Do Botanophila flies provide reproductive isolation between two species of Epichloë fungi? A field test


Author for correspondence:
Thomas L. Bultman
Tel: +1 616 395 7372


  • Epichloë spp., fungal endophytes of cool season grasses, produce collars of mycelium (stromata) on host stems that Botanophila flies visit for egg laying. Flies transfer fungal gametes among stromata and thereby serve to cross-fertilize fungi. Hence, the interaction is analogous to insect pollination in angiosperms. While most Epichloë species are not interfertile, Epichloë typhina and Epichloë clarkii can hybridize.
  • We investigated whether Botanophila flies play a role in the reproductive isolation of the two Epichloë species at a field site in southwestern Switzerland. We estimated the density of stromata and collected fly larvae and stromata occurring on plants.
  • While most ascospores collected from both species indicated intraspecific mating, 9.3% of fungal fruiting bodies contained spores of hybrid origin. Two species of Botanophila larvae occurred on stromata and both preferred E. typhina. Yet, both fly species laid eggs on both fungal species.
  • While preferences by Botanophila flies should influence reproductive isolation between the fungi, other mechanisms are likely more important. Our data, which show hybrid ascospores are produced, suggest postzygotic isolating mechanisms are an important means of reproductive isolation.


Speciation is the fundamental process that generates biological diversity and thus is of central interest to biologists. Biological species are defined by a lack of interbreeding between species and therefore traits that reduce gene flow between populations are critical in understanding species formation and maintenance. Reproductive isolating barriers, as traits that mediate reductions in gene flow, are central to understanding the process of speciation and, as such, are widely studied (Widmer et al., 2009). Reproductive isolating mechanisms can be divided into pre- and postzygotic barriers (Dozhansky, 1937; Coyne & Orr, 1989). Identifying the different types of isolating barriers and assessing their relative contribution to reproductive isolation are major challenges facing evolutionary biologists (Coyne & Orr, 1997; Widmer et al., 2009).

Fungi are well suited for this type of investigation because they display a wide array of life cycles and numerous species complexes are known to occur, yet studies of reproductive isolation in fungi are underrepresented in the literature relative to those on plants or animals (Le Gac & Giraud, 2008). Fungal examples include the anther smut fungus, Microbotryum violaceum, which infects multiple host species in the Caryophyllaceae (van Putten et al., 2007). Investigators studied transfer of host pollen (as fungal spore mimics) using fluorescent dye and found that insect vectors differed somewhat between plant host species. The researchers concluded that insect vector behavior (as a form of premating isolation in the fungus) contributed to host race differentiation in the fungus, but that a combination of factors were involved. Others found that premating isolation occurs due to temporal shifts in reproduction in yeasts (Murphy et al., 2006) and assortative mating in Homobasidiomycota (Le Gac & Giraud, 2008). Postmating isolation has also been found in fungi. Microbotryum (Le Gac et al., 2007) and Neurospora (Dettman et al., 2003) both show lower viability following heterospecific compared with conspecific matings.

A fungal system that is well suited to exploring questions about reproductive isolating mechanisms is the intriguing association between ascomycetous fungi (Epichloë spp.) and anthomyiid flies (Botanophila spp.). The interaction is a mutualism that is functionally similar to pollination (Bultman & White, 1988). Epichloë spp. live within grasses and in early summer produce collars of mycelium (stromata) on the surface of host stems (Schardl, 1996). Stromata prohibit (or choke out) development of the inflorescence on that culm and thus the infection is termed ‘choke’. The fungus produces gametes (spermatia) in abundance on the surface of stromata. Sexual reproduction in Epichloë generally requires mating between different individuals; that is, the fungus is heterothallic (White & Bultman 1987; but see Rao & Baumann, 2004). Development of the ascospore phase requires transfer of spermatia of one mating type to an unfertilized stroma of the opposite mating type (infected grasses harbor just one fungal individual). Transfer is accomplished by Botanophila spp. flies (Bultman & White, 1988) that visit stromata for feeding and egg laying (Kohlmeyer & Kohlmeyer, 1974). Immediately following oviposition, flies spread their feces along the stroma, and because feces contain viable spermatia (collected from multiple stromata of both mating types), the behavior results in cross-fertilization of the fungus (Bultman et al., 1998).

Following cross-fertilization, perithecia, the ascocarps (fruiting bodies) in which meiotically-formed ascospores are produced, begin to develop on the surface of stromata. Ascospores, as infective propagules, act like seed analogs (Bultman et al., 1998). Botanophila eggs hatch in 4 d and feed on perithecia. Even though larvae feed on developing perithecia, the net effect of Botanophila visitation on Epichloë reproduction is positive (Bultman et al., 1995, 2000). Because cross-fertilization results in proliferation of fungal tissue (the perithecia), one would expect larvae feeding on fertilized stromata to have more food to eat and thus should perform better than those feeding on stromata lacking perithecia. Thus, one might expect selection for female flies that are species-specific in their visitation patterns as monolectic flies should provision their young with perithecia and ascospores (Bultman & Leuchtmann, 2008).

Despite this expectation, previous studies aimed at assessing Botanophila visitation preferences have been inconclusive. An investigation of fecal contents of field-caught flies showed that individuals often carried spermatia of more than one fungal species in their gut (Bultman & Leuchtmann, 2003). Yet, based on spermatia harbored in the gut, flies did show some selectivity as they preferred some fungal species and discriminated against others (Bultman & Leuchtmann, 2003). A shortcoming of the study was that flies were collected at a common garden site where grasses infected by several species of Epichloë had been planted; thus flies were exposed to an unnaturally high diversity of co-occurring Epichloë species. Bultman & Leuchtmann (2003) also showed that flies oviposited on multiple species when in small screen enclosures. Yet, fly oviposition behavior when confined to enclosures might not reflect what happens naturally in the field.

If flies discriminated in host selection, then one would expect that they would be able to distinguish among Epichloë species. Evidence for this has been collected by Steinebrunner et al. (2008) who showed that two compounds (the sesquiterpene alcohol chokal K and a methylester) found in the headspace above Epichloë stromata were attractive to Botanophila flies and that the relative concentration of these compounds differed among several Epichloë species. This situation resembles Australian Chiloglottis orchids in which species vary in their floral odour and pollinators appear to have driven speciation among the plants (Peakall et al., 2010). However, Botanophila fly species did not exhibit preferences for different blends of the two fungal compounds (Steinebrunner et al., 2008). Thus, it is possible that flies are generally attracted to Epichloë stromata and subsequent decisions about oviposition are made based on tactile, not olfactory, cues.

While most Epichloë species are not interfertile, morphologically distinct Epichloë typhina and Epichloë clarkii can hybridize (Leuchtmann & Schardl, 1998). Despite this, hybrids of the two species are not observed in the field (F. Steinebrunner & A. Leuchtmann, unpublished) and thus, mechanisms must exist to reproductively isolate them. We investigated whether Botanophila flies play a role in the reproductive isolation of the two Epichloë species at a field site in southwestern (SW) Switzerland where both species coexist. At the site E. typhina infects Dactylis glomerata, while E. clarkii infects Holcus lanatus. Our study is the first to investigate this question in the field with naturally occurring plants, fungi and insect visitors.

Materials and Methods

Study area

Field work was conducted at the Aubonne Arboretum, Canton Vaud, in SW Switzerland. The site is a hillside that is dominated by grasses of the genera Anthoxanthum, Dactylis, Holcus and Poa.

Assessing stroma density

A 30 m tape was placed from the top to the bottom of the hillside and we recorded every culm of D. glomerata and H. lanatus exhibiting choke within 1 m of either side of the tape. We recorded the fungal species and the number of stromata present on each clump. Three vertical transects were completed, each separated by 3 m.

Sampling ascospore progeny

The frequency of conspecific and heterospecific matings in E. typhina and E. clarkii was assessed by collecting stromata with perithecia along transects placed horizontally on the study hillside. Using stroma density data from vertical transects (above) we divided the hillside up into sections that accentuated differences in stroma density of Epichloë species along the hillside. Those transects, ‘A1’, ‘A2’, ‘B’ and ‘C’, were 4, 8, 16 and 24 m, respectively, from the top of the hill. Each transect was 30 m long. One to four stromata of E. typhina and E. clarkii were collected from each clump of D. glomerata and H. lanatus within 2 m of either side of the tape. Ascospores within individual perithecia from three locations on stromata (basal, middle and upper portions) were observed microscopically. Ascospores of the two species are easily distinguished: E. clarkii produces spear-shaped part-spores while E. typhina produces much larger whole-spores (White, 1993).

Botanophila host use

Host use of ovipositing Botanophila flies was assessed by collecting larvae hatched from eggs they deposited on stromata along three 30 m transects oriented horizontally on the hillside study site at positions identical to those used to sample ascospores (transects A1, A2, B and C; see the previous section, ‘Sampling ascospore progeny’). The grass clump with stromata closest to the tape at every 2 m interval was sampled.

Relative production of spermatia produced by each of the two fungal species along transects A1, A2 and B was estimated using its stroma abundance (above) and multiplying this by its average stroma length (determined by measuring stromata on 10 potted grasses each of D. glomerata (= 123 stromata) and H. lanatus (= 107 stromata) using a digital caliper).

Taxonomic identification of Botanophila larvae

Taxonomic keys for Botanophila are based on male genitalia (Collin, 1967; Ackland, 1972). Thus, we used DNA sequence data to separate larvae collected from stromata (see the previous section, ‘Botanophila host use’) into species and, when possible, compared these with DNA sequences from identified adult males (Leuchtmann, 2007). Species identification was based on cytochrome oxidase subunit II (COII) sequence.

Total genomic DNA was prepared as in Leuchtmann (2007). A total of 141 larvae yielded DNA that was useable for sequencing. The PCR conditions, primers used and purification of DNA products were as previously described (Leuchtmann, 2007). Sequencing reactions were performed with the BigDye Terminator Cycle Sequencing Kit (PE Applied Biosystems, Carlsbad, CA, USA) and products separated on a capillary 3130XL ABI Genetic Analyzer (Frankfurter, Germany). Sequences were aligned by eye using sequencher 4.8 (Gene Codes, Ann Arbor, MI, USA).

Evolutionary distances (number of base substitutions/site) between 141 individuals and three reference sequences (Botanophila phrenione (Genbank accession number EF064349), Botanophila‘taxon 1’ AL-2006 isolate 69 (EF064348) (these two taxa were selected because they match the species found in our samples) and Egle parva (EF064366)) were computed using the Maximum Composite Likelihood method (Tamura et al., 2004) and the evolutionary history was inferred with the neighbor-joining method (Saitou & Nei, 1987) using mega4 (Tamura et al., 2007). Bootstrap percentages for support of the B. phrenione and Botanophila‘taxon 1’ clades were calculated by maximum likelihood analysis using the dnaml program in phylip (Felsenstein, 2005) with 500 replicates.

Determination of gut transit time in Botanophila adults

Adult Botanophila were collected at Aubonne and transferred to glass vials with cut culms of H. lanatus bearing E. clarkii stromata. Stromata had red food coloring applied to them. Flies were allowed to feed on the stained stromata for 15 min after which stromata were removed. Flies were then observed every 30 min for 3 h to determine if and when red-stained frass was excreted.


Stroma density

We found that the density of stromata of both Epichloë species varied along the hillside study site. Stromata of E. clarkii were much more common than those of E. typhina (12.7 m−2 vs 0.32 m−2, respectively; Mann–Whitney = 9.0, < 0.05). Epichloë clarkii stromata tended to be more common along the bottom half of the hillside, while E. typhina stromata were rare on the bottom half and more common on the upper half of the hillside (Table 1).

Table 1.   Mean densities of Epichloë clarkii and Epichloë typhina on three transects arranged vertically along the hillside at the Aubonne (Switzerland) study site
Fungal speciesTransect sectionEndpoints (m)Stroma frequencyStroma density (per m2)
  1. Section ‘A’ began at the top of the hill and section ‘C’ ended near the bottom of the hill.

E. clarkiiA0–15612.02
E. typhinaA0–15190.69

Ascospore progeny

We found that most ascospores collected from mature stromata had morphology consistent with that of the stromal parent (Table 2). That is, ascospores produced within perithecia on E. clarkii stromata formed spear-shaped part-spores typical of E. clarkii (White, 1993) and ascospores from E. typhina stromata remained whole. However, five (of 54 or 9.3%) perithecia produced both part-spores and whole ascospores within single asci, signifying that interspecific mating had occurred.

Table 2.   Ascospore type from perithecia collected along four transects across the hillside study site at the Aubonne Arboretum (Switzerland)
HostTransectNumber of grass clumpsNumber of peritheciaSpore type(s)
  1. Transect ‘A1’ was closest to the top of the hill, while transect ‘C’ was closest to the bottom. Note: transects ‘A1’ and ‘A2’ were both within the upper hillside transect section (‘A’) from Table 1, while transects ‘B’ and ‘C’ corresponded to sections ‘B’ (middle hillside) and ‘C’ (lower hillside) of Table 1, respectively.

DactylisA123Epichloë typhina
DactylisA2716E. typhina
   1E. typhina and Epichloë clarkii
DactylisB21E. typhina
   4E. typhina and E. clarkii
HolcusB512E. clarkii
DactylisC11E. typhina
HolcusC616E. clarkii

Botanophila identification

Sequence data from the COII gene showed that Botanophila larvae belonged to one of two clades (Fig. 1). Including sequence data from adult fly species identified previously using morphological characters allowed us to identify one clade as B. phrenione. The other clade matched to a second species that has not yet been identified and was referred to as ‘taxon 1’ by Leuchtmann (2007).

Figure 1.

 Phylogram of Botanophila larvae collected from Holcus lanatus and Dactylis glomerata at Aubonne, Switzerland. The phylogram is based on sequences of the mitochondrial cytochrome oxidase gene (COII). It is inferred from a maximum-likelihood analysis. Branch lengths are scaled to the evolutionary distances (number of base substitutions/site, see scale bar) used to infer the tree (Tamura et al., 2004). Branches supported by > 95% of the bootstrap replicates are shown next to the corresponding branch. Larvae collected from Dactylis are indicated with circles and those from Holcus are indicated with stars.

Botanophila host use

We found that ovipositing B. phrenione strongly preferred stromata of E. typhina in areas of the study site (transects ‘A1’, ‘A2’ and ‘B’) where co-occurrence of both fungal species was greatest (Fig. 2, χ2 = 96.95, df = 1, < 0.005). Similarly, taxon 1 also preferred E. typhina, but its preference was not as strong (Fig. 2, χ2 = 11.89, df = 1, < 0.05). Even with these preferences, both fly species laid eggs on both fungal species.

Figure 2.

 Frequency of occurrence of larvae of Botanophila larvae on stromata on Holcus lanatus and Dactylis glomerata along transects where both plant species co-occurred in the middle and upper part of the hillside study area (transect sections ‘A1’, ‘A2’ and ‘B’ from Table 1). Expected frequencies were a function of stroma abundance and stroma size.

Gut transit time

We found that captive flies fed stromatal mycelium stained with food coloring excreted the colored frass within 30 min.


Our results show that both Botanophila species at Aubonne preferred E. typhina for egg laying. However, their preferences were by no means absolute, as both species laid eggs on both E. typhina and E. clarkii. This result mirrors that of Bultman & Leuchtmann (2003) who collected Botanophila (species not certain) flies from a common garden containing several Epichloë species. They found that individual flies carried spermatia from one to several Epichloë species in their gut, but, nonetheless, exhibited a preference relative to the host species available. Interestingly, the preference was for E. typhina and not E. clarkii (Bultman & Leuchtmann, 2003), which is the same pattern we found collecting larvae from stromata at Aubonne. Epichloë typhina infects several grass species (whereas other Epichloë species are generally restricted to one host species; Schardl et al., 2004) and may often be the most common fungus Botanophila flies encounter in Europe. A preference for E. typhina may reflect the higher frequency of this host in the field.

Preferences by both fly species for E. typhina should reduce heterospecific transfer to E. clarkii relative to what would be expected if flies showed no preference. Yet, given that fly preference was far from absolute, the contribution flies made to reproductive isolation between the fungi was likely low.

Nonetheless, while our results reveal that Botanophila species visit multiple Epichloë species, flies could still provide prezygotic reproductive isolation if individual flies display a high degree of constancy in their visitation (Proctor et al., 1996). That is, if a fly tends to visit stromata of one species in sequence and then switches to another species on which it ‘majors’ for a subsequent sequence of visitations, then the fly could predominantly transfer conspecific spermatia to stromata. However, if the transit time for spermatia to travel through the gut is long compared with duration of a majoring visitation sequence, then this scenario would be less likely to provide reproductive isolation among the fungal species. Our assessment of transit time revealed that spermatia spend relatively little time in the gut and can travel through the digestive tract in as little as 30 min. As flies may spend up to 30 min feeding on a single stroma (T Bultman, pers. obs), it is possible that much of what a fly consumes on a stroma could be deposited on the next one or two stromata that the fly visits.

Even though flies visited both fungal species, the prevalence of hybrid fungal offspring was low (Table 2). One possible explanation for infrequent hybrid offspring is that when a stroma receives a mixture of E. clarkii and E. typhina spermatia, intraspecific mating reactions (plasmogamy and subsequent proliferation of tissue to form perithecia) interfere with the interspecific ones, as suggested by Chung & Schardl (1997), who hand-crossed several Epichloë species. While the two species are interfertile (Leuchtmann & Schardl, 1998), there still may be competitive interactions between intraspecific and interspecific matings that might occur on the stromal surface.

While hybrid offspring were rare, we did find some mature stromata that contained ascospores produced from interspecific matings (Table 2). This contrasts with extensive sampling of vegetative tissue from Epichloë infecting H. lanatus and D. glomerata at the Aubonne site, which has never revealed hybrids (F Steinebrunner & A Leuchtmann, unpublished). Thus, when formed, hybrid ascospores must suffer from poor viability or inability to infect either grass host species. Our results are the first to implicate postzygotic isolation as a mechanism isolating the two Epichloë species.

In summary, we show that Botanophila flies show some host preferences in a natural setting and thus affect the level of interspecific mating between E. typhina and E. clarkii (relative to what would occur if flies showed no preferences). Yet, flies do not provide complete reproductive isolation between the two fungal species. Fungi likely attract flies carrying spermatia of both (and perhaps even more) fungal species (as found in Bultman & Leuchtmann, 2003). Thus, flies are not the primary mechanism of isolation, but rather must act in concert with mating reaction interference (Chung & Schardl 1997), which should ‘weed out’ most heterospecific spermatia vectored by flies, and postzygotic inviability or host incompatibility, which would filter out any hybrids that were produced.


B. Blattmann offered technical support for molecular work. The Association de l’Arboretum national du Vallon de l’Aubonne kindly permitted us to conduct research on their grounds. D. Schmidt provided information on location and timing of stroma formation at Aubonne. The Institute of Plant Ecological Genetics at ETH provided a welcoming and stimulating environment for TLB, TJS and AD during their stay in Zurich. Three anonymous reviewers offered helpful and insightful suggestions. Financial support was provided by NSF-CRUI (DBI-03300840) award to TLB.