Evidence for genetic differentiation between choke-inducing and asymptomatic strains of the Epichloe grass endophyte from Brachypodium sylvaticum.
Epichloe is the causative agent of choke disease. It forms fruiting structures (stromata) producing conidia (asexual spores) and ascospores (sexual spores) around developing grass inflorescences, thereby preventing host flowering and seed set. The ascospores, which are formed in perithecia after "fertilization" of the primary stroma by conidia of another mating type (White and Bultman 1987), are ejected into the air and may serve as means for horizontal transmission to new host plants (Western and Cavett 1959; Kohlmeyer and Kohlmeyer 1974; White and Morgan-Jones 1987). The negative impact on host fitness by host sterilization suggests that these associations are antagonistic.
Acremonium, on the other hand, has no sexual stage in the life cycle but grows as a symptomless endophyte within its host plant and is transmitted vertically by infecting host ovules and seed (Sampson 1933; Latch et al. 1984). The absence of host sterilization together with the beneficial effects mentioned above and prevalences up to 100% in certain associations indicate that asymptomatic infections by Acremonium endophytes are beneficial.
In some grass/endophyte associations intermediate degrees of choking occur, which allow for both strategies of fungal reproduction (White 1988; Leuchtmann et al. 1994). In these associations, only some of the flowering tillers produced by an infected plant bear stromata, whereas other tillers are healthy and give rise to infected seed. In many populations seed-transmission may be the only means of dispersal in a subset of individuals or in all of the infected plants.
If infected, grass species are usually associated with endophytes exhibiting only one of the three life history strategies described (White 1988). Differences in the degree of host sterilization might therefore be controlled by the genotype of either partner of the symbiosis or by environmental factors. Differential growth of the fungus and the host during a critical stage of development has been suggested to determine the degree of disease expression (Kirby 1961). In fact, additions of gibberellic acid causing faster growth of the inflorescences have been shown to reduce the amount of stroma formation in Dactylis glomerata plants (Emecz and Jones 1970). High levels of nitrogen fertilizer added to the soil had a similar effect and reduced the number of stromata in Festuca rubra, while some individuals never exhibited choke under both treatments (Sun et al. 1990). However, if the patterns of stroma formation found in nature result from an evolutionary process, the degree of disease expression must have a genetic basis in at least one of the partners of the symbiosis. Some evidence is provided by studies comparing growth rates of endophytes isolated from stromata-bearing and symptomless hosts in vitro. These studies suggest that stromata-forming strains grow faster than asymptomatic strains, both from the same or from different host species, and that differences in growth rates on certain sugars are correlated with disease expression (White and Chambless 1991; White et al. 1991 a,b, 1993b).
In this study we report on genetic differentiation of stromata-forming and asymptomatic strains of the Epichloe endophyte of Brachypodium sylvaticum based on allozyme markers, compare patterns of genetic variation within and between endophyte populations with different levels of sexual reproduction, and discuss factors that can contribute to the relative success of the different strategies of propagation. Brachypodium sylvaticum forms a widespread association with its endophyte, where both modes of reproduction occur in various degrees and varying among individual plants or plant populations. Therefore, it is an ideal system to examine the evolution of disease expression. Results of this investigation will contribute to our understanding of the origin and evolution of endophyte/grass symbioses.
MATERIALS AND METHODS
Host and Life Cycle of Study Organism
Brachypodium sylvaticum is a perennial, caespitose grass of the tribe Triticeae (Clayton and Renvoize 1986), which is native to temperate Eurasia but has also been introduced to other parts of the world (e.g., North America). It is very common in Swiss woodlands, particularly in older clearings or along roads of mixed deciduous forests dominated by beech, where it grows on humid, usually calcareous, soils.
During the vegetative phase of grass development, the endophyte of B. sylvaticum grows intercellularly in aboveground plant parts and systemically invades all vegetatively formed tillers. At the time of host flowering, the endophyte may exhibit two alternative modes of reproduction: (1) vertically through seed transmission, or (2) horizontally through spores in the sexual cycle. Seed-transmitted strains do not undergo a sexual cycle. Hyphae grow into ovules as the inflorescences develop and proliferate in the nucellus tissues, where they become incorporated into seeds (White et al. 1991b). Flowering and reproduction of the host is not prevented in this case. Horizontal transmission involves a sexual cycle, with endophytes forming external fruiting structures (stromata) on and around developing inflorescences of the host and effectively preventing flowering and seed-set of the host (choke disease). The stromata first produce conidia that are functional spermatia (male gametes). Transfer of conidia of opposite mating type is required for fertilization (heterothallic mating), for which a specialized insect is the main vector (Kohlmeyer and Kohlmeyer 1974; White and Bultman 1987). After fertilization of the stromata, perithecia with asci and ascospores develop. Filamentous ascospores are wind transmitted and may serve as inocula to infect other plants either directly or indirectly through iterative germination (Webster 1980; Bacon and Hinton 1988). The mode of reproduction of the endophyte and the degree of choking greatly varies among different sites and different grass individuals.
Previous data and additional surveys herein indicated that populations of B. sylvaticum in Switzerland are always infected and that infection rates usually reach 100% (Leuchtmann 1992; pers. obs.). Most populations of B. sylvaticum are asymptomatic and stroma formation in a given population is restricted to a subset of the infected plants. Stromata-forming individuals are usually only partly sterilized (ranging from 20% to 80%) allowing for both sexual and asexual reproduction of the endophyte.
The endophyte of B. sylvaticum appears to be strictly host specific and reproductively isolated from the endophytes of other host grasses (Leuchtmann 1992). Mating tests have further confirmed that it constitutes a separate biological species which therefore should no longer be included in Epichloe typhina (Leuchtmann, unpubl. results).
Source of Isolates
Isolates were obtained from endophyte-infected plants collected at 10 different sites in Switzerland (see [ILLUSTRATION FOR FIGURE 1 OMITTED] for locations) and were isolated from surface sterilized tissues, as described by Leuchtmann and Clay (1988), or as single ascospores (Table 1). Each isolate was sampled from a different plant at least 1 m apart along randomly chosen transects. In addition, multiple ascospore isolates were derived from single stromata (5-10 per stroma) at two sites (Albisguetli and Sihlwald). Collection sites, sexual expression of endophyte on sampled plant, source of isolates, and the associated isozyme genotypes with numbers of isolates are listed in Table 1. Stromata-forming endophyte strains were found only at three collection sites (called mixed sites), whereas at all other sites no choked plants were observed during the period of the study. Our primary focus was concerned with mixed collection sites where between 19 and 129 isolates per site were sampled. The other sites were included for comparison and were often represented by only a few isolates.
Sample preparation and techniques of horizontal starch gel electrophoresis followed standard procedures previously applied to Acremonium and Epichloe endophytes (Leuchtmann and Clay 1990; Leuchtmann 1994). Mycelium for enzyme extraction was grown in axenic cultures on a liquid V-8 medium for 14 to 21 d and extracts of lyophilized samples subjected to starch gel electrophoresis. Gel slices were stained for enzyme activity following published protocols (Soltis et al. 1983; Wendel and Weeden 1989). Eleven enzyme systems representing 13 presumed loci were routinely resolved: Acid phosphatase (ACP; EC no. 220.127.116.11), Aconitase (ACO; EC no. 18.104.22.168), Aldolase (ALD; EC no. 22.214.171.124), Diaphorase (DIA; EC no. 126.96.36.199), Glucose-6-phosphate dehydrogenase (G6P; EC no. 188.8.131.52), Leucine aminopeptidase (LAP; EC no. 184.108.40.206), Malate dehydrogenase (MDH; EC no. 220.127.116.11), Phosphoglucose isomerase (PGI; EC no.18.104.22.168), Phosphoglucomutase (PGM; EC no. 22.214.171.124), 6-Phosphogluconate dehydrogenase (6PG; EC no. 126.96.36.199), and Triosephosphate isomerase (TPI; EC no, 188.8.131.52). ACP, G6P, and 6PG were run on citric acid/histidine HCL buffer at pH 7.0 (System I); ACO, ALD, and LAP on tris citric acid/tris citric acid buffer at pH 8.0 (System II); DIA, MDH, PGI, PGM, and TPI on tris citric acid/tris citric acid buffer at pH 7.2 (System V) (for details see Soltis et al. 1983).
Alleles were designated by the electrophoretic mobility of their homomeric protein products relative to that of the alleles of the most frequent isozyme genotype that were assigned Rf-values of 100 (Table 2). In all but one enzyme system, either a single band or two independently varying bands were resolved for all isolates; consistent with a haploid organism with a single locus or two isozyme loci. The exception was aconitase where twelve isolates exhibited a triple-banded phenotype instead of the expected single banded phenotype. This banding pattern was interpreted as the phenotype of a dimeric enzyme with duplicated genes and two different alleles. Since all isolates with this aconitase phenotype were derived from single ascospores, heterokaryosis or aneuploidy as explanation for multiple gene copies seems unlikely. The allelic combination (multilocus genotype) of an isolate is termed its allozyme genotype.
Analysis of isozyme data was based on allele frequencies of isolate samples at each collection site (population), while for three populations (Albisguetli, Pilatus, and Sihlwald) a subdivision into stromata-forming (S) and asymptomatic (NS) isolates was made. Cluster analyses was carried out on Rogers' genetic distance (Rogers 1972) using the unweighted pair-group method with arithmetic averaging (UPGMA). The total gene diversity [H.sub.T] (Nei 1973) was calculated using the mean allele frequency over all isolates and all loci. Gene diversity within populations ([H.sub.S]) was measured with the biased estimate (Nei 1973) while [Mathematical Expression Omitted] was the mean of [H.sub.S] over all populations. Genetic differentiation among populations ([G.sub.ST]) was the proportion of the total gene diversity due to differentiation among populations, calculated as [Mathematical Expression Omitted] (Nei 1973). To estimate genetic differentiation among collection sites and between different S and NS subpopulations, gene diversity analysis was applied on different levels of subdivision (Table 3). All calculations, including cluster analysis, were performed with the statistical package BIOSYS-1, Vers. 1.7 (Swofford and Selander 1989).
To estimate gene flow, which is an indication of the mean number of migrants per generation, Nm (N is the population size, m is the migration rate) was calculated assuming that Nm = 1/2 ([1/[G.sub.ST]] - 1) in haploid organisms (McDermott and McDonald 1993).
To measure clonality of stromata-forming and asymptomatic endophyte subpopulations, we used a statistical test designed to detect associations between alleles at different loci (Brown et al. 1980; Maynard Smith et al. 1993). For reasons of sampling design, this test was only performed on isolates collected from host tissues at the Albisguetli site. The analysis does a pairwise comparison between the electrophoretic types of all isolates while recording the number of loci K at which they differed. With n isolates there are a total of n(n - 1)/2 such pairs. The probability that two isolates bear different alleles at the jth locus is [h.sub.j] = 1 - [[Sigma].sub.[p.sub.ij]], while [p.sub.ij] is the frequency of the ith allele at the jth locus. If the alleles at different loci are independent (i.e., if there is no linkage disequilibrium), the expected variance of K becomes [TABULAR DATA FOR TABLE 1 OMITTED] [V.sub.E] = [Sigma][h.sub.j](1 - [h.sub.j]). Maynard Smith et al. (1993) introduced the index of association between loci ([I.sub.A]) which compares the observed variance of K (= [V.sub.O]) with the expected variance of K (= [V.sub.E]) defined as [I.sub.A] = [V.sub.O]/[V.sub.E] - 1. If isolates originate from a single random mating population, the expected value of [I.sub.A] is zero. To test whether the observed variance of K ([V.sub.O]) was significantly greater than [V.sub.E], we used a permutation test creating 1000 populations with the exact allele frequencies and sample sizes observed (but without assuming linkage). For each population, the variance of K was calculated and compared with [V.sub.O]. If variances are equal or greater than [V.sub.O] in less than 5% of the random permutations, then the difference between [V.sub.O] and [V.sub.E] is considered statistically significant, which means there is a linkage disequilibrium [TABULAR DATA FOR TABLE 2 OMITTED] (for a discussion of permutation tests, see Potvin and Roff 1993).
Among the 266 isolates derived from B. sylvaticum, there were 19 allozyme genotypes detected (Table 2). A genotype is defined as the allelic combination of enzymes at the six variable loci. An additional seven loci (ACP, ALD, G6P, LAP, MDH-1, PGI, and 6PG) used for calculations of genetic distances were monomorpic among all isolates and are not included in the table. The polymorphic loci exhibited two to four alleles per locus adding up to 21 alleles in total. The most variable enzyme was aconitase (ACO), where four alleles were resolved. In several genotypes two alleles appeared to be present for the ACO locus indicated by a double or triple banded pattern on the gel. For calculations these patterns were treated as heterozygous. No enzyme activity was detected at the MDH-2 locus in genotype C.
The distribution of endophyte genotypes found at the 10 different sites [ILLUSTRATION FOR FIGURE 1 OMITTED] is shown in Table 1. At each site 100% of the B. sylvaticum plants surveyed were infected. However, stroma formation was observed only at four sites (Albisguetli, Sihlwald, Pilatus, and Weissenstein) and usually in less than 10% of the plants occurring in small clusters within the site (from site Weissenstein no choke-inducing strains were isolated). The most common genotype (A) was present at all [TABULAR DATA FOR TABLE 3 OMITTED] sites. It was the only genotype found at six of the sites and the predominating genotype at all other sites, except Sihlwald, where genotype O was more abundant. All other genotypes were represented by only a small fraction of the isolates accounting for between 10% and less and than 1% of the isolates.
Fourteen genotypes were found among S, sexual isolates and 10 among NS, asexual isolates (Table 2). Of the NS isolates, five were also found among the S isolates and included all of the more common genotypes (A, B, E, F, G) of asexual isolates. Only the rare NS genotypes were not represented among the sexual isolates.
The Albisguetli site, from which the largest sample (129 isolates) was collected, was by far the most diverse with 14 genotypes. At this site, six genotypes were associated with asymptomatic plants and 13 genotypes with stromata-forming plants, while five further genotypes were common to both plant groups (Table 1). Several of the genotypes from S plants were characterized by private alleles not found among isolates from NS plants; these included alleles at the ACO locus (allele 86), PGM-1 locus (106), PGM-2 locus (63), and TPI locus (76). Furthermore, isolates derived from ascospores exhibited a higher diversity (0.23 genotypes per isolate) than isolates from plant tissues (0.17 genotypes per isolate).
In contrast, at the Sihlwald site, there were more genotypes among isolates from asymptomatic plants than among those derived from ascospores. This may, however, be due to the [TABULAR DATA FOR TABLE 4 OMITTED] smaller number of plants collected at this site. The site exhibited a private allele at the DIA locus (141), which dominated in the ascospore isolates (genotype O).
Of the isolates from the Pilatus site the same two genotypes were found on S and on NS plants, but in unequal frequencies. Genotype A dominated on NS plants whereas genotype F was more common on S plants, but the genotype/isolate ratio was much greater in choke-inducing strains (0.29) than in asymptomatic strains (0.07). At all three sites, the isozyme genotypes were not independent of presence or absence of stroma formation (G-test of most frequent genotype (A) versus all others: Albisguetli: G = 62.6, P [much less than] 0.001; Pilatus: G = 18.5, P [much less than] 0.001; Sihlwald: G = 4.4, P [less than] 0.05).
Among the sites where only symptomless strains were examined, only the 26 isolates from the Weissenstein site exhibited variation. At this site, however, stroma formation was observed on plants not sampled for the study. Two of the six genotypes (R and S) found at the Weissenstein site had double alleles at the ACO locus unique to this site, while the remaining genotypes were found at several other sites as well. All 32 isolates from the Hubweg site and from the remaining sites represented by four to five isolates (Greifensee, Brugg, Vattis, Pfafers, and Lalden) were uniform of the single genotype A (Table 1).
Genetic relatedness of genotypes based on Rogers's genetic distances is shown in the cluster diagram (cophenetic correlation = 0.692) of Figure 2. Genotypes of NS isolates did not group separately from genotypes of S isolates, but occurred in several subclusters of the dendrogram together with the S genotypes. This is suggesting that clonal NS genotypes may have arisen independently several times.
A linkage analysis of allozymes was performed with the isolates from the Albisguetli site only (Table 4). Within the NS subpopulation and within the total population, the analysis indicated a significant association of the alleles at the variable loci (ACO, PGM-2, and MDH-2) with [I.sub.A] = 0.689 (P [much less than] 0.001) and [I.sub.A] = 0.675 (P [much less than] 0.001), respectively, where as in the S subpopulation a linkage among loci was not significant ([I.sub.A] = 0.114, P = 0.21).
The cluster diagram based on genotype frequencies (cophenetic correlation = 0.98) of endophyte populations (or S and NS subpopulations) at the 10 sites clearly showed that the S and NS isolates belong to genetically differentiated subpopulations [ILLUSTRATION FOR FIGURE 3 OMITTED]. All NS populations and NS subpopulations from Albisguetli and Pilatus were genetically very similar and grouped in a single main cluster distinct from the S subpopulations. Only the NS subpopulation from Sihlwald did not fall within this main cluster, but was more similar to the S subpopulations from Pilatus and Albisguetli. Concurrently, the stromata-forming strains from Sihlwald were also the most divergent subpopulation of all sites at a distance of approximately 0.17. The relative distances between the S and NS subpopulations were in the same order of magnitude for all three sites.
Statistical analysis of gene diversity (Nei 1973) was performed to quantify genetic variation within and between sites, and between S and NS subpopulations (Table 3). In the comparisons made among sites, the average gene diversity [Mathematical Expression Omitted] was greater within S subpopulations (0.062) than within NS subpopulations (0.017) indicating a higher amount of genetic variation present in S strains compared to NS strains. The proportion of gene diversity due to differentiation among sites ([G.sub.ST]) was smaller between S subpopulations (0.728) than between NS subpopulations (0.846). Accordingly, the derived migration rates (Nm) were more than twice as high for S subpopulations (0.187) compared to NS subpopulations (0.091, Table 3).
Comparing the gene diversity between S and NS subpopulations at the three sites at which stromata-forming isolates were analyzed, [H.sub.T] values were greatest at the Sihlwald site (0.497) and smallest at the Albisguetli site (0.274) with the [G.sub.ST] value reaching 97.6% at the Pilatus site (Table 3). These values were much higher than those calculated for among S sites and higher or at least comparable to the among NS-sites values. Likewise, derived migration rates (Nm) were lower for the Pilatus site (0.012) and slightly higher for the Sihlwald (0.093) and the Albisguetli site (0.178) compared to the among NS sites. This clearly indicated that within a site there is a tendency for genetic differentiation of S and NS subpopulations, so that the subpopulations may be, at least in part, reproductively isolated. These findings are also in concordance with the relatively large genetic distances calculated between subpopulations [ILLUSTRATION FOR FIGURE 3 OMITTED].
Results from the isozyme analysis clearly showed that stromata-forming (S) and asymptomatic (NS) isolates of Epichloe endophytes from B. sylvaticum do not belong to a single randomly mating population at sites where both occur, but form genetically differentiated subpopulations. Genetic differentiation was expressed in the large genetic distances between S and NS subpopulations at the same site, as well as in the relatively low estimates of migration rates. In addition, isozyme genotypes were distributed nonrandomly among NS and S isolates and significant linkage within the NS subpopulation was present at the Albisguetli site.
Evidence for differences in the mode of reproduction between NS and S strains is provided by the linkage analysis. The significant linkage among the NS strains (in contrast to the S strains) is consistent with the view that asymptomatic strains are clonal (and will therefore exhibit linkage disequilibrium), while stromata-forming strains are not. The absence of linkage association among alleles in the S strains suggests that strains from stromata-bearing hosts propagate sexually (i.e., are transmitted horizontally) at least occasionally or have lost sexuality and horizontal spread only recently. Significant linkage among the NS strains, however, could also be caused by a Wahlund effect implying the presence of (at least) two genetically differentiated populations (Hartl and Clark 1989). Similarly, the significant linkage occurring when combining the isolates of both S and NS subpopulations can either be due to genetic differentiation among S and NS strains or due to the effect of linkage within the NS strains.
Differences in allozyme diversity within populations of either mode of reproduction may be interpreted in several ways. First, the higher allozyme diversity within S isolates compared to NS isolates (Table 3) supports the idea that asymptomatic endophytes originated from sexual strains since the NS strains harbor only a subset of the alleles found among S isolates. A higher variability of isozymes among Epichloe than among Acremonium endophytes was also found by Leuchtmann and Clay (1990), who analyzed endophytes from different host species. Further evidence that the asexual Acremonium endophytes are derived from Epichloe is provided by the observation that Acremonium species generally have a narrower host range than Epichloe (Clay 1989; Leuchtmann 1992). Moreover, sequence analysis of the internal spacer region 2 of the ribosomal DNA suggests that certain Acremonium strains evolved recently from Epichloe endophytes (Schardl et al. 1991; An et al. 1992; Schardl and Siegel 1993). On the other hand, lower genetic diversity within the NS strains could result from loss of alleles in asexually reproducing populations by random genetic drift (Hartl 1989).
The different degrees of genetic differentiation between sites in S and NS populations (Table 3) suggest differences in the amount of gene flow among populations with the two modes of reproduction. Stromata-forming subpopulations appeared to be genetically less differentiated than NS subpopulations. This might be caused by a higher amount of gene flow among S subpopulations by means of ascospores. Asymptomatic strains lack an infective stage and can only migrate within seeds. Because of their unequal weights and different dispersal, ascospores and seeds may differ in their migration distances, which might account for the differences in genetic differentiation observed among S and NS subpopulations. All these findings suggest that there are differences in the ability of stroma formation among fungal strains, since disease expression could be correlated with allele and genotype frequencies.
However, several genotypes or alleles, especially the more frequent ones, were found in both symptomless and stromata-bearing plants (Table 2) and the NS subpopulations exhibited merely a subset of the alleles found in the S subpopulations with no private alleles. This observation could be interpreted in several ways. First, allozyme genotypes might represent clones, which differ in their ability to form stromata. Hence, genotypes that are predominantly isolated from symptomless hosts (e.g., genotype A) might have a higher threshold for sexual reproduction than genotypes isolated mostly from choked plants (e.g., genotype E). Whether this threshold is reached might be determined by the host genotype and/or environmental conditions to which infected plants are exposed. In rare occasions and under special circumstances, NS strains could become sexual (or S strains asexual), which would allow for gene exchange between S and NS subpopulations.
An alternative explanation for the seemingly incomplete genetic isolation between S and NS subpopulations might be the fact that only a small number of loci have been analyzed and that differences in other loci might exist. Furthermore, NS subpopulations could have only arisen in recent evolutionary history and are therefore genetically little differentiated. In both hypotheses, reproductive isolation and genetic drift are most likely a major factor for the differences in allele and genotype frequencies between S and NS subpopulations at mixed sites.
Evidence to infer evolutionary patterns of the fungal life cycle based on the population structure found in this study is ambiguous. The close similarity of NS subpopulations [ILLUSTRATION FOR FIGURE 3 OMITTED] and the predominance of genotype A among NS isolates at most sites suggest spreading of one or few asymptomatic strains from single origins, whereas distance relationships of individual genotypes [ILLUSTRATION FOR FIGURE 2 OMITTED] would indicate that asexual clones have arisen independently many times. Genotypes of NS isolates occurred in several subclusters of the dendrogram together with the S genotypes. Many of the rare NS genotypes (e.g., J and P), however, might have a lower threshold value for stroma formation allowing them to change their mode of reproduction very easily. Further investigations using more sensitive genetic markers and cross inoculation experiments controlling for host genotype and environmental conditions are needed to elucidate the role of the fungal genotype in disease expression.
Genetic differences among fungal strains in relation to stroma formation raises questions about the direction of evolution and the selective factors that might favor one or the other mode of reproduction. The fitness of stromata-forming strains depends mainly on the efficiency of horizontal spread, since vertical transmission is reduced (or absent) compared to completely asymptomatic strains. Contagious spread of Epichloe under natural conditions is not very well documented, but appears to occur at a very low rate (Large 1954; Leuchtmann and Clay, in press). Artificial infection of the grass Dactylis glomerata by ascospores of Epichloe typhina was only successful when the spores were applied to cut flowering stems under high humidity yielding infection rates of 10% or less (Western and Cavett 1959). This indicates that horizontal spread may be inefficient, at least in dry habitats. Moreover, since maintaining a functioning stroma is a moisture-consuming process, stroma formation may only be advantageous where abundant moisture is available (i.e., in cool, moist climate; White et al. 1993a). Therefore, associations with intermediate disease expression might be adaptations to drought prone areas whereby the fungus can disperse through the seeds when environmental conditions are unfavorable for successful stroma formation. On the other hand, Clay (1993) suggested that environmental conditions found at higher latitudes or elevations, where cool season grasses dominate, might reduce the chances of contagious spread and therefore may select for seed transmission.
In associations with partially sterilizing and seed-borne endophytes, the fitness of stromata-forming strains also depends on the efficiency of vertical transmission and the frequency of multiple infections. Provided symptomless fungal strains are transmitted to all offspring and hosts cannot be infected with more than one strain, sterilizing endophytes could not compensate for their loss in vertical transmission by horizontal spread because more and more hosts would be occupied already by an asymptomatic endophyte. The efficiency of vertical transmission is not known and may differ among associations. In B. sylvaticum, vertical transmission can be assumed to be very high since infected plants always produce endophyte containing seeds, and usually 100% of the plants in natural populations are infected (pers. obs.). On the other hand, the observed population structure conforming with random mating among stromata-forming strains suggests that at least occasional horizontal transmission does occur. Therefore multiple infections of a grass individual by different strains cannot be excluded. Multiple infections was also an assumption to explain the origin of a heterokaryotic Acremonium strain in Lolium perenne (Schardl et al. 1994).
Another selective factor for or against stroma formation may be the life history of the host. Clay (1988a) observed a higher frequency of symptomless infections in caespitose compared to rhizomatous grass species. This is interpreted as having evolved because sterility of rhizomatous grasses reduces their fitness less than it does to caespitose plants. The parasite's fitness, however, does not depend on the host's fitness as long as there is horizontal transmission that is frequent enough to guarantee the colonization of new hosts (Toft and Aeschlimann 1991; Herre 1993). Furthermore, completely sterilizing endophytes of rhizomatous grasses do not depend on horizontal transmission exclusively, because they are also transmitted vertically through shoots of the grass.
Stroma formation not only affects the mode of transmission but also the mode of reproduction of the endophyte (sexual or asexual), which may be tightly linked to whether the relationship between host and endophyte is antagonistic or mutualistic. It is assumed that asymptomatic associations are mostly beneficial based on numerous studies with other grass hosts, whereas stroma formation reduces the host's ecological fitness (Clay 1988a; Siegel 1993). Sexual reproduction has been shown to be absent in various mutualistic symbionts that live within their partners. Its absence was explained by the constant environment provided by the host tissue that "generates pressures against continuing evolutionary change so that activities such as sex which generate variation are selected against" (Law and Lewis 1983). Since sex is considered to be costly (Lewis 1987), asexual mutants should spread unless sex provides some advantages compared to asexual reproduction. In antagonistic symbioses, host genotypes are selected for resistance and offer a continuously changing, that is, deteriorating environment. If genetic recombination helps overcoming host resistance, parasites that reproduce sexually may be fitter than asexual ones (Jaenike 1978). In mutualistic symbioses, however, susceptible host genotypes increase in frequency because of their higher fitness compared to genotypes incapable of symbiosis (Law 1985), which leads to a more and more favorable environment for endophytes, in which sexual reproduction has no advantages and may only be costly in the short term. Similarly, selection against recombination by reproductive isolation, which maintains host specificity of the endophyte strains, has been suggested as a potential force acting against stroma formation (Clay 1993).
However, the mode of transmission could be uncoupled from that of reproduction, for example by loss of ascospore formation but maintenance of horizontal transmission by conidia. Incompletely developed stromata without ascospores have been observed on a population of Ammophila breviligulata along the New Jersey shore (White et al. 1993c). The absence of perithecia in this population, however, may be due to the absence of the opposite mating type, especially since A. breviligulata is cultivated (White et al. 1993c). Few other host species have been observed to bear stromata without perithecia, for example Holcus lanatus in southern England. The Epichloe population on that host does not seem to be mating type limited, because Holcus plants with perithecia are quite common in the region (White pers. comm.). The rarity of such observations suggests that selection on the mode of reproduction might not be the driving force acting against stroma formation.
The evidence for genetic differences in the ability of stroma formation prompted a discussion about the adaptive significance of fruiting structures in grass endophytes. Stroma formation is linked to three different factors on which selection might act, the way of transmission, the mode of reproduction, and the virulence of a fungal strain. It will therefore be necessary to evaluate each factor independently by experimental or comparative approaches to estimate its significance in the evolution of life cycles of grass endophytes.
We are grateful to P. van Tienderen for writing a computer program for the randomization tests performed on the linkage data. We thank L. Gygax for assistance in the field and J. A. Shykoff for helpful discussions and valuable comments on the manuscript.
AN, Z.-Q., J.-S. LIU, M. R. SIEGEL, G. BUNGE, AND C. L. SCHARDL. 1992. Diversity and origins of endophytic fungal symbionts of the North American grass Festuca arizonica. Theor. Appl. Genet. 85:366-371.
ARACHEVALETA, M., C. W. BACON, C. S. HOVELAND, AND D. E. RADCLIFFE. 1989. Effect of the tall rescue endophyte on plant response to environmental stress. Agron. J. 81:83-90.
BACON, C. W., AND D. M. HINTON. 1988. Ascosporic iterative germination in Epichloe typhina. Trans. Brit. Mycol. Soc. 90: 563-569.
BACON, C. W., J. K. PORTER, J. D. ROBBINS, AND E. S. LUTTRELL. 1977. Epichloe typhina from toxic tall fescue grasses. Appl. Environ. Microbiol. 34:576-581.
BROWN, A. H. D., M. W. FELDMAN, AND E. NEVO. 1980. Multilocus structure of natural populations of Hordeum spontaneum. Genetics 96:523-536.
CLAY, K. 1987. Effects of fungal endophytes on the seed and seedling biology of Lolium perenne and Festuca arundinacea. Oecologia 73:358-362.
-----. 1988a. Clavicipitaceous fungal endophytes of grasses: Coevolution and the change from parasitism to mutualism. Pp. 79-105 in D. L. Hawksworth and K. Pirozynski, eds. Coevolution of fungi with plants and animals. Academic Press, London.
-----. 1988b. Fungal endophytes of grasses: A defensive mutualism between plants and fungi. Ecology 69:10-16.
-----. 1989. Clavicipitaceous endophytes of grasses: Their potential as biocontrol agents. Mycol. Res. 92:1-12.
-----. 1993. The ecology and evolution of endophytes. Agric. Ecosyst. Environ. 44:39-64.
CLAYTON, W. D., AND S. A. RENVOIZE. 1986. Genera Graminum. Grasses of the world. Kew Bulletin Additional Series XIII, London.
EMECZ, T. I., AND G. D. JONES. 1970. Effect of gibberellic acid on inflorescence production in cocksfoot plants with choke (Epichloe typhina). Trans. Brit. Mycol. Soc. 55:77-82.
HARTL, D. L., AND A. G. CLARK. 1989. Principles of population genetics. Sinauer, Sunderland, MA.
HERRE, E. A. 1993. Population structure and the evolution of virulence in nematode parasites of fig wasps. Science 259:1442-1445.
JAENIKE, J. 1978. An hypothesis to account for the maintenance of sex within populations. Evol. Theory 3:191-194.
KIRBY, E. J. M. 1961. Host-parasite relations in the choke disease of grasses. Trans. Brit. Mycol. Soc. 44:493-503.
KOHLMEYER, J., AND E. KOHLMEYER. 1974. Distribution of Epichloe typhina (Ascomycetes) and its parasitic fly. Mycologia 66:77-86.
LARGE, E. C. 1954. Surveys for choke (Epichloe typhina) in cocksfoot seed crops, 1951-53. Plant Pathol. 3:6-11.
LATCH, G. C., M. J. CHRISTENSEN, AND G. J. SAMUELS. 1984. Five endophytes of Lolium and Festuca in New Zealand. Mycotaxon 20:535-550.
LATCH, G. C., W. F. HUNT, AND D. R. MUSGRAVE. 1985. Endophytic fungi affect growth of perennial ryegrass. NZ J. Agric. Res. 28: 165-168.
LAW, R. 1985. Evolution in a mutualistic environment. Pp. 145-170 in D. H. Boucher, ed. The biology of mutualism. Oxford Univ. Press, New York.
LAW, R., AND D. H. LEWIS. 1983. Biotic environments and the maintenance of sex - Some evidence from mutualistic symbiosis. Biol. J. Linn. Soc. 20:249-276.
LEUCHTMANN, A. 1992. Systematics, distribution, and host specificity of grass endophytes. Nat. Toxins 1:150-162.
-----. 1994. Isozyme characterization, persistence, and compatibility of fungal and grass mutualists. Pp. 21-33 in C. W. Bacon and J. F. White Jr., eds. Biotechnology of endophytic fungi of grasses. CRC Press, Boca Raton, FL.
LEUCHTMANN, A., AND K. CLAY. 1988. Atkinsonella hypoxylon and Balansia cyperi, epiphytic members of the Balansiae. Mycologia 80:192-199.
-----. 1990. Isozyme variation in the Acremonium/Epichloe complex. Phytopathology 80:1133-1139.
-----. In press. The population biology of grass endophytes. In G. Carroll and P. Tudzynski, eds. The Mycota IV. Springer, Berlin.
LEUCHTMANN, A., C. L. SCHARDL, AND M. R. SIEGEL. 1994. Sexual compatibility and taxonomy of a new species of Epichloe symbiotic with fine fescue grasses. Mycologia 86:802-812.
LEWIS, W. M. J. 1987. The cost of sex. Pp. 33-55 in S. C. Stearns, ed. The evolution of sex and its consequences. Birkhauser, Basel, Switzerland.
MAYNARD SMITH, J., N. H. SMITH, M. O'ROURKE, AND B. G. SPRATT. 1993. How clonal are bacteria? Proc. Nat. Acad. Sci. USA 90: 4384-4388.
MCDERMOTT, J. M., AND B. A. McDONALD. 1993. Gene flow in plant pathosystems. Annu. Rev. Phytopathol. 31:353-373.
NEI, M. 1973. Analysis of gene diversity in subdivided populations. Proc. Nat. Acad. Sci. USA 70:3321-3323.
POTVIN, C., AND D. A. ROFF. 1993. Distribution-free and robust statistical methods: Viable alternatives to parametric statistics? Ecology 74:1617-1628.
ROGERS, J. S. 1972. Measures of genetic similarity and genetic distance. Stud. Genet., Univ. Texas Publ. 7213:145-153.
SAMPSON, K. 1933. The systemic infection of grasses by Epichloe typhina (Pers.) Tul. Trans. Brit. Mycol. Soc. 18:30-47.
SCHARDL, C. L., AND M. R. SIEGEL. 1993. Molecular genetics of Epichloe typhina and Acremonium coenophialum. Agric. Ecosyst. Environ. 44:169-185.
SCHARDL, C. L., J.-S. LIU, J. F. WHITE JR., R. A. FINKEL, Z.-Q. AN, AND M. R. SIEGEL. 1991. Molecular phylogenetic relationships of nonpathogenic grass mycosymbionts and clavicipitaceous plant pathogens. Plant Syst. Evol. 178:27-41.
SCHARDL, C. L., A. LEUCHTMANN, H.-F. TSAI, M. A. COLLETT, D. M. WATT, AND D. B. SCOTT. 1994. Origin of a fungal symbiont of perennial ryegrass by interspecific hybridization of a mutualist with the ryegrass choke pathogen Epichloe typhina. Genetics 136:1307-1317.
SIEGEL, M. R. 1993. Acremonium endophytes: our current state of knowledge and future direction of research. Agric. Ecosyst. Environ. 44:301-321.
SOLTIS, D. E., C. H. HAUFLER, D. C. DARROW, AND G. J. GASTONY. 1983. Starch gel electrophoresis of ferns: A compilation of grinding buffers, gel and electrode buffers, and staining schedules. Am. Fern J. 73:9-27.
SUN, S., B. B. CLARKE, AND C. R. FUNK. 1990. Effect of fertilizer and fungicide applications on choke expression and endophyte transmission in chewings rescue. Pp. 62-63 in Proceedings of the International Symposium on Acremonium/Grass Interactions. Louisiana Agricultural Experimental Station, Baton Rouge, LA.
SWOFFORD, D. L., AND R. B. SELANDER. 1989. BIOSYS-1, a computer program for the analysis of allelic variation in population genetics and biochemical systematics. Ill. Nat. Hist. Surv., Champaign.
TOFT, C. A., AND A. AESCHLIMANN. 1991. Introduction: coexistence or conflict? Pp. 1-12 in C. A. Toft et al., eds. Parasite-host associations: Coexistence or conflict? Oxford Univ. Press, New York.
WEBSTER, J. 1980. Introduction to fungi. Cambridge Univ. Press, Cambridge.
WENDEL, J. F., AND N. F. WEEDEN. 1989. Visualization and interpretation of plant isozymes. Pp. 5-45 in D. E. Soltis and P. S. Soltis, eds. Isozymes in plant biology. Dioscorides Press, Portland, OR.
WESTERN, J. H., AND J. J. CAVETT. 1959. The choke disease of cocksfoot (Dactylis glomerata) caused by Epichloe typhina (Fr.) Tul. Trans. Brit. Mycol. Soc. 42:298-307.
WHITE, J. F., JR. 1987. Widespread distribution of endophytes in the Poaceae. Plant Disease 71:340-343.
-----. 1988. Endophyte-host associations in forage grasses. XI. A proposal concerning origin and evolution. Mycologia 80:442446.
WHITE, J. F., JR., AND T. L. BULTMAN. 1987. Endophyte-host associations in forage grasses. VIII. Heterothallism in Epichloe typhina. Am. J. Bot. 74:1716-1721.
WHITE, J. F., JR., AND D. A. CHAMBLESS. 1991. Endophyte-host associations in forage grasses. XV. Clustering of stromata-bearing individuals of Agrostis hiemalis infected by Epichloe typhina. Am. J. Bot. 78:527-533.
WHITE, J. F., JR., AND G. MORGAN-JONES. 1987. Endophyte-host associations in forage grasses. XI. Concerning Acremonium typhinum, the anamorph of Epichloe typhina. Mycotaxon 29:489-500.
WHITE, J. F., JR., J. P. BREEN, AND G. MORGAN-JONES. 1991a. Substrate utilization in selected Acremonium, Atkinsonella and Balansia species. Mycologia 83:601-610.
WHITE, J. F., JR., A. C. MORROW, G. MORGAN-JONES, AND D. A. CHAMBLESS. 1991b. Endophyte-host associations in forage grasses XIV. Primary stromata formation and seed transmission in Epichloe typhina: Developmental and regulatory aspects. Mycologia 83:72-81.
WHITE, J. F., JR., A. E. GLENN, AND K. F. CHANDLER. 1993a. Endophyte-host associations in grasses. XVIII. Moisture relations and insect herbivory of the emergent stromal leaf of Epichloe. Mycologia 85:195-202.
WHITE, J. F., JR., G. LEWIS, S. SUN, AND C. R. FUNK JR. 1993b. A study of distribution of Acremonium typhinum in populations of red fescue in southwest England and in vitro growth comparisons to isolates from North America collections. Sydowia 45:388-394.
WHITE, J. F., JR., G. MORGAN-JONES, AND A. C. MORROW. 1993c. Taxonomy, life cycle, reproduction and detection of Acremonium endophytes. Agric. Ecosyst. Environ. 44:13-37.
|Printer friendly Cite/link Email Feedback|
|Author:||Bucheli, Erika; Leuchtmann, Adrian|
|Date:||Oct 1, 1996|
|Previous Article:||Effects of differential pollen-tube growth on hybridization in the Louisiana irises.|
|Next Article:||Identical major gene loci for heavy metal tolerances that have independently evolved in different local populations and subspecies of Silene vulgaris.|