Taxonomic revision of the genus Copelatus of Madagascar (Coleoptera, Dytiscidae, Copelatinae): the non- erichsonii group species

Abstract The genus Copelatus Erichson, 1832 (Coleoptera, Dytiscidae, Copelatinae) of Madagascar is revised in two parts. This review is restricted to the Copelatus species that have fewer than ten elytral + one submarginal stria, including all species except those of the erichsonii species group. Both morphological and molecular (mitochondrial COI) data are used in an integrative taxonomic approach. Thirteen species are recognised, of which five are described as new: Copelatus ankaratrasp. nov., Copelatus kelysp. nov., Copelatus pseudostriatussp. nov., Copelatus safiotrasp. nov. and Copelatus vokokasp. nov.Copelatus unguicularis Régimbart, 1903 and Copelatus apicalis Fairmaire, 1898 are both transferred to the genus Madaglymbus Shaverdo & Balke, 2008 (comb. nov.). Copelatus mimetesGuignot 1957 is a junior synonym of the widespread Afrotropical–Arabian Copelatus pulchellus (Klug, 1834) (syn. nov.). Copelatus marginipennis (Laporte, 1835) is reinstated (stat. nov.) as a valid species with Copelatus aldabricus Balfour-Browne, 1950 and Copelatus aldabricus var. simplex Guignot, 1952 as junior synonyms (syn. nov.). We designate lectotypes for Colymbetes marginipennis Laporte, 1835 and Copelatus obtusus Boheman, 1848. Copelatus peridinus Guignot, 1955 is recorded for Madagascar for the first time and Copelatus nodieri Régimbart, 1895 is rejected as a species present in Madagascar.


Introduction
The subfamily Copelatinae of diving beetles is a diverse group of aquatic beetles represented by eight genera: Agaporomorphus Zimmermann, 1921, Aglymbus Sharp, 1880, Capelatus Turner & Bilton, 2015, Copelatus Erichson, 1832, Exocelina Broun, 1886, Lacconectus Motschulsky, 1855, Liopterus Dejean, 1833, and Madaglymbus Shaverdo & Balke, 2008. This subfamily is relatively homogeneous morphologically, and only the single tribe Copelatini is recognised (Miller 2001;Miller andBergsten 2014, 2016). They are one of the most commonly encountered water beetles in many wet tropical and subtropical forests (Balke et al. 2004). After the revised circumscription of Aglymbus and the erection of the new genus Madaglymbus (Shaverdo et al. 2008;Miller and Bergsten 2016), this subfamily is represented by two genera on Madagascar: Copelatus and Madaglymbus. Copelatus is the most species-rich genus of Dytiscidae in the world, distributed throughout all zoogeographical regions and currently comprises 442 recognised species (Nilsson and Hájek 2018). Rocchi (1991) listed 22 species of Copelatus from Madagascar, 18 as endemic, after which only one new endemic species has been added by Pederzani & Hájek (2005) and one originally mislabelled species removed from the fauna by Balke et al. (2014).
To be able to deal with this diversity, Copelatus species have traditionally been organised in a number of species groups based on the number of elytral and submarginal striae following Sharp (1882;Zimmermann 1919Zimmermann , 1934Balfour-Browne 1939;Guignot 1961;Guéorguiev 1968;Nilsson et al. 1997;Nilsson 2001). This is today a practical and not always a phylogenetically sound division (Balke et al. 2004). In fact Balke et al. (2004), reconstructing a phylogeny of Copelatinae with mitochondrial CO1 and 16S, found the number of elytral striae to be a highly variable and homplastic character. High variability in number of striae, both intraspecifically and between closely related species, was recently documented in Antillean species (Manuel et al. 2018).
The species that occur in Madagascar fall into five of these species groups: the hydroporoides (= formerly haemorrhoidalis, no elytral striae), longicornis (3, 4 or 5 elytral striae), irinus (6 elytral + 1 submarginal stria), consors (10 elytral striae) and erichsonii (10 elytral + 1 submarginal stria) groups. About half of the described species are members of the erichsonii species group, and the other half of the remaining species groups with fewer striae. In this paper, we will treat the Copelatus species of Madagascar with fewer than 10 elytral and one submarginal stria, i.e., all the non-erichsonii group species. Again, this is a practical rather than a phylogenetic division and in fact Copelatus safiotra sp. nov., described below, we believe is most closely related to the species of the erichsonii group due to the shape of male genitalia.
This study is motivated by recent collecting efforts of aquatic beetles in Madagascar 2009-2018 in the Water Beetles of Madagascar Project. The project is a collaboration between the Swedish Museum of Natural History and the University of Antananarivo. The current study is based on this rich new material containing five new species of the irinus species group, together with type material and some additional museum accessions. We use both morphological and molecular data with a species delimitation anal-ysis, in an integrative taxonomic approach to the revision. We provide an identification key and for each species a diagnosis, description, known distribution and habitat and ecology notes. Each species is also illustrated with a dorsal habitus photograph, and ventral and lateral views of male penis and parameres.

Fieldwork
New collecting efforts of aquatic beetles were conducted mainly in National Parks, reserves and natural forests but also in degraded forests, open areas and along main roads from all parts of Madagascar, except scant from the very south (Fig. 1A). The fieldwork was supported by permits from the "Ministère de l'Environnement, de l'Ecologie et des Forêts". We collected in various habitats ranging from large rivers to small streams, cascades and hygropetric rocks, forest pools, rock pools, ditches and canals, ponds, marshes, and lakes. Copelatus can be found in diverse aquatic habitats, both running (lotic) and standing (lentic) waters. Along forest streams they are best sought for in stagnant parts such as side-pools and rock pools with dead leaves or vegetation or in residual pools in a partly dried out streambed. Leaf-choked forest pools, vegetation-rich edges of ponds and lakes and marshes are also very good habitats for Copelatus. Some species are very good fliers and readily come to light, but the proportion of light catches (using a 22W black-ringlight) in the studied material is relatively small. A surprising number can be found even in minute seeps especially if visited with head lamp at night when they are more active and swim around. Copelatus was sampled mainly using white pans and GB water nets and sieves with various sizes depending on waterbody type. GB water net is often best for larger habitats while sieves were used for sampling smaller habitats like rockpools. Specimens were collected into plastic tubes with 95% ethanol for conservation.
Each locality was given a collecting event code and associated metadata included geographic name(s), forest type, waterbody type, habitat description, eventual disturbance, collecting date and collectors. Altitude, latitude and longitude were recorded with a handheld GPS (Garmin). Each locality was also documented with photographs using compact Canon and Olympus digital cameras.

Madagascar administrative divisions
Madagascar is since the 2007 revision of the constitution divided into 22 regions as a first level administrative division, followed by districts as a second level, a recent change from the former six provinces as first level (Fig. 1B). Studied material is listed in sections preceeded by geographical divisions (former province. region: district1, dis-trict2.) to give records by both current and, the faunistically much longer used, former first level divisions.

Depositories
Most of the studied material came from the new fieldwork and is shared between the Swedish Museum of Natural History (NHRS), Parc Botanique et Zoologique de Tsimbazaza/ Madagascar Biodiversity Center (PBZT/MBC) and Department of Entomology, Antananarivo University (DEUA). Some material originated from efforts organised by the Natural History Museum in London (NHMUK) 2004-2007. Additional studied material came from earlier expeditions housed at museum collections in Museum National d'Histoire Naturelle, Paris (MNHN), Museum of Natural History, London (NHMUK), California Academy of Sciences (CAS), Naturhistorisches Museum, Vienna (NMW), and National Museum (Natural History), Prague (NMPC). Studied older types are housed in Paris (MNHN), London (NHMUK), Stockholm (NHRS) and Prague (NMPC), while holotypes of newly decribed species are housed in Stockholm (NHRS).
Depositories of studied specimens are referred to by above abbreviations (see Suppl. material 1). Long paratype series for new species with "NHRS, DEUA & PBZT/ MBC" given as depository will also be shared with other central institution collections.

Morphology
Specimens were examined under dissection microscopes from Leica (M165C and MZ12.5). Genitalia were extracted with a fine forceps or pin from the tip of the abdomen and glued onto cards on the same pin as the specimen. Dry-preserved specimens were first relaxed in warm water for 5-20 minutes before genitalia were carefully extracted. Photos of habitus were taken with a Canon EOS 5D Mark II DSLR camera equipped with a Canon MP-E 65mm 1-5X super macrolens and mounted on a motorised rail (Stackshot) from Cognisys. The system was operated using Canon EOS Utility and Zerene Stacker (Zerene Systems) softwares, the latter also used for stacking the Z-stack of captured images with the PMax or DMap algorithms. Photographs of dry-mounted genitalia were taken with a Canon EOS 7D DSLR camera mounted on a BALPRO 1 Universal bellow from Novoflex with a long working distance 10X Plano apochromatic microscope objective from Mitutoyo. The bellow was mounted on a motorised rail (Stackshot) from Cognisys and operated with the same software given above.
In describing the male penis we use the terminology suggested by Miller and Nilsson (2003).
Label data are given as written and separated by "//" if on separate labels and "|" if on different rows on the same label. Text within square brackets "[]" are our comments, explications or interpretations. Most examined specimens (individual mounted specimens, or single alcohol tubes with multiple specimens) have been given unique catalogue numbers and these are listed first, starting with "NHRS", "BMNH" (=NHMUK) or "CAS", followed by a number. A series of specimens with consecutive catalogue numbers are given as a range. The following additional abbreviations are used:

Alc.
in alcohol tube, Ex.
exemplars (number of individuals), GP (Genital Preparation) male genitalia have been examined, HT Holotype,

DNA extraction and PCR
DNA was extracted from one mesoleg or from soft abdominal tissue retrieved in association with dissection of male genitalia. The leg or soft tissue was incubated in lysis buffer at 56 °C overnight. Post-incubation protocol followed the cell and tissue DNA kit on a KingFisher Duo Prime. This system provides automated nucleic acid purification at a running time of approx. 25 minutes. We used ready-to-go beads to prepare 25 μl PCR reactions consisting of 21 μl of water, 1 μl of each primer and 2 μl of DNA template. We used the primers Jerry (F, 5'-CAA CAT TTA TTT TGA TTT TTT GG-3') and PatDyt (R, 5'-TCA TTG CAC TAA TCT GCC ATA TTA G-3') to amplify an 825 bp fragment of mitochondrial cytochrome c-oxidase subunit 1 gene (COI or cox1) (Simon et al. 1994;Isambert et al. 2011). For two dry-mounted specimens, Copelatus sp. female (Bemaraha) (CASENT-8135000) and Copelatus sp. female (Andasibe) (NHRS-JLKB000065698), we used the primers Jerry (F, 5'-CAA CAT TTA TTT TGA TTT TTT GG-3') -Hal 1450rw (R, 5'-GGA AAT CAT TGA ATA AAT CCA GCT-3'); and Hal 1450fw (F, 5'-AGC TGG ATT TAT TCA ATG ATT TCC-3') -PatDyt (R, 5'-TCA TTG CAC TAA TCT GCC ATA TTA G-3') to amplify the same segment but in two shorter sections (Wallin et al. 2017). The PCR programme started with 95 °C for 5 min, followed by 40 cycles of 95 °C for 30 s, 50 °C for 30 s, and 72 °C for 50 s. A final extension step of 72 °C for 8 min followed the 40 cycles, and reactions were then stored in the block at 12 °C. Agarose gel electrophoresis was used to verify amplifications, including a negative control.
Successful PCR products were purified using EXOSAP Clean-up mix of two enzymes (Exonuclease and Shrimp Alkaline Phosphatase) and run in a PCR machine with the programme 37 °C for 30 min followed by 80 °C for 15 min and finally 12 °C (∞). PCR products were sent to Macrogen for sequencing.
Gene regions were sequenced in both directions and sequence chromatograms were edited with SEQUENCHER version 4.10.1 (Gene Codes Corporation). The contigs were assembled from forward and reverse reads, and primer regions trimmed. New sequence data were then exported in fasta format and aligned together with Gen-Bank sequences of Copelatus from Isambert et al. (2011) in CLUSTALX using default settings. All new sequences are available in Genbank under accession numbers MK878825-MK878877 (Table 1).

Phylogenetic analysis
We first performed a Bayesian phylogenetic analysis to produce a CO1-genetree. As the taxon selection here is not aimed at producing a phylogeny, but to interprete genetic variation in light of morphological delimitations of a diagnosable set of species in a certain geographic region (Madagascar), the genetree was artificially rooted using Copelatus befasicus. We used PARTITIONFINDER Ver. 2.1.1 (Lanfear et al. 2016) to infer a suitable partitioning scheme (three codon-position specific partitions defined as input) and for each partition, a suitable model. The choice of models was limited to those available in MrBAYES, and selection was based on AICc scores. We ran 10M generations in MrBAYES Ver. 3.2.6 (Ronquist et al. 2012) to infer a non-clock CO1 genetree under selected partitioning scheme and models. Two runs, each with one cold and three heated chains were sampled every 1000 th generation. A majority-rule consensus tree was calculated from both runs after removal of 25% as burn-in from each.
To explicitly compare our morphological delimitations with a single-locus species delimitation method we implemented the GMYC-method (Fujisawa and Barraclough 2013). An ultrametric strict clock tree was calculated with BEAST Ver. 1.8.4 (Drummond et al. 2012) using the same partitioning and model scheme selected by PARTI-TIONFINDER above. Rooting was here done with the strict clock model. 2×50M gen- erations were sampled every 1000 th generation, and treeannotator was used to calculate a maximum clade credibility tree with median node heights, removing 10% as burn-in from each run. The GMYC analysis was carried out in R using the SPLITS package (Ezard et al. 2009) on this tree, under the single-threshold method. Convergence and mixing of MCMC runs from MrBAYES and BEAST were checked by the statistics provided by respective programmes and with TRACER Ver. 1.7.1 (Rambaut et al. 2018).

Molecular results
Our amplification of CO1 was successful for 53 samples which, together with sequences downloaded from GenBank, gave 77 terminals (Table 1). Alignment length was 825 bp and was gap-free but some sequences were unreadable near primer regions and therefore slightly shorter. The amplification of two shorter fragments from the degraded DNA originating from the two dry-preserved specimens was partially successful and gave 311 bp (CASENT-8135000) and 447 bp (NHRS-JLKB000065698) respectively and enabled inclusion in the phylogenetic analyses. Specimens of morphologically identified species all clustered as monophyletic in the Bayesian analysis except C. insuetus and C. kely sp. nov. (Fig. 2). These are part of a closely related group of species we will refer to as the Copelatus insuetus complex with interspecific genetic divergences ranging between 0.1-7.1%, but several not more than 2-3% (Table 2). Morphological analysis was inconclusive for some female specimens or populations in this group and these are discussed under respective species.   Result of the single-locus GMYC species delimitation using an ultrametric CO1 genetree from BEAST. Black branches indicate interspecific divergences, red branches represent intraspecific coalescence events. Values above interspecific nodes indicate posterior probability from the Beast analysis under a strict clock model. The * indicates two nodes of further splitting from the GMYC analysis that is within the confidence interval of 2Log likelihood units from the optimal solution.
The GMYC species delimitation analysis of the strict clock tree resulted in 11 separate evolutionary units that were largely but not entirely consistent with our morphological delimitation (Fig. 3). Especially, the Copelatus insuetus complex of four morphologically delimited species where merged into one unit in the GMYC analysis. But in contrast the non-named females that we could not identify morphologically were separated in the GMYC analysis and these very likely represent one or two new species of which we have yet to discover the male. In no case, had the GMYC analysis split groups of individuals that were morphologically considered the same species. The approximative 2log-likelihood confidence interval included between 5-13 units and in the case (13) closest to the morphological delimitations, the Copelatus insuetus complex was divided into two species, but also the northernmost Copelatus safiotra sp. nov. specimen was separated from geographically more southern populations. The drypreserved female from Bemaraha (CASENT-8135000) undoubtedly represents a fourth Malagasy species from the Copelatus pulchellus complex. However, the female from Andasibe (NHRS-JLKB000065698) was merged with Copelatus peridinus, despite a genetic distance of approximately 3%, and this is discussed below.

Identification key
Key to Copelatus species of Madagascar with fewer than ten discal and one submarginal elytral striae. Note that for some species males are necessary.

1
Impressed elytral striae absent (rows of points may be present) (Fig. 7A, B)  First, third, and fifth elytral striae distinctly abbreviated anteriorly; dorsal colouration largely black (Fig. 7D); penis distinctive in lateral view resembling a "pumpjack" with a long sub-erect neck and a downturned apex (Fig. 4B)  Elytra with irregular traces of intermediate striae or "pseudostriae" between first and second stria and between second and third stria; body length larger, 5.3 to 5.6 mm (Fig. 10B); penis curvature in lateral view rather weak (Fig. 4C)  Body subparallell and head with a broad interocular distance; pronotum without strioles (Fig. 10C); penis in lateral view with a distinct expansion subapically followed by a narrow apical blade (

Taxonomy
The following three taxa were described as Copelatus species from Madagascar and were listed as such by Rocchi (1991):  Guignot, 1955 is a synonym of Exocelina subjecta (Sharp, 1882) following Balke et al. (2014). Copelatus bilunatus was described by Guignot (1955a) from Zimmermann's collection based on a "♀ unique" from "Madagascar" without any further locality data. The species was assigned to Copelatus based on the presence of complete metacoxal lines and placed in the Copelatus hydroporoides species group, as it lacked impressed elytral striae. Balke et al. (2014) concluded that the specimen is identical to Exocelina subjecta (Sharp, 1882), a common species on New Caledonia, and synoymised the two assuming the specimen was mislabelled. Apparently the holotype is a male in contrast to what is stated in the original description. Balke et al. (2014) document the presence of a stout spine-like setae on the anterodistal angle of protarsomere IV, which is characteristic of the Copelatinae genus Exocelina (but also of Malagasy Madaglymbus). We have not examined the type ourselves but it is clear that in any case it is not a Copelatus species. ( (Guignot 1961;Nilsson et al. 1997;Nilsson 2001;Nilsson and Hájek 2018). We have studied a male syntype of this species which was not found by Guignot (1961: note 638) and conclude that is not a Copelatus. The syntype male has a stout spine-like setae on a protruding anterodistal corner of protarsomere IV which places it in Madaglymbus (Shaverdo et al. 2008) and we hereby transfer it to that genus: Madaglymbus apicalis (Fairmaire, 1898) comb. nov.;in fact Guéorguiev (1968: 32) suggested it might belong to Aglymbus "Peut-être c'est une Aglymbus?" [Eng. translation "Maybe it is an Aglymbus"], which was correct since this was before Madaglymbus was erected for the Madagascar species of Aglymbus (Shaverdo et al. 2008). (Régimbart, 1903), comb. nov. Régimbart, 1903:19.  Régimbart | 1908 // C. unguicularis | Type // unguicularis Rég. // Remarks. Copelatus unguicularis was described by Régimbart (1903), based on a single male from Suberbieville [= Maevatanana]. It was classified by Régimbart in the fifth species group of Sharp (1882) (the consors species group sensu Guignot 1961) based on ten discal but no submarginal elytral stria where it is still classified today (Guignot 1961;Guéorguiev 1968;Nilsson et al. 1997;Nilsson 2001;Nilsson and Hájek 2018). We have examined the male holotype and like the type of Copelatus apicalis, it has a stout spine-like setae on a protruding anterodistal corner of protarsomere IV and is not a Copelatus. In addition, the elytral striae are very irregular, more like very elongate and deep strioles and possibly not homologous to the regular impressed striae in Copelatus. We here transfer it to Madaglymbus: Madaglymbus unguicularis (Régimbart, 1903) comb. nov.

The Copelatus hydroporoides group
This group is defined by the lack of elytral striae (Sharp 1882;Guignot 1961;Guéorguiev 1968). It is certainly not a natural group. A number of species formerly placed in this group has lately been moved into other genera. Perhaps it still remains useful as a "trash can" for Copelatus species lacking elytral striae, but it is very probably not monophyletic and likely still mixed with some copelatine species that should be transferred to other genera. Guignot, 1955 Fig. 7B Copelatus baculiformis Guignot, 1955b: 193. Type locality. Madagascar, Massif Ankaratra, Manjakatompo, alt. 1700-1800 m.

Copelatus baculiformis
Type information from original description. based on a single female specimen (holotype), collected December 1951 by R. Benoist.
Description. (based on holotype ♀): Body length 4 mm. Body shape elongate oval and dark brown to blackish ferrugineous. Head uniformly dark brown ferrugineous to slightly darker posteriorly inside eyes, with thin sparse punctation. Pronotum dark brown ferrugineous, same colour medially and laterally but darker along anterior and posterior third. Disc of pronotum less densely punctuated, posterolateral corners with dense superficial strioles. The entire dorsal surface covered with a microsculpture. Elytra uniformly coloured in same dark brown to blackish ferrugineous colour as anterior and posterior parts of pronotum ( Fig. 7B). Impressed striae absent but four elytral rows of impressed points present. Innermost, a presumed first row completely lacking, row 2, 3, and 4 visible from base but row 3 less distinct than 2 and 4. Row 5 very vague and only visible posteriorly. Elytra densely covered with punctures which laterally of approximately the third row of points is replaced by strioles (probably only in the female). Appendages testaceous.
Habitat and ecology. Unknown, but according to original description collected at at an altitude of 1700-1800 m. See Hjalmarsson et al. (2013) for a description of the locality Manjakatompo and its conservation priority.
Comments. No other specimen than the female holotype is known of this species and it is a bit of a "mystery species". We have conducted fieldwork at the type locality Manjakatompo multiple times (2011, 2012, 2014, and 2016) but never found any specimens resembling this species. The species belongs in the hydroporoides species group of Copelatus, but two other Malagasy species placed in this group have turned out to be misidentified Madaglymbus or Exocelina (see above). Copelatus baculiformis was described by Guignot the same year (1955) that he described C. bilunatus, considered mislabeled (Balke et al. 2014) from the Zimmermann collection. It is certainly possible that also C. baculiformis is based on a mislabeled specimen, but in contrast to C. bilunatus the type locality data is more exact; Massif Ankaratra (mountain), Manjakatompo (locality), 1700-1800 m (altitude in meters), XII-51 (collecting month and year), and R. Benoist (collector), speaking against such a mistake. Based on general external morphology, body shape and lack of striae the species resembles C. peridinus. It is much smaller and therefore likely not conspecific, but they could be closely related, as well as with the female sequenced from Andasibe (NHRS-JLKB000065698). Guignot, 1955 Figs 4A, 7A Copelatus peridinus Guignot, 1955c: 188. Copelatus seydeli Guignot, 1958 Diagnosis. Similar to C. baculiformis based on shape and colouration, but C. peridinus is bigger with body length between 5.7 and 6.6 mm (Fig. 7A). Elytra lacking deeply impressed striae, but elytral rows of impressed points present out of which two rows are most distinct. Penis in lateral view curved with two points at which it curves slightly more abruptly creating two gentle angles (Fig. 4A); parameres rather broad at base and evenly curved (Fig. 4A).

Copelatus peridinus
Description. Body length 5.7-6.6 mm. Body shape elongate oval, from midpoint uniformly tapering towards a rather pointed apex. Head and pronotum both in the same dark brown to blackish ferrugineous colouration; elytra anteriorly in the same colour as pronotum, but posterior part brown ferrugineous with largely testaceous lateral margins (Fig. 7A). Head, pronotum and elytra with thin dense punctation, in addition pronotum and elytra with small and shallow punctures, and microsculptured. At posterolateral corners of pronotum, punctures almost a little joined and corrugated. Elytra lacking deeply impressed striae. Two continuous rows of punctures are most obvious and divide the elytra in three more or less equal intervals. A third and a fourth row of punctures also present, albeit less distinct and more fragmented. These occur between the two first distinct rows and between the second distinct row and the elytral margin. A fifth row suggested between the two outermost rows by a few spaced out punctures, almost unidentifiable in most Malagasy speci-mens seen but more distinct in specimens seen from mainland Africa. Appendages testaceous to rufus.
Ventral side dark ferrugineous. Prosternal process strongly carinate anteriorly and with a rather short process. Lateral parts of the metaventrite medium-broad. Metacoxal lines short, anteriorly diverging and abbreviated well before metaventral margin. Metacoxa with short fine strioles continuing onto abdominal ventrites.
Male: first three pro-and mesotarsomeres widened. Protibia modified, narrow at base and with an early abrupt bend, extended and broadened towards middle with a straight ventral side but angled dorsal side. Pro-and mesotarsal claws unmodified.
Penis in ventral view with apical part more or less straight and even, gently pointed at apex but with the very apex minutely twisted to the right (Fig. 4A). Penis in lateral view curved with two points (1/3 rd and 2/3 rd from base) at which it curves slightly more abruptly creating two gentle angles (Fig. 4A). Parameres rather broad at base and evenly curved and tapering towards apex (Fig. 4A).
Female: dorsal sculpture similar to male. Distribution. Likely a more widespread distribution in at least central and eastern continental Africa than the current records (Lubumbashi and Kivu in DR Congo) indicate. In Madagascar, only know from the eastern central parts: Betongolo (Antananarivo), the Analamazaotra NP, and P.K.57 Route d'Anosibe [RN23] (Fig. 11C).
Habitat and ecology. A series of six specimens collected with light trap ("piège lumineux") in the capital Antananarivo 1946, indicates flight capacity and anthropogenically disturbed habitats. All records from DR Congo are also from light trap catches (Bilardo 1982). We collected a second series of teneral specimens in a shallow forest pool with vegetation, near the entrance to the Analamazaotra NP.
Comments. This species was described from Lubumbashi, DR Congo, and has not been recorded from Madagascar before. Earliest record found is from November 1946. It seems to be a dispersive good flier and often collected at light so its presence in Madagascar is therefore not surprising. However, it is not widespread in Madagascar as far as we know. In fact, the known distribution is restricted to the surroundings of the capital and east of the capital along the main national route towards Toamasina, which could suggest a recent incidental human-mediated introduction from mainland Africa.
Note that it may be that this is a species with intraspecific variation with regards to elytral striation, ranging from five puncture lines out of which two are more distinct, to five weakly impressed striae (see further discussion under Copelatus sp. 2 below).

The Copelatus longicornis group
Sharp (1882) defines this group as those having 3, 4, or 5 discal striae on elytra but lacking a submarginal stria. There is only one species in Madagascar from this group, Copelatus befasicus Guignot, 1956, and it has five discal striae.
Type information from original description. based on an unknown number of female type specimens but holotype and paratypes are distinguished in introduction. Diagnosis. Similar to C. insuetus and related species in habitus by being small, elongate, and subparalell, but C. befasicus is distinguished from C. insuetus and from all other Malagasy Copelatus species by the presence of only five elytral striae, and without submarginal striae. In addition, the first stria is shortened, present only in the posterior third ( Fig. 10D).
Description. Body length 4.1-4.2 mm. Body shape elongate and subparallel, dorsal surface reddish brown with a lighter elytral base. Head and pronotum uniformly rufus brown. Head, pronotum and elytra with dense punctation. Elytra and pronotum covered with dense punctures and the whole dorsal surface with a microsculpture. Lateral sides of pronotum striolate with the widest striolate area in the posterior corners. Elytra light brown with a distinct testaceous band basally (Fig. 10D). First elytral stria shortened and present only in posterior third. Second to fifth elytral stria starting more or less at base and all striae approaching the apex of elytron, but the second and fourth a little shorter. Submarginal striae absent. Appendages testaceous.
Ventral side light brown. Metacoxa and abdominal ventrites punctate and striolate. Prosternal process rather short and medially raised, triangular in cross-section. Lateral parts of metaventrite medium broad. Metacoxal lines anteriorly diverging but rather weakly so, abbreviated well before metaventral margin. Antennae, palps and legs testaceous.
Male: unknown. Distribution. Known only from two localities in the western part of Madagascar, the deciduous forest south of Befasy, Morondava, and at one locality between Morafenobe and Beravina village (Fig. 12D).

Habitat and ecology.
Paulian collected the species in 1956 in the western dry deciduous forest south of Befasy, Morondava. We rediscovered the species in 2009, when we found one female specimen of C. befasicus also in the western part but a bit further north than the type locality, along the road between Morafenobe and Beravina village at an altitude of 290 m. This locality consisted of dry savannah with mixed wood and grassland ecosystem after deforestation. The habitat consisted of muddy/sandy residual pools with some dead leaves in a temporary stream after the rainy season. The dry deciduous forest ecosystem in western Madagascar has suffered immensely from deforestation and very little of this habitat remains (Ganzhorn et al. 2001;Whitehurst et al. 2009). The species is rare and likely threatened due to the disappearance of western deciduous forests in Madagascar. That all four known specimens are females may suggest an uneven sex ratio as a test of equal sex ratio is marginally non-significant (p = 0.0625) if considered randomly picked from the population.
Comments. This is the only species in the Copelatus longicornis group from Madagascar. The longicornis species group currently contains 38 species distributed mainly in the Neotropical and Afrotropical regions but also with few species present in Japan, New Guinea, and Fiji Islands (Nilsson and Hájek 2018). The group as currently defined is certainly artificial from a phylogenetic perspective and the only character they have in common is the low number of elytral striae.
Lateral margin of pronotum rusty ferrugineous, with short sparse strioles. Pronotum with puncture rows and microsculpture. Elytra narrowly testaceous to ferrugi-neous posterolaterally. Six elytral striae present and one submarginal stria (Fig. 7D). Striae 2, 4, and 6 longer than striae 1, 3, and 5, at least as continued fragments; first stria shortest, present only in posterior fourth to posterior third. Second and third striae present as clearly impressed striae in posterior half or a little longer, but second stria continues, albeit fragmentary and less impressed, more distinctly all the way to base. Fourth stria the most complete, more or less present to base but less deeply impressed or fragmentary in anterior third. Fifth stria abbreviated a little less than third stria, present in approximately posterior two thirds, and like stria 3 has small isolated point remnants of stria more anteriorly. Sixth stria longer than fifth and with more distinct remnants of striae present anteriorly. Elytra covered with punctures in apical part.
Male: protibia slightly widened at apex, somewhat curved and angulate basally. Penis in lateral view curved with two points where curvature is more abrupt, constricted before apex at a narrow "neck" and expanding to apical part (Fig. 4B). As a whole in lateral view, the penis bears resemblance to the silhouette of a "pumpjack". Penis apex in ventral view left-turned. Parameres broad at base, curved and tapering towards apex, apical part with numerous setae on ventral margin (Fig. 4B).
Female: dorsal sculpture similar to male. Distribution. Occurs on Madagascar and Mauritius (Gomy 2016). Guignot (1948) tentatively refer a single female specimen from Lulua in DR Congo to C. distinguendus but this was very likely a different species. In Madagascar, it has been recorded from the provinces of Antananarivo, Fianarantsoa Toamasina, and Antsiranana; more specifically from Anjozorobe Angavo reserve, around Ankazobe, Manankazo, Andranofeno Sud, around Ambohitantely reserve, Manjakatompo Ankaratra reserve; also from the eastern central area around Moramanga, Analamazaotra NP; in the southeast from Ambositra in col de Tapia, Ankazomivady forest, Ranomafana NP, Andringitra NP; and from the northeast in Anjanaharibe Sud reserve (Fig. 11D).
Habitat and ecology. This species has been collected in various localities, mostly from open, partly deforested areas at mid-to high altitudes or open forest marshes. It occurs at altitudes above 900 m and is often associated with grass vegetation along lake shores and in marshes, found by stamping, and at vegetation-rich margins of rivers.
Comments. Régimbart (1895) described this species under the name Copelatus duodecimstriatus Aubé but realised later (1903) that what he had described from Madagascar was not conspecific with Aubé's C. duodecimstriatus from the Mascarene islands and he gave it the new name C. distinguendus, referring to his description from 1895. Copelatus duodecimstriatus lacks submarginal striae, has a distinctly different penis shape, and is endemic to Mauritius (Vinson 1967;Guignot 1961;Gomy 2016). The closest relatives of Copelatus distinguendus are likely a group of species on continental Africa with similar genitalia (e.g., C. ateles Guignot, 1955, C. sylvaticus Guignot, 1952, C. andreinii Régimbart, 1905, and additional species described by Bilardo and Rocchi 2013). These have a variable number of elytral striae from five discal to ten discal plus one submarginal, but similar "pumpjack"-shaped genitalia (Bilardo and Rocchi 2013). Copelatus distinguendus differs from these especially by the lack of a lateral tooth on the male penis in ventral view. (  Diagnosis. Similar to C. marginipennis (Laporte, 1835) and C. mahajanga in overall habitus, but body shape more like the elongated shape of C. mahajanga. Copelatus pulchellus on Madagascar is ferrugineous black in overall colouration with or without a rather narrow or vague testaceous band basally on elytra (Fig. 8D). Instead testaceous regions on elytra are limited to posterolateral and apical parts. The first elytral stria almost full length. Penis in lateral view rather abruptly curved past middle so that basal and apical parts are approximately at right angles (Fig. 5C). Penis overall simpler compared to that of C. marginipennis and C. mahajanga and apex slightly twisted.
Pronotum dark brown to ferrugineous black with testaceous anterolateral corners. Disc covered with fine microsculpture forming regular cells and regularily spread small punctures of about same size as cells. Punctuation becomes coarser in posterolateral corners with a weak tendency to corrugate.
Elytra predominantely dark brown to ferrugineous black on disc and along striae with or without a rather narrow and vague basal testaceous band (Fig. 8D). Outer intervals and apical region testaceous to varying degrees, often a testaceous interval 4 stands out. Elytral surface covered with same type of microsculpture and punctures as on pronotum. Six clearly impressed elytral striae present on disc and one submarginal stria: second to sixth stria starting at base, first stria somewhat abbreviated anteriorly (by approximately 1/8 th the length of elytra); submarginal stria starting at approx. middle.
Female: elytral striolation limited to the medial parts of the outer three elytral intervals in the single female studied from Madagascar. From other parts of the distribution a female form is known that has the entire elytra striolated (Sharp 1882;Hájek and Reiter 2014).
Male: protibia bisinuate and angled at base, distally expanded. Penis thin, strongly angled at middle in lateral view, and apex somewhat twisted to the left in ventral view (Fig. 5C). Parameres more narrowly triangular apically compared to those in C. marginipennis and C. mahajanga.
Distribution. As the species C. pulchellus is currently interpreted, this is a very widely distributed Afrotropical and Middle Eastern species. Balfour-Browne (1950) recorded the species from Senegal in the west, Tanzania in the east, South Africa in the south, and Mesopotamia (Iraq and adjacent regions) in the Middle East. Hájek and Reiter (2014) recorded the species from Yemen and stated that the likely distribution in the Middle East included the entire Arabian Peninsula. Guignot (1961) also gives India and Ceylon (= Sri Lanka) but this was likely based on the misidentification of closely related oriental species (see Ghosh and Nilsson 2012;Sheth et al. 2018). Copelatus pulchellus was not previously recorded from Madagascar following the revision by Balfour-Browne (1950) but as we synonymise C. mimetes with C. pulchellus, Madagascar now forms part of the distribution. From Madagascar we have seen specimens from Antsiranana, Mahavavy Kinkony Reserve, Tsingy de Bemaraha National Park (Bekopaka), and from Lambomakandro, Sakaraha (Fig. 11C). It can likely show up anywhere in lowland Madagascar, but especially in the western lowlands.
Habitat and ecology. On Madagascar we have collected the species associated with a small forest stream with sidepools in a karstic limestone area ("tsingy") and in a muddy stagnant pool in a dried-out river bed. Both localities are in dry deciduous forests of lowland western Madagascar.
Comment. Copelatus pulchellus forms part of a diverse species group with many externally very similar species. Balfour-Browne (1950) admitted that his previous treatment of the species (Balfour-Browne 1939) was entirely wrong as he had then not studied the male genitalia. When he did so in 1950 it resulted in the description of several new species previously lumped under C. pulchellus.
Copelatus pulchellus is now interpreted as a widespread Afrotropical and Middle Eastern species with the male penis similar to that illustrated in Figure 5C. Dorsal colouration is interpreted as very variable, even consisting of several distinct colour forms such as the darker forms described as C. obtusus Boheman and later as C. africanus Sharp (compare figure 24 with photographs in Perissinotto et al. (2016: fig. 24) and Hájek and Reiter (2014: fig. 30)). Females are also interpreted as variable in the striolation pattern on the elytra. Sharp described a female form from Mesopotamia that was entirely striolated over the elytra under the name C. strigulosus, which is interpreted today as intraspecific variation of C. pulchellus (see photograph in Hájek and Reiter (2014: fig. 30)). We have seen three males and one female from Madagascar. All males are of the darker colour form, lacking a basal transverse testaceous band on elytra, similar to Boheman's C. obtusus. The female has a weak basal testaceous band but is otherwise also most similar to the dark colour form. Striolation is limited to the middle of the outer three intervals. We have no doubt that the three non-type specimens examined are conspecific with C. mimetes Guignot. But after our comparison of the male genitalia of the type with C. pulchellus specimens from mainland Africa, we concluded that C. mimetes cannot be upheld as a separate species given how C. pulchellus is interpreted. We noted very minor differences in the penis apex which are not greater than differences between east, south, and west African specimens of C. pulchellus. The penis figured by Guignot (1957: fig. 2, reproduced by Nilsson et al. 1997: fig. 8) is very inaccurate, depicting a shape quite different from C. pulchellus but that does not correspond with the holotype. Copelatus pulchellus may still be a species complex for future studies to solve but currently it is a widespread and variable species distributed over Africa, Madagascar, and the Middle East.
Copelatus basalis Boheman, 1848 was synonymised with C. pulchellus by Omer- Cooper (1965). We have studied the same type material as Omer-Cooper and agree with this conclusion. However, we are not convinced that the material housed at NHRS as these types are the correctly identified types of the name C. basalis. Boheman (1848) describes in Latin four species from Johan August Wahlbergs collectings in Caffraria (South Africa) of which three are new, numbered 259 -Copelatus pulchellus Klug; 260 -Copelatus obtusus; 261 -Copelatus striatellus; and 262 -Copelatus basalis. For C. pulchellus the disc of elytra is described as having six striae. For the longer description of C. obtusus this is further elaborated to detail that there are six discal stria but entire elytra has seven striae "septemstriata" (six discal and one submarginal). This pattern and number of striae is consistent with the Copelatus pulchellus species group. The following species is correctly identified as C. striatellus with nine discal striae, and is clearly stated as such: "disco striis 9 tenuibus", of which the innermost is much abbreviated. This description matches very well with the types preserved at NHRS. Finally, C. basalis is described as the last Copelatus species in Boheman's work and elytra is described as "12-striata". The supposed types at NHRS for both C. obtusus and C. basalis have six discal and one submarginal striae. The type for C. obtusus matches the original description of "septem-striata" but the three types of C. basalis with the same number of striae does not match the original description of "12-striata". It would be very inconsistent of Boheman to describe the total number of striae on one elytron in one case "elytra…septem-striata", and in the other only the discal striae but summing up the number from both elytral halves "elytra…12-striata". It would also be very illogical to place C. basalis after the 9-striated C. striatellus if it has the same number of striae as C. pulchellus and C. obtusus, which come first. Similarily, Guignot (1961) was confused about Boheman's C. basalis and listed the name both under the 12-striated C. mocquerysi Régimbart with a question mark, and under C. pulchellus. We consider the status of C. basalis Boheman as uncertain but we have not found any alternative potential type material at NHRS. We designate the single undoubted syntype of C. obtusus Boheman, 1848 in the NHRS collection as lectotype to preserve the stability of the name. ( Type information from original description. housed in Buquet collection and originating from Senegal; of aldabricus: based on male (holotype), J.C. Fryer collection, collected 1908-9 from Aldabra; of simplex: based on male and female syntypes from Madagascar without further locality data. Diagnosis. Copelatus marginipennis is distinguished from all other Malagasy Copelatus except C. pulchellus and C. mahajanga by the presence of six discal elytral striae and a broadly oval body shape. Copelatus marginipennis is most easily separated by the distinct shape of the male penis in lateral view (Fig. 5A). Copelatus pulchellus (from Madagascar) lacks a transverse testaceous band basally on the elytra and females of C. marginipennis can usually be distinguished from C. mahajanga by the normally narrower testaceous band, a longer first elytral stria, and a broader body shape.

Copelatus marginipennis
Description. Body length 5.2-6.6 mm. Body shape oval, rather convex and attenuate posteriorly, dark brown to brown ferrugineous. Head, pronotum and elytra in the same dark brown ferrugineous, covered with fine dense punctation. Lateral sides of pronotum more brownish, with short strioles and the widest striolate area in the posterior corners. Elytra dark brown to brown ferrugineous, with a testaceous transverse band at base (Fig. 8A). Six clearly impressed elytral striae present and one submarginal: first stria abbreviated and slightly shorter, starting at approx. 1/7 th posterior of elytral base, second to sixth full length, and submarginal stria starting at approx. the middle; all striae approaching the apex of elytron except the sixth, which is abbreviated posteriorly and a little shorter. Sometimes, the first and second elytral striae, like fifth and sixth, unite posteriorly. Elytral surface covered with dense punctures.
Ventral side brownish to ferrugineous, metacoxa with microsculpture, densely and finely punctate. Metacoxa and abdominal ventrites striolate. Prosternal process rather short and spear-shaped, medially only weakly raised and rounded. Lateral parts of metaventrites rather broad. Metacoxal lines short and rather strongly diverging anteriorly. Antennae and palps both in the same brown colour. Pro-and mesothoracic legs brown ferrugineous. Metathoracic legs dark brown ferrugineous.
Male: Protibia strongly angled basally and expanded in apical two thirds. Penis in ventral view with a small preapical tooth on right side; in lateral view very characteristic with a subbasal dorsal knob, and post-middle with a deep ventral invagination (Fig. 5A).
Females from Madagascar usually with elytral striolation rather weak and restricted to outer intervals, but rarely the elytra are entirely and distincly striolate. Females on average smaller than males.
Distribution. Endemic to the western Indian Ocean islands as far as is modernly known, but the 1835 type locality of the original description and labels read "Senegal", which indicates either mislabeling or that the species in fact also occurs on continental Africa. Known from Madagascar, Reunion, Comores, Seychelles, and Aldabra Island (Gomy 2016). Widespread in Madagascar and recorded from Montagne d'Ambre NP and Masoala NP in the north and northeast to Kirindy in the southwest (Fig. 11A), but apparently lacking from most of the central plateau. It can probably show up anywhere around the island below 1000 m in altitude.
Habitat and ecology. This is a lowland species that seems to be most common in the deciduous western parts of Madagascar. It was common in the newly designated protected areas of Mahavavy Kinkony when we visited it in 2009. In the deciduous forest biome it has been recorded from Kirindy in the south to Ankarana NP in the north. The species seems to be a generalist and can as well show up on lowland humid east coast and at midaltitudes up to at least 1000 m. It is an apt flier collected with light traps and often in very temporary and small shallow pools including water-filled wheel tracks. Found in all kinds of temporary pools, as well as in streams and in residual pools in dried-up riverbeds. It occurred sympatrically with C. mahajanga in Betsabora river near Moramanga and at a 22W black light trap by a forest pool in Ankarafantsika NP.
Comments. Copelatus aldabricus was described by J. Balfour-Browne (1950) from Aldabra Island in the Seychelles. Guignot (1952) named a non-striolated female variety from Madagascar as var. simplex, and again in 1961 as var. aequabilis (non-available infrasubspecific name) from Andranofotsy (E of Maroantsetra NE Madagascar), and stated (1961) that females of C. aldabricus are normally striolated, but that on Madagascar the non-striolated var. aequabilis is most common or even ubiquitous. We have found at least one female on Madagascar (from Masoala NP) with the elytra entirely striolated (the "normal" variety sensu Guignot 1961), but indeed the form where striolation is rather weak and only at outer elytral intervals is most common. Guignot's (1961) distinction between female forms either having or lacking strioles is, however, a simplification. All females we have seen have at least some weak tendencies of striolation on outer intervals and the extent can vary significantly between individuals.
Copelatus marginipennis (Laporte, 1835) has been treated as a junior synonym of Copelatus pulchellus Klug, 1834 since Aubé (1838). We discovered the type material of Laporte´s C. marginipennis in MNHN, Paris. This could be identified based on a folded blue label bearing the information "Copelatus marginipennis Buquet, pulchellus var Aubé, h. in Senegal, D. Buquet", which fits perfectly the description by Laporte (1835). The genitalia of this male show without any doubt that it is a synonym of the younger name C. aldabricus and not of C. pulchellus. The name marginipennis has to our knowledge not been used as a valid name after 1899, fulfilling the first of the two criteria for protecting a younger currently used synonym from an older available name (ICZN 23.9.1.1). However, Copelatus aldabricus is a relatively recent name from a region with a low-intensity taxonomic research and we do not believe it fulfils the second criteria of at least 25 papers published by at least ten authors in the immediately preceding 50 years and encompassing a span of not less than ten years (ICZN 23.9.1.2). Therefore it is not justifiable to protect the younger synonym C. aldabricus and the older C. marginipennis is therefore brought back as the valid name of this species. We designate the discovered syntype of Colymbetes marginipennis Laporte, 1835 in the MNHN collection as lectotype to preserve stability of the name. At the same time as this type specimen was discovered, also a likely "type" specimen of the nomen nudum Copelatus fimbriolatus Dejean, 1837 from "Ile de France" [= Mauritius] was found with the same type of blue folded label (see supplementary file 1: NHRS-JLKB000065420). This nomen nudum is currently considered a synonym of C. pulchellus Klug (Nilsson and Hajek 2018). Unfortunately, it is a female specimen, which is why its identity cannot be established with certainty, but it is more likely identical with C. marginipennis since C. pulchellus is not yet known from any of the Mascarene Islands (Gomy 2016  Diagnosis. Close to C. marginipennis in habitus appearance but body shape more elongate. On average C. mahajanga have a broader transverse testaceous band at base of elytra (Fig. 8B), and the first elytral stria is more abbreviated anteriorly, starting at approx. one quarter to one half posterior of base. Penis shape is unique in C. mahajanga, bisinuate in ventral view with an expanded apex, not resembling that of any other Copelatus species in the irinus group from Madagascar.
Description. Body length 5.2-6.2 mm. Body shape oblong oval, rather convex, brown to dark brown. Head ferrugineous brown, paler in front, sometimes darker around eyes, finely microreticulate and punctate, two shallow depressions between eyes. Pronotum dark brown, paler at sides. Dorsal surface with fine microsculpture and scattered punctures, lateral sides of pronotum striolate with the widest striolate area in the posterior corners. Elytra brown, paler at sides and at apex, with a broad testaceous transverse band at base (Fig. 8B). Elytral surface finely reticulate and punctate. Six clearly impressed elytral striae present and one fairly long submarginal stria; first stria abbreviated anteriorly, starting at approx. one quarter to one half posterior of elytral base; other striae (second to sixth) starting more or less at base; submarginal stria starting around middle; striae 1-5 approaching apex of the elytron, but the second and fourth are a little shorter; the sixth stria much abbreviated posteriorly and sometimes the sixth unites with fifth apically approx. 1/5 from apex.
Ventral side ferrugineous brown, metacoxa and abdominal ventrites striolate and punctate. Prosternal process similar to C. marginipennis but lateral parts of metaventrite slightly narrower and metacoxal lines less divergent anteriorly. Antennae, palps, pro-and mesothoracic legs brown, but metathoracic legs ferrugineous dark brown.
Male: protibia modified, widened in front, strongly angled after base, with several long spines on the outer side of distal half. Tarsomeres I-III of pro-and mesolegs enlarged, with pads of numerous setae. Penis unique: in ventral view with a bisinuate shape and a widening asymmetrical apex; in lateral view abruptly curved near middle, with a right angle of more or less 90 degrees, the basal part robust (Fig. 5B). Parameres broad (Fig. 5B).
Female: similar but smaller than the male, legs not modified.
Distribution. Endemic to Madagascar. Known from several places in Mahajanga province and near Moramanga (Fig. 11B). We have seen material from around Mahajanga (Type series), Mahavavy Kinkony Reserve, Namoroka NP, Tsingy de Bemaraha NP, Ankarafantsika NP, Kamoro River south of Ankarafantsika NP and collected the species both in 2011 and 2014 at Betsabora River 6 km W of Moramanga which expanded the known distribution significantly. Note that the name Betsabora is uncertain and recovered from locals on site for this river which is a tributary to Mangoro River. Ampanihifana may be the correct name.
Habitat and ecology. Type series probably collected by light trap (Pederzani and Hájek 2005), and we collected it with black light by a leaf-choked forest pool in Ankarafantsika NP. Also collected with pitfall traps and from leaf-litter in Namoroka NP before the rainy season in November, indicating the species can tolerate periods when the habitats dry up without immediately seeking new water. We collected the largest series at Betsabora River at an elevation of 940 m. This small river runs through a savannah type area with mixed woodland-grassland ecosystem subsequent to degradation and deforestation. The river site consisted of grassy, vegetation-rich margins, bays and side pools, further described in Michat et al. (2017). Around the Mahajanga area all sites are lowaltitude, 100 m or lower.
Comments. Copelatus mahajanga was the most recently described Copelatus species from Madagascar (Peredzani and Hájek 2005) and was previously only known from the type series and lowland west type locality "Mahajanga env." without further details. The discovery of the species in Betsabora River at mid-altitude in the east indicates that the species may have a much wider distribution and ecological niche. As the species is very similar to C. marginipennis habitually, it might be misidentified as such in collections if male genitalia were not examined. Guignot, 1941 Figs 6A, 9A Copelatus insuetus Guignot, 1941: 39. Diagnosis. Body shape elongate oval, elytra brown with basal testaceous band, with six discal and one submarginal stria (Fig. 9A). Copelatus insuetus can be distinguished from the following species based on the penis ventral outline in lateral view with a distinct "shoulder" (Fig. 6A), an apex in left lateral view with a dorsal ridge crossing posterior dorsal margin and medial part longitudinally coarsely sulcate ventrolaterally in right lateral view (Fig. 6A).

Copelatus insuetus
Description. Body length 4.3-5.0 mm. Body shape elongate, weakly oval, some specimens appearing subparalell. Head rufotestaceous anteriorly and posteriorly but infuscated by two blotches inbetween the eyes. Blotches usually meet in middle and form an M-mark, but sometimes infuscation reduced to nearly absent. Pronotum usually with extensive medial infuscation leaving only lateral, anterolateral and posterolateral areas testaceous. Some specimens with less extensive medial infuscation (Fig. 9A). Elytra brown with an irregular testaceous band at base. Testaceous band with longest projection posteriorly in second interval. All appendages testaceous.
Elytra with six clearly impressed discal and one submarginal stria. Fifth and sixth striae abbreviated anteriorly, and also first stria may be somewhat abbreviated. Submarginal stria starting between one third from the base and midway between base and apex. Head, pronotum, and elytra microreticulate and finely micropunctate. Posterolateral corners of pronotum with strioles that extend along posterior surface, reduced or absent medially.
Ventral side ferrugineous brown, except prothorax, epipleura, appendages, and gula of head which are testaceous. Area around metacoxal lines lighter (not all individuals) and also spots laterally on abdominal sternites II-VII and along some sternite margins are lighter. Metacoxa and abdominal sternites II-IV marked with strioles. Prosternal process short, broad, with blunt apex. Lateral parts of metaventrite broad. Metacoxal lines long, abbreviated only slightly before metaventral margin, diverging anteriorly.
Male: first three pro-and mesotarsomeres widened and ventrally equipped with suction cups; number of suction cups per segment for I-III: 7:4:4 for both pro-and mesotarsus. Protibia modified, narrow with a bisinuate angulate ventral margin at base, broadened distally. Pro-and mesotarsal claws unmodified. Penis in ventral view thin and simple, apical part slightly left-turned. Penis in lateral view angled after basal third forming a "shoulder". Apex in left lateral view with a characteristic dorsal ridge crossing posterior dorsal margin (Fig. 6A). Penis serrated with fine transverse ridges preapically on the left ventral ridge and halfway down the left side. Penis medially strongly longitudinally sulcate ventrolaterally in right lateral view (Fig. 6A). Parameres as in Figure 6A.
Female: anterior half of elytra finely striolated from second or third elytral interval to external margin. Degree of elytral striolation in females quite variable between specimens. These strioles are finer than the strioles on pronotum found in both sexes.
Distribution. Endemic to Madagascar. Known with certainty from the eastern escarpment, Ranomafana NP in the south, Analamazaotra NP, and Mantadia NP in the central region, and Zahamena NP and Andilamena further north (Fig. 12A). We have also seen older material including males from Paris museum from lowland Lokobe NP (island of Nosy Be) in NW and Maroantsetra in NE (not reconfirmed) that correspond to the same species. Finally, two females (NHRS) from Ankarafantsika NP is likely this species.
Habitat and ecology. We have collected this species in Analamazaotra NP in stagnant forest pools with vegetation or with plentiful of dead leaves. Also found in a sidepool next to a river in semi-open degraded area near Moramanga. The eastern escarpment localities range in altitude between 900-1300 m, but if the localities of Nosy Be and Maroantsetra are correct this species can also occur at lowland sea level altitudes. It seems to be most abundant in tropical eastern forests, but if the Ankarafantsika females belong to this species, then it can also occur in western deciduous forests.
Comments. Eventual older records of C. insuetus should be regarded with caution in light of the habitually very similar new species described below. Diagnosis. Similar body shape to C. insuetus but C. kely is smaller than all other species in the Copelatus insuetus complex on Madagascar; body length 3.8-4.4 mm (Fig.  9C). The penis is most similar to that of C. insuetus but the "shoulder" in lateral view is less distinct; there is no crossing of a dorsal ridge against the posterior dorsal margin at apex in left lateral view and longitudinal ventrolateral sulcation is lacking (Fig. 6B).

Copelatus kely
Description. Body length 3.8-4.3 mm. Body shape elongate oval to subparallel. Head rufo-testaceous and only vaguely infuscated medially and around eyes. Pronotum largely rufotestaceous with only faint infuscation medially. Lateral margins lighter testaceous but pronotum also medially lighter than elytra. This gives the habitus appearance of a lighter more rufous anterior part of body contrasting with brown elytra. Elytra brown with a testaceous band basally. Testaceous band narrower and with less of a tendancy to be extended posteriorly in second interval (Fig. 9C) compared with C. insuetus. All appendages testaceous.
Elytra with six impressed discal and one submarginal striae. First to fourth striae full length or first stria slightly abbreviated at base; fifth and sixth striae abbreviated anteriorly; submarginal stria starting approx. one third to one half from the base. Head, pronotum, and elytra microreticulate and finely micropunctate. Striolation of pronotum rather restricted and present only in posterolateral corners and somewhat inwards along posterior margin but not posteromedially.
Ventral side largely testaceous to faintly infuscated, similar to C. insuetus but lighter. Prosternal process and lateral parts of metaventrite similar to C. insuetus but metacoxal lines less strongly diverging anteriorly.
Male: first three pro-and mesotarsomeres widened, but less so than in C. insuetus, and ventrally equipped with suction cups (same constellation as in C. insuetus). Protibia modified, bisinuate and angled basally and broadened distally. Pro-and mesotarsal claws unmodified. Penis thin and simple, in ventral view with apical part slightly leftturned; in lateral view slightly angled after basal third giving a suggested "shoulder" (Fig. 6B) but which is less distinct compared with C. insuetus. Apex in left lateral view without a dorsal ridge crossing posterior dorsal margin, but finely serrated preapically (only visible at high magnification). Parameres as in Figure 6B.
Female: with very weak, faint and dispersed strioles on anterior half of elytra from third or fourth interval to the lateral margin.

Etymology.
The new species is named after the Malagasy word for small, "kely", referring to the small body size. It is the smallest species so far known from the C. insuetus species complex on Madagascar. It is a non-latinised adjective.
Distribution. Known from the eastern central part of Madagascar, at Ambohidray Reserve and in Analamazaotra NP (Fig. 12B).
Habitat and ecology. This species occurs in the eastern central rainforest. Most specimens were collected from small waterfilled pitfalltrap holes for Mantella frogs, but we've also found it in muddy, stagnant, forest pools with dead leaves. The altitude of known localities ranges from 930 to 1050 m. At Analamazaotra NP, we found one specimen occurring sympatrically with Copelatus insuetus at the same locality.
Comments. This species may be endemic to a very limited area and apart from one specimen found in Analamazaotra NP, we collected the remaining series from the Ambohidray reserve. Most specimens were in fact found in water-filled pitfall trap holes set for a microendemic Mantella frog species. Copelatus kely adds to the importance of this reserve for conservation of rare and microendemic eastern rainforest species. Ambohidray reserve was established in 2013 and managed through collaboration between the local people association (VOI MMA) and Antananarivo University. During fieldwork at the reserve in April 2018 however, we observed worrying signs of disturbances; "tavy" or slash and burn of the forest for agriculture, the cutting of woods for charcoal, signs of zebu-cattle along forest paths. These factors could cause a serious threat to the aquatic insect fauna of the reserve. The reserve of Ambohidray harbours some species not known from anywere else on Madagascar. If the reserve has any ambition to serve as a refugium for these species, activities destroying or degrading the forests or aquatic habitats should be avoided. Copelatus kely is very close to C. insuetus, and the two species were not reciprocally monophyletic in the CO1 gene tree (Fig. 2). Diagnosis. Body shape slightly shorter, less elongate and slightly more oval than C. insuetus, and eyes smaller. Elytral striae more deeply impressed and intervals therefore slightly more convex (Fig. 9B). Penis shape in lateral view gently curved from base to apex (Fig. 6C), lacking the defined "shoulder" of C. insuetus and C. kely. Penis more or less straight in ventral view, which separates the species from all other in the insuetus complex.

Copelatus vokoka
Description. Body length 3.9-4.5 mm. Body shape elongate oval, but slightly less elongate compared with C. insuetus. Head rufotestaceous with only a faint suggesstion of an M-shaped infuscation between eyes. Pronotum rufotestaceous as the head, with weak medial infuscation Elytra dark brown, with testaceous band at base. Testaceous band generally broader and more diffusely transitioning into the darker elytral colour posteriorly (Fig. 9B) compared with C. insuetus. Antennae, palps and legs testaceous.
Elytra with six discal and one submarginal striae. First to fourth more or less full length, fifth and sixth slightly abbreviated anteriorly; submarginal stria starting approximately one half to one third from base. Striae more deeply impressed and intervals therefore slightly more convex compared to C. insuetus. Head, pronotum, and elytra microreticulate and finely micropunctate. Pronotum extensively and coarsely striated on posterior surfaces, although reduced posteromedially. On some specimens striolation covers most of pronotum, also on anterior surfaces, but is reduced medially.
Ventral side entirely testaceous except metacoxal plate which is variably infuscated, especially laterally in some individuals; metacoxa and abdominal sternites II-IV striolate. Prosternal process with a slightly more pointed apex than in C. insuetus and C. kely, and lateral parts of metaventrite not as broad. Metacoxal lines long and not strongly diverging, as in C. kely.
Male: first three pro-and mesotarsomeres widened, ventrally equipped with suction cups (same constellation as in C. insuetus). Protibia modified, bisinuate and angled basally and broadened distally. Pro-and mesotarsal claws unmodified. Penis long, thin and simple; apex in ventral view straight and not leftturned; in lateral view rather evenly arched, lacking the distinct "shoulder" characteristic of C. insuetus but with a different type of postmedial and preapical suggested humps in the curvature. Apex in lateral view also broder closer to apex. Apex in left lateral view with a dorsal ridge crossing posterior dorsal margin but at a more acute angle (Fig. 6C) compared with C. insuetus. Preapically left side with fine transverse ridges. Right lateral and ventral side with fine longitudinal microsculpture but not forming coarse sulci as in C. insuetus. Parameres as in Figure 6C.
Female: very faint to no striolation on outer elytral intervals in the series from Ivohibe. All females from Ranomafana NP were densely striolated over the entire elytral surface except at the apical part, and entire pronotum. Density of striation approx. 5-8 striae in breadth across each elytral interval.
Etymology. Vokoka is a Malagasy adjective for curvature, also used for an old person with a hunched back. Here it refers to the even curvature of the male aedeagus in lateral view where the even curvature sets it apart from C. insuetus. It is a non-latinised adjective.
Distribution. Known from the mountainous escarpment of southeastern Madagascar at Ivohibe Special Reserve and Ranomafana NP (Fig. 12B). Our sampling is rather scant in the humid forests south of Ivohibe, for instance down to Andohahela NP so it is possible the species distribution continues further south.
Habitat and ecology. We collected this species in 2013 from forest pools with dead leaves next to streams at the Ivohibe reserve in pristine humid forest at an altitude of 870 m. The reserve of Pic d'Ivohibe was established in 1964 and is managed through collaboration between Madagascar National Parks, local people associations, and other partners. During our visit in 2013, there were little signs of degradation except at the entrance and at the west edge of the reserve. Local people sometimes take zebu cattle with them on a path through the forest. At a larger scale zebu excrement could influence freshwater quality and species assemblages through eutrophication, but there were no signs of this inside the intact pristine forest. In Ranomafana NP it was collected in a shallow clear-water shaded forest pond with large quantity of dead tree leaves and some Poaceae (M. Manuel pers. comm.).
Comments. Copelatus vokoka sp. nov. falls in the irinus group, based on the number of elytral striae. The new species is most closely related to C. ankaratra according to the mitochondrial gene CO1 and belongs to the C. insuetus species complex on Madagascar. Description. Body length 4.4-5.2 mm. Body shape very elongate and subparallel. Head infuscated around eyes and medially, testaceous on clypeus and as a posterior band. Pronotum entirely infuscated except lateral margins and especially the anterolateral corners more broadly testaceous. Elytra in same dark brown colour as infuscated parts of head and pronotum, except an irregular basal testaceous band (Fig. 10A). Testaceous basal band overall narrower than in C. insuetus. All appendages testaceous.

Copelatus ankaratra
Elytra with six clearly impressed discal and one submarginal stria. First to fourth elytral striae more or less full length, fifth and sixth striae slightly abbreviated anteriorly; submarginal stria starting 1/3 rd to 1/2 posterior of base and does not reach apex. Head, pronotum, and elytra microreticulate and finely micropunctate. Posterior third to posterior half of pronotum striolate. Strioles on average coarser than in C. insuetus and in some specimens more extensive onto disc and posteromedially.
Ventral side similar to C. insuetus, slightly darker on average so that medial light area around metacoxal lines may be more contrasting. Strioles on metacoxa rather short. Prosternal process more rhomboid and apex more strongly raised medially than in C. insuetus. Lateral parts of metaventrite rather broad. Metacoxal lines long and anteriorly diverging, similar to C. insuetus.
Male: first three pro-and mesotarsomeres widened and ventrally equipped with suction cups (same pattern as in C. insuetus). Protibia modified, bisinuate, angled basally, and broadened distally. Pro-and mesotarsal claws unmodified. Penis in posteroventral view distinctly angled at base (Fig. 6D) so that main blade appears tilted. Penis in lateral view evenly curved, lacking a distinct "shoulder", long, thin, and simple apex without a dorsal ridge crossing posterior dorsal margin in left lateral view. Faint longitudinal microsculpture visible at high magnification but lacking coarse longitudinal sulci ventrolaterally and lacking serrations as in C. insuetus. Parameres as in Figure 6D.
Female: all but one specimen examined lack finer elytral striolation and is in elytral structure similar to males. However, the entire pronotum and elytra except apical quarter are coarsely longitudinally striolate in one female specimen examined. Striolation coarser and made up of longer irregular but connected strioles and very different to the short separate fine strioles seen in external intervals of C. insuetus females. Density approx. 5-7 strioles in breadth across an elytral interval.
Etymology. The new species is named after the mountain massif Ankaratra where it was found and in honour of the newly created Ankaratra Massif Reserve in 2015. The epithet is a noun in apposition (ICZN 11.9.1.2).
Distribution. Known only from a few localities in the Ankaratra Massif Reserve on the central highland plateau of Madagascar (Fig. 12C).
Habitat and ecology. This new species was collected in the mountains of Ankaratra at altitudes above 2000 m. The first locality, Anosiarivo forest, consisted of water from a source flowing over grass vegetation at an altitude of 2070 m. The second locality, Tsiafajavona Mountain, was a small stream with grass vegetation downstream but very near to the source at an altitude of 2600 m near Ankaratra peak. Copelatus ankaratra seems to be a high-altitude alpine species associated with spring water.
Comments. Copelatus ankaratra sp. nov. belongs to the irinus group, based on the number of elytral striae. It belongs to the Copelatus insuetus species complex radiation on Madagascar and based on its CO1 it is most closely related to C. vokoka from Ivohibe. Notably, both these species have a densely striolated female form. There was a surprisingly large genetic distance (2.3-2.5% uncorrected p-distance) between the locality near the peak (2600 m) and the locality in the forested region ca. 5 km away at a lower altitude (2070 m). We find the male genitalia and other characters very similar and treat them here as conspecific.
The Ankaratra Massif is an area known for several microendemic species not known from anywhere else on Madagascar. This includes two critically endangered micro-endemic frogs, Boophis williamsi (Guibé, 1974) and Mantidactylus pauliani Guibé, 1974 only found in montane streams at elevations above 2000 m.
Manjakatompo forestry station was established in 1923 in the forested part of the mountains to preserve an area of 8320 ha, out of which only 650 ha is natural forest and 2300 ha has been replanted with exotic trees (Nicoll and Langrand 1989;Goodman et al. 1996). Even the natural forest part is largely composed of secondary forest mixed with exotic trees. The forests still support endangered highland fauna and are important sites for some of the last remaining central plateau forests (Hjalmarsson et al. 2013). The higher elevation of the Ankaratra Massif has, until recently, lacked any protection despite the unique faunal components and characteristics (Vences et al. 2002;Goodman et al. 1996). The area has suffered severe degradation due to anthropogenic activities, mainly heavy deforestation and fire (Rabemananjara et al. 2012), but also overgrazing by livestock and expanding potato farming (IUCN SSC Amphibian Specialist Group 2016). However, in 2015 the Ankaratra Massif Reserve was created which encompasses Manjakatompo Special Reserve and the higher elevations where the endangered montane frogs are found (Moore 2015). It is managed by the local Malagasy conservation association (VIF) in collaboration with Ministère de l'Environnement, de l'Ecologie et des Forêts. This is one of the results of an ambitious conservation programme involving habitat restoration, alternative livelihood initiatives, and public awareness (Rabemananjara et al. 2012). Nevertheless, during a visit as late as 2016, there were many signs of continuous degradation from "tavy" slash and burn agriculture, commercial logging, and charcoal extraction exposing forest streams. Very large parts of the forest were burnt, causing erosion that spills into the streams. Manjakatompo forest and the Ankaratra Massif are very important forest and montane refugia of highland fauna (Vences et al. 2002;Goodman et al. 1996;Hjalmarsson et al. 2013;Andreone et al. 2014), and it should be highly prioritised for protection; we hope that the new reserve status on paper will lead to actual changes in practice. Diagnosis. The best diagnostic character for the species is the pseudostriae between first and second, and second and third striae (Fig. 10B), which separates the species from all other Copelatus species from Madagascar; the penis is gently curved from base to apex and parameres are long and thin (Fig. 4C).

Copelatus pseudostriatus
Description. Body length 5.3-5.6 mm. Body shape elongate oval. Head and pronotum of all three specimens exposed to DNA extraction lysis buffer which has discoloured them slightly. Colour descriptions of head and pronotum below should therefore be taken with caution and can differ somewhat from other specimens. Head and pronotum rather uniformly brown but head likely infuscated between and around eyes (more visible in one paratype) and pronotum may have been infuscated medially prior to exposure to lysis buffer. Elytra uniformly testaceous brown with a faint suggestion of a darker transverse field preapically (Fig. 10B). Appendages testaceous.
Elytra with six clearly impressed striae and one submarginal stria. Stria five distincly shorter basally and the submarginal stria starts 1/4 th to 1/3 rd posterior from the base. Between first and second, and between second and third striae, there are irregular traces of intermediate striae, or "pseudostriae", extending from just after base until 1/4 th from the apex (Fig. 10B). Anteriorly they are continuous or almost so and posteriorly they become fragmented. Few traces of pseudostriae can also be seen between the third and fourth striae but only anteriorly, and these are fragmented. Head, pronotum, and elytra microreticulate and finely micropunctate. Pronotum rather weakly striolate along posterior margin, slightly more at posterolateral corners, and sporadically anteriorly and on disc.
Ventral side largely ferrugineous, a little lighter testaceous-ferrugineous around metacoxal processes, medially on the metaventrite and on sternite II. Prosternal process short, medially raised and rounded and with a fairly pointed apex. Lateral parts of metaventrite medium broad. Metacoxal lines anteriorly diverging and abbreviated well before metaventral margin. Metacoxal plate distinctly striolate with short strioles.
Male: first three pro-and mesotarsomeres widened. Protibia modified, narrow but not bisinuate with an angulate ventral margin at base, broadened, almost club-like, distally. Protarsal and mesotarsal claws unmodified. Penis in ventral view narrowed one third from apex and the very last apical tip angled to the left; in lateral view evenly and weakly curved from base to apex (Fig. 4C). Parameres as in Figure 4C, long and thin and form an evenly curved elongated triangle.
Female: elytral structure similar to male. Etymology. The name pseudostriatus is a compound word formed from pseudo-(false) and striatus (furrowed or striated) and refers to the intermediate non-complete striae in-between the complete continuous striae on the elytra in this species. It is the only species of Copelatus on Madagascar with this characteristic. The word striatus (masculine) is a participle (verb as adjective) in the nominative singular (ICZN 11.9.1.1).
Distribution. Endemic to Madagascar, only known from the type series from Tsaratanana Massif (Fig. 12A).
Habitat and ecology. Not known, but the type series of specimens were collected in 2004 likely from a stream, near Antetikalambazaha Camp at an altitude of 1700 m.
Comments. Species group assignment of Copelatus pseudostriatus sp. nov. is hardly possible: based on the complete striae it would fall in the irinus group, but the incomplete pseudostriae are likely reduced striae of an ancestor with a higher number of complete striae. This is a very distinct species with no recognisable close relatives, either based on genitalia or the CO1 gene. Tsaratanana Massif contains the highest peak in Madagascar at 2876 m and possibly this species is endemic to the Tsaratanana Massif. However, Hjalmarsson et al. (2013) showed that a high-altitude diving beetle species, Rhantus manjakatompo Pederzani & Rocchi, 2009, collected at the same place and time as the type series of C. pseudostriatus, is a species shared between Tsaratanana and Ankaratra Massif in central Madagascar. Diagnosis. Somewhat similar to C. insuetus on habitus appearance, but sturdier, broader pronotum and head with a much greater interocular distance compared to width of eyes, and subparalell along a longer distance of body with more rapidly attenuating anterior and posterior ends; easily distinguished from all other irinus group species of Madagascar by the penis shape, which is of a type otherwise found in species closely related to Copelatus owas; penis has a large medial expansion in lateral view followed by an apical blade (Fig. 6E).
Description. Body length 4.3-5.2 mm. Body shape elongate and subparallel along a very long part of the body. Pronotum and head broad and eyes small creating a very wide interocular distance. Maximum width of pronotum clearly in front of hind corners. Head rufotestaceous with weak or absent infuscation in between and posterior of eyes. Pronotum infuscated medially with broadly testaceous lateral sides. Elytra brown, with a broad testaceous transverse band basally (Fig. 10C). Testaceous band broader and transitioning posteriorly more diffusely into the brown colour compared with C. insuetus. Antennae, palps and legs testaceous.
Elytra with six discal and one submarginal stria. Fifth stria abbreviated anteriorly, variably also first and third striae. In some individuals especially fifth but also first and sixth striae are rudimentary or with very shallow impressions. Interval between fifth and sixth striae narrow, approx. half interval between first and second striae. Submarginal stria short, starting around middle. Head, pronotum, and elytra microreticulate and finely micropunctate. Pronotum not striolate.
Ventral side entirely testaceous except last three abdominal ventrites may be vaguely infuscated and have lighter lateral spots. Metacoxa and abdominal sternite II-IV striolate, but strioles shallower and finer than in C. insuetus. Compared with C. insuetus, the metacoxal lines are shorter, and the anterior traces suggest an inward curve towards the posterior metaventral margin. The lateral part of metaventrite is narrower than in C. insuetus at level of mesocoxa, equal to the width of mesofemur at middle. Posterior metaventral margin not straight but slightly angular at level of apex of mesotrochanter. Prosternal process is slightly more elongate.
Male: first three pro-and mesotarsomeres widened, ventrally equipped with suction cups. Pattern of suction cups same as for C. insuetus but tarsomeres not as wide and less developed as an integrated protarsal palett. Protibia modified, bisinuate, angled basally, and broadened distally. Pro-and mesotarsal claws unmodified. Penis very characteristic, rather broad and short and in lateral view with a medial expansion followed by a sharp constriction before the narrow blade-like apex (Fig. 6E) which is characteristic of several species related to C. owas Régimbart, 1895; apical blade curved leftwards in ventral view and serrated by transverse ridges on the convex right side. Parameres as in Figure 6E, rather long and thin with a medial emargination on the concave side.
Female: on average smaller than males (Table 3). At least some females with elytral microreticulation slightly more strongly impressed than in males and therefore appearing more matt.
Etymology. The species name safiotra is a Malagasy noun for hybrid, here referring to the unusual combination of a male genitalia type, typical of the Copelatus owas species complex, in a body with a 6+1 striated elytra very much resembling the C. insuetus complex of species. It is a non-latinised noun in apposition.
Distribution. Endemic to Madagascar. This species has a rather large distribution in the eastern humid forest from Anjanaharibe Sud reserve in the NE, all along the eastern escarpment including Zahamena NP, Mantadia NP, and Ranomafana NP, and even extending to the rather isolated western patch of subhumid forest at Isalo NP (Fig. 12D).
Habitat and ecology. This species seems to be strongly associated with clean streams having sandy substrate in humid forests. At these localities, individuals can be found in sidepools, at margins or sites protected from waterflow (e.g., by fallen logs) where dead leaves and debris accumulate. The species has been found at altitudes between 700 and 1300 m but most numerous at elevations above 900 m in primary humid forest. The discovery in Isalo NP, in a sandy river running through a Canyon, indicates that subhumid forests may also be part of the species' niche.
Comments. Copelatus safiotra sp. nov. falls in the irinus group, based on the number of elytral striae (six discal and one submarginal striae). However, the genitalia is of a very different type and characteristic of the complex of species close to C. owas in the erichsonii group with ten discal and one submarginal striae. We hypothesize that this species, despite the body shape and number of elytral striae, is not related to the other species treated here, but belongs to the radiation of species around C. owas. This would reinforce the idea that the number of elytral striae is a very homoplastic character and not reliable to create phylogenetically sound groups (Balke et al. 2004).

Additional non-erichsonii group species of Madagascar
We are aware of several Copelatus species not of the erichsonii group on Madagascar but of which we have only seen females. Some of these are undoubtedly new, confirmed with DNA data (Figs. 2, 3), but we refrain from formally naming these prior to the discovery of males. We discuss these below and list the examined material. more likely a new endemic species of Madagascar, the male of which should be sought for SE of Antsalova, Tsingy de Bemaraha NP, western Madagascar. We managed to sequence partial CO1 (311 bp) of this female which differed substantially from the three other species in the complex (K2P: 8.2-13.6%). DNA matching of the male, once discovered, would be straightforward. Description. This species has a configuration of elytral striae not found in any of the other species treated here. It has five discal striae and is lacking a submarginal stria and would fall in the longicornis species group together with C. befasicus. However, the five striae are likely not homologous as C. befasicus has an abbreviated first stria, whereas the first stria is entirely lacking in this specimen, therby creating an interstriae space double in width compared to the outer intervals. But despite the lack of the first stria, it has five striae on the central and lateral parts of elytra where C. befasicus only has four. In addition, the striae are very faintly impressed, intermediate between real striae and the puncture lines found in C. peridinus. The colouration is uniformly brownish black like in C. peridinus and the body size is also similar. This specimen is possibly a different species compared to all others presented here; however, we cannot rule out that intraspecific variation of C. peridinus ranges from two defined puncture lines to five weakly impressed striae. They are very similar in all other aspects. We managed to sequence a partial fragment of CO1 (447 bp) from the Andasibe specimen (NHRS-JLKB000065698) and this indeed showed that this specimen is closely related to, or possibly conspecific with, C. peridinus. Genetic distance between them was 3.0% (K2P). Such a level of intraspecific variation is not impossible but unlikely given they were from the same locality (Analomazaotra reserve). Together with the morphological characters we believe this may be a different and distinct species, but we await the discovery of a male before the identity can be established with confidence. The second female from Namorona River near Ranomafana NP (NHRS-JLKB000065762) is substantially larger (6.4 mm) than the female from Andasibe (5.6 mm) but otherwise shares the same characteristics (for distribution see Fig. 11B).  Comments. The DNA data revealed that these females represent one or possibly two additional new species in the C. insuetus complex (Figs 2, 3). In fact, the CO1 data reveals that they are the most divergent in that group and are sister to a clade with all the other species: C. insuetus, C. vokoka, C. kely, and C. ankaratra. The genetic distance between members of these two clades ranges from a minimum of 4.5% to a maximum of 7.1%, strongly indicating a separately evolving unit. The genetic distance between the specimens from Ivohibe and those from north of Toamasina was 2.3-2.4% (K2P), a distance that does not rule out conspecificity as the geographic and altitudinal distance are substantial between these localities (for distribution see Fig. 12C). It is also on the same level as the intraspecific distance found within C. ankaratra between a peak population and a population at lower altitude of the Ankaratra Massif between which we do not find any morphological character differences to justify further separation. On the other hand, C. kely and C. insuetus are indistinguishable based on CO1 squences alone (Figs 2, 3). We refrain from describing these as a new species since the shape of male genitalia is very important for identification in this group. Morphologically we note the following based on the females: in body size this species is similar to C. vokoka, slightly larger than C. kely but slightly smaller than C. insuetus and C. ankaratra. Compared with females of C. vokoka, Copelatus sp. 3 has a narrower testaceous basal band, flatter elytral intervals between striae, and more limited striolation on pronotum, restricted to posterolateral corners. Copelatus ankaratra can be distinguished based on its dark colour and extremely elongate body shape. Small C. insuetus females can often be distinguished on the posteriorly extended testaceous basal medial band. It is most difficult to distinguish Copelatus sp. 3 females from large female specimens of C. kely.

Discussion
Madagascar is known for an extremely rich endemic flora and fauna which, together with the unfortunate level of deforestation, has rewarded the island with a top spot among the world's biodiversity hotspots (Myers et al. 2000). This high level of endemism, also among insects, is particularly manifested at species level but also at higher taxonomic level such as endemic genera (Goodman and Benstead 2003). For diving beetles (family Dytiscidae), there are currently three endemic genera. Other Dytiscidae genera like Copelatus are not endemic but may still contain endemic species radiations within. Here we have revised the species of Copelatus on Madagascar excluding the erichsonii species groups. We recognise 13 species with names and three additional nonnamed species based on females. Of the named species nine are endemic to Madagascar, two (C. marginipennis and C. distinguendus) are regional endemics to Madagascar and nearby west Indian Ocean islands, and two (C. pulchellus and C. peridinus) also occur on the African continent. This gives an endemic proportion at species level of approximately 70%, an intermediate level compared to other insect groups on Madagascar (Goodman and Benstead 2005).
The Copelatus diversity on Madagascar represents four of the traditional species groups in the genus based on the number of elytral striae (Sharp 1882;Balfour-Browne 1939;Guignot 1961;Guéorguiev 1968;Nilsson et al. 1997;Nilsson 2001): the hydroporoides (2 species), longicornis (1 species), irinus (10 species), and erichsonii (> 20 species) species groups. Copelatus unguicularis of the consors species group turned out to be a species of the genus Madaglymbus. As has been flagged before, these species groups are commonly not monophyletic (Balke et al. 2004), and in fact the number of elytral striae is a highly homoplastic character which can even vary within a single species (Bilardo and Rocchi 2008; personal observations). A better system based on phylogenetic relationships can likely be approached by using the shape of male aedeagus, and by using genetic data. We informally refer to some species complexes which we believe are groups of closely related species based on the shape of male aedeagus. Hence, the C. insuetus complex contains four named and one or two additional unnamed species where males are currently unknown, and we hypothesise that these constitute a hitherto unrecognised monophyletic radiation on Madagascar, to be tested with a larger sample. It is also highly likely that additional species of this complex exist on Madagascar and will be discovered in the future. This complex includes young species not yet delimitable based on the mitochondrial CO1 gene (Fig. 3). This can be due to incomplete lineage sorting or hybridisation, but they are recognisable morphologically. Copelatus safiotra sp. nov., in contrast, although in habitus similar to the C. insuetus complex, did not belong to this complex. Despite the low number of elytral striae, the male aedeagus is of the C. owas complex type, with a subapical expansion followed by an apical blade, and likely, phylogenetically, belongs with this group.
Copelatus as a genus is widespread all over Madagascar. Species can be found from the humid forest in the east to the dry forest in the west and from lowlands to the highest peaks. But the different species complexes have particular niches. Species in the C. insuetus complex are predominantely inhabitants of the eastern humid forests. Copelatus distinguendus and related species (C. peridinus, C. sp. 2, and likely C. baculiformis) mostly occupy open, often anthropogenically disturbed, habitats of the Central Highlands. The C. pulchellus complex (C. marginipennis, C. mahajanga, C. pulchellus, and C. sp. 1), and C. befasicus seem to be most abundant in the dry deciduous western forests, lowlands with open to semiopen landscapes. Finally, some species are specialists like the high altitude crenophile C. ankaratra sp. nov. and the sandy stream specialist C. safiotra sp. nov. Not until we have this knowledge -what are the species, how can we recognise them, where do they occur and how do they live -can we attempt to protect them and their habitats in the face of constantly increasing human pressure on the environment.