Sequences acquired from previous molecular phylogenetic studies of Erebidae (Zahiri et al. 2011; Zahiri et al. 2012; Zahiri et al. 2013b; Scott et al. 2014; Zenker et al. 2016; Scott Chialvo et al. 2018) were downloaded from GenBank and assembled into single gene matrices. Preliminary phylogenetic analyses of the aligned sequences were conducted to determine which gene markers appeared to be most phylogenetically informative and would provide the most complete taxon sampling for our analyses. Based on the results of these preliminary analyses we chose to proceed with one mitochondrial protein-coding gene, cytochrome oxidase I (COI); one nuclear protein-coding gene, ribosomal protein S5 (RPS5); and one nuclear structural gene, the large subunit rRNA D2 loop (28S). Sequences from five species classified in the Erebidae subfamily Aganainae, and representative species of the Arctidinae tribes Arctiniini, Syntomiini, and Amerillini were downloaded from GenBank and included in the single gene matrices as outgroups. Molecular sequence data for 31 additional species, representing 16 genera from the southwestern United States were added. All voucher specimens have been deposited in the University of Arizona Insect Collection (UAIC).

Total genomic DNA was extracted from the right mesothoracic leg or the abdomen of single specimens using the Qiagen DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA) according to manufacturer suggested protocols. Total genomic DNA was stored in buffer at -80 °C.

Gene fragments were PCR amplified for COI using the primers LCO1490 and HCO2198 (Hebert et al. 2003); the nuclear protein-coding gene RPS5 and nuclear large subunit 28S were amplified using primers and PCR protocols as provided in Scott et al. (2014). PCR products were cleaned, quantified, normalized and sequenced in both directions at the University of Arizona’s Genomic and Technology Core Facility using a 3730 or 3730XL Applied Biosystems automatic sequencer. Chromatograms were assembled into contigs and initial base calls were made for each gene with Phred (Green and Ewing 2002) and Phrap (Green 1999) as orchestrated by Chromaseq ver. 1.5 in Mesquite ver. 3.6 (Maddison and Maddison 2017, 2018). Final base calls were made in Mesquite and ambiguous bases were designated by standard ambiguity codes. GenBank accession numbers for all sequences used in this study are listed in Table 1.

Single gene matrices were aligned using default settings in MAFFT v7.474 (Katoh and Standley 2013) and were concatenated in Mesquite. Maximum likelihood analyses were conducted on each gene individually and on the concatenated dataset using IQ-TREE ver. 1.6.10 (Nguyen et al. 2015), as orchestrated by Mesquite. The ModelFinder feature within IQ-TREE (Kalyaanamoorthy et al. 2017) was used to find the optimal character evolution models. The MFP model option was used for 28S, and the TESTMERGE option for the protein-coding genes. The TESTMERGE option sought the optimal partition of sites, beginning with the codon positions in different parts. Analyses of the concatenated data matrix were conducted using the TESTMERGE option, beginning with each codon position for each gene as a separate part (thus, the analysis began allowing for up to 7 parts, three for both of the protein-coding genes and one for 28S). One hundred searches were conducted for the maximum likelihood tree for each matrix. One thousand replicates were used for bootstrap analyses.

Lithosiini monophyly is supported by two morphological apomorphies found in the larvae. Both the unique arrangement of labral setae M1 and M2 and the mandibular mola were present in all Lithosiini larvae reared as part of this study, many of which were previously unknown, including Agylla septentrionalis Barnes & McDunnough, Cisthene kentuckiensis (Dyar), Gardinia anopla Hering, Crambidia myrlosea Dyar, Eudesmia arida (Skinner), Hypoprepia lampyroides Palting & Ferguson, Inopsis modulata (Edwards) and Lycomorpha fulgens (Edwards).

Acsala anomala Benjamin occurs on a long branch by itself, supporting the placement of this species in a monotypic subtribe Acsalina. This enigmatic species was placed among the Lymantriidae, however following description of the larval stages feeding on lichens and the presence of a mandibular mola (Lafontaine et al. 1982) it was considered Lithosiini. Bendib and Minet (1999) list many unique apomorphies of the Acsalina, including flightless females, translucent wing vestiture, compound eyes with interommatidial setae, and a primitive hindwing ground plan not found among other Lithosiini. From a biogeographic perspective it is interesting that the Acsalina does not seem to be recently derived from any other temperate lineage contrary to virtually all other Lepidoptera endemic to the Arctic.

When Bendib and Minet (1999) erected the tribe Cisthenini they divided it into the Cistheniti (containing Cisthene Walker, Clemensia Packard, Hypoprepia Hübner, Lycomorpha Harris, Lycomorphodes Hampson, and Rhabdatomis Dyar) and Clemensiiti (containing Clemensia Packard, Pronola Schaus, Siccia Walker, Hyposiccia Hampson and Parasiccia Hampson). They noted Cistheniti have an unusual resting posture with the antennae facing forward, while Clemensiiti exhibit the plesiomorphic folding backwards of the antennae. They also noted that Clemensiiti rested with the wings flattened rather than roof-like over their back as in Cistheniti.

We find strong support to remove Clemensia and Pronola (as discussed below) and include thirteen Neotropical genera that were unplaced by Bendib and Minet (1999). These include nine genera (Balbura Walker, Cloesia Hampson, Dipaenae Walker, Dolichesia Schaus, Ilice Walker, Metalobosia Hampson, Nodozona Druce, Ptychoglene Felder, and Talara Walker) which were found to be cisthenines in previous studies molecular-based studies (Zenker et al. 2016; Scott Chialvo et al. 2018) as well as four genera (Abrochocis Dyar, Bruceia Neumögen, Chrysochlorosia Hampson, Meterythrosia Hampson) which we include in a molecular-based study and classify for the first time (Table 2). We were not able to obtain fresh specimens of Haematomis Hampson for inclusion in our phylogeny, however we speculate based on its small size and resting posture that Haematomis belongs to Cisthenina. The wing pattern, particularly the distinctive pink basal wing markings, are consistent with other cisthenines that are thought to be Mullerian mimics of lampyrid beetles, especially some Hypoprepia species (H. lampyroides and H. inculta, for example).

Contrary to previous classifications, this study finds support to include two genera, Eudesmia and Gardinia, within Cisthenina. Bendib and Minet (1999) classified Eudesmia in subtribe Eudesmiina along with three other Western Hemisphere genera: Euryptidia Hampson, Josiodes Felder, and Paratype Felder. This is the first molecular-based phylogenetic study to include a member of the subtribe Eudesmiina. Results indicate that recognizing Eudesmiina as a valid subtribe would render Cisthenina polyphyletic. Therefore, as first revisers we place Eudesmiina in synonymy under Cisthenina, with an expanded concept of the latter. Both names, Cisthenina and Eudesmiina, were published at the same time (Bendib and Minet 1999), so neither name has priority (ICZN Article 24.2). We choose Cisthenina since it is a much more diverse lineage. Given the apomorphies presented by Bendib and Minet for uniting the four genera they placed within the Eudesmiini, we tentatively place them all within Cisthenina (Table 2).

While Gardinia was unplaced by Bendib and Minet (1999) it was treated as a member of Lithosiina by several authors (Scott and Branham 2012; Scott et al. 2014). Gardinia is a Neotropical genus containing five species, including one species from southeastern Arizona, Gardinia anopla Hering (Fig. 1E). Gardinia anopla is the largest lichen moth in North America with an average forewing length of 25 mm, making it more than twice as large as other Cisthenina (typically with forewing lengths of 10 mm or less). When captured, adults of this nocturnal species produce audible clicks. Among Cisthenina, adults of Cisthene martini are known to produce clicks in response to bat echolocation and are generally avoided by bats (Dowdy and Conner 2016). The clicks of Gardinia might be used similarly to warn bats of their distastefulness.

Cisthenina larvae generally have short, sparse setae and they lack verrucae, which was proposed as an apomorphy for the subtribe (Bendib and Minet 1999). However, the larvae of Gardinia (Fig. 5A) and of Eudesmia (Fig. 5B), reared as part of this study, possess verrucae making species in these genera the only members of Cisthenina known to have them. Eudesmia arida (Skinner) larvae possess exceedingly long, soft setae, unlike the short, stiff setae characteristic of other Cisthenina (Fig. 5B).

Figure 5. 

Representative Cisthenina larvae A Eudesmia arida (Skinner) B Gardinia anopla Hering.

All members of the Cisthenina, as defined here, are endemic to the Western Hemisphere. Among Cisthenina adults, apodemes on the second abdominal sternite are long and the anterolateral processes are fused with the rest of the sternum (Fig. 6). We find that these character states are present in Eudesmia and Gardinia (Fig. 6) which hold as a strong apomorphies of Cisthenina as redefined in this study.

Figure 6. 

Dissections of the second abdominal sternites of adults illustrating two alternate states found among subtribes of Lithosiini A apodemes of Pygarctia roseicapitis (Neumoegen & Dyar) are relatively short and the anterolateral processes articulate with the sternal plate as found in most members of Lithosiina (other than Agylla) and in Arctiini B apodemes of Gardinia anopla Hering are relatively long and the anterolateral processes are fused with the rest of the sternum as found in members of the Cisthenina, including Eudesmia and Gardinia.

Type-genus: Clemensia Packard.

When Bendib and Minet (1999) erected the tribe Cisthenini they divided it into the Cistheniti and Clemensiiti, with the latter housing Clemensia, Pronola, Siccia, Hyposiccia and Parasiccia. In addition to the differences they noted in resting posture between the subtribes, they noted three apomorphies of the Clemensiiti, including the presence of a pair of metascutal membranous areas, sternite A2 possessing curved movable anterolateral processes and the presence of a corethrogyne in females.

In this study Clemensia falls outside Cisthenina and it forms a highly supported clade with the small neotropical genus Pronola Schaus (5 species), the adults of which are similarly sized and have a similar peculiar rounded wing shape. Zenker et al. (2016) found Clemensia + Pronola were the sister group of the Oriental genus Garudinia Moore. Additional taxon sampling from around the world will be needed to determine the extent of this clade, with the genera Sicia, Hyposiccia and Parasiccia from the Western Hemisphere likely to be included within it. Future research on this clade is likely to be fruitful. Not only is a larger molecular and morphological analysis required, the limited information available on the immatures of species in this clade (McCabe 1981) suggests they are strictly algivores, refusing to feed on lichen at all. This observation, combined with their somber coloring, might indicate they do not sequester lichen phenolics for protection as do all other lithosines.

Bendib and Minet (1999) did not treat or assign genera to the Lithosiina. Results of this study indicate the following 13 genera are included in this well-supported clade: Agylla Walker, Apistosia Hübner, Areva Walker, Atolmis Hübner, Crambidia Packard, Cybosia Hübner, Eilema Hübner, Gnamtonychia Hampson, Hiera Druce, Inopsis Felder, Lithosia Fabricius, Manulea Wallengren, Mintopola Hampson.

This is the first study to include specimens of Gnamtonychia Hampson and Inopsis Felder in a molecular-based phylogenetic analysis. Including them in Lithosiina is also supported by the shape of the second abdominal sternite (Fig. 6A) and their resting posture with their wings held somewhat flattened and rolled around their abdomen, two traits typical of Lithosiina. Gnamtonychia ventralis Barnes and Lindsey occurs in southeastern Arizona and New Mexico and Inopsis modulata Edwards occurs in Mexico and is rarely found in southeastern Arizona. These two species are remarkably similar in external appearance as adults, however side-by-side I. modulata is a slightly smaller moth with shorter, more rounded wings than G. ventralis (Fig. 3B, C). Both species are evidently part of a Mullerian mimicry complex that includes the arctiine Pygotenucha terminalis (Walker) (included here as an outgroup), which is similarly colored and a toxic milkweed-feeder in the larval stages. The larvae of G. ventralis are unknown. The larvae of I. modulata have distinctive orange to red verrucae and a dark bodies (Fig. 5B), making them conspicuous as they feed on lichens growing on tree branches.

In agreement with Zenker et al. (2016), the molecular phylogeny in this study places Agylla within Lithosiina. Agylla represents the single largest radiation among Western Hemisphere Lithosiini, with 101 described species found in the Neotropics. Primarily an Old World group, Zenker et al. (2016) proposed the Lithosiina colonized South America from the Holarctic in one or more events. In fact, results presented here likely confirm that there have been at least three incursions from the Old World to the New World (assuming the group originated in Asia as proposed by Zenker et al. 2016). In our analyses we added A. septentrionalis Barnes and McDunnough (Fig. 1A), which is restricted to the mountains of southeastern Arizona. Adults of this species look similar to some European members of the genus Lithosia Fabricius such as male L. quadra Linnaeus included in our analyses. However, the results of this analysis show that these two genera are not closely related, Crambidia is instead closely related to Manulea+Eilema (Palaearctic/Oriental), whereas Agylla is part of a Neotropical clade. We note that adults of A. septentrionalis hold their wings “tent-like” over the body rather than flattened and rolled around their abdomens like most Lithosiina. In addition, the adults possess a Cisthenina-like second abdominal sternite (Fig. 6B). Thus, the placement of Agylla within Lithosiina, means that these morphological characteristics are more labile than previously thought.

With a tribe as large as Lithosiini, it is surprising that a subtribal classification was neglected for so long, yet understandable given their worldwide diversity and confounding variation of morphological characters. Beginning with Bendib and Minet (1999), we started to conceptualize how Lithosiini genera might be related. Some of our placements here, such as Gardinia among Cisthenina, show that the appearance of the adults does not belie their phylogenetic relatedness. With the apparent lack of morphological apomorphies identified thus far that support subtribal alliances, molecular techniques provide a useful tool for understanding how their diversity evolved. As additional molecular data are published and made available, their evolutionary relationships will become more apparent and hopefully lead to the discovery of morphological apomorphies in both larvae and adults. Presently the whole life history is known for only a small percentage of species. Thus, we have barely scratched the surface in understanding these remarkable lepidopterans and their unique relationship to their lichen hosts and to each other.