Introduction

Among grass species (Poaceae family) in the Polish grasslands with high importance for agriculture are: perennial ryegrass (Lolium perenne L.), meadow fescue (Festuca pratensis Huds.), red fescue (Festuca rubra L.) and tall fescue (Festuca arundinacea Schreb.). These species are known to host symptomless endophytic fungi of Epichloë species (Petroni 1986). Until recently the dual naming system existed: sexual stage Epichloë (telemorph) and asexual Neotyphodium (anamorph). It has been proven to be more of an impediment than a benefit for most scientists working on the Epichloë species, so they were combined in one genus of Epichloë (Leuchtmann et al. 2014).

No obvious symptoms of endophyte infection are visible over the whole stage of the plant development. As inflorescences develop, the mycelium grows into ovules and within seeds it colonizes the scutellum and embryo axis before the seed germination phase (Philipson and Christey 1986). Epichloë endophytes are widely distributed in grasses (White 1987; Moon et al. 2002; Cheplick and Faeth 2009). They were found also in Europe, with average frequency of wild grass population colonization ranging from 1 to 100 % (Wäli et al. 2000; Zabalogogeazcoa et al. 2003; Lewis 2000; Hahn et al. 2007). Endophyte infection (from 20 % to 60 %) was noted in Polish grasslands by Pfannmöller et al. (1994) and by Lewis (2000). Endophytes were found also both in seeds and in plants of common grass species in Polish grasslands by Pańka and Żurek (2005); Wiewióra (2011) and Wiewióra et al. (2008). Epichloë fungi were present in ecotypes of Festuca pratensis, F. arundinacea, F. rubra, Lolium perenne and Deschampsia cespitosa collected from permanent grasslands in Mazovia region (Żurek et al. 2010a). A positive effect of endophyte colonization on persistency and aesthetic value of Festuca rubra cultivars was noted (Pańka and Żurek 2005; Prończuk and Prończuk 2000). It was also concluded that endophyte colonization increases along with grass plantation age (Wiewióra et al. 2006).

For agricultural practice, endophyte – plant symbiosis is both positive and negative. Endophyte colonized (E+) grasses express a range of adaptations to abiotic (drought, mineral imbalance, soil acidity) and biotic (disease, pest or animals) stress (Funk et al. 1994; Cheplick et al. 2000; Malinowski and Belesky 2000; Zaurov et al. 2001; Hesse et al. 2003; Hesse et al. 2004; Hesse et al. 2005; Cheplick 2004). As a result, E+ grasses may be more compatible than non-colonized grasses and thrive better under limited resources (Bacon 1993; West 1994; Belesky and Malinowski 2000; Schardl et al. 2004). However, in some cases, negative effects of the symbiosis in the plant were observed (Hesse et al. 2004; Cheplick et al. 1989). In certain circumstances endophytes produced ergovaline, lolitrem B, peramine and loline. These alkaloids play a role in defending the host plant against herbivores (peramine and loline) and have been linked with decreased animal production and health problems (ergovaline and lolitrem B). “Ryegrass staggers syndrome” and “fescue toxicosis” are the most common animal diseases caused by E+ grass. Ryegrass staggers syndromes (decrease of milk production, weight loss, disorders of nervous system) are caused by presence of large quantities of lolitrem B in animal forage (Fletcher 1993; Fletcher and Sutherland 1993; Cross 2000; Miyazaki et al. 2004; Ball 2007). The fescue toxicosis, connected with ergovaline production is manifested by reduced forage intake, excessive salivation and reduced reproductive performance (Cross 2000; Miyazaki et al. 2004). In addition to the animal diseases described above, high mortality was also noted (Foot et al. 1988). European research concerning ergovaline contents in endophyte-colonized grasses is becoming increasingly specialized. This research involves not only ergovaline levels produced by different grass species (Leuchtmann et al. 2000; Vazquez de Aldana et al. 2001; Vazquez de Aldana et al. 2003), but also intraspecific variability of the production of mycotoxin and relationship between the genetic distance among endophyte isolates and the alkaloid contents (Bony et al. 2001; Vazquez-de-Aldana et al. 2010), differences in ergovaline contents between different types of plant tissue (Vazquez de Aldana et al. 2007) or genetic diversity of endophyte (Dahl Jensen et al. 2007). However, in Poland studies examining ergovaline in endophyte hosting grasses are rare. Some analyzes of mycotoxins produced by endophytic fungi in swards of permanent grasslands in the Mazovia region were performed by Żurek et al. (2010b). They showed that endophytes colonizing L. perenne, F. arundinacea, F. pratensis, F. ovina and F. rubra have the ability to produce ergovaline.

The alkaloid type and their concentrations are affected by different factors, such as plant and fungus species and genotypes, plant growth conditions, season and plant part (Leuchtmann et al. 2000; Siegel and Bush 1996). Current knowledge of the nature of the major forces driving endophyte presence, infection frequency and toxin production in seminatural permanent grassland is still very limited. Many studies make the assertion that climatic conditions are important factors that determine the expression of plant/endophyte interaction (Compant et al. 2010; Vazquez-de-Aldana et al. 2010; Repussard et al. 2014a; Repussard et al. 2014b; Reed et al. 2011). It is therefore possible that processes related to host-fungus symbiosis are dependent on local bioclimatic conditions with possible relations to spatial distribution of trait performance. To describe and compare spatial structure of data, spatial autocorrelation can be used (Długosz et al. 2014). Spatial autocorrelation refers to the dependencies that exist among observations that are attributable to the relative locations or underlying two dimensional ordering of variable values in geographic space (Dale and Fortin 2002). In turn, these dependencies produce clustering of similar (positive spatial autocorrelation) or dissimilar (negative spatial autocorrelation) values, and hence induce some map patterns. In its most general sense, spatial autocorrelation is concerned with the degree to which objects or activities at some place on the earth’s surface are similar to other objects or activities located nearby (Sokal and Oden 1978; Escudero et al. 2003). One often used autocorrelation measures is Moran’s Index (Moran’s I) (Moran 1950). In general, a Moran’s I value near +1.0 indicates clustering while an index value near −1.0 indicates dispersion, provided there is statistical significance of calculated values. A zero value indicates a random spatial pattern.

The aim of the current study was to describe relations between endophyte colonization and ergovaline contents in grasses from different regions of Poland to local climatic conditions, taking into account spatial distribution of observed traits.

Materials and methods

Sampling

More than 740 ecotypes of 26 grass species were collected from 226 localities and further tested for endophyte presence and ergovaline contents. We use the term ‘ecotypes’ to describe a group of plants within a species that is adapted to particular environmental conditions (locality) and therefore exhibit structural and physiological differences from other members of the same species. Ecotypes were collected in a form of living plants from permanent grassland in most cases used for cattle breeding. Only localities with grasses dominating in swards were chosen for sampling. Ecotypes were collected according to species abundance in particular locality, i.e. rarely found species were not collected from this site. From 5 to 10 individuals were collected per one ecotype per species in each locality, with a distance of 5 – 10 m left between sampled plants to avoid sampling clones. For further analysis, plants were maintained for the next year in the field collection in Radzików. Plants were grown in a spaced plant nursery, with 0,5 m distances between plants. No additional treatments (fertilization, watering, chemical weed control) were applied during plant growth. During plant growth weeds were hand removed, and at the end of the growing season aboveground parts of plants were cut off with a pasture mower at the height of 15–20 cm. Sampling for analysis consisted of aboveground biomass (leaves, stems and inflorescence) of an individual plant.

Detection of endophytes

Endophyte infection was determined by a rapid staining method according to Saha et al. (1988). Ten tillers from each plant were investigated. Small epidermal strips were peeled off the adaxial surface of the leaf sheaths and placed into a drop of 0.5 % Bengal rose staining solution (5 % of ethyl alcohol solution). Stain drops with epidermal strips were further covered with a glass slip and examined under a light microscope (magnification of ×100) for the presence of fungal hyphae. The endophyte presence in an ecotype was noted when hyphae were found in at least one tiller per plant per ecotype. The endophyte appeared as an intercellular, long and convoluted hypha running parallel to the leaf-sheath axis of the plant cell without forming haustorial structures (Clay and Holah 1999).

Endophyte frequency (EF) was calculated separately for each grass species per locality and expressed as a percentage of inhabited plants in all plants examined.

Analysis of ergovaline

All grass samples for ergovaline (ERV) analysis were collected the same day after one year of growth in the same field conditions so as to reduce the effect of variation of local climatic conditions during grass sampling (elimination of the seasonality of ergovaline production, homogeneity of soil and environmental conditions). It is therefore expected that differences in ERV expression could be the results of adaptations to their initial habitat and specific interaction between plant and endophyte that resulted in ongoing changes in the expression profile of the plant/endophyte association. The ERV analysis was conducted according to Craig et al. (1994). The 0,2–1,0 g of finely ground air-dried sample (three samples per ecotype) were weighed into a glass screwcap bottle with 10 ml of chloroform, 1 ml of internal standard (ergotamine d-tartrate from Sigma-Aldrich) and 1 ml of 1 M NaOH and gently shaken for 24 h. After 15 min of centrifugation, the supernatant was pulled out. The supernatant was cleaned on an SPE column (1,0 g silica, 0,5 g sodium sulfate, separated with paper disc). The column was conditioned with 5 ml of CHCl3, before adding 5 ml of supernatant. After that, the column was washed twice with 1 ml of chloroform: acetone solution (3:1) and 1,5 ml of MeOH. Elution was made with 2,5 ml MeOH. Eluent was concentrated with N2 and reconstituted in 500 μl of MeOH. Analysis was performed using an HPLC system with a divinyl-benzene column, with fluorescence detection (excitation = 250 nm, emission = 420 nm). Mobile phase: ACN/2,6 mM ammonium carbonate (70:30), flow rate: 1,0 ml/min, injection volume: 20 μl.

Mean ergovaline content for grass species per locality was calculated as an average value from all plants of given species examined for ERV contents.

Bioclimatic data

For each location, GPS coordinates were taken during sampling. Further, using DIVA-GIS ver. 7.1.7 software (http:///www.diva-gis.org) 21 bioclimatic variables from WorldClim database (http://www.wordclim.org/current) were ascribed to each collection site. Mentioned variables were compiled on the basis of monthly averages of climate as measured at weather stations from a large number of global, regional, national, and local sources, for the 1950–2000 period (Hijmans et al. 2005).

The following bioclimatic variables were used:

  1. a)

    Temperature realted data:

BIO 1 - minimal temperature (°C);

BIO 2 - maximal temperature (°C);

BIO 3 - annual mean temperature (°C);

BIO 4 - mean monthly temperature range (°C);

BIO 5 - isothermality [(BIO 4/BIO 9)·100] (%);

BIO 6 - temperature seasonality [CV] (%);

BIO 7 - maximal temperature of the warmest month (°C);

BIO 8 - minimal temperature of the coldest month (°C);

BIO 9 - temperature annual range [BIO7 - BIO8] (°C);

BIO 10 - mean temperature of the wettest quarter (°C);

BIO 11 - mean temperature of the driest quarter (°C);

BIO 12 - mean temperature of the warmest quarter (°C);

BIO 13 - mean temperature of the coldest quarter (°C);

  1. b)

    Percipitation related data:

BIO 14 - annual precipitation (mm);

BIO 15 - precipitation of the wettest month (mm);

BIO 16 - precipitation of the driest month (mm);

BIO 17 - precipitation seasonality [CV] (%);

BIO 18 - precipitation of the wettest quarter (mm);

BIO 19 - precipitation of the driest quarter (mm);

BIO 20 - precipitation of the warmest quarter (mm);

BIO 21 - precipitation of the coldest quarter (mm).

Statistical analysis

The following data were ascribed to each locality:

  • Longitude, given in decimal format;

  • Latitude, given in decimal format;

  • EF, endophyte frequency per species - calculated separately for each grass species per locality and expressed as a % of inhabited plants in all plants examined.

  • ERV, ergovaline concentration per species - calculated as average value from all plants of a given species examined for ERV contents, expressed in mg·kg−1.

Considering spatial orientation of data in our research (GPS coordinates for all locations), SAM v. 4.0 (Rangel et al. 2010) software was used to calculate: linear regression analysis, Pearson’s spatial correlation coefficients and spatial autocorrelation index (Moran’s I). Pearson’s spatial correlation between variables were calculated using Dutileull estimator to correct degrees of freedom. Default number of classes and default distance classes were used while calculating Moran’s I separately for each grass species. Geographic distances were computed on the metric notation basis of coordinates (longitude and latitude), therefore distance classes were further given in kilometers. Autocorrelograms were drawn on the basis of distance classes (OX-axis) and calculated Moran’s I coefficients (OY-axis).

Results

Ecotype colonization

Grass ecotypes were collected at 226 localities from permanent grasslands in Poland. Most of the explored localities were located in the vicinity of water (ditch, stream, river, lake). A total number of 747 ecotypes of 26 grass species were collected in the form of a living plant, mostly of Festuca (F. rubra - 173 ecotypes, F. pratensis - 139, F. arundinacea - 31, F. ovina - 49, F. tenuifolia - 9 and F. gigantea - 2), but also of Lolium perenne − 152 ecotypes, L. multiflorum - 8, Deschampsia cespitosa - 103, Phleum pratense – 1, and Poa pratensis - 45 ecotypes (Table 1). Besides, the following minor grass species (grass species not abundant on permanent grasslands in Poland) were also collected: Agrostis gigantea – 3, Alopecurus pratensis – 3, Arrhenatherum elatius – 3, Bromus inermis – 2, Cynosurus cristatus – 3, Deschampsia flexuosa – 10, Festuca heterophylla – 1, Glyceria fluitans – 1, Holcus sp. – 1, Koeleria macrantha – 2, Phleum boehmerii – 2, Poa compressa – 1, Poa nemoralis – 1, Poa palustris – 1 and Sieglingia decumbens – 1.

Table 1 Endophyte occurrence (E+) in grass ecotypes and endophyte frequency per locality
Full size table

Endophytes were found associated with 158 (69.9 %) of 226 sampled localities. More than 34 % of the examined ecotypes were infected by Epichloë fungi. Endophytes were found in plants of F. rubra, F. pratensis, F. arundinacea, F. ovina, F. tenuifolia and F. gigantea and also of L. perenne, L. multiflorum, D. cespitosa, P. pratense and P. pratensis. The most colonized species were: F. pratensis (74.1 %), F. rubra (41.0 %), L. perenne (32.9 %) and F. arundinacea (32.2 %) (Table 1). No endophyte presence was noted in ecotypes of minor grass species.

All plants of timothy (Phleum pratense) were infected by endophytes, but for this species only one ecotype was analyzed. Mean endophyte frequency for grass species sampled from more than 10 localities ranged from 35.3 % for D. cespitosa to 90.0 % for F. arundinacea (Table 1).

Ergovaline contents

In our study it was found that endophyte infected (E+) plants were collected from 69.9 % of all localities. The majority (144 of 259, i.e. 55.6 %) of naturally infected grass ecotypes contained detectable amounts of ergovaline. Ergovaline contents ranged from 0.031 up to 1.517 mg·kg−1 (mean 0.168 mg·kg−1) (Table 2).

Table 2 Ergovaline contents in examined grass species
Full size table

The highest amounts of ergovaline (mean 0.998 mg·kg−1, range from 0.577 to 1.517) were produced by endophyte-infected F. arundinacea ecotypes. Ergovaline was found also in L. perenne, F. ovina and F. pratensis E+ ecotypes. However, mean alkaloid contents in these species (0.200; 0.145 and 0.125 mg·kg−1, respectively) were lower than for F. arundinacea. Very low amounts of ergovaline were detected for single ecotypes of F. gigantea and F. tenuifolia (0.031 and 0.054 mg·kg−1, respectively). Ergovaline was not detected in E+ ecotypes of D. cespitosa, P. pratense, L. multiflorum and P. pratensis.

From 144 ecotypes containing endophytes capable of producing ergovaline, 26.4 % produced this alkaloid in amounts higher than claimed to be chronic toxic for feeding animals (more than 0.2 mg·kg−1 according to Bony and Delatour 2000). The greatest number of these ecotypes was observed for F. pratensis (15 ecotypes, 21.7 %) and in all E+ ecotypes of F. arundinacea (10 ecotypes, 100 %). It should be noted that endophytes colonizing F. arundinacea ecotypes produced ergovaline in amounts far in excess of 0.2 mg·kg−1 (Table 2).

Endophyte frequency and ergovaline contents with relation to local climatic conditions

The overall variations of EF and ERV across all sampled locations were explained in small part by bioclimatic data (Table 3). The variation of EF was explained from 6 % to 15.2 % by temperature-related data and from 2.4 % to 11 % by precipitation-related data, as well as ergovaline concentration in plant tissues - from 3.5 % to 11.5 % and 5.6 % to 11.4 %, respectively.

Table 3 Results of linear regression analysis for the effect of bioclimatic variables on the overall variation of endophyte frequency and ergovaline concentration. Values of determination factor (R(Bacon 1993)) were given together with significance for linear regression analysis
Full size table

EF in Festuca arundiacea and F. pratensis was not significantly correlated with any from thirteen temperature–related bioclimatic variables (Table 4). The incidence of endophytes in Festuca rubra and Lolium perenne was positively related to temperature seasonality and temperature annual range. Extreme winter temperatures may additionally affect endophyte frequency in Festuca rubra. Frequency of plants of mentioned species inhabited by endophytes increased with decreasing winter extreme temperatures. Contrary to this, endophyte frequency in Lolium perenne ecotypes increased along with increasing maximal temperature of the warmest month.

Table 4 Pearson’s spatial correlation coefficients between bioclimatic variables of location sites and endophyte frequency in grass ecotypes. Statistically significant coefficients were marked with * (P > 0.05)
Full size table

Endophyte frequencies in Festuca pratensis and Lolium perenne plants were negatively correlated with seven and three of eight bioclimatic variables related to precipitation, respectively (Table 4). For Festuca arundinacea it has been calculated only for one precipitation-related variable (precipitation seasonality) and for Festuca rubra and F. ovina for none.

For mean ERV content some significant correlations were calculated only for Festuca pratensis for precipitation during summer months (Table 5). As precipitation decreased the frequency of ecotypes able to produce ergovaline increased. From all Festuca pratensis ecotypes hosting ergovaline producing endophytes, relatively high amounts of ERV (more than 0.2 mg·kg−1) were found in ecotypes collected in localities with average precipitation of the wettest month 77.2 mm, as compared to 82.3 mm for localities were ERV concentration was less than 0.2 mg·kg−1 (Table 6). This relation was similar in the case of other endophyte-hosting species, despite the insignificant correlation coefficients calculated.

Table 5 Pearson’s spatial correlation coefficients between bioclimatic variables of location sites and mean ERV contents in E+ grass ecotypes. Statistically significant coefficients were marked with * (P > 0.05)
Full size table
Table 6 Selected precipitation averages for localities with grasses hosting endophytes able to produce ergovaline above (more than 0.2 mg·kg−1) or below (less than 0.2 mg·kg−1) animal hazard level
Full size table

Spatial analysis of endophyte frequency and ergovaline contents

Distribution of endophyte frequency in Festuca plants were most probably random over the sampled area. In case of Lolium perenne low order (distance 40.5 km) as well as high order (distance 365.4 km) significant autocorrelations were calculated (Fig. 1 a, b, c, d, e).

Fig. 1
figure 1figure 1

Correlograms drawn on the basis of Moran’s I (spatial autocorrelation index) calculated for (a) Festuca pratensis, (b) Festuca arundinacea, (c) Lolium perenne, (d) Festuca rubra, (e) Festuca ovina and (f) bioclimatic variables averaged for temperature related data (BIO 1 – BIO 13, yellow bar) and precipitation related data (BIO 14 – BIO 21, blue bar). OY axis – Moran’s I values, OX – axis – distance classes (in km), asterisks (*) and dashed lines – significance of autocorrelation with P > 0.05 and P > 0.001, respectively. On Fig. 1a. – 1e. black bar represents endophyte frequency data and red bar – ergovaline concentration data

Full size image

Despite the random distribution of EF sampled Festuca species, ERV concentration low order (36.5 and 37.4 km) autocorrelations were significant for Festuca pratensis and Festuca rubra, respectively. Ecotypes with endophytes able to produce ergovaline tended to aggregate in nearby locations. In Lolium perenne clustering of ecotypes with ERV producing endophytes was much wider and reached 103 km.

Spatial analysis of bioclimatic data

Spatial distribution of bioclimatic variables over the sampled localities was similar up to a distance of 458 km (Table 7, Fig. 1 f). Low order (short distance - 36 km) positive autocorrelations decreasing gradually to 194–223 km, and approaching zero at distances 283–313 km and significant but negative autocorrelations were calculated up to 458 km. High order (long distance – 615 km) autocorrelations were positive for eight bioclimatic variables, mostly related to high summer temperatures and high winter precipitation values. Such long distances within the sampled area include localities near the sea and in mountains regions, where summer temperatures and winter precipitation fall within the same range of values. For other variables high order Moran’s I were significant and negative.

Table 7 Moran’s I autocorrelation coefficients for bioclimatic variables of area sampled in our research
Full size table

Discussion

This study revealed the presence of endophytic fungi of the genus Epichloë in Lolium perenne, L. multiflorum, Festuca pratensis, F. rubra, F. arundinacea, F. ovina, F. capillata, F. gigantea, Poa pratensis, Deschampsia cespitosa and Phleum pratense. National surveys on the prevalence of grass endophytes are sparse, however they confirm the presence of these fungi in the native ecotypes of the above-mentioned grass species (Pańka and Żurek 2005; Żurek et al. 2010a; Wiewióra et al. 2006; Żurek et al. 2012; Pańka and Żurek 2008). Results of the studies conducted for other regions of Europe and the World are also consistent with those obtained in this work, i.e. the grass species most often infected by endophytes are: L. perenne, F. arundinacea and F. pratensis (Zabalogogeazcoa et al. 2003; Lewis 2000; White and Cole 1985; Schardl and Philips 1997; Guillaumin et al. 2000). Lewis (2000) reported that endophyte infected grasses originated from 22 European countries. Most of the records are for L. perenne and overall 49 % of plants of these populations contained endophytes. The percentage of the populations that were infected ranged from 1 to 100 %, although populations with 100 % infection were uncommon. Some reports about endophyte infection in F. arundinacea show that 95 % of populations were infected and populations with 100 % were noted in all regions from which data was collected.

Some authors argue that the main factor influencing the variations in the frequency of endophytes in plant tissues is temperature (Ju et al. 2006). Their research showed a lower EF in F. arundinacea, where the average monthly temperatures are often below the minimum temperature of endophytes growth and above the minimum temperature for the growth of plants. It has also been confirmed in our previous studies (Żurek et al. 2013) that grass-endophyte symbiosis was more frequent in locations with higher temperature fluctuations over the year, with a wider range of extreme annual temperature conditions and with relatively lower precipitation. This is also consistent with the results obtained in this study, which indicate that higher endophyte frequencies in F. pratensis plants are observed in local dry conditions. However, Brosi et al. (2010) found no effect of elevated temperature on endophyte frequency in tall fescue.

Type and amount of alkaloid produced in plant tissues depends on the following factors: endophyte species, season, infected grass genotype, environmental conditions (the content of nitrogen in the soil, drought, local climatic conditions) (Hahn et al. 2007; Leuchtmann et al. 2000; Vazquez-de-Aldana et al. 2010). From the point of view of the production of alkaloids by endophytic fungi, important determinants of this process are the climatic conditions (mainly the amount of precipitation and air temperature) prevailing in the area, which are inhabited by grass endophytes (Reed 2002; Salminen et al. 2005). Increased production of alkaloids by endophytic fungi frequently occurs when the plant inhabited by the endophyte is subjected to drought stress (Reed et al. 2011; Arechavaleta et al. 1992; Hahn et al. 2008). These relationships have been demonstrated in our research in the case of ecotypes of F. pratensis (Table 4). The wettest quarter in Poland is usually the summer months i.e. July, August and September, with August being the wettest month (Lorenc 2005). The production of ergovaline in Festuca pratensis as a process was not determined by precipitation but it was observed increased of this alkaloid production while precipitation decreased. Ecotypes with endophytes able to produce higher amounts of ergovaline were much more frequent in regions of lower summer precipitation, i.e. Central Poland. Current research further confirms a strong relationship between host species, fungal endophytes and local climatic conditions. Similar conclusions were made by Bony and Delatour (2000), who found that the content of this toxin in European grasses occurs at a high level, however, this trait is highly variable, depending on the development phase of the plants and the climatic conditions prevailing in the study area. Also Vazquez-de-Aldana et al. (2010) and Eerens et al. (1998) observed significant interaction between ergovaline concentration and lower water availability. The first of them have observed that lower ergovaline content detected in 2001 could be related to the higher precipitation during the first half of this year (398 mm), in comparison to the precipitation noted during the same period in 2002 (222 mm) and 2003 (245 mm). Contrary to this, Repussard et al. (2014a, 2014b) confirmed positive correlation between ERV level and cumulative degree-days, whereas rainfall had no effect. Similar conclusions considering precipitation were also raised by Brosi et al. (2010).

Our results show that the highest amount of ergovaline was detected in F. arundinacea ecotypes. Similar results were reported by Leuchtmann et al. (2000) and Vazquez de Aldana et al. (2003). This may indicate that the endophytic species associated with this grass species are genetically determined to produce larger amounts of ergovaline. Chronic toxic amounts of ergovaline in forage causing weight loss, milk production decrease and daily weight increase reduction range from 0.2–0.4 mg·kg−1 forage dry matter, while disease symptoms may occur at 0.3–0.5 mg·kg−1 for horses or at 0.4–0.7 mg·kg−1 for cattle (Jensen et al. 2007). The concentration of ergovaline can increase with water stress as reported in F. arundinacea (Arechavaleta et al. 1992) and L. perenne (Hahn et al. 2008), although the ability to produce the alkaloid was also confirmed for the edophytic fungi colonizing F. rubra and F. pratensis (Cagaš et al. 1999). Until recently it was thought that Epichloë uncinata infected plants of F. pratensis is incapable of producing ergot alkaloids (Panaccione et al. 2001), but some result from the Polish and Czech regions shown that ergovaline is also quite often detected in these plants (Cagaš et al. 1999; Mikołajczak et al. 2005; Podkówka et al. 2011; Pańka et al. 2013 and Pańka et al. 2011). According to the latest research E. uncinata is a hybrid between E. typhina and E. bromicola and the same species of grasses can independently host more than one Epichloë species (Saikkonen et al. 2016). Therefore, the infection of Festuca pratensis by E. typhina or E. festucae cannot be excluded.

Fleetwood et al. (2007) have shown that ergovaline biosynthetic genes are expressed in planta and suggest that specific plant conditions may be required for the induction of ergopeptine biosynthetic genes in the fungal endophyte. Alternatively, it is not clear to what extent differences in alkaloid concentration depend on the metabolic activity of the fungus. Maybe are they the result of changes in fungal gene expression or enzyme activities to different and changeable concentrations of fungal tissue in the fungus/host association (Rasmussen et al. 2009).

In our research 26.4 % of E+ ecotypes contained amounts of ergovaline higher than 0.2 mg·kg−1. In a few cases the average amount of ergovaline was much higher than for other localities and was detected in more than one species per locality. Also, in case of some localities and grass species, none from the tested E+ ecotypes contained any detectable amount of ergovaline. However, due to the specific nature of Polish grasslands the threat to animals associated with ergovaline content in swards is rather low. In swards of the majority of Polish grasslands, up to 25–30 dicotyledonous herb species grow together with grasses (Grzegorczyk 2010; Trąba and Wolański 2011; Warda and Kozłowski 2012). Therefore, even a high concentration of toxins in only a few plants per grassland will be ‘diluted’ in green sward or in hay. However, due to the increasing area of highly productive short rotation meadows, the possibility of toxic concentrations of ergovaline or other alkaloids becomes quite real. Very important are also local stress factors such as drought, which may induce a significant increase of alkaloids in grasses (Zabalgogeazcoa and Bony 2005). In the last 25 years, a lot of information can be found about frequently occurring droughts in our country, which included large areas. Regions located in Central Poland are especially prone to frequent droughts (Łabędzki 2004).

Low order (short distance) positive autocorrelation can be due to migration of organisms and is probably not necessarily associated with gene flow over the distances involved (Sokal and Oden 1978). This relationship has been found in the case of EF for Lolium perenne and Festuca ovina as well as ERV for Festuca arundinacea, Lolium perenne and Festuca rubra. Short distance positive autocorrelation could also be found with environmental aspects, as can be seen in the case of the analysis of spatial structure of climatic data in our research. Therefore, two mechanisms could be involved in the observed spatial structure of endophyte frequency and ergovaline production. First - plant population survival is similar in similar climatic conditions (Pearson and Dawson 2003). Second – endophytes and grasses may be transmitted by seeds even for a distances of kilometres. Sufficiently mobile organisms can be expected to track the geographical position of their bioclimatic conditions (Collingham et al. 1996; Collingham et al. 2000). Therefore a combination of both factors (survival rate and seed transmission) may result in short distance positive autocorrelation.

High order (long distance) negative autocorrelation (EF and ERV for Festuca arundinacea and Lolium perenne, EF for Festuca pratensis and F. ovina) is explained by the nature of the most different localities that are farthest apart (Sokal and Oden 1978). Areas sampled in our research includes localities from North of Poland (near the Baltic sea) and localities in the South of Poland (mountains regions). There is a distance of more than 750 km between these localities, therefore it is probably true that the most different localities are also the farthest apart. And when this high order negative autocorrelation is coupled with low order positive autocorrelation, this should be the most common correlogram profile observed in nature over long distances (Sokal and Oden 1978). A correlogram of this shape has been shown for averaged temperature and precipitation bioclimatic data in our research (Fig. 1f).

Conclusions

We explored spatial variation of mutualistic interactions between a host organism (grass plant) and infecting fungus (endophyte) by means of its intensity (endophyte frequency per locality) and one from many outcomes, i.e. ergovaline production. Mutualism, as one among many interaction classes, is most likely to change in different ecological contexts (Chamberlain et al. 2014). It is possible that the variation of climatic conditions in Poland is too narrow to show a wide range of possible plant-fungus interaction outcomes, this may explain why the proportion of overall variation of EF and ERV ascribed to climatic condition was rather low.

Ergovaline is quite common under conditions of semi natural grassland in Poland. Grasses of nearly half of localities under study (47.3 %) were found to host endophytes able to produce ergovaline. Animal hazard associated with the increased toxin amount in grass was found on 25 (11.1 %) of all the localities, mostly of relatively low precipitation in the Summer. Spatial structure of endophyte incidence on grassland in Poland is most probably random, however short distance aggregation occurs mostly in case of Festuca pratensis, Festuca rubra and Lolium perenne ERV concentration.

Considering recent projections and analyses of climate change, there are signs of deteriorating agroclimatic conditions and a need for adaptive measures, to either increase soil-water availability or to develop crop drought resistance in the majority of European climatic zones (Trnka et al. 2011). Therefore, if our grassland will suffer from summer drought periods, percentage of endophyte infected plants and average amounts of ergovaline may increase. And this will directly influence animal health.