**4. Discussion**

Our results sugges<sup>t</sup> that potentially entomopathogenic fungi are invariably present in insect populations. Fungal infections were recorded in all subsets of our laboratory rearing data (host species \* year), with just one exception (*C. pusaria* in 2014, but note the low samples size, Table 2). The detected diversity of fungi was notably high. Altogether, 25 species of fungi from 7 orders were isolated from the laboratory and the field experiments, with two-thirds (16 species) belonging to the Hypocreales (Table 1). Six of the collected fungi could not be unambiguously affiliated to a described species on the basis of their ITS DNA sequences (Table 1), and UNITE species hypotheses with respective DOI codes [22] are used for communicating on these. This might have been due to the possibility that cryptic undescribed species were involved, or that voucher specimens of known species have not ye<sup>t</sup> been sequenced, or the inapplicability of the ITS region for discriminating sibling species. Greatest fungal diversity was found on *Ematurga atomaria* (17 spp.), followed by *Cabera pusaria* (11), *Hypomecis punctinalis* (6), *Chiasmia clathrata* (2), and *Orthosia gothica* (2). However, the numbers of species recorded for each host are well consistent with species-specific sample sizes (Table 2) so that these figures should not be interpreted as an indication of differences in the community of fungi associated with different moth species.

Among the unambiguously entomopathogenic fungi, here defined as members of Cordycipitaceae, we identified two abundant (*Akanthomyces muscarius* and *Cordyceps farinosa*) and five scarce species (Table 1). In addition, members of the families Hypocreaceae, Nectriaceae, Aspergillaceae, and Mucoraceae were frequently found growing on dead insect pupae. These fungi were especially common in the field experiment, often accompanying a species of the Cordycipitaceae. Therefore, we may consider such fungi to primarily take advantage of pupae killed by other pathogens, but it cannot by any means be excluded that some of these may still possess thus far unrecognized opportunistic abilities to cause the death of the insects (see [24]). Further studies are needed to establish the nutritional strategies of fungi that are repeatedly found on dead insects.

Our study focused on fungal infections of folivorous larvae, which can only be brought about by plant- and air-borne propagules. In particular, the hosts were not in contact with soil and leaf litter, which are the environments considered to constitute reservoirs for entomopathogenic fungi [2,4]. This may explain why our samples did not include some well-known and common entomopathogens, such as *Beauveria* and *Metarhizium* spp., and suggests that the full spectrum of the fungi associated with natural populations of the studied insects may be considerably broader than recorded in the present study. Several of the presumable saprotrophs/potential entomopathogens, for which we have identified no or just very few previous records on insects, have been reported from plants, soil, or also from air and water. The fact that such fungi were more common on pupae from the field than from the lab experiment suggests that prolonged exposure to the natural environment favors the deposition of different fungi on insects and/or their host plants (but see also [2]).

Our laboratory-derived data do not sugges<sup>t</sup> any strong specialization of the fungi to particular host species. However, the just moderate amount of data available did not allow us to perform any formal analyses of specialization patterns. Indeed, 9 of the 13 fungal species that were found only on one host were represented by just a single observation. Nevertheless, it should be noted that the most abundant species—*Cordyceps farinosa* and *Akanthomyces muscarius*—were both found as readily infecting all the three most numerous hosts (*C. pusaria*, *E. atomaria*, and *H. punctinalis*), with no evidence of preferring one species over the other.

The overall prevalence of fungal infections in our laboratory rearings varied among different subsets of the data from 0 to 6% (excluding subsamples with less than 100 pupae). This value was 13.3 and 55.5% for the 2 years of the field experiment (0 to 6% vs. 2.2 and 24%, if to consider Cordycipitaceae only), indicating 10 times higher incidence of fungal infections than in the laboratory rearings. These values of prevalence can alone be interpreted as evidence of a non-negligible role of the insect–fungus interactions in the ecology of studied moths.

The lab-based estimates can underestimate the prevalence of entomopathogens in nature as the insects in the laboratory culture should be less exposed to various potential sources of infection than in the field, e.g., soil or infected insects [23,27]. Alternatively, laboratory mass rearings of insects might be prone to disease outbreaks, leading to higher prevalence values in the lab compared to the field. This appears not to have been the case, as epizootic levels were not reached in any of the subsets of our data. Additionally, the diversity of fungal pathogens recorded in the lab, as well as their similarity with field collections, provides evidence against outbreak of a particular fungus in our experimental facilities. The observation that fungal pathogens are always present at low frequencies is well consistent with the decades-long experience of insect rearing by some of the authors (but see [28]). Such a pattern might indicate that the presence of fungal conidia is not a limiting determinant of the prevalence of fungal diseases (see also [29]) but instead, the condition of the host may be decisive—only the weakest individuals are unable to resist the infection [4,30].

Currently, ecological knowledge about entomopathogenic fungi is primarily based on studying a few well-known species of fungi and isolating entomopathogens mainly from soil rather than describing complete fungal communities of particular insect species [3]. This has produced a skew in our knowledge, with the conclusions mainly based on a few members of the Hypocreales such as *Beauveria* and *Metarhizium* species [8]. The present study is one of the first that has aimed to document a full set of fungi isolated from several lepidopteran species (see also [24,25]). The detected diversity should inspire further studies—in addition to considerable bionomic data to be gained, the deeper knowledge would allow us to address the thus far little understood ecological role of pathogenic fungi in natural insect populations.

**Author Contributions:** Conceptualization, K.P., T.T., and R.G.; methodology, K.P., T.T., and H.M.; investigation, R.G. and H.M.; data curation, R.G. and H.M.; writing—R.G., T.T., and K.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Estonian Research Council grants PRG741 and IUT20-30 and the European Union through the European Regional Development Fund (the Center of Excellence EcolChange).

**Conflicts of Interest:** The authors declare no conflict of interest.
