Phylogenetic relationships of the woodlouse flies (Diptera: Rhinophorinae) and the cluster flies (Diptera: Polleniidae)

Phylogenetic relationships within the oestroid subclades Rhinophorinae (Calliphoridae) and Polleniidae were reconstructed for the first time, applying a Sanger sequencing approach using the two protein-coding nuclear markers CAD (carbamoyl-phosphate synthetase 2, aspartate transcarbamylase, and dihydroorotase; 1794 bp) and MCS (molybdenum cofactor sulfurase; 2078 bp). Three genera of Polleniidae and nineteen genera of Rhinophorinae were analyzed together with a selection of taxa representing the major lineages of Oestroidea (non-rhinophorine Calliphoridae, Oestridae, Sarcophagidae, Tachinidae). The selected markers provide good resolution and moderate to strong support of the distal branches, but weak support for several deeper nodes. Polleniidae (cluster flies) emerge as monophyletic and their sister-group relationship to Tachinidae is confirmed. Morinia Robineau-Desvoidy as currently circumscribed emerges as paraphyletic with regard to Melanodexia Williston, and Pollenia Robineau-Desvoidy is the sister taxon of the Morinia–Melanodexia clade. We propose a classification with two subfamilies, Moriniinae Townsend (including Morinia, Melanodexia, and Alvamaja Rognes), and Polleniinae Brauer & Bergenstamm (including Pollenia, Dexopollenia Townsend, and Xanthotryxus Aldrich). Anthracomyza Malloch and Nesodexia Villeneuve are considered as Oestroidea incertae sedis pending further study. Rhinophorinae (woodlouse flies) emerge as monophyletic and sister to a clade composed of (Ameniinae + (Ameniinae + Phumosiinae)), and a tribal classification is proposed with the subfamily divided into Rhinophorini Robineau-Desvoidy, 1863 and Phytonini Robineau-Desvoidy, 1863 (the Stevenia-group and the Phyto-group of authors, respectively). Oxytachina Brauer & Bergenstamm, 1891, stat. rev. is resurrected to contain nine Afrotropical rhinophorine species currently assigned to genus Rhinomorinia Brauer & Bergenstamm, 1891: Oxytachina approximata (Crosskey, 1977) comb. nov., O. atra (Bischof, 1904) comb. nov., O. bisetosa (Crosskey, 1977) comb. nov., O. capensis (Brauer & Bergenstamm, 1893) comb. nov., O. scutellata (Crosskey, 1977) comb. nov., O. setitibia (Crosskey, 1977) comb. nov., O. verticalis (Crosskey, 1977) comb. nov., O. vittata Brauer & Bergenstamm, 1891, and O. xanthocephala (Bezzi, 1908) comb. nov.


Introduction
Oestroidea comprise a diverse clade of true flies comprising some of the most familiar insects, such as blow flies and flesh flies.The group accounts for about 15,000 known species [1], but estimates suggest that the true number may be at least twice as many [2,3].As holometabolous insects, their larval stage is morphologically and functionally entirely different from the adult stage, and whereas adults are often flower visitors, oestroid larvae can be general scavengers; vertebrate or invertebrate necrophages; vertebrate coprophages; vertebrate parasites; invertebrate parasitoids; predators of frog spawn, molluscs, earthworms, termites, grasshopper eggs or spider eggs; and even mycophages and palynophages [2,4,5].
Polleniidae (Fig 1)-also known as cluster flies due to the tendency of adults of some species to cluster indoors for overwintering-is a family of earthworm parasitoids [19][20][21][22][23].The  (Shannon, 1926).C Melanodexia tristina (Hall, 1948) polleniids have been treated either as a subfamily (Polleniinae) or tribe (Polleniini) within the Calliphoridae [24][25][26][27] or given family rank [28].Polleniids have recently become well-established as the extant sister taxon of the megadiverse parasitoid clade Tachinidae [11,12,18,[29][30][31][32][33], even though morphology or other character systems so far have provided few clues in support of this relationship.Studies of the phylogenetic relationships among polleniid genera are limited and include only sparse taxon sampling [28,34,35].Currently the family contains some 150 named species in eight genera, with the bulk of diversity in the Palaearctic Region [36].Polleniids are also widespread and abundant in the Oriental and Australasian regions, but native species are confined to smaller areas in the Nearctic (West Coast of the USA) and the Afrotropics (southern Africa), and are entirely absent from the Neotropics [36,37].A few species have become widely distributed, possibly due to individuals diapausing in shipping containers and with the widespread establishment of introduced populations of their host earthworms [38].
The present paper provides a comprehensive phylogeny of the Rhinophorinae and the Polleniidae, involving an extensive taxon sampling and employing two nuclear protein-coding genes, CAD and MCS, previously evaluated as having high phylogenetic informativeness [31,72].
CAD (carbamoyl-phosphate synthetase 2, aspartate transcarbamylase, and dihydroorotase) and MCS (molybdenum cofactor sulfurase) were chosen for their phylogenetic information and reliability for Mesozoic-Cenozoic-aged explosively radiated groups such as the Oestroidea [72], as well as for ease of comparisons with results and integration of sequences from analyses of other available datasets.
Three legs were removed from ethanol-preserved specimens and stored in ethanol until extraction.Extractions were performed using the DNeasy Blood and Tissue Kit (Quiagen, Venlo, the Netherlands) with the following modifications of the manufacturer's protocol: legs were placed entire in the digestion buffer and Buffer ATL was replaced with a digestion buffer as described by Gilbert et al. [73] but modified to consist of 10 mM Tris-HCl (pH 8), 10 mM NaCl, 5 mM CaCl 2 , 2.5 mM EDTA, 1% sodium dodecyl sulphate (SDS), 250 μg/mL proteinase K, and 40 mM dithiotreitol (DTT) (final concentrations).PCR amplification reactions (total volume 25 μL) were composed of 18 μL deionized water, 4 μL of 5X HOT FIREPol Blend Master Mix (Solis BioDyne), 0.5 μL of each primer (final concentration of 0.2 μM), and 2 μL of DNA solution.The most effective program among all the experimental variations was a PCR protocol consisting of an initial denaturation stage of 12 min at 95˚C; 35 cycles of 95˚C for 30 sec, variable annealing temperature (depending on the primers used, see Table 1) for 1 min, 72˚C for 2 min, and a final extension time of 10 min at 72˚C.After visualization on a 2% Agarose gel, PCR products were sent to Macrogen Europe (Amsterdam, the Netherlands) for PCR product cleanup and sequencing.
Sequencing output files were assembled and trimmed using Geneious 9.1.8(Biomatters Ltd., Auckland, New Zealand).FASTA files of sequences were aligned using MAFFT (v.7.017) with the G-INS-i algorithm using the default parameters [74,75] (see S1 Dataset).The resulting alignments were checked for accuracy by looking for stop codons and spurious gaps once the alignments were translated into proteins.The single-gene alignments were then concatenated using the "Concatenate alignments" tool in Geneious (see S1 Dataset).
PartitionFinder v2 [76,77] was used to find the best-fitting partitioning scheme and to select substitution models for each partition without overparameterization, evaluated by the information-theoretic metric BIC (Bayesian Information Criterion).The initial 6 data blocks were the first, second, and third codon positions of each of CAD and MCS, and the program was set to perform a greedy search to compare all possible partitioning schemes.The best-fit scheme grouped all data blocks in one single partition.The model GTR+I+G was selected as the substitution model for this partition.
Likelihood analyses were conducted using RAxML version 8.2.12 [78] on XSEDE (Extreme Science and Engineering Discovery Environment) through the CIPRES (Cyberinfrastructure for Phylogenetic Research) Science Gateway [79].The tree of highest likelihood from 100 replicate runs was selected for plotting the bootstrap values from 250 ML rapid bootstrap replicates obtained through a GTR+G+I approximation.Trees from all analyses were visualized using FigTree [80].

Results
Analysis of the concatenated matrix from CAD (1794 bp) and MCS (2078 bp) resulted in a well-resolved ML topology, although with some branches having low bootstrap support values (henceforth b.v.) (Fig 3).
The ML tree (Fig 3 ) shows overall strong support for shallow nodes within families, but somewhat lower support for many of the deeper branches representing relationships among families and other major clades.Relationships among the genera within the target cluster flies and woodlouse flies are generally robust and well-resolved.DNA sequence data for

Discussion
A fully resolved and well supported phylogeny of oestroid flies has proved difficult to attain through both morphological and Sanger-generated molecular data.However, a consensus on the topology of the backbone is now emerging through phylogenomic and phylotranscriptomic approaches [11][12][13].Conflicts in the deeper splits, i.e., in the position of families and subfamilies between the present study and the more recent phylogenomic studies are here considered as most likely resulting from our use of data from only two nuclear loci.Many deep nodes received low statistical support values, and they are not discussed further.
Despite recent study, the phylogenetic relationships among polleniid genera are still tentative.Employing a combination of morphological characters and fragments of three nuclear markers (CAD, MCS, MAC) on a selection of Morinia species, one Pollenia and the monotypic Alvamaja Rognes, Cerretti et al. [28], reconstructed Morinia as monophyletic and sister to Alvamaja, this clade being sister to Pollenia.Recently, Johnston et al. [35] presented a mitogenomic analysis of 21 polleniid taxa, including a broad representation of West Palaearctic Pollenia and one species each of Melanodexia, Morinia and Dexopollenia Townsend.The study retrieved Dexopollenia as sister to a clade composed of Morinia and Melanodexia, with this clade in turn sister to Pollenia.Johnston et al. [35] performed further analyses by using COI sequences (i.e., not the entire mitogenome) of Xanthotryxus mongol Aldrich and an additional species of Morinia, and recovered Dexopollenia + Xanthotryxus Aldrich as sister to the remaining Polleniidae, and the latter resolving as Pollenia being sister to Morinia + Melanodexia.These relationships come with low support, which is partly obscured by Johnston et al. [35] incorrectly using support values for particular nodes to indicate support for the basal dichotomy.However, the present ML topology is largely consistent with the results in Johnston et al. [35] (except for retrieving Pollenia as paraphyletic), but the limited taxon sampling does not allow for testing the phylogenetic position of Dexopollenia and Xanthotryxus.The sparse morphological evidence tends to support a Dexopollenia-Xanthotryxus-Pollenia-clade.All species of Xanthotryxus and most species of Dexopollenia and Pollenia share the presence of golden, wavy, hair-like setae on parts of the body, and the morphologically very similar Pollenia and Xanthotryxus also share a subcostal sclerite with a bundle of long, black or yellow setae among the micropubescence.However, comparative morphology of polleniids needs much more study, and differing phylogenetic topologies obscure interpretations of character state polarities.For instance, the Australian Pollenia nr.stolida examined here differs from the Palaearctic and New Zealand species of Pollenia by lacking the first presutural intra-alar seta and by the three preapical setae of the hind tibia (anterodorsal, dorsal and posterodorsal) being subequal in size; both these character states are shared with Morinia and Melanodexia and may represent plesiomorphic conditions.Interestingly, we found this species taking up the position of sister taxon to all other polleniids included in the analyses, although with weak support (Fig 3).
The Morinia-Melanodexia clade is supported by the following putative morphological autapomorphies: i) narrow, tongue-shaped, lower calypter, ii) posterior spiracle with reduced posterior lappet (rhinophorine-like) and iii) node at base of R 4+5 bare.Morinia is here represented by the type species M. doronici (Scopoli) (Palaearctic) and by an undescribed species from South Africa [19,29].Our phylogeny reconstructed the two included species of Morinia as paraphyletic with respect to Melanodexia.Indeed, our careful examination of Morinia from both Palaearctic and Afrotropical regions has not revealed any strong evidence supporting their monophyly, except for sharing a slim, narrow, body shape, which contrasts with the stouter body characterizing the other polleniids, except Alvamaja.Despite this, we consider it premature to lump species currently assigned to Morinia and Melanodexia under the same genusgroup name as long as there is inconclusive data on the phylogenetic position of Alvamaja, Anthracomyza Malloch, Dexopollenia Townsend, Nesodexia Villeneuve and Xanthotryxus, all of which may belong in the Polleniidae [36].Alvamaja presents a unique combination of character states and could belong to the Morinia-Melanodexia clade based on its rhinophorinelike (i.e., non-operculate) posterior spiracle and narrow lower calypter.This relationship is supported by the morphological evidence presented in Cerretti et al. [28].Dexopollenia and Xanthotryxus share several, derived character states with Pollenia, including the golden, wavy hair-like setae particularly abundant on the thorax, and the cluster of long black or yellow setae on the subcostal sclerite.No progress has been made so far in resolving the phylogenetic placement of Anthracomyza and Nesodexia.These are both monotypic genera and no molecular sequence data have been obtained from them.Although several polleniid nominal genusgroup taxa remain to be included in a phylogenetic analysis, we are here proposing to apply a subfamily classification with the Polleniidae composed of two subfamilies, as follows: Moriniinae Townsend: including Morinia, Melanodexia, and Alvamaja; Polleniinae Brauer & Bergenstamm: including Pollenia, Dexopollenia, and Xanthotryxus.
We here treat Anthracomyza and Nesodexia as Oestroidea incertae sedis pending further study.
Our study is the first attempt at resolving the phylogenetic relationships within the woodlouse flies using molecular data.Analyses support both the monophyly of the subfamily and its division into two subclades, which we propose here as tribes: Rhinophorini and Phytonini (i.e., the Stevenia group and the Phyto group of Pape & Arnaud [81], respectively).By integrating our results (Fig 3) with those from previous phylogenetic reconstructions, deduced from morphological data and a larger taxon sampling [61,  Within Rhinophorini, all the included non-monotypic genera emerged as monophyletic except for "Rhinomorinia".This nominal genus was retrieved as polyphyletic, being divided into two well-supported lineages, which are characterized by distinctive morphological features [60-62; see also the key below].One clade comprises exclusively Afrotropical species (here placed in the resurrected nominal genus Oxytachina stat.rev., see below) and was retrieved as sister to the remaining Rhinophorini with moderate support.The other clade comprises two Palaearctic/western Oriental species, of which Rhinomorinia sarcophagina (Schiner) (type species of the genus) was included and clustered within the Rhinophorini as sister to Oplisa with strong support.Under ML, Rhinophora lepida (Meigen) was recovered as sister to all Rhinophoriini except Oxytachina, differing from the morphology-based phylogeny of Cerretti et al. [61] that retrieved Rhinophora joining a primarily Palaearctic subclade composed of Rhinomorinia, Macrotarsina, Oplisa and Stevenia, which also contained some Afrotropical and Oriental species for which molecular data are not available.The position of the Afrotropical genus Ventrops within an otherwise Neotropical clade is biogeographically challenging, but the low support indicates that this hypothesis needs further testing.Support values for the genus-level reconstruction within this clade were weak to moderate.Interestingly, the morphology-based phylogeny of Cerretti et al. [61] had Ventrops as the sister taxon of a clade containing all the Neotropical taxa and the Australian genus Bixinia.Analyses of morphological data separated the genus Bixinia widely from the other Australasian taxa [61], while the present molecular analysis of Axinia and Bixinia places these genera within the same tribe but separated by multiple intervening genera (Fig 3).

Conclusions
Until recently, the taxonomic boundaries and phylogenetic affinities of Polleniidae and Rhinophorinae-two key groups of parasitoids of soil-dwelling organisms-remained controversial.Our analysis, despite being limited to two protein-coding nuclear genes, confirmed previous hypotheses on the relationships between the two groups and provided new insights into their internal phylogenetic relationships.These results allowed us to formally propose a subfamilial and tribal classification for the polleniids and rhinophorines, respectively, and to resurrect the genus Oxytachina to include five Afrotropical species previously assigned to the genus Rhinomorinia, which thereby is restricted to two Palaearctic species.