Associations between fungi and insects are manifold, ranging from pathogenic (Hajek and Leger, 1994) to mutualistic (Biedermann and Vega, 2020). The most well-known mutualistic association in ants is the nutritional relationship between fungus growing ants (tribe Attini, subtribe Attina) and basidiomycete fungi (Agaricales: Agaricaceae and Pterulaceae) which are usually cultivated as mycelia and used as the primary food source for the ant colony (Weber, 1972; Schultz and Brady, 2008). However, there are other less well-known mutualistic interactions between ants and ascomycete fungi identified as Chaetothyriales, with Capnodiales also occurring frequently (Defossez et al., 2009; Mayer and Voglmayr, 2009; Ruiz-Gonzalez et al., 2011; Voglmayr et al., 2011; Nepel et al., 2014; Kokolo et al., 2016; Vasse et al., 2017; Gegenbauer et al., 2023). Both Chaetothyriales and Capnodiales belong to the so-called “Black Fungi”, a diverse group of slow-growing ascomycetes with melanized hyphae. Black Fungi have been found in mutualistic association with Old World Lasius ants on the walls of nests inside dead trees (Lagerheim, 1900; Maschwitz and Hölldobler, 1970; Schlick-Steiner et al., 2008), with mound-building Formica ants (Lindström et al., 2018; Lindström et al., 2021), and with various arboreal ants in the tropics. The latter association between arboreal ants and fungi is the focus of the present review because this type of association has received much attention within the past 15 years.

Arboreal ants (= species that forage and nest mostly or entirely on plants) make up about a third of the total ant fauna in tropical rainforests (Longino and Colwell, 2020). A time-scaled ant phylogeny suggests that arboreal foraging evolved in the early Cretaceous followed by arboreal nesting in the late Cretaceous (Nelsen et al., 2018). Under conditions of limited nesting space availability in arboreal ant communities in tropical forests (Wilson, 1987; Fiala and Maschwitz, 1992; Fonseca, 1993; Camarota et al., 2020) that accompanies the high diversity of ants (Kass et al., 2022; Schultheiss et al., 2022) niche differentiation by development of different nesting strategies may be a crucial factor for arboreal ant communities to allow a broader use of limited resources.

Three categories of arboreal ants differing in their nesting strategy can be defined: (1) a group of opportunists that use a range of cavities and crevices in plants as incidental nest sites to shelter the colonies (Hölldobler and Wilson, 1990; Novais et al., 2017; Philpott et al., 2018), many of them making constructions with fibers, resin, soil, sand, and organic matter to reduce entrance sizes and improve the cavity defense (Priest et al., 2021). (2) a group of carton-builders that make de novo constructions of carton nests and/or galleries independent from existing cavities; construction materials are plant derived bark and wood fragments, trichomes, mosses and epiphylls, and hyphae of ascomycete so-called “Black Fungi” (Weißflog, 2001; Mayer and Voglmayr, 2009; Ruiz-Gonzalez et al., 2011). Ant-constructed carton with fungi can constitute walls of free-hanging nests on the lower surface of leaves as in Crematogaster and Technomyrmex (Weißflog, 2001; Voglmayr et al., 2011) (see Figure 19 in Weißflog, 2001) ( Figure 1H ), runway galleries along stems, branches and leaf surfaces as found on Tetrathylacium macrophyllum built by Azteca brevis ants (Mayer and Voglmayr, 2009; Mayer et al., 2017) ( Figures 1I, O ), or cardboard-like “carton” structures sheathing entrances to their nest side, protecting flocks of scale insects, or partitioning the nest interior ( Figure 1A ). Interwoven plant hairs manured with the excrements of the ants like in Monomorium sp. (Weißflog, 2001) ( Figure 1F ), or in the Hirtella physophora - Allomerus decemarticulatus association (Dejean et al., 2005) ( Figure 1M ) are also used. (3) a category of plant-ants that live inside “myrmecophytes”, plants with organs that have been transformed into hollow cavities called “domatia” which can be derived from stems ( Figures 2A, C-E, K, L, N ), leaf pouches, leaf petioles, stipules, root tubers, rhizomes, or hypocotyls (Chomicki and Renner, 2015), being induced by the plants themselves without any induction from plant-ants (but see Blüthgen and Wesenberg, 2001; Gaume et al., 2005). In addition to shelter, some myrmecophytes offer an immediate source of nutrients for the inhabiting ant colony through carbohydrate-rich extrafloral nectaries (Heil et al., 2000; Blüthgen et al., 2004; Gonzalez-Teuber and Heil, 2009) and, in some cases, protein- and lipid-rich food bodies (Folgarait and Davidson, 1994, Folgarait and Davidson, 1995; Heil et al., 1998; Fischer et al., 2002; Hatada et al., 2002; Heil et al., 2004; Webber et al., 2007; Bischof et al., 2013). Myrmecophytes are found almost exclusively in the tropical regions of Australasia, Africa and the New World (Chomicki and Renner, 2015), where the pressure of pathogens and herbivores is high, and nutrients are limited (Davidson, 1997; Kaspari and Yanoviak, 2001). Like in free nest structures, Black Fungi are also present in domatia ( Figure 2 ). This has been documented for a long time (Miehe, 1911; Bailey, 1920), but due to the unspectacular appearance, difficulties in cultivation and taxonomic assignment, and lack of conclusive evidence of mutualism between fungi and the myrmecophytic partners, fungi in ant-plant associations did not receive much attention. Only recently, the development of molecular techniques has allowed the taxonomic assignment of fungi found in many unrelated ant-plant associations in the tropics worldwide ( Figure 3 ).

Black Fungi associated with ants can be subdivided into two main functional groups that differ substantially: group 1 are those that occur on ant-constructed free-hanging walls of nests and galleries built by the ants, whilst group 2 are those that occur inside the domatia of myrmecophytes ( Figure 4 ).

Character state analyses based on phylogenies indicate that the ancestral Chaetothyriales were slow-growing colonizers of rock surfaces characterized by extreme oligotrophic conditions, and in these ancestral ecological niches, they had to cope with co-occurring toxin-producing lichens and cyanobacteria (Gueidan et al., 2008; Quan et al., 2020). Therefore, extremotolerance (temperature, drought, poor nutrient availability) and toxin resistance are crucial ancestral evolutionary adaptations that subsequently opened a window of opportunity to colonize a surprisingly diverse range of extreme habitats that are unable to support the growth of most other potential competitors (Quan et al., 2020). As a result, Chaetothyriales may be predisposed to interact with ants because of their tolerance for various toxic chemical compounds (Teixeira et al., 2017) that are produced by ants for communication (Jackson and Morgan, 1993) and to control diseases (Yek and Mueller, 2011). The production of antibiotic and antimycotic substances (e.g. formic and acetic acid, Tragust et al., 2013) are vital for social insects to ensure health and survival of the colony (Penick et al., 2018), which in turn requires that associated fungi are resistant to these substances as a pre-requisite to inhabit this niche and to establish mutualisms. It therefore makes sense that Chaetothyriales might produce bioactive substances against competing fungi (El-Elimat et al., 2013). In addition, Chaetothyriales appear to be pre-adapted to manipulation by specialized ants, as they thrive under constant and frequent mechanical disturbance due to their pronounced ability of regeneration. The combination of these factors may account for the close relationship between ants and those fungi.

Much less is known about the evolutionary and genomic background of ant-associated Capnodiales. However, they seem to occupy somewhat different niches than the Chaetothyriales, as in different ant-plant fungal systems either the Chaetothyriales or the Capnodiales are dominant. The Capnodiales sensu lato represent a morphologically and ecologically highly diverse and speciose lineage with different lifestyles and modes of nutrition, viz. saprobes, plant and human pathogens, mycoparasites, rock-inhabiting fungi, lichenized, epi-, ecto- and endophytes (Abdollahzadeh et al., 2020). Several families of the former Capnodiales sensu lato were recently elevated to distinct orders (Abdollahzadeh et al., 2020). As a result, Capnodiales sensu stricto were restricted to the sooty molds which colonize plant surfaces covered by sugar-rich honeydew excreted by sap-sucking hemiptera (Abdollahzadeh et al., 2020). Voglmayr et al. (2011) observed that Capnodiales were prominent particularly in “carton” of Crematogaster ants, which differed markedly from other ant nests by extremely hard, dense and tough texture, with the fungi only growing on the carton surface. Considering the close association of ants with hemiptera which are tended by the ants in return for carbohydrates, honeydew-inhabiting Capnodiales may be predestined to colonize suitable surfaces within ant nests, if carbohydrates are accidentally or intentionally deposited on them. However, additional investigations are needed to ascertain the nutritional requirements of these Capnodiales isolates.

As phylogenetic analyses of ant-associated Capnodiales are not available to date, only Chaetothyriales are considered here. Interestingly, the phylogenetic analyses suggest that the strains of the two functional groups exhibit distinct evolutionary patterns. Within Chaetothyriales, species associated with nest-walls and galleries (functional group 1) appear more scattered in the phylogeny compared to species associated with ants living in domatia (functional group 2) (Voglmayr et al., 2011; Nepel et al., 2014; Vasse et al., 2017; Quan et al., 2020). Chaetothyriales of the first group belong to Trichomeriaceae, Herpotrichiellaceae and Cyphellophoraceae and are particularly well represented in the families Trichomeriaceae and Herpotrichiellaceae. This pattern suggests that life as mutualists of ant-made carton evolved multiple times independently, even across different orders. It is not clear whether these species are ant specialists, or if they are opportunistic fungi that primarily live in other environments but are specifically recruited by the ants, or if at least some of them are commensals that arrived without any relation to ants.

In contrast, most domatia-associated Chaetothyriales cluster close to Cyphellophoraceae and Paracladophialophoraceae (Quan et al., 2023) in a monophyletic clade that is composed of domatia-associated species only (“domatia symbiont clade”) and that could potentially constitute a distinct family (Voglmayr et al., 2011; Vasse et al., 2017; Quan et al., 2020). The substantial genome (and gene) reduction reported for species in the domatia symbiont clade (Moreno et al., 2019) suggests a long evolutionary specialization. However, there are also several species of domatia-associated Chaetothyriales that do not belong to the monophyletic main clade but are scattered throughout the phylogeny (Vasse et al., 2017). This may indicate that the ant-associated lifestyle evolved several times. Alternatively, the association with mutualistic ant-plant interactions may be an optional, opportunistic strategy. One example of the latter is Exophiala oligosperma, which is known to be a human pathogen found in low-nutrient substrates (De Hoog et al., 2003), and has recently been discovered in ant-occupied domatia of various ant-plants in South-East Asia, and Central and South America (Vasse et al., 2017; Blatrix et al., 2021). In addition, several ant-plant mutualisms on different continents have been found to share a single species of Chaetothyriales, showing the ability of such species to change hosts and continents (Nepel et al., 2016; Vasse et al., 2017; Blatrix et al., 2021; Greenfield et al., 2021). In summary, the evolutionary pattern of domatia-associated Chaetothyriales seems to involve one main radiation possibly linked to specialization on the one hand, and opportunistic nonspecific associations on the other hand. The opportunistic strains isolated from the ant nest environment may not all be involved in interaction with ants.

The phylogenetic patterns of ant-plants and plant-ants are markedly different, as myrmecophytism (i.e. bearing domatia) in plants is distributed throughout the plant phylogeny, with at least 158 independent evolutionary origins, each occurring less than 20 Mya ago (Chomicki and Renner, 2015). In ants the obligate association with myrmecophytes has evolved many times independently, but with three times fewer origins than in the ant-plants (Chomicki and Renner, 2015). Co-diversification between plants and ants seems not to be a driver in the evolution of ant-plant mutualisms. According to the latest phylogenetic reconstruction, the monophyletic clade containing most of the domatia-inhabiting Chaetothyriales evolved at least 27 Mya ago (Quan et al., 2020), that is, earlier than any of the extant lineages of myrmecophytes (Chomicki and Renner, 2015). Two non-exclusive hypotheses, outlined in the following two paragraphs, are proposed to explain how the contemporary associations might have developed despite the distinct evolutionary history of domatia fungi and myrmecophytic plant lineages.

The first hypothesis suggests, that myrmecophytism may have existed much earlier than the recent plant lineages. The development of a domatium is most likely an evolutionarily labile trait that is easily acquired and lost for the following reasons: (i) it has been independently acquired at least 158 times and lost at least 43 times (Chomicki and Renner, 2015), (ii) there are no known plant fossils with domatia, and (iii) in some species, this trait is highly variable (Michelangeli, 2005; Shenoy and Borges, 2010;, Kokolo et al., 2020). Consequently, although the extant lineages of myrmecophytes are less than 20 Mya old, myrmecophytism may have existed much earlier but without leaving any detectable evidence. If a more than 27 Mya old specialized clade of domatia-associated Chaetothyriales exists, as suggested by the most recent dated phylogeny (Quan et al., 2020), it could have originated from myrmecophytic plant lineages that are now extinct. The adaptation to ant-inhabited domatia as habitat could have been maintained through host-switching or opportunistic life style outside domatia, both of which are strategies that do not seem to be rare in extant species (see previous paragraphs).

The second hypothesis suggests, that fungi associated with ants may have originated from epiphyllous species. In the current state of knowledge, most ant-associated melanized fungi in the tropics interact with arboreal ants. It is noteworthy that the clade of domatia-associated Chaetothyriales is sister to a pair of species known to be epiphyllous (i.e. living on the surface of tree leaves). In addition, the closest relatives of these clades are the families Cyphellophoraceae and Phaeosaccardinulaceae, mostly composed of plant-associated species (Quan et al., 2020). Domatia species also occur in Cyphellophoraceae, but also in Herpotrichiellaceae and Trichomeriaceae which have many epiphyllous ones (Quan et al., 2020; Quan et al., 2023). The scraping of leaf surfaces by arboreal ants — most probably to collect epiphylls for larval feeding, as observed today (Davidson et al., 2003; Davidson et al., 2016) — may have provided many opportunities over time for epiphyllous Chaetothyriales and Capnodiales to adapt and colonize in the environment of newly emerging domatia, where their tolerance to different chemical compounds would give them an advantage over competitors. However, Baker et al. (2017) found a marked difference between the fungal community in the domatia of African Vachellia drepanolobium and the one on the surface of leaves of the same plants. A comprehensive study of the epiphyllous melanized fungi would probably substantially improve our understanding of the evolution of the ant-plant associated species.

Similarly, most carton-associated Chaetothyriales from the family Trichomeriaceae are embedded in a clade of epiphyllous species, and Capnodiales are known to be epiphyllous and colonizers of honeydew on leaf surfaces (Abdollahzadeh et al., 2020).

Are the fungi more dependent on ants or plants species, or both? Globally, the presence of fungi in ant-plant interactions appears to depend more on the ants than on the plants. In Azteca – Cecropia associations, the identity of the associated fungi was not dependent on the host plant species but on the associated ant species (Nepel et al., 2016). Similarly, when Cordia nodosa is inhabited by Azteca sp. cf. depilis, a mutualist that does not build galleries on the plant like A. octoarticulatus does, no Chaetothyriales have been found (Ruiz-González et al., 2019). Instead, two fungal strains, from Pleosporales and from Trichosporales, have been identified in the associations involving Azteca sp. cf. depilis but not if inhabited by Allomerus spp. (Ruiz-González et al., 2019). Also in Barteria dewevrei we find this phenomenon: Chaetothyriales could only be isolated when inhabited by the mutualists Tetraponera aethiops or T. latifrons, but not if colonized by Crematogaster sp., a parasite (Kokolo et al., 2016). The host plant provides the environment, but the presence of Chaetothyriales as mutualistic partners seems to depend on the ant species.

The progress of sequencing technologies opened the door to a world of less visible associates of ants. It allowed us to realize how frequent and widespread ecological relationships between Black Fungi and many different ant species are. Ant nests are important, previously overlooked habitats for Chaetothyriales and Capnodiales, the vast majority of which are still undescribed or only known as sequence.

In Chaetothyriales, the phylogenetic pattern (one monophyletic and well separated clade) and functional considerations (possible morphological adaptation and reduced gene functions) suggest that the domatia-associated species have coevolved with ants over a long period of time and have developed specific adaptations to this lifestyle, possibly representing an adaptive radiation. Further research, including a formal quantification of this trait and its mapping onto the phylogeny, would be a valuable aid for understanding the evolution of the ant-fungus mutualism in relation to the lifestyle of both partners.

There are, however, many open questions. Not all Chaetothyriales and Capnodiales detected may be mutualists, but to determine this requires a better understanding of the functional role of the associated fungi. Understanding the functional role also helps in understanding the evolutionary drivers which lead to a beneficial mutualism between ants and Chaetothyriales. Further, we assume that ant-associated Chaetothyriales growing in domatia have a long common history, but the origin of this fungi - ant nest mutualism is unclear. It would not be surprising to find that the ants have recruited the fungi from epiphyllous communities in their foraging environment, or that the fungi migrated into ant nests as a suitable niche and recruited the ants as their dispersal vector. A comprehensive study of the epiphyllous Chaetothyriales and Capnodiales as well as a higher number of genome analyses in ant-associated fungal species, would substantially improve our understanding of the evolution and functional role of the fungal species associated with ant and plant interactions.

Another still unexplored aspect is the defensive benefit of domatia and carton fungi for ants. While the antimicrobial activities of substances from bacterial symbionts in ants have been investigated (Seipke et al., 2012; Seipke et al., 2013; Hanshew et al., 2015; Fukuda et al., 2021), the role of fungi for protection against pathogens or otherwise detrimental fungi is yet to be considered. There are remarkable examples in other insects where the occurrence of Penicillium protects the larvae (Wang et al., 2015b). Though gene clusters in domatia-associated fungi point to a protective role, it is unknown whether and which substances are released and what their effect might be.

VM wrote chapters 1, 2, 6 and 8. HV chapter 3. RB chapter 4. CL & JO chapters 5 and 7. All authors contributed to the article and approved the submitted version.