Beaded lacewings – a pictorial identification key to the genera, their biogeographics and a phylogentic analysis (Insecta: Neuroptera: Berothidae)

The present paper comprises the first illustrated key for the genera of the Berothidae. Distribution maps for all genera are provided and distribution areas are discussed. A phylogenetic analysis based on the matrix of morphological characters of Aspöck and Nemeschkal (1998) using additional taxa and a modified approach concerning the homology of the genital sclerites (Aspöck and Aspöck 2008) yielded the following conflicting results: (i) The Trichomatinae are nested within the Nosybinae – which is considered an implausible position. (ii) Within the Berothinae, the genus Podallea appeared as the sister group of Asadeteva + Nodalla, whereas previously it was the sister group of the Isoscelipteron clade. In addition, the phylogenetic relevance of a peculiar phenomenon concerning the sclerites of the seventh abdominal segment of the females is discussed.


Introduction
The Berothidae Handlirsch, 1908, are a small neuropteran family that comprises about 110 known species in 24 genera. The adults superficially resemble hemerobiids ( Fig. 1) but can clearly be distinguished from them by a pair of large haired tubercles on the vertex and the long "running" CuA of the hindwing (Tjeder 1959). The term beaded refers to the encrusted appearance of the setae on the wings of females that is caused by a secretion (Penny et al. 1997).
The fossil record of Berothidae dates back as far as the Middle Jurassic, but the family is assumed to be much older (Makarkin et al. 2011). Most of the fossil material was described from the Cretaceous e.g. in Lebanese amber (Whalley 1980), Burmese amber (Engel and Grimaldi 2008), New Jersey amber (Grimaldi 2000) and Canadian failed to corroborate recognization of the subfamilies Nyrminae and Cyrenoberothinae, which were however retained as a working hypothesis.
The Rhachiberothinae were primarily described as a subfamily of the Berothidae (Tjeder 1959(Tjeder , 1968, later interpreted to be a subfamily of the Mantispidae (Willmann 1990), and finally elevated to family rank as sister group of the Berothidae within the clade Dilaridae + (Mantispidae + (Berothidae + Rhachiberothidae)) (Aspöck and Mansell 1994). Although this arrangement could not be re-established in a molecular analysis of the Neuropterida (Haring and Aspöck 2004), the sister group relationship Berothidae + Rhachiberothidae was corroborated in the phylogenetic analyses of morphological characters by Aspöck et al. (2001), Beutel et al. (2010), Zimmermann et al. (2011), Randolf et al. (2013Randolf et al. ( , 2014 and in the treatment of genital sclerites by . The present approach is based on the matrix of Aspöck and Nemeschkal (1998), amended by a new homologization of male and female genital sclerites  and the data for three additional genera: Ormiscocerus Blanchard, 1851, which was rediscovered by Penny and Winterton (2007), Berothimerobius, which was newly described by Monserrat and Deretsky (1999) and Tanzanberotha U. Aspöck & Hynd, 1995. Tanzanberotha was not included in the analysis of Aspöck and Nemeschkal (1998) because females were unknown, but it was suggested to be part of Nosybinae (Aspöck and Nemeschkal 1998). Females are still unknown, but the genus was included in the present analysis to test if male characters corroborate Tanzanberotha as part of the Nosybinae.
Berothidae occur throughout most biogeographical regions. Cyrenoberothinae are known from southern South America and southern Africa, Nosybinae from the Afrotropical region and South America, Berothinae from most parts of the world with a high diversity in Australia and Africa (Grimaldi and Engel 2005). Nyrminae are known from Anatolia (Aspöck 1989) and Chile Winterton 2007, Monserrat andDeretsky 1999). In the present paper the Berothimerobiinae with the single genus Berothimerobius are interpreted to be part of the Nyrminae [nov. syn.!]. The Protobiellinae are restricted to New Zealand and Australia, and the Trichomatinae to Australia Nemeschkal 1998, Grimaldi andEngel 2005).
Until now the only identification keys available are for the species of a single genus (e.g. Ardila-Camacho 2013, Winterton 2010) or for the genera or species of a single geographic region (e.g. Aspöck et al. 2013). The present work gives an identification key for all known genera of the Berothidae for the first time. In addition, distribution maps of the genera based on data from literature are provided.  already published figures served as template (Tab. 1, Supplementary file 1) except Figures 19, 22 and 23, which were drawn from photographs of the material in the collection of the Natural History Museum Vienna. The identifying characteristics used in the key are highlighted in grey and/or marked with an arrow.

Distribution maps
Localities were taken from original literature (Supplementary file 2) and listed continuously with a number for each locality in MS Excel 2010 (Supplementary file 3). Plausible locations were set in MS Encarta Professional 2004 (version 13.0.0.0531) and then transferred to copies of Encarta maps with Photoshop Elements 8 (version 8.0) since these cannot bee illustrated in Encarta directly. Some of the location marks are super-imposed due to the vicinity of the localities.

Cladistic analyses
The phylogenetic analysis comprises 50 characters for 24 genera and 5 outgroup taxa. The character matrix of As-pöck and Nemeschkal (1998) was amended to account for newly available data: the homologized characters of the genital sclerites  and data for Ormiscocerus, Berothimerobius and Tanzanberotha from the literature (Penny and Winterton 2007, Monserrat and Deretsky 1999, Aspöck and Hynd 1995. The matrix is consistent on the genus level (Supplementary file 4).
Only informative characters were included in the data matrix. The cladistic analyses were performed with TNT (Goloboff et al. 2008). Space for 500000 trees was reserved in the memory. Analyses were performed under equal and implied weights. For implied weighting (Goloboff 1993), the concavity constant K was set from 3 -15. For each weighting scheme, traditional analyses with 10 000 replications and TBR saving 20 trees per replication were conducted. The root was Nallachius (Dilaridae). Bootstrap values and the Bremer support values were calculated with heuristic search (100 000 replications, 1000 TBR branch swapping replications). For character optimization Winclada (Nixon 2002) was used.
A list of the currently known valid genera and species of Berothidae is provided (Supplementary file 5).

Cladistic analyses
The analysis using equal weights yielded twenty most parsimonious trees (MPTs) with a length of 209 steps, a consistency index (CI) of 0.34, and a retention index (RI) of 0.6. Bremer Support values and Bootstrap values are indicated in Fig. 57. In the MPTs the position of Nyrma is very unstable: it took any possible position within the Nyrminae, and in two trees the genus was sister group to the Cyrenoberothinae. Naizema, belonging to the Nosybinae, consistently resulted as sister group of the Trichomatinae in the MPTs. Spiroberotha, another nosybinae genus, switched between a position as sister group of Nosybus + Tanzanberotha within the Nosybinae, and a position as sister group of Naizema + Trichomatinae. Within the Berothinae, Stenobiella switched between being sister to all except Berlekrumyia and sister to all except Berlekrumyia plus the clade (Podallea + (Nodalla + Asadeteva)).
Three different topologies were retrieved by implied weighting with K3-15, none of which represents a MPT. Weighting with K3-4 and K6 yielded a tree that is 212 steps long, as did weighting with K5. These two tree topologies differ only in the position of Stenobiella which is the sister group of Nodalla + Asadeteva under K5, and sister group of a larger clade comprising Nodalla + Asadeteva as well as Podallea and the Isoscelipteron clade. The two tree topologies are consistent in that Podallea is the sister group of the Isoscelipteron clade, and in Trichomatinae being the sister group to Nosybinae + Berothinae. Implied weighting with K7-15 yielded a tree with 210 steps. In that tree, the subfamily Nosybinae was not retrieved as monophyletic since the nosybinae genus Naizema was the sister group of the Trichomatinae as in all MPTs.

Biogeography
The distribution areas of berothid genera are extremely diverse. At present they cannot be assigned to known general patterns, although they certainly contribute to an understanding of their biogeography.
The Australian region (Figs 47,48) comprises 8 genera (representing 27 described species), all of them -with only one exception, Isoscelipteron -are endemic. Some have undergone an amazing radiation, e.g. Stenobiella, with 12 described species and more awaiting description. Stenobiella has the largest distribution area, mainly occurring in the eastern half of the continent but with two vicariant species in the west. Trichoberotha, comprising two species, is restricted to northwestern Australia, the vicariant monotypic sister group Trichoma has a much larger distribution area in the southern half of the continent. Quasispermophorella is characterized by vicariant distribution areas of its two species in south-western and south-eastern parts of the continent. Spermophorella comprises six species which are disjunctly distributed along northern, eastern and southern borders of the continent. Isoscelipteron is distributed in the northeast, nested between the disjunct distribution areas of Spermophorella. The origin of the distribution pattern is not yet understood for the monotypic Austroberothella with two species in the southeast of the continent and its monotypic sister group Protobiella from New Zealand.
The Oriental region (Figs 49,54,55) comprises four genera with altogether 16 species. The most diverse genus Berotha with altogether five species is endemic to the Oriental region and is scattered over large areas of this realm. The genus Isoscelipteron is recorded with four species from several islands in the Oriental realm and apparently has its main diversification in this region (with only two other species in the Palearctic and one in the Australian region). Lekrugeria occurs in the northwestern transition-zone to the Palearctic region with two species (and another in the Afrotropical region, see below). The record of one species of Nodalla (predominantly distributed in Palearctic Africa, see below) from Sri Lanka might be mislabelled.
The Palaearctic region (Figs 50, 54, 55) comprises five genera representing 15 described species. Nodalla is an element of the Eremic and distributed within the West Palearctic desert belt from northwestern Africa throughout the Arabian Peninsula to Afghanistan. With eleven known species it is the most diverse Palearctic genus (Aspöck and Aspöck 1998). Whether the intra-generic differentiations date back to the Pleistocene or are much older remains speculative at present. Isoscelipteron, which is predominantly known from the Oriental (see above) and Australian regions (one species), is represented by a western-mediterranean and an eastern-mediterranean species. Asadeteva is known from two species in Pakistan and Afghanistan, respectively, both from high altitudes (1400 m -1800 m). The species are certainly Palaearctic elements, however occupy a transition zone to the Oriental realm. The monotypic enigmatic genus Nyrma which is so far only known from Anatolia has a highly disjunct Neotropical sister group (Ormiscocerus + Berothimerobius Fig.  53). This vicariance pattern is not typically Gondwanean and is not yet understood.
The Afrotropical region (Figs 51, 52, 55) comprises six genera, representing 26 species. The most diverse genus Podallea is represented by 15 species. Its distribution is confined to the Afrotropical region, including the southwestern part of the Arabian Peninsula, Madagascar and the Comoro Islands. All Podallea species are arboreal elements. The extant distribution patterns are probably mainly the result of vicariance events apparently long before the Pleistocene Aspöck 1981, 1996).The five species from Madagascar are closely related or identical with the Podallea species from the African mainland and are certainly the result of not very old colonizations . Nosybus (comprising seven species) and its monotypic sister group Tanzanberotha from Tanzania and Kenya are endemic to the Afrotropical region. Their common Neotropical sister group (Spiroberotha + Naizema) demonstrates a classical Gondwanean vicariance pattern (but see the above mention of the conflict with the Australian Trichomatinae). Lekrugeria comprises one Afrotropical species which is distributed along a sub-Saharan belt and two other species occurring in Northern India and Nepal in a small transition zone of the Palearctic and the Oriental regions. The genus represents a Palaearctic-Subsaharan-Oriental pattern, which is unique among Berothidae (Aspöck and Aspöck 1986 c). The monotypic endemic genus Manselliberotha from southern Africa exemplifies another Gondwanean vicariance pattern with its sister group, the Neotropical Cyrenoberotha. The monotypic isolated genus Berlekrumyia which is recorded from southern Africa is endemic to the Afrotropical region.
Berothidae of the Nearctic and Neotropical regions (Figs 53, 55) predominantly occupy the southern parts of NorthAmerica, as well as parts of Central and South America. The most common taxon is Lomamyia, vicariant to all other Berothinae which are restricted to other continents. The genus comprises 12 species, 10 in North America, one in Central America and one in eastern South America. All other genera of this region represent relic taxa, with one or two species only; however, they are characterized by vicariant sister groups in the Old World. The monotypic genera Ormiscocerus and Berothimerobius are endemic to Chile. The transoceanic vicariance with their joint sister group Nyrma from Anatolia does not fit into any of the typical Gondwanean patterns. The monotypic relic genus Cyrenoberotha is endemic to Chile, representing the Neotropical counterpart to Manselliberotha from the Afrotropical region. It may be interpreted as a typical Gondwanean pattern -however, its origin is not yet understood. Spiroberotha (two species) from Central America and northern South America and the vicariant Naizema (two closely related species or monotypic) from southern South America represent an unknown pattern (both belonging to the otherwise Afrotropical Nosybinae).
In summary, it can be stated that the distribution patterns of berothid genera are extremely heterogeneous. Their origin, biogeographic and biological backgrounds and their congruence with general patterns are far from being understood. 1) Two types of Gondwanean patterns are distinguishable: Neotropical / Afrotropical vicariance on the one hand, represented e.g. by Cyrenoberotha / Manselliberotha (Figs 51,53), and Afrotropical / Australian vicariance and on the other hand, e.g. represented by Podallea (Fig. 52) and the Isoscelipteron clade (Fig. 47). 2) One genus, Lekrugeria, marks a sub-Saharan Palaearctic belt which extends to the Palaearctic/Oriental transgression zone (Fig. 50).
3) The genus with the largest and highly vicariant distribution area, Isoscelipteron, comprises the Palearctic, Oriental and Australian regions (Fig. 54). 4) Finally: Endemism is a very common phenomenon among Berothidae, with radiations of genera within biogeographic realms, on one hand, or as monotypic enigmatic locus typicus relics, on the other hand.

Phylogeny
The results of our cladistic analyses will be discussed in the context of previous results reported in Aspöck and Nemeschkal (1998), on one hand, and concerning their plausibility, on the other hand. Prima vista it can be emphasized that the Rhachiberothidae (Figs 56, 57: Mucroberotha + Rhachiberotha), which in the present analysis was part of the out-group, is corroborated as a discrete family, which has already been advertised by Aspöck and Mansell (1994). This contrasts with the interpretation of the rhachiberothids as a subfamily of the Berothidae which was originally proposed by Tjeder (1959) and more recently advocated by Winterton et al. (2010). Generally, it is noteworthy that the trees retrieved with strong implied weighting (K3-6) are mostly congruent with the previous analysis by Aspöck and Nemenschkal (1998), while major conflicts emerge in the MPTs and the trees with implied weighting above K7. Conflicts in the shorter trees (MPTs and K7+) concern the Trichomatinae ( Fig. 56: TRI) appearing as part of the Nosybinae (Fig. 56: NOS) and the position of the genus Podallea as sister to Nodalla + Asadeteva within the Berothinae (Fig. 56: BER).
Within the Nosybinae the genus Nosybus (comprising seven species) and the monotypic Tanzanberotha, both from the Afrotropics (Fig. 51), emerged as the sister group of Spiroberotha (two species) from the Neotropical region (Fig. 53). These three genera are the sister group of the Neotropical Naizema (two species). Yet, the Trichomatinae appear as sister group of Naizema and thus within the Nosybinae in all shorter trees. However, this result lacks plausibility, since the complex of the gonocoxites 10 in the Nosybinae is highly differentiated, enlarged and rolled up, whereas in the Trichomatinae the complex is simple. In Aspöck and Nemeschkal (1998) and in the trees obtained by implied weighting with K3-6, the Trichomatinae are the sister group of Nosybinae + Berothinae. The Trichomatinae are endemic in the Australian region (Fig. 47) and comprise two genera: the monotypic Trichoma and Trichoberotha (with two species).
The Berothinae represent the largest clade of Berothidae and emerged as a monophylum in all trees (Figs 56,57). They comprise 11 genera: Asadeteva (two species from Asia, Fig. 50), Nodalla (11 species from northern Africa, Fig. 50), Stenobiella (12 species endemic to Australia, Fig. 47), Berotha (five species, Oriental region, Fig. 49), Lekrugeria (three species, Subsahara and Asia, Fig. 49, 50), and Lomamyia (12 species, Neactic and Neotropical region, Fig. 53). The monotypic Afrotropical Berlekrumyia appears as the sister group to the rest in all MPTs, as in Aspöck and Nemeschkal (1998), but not in the trees obtained by implied weighting with K3-6, where Lomamyia holds this position. The Podallea clade comprises the Afrotropical Podallea (with 15 species), the Isoscelipteron clade with Isoscelipteron (Palaearctic, Oriental and Australian regions with 13 species) and Quasispermophorella (two species) + Spermophorella (six species) with the latter two genera being endemic to the Australian region (Fig. 48). The monophyly of the Podallea clade was one of the highlights in the study by Aspöck and Nemeschkal (1998) and was a most convincing result. In the present analysis this relationship was obtained by implied weighting with K 3-6. However, it is most surprising and lacks plausibility that in the MPTs and under implied weighting with K7-15 (compare strict consensus Fig. 57), Podallea emerged consitsently as the sister group of Asadeteva + Nodalla. The characters 33.1 (male: complex of gonocoxites 10 looped) and 47.2 (female: ductus receptaculi coiled), which in our understanding are clearly synapomorphies of Podallea and the Isoscelipteron clade, occur as homoplasies in those trees.
One character complex that is highly interesting in the context of Berothinae and Nosybinae phylogeny is a conspicuous formation of the female sternal region of the abdominal segment seven (chars. 39 and 40): Hairy domed lateral sclerites, interpreted as gonocoxites seven, accompanied by smooth and weakly sclerotized discs which are interpreted as gonapophyses seven   question remains whether it is in fact a synapomorphy of Nosybinae + Berothinae with secondary reductions in some genera, or whether Spiroberotha, a genus which also switches positions in the present analysis, might belong to the Berothinae. The condition of the seventh female abdominal segment is a phenomenon restricted to this subfamily. In a larger context it rises the question whether this differentiation of the seventh sternite is a re-expression tracing back to the combined nature of the abdominal sternites . The Nyrminae (Fig. 55, Fig. 56: NYR) comprise the hitherto monotypic Nyrma from Anatolia (Fig. 50) and Ormiscocerus + Berothimerobius, both monotypic genera from Chile (Fig. 53). All three genera are apparently old relics from Gondwana (although they do not show a typical Gondwanean vicariance pattern). Character 16.1 (more than three sc-r crossveins in the forewing) is a synapomorphy of Nyrma, Ormiscocerus and Berothimerobius (Fig. 56). The inflated forecoxae of the males (char. 50.1) of Ormiscocerus and Berothimerobius are a very striking synapomorphy of the two genera (Fig. 56) which are geographically very close together (Fig. 53).

Concluding remarks
The discussion of the distribution areas of the genera on a worldwide scale serves as a brain storming excerise for a forthcoming biogeographic analysis which would be premature at the present state of berothid phylogeny. Nevertheless, we expect a positive cybernetic aspect in presenting both approaches. That our phylogenetic tree should be understood as starting point for falsification does not need to be emphasized.

Acknowlegements
We greatfully acknowledge Harald Bruckner (NHMV) and Franziska Denner (NHMV) for the data acquisition and Harald Bruckner additionally for the drawings of the distribution maps, Dominique Zimmermann (NHMV) for her help with tricky TNT problems, Wolfgang Brunnbauer (NHMV) for acquisition of literature, Horst Aspöck (Medical University of Vienna) for permanent discussion on berothids and valuable comments on the manuscript, and John Plant for linguistic improvements and critically reading   the manuscript. We want to thank the two reviewers for their essential input and the Museum für Naturkunde Berlin for publishing this article in open access and free of charge.