Resolving the taxonomic conundrum in Graphoderus of the east Palearctic with a key to all species (Coleoptera, Dytiscidae)

Abstract The Holarctic diving beetle genus Graphoderus (Dytiscinae, Aciliini) contains relatively few and well-known species but these may still be difficult to identify based on external characters. A taxonomic problem in the eastern Palearctic was discovered that relates to the Palearctic Graphoderus zonatus (Hoppe, 1795) and the Nearctic Graphoderus perplexus Sharp, 1882. Based on qualitative and quantitative characters, especially on male genitalia which have been poorly studied in the past, it is shown that eastern Palearctic specimens identified by previous authors as either of the two species in fact belongs to a third species. The synonymized name Graphoderus elatus Sharp, 1882, is reinstated as a valid species (stat. n.) and a lectotype is designated from the mixed syntype series. The male genitalia of all known Graphoderus species have been examined and an illustrated identification key to the genus is provided. The three species in the complex of focus, Graphoderus elatus, Graphoderus zonatus and Graphoderus perplexus are found to have allopatric distributions; Graphoderus perplexus in the Nearctic region, Graphoderus zonatus in the west Palearctic region and eastwards to the Yenisei-Angara river and Graphoderus elatus east of the Yenisei-Angara river. All previous records of either Graphoderus zonatus or Graphoderus perplexus in the east Palearctic, east of the Yenisei-Angara river turned out to be misidentified Graphoderus elatus. This conclusion also brings with it that dimorphic females, thought only to be present in the single subspecies Graphoderus zonatus verrucifer (CR Sahlberg, 1824), proved to be present also in a second species, Graphoderus elatus. The dimorphic female forms is either with dorsally smooth elytra and pronotum or conspicuously granulated elytra and wrinkly pronotum. As has been shown in Graphoderus zonatus verrucifer there is a correlation between the occurrence of granulate female forms in a population and an increase in the number of adhesive discs on pro- and mesotarsus in males within Graphoderus elatus.


Introduction
The genus Graphoderus Dejean, 1833 consists of medium sized (10-16 mm) diving beetles of the family Dytiscidae. Adults are dorsally testaceous to rufous with black irrorations and except for one species they all have two transverse black bands across the pronotum Holmen 1995, Larson et al. 2000). Both adults and larvae are found in ponds, smaller lakes, bogs or wetlands Holmen 1995, Larson et al. 2000). Larvae feed mainly on planktonic microcrustaceans but also on larvae and pupae of mosquitoes (Culicidae) (Galewski 1975(Galewski , 1990. Adults are predatory on crustaceans, insects and worms (Galewski 1990). Oviposition, and larval and pupal development takes place from April or May to October, and in some years there is also a partial second generation (Galewski 1990, Hilsenhoff 1993, Nilsson and Holmen 1995, Foster 1996. Adults overwinter in aquatic habitats, under Sphagnum in bogs or hidden in bottom sediments of water-bodies which do not dry out (Galewski 1990, Hilsenhoff 1993. Males have enlarged protarsomeres I-III forming a palette with adhesive discs and mesotarsomeres I-III may be with or without adhesive discs (Nilsson and Holmen 1995).
This order was shaken when Nilsson et al. (1999) reported the Nearctic species G. perplexus to also occur in the east Palearctic and we started to discover additional G. perplexus-like specimens from the east Palearctic.

Graphoderus species in the spotlight
One of the species, G. bilineatus, has received increased attention after it was put on the EU list of species in Annex II under the Habitats Directive. EU member states were required to designate special areas of conservation for the species in Annex II and report on their conservation status. Environmental agencies in several EU countries have since made focused inventories of this species to get better data on its occurrence, distribution and abundance (Hájek 2004, Hendrich and Balke 2005, Kalniņš 2006, Hendrich and Spitzenberg 2006, Naturvårdsverket 2011, Hendrich et al. 2011, Iversen et al. 2013, Przewoźny et al. 2014, Scheers 2015. Results of the inventories have also contributed to new studies investigating the distribution of G. bilineatus in neighboring countries (Kálmán et al. 2011, Csabai et al. 2015. Another Graphoderus species which has received substantial attention lately is G. zonatus and in particular its subspecies G. zonatus verrucifer, as its females are dimorphic with one morph dorsally smooth like the male and the other morph with a peculiar wrinkly pronotum and roughly granulated elytra (Nilsson 1986, Nilsson and Holmen 1995, Bergsten et al. 2001, Härdling and Bergsten 2006, Karlsson-Green et al. 2013. The two morphs co-occur in varying proportions in different populations and it was shown by Bergsten et al. (2001) that the proportion of the granulate morphs in a population was significantly correlated with suction cup characteristics of the male's pro-and mesotarsal palettes. With a higher proportion of the granulated female morph male pro-and mesotarsal palettes got wider, the three large suction cups got larger and the smaller suction cups got smaller and more numerous (Bergsten et al. 2001). The female granulate morph was interpreted as an antagonistic character evolved in an arms race with male suction cups over aspects in the mating, e.g. frequency and timing (Arnqvist and Rowe 2005). The antagonistic nature of the dorsal female sculpture to the function of the mechanically working male suction cups was inferred mathematically from first principles by Bergsten and Miller (2007) and later shown empirically (Karlsson-Green et al. 2013). Both theoretical (Härdling and Bergsten 2006) and empirical (Karlsson-Green et al. 2014) work has been conducted to try to understand the role of selection and drift for the distinct morphs to be able to co-occur over time.

The taxonomic conundrum in the east Palearctic
In 1882, Sharp (1882) described a Graphoderus species with Holarctic distribution, Graphoderus elatus Sharp, 1882, based on material from both Canada and east Siberia. Sharp (1882) separated this species from the Nearctic species G. perplexus, described in the same work, based on the number of adhesive discs on pro-and mesotarsus. Both species were distinguished from the Palearctic species G. zonatus by being narrower in front and having wider epipleura. Horn (1883) was not convinced that the male tarsal characters given by Sharp (1882) indicated separate species and made G. fascicollis, G. perplexus and G. elatus synonyms of the Palearctic G. cinereus. Zimmermann (1920) disagreed with Horn (1883) and instead placed G. perplexus and G. elatus as synonyms of G. zonatus. We will refer to this "species group" (G. zonatus, G. perplexus and G. elatus) which might not be closely related, as the zonatus-species complex and it is distinguished from other Graphoderus species in that the black pronotal bands mostly neither reach the anterior nor the posterior pronotal margins Holmen 1995, Larson et al. 2000). Further, Gschwendtner (1937) mentioned that G. perplexus and G. elatus were not particularly different from G. zonatus. Instead Wallis (1939) was the one who separated G. perplexus and G. elatus from G. zonatus by the same characters which Sharp (1882) used -wider epipleura and a body shape more narrowed in front. Also, Wallis (1939) specified differences between G. zonatus and G. perplexus/G. elatus in the bifurcation of the chitinous ring enclosing the genitalia. However, after having examined over 100 males, he proposed that G. perplexus and G. elatus could not be separated from each other based on the number of adhesive discs and he synonymized them as the same species with perplexus as the valid name. Zaitsev (1953) (English translation (Zaitsev 1972)) described the Russian fauna of Dytiscidae in 1953 and did not mention G. perplexus as occurring in the Palearctic region. However Zaitsev (1972) put Sharp's G. elatus type from east Siberia as a synonym of G. zonatus and also reported G. zonatus to be distributed in North America. Later, in the treatment of the Russian Far East Dytiscidae by Lafer (1989), the G. elatus type from east Siberia was not referred to at all, instead Lafer only mentioned G. zonatus from the zonatus-species complex to occur in the Palearctic. Yet, in 1999Nilsson et al. (1999 studied newly collected material from Kamchatka (Russian Far East) and identified the material of a Graphoderus species as G. perplexus based on the number of tarsal adhesive discs which was significantly fewer than the material from Urup, Kuril Islands, instead identified as G. zonatus (Nilsson et al. 1997). This resurrected the old hypothesis by Sharp (1882) of a Graphoderus species occurring in both the Nearctic and east Palearctic realms, now under the name G. perplexus. This was also transferred to the world catalogue (Nilsson 2001(Nilsson , 2015 and the Palearctic catalogue (Nilsson 2003a, Nilsson andHájek 2015). There is however a problem with using the number of tarsal suction cups as a sole distinguishing character because of its variation and correlation with female counter-adaptions (Bergsten et al. 2001). The number of suction cups was used by Nilsson (1986) to delimit the two subspecies G. zonatus zonatus and G. z. verrucifer but only based on Swedish material. In Sweden the two subspecies could be delimited geographically in that south of a diagonal line across southern Sweden no granulate females were known (Nilsson 1986). On average the number of both pro-and mesotarsal suction cups were higher in the northern subspecies (G. z. verrucifer) but the variation overlapped (Nilsson 1986). Populations of G. z. verrucifer occur as well in Italy but these populations were earlier considered to be a variety of G. cinereus (var. bertolinii Seidlitz, 1887), which Pederzani (1986) instead reported to belong to a relict population of G. verrucifer (CR Sahlberg, 1824), new to Italy. The same year Nilsson (1986) changed the status of G. verrucifer to a subspecies of G. zonatus. Yet, how to delimit the subspecies of G. zonatus in the eastward extension of the distribution across Russia and Siberia to Japan is poorly understood (Nilsson and Holmen 1995). Both subspecies of G. zonatus are listed for Mongolia and all records of G. zonatus from China and Japan are considered to be G. z. zonatus whereas all western and eastern Siberian as well as Far East Russian G. zonatus are considered to be G. z. verrucifer (Nilsson 2003a, Nilsson andHájek 2015). The number of tarsal suction cups in the Kamchatka material is in fact partly within the documented range for G. z. zonatus (Nilsson 1986, Nilsson et al. 1999 which was probably excluded only on the basis of a presumed more southern distribution. In summary then, there is an unresolved taxonomic conundrum in the east Palearctic. Is there really a partly Holarctic Graphoderus species spanning both sides of Beringia? Are there other distinguishing characters apart from the number of males' suction cups to shed light on the zonatus-species complex? We are especially interested in evaluating the diagnostic power of the male genitalia because the genitalia in Graphoderus have not been as extensively used as in e.g. Agabini due to their partly soft-tissue nature (Nilsson and Holmen 1995). There is no reason to believe that it is less informative than the genitalic characters in the sister-group Acilius Leach, 1817 (Bukontaite et al. 2014) which recently helped to solve a taxonomic confusion in the Nearctic region (Bergsten and Miller 2006). If there is no Holarctic Graphoderus how should we treat G. elatus based on the syntype series including both continents? In the updated world catalogue (Nilsson 2015), G. elatus is listed with a "?" as a synonym under G. perplexus and in the updated Palearctic catalogue (Nilsson and Hájek 2015) G. perplexus is listed as occurring in far east Russia and Nearctic but without the synonym G. elatus.
The aim of this study is to 1) resolve the taxonomic conundrum of the zonatusspecies complex in the east Palearctic, 2) evaluate the usefulness of the male genitalia for species identification and delimitation in Graphoderus and 3) to provide an identification key and iconography with habitus and male genitalia images of all Graphoderus species.

Methods
Examined material came from the following collections referred to by their abbreviation:

BMNH
The

Measurements
Genitalia were prepared from 57 male specimens of Nearctic G. perplexus and from sixteen males of the G. perplexus-like specimens from east Palearctic, shown in this paper to be G. elatus Sharp, 1882. Six measurements were then taken (in micrometers, µm) from photographs of the genitalia, with focus at the anterior lobes of the penis. The camera was an infinity X, mounted on an Olympus SZX12 stereomicroscope using the program DeltaPix InSight v4.0.9. The measurements were; PW = penis width (at midway between apex and base), PL = penis length, PCLL = penis central lobe length, PLLL = penis lateral lobe length, PCLWb = penis central lobe width at base and PCLWt = penis central lobe width at level of lateral lobe apex (Fig. 1a). Another five measurements were also taken from seventeen males and ten females of G. elatus from the east Palearctic; TL = total body length, EL = elytral length, MEW = maximum elytral width, PrL = pronotal length and PrWb = pronotal width at base (Fig. 1b). Total body length (TL) was also measured for the 57 Nearctic male specimens of G. perplexus.
Four ratios, PL/TL; PW/PL; PLLL/PCLL and PCLWt/PCLWb, were calculated from the measurements to test the hypothesis that there were no differences in ratios between the populations from the two continents. The ratios were tested with separate, independent 2-group Mann-Whitney tests with default settings in R version 3.2.2 2015-08-14 (R Team Core 2015). Bonferroni correction, a multiple-test correction, was used to compensate for testing several measurements from the same individuals against each other (Dunn 1961). Boxplots of the ratios with median and ± 25% of the ratio values (whiskers showing minimum and maximum ratios), were also made in R.
The number of adhesive discs on pro-and mesotarsus was counted for seventeen male specimens of Nearctic G. perplexus and fourteen males of G. elatus from the east Palearctic, excluding males from populations with or possibly with granulated females. The average number of adhesive discs from left and right pro-and mesotarsus of each specimen was calculated and the numbers were tested against the hypothesis that there were no differences between the populations from the two continents. This was tested with independent 2-group Mann-Whitney tests with default settings in R, in which also boxplots were made. The tests were repeated, this time including males from populations with granulated females which did not alter the results (not shown).

Images
Habitus photographs of the species in dorsal view were taken with a Canon EOS 5D DSLR, and a Canon 100mm 2.8L Macro lens mounted on a motorized rail (Stackshot) from Cognisys. The photos were stacked in Zerene Stacker v1.04 and edited in Digital Photo Professional v3.13.20 and Adobe Photoshop CS5. Photographs of the male genitalia in dorsal and lateral view (slightly dorsolateral to avoid penis apex being hidden by tips of the parameres) were taken with the same infinity X camera and Olympus SZX12 stereomicroscope as above. The photographs shown in Figure 7, 10 and 11 were taken with a Leica M125 stereomicroscope + Canon EOS 550D digital camera in the Sackler Bioimaging Laboratory of the Natural History Museum, London. They were stacked using Helicon Focus software. The photographs of the genitalia in dorsal view were edited in Adobe Photoshop CS5 and the photographs of the lateral view were used as aid to make line drawings of each species penis in Adobe Illustrator CS5. The final plates were made in Adobe Photoshop CS5.

Results
Examination of male genitalia in the zonatus-species complex revealed that the lateral view of the central lobe at the penis' trifid apex was highly diagnostic to separate G. zonatus from G. perplexus and G. elatus (Fig. 2). In G. zonatus, in lateral view the dorsal margin of the central penis lobe forms an even convex curve (Fig. 2x). In G. perplexus and G. elatus the same view shows a concave shape (Fig. 2v, l). The shape was identical in both subspecies of G. zonatus, including material from the population in the Italian Alps, monomorphic for the granulated female morph. Examination of east Palearctic material based on this character showed that all material east of the Yenisei river and its headwater tributary Angara river was not G. zonatus although it had often been misidentified as G. zonatus (Zaitsev 1972, Lafer 1989, Mori and Kitayama 1993, Nilsson 1995, Nilsson et al. 1999. As well, all records from the northernmost peninsula on Hokkaido in Japan of G. zonatus are based on misidentifications (Mori and Kitayama 1993). The easternmost record of G. zonatus we have studied is a male taken by RB Angus in 1970 at Dachnaya, just west of Irkutsk Lat. 52.1220°N Lon. 104.0840°E.
Reported G. zonatus from Sakhalin and Kuril Islands (Nilsson et al. 1997(Nilsson et al. , 1999) are based on misidentifications as shown by reexamination of the material in Nilsson's collection donated to NHRS in 2013. Records of G. zonatus from northeast China are with all certainty also based on misidentifications (Nilsson 1995), as we found material from Inner Mongolia, Heilongjiang and Qinghai to have the concave shape of the central penis lobe. This shape was also identified in beetles from northeast Mongolia, Onon river, but the distribution of G. zonatus in the north-central part of Mongolia (Shaverdo et al. 2008) is more uncertain. Apart from the male genitalia, G. zonatus can also be distinguished from G. perplexus and G. elatus by having narrower epipleura posteriorly.

Granulate females and adhesive discs
The realization that true G. zonatus could be distinguished by the shape of male genitalia in lateral view and based on this also could be inferred not to occur east of Yenisei-Angara river brought about an enticing novelty. G. zonatus was no longer the only Graphoderus species with dimorphic females, one morph of which had elytra granulated and pronotum wrinkled and the other morph which had smooth elytra like the males. Granulated females have been reported east of the Yenisei-Angara river, e.g. from the Kuril Island Urup, which has been seen as evidence for the subspecies G. zonatus verrucifer (Nilsson et al. 1997). But our examinations of the same Kuril material showed that the males from that population were not G. zonatus. We have also found granulated females from Inner Mongolia (leg. Li, Chunyuan and Chaoqun) and Yakutsk with a male from the latter population rejecting the identity as G. zonatus. As well, in the material from North Sakhalin we found females with smooth elytra, however the male from the same population had a higher number of suction cups indicating that female dimorphism can occur within the same population. Closer comparison of the granulated females showed a somewhat more irregular granulated structure on the elytra, more elongated convexities of the granules and a stronger tendency towards forming longitudinal lines in the east Palearctic material (Fig. 3) but this difference is so far based on too few granulate females to be considered conclusive. Just as predicted by the arms race hypothesis males from granulated populations showed a higher number of adhesive discs on the tarsi. Males from the North Sakhalin and Urup populations had 59-66 protarsal suction cups and 29-31 mesotarsal suction cups. This should be compared with 28-47 protarsal and 14-20 mesotarsal suction cups from localities without granulated females. Granulated Graphoderus females have never been found in the Nearctic region but females of the Nearctic species may sometimes have wrinkles on the pronotum.

Quantitative morphometrics
As Nearctic G. perplexus and east Palearctic G. elatus have a similar concave shape of the male central penis lobe in lateral view, various characteristics of the male genitalia were quantified to test for con-or heterospecificity (see Methods). In particular we had from initial examination noted that the penis, as well as the entire genitalic package with parameres, seemed to be notably longer in the east Palearctic specimens (compare Fig. 2k with 2u). Two separate Mann-Whitney tests showed that G. elatus from the east Palearctic are distinguished from Nearctic G. perplexus in the male genitalia by significant difference in the ratio PL/TL (relative penis length, P < 0.001) (Fig. 4a) and in PW/PL (penis shape, P < 0.001) (Fig. 4b) (Table 1). Two separate Mann-Whitney tests also showed that Nearctic G. perplexus had significantly fewer adhesive discs on both pro-and mesotarsus (P < 0.001) (Fig. 5). Figure 6 illustrates the difference of pro-and mesotarsus between G. perplexus and G. elatus.
No statistical significance were found in the ratios PLLL/PCLL (relative extension of central penis lobe to lateral lobes, Mann-Whitney test, P = 0.8051) or in PCLWt/ PCLWb (anterior narrowing of central penis lobe, Mann-Whitney test, P = 0.2628) ( Table 1). The four genitalia ratio tests were conducted using Bonferroni correction to adjust alpha to 0.0125 (0.05/4). The results clearly reject the hypothesis that G. perplexus and G. elatus are the same species.      Kamiya (1940), Balfour-Browne (1946), Zaitsev (1972), Lafer (1989), Zeng (1989), Mori and Kitayama (1993), Nilsson (1995), Nilsson et al. (1997); Graphoderus zonatus zonatus sensu Nilsson (2003a) (in part), Nilsson and Hájek (2015) (in part);  Lectotype justification. Sharp (1882) based his description of Graphoderus elatus on two male specimens, one from "Eastern Siberia (Amurland)" and the other from "North America (Red River)". He gave G. elatus the number 995. These two syntype specimens, present in Sharp's collection (BMNH), were studied and genitalia extracted. Both are pinned through cards, these cards being mounted on longer pins. The Siberian specimen is labeled on the face of the card, in Sharp's handwriting "Eastern Siberia 995 elatus", while the North American specimen is labeled "Red River. Am. Bor. 995 var". The syntype specimen from Red River Am. Bor. we consider conspecific with G. perplexus. The Lectotype of G. perplexus (designated by Larson 1975) was studied at BMNH but is a female so the shape of male genitalia could not be compared. The characters Sharp (1882) used to distinguish G. elatus from G. perplexus are unconvincing as already argued by Wallis (1939). The syntype specimen from Siberia has the concave outline of the penis' apex central lobe in lateral view. The penis is also of the longer type (Fig. 2k) and belongs to what we initially called G. perplexus-like specimens from the east Palearctic. We therefore designate the Siberian specimen as lectotype of Graphoderus elatus Sharp, 1882, and have so labeled it. What we initially called G. perplexus-like specimens from the east Palearctic belong to Graphoderus elatus which is here reinstated as a valid species (stat. n.). The American specimen although belonging to G. perplexus, is a paralectotype of G. elatus. As Sharp's description of G. elatus is very short and was based on a mix of two species we provide a redescription and documentation (Fig. 7) based on the designated lectotype followed by a discussion of the intraspecific variations as here interpreted from all examined material ( Table 2).
Description of the Lectotype ♂. Body length 14.0 mm; maximum elytral width 8.3 mm.
Head (Fig. 7a) dorsally testaceous; basal black band extending between eyes, retracted under anterior of pronotum, extending apically to eyes with no apparent separation from eyes; V-shaped black marking anteriorly. Head ventrally testaceous; clypeus testaceous; maxillary palpi yellow with apical palpomere piceous distally. Labial palpi yellow with apical palpomere darkened along its inner margin. Antenna testaceous with antennomeres more or less piceous in distal half. Pronotum (Fig. 7a) 2.1 mm long; maximum width 6.5 mm; testaceous; transverse black bands separated from anterior margin by testaceous band and from posterior margin by narrow testaceous, more piceous band; black bands do not reach sides of pronotum; anterior black band at edges with narrow posteriorly directed projections; posterior black band with narrow lateral portions projecting from basal margin of band. Elytron 10.5 mm long; smooth, yellow with black irrorations; irrorations reduced along margins of elytra; sutural midline black; scutellum piceous.
Ventral side (Fig. 7b) testaceous-rufous, darkened due to age so that minor variations in its color not apparent. Forelegs testaceous; midlegs partly testaceous, mesotibia and meso tarsus rufous-testaceous with golden setae along edge; metatrochanter and metafemur testaceous; metatibia rufous with golden setae along edge; metatarsus rufous-piceous with long golden setae along edge. Elytral epipleuron testaceous; broad anteriorly, gradually tapering along edges of abdomen but relatively broad also in first part of posterior half.
Protarsal claws similar in size and shape, shorter than protarsomere V; mesotarsal claws similar in size and shape. Posterior metatarsal claws almost three times as long as anterior metatarsal claws. Protarsomeres I-III enlarged with three larger adhesive discs basally and about 32 smaller discs distally. Mesotarsomeres with two more or less regular rows of seven discs, left mesotarsus with one additional smaller disc on mesotarsomere I and a second on mesotarsomere II and right one with only one extra disc, on mesotarsomere II.
Penis in dorsal view (Fig. 7c) about 2.8 mm long; width 0.55 mm; apex trifid with three distinct lobes. Penis in lateral view (Fig. 7d) with lateral apical lobes fairly slender; central penis lobe sclerotized along edge, longer than side lobes, its dorsal margin concave (Fig. 7e). Parameres with external margins straighter medially, their apices convergent (Fig. 7f). Lappets of aedeagal ring sclerite short and wide, their outer apical margins rounded (Fig. 7g).  Intraspecific variation. Body length between 13.9 and 16.3 mm; maximum elytral width between 8.0 and 9.6 mm. Pronotum length 1.8 to 2.5 mm long; width 5.3 to 7.2 mm; smooth in males; either deeply wrinkled (when also elytra granulated) or smooth in females; in smooth specimens either shining with anterior row of impressed punctures very distinct, or matt with puncture-row less distinct; anterior black band of pronotum mostly continuous, sometimes thin and weak or non-continuous; shape of posterior black band of pronotum varies, separated from posterior margin by testaceous band which sometimes is partly piceous. Elytron between 10.4 and 12.7 mm long; smooth in males; smooth or granulated in females.
Male posterior metatarsal claws almost three times as long as anterior metatarsal claws; female posterior metatarsal claws less than twice as long as anterior metatarsal claws which are slightly curved apically. Protarsomeres I-III enlarged in males with three larger adhesive discs basally and 28-66 smaller discs distally; mesotarsomeres in males with irregular rows of 14-31 adhesive discs; in populations with granulated females, number of adhesive discs in males are in upper range. Penis in dorsal view between 2.4 and 2.8 mm long; width between 0.6 and 0.7 mm. Shape of lappets in aedeagal ring sclerite variable which also applies to outer apical margin.
Distribution. (Fig. 8) The distribution covers Russia, east of the Yenisei-Angara river to the pacific coast, north to the East Siberian Sea and south to Qinghai in China and Hokkaido in Japan. Specimens from the following regions in east Palearctic were examined: Russia: Yakutia, Kamchatka, Chita Region, Amur Region, Kuril Islands, North Sakhalin, Verkhoyansk and Indigirka. Japan: Hokkaido. Mongolia: Onon river. China: Heilongjiang, Inner Mongolia and Qinghai.

Key to Graphoderus species
As the resolved situation in the east Palearctic means that there are no species in common between Nearctic and Palearctic the key is constructed with a first dichotomy between the continents for ease of use. In order for both males and females to be identifiable, each key step has multiple characters and characters of the pronotal black bands are included as they are often very useful albeit not always absolutely trustworthy. Mesotarsal formula, e.g. 6-4-4, refers to six adhesive discs on mesotarsomere I, four on mesotarsomere II and four on mesotarsomere III. Note that there are errors in the Graphoderus genitalia figured in Larson et al. (2000); 190c labeled as G. perplexus is more likely G. fascicollis and 190b labeled as G. fascicollis is possibly G. manitobensis, and true G. perplexus genitalia does not seem to be included in the figure. Head and pronotum yellow to reddish brown with no defined black markings (Fig. 9h); body length 10.4 to 12.4 mm; male genitalia simple with barely trifid apex (Fig. 2o-p)  Posterior black band of pronotum not reaching posterior margin, or sometimes separated from margin by a piceous-reddish area, anterior black band separated from anterior margin (Fig. 9k); male protarsus with 25-35 adhesive discs, male mesotarsus with 13-20 discs; male penis deeply trifid, invaginations separating lateral lobes from central lobe distinctly deeper than width of lateral lobes of penis apex, as in Figure 2u Anterior black band of pronotum contiguous with anterior margin (Fig. 9j); female pronotum with weak corrugated sculpture; male mesotarsus not dilated and lacking adhesive discs; male protarsal claws different in shape and size, posterior claw with sinuate ventral margin and about 2/3 length of anterior claw; male parameres very long, at least 1/4th longer than penis (Fig. 2s) .....G. occidentalis -Anterior black band of pronotum mostly separated from anterior margin by a more or less evident reddish area; female pronotum with conspicuous corrugated sculpture; male mesotarsus dilated with adhesive discs on ventral surface; male protarsal claws equal or anterior claw only slightly longer than posterior which does not have a sinuate ventral margin; male parameres shorter, not more than 1/5th longer than penis (Fig. 2m, q)  Metanepisterna ("metasternal wing") broad, width between 0.48 and 0.60 mm; female elytron at shoulder with less pronounced striolate punctures; male mesotarsus with 12 discs in two rows; central penis lobe of trifid apex much shorter than lateral lobes (Fig. 2m-n)  Posterior black band of pronotum narrow, equal to only 1/3 to 1/2 of medial yellow area, contiguous with posterior margin (Fig. 9d); epipleuron broader at level of abdominal ventrite II than at level of ventrite I, body "pear-shaped" due to posteriorly widened epipleura; male genitalia as in Figure 2g Ventral side of body mostly piceous; metatibia and metatarsus dark brown to black; female pronotum with conspicuous corrugated sculpture; anterior black band of pronotum continuous with anterior margin, in males this band is narrow and equal to about 1/3 of medial yellow band (Fig. 9a); male genitalia as in Figure 2a- Transverse black bands of pronotum contiguous with anterior and posterior margin, respectively (Fig. 9b); anterior mesotarsal claw longer than posterior claw, strongly in males weakly in females; female pronotum with weak corrugated sculpture; male mesotarsus not dilated and without discs; trifid apex of male penis very shallow, parameres very long, at least 1/4th longer than penis ( Fig. 2c-d)  Epipleura rather wide at level of abdominal ventrites I-III (Fig. 10f-h); central lobe of male trifid apex in lateral view concave (Fig. 2l) Fig. 10a-b); central lobe of male trifid apex in lateral view convex (Fig. 2x) Minimum distance between meso-and metacoxae almost same as width of metaventral process between mesocoxae (Fig. 11a); female posterior metatarsal claw about 1.8 the length of anterior claw, which is not strongly curved apically; male mesotarsus with 12 discs in two rows ventrally, formula 4-4-4; male penis apex less deeply trifid and central lobe in lateral view more abruptly raised (Fig. 2e-f) Minimum distance between meso-and metacoxae clearly less than width of metaventral process between mesocoxae (Fig. 11b); female posterior metatarsal claw about 1.5 the length of anterior claw which is strongly curved apically; male mesotarsus with 14 discs in two rows, formula 6-4-4; male penis deeply trifid and central lobe in lateral view describing a long evenly convex curve ( Fig. 2i-j)

Discussion
The within species variation in the shape and extension of the transverse black bands on the pronotum in G. zonatus (Nilsson 1986) was observed to exist in G. perplexus and G. elatus as well. Rarely, G. zonatus specimens are found where the basal black band reaches all the way to the posterior margin. Initially, we were struck by the very thin black bands in several specimens of G. elatus, and while this color pattern seems more common in G. elatus than in either G. zonatus or G. perplexus we have in some specimens found the same coloration in the latter species too. We therefore consider this character as less reliable for species diagnosis within the zonatus-species complex. We also examined the shape of the bifurcation in the chitinous ring around the male genitalia, which Wallis (1939) indicated as diagnostic. We found it to be informative, probably significantly so if the shape had been quantified and tested statistically but we observed within species variation in all three species. It has also been seen in G. zonatus that the coloration of elytra can probably be connected to the habitat e.g. darker speci-  mens have more often been found in dystrophic water (Nilsson 1986) and this might indicate that specific coloration may also be an inferior character to separate species. Instead the great interspecific variation in the male genitalia, especially at the top of the penis, is commonly used to separate species within Dytiscidae (Nilsson and Holmen 1995, Larson et al. 2000, Miller 2001, Bergsten and Miller 2006, and we found this as the most informative character (Fig. 2). The shape in lateral view of the central lobe at the penis' trifid apex was a qualitative non-overlapping character we found the most reliable to separate G. zonatus from G. perplexus/G. elatus. The male penis (and parameres) were significantly longer relative to body length and also had a more elongate shape as measured by a length to width ratio in G. elatus compared to G. perplexus. Though significant, variation was overlapping for these characters (Fig. 4). Although not quantified, the epipleura are also in general wider in G. elatus not only compared to G. zonatus, but also compared to G. perplexus (Fig. 10). Since G. elatus was described by Sharp (1882) it has mostly been treated as a synonym of different species (Horn 1883, Zimmermann 1917, Wallis 1939) rather than a valid species. Wallis (1939) was the one who concluded that its diagnostic character was not enough to retain both G. perplexus and G. elatus, but neither the size nor the shape of the male genitalia had been considered (Sharp 1882, Wallis 1939. However, when Nilsson et al. (1999) later identified five adult specimens from Kamchatka as G. perplexus, they somewhat misleadingly stated this to be the first Palearctic record of the species despite the fact that in synonymizing G. elatus with G. perplexus Wallis (1939) defined G. perplexus as a species occurring in both Nearctic and east Siberia of Russia. It was not until the compilation of the world catalogue two years later that Nilsson (2001) brought back the somewhat forgotten name G. elatus. As Nilsson's collection now is housed at the NHRS, we have re-examined all Graphoderus material from Kamchatka (Nilsson et al. 1999) and Urup, Kuril Islands (Nilsson et al. 1997) as well as material from North Sakhalin collected by Minakawa in 2002. All the males proved to have the concave shape of the central penis lobe that separates G. elatus and G. perplexus from G. zonatus (Fig. 2) and the genitalia also proved to be larger than G. perplexus, fitting into the diagnosis of G. elatus.
From our results we here propose a strict allopatric distribution of the three species in the zonatus complex. We propose that 1) G. perplexus only occurs in the Nearctic region, 2) G. elatus only occurs in the east Palearctic region, east of the Yenisei-Angara river and 3) G. zonatus occurs in the Palearctic region from central and north Europe through Turkey, Caucasus and eastwards up to the Yenisei-Angara river. It is likely that there is a contact zone where G. zonatus and G. elatus meet, but whether any hybridization occurs is unknown. All Graphoderus zonatus records from Amur, Chita, Verkhoyansk, Indigirka, Kamchatka, Sakhalin and Kuril Islands are Graphoderus elatus. Also material from Magadan, Khabarovsk and Primorsky Kray reported by Lafer (1989) is most likely G. elatus (not examined). Material we have examined from Yakutia has also been G. elatus and records referred to East Siberia are probably mostly G. elatus but could refer to G. zonatus. The material we studied from northeast Mongolia was G. elatus and the reported G. zonatus from the north-central and western part of the country (Shaverdo et al. 2008) might be misidentified G. elatus. Likewise, the newly collected material from Gangca in China was G. elatus as well as material from Heilongjiang and Inner Mongolia, but G. zonatus may occur in the more western parts Figure 11. Ventral view showing meso-and metathorax of G. bieneri (a) and G. cinereus (b). To indicate the relative distances between the mesocoxae (1) and the meso-and metacoxae (2). of China. What was thought to be G. zonatus in Japan was first discovered on Hokkaido by Matsumoto in 1986 (Mori and Kitayama 1993). We here conclude that all material from Japan referred to G. zonatus is instead G. elatus.
Our easternmost record of true G. zonatus was collected outside Irkutsk, just west of the Yenisei-Angara river. Strikingly this is also the easternmost record of several other west Palearctic aquatic beetles like Helophorus granularis (Linnaeus, 1761), H. strigifrons Thomson, 1868and H. pumilio Erichson, 1837(Angus 2011. The sister species pair in the Agabus lineatus Gebler, 1848 species group is also separated east and west of Yenisei-Angara river (Nilsson 2003b). As well, the Palearctic Colymbetes paykulli Erichson, 1837 and the Holarctic C. dahuricus Aubé, 1837 are separated in this area (Drotz et al. 2015). All other examined "G. zonatus" from the Palearctic, east of the Yenisei turned out to be G. elatus but it is important to note that the number of adhesive discs on male pro-and mesotarsus are not reliable characters to separate G. zonatus from G. elatus. In the populations of G. elatus from Urup, Inner Mongolia (leg. Li, Chunyuan and Chaoqun) and Yakutsk (18 km E. of river Lena) the females had granulated elytra and the males from Urup had a significant larger number of adhesive discs. In the material examined from Kamchatka, Hokkaido, Inner Mongolia (leg. Jia), Gangca and Heilongjiang females had smooth elytra and males a lower number of adhesive discs. Yet our examined material from North Sakhalin indicates that the two morphs can exist within the same population. However, the question arises whether these dorsal female forms could have arisen twice, completely independent of each other or if they are sister-species and inherited the polymorphism from a common ancestor. Based on the male genitalia it seems more likely that G. elatus and G. perplexus are sister-species, which would require either a loss of the polymorphism in G. perplexus or independent gains in G. zonatus and G. elatus. We are inclined to believe that a loss of polymorphism is more likely than independent gains, but phylogenetic studies (in progress) are first required to establish sister-species relationship in Graphoderus.

Conclusions
From qualitative and quantitative characters of the male genitalia the earlier synonymized name G. elatus Sharp, 1882 is reinstated as a valid species and a lectotype has been designated. We propose an allopatric distribution of species in the zonatus-species complex where G. perplexus is Nearctic, G. elatus only occurs in east Palearctic, east of the Yenisei-Angara river and G. zonatus occurs only west of the same river. All previous G. zonatus records from east of this river are misidentified G. elatus. Surprisingly, there are now two Graphoderus species with distinctly dimorphic females, G. zonatus and G. elatus. Finally, by providing male genitalia illustrations together with a new identification key to all Graphoderus species we hope to aid future identification work and taxonomic endeavors in the group.