﻿A species-group key and notes on phylogeny and character evolution in New Guinean Exocelina Broun, 1886 diving beetles (Coleoptera, Dytiscidae, Copelatinae)

﻿Abstract Detailed information about the known species groups of Exocelina Broun, 1886 from New Guinea is presented, including species numbers, distribution, and references of species-group diagnoses, keys to the species, and species descriptions. An identification key to all species groups is provided. Phylogeny and morphological character evolution are discussed.


Introduction
Exocelina Broun, 1886 is a highly diverse genus of diving beetles. Most species occur in running-water habitats, especially low-order streams and habitats associated with wider mountain streams, throughout the Australian, Pacific and Oriental regions. Mainly lentic lifestyles also occur in four independent and not particularly species rich clades .
The genus was proposed by Broun (1886) for his new, most likely epigean species, Exocelina advena, described from Mokohinau Islands, New Zealand. Later, Broun (1893) recognised it as Copelatus and renamed it C. sharpi due to homonymy with the Neotropical C. advena Sharp, 1882. However, this name also turned to be a junior homonym of another Neotropical species (Branden 1884) and was synonymised with C. australis (Clark, 1863) by Zimmermann (1920).
Exocelina was infused with new taxonomic life under the name Copelatus (Papuadytes) . Papuadytes was erected based on morphological characters for 31 New Guinean species, with subsequent addition of a Chinese species (Balke and Bergsten 2003) and seven additional New Guinean species (Balke 1999;Shaverdo et al. 2005). The monophyly and generic status of the group were supported following analyses of copelatine phylogeny based on DNA sequence data (Balke et al. 2004(Balke et al. , 2007Bilton et al. 2015). Transferring more and more Australian Copelatus species to Papuadytes (Nilsson & Fery, 2006) led to the inclusion of C. australis (= Exocelina advena) in Papuadytes, and the latter name was recognised as being a synonym of Exocelina (see Nilsson 2007). Further investigation showed that the genus has a wide distribution in Australasia: 37 species (including 27 new ones) were recorded from New Caledonia (Wewalka et al. 2010;Balke et al. 2014); one species, E. cheesmanae (J. Balfour-Browne, 1939), from Vanuatu (Bilton et al. 2015); one species, E. parvula (Boisduval, 1835), from Hawaii (Nilsson 2007); two new interstitial species from Australia (Watts and Humphreys 2009;Watts et al. 2016); one new subterranean species, E. sugayai, from Malaysia (Balke and Ribera 2020).
However, New Guinea is the core of species diversity of the genus and, therefore, was the focus of our taxonomic project started in 2012. Since the publication of Shaverdo et al. (2005) on Exocelina of the island, 116 new species have been described (Table 1), increasing the number of Exocelina in New Guinea to 152 species and the number of Exocelina worldwide to 209 species (Nilsson and Hájek 2022). We believe that further extensive fieldwork in New Guinea and careful taxonomic investigation of the group might reveal the existence of more new species.
This paper aims to unite and discuss all known information on systematics of the New Guinean Exocelina provided in our previous studies (Table 1), focusing on the infrageneric structure of the group. Since all species groups were treated in numerous separate publications, we believe that this paper will provide better orientation in this species-rich genus, and easier species identification. Additionally, since the proposed species-group structure is based not only on morphological characters but also supported by molecular analyses, we believe that it is a good tool for understanding New Guinean Exocelina phylogeny and character evolution.

Materials and methods
Our study is based on published articles on the taxonomy of New Guinean Exocelina. In cases where specimen study was necessary, we followed the methods described in detail in our previous articles (Shaverdo et al. 2012. The results are presented as a species-group table and a key to species groups. The table includes all known species groups of New Guinean Exocelina with their numbers of species and subspecies, species-group distribution, and references for each group: species-group diagnoses, keys to species identification and species descriptions. The key provides identification to the species-group level and is meant to be a start point in the determination of New Guinean Exocelina. To illustrate the key, figures from our published articles are used, as indicated for each figure in the captions.

Species-group structure
We recognise 26 species groups of New Guinean Exocelina. The groups were proposed based on our study of morphological characters of the species and data from molecular phylogenetic analyses, where the main diagnostic criteria were structure of the genitalia and relative position of the species in the phylogenetic trees.
Most of the species in New Guinea are lotic, that is, associated with running water habitats. All of these species form one monophyletic group and are, thus, endemic to the island. The only exception is the stagnophilous species E. baliem Shaverdo, Hendrich & Balke, 2013, which belongs to the E. ferruginea group. This group has two other representatives: the Australian E. ferruginea (Sharp, 1882) and the New Caledonian E. inexpectata Wewalka, Balke & Hendrich, 2010(Shaverdo et al. 2013.

Key to New Guinean species groups of Exocelina
The key is proposed for identification of the species groups and species in the case of monotypic groups. The keys to species of individual groups can be found in the publications listed in Table 1.
The key is mostly based on male characters, but organised in a way to get one as far as possible with female identification. In many cases, females cannot be assigned to species due to the similarity of their external and internal structures (for female genitalia see figs 17a, b in Shaverdo et al. (2005) and fig. 7 in Shaverdo et al. (2013)). Some species are rather similar in external morphology and, therefore, in most cases, the male genitalia need to be studied for reliable species identification. However, for some groups, identification of the females is possible to the species group and even to species. The important point here is not to separate females from males from the same locality. Their identification should follow identification of the males of all species from the chosen locality. If co-occurring species are not numerous (2-4 species), successful identifications of females are highly possible. Male protarsomere 5 strongly modified: concave ventrally, sometimes with anteroproximal setae enlarged. Male protarsomere 4 with anterolateral hooklike seta small, not developed ( Median lobe very broad, robust, almost parallel-sided, with weak median constriction in ventral view; lateral sides strongly thickened; apexes of ventral sclerites very unequal: right one much longer than left one (Fig. 10)  Paramere with distinct dorsal notch and subdistal part well developed (Fig. 11A)  Median lobe almost parallel-sided, often narrowed distally before or to apex or broadened subdistally; its apex usually with thickened sides, slightly or distinctly enlarged ("swollen", often in shape of a baby pacifier), rounded, truncate, or slightly concave in ventral view (Fig. 15)  Median lobe thinner in apical half; in ventral view, evenly attenuated to pointed apex and, in lateral view, evenly broad, with rounded apex; its lateral margins slightly thickened (Fig. 16A)  Median lobe in lateral view broader, more strongly curved, more or less evenly attenuated to thinner apex; in ventral view, apex bluntly pointed (Fig. 16D) Median lobe with discontinuous outline in ventral and often in lateral views (Fig. 4)  Apex of median lobe with two lateral and one dorsal prolongations (Fig. 18)  Paramere with numerous small spines, no long setae. Apex of median lobe thick, short and slightly curved downwards, its minuscule tip curved upwards in lateral view (Fig. 19) . Apex of median lobe with thickened sides, often distinctly enlarged ("swollen"), in lateral and ventral views often of shape of a baby pacifier, rounded, truncate, or slightly concave in ventral view (Fig. 20C, D) fig. 1) 2 Structure of male protarsomeres 4 and 5 of E. aipo             fig. 12A, B) B E. pui Shaverdo & Balke, 2017(Shaverdo et al. 2017b fig. 16A, B) C E. casuarina   (Shaverdo et al. 2018: 54, fig. 26A, B) D E. keki Shaverdo & Balke, 2018(Shaverdo et al. 2018 fig. 30A, B).

Phylogeny and infrageneric structure
The infrageneric structure of New Guinean Exocelina is largely based on the molecular phylogeny of the group, most of the species groups being represented as monophyletic clades on the phylogenetic tree (Fig. 22). We consider this approach to be very useful for understanding the taxonomy and evolution of such a species-rich group. Earlier phylogenetic analyses based on molecular data substantiated the lotic New Guinean Exocelina as a monophyletic group, which emerged from a single colonization event by an Australian lineage and led to a rich species radiation on the island (Balke et al. 2004(Balke et al. , 2007. More recent investigations suggested an origin of New Guinean Exocelina during the late Miocene, ca 5 or 9 million years ago (Toussaint et al. 2014, or even in the mid-Miocene, ca 17 Ma, when the New Guinean orogeny was at an early stage (Toussaint et al. 2021) and inferred a constant process of lineage diversification with a continuous slowdown in speciation.
A second colonization event was by a lentic species, evident from the presence of only one extant species, i.e., E. baliem from wetlands in the Baliem Valley of Papua Province (Shaverdo et al. 2013). According to an unpublished molecular phylogenetic analysis, this species forms a clade with the Australian E. ferruginea and the New Caledonian E. inexpectata and is placed together with them in the E. ferruginea group (Fig. 23).
The 151 lotic New Guinean Exocelina species form a monophyletic group, which contains two clades: the smaller clade I with only six species groups and the distinctly larger clade II with 19 species groups (Fig. 22). In clade I, only the monophyly of the E. ullrichi group and two monotypic groups (E. mekilensis group and E. koroba group) is well resolved. The majority of the remaining species are placed in the E. casuarina group, whose phylogeny is discussed in details in Shaverdo et al. (2018). With 24 species, this group is the second largest species group of New Guinean Exocelina. Interestedly, the E. aipo and E. okbapensis groups together form a monophyletic clade despite having rather distinct morphologies.
Clade II itself also consists of two large subclades (1 and 2 in Fig. 22). Subclade 1 is very heterogeneous and includes 10 species groups (all species-poor); seven represented as monophyletic clades. The E. danae group, the most speciose group of the clade, is inferred as polyphyletic, and the E. bacchus and E. warasera groups are both paraphyletic. Subclade 2 is the most species-rich clade since it contains the largest species group of New Guinean Exocelina, the E. ekari group. This group includes 62 species and is monophyletic, forming a monophyletic clade with the E. skalei group. The remaining seven groups also form a monophyletic clade (the E. ascendens complex) and represent rather different morphological lineages. Whilst species placement into species groups using morphology worked well for the other groups and was later confirmed by phylogenetic analysis, species of the E. ascendens complex can mainly be placed using molecular data. Without these data, the groups would probably never be organised in a way that reflects their evolutionary history (see below).

Notes on character evolution
More than 20 different morphological characters were used to describe species and organise species into a species-group structure. Some characters are very diverse and have more than 10 different states, e.g., antennal shape (9 states), shape of the median lobe (14 states), setation of the dorsal side of the parameres (12 states). Here, we briefly discuss characters, which we think are the most taxonomically and phylogenetically important and worthy of further study, not only as separate characters but also in combination.

Structure of the male genitalia
The shape of the median lobe and paramere and their setation are very diverse and serve as the basic characters for species-group structure in New Guinean Exocelina. These characters were primarily used to group the species. The most divergent on these characters is the E. ekari group with a discontinuous outline of the median lobe, which could be considered an autoapomorphic character; only E. skalei with its slight apical discontinuity of the median lobe belongs to the E. skalei group. Together with the recently described E. mimika Shaverdo & Balke, 2020, this former member of the E. ekari group, was placed into the E. skalei group based on the reduced setation of its paramere. Representatives of E. ekari group have the most complicated and diverse shape and setation of the median lobe and paramere of all New Guinean Exocelina. Most likely, this results from strong sexual selection, or adaptive evolution for sexual isolation, since many species of the group co-occur (up to six species) which is not the case in other species groups.
Almost every species group has its own characteristic shape of the median lobe and paramere and their setation or combination of these characters. As already mentioned above, the most problematic was the placement of species of the E. ascendens complex that have male genitalia similar to some species of the E. casuarina-, E. aipo-and E. okbapensis groups.

Lateral bead of the pronotum
Presence and part or complete reduction of the lateral pronotal bead are states of this actively used in the key character. It is helpful for species identification, but could not be used reliably for phylogenetic purposes. Absence of the lateral pronotal bead is obviously homoplastic. It has developed independently probably up to eight times within New Guinean Exocelina (Fig. 24). Interestingly, absence of the lateral pronotal bead is characteristic for some representatives of the largest species groups: E. ekari group and E. casuarina group. A few species demonstrate a very narrow pronotal lateral bead or presence of its traces.

Modification of the male antennae
New Guinean Exocelina includes more species with modified antennae than any other genus of Dytiscidae; 45 species have them, mainly in males. The degree of modification and number of antennomeres involved are specific for certain species and/or species groups and strongly vary (up to nine different character states) from almost all antennomeres slightly stout to some of them extravagantly enlarged or extremely reduced (Fig.  17). Half of the species (31 spp.) of the E. ekari group have modified antennae, whilst this character is absent in the second largest group, the E. casuarina group. For the E. ullrichi group, it is a group-diagnostic character, as well as for the E. miriae subgroup of the E. danae group (Fig. 7). Modified male antennae evolved independently up to 10 times in different groups, including five different lineages within the E. ekari group (Fig.  24) and could be used for delimitation of the subgroups within it. It is worth noting that, in some species, modification of the antennae is correlated with stronger dorsal surface structure (especially in females) or/and sometimes with diminution of the hook-like setae of the male protarsomere 4. This may indicate association with sexual processes.

Anterolateral seta of the male protarsomere 4
Hook-like anterolateral seta of male protarsomere 4 is the main diagnostic character of the genus Exocelina and its unique morphological autoapomorphy. However, secondary diminution or differences in shape are observed in many New Guinean species of the different species groups and have obviously independently involved (Fig. 24). It is currently impossible to postulate why certain species show such characters, although as with other features, sexual selection is likely involved. In the E. ekari group, diminution of the seta often occurs in species with enlarged male antennomeres and sometimes also with stronger dorsal surface structure, e.g., species close to E. polita (Sharp, 1882). In the E. casuarina-, E. danae-, E. jaseminae-, E. warasera-, or E. bacchusi groups, representatives of which have simple antennae, reduction of the hook-like seta does not correlate with this character, however, and can be found in species with shiny or matt dorsal surfaces. Although all representatives of the E. ekari group without lateral pronotal bead have rather strongly developed hook-like seta on male protarsomere 4, diminution of the hook-like seta was observed in some species of the E. casuarina group without the pronotal bead.

Conclusion
New Guinean Exocelina represent a large and diverse group of Copelatinae beetles. Here, and in our previous publications (Table 1), we provide comprehensive taxonomic and faunistic treatments for this radiation. Further investigation of the group will definitely lead to more new species descriptions, some degree of restructuring of the species-group classification and better understanding of species distributions across the island. Having very diverse and intriguing character combinations, New Guinean Exocelina are an excellent potential model system for detailed studies on the evolution of homoplastic characters, co-evolution of different species, sexual dimorphism, and sexual conflict during mating.