transmission rates (water availability: Davitt, Chen & Rudgers 2011; soil resource addition: Gibert, Magda & Hazard 2012a; but see García Parisi et al. 2012). Furthermore, considering the use of a common garden experiment, differences observed in endophyte transmission rates between F. eskia populations have a genetic basis and/or are derived from a maternal effect. However, the risk of a maternal effect in our experiment was notably reduced; plants were acclimated to a single environment prior to the experiment; each clone was generated from only five tillers, we focused exclusively on neoformed tillers and spikes; and finally, the length of our experiment (i.e. 2 years) allowed bud differentiation and anthesis under common garden conditions. Thus, this is the first study where population endophyte transmission rates appear to have a genetic basis. Our results are inconsistent with conclusions by Gundel et al. (2011a), where host genetic background effects on endophyte transmission rates were not detected at the individual plant level. By crossing different genotypes with a highly endophyte-infected (95%) population in Lolium multiflorum, Gundel et al. (2011a) showed the absence of significant differences in endophyte transmission rates between daughter plants that originated from different pollen. Beyond species differences between these two studies (i.e. F. eskia and L. multiflorum), an explanation to these seemingly discordant results should be that the endophyte transmission rate might be determined by the genetic background of the host mother plant.

From an evolutionary point of view, why would a host plant not transmit a beneficial fungal endophyte? Theoretically, selection effects on the host should act to reduce the transmission of symbionts acting as antagonist partners (e.g. Fine 1975; Ewald 1987). Here, even in F. eskia populations with low endophyte transmission rates (15%), evidence of an antagonist effect of symbiosis was not detected. Another explanation is transmission rate evolution might be constrained by selection for another trait in the host, i.e. a trade-off between symbiont transmission rate and another host trait. Recently, Asplen et al. (2012) proposed that trade-offs often generate constraints on the evolution of traits important to the fitness of interacting organisms. In our study, we hypothesize a trade-off between endophyte transmission rates and plant resource acquisition. Indeed, endophyte transmission rates in F. eskia varied along a ‘leaf economics trade-off’, explained by LDMC and SLA (Grime 1977; Wright et al. 2004). The Puymorens and Guzet populations exhibited a resource conservation strategy, including the lowest SLA and highest LDMC at both the site of origin and the common garden. Puymorens and Guzet also showed the highest endophyte transmission rates (86% and 63%, respectively). In contrast, Ansabère and Rioumajou populations demonstrated a resource acquisition strategy with higher SLA and lower LDMC values, and the lowest endophyte transmission rates (15% and 34%, respectively). The establishment of a ‘conservation syndrome’ in the host plant appears more favourable to plant colonization by the fungus than an ‘acquisition syndrome’ due to differences in plant relative growth rate: the conservation syndrome results in low relative growth rates, whereas the acquisition syndrome corresponds to plants with high relative growth rates (Lambers & Poorter 1992).

Conclusion

Hereditable symbioses are probably intermediate states towards fixation of symbioses in hosts. However, researchers have primarily studied the mechanisms that maintain variation in symbioses from an evolutionary perspective by traditionally focusing attention on the benefits of a symbiosis to host fitness and overlooking the symbiont vertical transmission mechanism(s). The primary objective of our study was to determine the relative importance of both mechanisms in relation to the maintenance of variable symbiont frequencies in host populations. Using the grass–endophyte symbiosis as a model of hereditary symbiosis, we showed that: (i) variation in symbiont frequencies in native grass populations is principally explained by endophyte transmission rates rather than the fitness differential between S and NS plants on an ecological time scale, (ii) endophyte transmission rate appears to be a genetically based mechanism at the host population level, and (iii) these two mechanisms involved in hereditary symbiont fixation are not directly linked on an ecological time scale. We propose that this apparent mismatching is a trade-off with another host trait that will generate constraints on the evolution of endophyte transmission. Finally, we suggest the symbiont transmission rate is a key mechanism in the symbiont fixation process, and we encourage further study to identify the mechanisms underlying the variation in vertical transmission rates.

Acknowledgements

The French government (FUI) and the state of Midi-Pyrénées supported this work. We thank Paul Laurent, Benoit Gleizes, Paul Servanty and Rouger Romuald for their technical assistance. We are very grateful to Dr. Robert Faivre for his help with statistical analyses and Joanna Schultz from www.writescienceright.com for English corrections.

Ancillary