Cultivar genotype, application and endophyte history affects community impact of Schedonorus arundinaceus

Authors


Correspondence author. E-mail: jonathan.newman@uoguelph.ca

Summary

  1. Cultivars of several cool-season grasses used in forage and turf applications are developed to contain low (E−) or high (E+) seed fungal endophyte presence, and these characteristics may influence their competitive ability and effects on communities.
  2. We established a long-term field experiment to test the predictions that Schedonorus arundinaceus (SA) tall fescue forage cultivars differ in their effects on communities from turf cultivars and that E+ cultivars differ in their effects on communities from E− cultivars. Two of the three E+ turf cultivars were low endophyte, and turf cultivars were therefore defined as ‘E+ history’ and ‘E− history’.
  3. Forage (E−) plots contained more SA than turf (E− history) plots and differed in community structure and function from turf (E− history) plots. Cultivar identity also influenced community structure and function in pairwise comparisons between forage (E−) and turf (E− history) plots.
  4. Within the forage cultivars, E+ plots contained more SA and were marginally less dominated by other grasses than E− plots, but these differences were not consistent in comparisons between specific E+ and E− forage cultivars. Moreover, E+ and E− forage plots were similar in other aspects of community structure and function, indicating that a cultivar's endophyte status does not consistently determine its effects.
  5. Despite low endophyte presence in two of three cultivars, E+ history turf plots differed in measures of composition, structure and nutrient cycling from E− history turf plots, indicating that there are genetic differences between E+ and E− cultivars which can influence their effects on communities.
  6. Synthesis and applications. Schedonorus arundinaceus cultivars vary in their competitive ability and effects on communities, and this variation may be explained by the application for which the cultivars were developed, their endophyte status and genetic differences among cultivars within application categories. We recommend preferentially selecting cultivars developed for turf applications and cultivars with low seed endophyte presence to minimize negative effects of seeding a non-native grass on communities. However, final selections should be based on common trial performances because genetic differences between cultivars will affect their performance independently of their application category and endophyte status.

Introduction

Many cool-season grasses developed for forage and turf applications are of conservation interest because of their ability to establish naturalized populations outside of managed areas. Several of these grasses contain asexual fungal endophytes within their above-ground tissues (Rudgers et al. 2009), and both application and endophyte presence may affect host competitive ability and effects on communities. Forage cultivars may be stronger competitors and have greater effects on communities than the turf cultivars as they are selected for use in less-intensively managed sites dominated by non-native species. Endophytes may increase host competitive ability (Lemons, Clay & Rudgers 2005) and alter host effects on communities (Clay & Holah 1999; Spyreas, Gibson & Middleton 2001) as a result of effects on plant physiology and growth (Rasmussen, Parsons & Newman 2009; Siegrist et al. 2010). However, given the phenotypic variation among commercially produced cultivars (Wilkins & Humphreys 2003; Bonos, Clarke & Meyer 2006), it is unclear whether the application for which cultivars are developed and their endophyte status would consistently explain variation in competitive ability and effects on communities among cultivars of a single species.

Different suites of traits, including endophyte presence, are selected when producing forage or turf cultivars. Forage cultivars are typically selected for dry matter yield, palatability, digestibility and grazing tolerance (Pedersen 1988; Wilkins & Humphreys 2003) and are usually planted in conjunction with non-native legumes (Leahy & Robinson 2000). By contrast, turf cultivars are selected for disease resistance, texture, colour and mowing tolerance (Funk, White & Breen 1993; Bonos, Clarke & Meyer 2006) and are often planted in monoculture or mixtures with other cool-season turf grasses. These different selection goals could affect ecologically important traits (i.e. leaf surface area, leaf tissue C:N ratios; Daehler 2003; Van Kleunen, Weber & Fischer 2010) that distinguish forage and turf cultivars. In terms of endophyte presence, forage cultivars are under strong selection to contain endophytes that produce lower amounts of defensive ergot alkaloid compounds, while alkaloid production is not necessarily an undesired trait in the turf application (Easton 2007). Given the different selection goals in the forage and turf industry, it is likely that forage and turf cultivars differ in their competitive ability and effects on communities.

Although endophyte presence is generally thought to increase host competitive ability and effects on communities in the agronomic grasses (Saikkonen et al. 2006), it is not clear whether E+ cultivars would be consistently different from E− cultivars within the forage and turf applications. Our understanding of the effects of endophyte presence on plant community structure and function has primarily arisen from studies on the effects of Neotyphodium coenophialum (Morgan-Jones & Gams) Glenn, Bacon and Hanlin within Schedonorus arundinaceus Schreb. Dumort (SA) tall fescue forage cultivars developed as E + . In the first study to investigate effects of endophyte presence on plant diversity, Clay & Holah (1999) seeded plots in an abandoned agricultural field with E+ and E− versions of the KY-31 forage cultivar. After four years, species richness was lower in plots containing E+ KY-31. Follow-up studies suggested that lower richness in E+ plots resulted from a combination of reduced species establishment, increased tree-seedling herbivory and plant–soil feedbacks that promote E+ SA invasion into soils previously dominated by other species (Matthews & Clay 2001; Rudgers et al. 2007; Rudgers & Orr 2009). In the only other study to investigate effects of endophyte presence in a SA population of unknown origins, species richness declined with increasing endophyte presence in unmowed plots and increased with increasing endophyte presence in mowed plots (Spyreas, Gibson & Middleton 2001).

Endophyte presence may also affect the host's influence on local nutrient cycling. Franzluebbers et al. (1999) found that net nitrogen mineralization (Nmin) was higher in E+ KY-31 soils than in E− KY-31 soils and attributed this outcome to either greater nitrogen immobilization in E− soils or release of accumulated N-containing compounds in E+ soils. Effects of endophyte presence on Nmin were mixed in follow-up studies. Nmin was marginally increased in E+ treatments in another study with KY-31 (Franzluebbers & Hill 2005), but in a field experiment with the cultivar Jesup (Franzluebbers 2006) and in a multi-site study with several cultivars (Iqbal et al. 2012),soil Nmin was similar between E+ and E− communities. These studies suggest that endophytes may affect plant community structure and function but because they primarily assessed effects of endophyte presence within E+ forage cultivars, it is unclear whether the same differences would occur between cultivars developed as E+ and E− for use in forage and turf applications.

Several factors should be considered when using results from endophyte-removal studies (i.e. those that remove the endophyte from E+ cultivars to study endophyte effects) to predict the ecological effects of seeding E+ and E− cultivars. First, the effects of endophyte presence reported in removal studies may not represent average endophyte effects across populations and species (Faeth 2002; Saikkonen et al. 2006). Endophyte effects can vary among individuals within a host population (Cheplick 1998, 2008; Assuero et al. 2000), and the cultivars commonly used in endophyte-removal studies may be more strongly affected by endophyte presence than other cultivars (Saikkonen et al. 2006). Second, although endophytes may affect the range of traits over which E+ cultivars can be selected (Easton 2007), this does not mean that they would be any more or less competitive than E− counterparts developed for use in the same application. With few exceptions, commercially produced E+ and E− cultivars are developed independently of one another, and E− cultivars are not endophyte-free versions of E+ cultivars. E+ cultivars differ in their endophyte presence and plant background from E− cultivars and as a result E− cultivars may be just as competitive as E+ cultivars. Finally, because desirability of endophyte presence differs between forage and turf cultivars, it is possible that the same endophyte could have different ecological effects within forage and turf cultivars of the same host species.

Schedonorus arundinaceus is one of the several species widely planted for forage and turf applications that contains fungal endophytes and is invasive in non-agricultural systems in North America (Cully, Cully & Hiebert 2003; Tunnell, Engle & Jorgensen 2004). We assessed whether tilled communities seeded with E+ and E− SA cultivars developed for forage and turf applications differ in community structure and function. We predicted that forage plots will contain more SA and differ in structure and function from turf plots and that plots seeded with E+ history forage and turf cultivars will contain more SA, will be less diverse and produce more plant-available nitrogen than plots seeded with E− history counterparts. We assess the consequences of introducing a new species to a site and leave future studies to assess how likely these cultivars are to colonize new locations from established populations. Additionally, while we ask whether there are ecologically important differences between communities seeded with E+ and E− cultivars, future studies will need to determine whether observed responses are directly attributable to the endophyte status of these cultivars.

Materials and methods

Study site

The long-term impacts of fungal endophytes experiment at the University of Guelph Turfgrass Institute (Wellington County, Guelph, ON, Canada; 43°32′56″N, 80°12′39″W) was established in late-spring 2007. The study site was previously developed as an apple orchard but was maintained with occasional mowing for 20 years prior to our study. Soils consist of Guelph Sandy Loams (Brunisolic Grey-Brown Luvisol) developed on loam till. Non-native species including Elymus repens L. Gould, Poa pratensis L. Taraxacum officinale F.H. Wigg and Cirsium arvense L. Scop dominated the site prior to the study. Although, P. pratensis and E. repens may also harbour fungal endophytes (Saikkonen et al. 2000; Wei et al. 2006), endophyte presence in these species was not tested.

Study design

The experiment consists of 240 2 × 2 m plots spaced 0·5 m apart in a 75 × 30 m grid in a completely randomized block design with 10 blocks. Prior to seeding in early June 2007, the area was tilled twice to break up existing vegetation. Herbicide was not applied prior to or during the study. After tilling, a subset of the plots were seeded (5 g m−2) with SA. This seeding rate is similar to the seeding rate in Clay & Holah (1999). Plots were watered 3–4 times a week in the first two months. Plots were not weeded or disturbed post-seeding (i.e. no mowing or grazing). Aisles among plots were mowed as needed.

Plots included in this study were seeded with one of five cultivars developed to contain high seed endophyte (E+) presence (N. coenophialum) or one of five cultivars developed to contain low seed endophyte (E−) presence (Table 1). ‘Low endophyte’ is an industry term used to describe cultivars thought to be endophyte-free, but where the producer does not guarantee its absence. Each endophyte category included cultivars developed for use in a forage (2 E+; 3 E−) or turf (3 E+; 2 E−) application (Table 1) which were selected based on seed availability. When seed was obtained from a supplier (spring 2007), it was dry-stored at −17 °C to prevent declines in seed viability and endophyte presence that could occur with warmer storage conditions (Welty, Azevedo & Cooper 1987). Starlet was harvested in 2003 and Rhizing Star and Savory were harvested in 2005. The harvest year for the remaining cultivars and the seed storage conditions between harvest and arrival for all cultivars are unknown. Seed germination and endophyte presence were not tested prior to seeding.

Table 1. High (E+) and low (E−) endophyte Schedonorus arundinaceus cultivars developed for forage and turf applications were used in the study. Tiller percentage is the percentage of tillers collected from study plots that tested positive for endophyte presence. Seed percentage is the percentage of seeds that tested positive for endophyte presence in June 2010. Germination percentage is the percentage of seed germinating in a 2010 greenhouse germination trial
Cultivar (abbr.)Endophyte historyTiller % (number of tillers)Seed %, n = 20Germination %, n = 30
  1. a

    Georgia Agricultural Experiment Station, Athens, USA.

  2. b

    Pickseed, Lindsay, ON, Canada.

  3. c

    Parsons Seed Ltd, Beeton, ON, Canada.

  4. d

    Seed Research of Oregon, Corvallis, USA.

Forage
Jesup (Jsp)aE+88 (100)10080
Georgia 5 (Ga5)aE+81 (100)10067
Carnival (Crn)bE−0 (100)094
Savory (Svr)cE−1 (97)090
Rahela (Rhl)cE−0 (94)0100
Turf
Rhizing Star (Rhz)cE+10 (77)10093
Mustang III (Mst)bE+0 (78)087
Starlet (Str)cE+0 (68)073
SR8500 (Sr8)E−0 (59)050
Crewcut II (Crw)dE−0 (74)097

Endophyte presence and germination in the stored source seed was evaluated in 2010. Twenty seeds remaining from the original seed lot of each cultivar were tested for endophyte presence using an immunoblot assay (Phytoscreen seed endophyte detection kit; Agrinostics Ltd. Co., Watkinsville, GA, USA). The assay tests for the presence of endophyte cell wall proteins but not necessarily endophyte viability (Hiatt et al. 1999). With exception of Starlet and Mustang III, seed endophyte presence was consistent with the cultivar endophyte history (Table 1). To test germination, five seeds of each cultivar were planted into six pots in a greenhouse in April 2010. Pots were watered to field capacity daily and germination was recorded every other day for two weeks. Germination was higher in the stored Starlet and Mustang III seed than in the E+ forage cultivars (Table 1), suggesting that factors that affected endophyte presence did not reduce germination.

Schedonorus arundinaceus tillers from each plot were tested for endophyte presence in June 2010. We collected up to ten tillers per plot spaced at least 20 cm apart. In cases where there were fewer individuals, less than ten tillers were collected. Tillers were frozen for no longer than two weeks and tested for endophyte presence using an immunoblot assay (Phytoscreen field tiller endophyte detection kit; Agrinostics Ltd. Co., Watkinsville, GA, USA) which yields results consistent with microscopic analysis (Hiatt et al. 1999). Tiller results were consistent with the cultivar endophyte history for all but the E+ history turf cultivars (Table 1). The tiller and seed results indicate that Rhizing Star was the only E+ history turf cultivar that was high endophyte at seeding. Seed endophyte presence declines over time and at high storage temperatures (Welty, Azevedo & Cooper 1987; Hill & Roach 2009), and both factors may have affected endophyte presence in Starlet and Mustang.

Sampling

Species within a plot were recorded and species relative abundances were quantified using non-destructive point–intercept sampling (Jonasson 1988; Brathen & Hagberg 2004) in July–August 2010. Point–intercept sampling was carried out by placing an elevated (1 m) rod diagonally across the plot and dropping narrow steel pins vertically through the vegetation at 10-cm intervals. To avoid edge effects associated with pathways among plots, vegetation was not sampled within 0·5 m of plot edges. The identity of each leaf and stem touching a pin was recorded and used to calculate species relative abundances (pi = number of touches for species i/total number of touches). These data were used to calculate Simpson's diversity ([1/D], where = ∑pi2) an aggregate metric of richness and evenness (Wilsey et al. 2005) and evenness ([1/D]/number of species recorded during point intercept sampling). Diversity measures took into account SA abundances. The total number of touches in a plot was used as a proxy for above-ground biomass production (Jonasson 1988).

Net Nmin was estimated using Plant Root Simulator (PRS) probes (Western Ag Innovation Inc., Saskatoon, SK, Canada) in root exclusion cylinders (15 × 20 cm polyvinyl chloride) in 2010. PRS probes are ion exchange membranes designed to simulate plant roots which continuously adsorb charged ions over a given burial period. PRS probes were inserted into the cylinders (two per plot) in early May and replaced every four weeks until September. Cylinders were kept plant free and contained holes at the soil surface to prevent water from pooling. One anion probe was used per cylinder in all blocks, with the addition of one cation probe per cylinder in seven of 10 blocks. At the end of an insertion period, probes were removed, washed with deionized water, stored at 4 °C and returned for analysis as a composite sample for each plot. Plant-available NH4+ was below detection limits (μg NH4+ 10 cm−2 burial period−1) for a majority of plots at each sample period and a negligible portion of total plant-available N, so only nitrate results are presented here. Monthly nitrate values (μg NO3 10 cm−2 28 days−1) were summed across sample periods to produce a season total value (μg NO3 10 cm−2 140 days−1). Because net Nmin is affected by soil moisture and temperature, soil temperature (°C) and soil percentage volumetric water content (Campbell Scientific HydroSense Soil Water Measurement System; Edmonton, AB, Canada) were sampled to 20 cm in the centre of each plot every other week.

Data analysis

It was unreasonable to follow the original study design by assessing the main effects and interaction of cultivar application and endophyte history with a two-way nested anova because endophyte presence was not consistent in the E+ history turf cultivars. To contrast responses among the application–endophyte categories of interest (Forage vs. Turf in E−; E+ vs. E− in Forage; E+ vs. E− in Turf), community structure and function responses (species richness, Simpson's diversity, evenness, cool-season (C3) grass relative abundances, total number of leaf hits, season total plant-available nitrate) were analysed with a one-way anova with block and cultivar as fixed effects and planned contrasts (proc glm; sas 9.2; SAS Institute, Cary, NC, USA). The scope of statistical inference was limited to the specific cultivars used within the study because cultivars for each category were developed by a limited number of possible producers and could not be considered representative of the entire population of cultivars for each application–endophyte category. Block was treated as a fixed effect as there was not some broader population of blocks over which we were intending to generalize findings (Newman, Bergelson & Grafen 1997). The first contrast tested for differences in abundances and impact between E− history forage and turf cultivars. The last two contrasts were orthogonal and tested for differences between E+ and E− history plots within each application category. Least significant difference post hoc tests (α = 0·05 = probability of a Type I error for any one comparison) were used to infer pairwise differences among cultivars. The post hoc tests are presented for direct cultivar comparisons and results from the post hoc tested should not be used when interpreting group differences.

Dominant species composition (S. arundinaceus, P. pratensis and Elymus repens relative abundances) was analysed with a manova (proc glm; sas 9.2; SAS institute) using model terms and contrasts as described previously. Monthly plant-available nitrate, soil moisture and soil temperature were analysed with repeated measures anova (proc glm; sas 9.2; SAS institute) with the model terms as described previously and a repeated statement. One Starlet plot was inadvertently skipped during sampling for soil moisture and temperature and the sample size for this analysis was reduced by one plot. Species relative abundances were arcsin square-root-transformed to meet normality assumptions prior to analysis. Plant-available nitrate was natural log-transformed to meet normality assumptions.

Results

Diversity and composition

After four growing seasons, the plots were dominated by non-native cool-season perennial grasses (grand mean 92 ± 0·09% of the touches). Forbs and warm-season grasses were less abundant and, with the exception of isolated occurrences of Cornus L. spp. (six plots) and Vitis L. spp. (three plots), woody species were absent. The most abundant species were the cool-season grasses P. pratensis, SA and E. repens. P. pratensis was present in all plots, and SA was present in all but eight plots. All of the plots where SA was absent were initially seeded with turf cultivars (four of each endophyte type; three, SR8500; three, Rhizing Star; one, Starlet and one, Crewcut II). Vegetation was not sampled in 2007 and it is unclear if SA failed to establish or if SA was excluded after establishment in these plots.

Plots seeded with different cultivars varied in their cool-season grass composition and this variation could be explained by cultivar application–endophyte groups (Table 2, Fig. 1). Forage (E−) plots contained more SA and less P. pratensis and E. repens than turf (E− history) plots. Within the forage cultivars, E+ plots generally contained more SA and marginally less P. pratensis and E. repens than E− plots. Within the turf cultivars, E+ history plots also contained more SA and less P. pratensis than E− history plots. Responses were consistent among E+ forage and E− history turf cultivars but were more variable among E− forage and E+ history turf cultivars (Fig. 1). Carnival plots were consistently similar to E+ forage plots and different from Savory and/or Rahela plots. SA was less abundant in Rhizing star plots than in Mustang III or Starlet plots (Fig. 1).

Figure 1.

Transformed cool-season grass relative abundances (mean ± SE) in plots seeded with forage or turf cultivars (see Table 1 for abbreviations) with a history of high (E+) or low (E−) seed endophyte presence. Means with the same letter are not significantly different.

Table 2. F values from a multivariate anova of relative abundances of the dominant C3 grasses, Schedonorus arundinaceus, Poa pratensis and Elymus repens. F values were approximated from the Pillai's trace multivariate test statistic for the multivariate analysis. Cultivars had a history of high (E+) or low (E−) seed endophyte presence
SourceUnivariateMultivariate
d.f. Schedonorus arundinaceus Poa pratensis Elymus repens d.f.Pillai's trace
  1. *< 0·05; **< 0·01; ***< 0·001; ‡< 0·10.

Block9,813·46**1·88‡11·73***27,2434·06***
Cultivar9,8114·11***6·26***3·14**27,2432·97***
Contrasts
E− Forage vs. E− Turf1,8143·38***27·56***4·95*3,7917·14***
E+ vs. E− in Forage1,8111·46**3·56‡3·39‡3,793·98*
E+ vs. E− in Turf1,816·12*9·61**0·013,793·86*

Plots seeded with different cultivars were similarly species rich (grand mean 8·0 ± 1·6) but differed in measures of diversity and productivity between and within cultivar applications (Table 2, Fig. 2). Forage (E−) plots were more even, more diverse and less productive than turf (E− history) plots. Within the forage cultivars, E+ and E− plots were similarly diverse and productive, but differed in species evenness (Table 2, Fig. 2). Within the turf cultivars, E+ history plots were similarly even, but less productive and more diverse than E− history plots. Diversity and productivity responses were consistent among cultivars within endophyte–application groups, with the exception that Rahela plots were consistently similar to Crewcut plots and more productive than Carnival and Savory plots.

Figure 2.

Simpson's diversity, evenness and total touches (a proxy for above-ground biomass) from point intercept sampling in August 2010. Plots were seeded with forage or turf cultivars (see Table 1 for abbreviations) with a history of high (E+) or low (E−) seed endophyte presence. Means with the same letter are not significantly different.

Soil nitrogen mineralization

Monthly plant-available nitrate production also differed among cultivar groups. In general, monthly plant-available nitrate production and soil moisture declined and soil temperature increased over the growing season (May–August; see Table S1 in Supporting Information; Fig. 3). Plant-available nitrate production was initially greater in forage (E−) plots than in turf (E− history) plots (Table S2; Fig. 3). Soil moisture was higher in forage (E−) plots than in turf (E− history) plots in early June (F1,80 = 3·87; = 0·05), August (F1,80 = 4·53; < 0·05) and September (F1,80 = 3·96; = 0·05) but was marginally drier in late June (F1,80 = 2·90, = 0·09) and July (F1,80 = 3·54; = 0·06). These dates did not correspond to time periods when Nmin differed between the application groups (Table S2). Within applications, plant-available nitrate production was similar between E+ and E− forage plots, but was greater in E+ than in E− history turf plots (Fig. 3).

Figure 3.

Monthly transformed plant-available nitrate in plots seeded with (a) turf cultivars (see Table 1 for abbreviations) with an E+ and E− history and (b) forage (F) and turf (T) cultivars with an E− history. Asterisks indicate months where responses between groups were significantly different.

Season total plant-available nitrate production marginally differed among plots seeded with different cultivars (Table S1). Season total plant-available nitrate production was greater in forage (E−) plots (157·53 mg NO3 10 cm−2 140 days−1) than in turf (E−) plots. Within forage cultivars, season total plant-available nitrate was similar between E+ and E− plots (Table S1). Within turf cultivars, E+ history plots produced more nitrate (150·66 mg NO3 10 cm−2 140 days−1) than E− history plots (85·80 mg NO3 10 cm−2 140 days−1).

Discussion

This study assessed the effects of seeding SA forage and turf cultivars selected to contain high (E+) or low (E−) seed endophyte presence into a tilled field with an established propagule pool. Seeding different cultivars affected final SA abundances and measures of community structure and function and the cultivar application–endophyte status explained some of this variation. The greatest differences occurred between communities seeded with forage (E−) and turf (E− history) cultivars (Fig. 1-3), and it is likely that communities seeded with high endophyte forage and turf cultivars would also differ. Within forage cultivars, E+ cultivars were more abundant than E− cultivars (Table 3; Fig. 1). However, other aspects of community structure and function were similar between E+ and E− forage communities (Fig. 2), indicating that a cultivar's endophyte history does not consistently determine its impact. Despite low endophyte presence in two of three E+ history turf cultivars, communities seeded with E+ and E− history turf cultivars differed in structure and function (Fig. 1-3). This indicates that there are genetic differences between these cultivars; however, it is unclear whether high endophyte presence would magnify or reduce these differences. It is important to note that results from the group comparisons represent a general trend across cultivars with the same status and not all of the cultivars between and within groups were different from one another in the pairwise tests. This indicates that while the application–endophyte status of a cultivar is informative, the ecological effects associated with seeding this non-native grass need to be considered on a cultivar-by-cultivar basis.

Table 3. F values from an anova of plant community structure metrics. Cultivars had a history of high (E+) or low (E−) seed endophyte presence. Total touches refers to the number of leaf touches recorded during point intercept sampling and is a proxy for above-ground biomass
Sourced.f.Species richnessSimpson's diversityEvennessTotal touches
  1. *< 0·05; **< 0·01; ***< 0·001; ‡< 0·10.

Block9,814·30***1·210·591·72‡
Cultivar9,810·703·20**2·52*2·83**
Contrasts
E− Forage vs. E− Turf1,811·0017·60***7·22**13·35***
E+ vs. E− in Forage1,811·640·984·14*0·03
E+ vs. E− in Turf1,810·138·55**2·355·34*

We found that SA's ability to invade a disturbed field depends on the application for which the cultivar was developed. Forage (E−) cultivars were consistently more abundant than turf (E− history) cultivars. Forage cultivars were probably more abundant because they are selected to co-exist in mixed-species pastures (Wilkins & Humphreys 2003), whereas turf cultivars are often planted in monoculture in more intensively managed settings. Given that the E+ forage cultivars were more abundant than the turf cultivar (Rhizing Star) that was high endophyte at seeding (Fig. 1), it is likely that high endophyte forage cultivars would also be more abundant than high endophyte turf cultivars. We would expect to find similar differences among cultivars of other cool-season grasses (i.e. Schedonorus pratensis Huds. P. Beauv. meadow fescue; Lolium perenne L. perennial ryegrass) developed for multiple applications. However, in systems where cultivars are typically produced for a single application (i.e. grasses produced for landscape restoration) responses should be similar among cultivars (Wilsey 2010).

While E+ forage cultivars were more abundant than E− forage cultivars, the effects that E+ forage cultivars had on community structure and function relative to E− counterparts were not consistent with predictions from endophyte-removal studies. Endophytes may affect the range of traits over which cultivars are selected (Easton 2007), and it appears that the endophyte status of a cultivar may be a meaningful predictor of competitive ability in the forage grasses. However, the endophyte status of a forage cultivar may not be a good predictor of the consequences of establishment. E+ cultivars marginally reduced co-occurring cool-season grasses (as in Clay & Holah 1999) over E− counterparts, which increased plant species evenness, but not richness. Although endophyte-removal studies indicated that unmowed E+ communities would be less species rich than E− communities (Clay & Holah 1999; Spyreas, Gibson & Middleton 2001; Rudgers, Fischer & Clay 2010), they were similarly species rich in our study. It is possible that either E+ and E− cultivars have similar effects on communities or that our findings were a consequence of the scale of the experiment and the local propagule pool. Our plots were similar in size to Spyreas, Gibson & Middleton (2001), but two orders of magnitude smaller than the plots in Clay & Holah (1999). This difference in size may have affected herbivore use of the plots, as they may not have been large enough to present a meaningful difference in vegetation types to the vertebrate herbivores that were prevalent at our site. Our site also had few woody species in the local propagule pool. Increased tree-seedling herbivory in E+ communities may have contributed to the effects of endophytes on richness in Clay & Holah (1999; Rudgers & Clay 2007), but this was probably not a mechanism affecting community structure in our study. Our hypothesis concerning effects of endophyte history on Nmin was not supported, suggesting that both types of cultivars had similar effects on the decomposer and microbial communities. It is unclear whether similarities in the responses between E+ and E− forage cultivars stems from minimal endophyte effects or from genotypic differences between E+ and E− plant hosts and future studies need to separate the role that these factors play in affecting cultivar performance.

Our results indicate that E+ forage cultivars may have similar effects on communities, but effects on communities may be more variable among E− forage cultivars. We included two common E+ cultivars and three E− forage cultivars in our study. The forage cultivars, Jesup and Georgia-5, were both developed by the Georgia Agricultural Experiment Stations (Bouton et al. 1993, 1997) for use in the south-eastern United States and are commonly used in endophyte-removal studies (e.g. Franzluebbers 2006; Antunes et al. 2008). Even though they are genetically distinct (Mian, Hopkins & Zwonitzer 2002), these cultivars were similar in their abundances and effects on communities. However, the responses among E− cultivars developed by several producers (Table 1) were more variable (Fig. 1). E− Carnival plots were consistently similar to the E+ forage plots, and Rahela plots were consistently similar to E− turf plots. The comparatively fewer differences among E+ forage cultivars than among E− cultivars is consistent with findings from Rice et al. (1994) that endophytes may reduce phenotypic variation among hosts, but this is not a definitive result because of the confounding effects of producer and differences in the number of cultivars in each category. If this pattern holds across cultivars developed from multiple sources, we would expect to find that additional E+ forage cultivars have similar effects on communities, but these effects would vary more widely among E− forage cultivars.

Endophyte effects can be variable (Cheplick 1998, 2008; Assuero et al. 2000), and it is unclear whether E+ and E− history turf cultivars would have differed if all of the E+ history cultivars were high endophyte at seeding. Despite the fact that two of the three E+ history turf cultivars were low endophyte at seeding, E+ history turf cultivars as a group were more abundant and reduced co-occurring cool-season grasses to the extent that E+ plots were more diverse than E− history plots. Compositional differences translated into differences in productivity and nutrient cycling between E+ and E− history communities. These results indicate that there are meaningful genetic differences between E+ and E− history turf cultivars. However, it is unclear whether high endophyte presence in the E+ cultivars would reduce or magnify these differences and whether this would result in consistent differences between E+ and E− turf cultivars.

Conclusions

Studies that manipulate endophyte presence within a cultivar are important for developing our understanding of potential endophyte effects in the cool-season grasses. However, findings from such studies should not be used to assess the conservation implications of seeding independently produced E+ and E− cultivars. E+ SA forage cultivars were more abundant than comparable E− cultivars, but E+ and E− cultivars had similar impacts, which indicates that the ecological consequences of seeding E+ over E− cultivars may not be as dramatic as would be suggested from endophyte-removal studies. Our results indicate that a cultivar's application–endophyte history should only generally be taken into consideration when selecting cultivars for land management. Where possible, we recommend preferentially selecting turf and E− cultivars because they may be generally less problematic from a conservation standpoint than forage and E+ cultivars, but warn that it is possible that cultivars in these categories may have similar ecological effects as forage and E+ counterparts. Final cultivar selections should be based on common trial performances because genetic differences between cultivars will affect their performance independently of their application category and endophyte status.

Acknowledgements

Peter Purvis, Ken Carey, the summer staff at the Guelph Turfgrass Institute and members of the Grass Endophyte Research Group at the University of Guelph helped to establish and maintain the study plots. Funding for this project was provided by grants from the Canadian Natural Science and Engineering Research Council to J.A.N., the Ontario Ministry of Agriculture, Food and Rural Affairs to J.A.N, H.M and J.N.K. and the North Dakota Experimental Program to Stimulate Competitive Research to K.A.Y.

Ancillary