Eschscholtz (1829) named three cardiophorine genera (Aptopus, Cardiophorus, and Esthesopus) in his initial division of genus Elater Linnaeus, 1758. Fifteen additional Cardiophorine genera were described between 1800 and 1900, of which four to seven were in synonymy at the outset of this study according to various authors (listed in synonymy). Candèze (1860) wrote the first genus level revision of the Cardiophorinae, in his four-volume total revision of Elateridae (Candèze 1857–63). The most recent genus level revision was published by Schwarz (1906). The monophyly and membership of all genus level groups remain untested hypotheses because phylogenetic analysis has never been applied to any of the genera.

The Cardiophorinae were divided into two tribes when Gurjeva (1974a) transferred the monotypic tribe Nyctorini from the Elaterinae into the Cardiophorinae. This transfer was made without comment, suggesting Gurjeva did not realize Nyctorini was incorrectly described as Elaterinae. So, division of the Cardiophorinae into tribes Cardiophorini and Nyctorini was perhaps accidental. Dolin (1975) placed Nyctorini in synonymy under Cardiophorini, effectively eliminating tribal level structure. Stibick (1979a) removed Nyctorini from synonymy because, although he dismissed two diagnostic characters as weak, he did not know of other Cardiophorinae with absent [short] adult prosternal lobes. This study will use phylogenetic results to assess whether tribe Nyctorini should be a synonym of Cardiophorini.

Although little-tested phylogenetically, the monophyly of the Cardiophorinae has never been questioned in the literature. Subfamily-level non-monophyly is however possible due to inconsistencies in the characters used to separate Cardiophorinae from Negastriinae. Several apparent synapomorphies unite Negastriinae and Cardiophorinae: closed mesocoxal cavities; hind wing without anal cell; and basally-fused parameres, articulated at their midlength (Douglas 2011). The characters used to distinguish these two subfamilies since their description (Candèze 1860, Negastriinae as part of Cryptohypnites) are the short cardiophorine prosternal process, the broad prosternum of most Negastriinae, and the heart-shaped scutellum of most Cardiophorinae. However, Stibick (1979a) noticed these were not universal and has omitted the prosternal width character and qualified the other two characters with the terms “usually” and “normally.”

Other putative evidence for cardiophorine monophyly comes from the distinctive cardiophorine larvae (Hyslop 1921, Ôhira 1962, Stibick 1979a, Calder 1996). Potentially synapomorphic characteristics include: deeply cleft mandibles, thread-like abdomen with extra pseudosegmentation, and digitate anal lobes (Stibick 1979a). Although this larval type is known from Europe (Palm 1972), Central Asia (Atamuradov 1993), northeast Asia (Dolin and Gurjeva 1975), Japan (Ôhira 1962), Australia (Calder 1996), New Zealand (collections only, without digitate anal lobes), North America (Tenhet 1941) and South America (Costa et al. 1988), larvae remain unknown for most genera and species. Thus, it remains unknown whether these probable synapomorphies are of the Cardiophorinae alone, of the Cardiophorinae and other taxa, or of only some Cardiophorinae. Although these strong larval morphological characters exist and are possible evidence for cardiophorine monophyly, too few larvae are known for them to yet be used to test monophyly.

A previous study (Douglas 2011), analyzing elaterid phylogeny using adult morphology, found: the included Cardiophorinae were closest to Margogastrius Schwarz, 1903 and Teslasena Fleutiaux, 1892 (Physodactylinae) and then Negastriinae. There the included Cardiophorinae were monophyletic, excluding Exoeolus Broun, 1893 and three fossil genera (Crioraphes Iablokoff-Khnzorian, 1961; Pseudocardiophorites Dolin, 1976 and Protocardiophorus Dolin, 1976). It also showed that Cardiophorinae may render Negastriinae paraphyletic. Furthermore, Tropihypnus Reitter, 1905 and a paraphyletic Hypnoidini (Dendrometrinae) were the sequential sister groups to Cardiophorinae + Negastriinae.

Douglas’ (2011) finding close relationship between the Cardiophorinae and the Negastriinae agrees with DNA sequence data-based results (Sagegami-Oba et al. 2007; Oba 2007; Kundrata and Bocak 2011 [with Platiana Schimmel, 1993 as sister to included Cardiophorinae, assigned to Dimini]; and Kundrata et al. 2016). Douglas (2011) also found strong support for Dendrometrinae: Hypnoidini as sister to Cardiophorinae + Negastriinae + Tropihypnus. This study uses Negastriinae, Hypnoidini, and Tropihypnus as outgroups for phylogenetic analysis of the Cardiophorinae, as the taxa identified as most likely (Douglas 2011) to render the Cardiophorinae non-monophyletic. Cardiophorine monophyly has not been demonstrated through analyses of larval or adult morphology to date, and requires testing.

Specimens examined for morphological coding belonged to 29 insect collections (Table 1). Codens listed here follow Arnett et al. (1993), except where collections preferred other codens. Among these specimens were 61 primary types, or paratype specimens, representing 41 species (Appendix III). Appendix III also includes lectotype designations for 29 species. These are designated to fix generic concepts and to ensure their universal and consistent interpretation. Types of 307 more species, which were not coded for phylogenetic analysis, were photographed at NHM (London), ISNB and MNHN. Types of 85 more North American cardiophorine species (listed in Douglas 2003) were also examined to ensure taxon sampling reflected much of the group’s morphological variation and assess new taxonomic placements.

Non-type specimens were identified by comparison with types (types were examined for 40 species) or specimens identified by experienced workers (three species, Appendix I). All non-type specimens examined were labelled with unique identifier numbers (Appendix II). Three distinctive undescribed species were included to better represent the Cardiophorinae. All identifications of non-types were evaluated using published keys and descriptions (Appendix I), and five species were identified using literature alone. Information from type specimens was often used in coding species. In most cases where a single name-bearing type did not already exist, a lectotype was designated for each species name (Appendix III).

For this study, 51 ingroup and 26 outgroup species were coded for phylogenetic analysis (Appendices I, IV). Of these, 21 were missing genitalic data for one sex. For two species, material was unavailable for both the aedeagal and female genitalic characters. Non-genitalic characters were coded from male specimens, except for six species, for which only females were available. Agrypnella was coded using males of the type species A. eburnea Champion, 1895, and a female identified as A. squamifer (Candèze, 1895), although it is unknown whether these two species are really distinct. Some species for which some characters could not be examined were coded partially from literature (Calder 1996, Gurjeva 1966, Ôhira 1963). A generic catalog including references and synonymies was assembled for the Cardiophorinae (Revised Synonymy below) to clarify genus level nomenclature.

Ingroup taxa included 53 cardiophorine species, including 5 taxa found to be near Cardiophorinae by Douglas (2011) (Pachyelater Lesne, 1897 (Physodactylinae or Dendrometrinae); Margogastrius (Physodactylinae); Negastrius americanus (Horn, 1871, Negastriinae); Teslasena (Physodactylinae), and an undescribed species from New Zealand). The type species of 25 of 32 valid and subjectively synonymised genus level names for the Cardiophorinae were included to ensure accurate inclusion of named genera and subgenera (Revised Synonymy below). Four genus-level names (Aptopus, Coptostethus Wollaston, Dicronychus, Triplonychoidus) were represented by non-type species only, and five genus level taxa were entirely unavailable for inclusion (Allocardiophorus Ôhira, Cardiophorellus: subgenus Parapleonomus Cobos, Cardiophorus: subgenus Lasiocerus Buysson, Mionelater Becker [fossil], Ryukyucardiophorus Ôhira (examined post-analysis)). Taxon sampling was most comprehensive for the genera with species occurring on multiple continents.

The outgroup (24 taxa) includes subfamilies Negastriinae, Elaterinae, Agrypninae, and Dendrometrinae. Previous studies indicated the Negastriinae were the sister group to (Sagegami-Oba et al. 2007; Oba 2007; Kundrata and Bocak 2011; and Kundrata et al. 2016), or rendered paraphyletic by (Douglas 2011) the Cardiophorinae. Because of this, exemplars of the type species 15 of 30 world negastriine genera, and one non-type species, were included to further test cardiophorine monophyly. These genera include seven of Stibick’s (1971) eight unnamed genus groups. Three genera of Hypnoidini including the type genus Hypnoidus Dillwyn, 1829, and Adrastus Eschscholtz, 1829, were included because Douglas (2011) found these were the most likely sister groups to Cardiophorinae + Negastriinae. Type species of Elater Linnaeus, 1758; Agriotes Eschscholtz, 1829; Athous Eschscholtz, 1829; and Agrypnus Eschscholtz, 1829 were also included, although only Elater was formally defined as outgroup in the analyses to root the trees.

Specimens were relaxed for examination by placement in nearly boiling distilled water for 10–30 minutes. Wings were photographed in water under a cover slip on a glass microscope slide. Male and female genitalia were prepared and examined as outlined in Calder (1996). Specimens were examined using a Leica Wild M-10 dissecting microscope and all structures examined were photographed using an attached Nikon Coolpix 995 digital camera. Measurements were made using either an ocular micrometer or from digital photographs using Corel Photo-Paint 12 software. Drawings were made using these digital photographs. Vestiture has been omitted from drawings except where taxonomically informative. Structural terms follow Douglas (2011). Figures of single proximal sclerites of the bursa copulatrix are meant to be of the right-sclerite viewed as from inside the bursa (internal view), unless stated otherwise. Sclerites of the bursa copulatrix are illustrated in internal view unless lateral view is specified.

Morphological characters were coded using majority coding of polymorphisms in order to use all available information and avoid bias. For qualitatively defined characters, majority coding was practiced by coding the character state most commonly observed in each species. For 27 quantitatively coded morphometric characters (Table 2), the value entered for each species was the mean of ratios of length measurements or mean angle. These values were ranked assigned to character-state bins “0” or “1” based on whether the measured value for that character for each species was above or below the median value for the character among all species.

Morphological character selection and coding were performed together. All observed variation was evaluated as a potential character source following one procedure. To be considered suitable, variation between homologous structures must allow diagnosis between at least one pair of species. All 136 characters that could not be described as length ratios, or counts, were treated qualitatively as binary or multistate characters. Qualitative characters included the presence or absence of structures, or objective shape descriptors (e.g., notched vs. uniformly convex). An exemplar species was assigned for each qualitatively defined character state in an effort to produce repeatable, standardized character state definitions (many follow Douglas 2011). Following these criteria, all qualitative characters identified as showing non-overlapping variation between at least two species were considered for possible use. Subsequent characters with apparent developmental or genetic non-independence were then excluded. Autapomorpic characters were also encoded because they provide branch length information for Bayesian analysis and diagnostic characters.

Phylogenetic analysis was conducted using 163 characters (Table 2), of which 27 were coded quantitatively (into binary pairs) and 136 qualitatively. These characters included 376 character states (after binary coding of quantitative characters), of which 40 were autapomorphic (Table 2). Qualitatively coded characters 6, 9, 22, 27, 29, 37, 44, 60, 85, 123, 133, 145, and 161 were treated as ordered multistate characters. Three characters, common to many fossorial Elateridae (Douglas 2011), were omitted from the analyses presented here, to avoid phylogenetic bias due to convergent evolution. These were characters 9, 34 and 46 which included the following character states apparently associated with fossorial adults: mandibular apex unidentate; prosternum with anterior edge short, exposing labium; and protibiae near apex with posterior surface flattened, concave, or broadened apically, apparently modified for digging. Character 162, riparian habitat association was also excluded from the analysis. Analyses including these characters (not presented) had similar topologies to those with them omitted but Bayesian posterior probability values (PP) were lower throughout the tree, supporting the hypothesis of convergence.

Although both parsimony and Bayesian analyses (as implemented by MrBayes v.3.1.2, Huelsenbeck and Ronquist 2001, Ronquist and Huelsenbeck 2003, using the model by Lewis 2001) were used here to infer phylogeny, results of the model-based Bayesian analyses were preferred for taxonomic inference. The major expected advantage of using Bayesian analyses for morphological data is that the Mkv model of Lewis uses branch length information while parsimony does not. Empirically, Wiens (2005) found that accuracy of Bayesian analyses equalled or exceeded that of parsimony. Parsimony analyses were also performed because parsimony remains widely accepted.

Bayes factors were used, as outlined by Sikes et al. (2006), to infer which of two evolutionary models best fit the data. These were the generalized Jukes-Cantor model for k states (Mkv), corrected for acquisition bias (Lewis 2001), with or without gamma distributed rate variation between characters (Mkv vs. Mkv + Γ). A Bayes factor of 1118 (2X ln L = -12837 for -Gamma and = -11719 for +Gamma) showed strong support (assessed as outlined by Kass and Raftery 1995) for models including gamma-distributed rate variation between characters over models that did not include gamma variation.

Phylogenetic analysis was performed using both parsimony and Bayesian criteria. For Bayesian analyses, prior probability distributions were at default values of MrBayes. Gamma distribution was approximated using the default setting of four rate classes. Settings for likelihood parameters used were Mkv (nst=1, coding=variable) and Mkv (nst=1, coding=variable, rates=gamma). Searches began with randomly selected starting trees and were run for 8 million cycles (until the average standard deviation of split frequencies between four parallel runs was below 0.01). Samples of trees from the MCMC chain were taken every 100 cycles, which resulted in 80 thousand trees. All but the first 20 thousand trees were used to compute a majority rule consensus tree assigning posterior probabilities of tree topology. The matrix was analysed three times to test repeatability. Because these differed slightly, the analysis with the highest average harmonic mean log likelihood was used for phylogenetic inference.

Parsimony analysis was performed using PAUP* (Swofford 2001). The heuristic search procedure was used with 1000 random replications of stepwise-addition, with the branch-swapping algorithm (maxtrees set to auto-increase, multrees option in effect, Appendix 5). Bootstrap values were generated through 1000 replicates of bootstrapping using the same settings but with maxtrees set to 1000 and the number of replicates of stepwise addition reduced to 10 to reduce processing time. Decay index scores were calculated using PRAP (Müller 2004) for decay analysis of a strict consensus of all trees found in initial parsimony analysis.

Zander (2004) found that Bayesian posterior probability (PP) values are predictably liberal at branch lengths typical for morphological studies (fewer than 35 changes). Because of this bias, PP values and other branch support metrics were corrected using Zander’s table 4 when assessing clade credibility for hypothesis testing.

Testing hypotheses of monophyly was done by determination of the PP of the focal clade. The hypotheses tested are either ones stated explicitly as such, or ones implied by the description of taxa.

A key to the genera of cardiophorine, and corresponding diagnoses were developed using: existing diagnostic characters, the phylogenetic matrix (Appendix IV), and examination of other species from each genus.). Where phylogenetic results were informative, classification was revised to reflect phylogenetic history through synonymy, description of new genera, changes of rank, and new generic placements. Existing generic concepts were maintained where phylogenetic results were inconclusive.

Bayesian (Fig. 36, Table 3) and also parsimony analyses (Fig. 37) show that the Cardiophorinae are a well-supported clade if several taxa are transferred into Cardiophorinae. Here, the Cardiophorinae can be corrected by adding a few species and genera from subfamilies Physodactylinae, Dendrometrinae and Negastriinae (Margogastrius, Negastrius americanus, Teslasena, and the undescribed species from New Zealand, and Pachyelater Lesne). Three of these also require further taxonomic alterations, as discussed below.

The resulting Bayesian tree (Fig. 36) also showed Negastriinae as paraphyletic (Nodes a–c) and strong support for monophyly of Cardiophorinae + Negastriinae (Node a). However parsimony analysis found weak support for a monophyletic core Negastriinae that is sister to the Cardiophorinae (Fig. 37). In the Bayesian analysis, support for negastriine monophyly was only 0.0001 (Table 3). The strength of this rejection is surprising, given that Bayesian analysis (Fig. 36) left Negastriinae mostly unresolved (Node a), and that support for the two nodes showing paraphyly of the Negastriinae was only 0.65 & 0.72 (Nodes b & c). Further phylogenetic analysis, including more genera of the Negastriinae, is important to further test the validity of the Negastriinae and membership of both subfamilies.

The existing tribal classification of the Cardiophorinae is incorrect according to both Bayesian and parsimony analyses. This is because the monotypic Nyctorini rendered the only other tribe, the Cardiophorini, paraphyletic (Table 3, Fig. 36). The only diagnostic characters of Nyctorini (Semenov-Tian-Shanskij and Pjatakova 1936), i.e., the short anterior prosternal lobe and sexual size dimorphism, are homoplastic characters found in many fossorial elaterids. For these reasons Nyctorini should be a junior synonym of Cardiophorini, effectively eliminating tribal level classification of the Cardiophorinae. Because of the generally pectinate shape of trees for Cardiophorinae (Figs 36, 37), no natural divisions were found for a convenient tribal level classification, and all Cardiophorinae should be placed in tribe Cardiophorini.

The only taxonomic change to Negastriinae is the transfer of Negastrius americanus from Negastrius Thomson, to Cardiophorinae (as a new genus). Within Negastriinae, three genera previously transferred from Cardiophorinae form a well-supported clade in both Bayesian (PP > 0.95 after correction for branch length, length = 27, uncorrected probability = 1.00) and parsimony (D = 5, BS = 75) analyses (Figs 36, 37). These distinctive genera, Agrypnella, Cardiohypnus and Rivulicola live in riparian habitats in the Neotropics, South Asia and Australia respectively. Rivulicola is unusual as the only Negastriinae known from Australia. These are recognizable among the Negastriinae because of their scale-like setae. All three genera were once placed in the Cardiophorinae because of their heart shaped scutella (with emarginate anterior edge), ovoid pronota and elytra, and short prosternal processes. They were transferred independently to the Negastriinae by three different authors (Dolin 1992, Golbach 1994, Calder 1996), at least in part, because of their convex prosternal sides. However, despite the similarities that these three genera share with Cardiophorinae, this group was not found sister to the Cardiophorinae here. Fleutiauxellus Méquignon, a negastriine appearing less like Cardiophorus is the most likely sister group to the Cardiophorinae according to Bayesian analysis, sharing with many Cardiophorinae the pedunculate anterior sac of the bursa copulatrix.

The required changes of classification among the Cardiophorinae are discussed beginning at the root of Cardiophorinae in the Bayesian tree. Some paraphyletic and polyphyletic genera are recognised here, in cases where phylogenetic results did not provide well-supported alternative to the prior classification. The most basal cardiophorine node is an eight-way polytomy (Fig. 36, Node d). Here, genus Aphricus is made paraphyletic in both analyses by at least the fossorial Australian genus Patriciella Van Zwaluwenburg, 1953 and an undescribed species similar to Aphricus from New Zealand. In order to avoid recognising a non-monophyletic genus, new genus Chileaphricus is established for Aphricus chilensis Fleutiaux. Since Aphricus (from California, USA) + Patriciella and the undescribed species from New Zealand, form a clade with moderate support (PP = 0.82, D = 1) they should be treated as a single genus (by synonymising Patriciella under Aphricus, its type species becoming Aphricus australicus Van Zwaluwenburg, 1947).

Blaiseus Fleutiaux, another basal cardiophorine, was found monophyletic here. This genus has 10 species distributed in Southeast Asia, South Africa, and Central and North America, (Douglas 2009). The type species (B. Bedeli Fleutiaux, 1931), and a male of the South African B. nothoafricanus Douglas, 2009 were included here. Although support for their monophyly was only 0.90 (also supported by parsimony, Fig. 37. Bremer support, (D) = 3, bootstrap support, (BS) = 81), the characters uniting them are distinctive. One such synapomorphy is their unique, split parameres (Fig. 25). The widespread distribution of the few known species, and the basal position of this genus within Cardiophorinae suggest Blaiseus is a long-separated lineage with a possibly relictual distribution.

Pachyelater Lesne, 1897 is a robust-bodied fossorial elaterid genus from Madagascar with sexually dimorphic males and females (Figs 50–53). Because this genus falls within the Cardiophorinae in both analyses, it should be transferred from Dendrometrinae to Cardiophorinae. Because females of Pachyelater are flightless and fossorially adapted the undiscovered females of the closely-related Aphricus may also share these traits (Fig. 36). Furthermore, females of Aphricus spp. may also be similarly larger than males, with reduced eyes and have the ovipositor and bursa copulatrix without sclerites. Margogastrius Schwarz, a genus known from only two damaged female type specimens from coastal Tanzania (also flightless and fossorial) was found with weak branch support as the sister to Blaiseus (although they are not closely related according to parsimony, Fig. 37). Examining internal genitalia of the remaining undissected type specimen might yield further phylogenetic information. No associated males have been identified with external morphology or distribution like these females. While the historically enigmatic species Negastrius americanus clearly belongs to the Cardiophorinae, the characters examined in both male and female specimens did not suggest placement in any other genus (Figs 36, 37). Therefore I propose to place it in the new monotypic genus Floridelater gen. n.

The remaining taxa in the polytomy of the Bayesian analysis (Fig. 36, Node d): Neocardiophorus Gurjeva, 1966; Nyctor Semenov-Tian-Shanskij & Pjatakova, 1936; and Cardiophorus subgenus Metacardiophorus Gurjeva, 1966 are all known from central Asia. Among these, only Nyctor is known from both sexes, so the discovery of females of the other two genera would provide important data for improved phylogenetic placement and on the evolution of flightlessness in the Cardiophorinae. Since subgenus Metacardiophorus is distantly related to subgenus Cardiophorus (Fig. 36, Node d, not h; Fig. 37), it should be raised to genus rank.

Among the genera historically confounded with Cardiophorus, the most basal is Paracardiophorus. This genus was found polyphyletic (Table 3, Fig. 36: Nodes e, f & i, Fig. 37) because it includes superficially similar species from Australia and Chile. It is argued below that those should be part of a new genus. The type species of Paracardiophorus forms a fully supported clade with two North American Cardiophorus species (PP > 0.95 after correction for branch length (12), uncorrected probability =1.00), which is also indicated by parsimony analysis (Fig. 27, D = 2, BS = 57). These and all other North American species with the same apparent synapomorphies should be transferred to Paracardiophorus. These are the North American Cardiophorus with the base of the female spermathecal gland duct sclerotised (Figs 34, 35), some species also have truncate pronotal hind angles (Fig. 4), and the aedeagal parameres spatulate (Fig. 65). Paracardiophorus is the most likely (PP = 0.67) sister group of the remainder of Cardiophorinae.

Beyond confusion with Paracardiophorus, genus Cardiophorus: subgenus Cardiophorus was paraphyletic at four nodes (Fig. 36, f–i). It is also paraphyletic at 2 or more nodes in parsimony analysis, with most forming a polytomy in the parsimony analysis (Fig. 37). This large genus, which contains half the described cardiophorine species, is paraphyletic because it also includes 21 other genera (Fig. 36, Node h). Unfortunately, because of this poor phylogenetic resolution, there is little basis yet for an improved definition of Cardiophorus. The monotypic Cardiophorus: subgenus Zygocardiophorus Iablokoff-Khnzorian & Mardjanian, 1981 was found to be sister to Cardiophorus + the remainder of Cardiophorinae (Nodes f, g) and thus should be raised to genus rank. The position of Cardiophorus: subgenus Lasiocerus Buysson is unknown, because no specimens were available for examination.

Dolin and Gurjeva (1975) described genus Paradicronychus based on larval characters only (although conspecific adults were also known), and without a formal designation of a type species. Because of IZCN regulations for genera described after 1930 (Art. 13.3), Paradicronychus is not an available name. Although larvae of many cardiophorines from the former USSR are known, larval morphology of the world fauna remains too poorly documented to define genera based on larvae alone. Both analyses placed C. inflatus Candèze, 1882, (considered Paradicronychus by Dolin and Gurjeva (1975)) within the broadly paraphyletic Cardiophorus. Because of this result, and because no adult characters were identified to distinguish it from Cardiophorus, the nomen nudum name Paradicronychus should be placed as synonym of the nominate subgenus of Cardiophorus (with its included species to Cardiophorus as C. inflatus Candèze, 1882, and C. nothus Candèze, 1865).

Two other Cardiophorus subgenera, Coptostethus Wollaston and Perrinellus Buysson are each based on a single evolutionarily labile character (reduction of flight wings, and narrowed base of scutellum respectively), and are probably not monophyletic (although not synonymised here). The first, Coptostethus, is a name historically applied to various short-winged Cardiophorinae. Some Cardiophorus from Africa and Eurasia possibly adapted for fossorial life have been grouped into the subgenus Perrinellus, which was not recovered as monophyletic in either analysis (PP = 0.06, Table 3). Evidence that numerous other cardiophorines have similar modifications for digging may be further evidence these characters are convergent and this assemblage is artificial. While these genera remain non-monophyletic and weakly defined, I do not recommend taxonomic changes until their positions are better resolved.

Globothorax Fleutiaux and Teslasena Fleutiaux (Physodactylinae) are a strongly supported (Fig. 36. Uncorrected PP = 1.00, branch length = 2; Fig. 37, Bremer support = 2) clade within Cardiophorus and the other genera rendering it paraphyletic (Node g). Like the other genera here, these two should not be synonymised under Cardiophorus. Teslasena should be considered a junior synonym of Globothorax, because of the well-supported monophyly of these two species. This synonymy means that included species Teslasena femoralis (Lucas, 1857), Teslasena foucarti Chassain, 2005, and Teslasena lucasi Fleutiaux, 1899 are transferred to Globothorax as Globothorax femoralis (Lucas, 1857, Anelastes); G. foucarti Chassain, 2005; and G. lucasi Fleutiaux, 1899 respectively. Their sympatry in Brazil further supports the hypothesis that the known specimens of Teslasena and Globothorax are dimorphic males and females of one genus (although not necessarily conspecific). I recommend this despite characters presented by Rosa (2014) distinguishing the two genera: these may be variation between species, or sexual dimorphism but their presence does not refute the hypothesis that they are best understood as congeneric.

Dicronychus Brullé was coded here based on D. cinereus Brullé, which was considered the senior synonym of the type species at the beginning of this study. Although within the Paraphyletic Cardiophorus according to both analyses (Figs 36, 37), Dicronychus should not be a synonym of Cardiophorus, at least until a monophyletic Cardiophorus can be defined. However, since Dicronychus is only distinguished from Cardiophorus by the presence of a second tarsal claw tooth (Fig. 21), there may be no basis on which to distinguish it from Cardiophorus even at the species level because of apparent intraspecific dimorphism. Such dimorphic claws may underlie the sympatric Cardiophorus aptopoides Candèze, 1865; and C. brevis (Candèze, 1859) from Mexico, which appear identical except the presence or absence of a basal claw tooth (including aedeagal shape and regional colour variants). A similar otherwise apparently identical Cardiophorus-Dicronychus pair of species (Cardiophorus varius, Cate et al., 2002 and D. hoberlandti Cate et al., 2002) from Iran also may be a single species with dimorphic claws. There is no evidence for the monophyly of genera Dicronychus and Platynychus Motschulsky, 1858 (PP = 0.07, Figs 36, 37, Table 3), therefore Platynychus should be removed from synonymy under Dicronychus, where it has been placed by some authors. Platynychus is distinguished from both Cardiophorus and Dicronychus by its closed procoxal cavities.

Genus Cardiophorellus Cobos also falls within the paraphyletic nominate subgenus of Cardiophorus (Fig. 36, Node h, not contracted by parsimony, Fig. 37). The type species of Cardiophorellus is much like Cardiophorus except the anterior edge of its scutellum is broadly concave and not angulately emarginate, and its mandibles are simple. Due to phylogenetic uncertainty, there is no evident best taxonomic placement for Cardiophorellus. Because of this uncertainty, and because Cardiophorellus is readily diagnosed, it seems best to continue to consider Cardiophorellus a valid genus. The type specimen of the monotypic subgenus Cardiophorellus (Parapleonomus) Cobos, 1970 was not found at MNHN (Paris), so I cannot comment on its validity or rank.

Although the hypothesis of Aptopus Eschscholtz monophyly was rejected (Table 3, Fig. 36 (Nodes h & k), Fig. 37), this only affects the placement of the species A. agrestis (Erichson). Apart from its pectinate claws, this species is like Horistonotus species with costate elytral intervals. Because parsimony phylogenetic analysis suggested A. agrestis was the most likely sister taxon to Horistonotus simplex LeConte, such Aptopus species with carinae following the lateral edge of the pronotum should be transferred to Horistonotus. However, because the type specimen of A. agrestis was not examined, this species is not transferred to Horistonotus here. The concept of Aptopus used here is from modern authors (e.g. Aranda 1998, also Section 1 of Candèze 1860) because the type specimens of the type species, A. tibialis Eschscholtz, 1829 are lost or were unavailable for examination and because the only published species description lacks detail (eight words only).

Genus Phorocardius Fleutiaux was described to include Cardiophorus-like species with apically bidentate tarsal claws (Fig. 22, not Fig. 21), however the nominate subgenus + subgenus Diocarphus Fleutiaux are not monophyletic (Table 3). Therefore Phorocardius and Diocarphus should be recognized as distinct genera despite uncertainty about their positions in the poorly resolved nodes near Cardiophorus (Figs 36, 37). Tropidiplus Fleutiaux, 1903 is a distinctive east African genus among the genera rendering Cardiophorus paraphyletic. The hypothesis (Schwarz 1906) that Tropidiplus is a synonym of Craspedostethus was clearly rejected (Table 3, Fig. 36, also Fig. 37). Similarly Displatynychus Ôhira was a subgenus of Platynychus until Ôhira (1987) raised it to genus rank. Bayesian analysis (Fig. 36) supports separation of Displatynychus from Platynychus (not contradicted by parsimony, Fig. 37).

Genus Cardiotarsus includes species from Africa, Mauritius, S. and E. Asia and Australia. These analyses included the type species (C. capensis Candèze, 1860, known here from females only), another (undescribed) African species and Cardiotarsus mjobergi, Australia’s only known species. Bayesian hypothesis testing (Table 2) rejected the hypothesis that even the two African species were monophyletic (also not recovered by parsimony, Fig. 37). C. mjobergi was placed at Node k of the Bayesian tree (Fig. 36) within the southern clade (Fig. 36 node j). I propose transfer of Cardiotarsus mjobergi to genus Cardiodontulus, from Papua New Guinea, because of this non-monophyly and it matches the Van Zwaluwenburg’s definition of that genus. This placement is also plausible, because both are from the Australian biogeographic region. Although the type specimen of C. mjobergi was not examined, I am confident in the identification of the specimens examined because they were from near the type locality, which is in a well-collected area near a major insect collection, and this species was also illustrated in Calder’s (1996) guide to Australian Elateridae. Otherwise, I propose no changes to the biologically inaccurate (but easily diagnosable) genus Cardiotarsus until the phylogeny of Cardiophorinae is better resolved.

The remaining apical southern clade (PP = 0.81, Fig. 36: Node j) is composed mainly of Australian and Neotropical species, plus two South Asian genera and one from Africa. This clade was also inferred by parsimony (Fig. 37, but with Globothorax and Teslasena added), and includes mostly species with bilobed or multilobed proximal sclerites of the bursa copulatrix, and many of the species with closed procoxal cavities, and lacking lateral expansions of the parameres. Among these, the monophyly of each of Odontocardus Fleutiaux, 1931; Triplonychoidus Schwarz, 1906; Paraplatynychus Fleutiaux, 1931; Triplonychus Candèze, 1860; Cardiodontulus Van Zwaluwenburg, 1963; Craspedostethus; and Buckelater were not tested. These genera remain unaltered, except as discussed for Cardiodontulus. Two large, mainly Neotropical genera (extending into temperate North America) Esthesopus and Horistonotus are both not monophyletic (Table 3). However their definitions and status should be maintained until better resolution is available. The definition of Horistonotus is broadened here to include species with multiple claw points.

Of the five species in the weakly supported apical clade (Fig. 36, Node l), two belong to the polyphyletic genus Paracardiophorus. These two species from Australia and Chile are rendered paraphyletic (also at low posterior probability) by Buckelater, from Brazil. Because the included Australian and South American Paracardiophorus are identical in most characters including the male and female genitalia, I propose placement of them in a new genus along with other species from both continents sharing their diagnostic characters. The type species of this new genus, Austrocardiophorus, is Cardiophorus humeralis Fairmaire & Germain, 1860 from Chile (recently in Paracardiophorus). This solution is considered preferable to placement in the currently monotypic Buckelater because its female genitalic characters remain unknown, which contributes to taxonomic uncertainty.

This section outlines some character state changes implied by the trees (Figs 36, 37), which may be diagnostically helpful. While these characters may be true synapomorphies of their groups, Bayesian analysis does not rely on identifying them unambiguously as such.

Three characters unite the Negastriinae + Cardiophorinae. The fusion of the parameres at their midlength into a tube (Char. 117, Figs 24, 25) appears unique among the Coleoptera (Iablokoff-Khnzorian and Mardjanian 1981), and universal among Cardiophorinae and Negastriinae. Examination of two other possible synapomorphies revealed more intrageneric variability than found by Douglas (2011). Firstly, the hind-wing membrane has an anal notch (Fig. 17, at AA4) in all examined Negastriinae except Migiwa Kishii, 1966, but this notch is present in only most Cardiophorinae (Char. 75). Secondly, all included Negastriinae, except Arhaphes Candèze, 1860, but only most cardiophorine genera had a tridentate lobe at the midline of the posterior edge of the pronotum. The only character to distinguish the Cardiophorinae from the Negastriinae, was an apparent reversal to straight-sided prosternum (alternative = convex, Char. 36). No variation from this character-state was found in Cardiophorinae.

No clear evidence was found for basal synapomorphies of Negastriinae not also shared by Cardiophorinae. As found by Douglas (2011), they were distinguished from Cardiophorinae by their convex lateral edges of the prosternum (near midlength). However, this character is an apparent symplesiomorphy shared with Hypnoidus and Tropihypnus according to the most likely topologies identified by Douglas (2011).

Several synapomorphies unite three brightly patterned riparian negastriine genera from the Neotropics (Agrypnella), the Himalayan foothills (Cardiohypnus), and Australia (Rivulicola). These are the only Negastriinae with sublateral pronotal incisions and carinae (Char. 29). They are also the only Negastriinae, except for Monadicus, with: scale-like setae (Char. 23); and the posterior edges of hypomeron mesad of hind angles with rectangular or semicircular indentations (Char. 32). Two of these, Agrypnella and Cardiohypnus, also have sides of pronotum overhanging the lateral carinae like in Cardiophorus.

Quasimus Gozis, 1886 + Yukoana Kishii 1959 (both Negastriinae, Quasimusini) share several possible synapomorphies: tarsomere 4 (and no others) is lobed on all legs (shared in Negastriinae by only Neoarhaphes Costa 1966, Char. 94); pronotal hind angles with dorsal angle carina reaching anterior edge of pronotum (shared in Negastriinae with Monadicus and Agrypnella, Char. 27); parameres with two setae each (shared in Negastriinae with Arhaphes, Cardiohypnus, and Agrypnella, Char. 123); and pronotosternal sutures with anterior ends grooved (shared in Negastriinae with Monadicus Candèze, 1860, and Zorochros Thomson, 1859, Char. 37). Arhaphes + Neoarhaphes share two unique characters: bottle-shaped (lageniform) apical segments of the labial and maxillary palpi (Char. 19); partially or completely fused prosternum and pronotum (Char. 38); and also a tubercle at the posterior end of the mesosternal cavity (shared in Negastriinae with Migiwa only, Char. 52).

The Cardiophorinae have only two apparent synapomorphies not shared with at least some Negastriinae: the straight-sided prosternum (Char. 36); and presence of paired proximal sclerites in the bursa copulatrix (absent in Blaiseus, Craspedostethus, Floridelater (formerly Negastrius americanus), and Pachyelater, Char. 152). A third possible synapomorphy, the presence of one or two pedunculate anterior sacs of the bursa copulatrix (Char. 143) is shared by all examined Cardiophorinae and their apparent sister-taxon, Fleutiauxellus.

Within Cardiophorinae, only a few groups were united by moderate to high branch support. Pachyelater + Aphricus + undescribed species from New Zealand + Patriciella share straight sides of the mesosternal cavity posterior to anterior edge of mesocoxae (Char. 52). The Palaearctic Paracardiophorus + the Nearctic Cardiophorus cardisce + C. luridipes all share dorsally truncate pronotal hind angles.

Combined diagnosis of Cardiophorinae + Negastriinae

If procoxal cavities not closed to mesepisternum and mesepimeron, then scutellum emarginate anteriorly; hind wing without wedge cell; male aedeagus with paramere bases fused together into tube both dorsally and ventrally, articulated apicad of bases, or rigid (Figs 24, 25); female ovipositor without styli.

Plates

Figures 1–13. 

1–2 Frontoclypeal area Cardiophorus Scale bar = 0.5 mm). 1 C. convexus (Say), (anteroventral view) 2 C. gramineus, (dorsal view). CE = compound eye; SC = supra-antennal carina; SG = supra-orbital groove. Figures captions include post-revision names 3–4 Latero-ventral view of hypomeron Scale bar = 1 mm) 3 Cardiophorus gagates Erichson 4 Paracardiophorus propinquus Lanchester. Ant = anterior; LC = lateral carina 5 Lateral view of prosternal process of Cardiophorus erythropus Erichson Scale bar = 0.5 mm). A length from procoxa to dorsal apex; Ant = Anterior B length from procoxa to end at halfway between dorsal and ventral apices C length from procoxa to ventral apex D vertical distance between dorsal and ventral apices. Dorsal and ventral apices are considered points where profile of respective surface is 45º from horizontal 6–9 Dorsal view of scutellum of Cardiophorinae Scale bar = 0.5 mm) 6 Esthesopus castaneus 7 Blaiseus bedeli 8 Cardiophorus gramineus 9 Floridelater americanus 10 Latero-ventral view of mesepimeron and mesepisternum of Cardiophorus fenestratus (LeConte) showing measurement of angle (a) of anterolateral corner of mesepisternum Scale bar = 0.5 mm). MST = mesepisternum; MRN = mesepimeron 11–13 Ventral view of left side of mesocoxal cavity of Elateridae (After Arnett 1960, scale bar = 1 mm) 11Elaterinae sp. 12 Hypnoidus sp. 13Cardiophorinae sp. Lat = lateral; MCX = mesocoxae MRN = mesepimeron; MST = mesepisternum; MS = mesosternum. Captions reflect the revised classification.

Figure 14–22. 

14 Lateral view of metasternum of Zygocardiophorus nigratissimus Buysson Scale bar = 0.5 mm). LC = lateral carina; MTP = metepisternum; MTS = metasternum15–16 Dorsal view of anterior end of left elytron of Cardiophorinae Scale bar = 0.5 mm). 15 Cardiophorus gramineus 16 Paracardiophorus cardisce (Say) 17–18 Hind wing of Elateroidea Scale bar = 1 mm) 17 Anelastes druryi (Kirby) 18 Blaiseus bedeli. Vein names follow Kukalova-Peck and Lawrence (2004)19–20 Left metacoxal plate and metepisternum of Elateroidea Scale bar = 0.5 mm). 19 Aulonothroscus punctatus (Bonvouloir) 20 Cardiophorus gramineus. MCX = metacoxal plate; MTS = metepisternum21–22 Metatarsal claws of Cardiophorinae Scale bar = 0.5 mm). 21 Dicronychus cinereus 22 Diocarphus solitarius. Captions reflect the revised classification.

Figures 23–35. 

23–25 Dorsal view of aedeagus of Elateridae Scale bar = 0.5 mm). 23 Hypnoidus riparius Fabricius, 1792 [not Cardiophorinae] 24 Paracardiophorus cardisce 25 Blaiseus bedeli 28–31 Proximal sclerites of the bursa copulatrix of Cardiophorinae, interior view (top end of sclerite extends furthest into bursa) Scale bar = 0.5 mm). 28 Paracardiophorus cardisce 29 Paraplatynychus mixtus 30 Globothorax chevrolati 31 Esthesopus parcus Horn 32–33 Distal sclerites of the bursa copulatrix of Cardiophorinae, interior view Scale bar = 0.5 mm). 32 Cardiophorus inflatus 33 Cardiophorus gramineus 34–35 Sclerotised base of spermathecal gland duct of Cardiophorinae, lateral view Scale bar = 0.5 mm). 34 Paracardiophorus cardisce 35 Paracardiophorus musculus. SD = spermathecal gland duct. Captions reflect the revised classification.

Figure 36. 

Inferred phylogeny of Cardiophorinae and Negastriinae based on 159 adult morphological characters. Tree is a 50% majority-rule phylogram with branch lengths estimated by MrBayes, of 120000 post-burnin trees. Model = Mkv+G. Values are posterior probabilities. Scale bar indicates 0.1 changes per site. Several nodes are labelled a-l to simplify discussion. Taxon labels D, P, and N are members of Dendrometrinae, Physodactylinae and Negastriinae within the Cardiophorine. Genus name “C.” indicates Cardiophorus.Names are pre-revision. Type species of genera are marked with “*”.

Figure 37. 

Inferred phylogeny of Cardiophorinae and Negastriinae based on 139 parsimony-informative adult morphological characters (left), with Bayesian topology (adapted from Fig. 36) on right. Parsimony tree is a strict consensus of 412 most parsimonious trees of 1083 steps, CI = 0.17. Values above branches are decay indices, values below branches are bootstrap indices (above 50%). D (and colour purple), P (and colour orange), and N (and colour blue) are members of Dendrometrinae , Physodactylinae and Negastriinae respectively within the Cardiophorine clade (shown by colour alone elsewhere). Genus name “C.” indicates Cardiophorus. Names are pre-revision. Type species of genera are marked with “*”.

Figures 38–49. 

38–41 Chileaphricus chilensis lectotype male 38, 39 pterothorax 40 head and prothorax 41 aedeagus 42–44 Margogastrius schneideri paralectotype female. 42, 43 habitus 44 spermatheca, lateral view 45–47 Floridelater americanus. 45 aedeagus 46, 47 male 48–49 Blaiseus bedeli lectotype male. Scale bars: 1 mm, 0.5 mm for detail photos. Captions reflect the revised classification.

Figures 50–62. 

50–53 Pachyelater madagascariensis 50 female 51, 52 male 53 aedeagus 54–56 Aphricus sp. 54 aedeagus 55, 56 male 57–58 Neocardiophorus mamajevi. 57 male 58 aedeagus 59–61 Nyctor expallidus. 59 aedeagus 60, 61 male 62 Metacardiophorus sogdianus, male paratype. Scale bars: 1 mm for habiti, 0.5 mm for detail photos. Captions reflect the revised classification.

Figures 63–78. 

63–64 Metacardiophorus sogdianus, paratype 63 male 64 aedeagus 65–68. Paracardiophorus musculus. 65 aedeagus 66 proximal sclerite 67, 68 female 69–73. Zygocardiophorus nigratissimus. 69, 70 male 71 elytral apex 72 proximal sclerite 73 aedeagus 74–76 Platynychus indicus, female lectotype 74–75 adult 76 distal and proximal sclerites of bursa copulatrix, lateral view 77, 78 Globothorax chevrolati lectotype female. Scale bars: 1 mm for full habiti, 0.5 mm for detail photos. Captions reflect the revised classification.

Figures 79–89. 

79, 80 Globothorax femoralis lectotype male 81–84 Cardiophorus gramineus 81, 82 female 83 bursa copulatrix with distal and proximal sclerites, lateral view 84 aedeagus 85–87 Cardiophorellus gracilicornis paratype. 85, 86 male 87 aedeagus 88–89 Dicronychus cinereus female. Scale bars: 1 mm for habiti, 0.5 mm for detail photos. Captions reflect the revised classification.

Figures 90–104. 

90–92 Dicronychus cinereus. 90 proximal sclerite 91 distal sclerites 92 aedeagus 93–97 Aptopus rugiceps 93, 94 adult 95 proximal sclerite 96 distal sclerite 97 aedeagus 98–99Cardiophorus (Perinellus) argentatus, Lectotype. 98 male 99 aedeagus 100Cardiophorus (Coptostethus) femoratus, Lectotype 101–103 Phorocardius florentini 101, 102 Lectotype 103 bursal sclerites 104 Tropidiplus tellinii, female. Scale bars: 1 mm for habiti, 0.5 mm for detail photos. Captions reflect the revised classification.

Figures 105–121. 

105–108 Tropidiplus tellinii 105 female 106 tarsal claw with basal seta, lateral view (Lectotype male) 107 distal & proximal sclerites, internal view 108 aedeagus (Lectotype male) 109–112 Displatynychus adjutor 109, 110 female 111 distal sclerite 112 aedeagus. 113–115 Diocarphus solitarius. 113 distal and proximal sclerites in bursa copulatrix, internal view 114, 115 female 116–118 Cardiotarsus capensis 116, 117 male 118 bursal sclerites lectotype 119–121 Odontocardus vitalisi 119 distal sclerites of bursa copulatrix, internal view 120, 121 male. Scale bars: 1 mm for habiti, 0.5 mm for detail photos, 0.1 mm for tarsal claw. Captions reflect the revised classification.

Figures 122–139. 

122 Odontocardus vitalisi. Aedeagus 123–125 Triplonychoidus trivittatus paralectotype. 123 aedeagus 124, 125 male 126–131 Esthesopus castaneus 126, 127 male 128 ventral view of tarsomeres 4&5 129 tarsal claw 130 proximal sclerite of bursa copulatrix 131 aedeagus 132–135 Aptopus agrestis 132, 133 male 134 tarsal claw 135 aedeagus 136–139 Horistonotus flavidus 136, 137 lectotype female 138 proximal sclerite of lectotype 139 aedeagus Scale bars: 1 mm for habiti, 0.5 mm for detail photos, 0.1 mm for tarsal claw. Captions reflect the revised classification.

Figures 140–154. 

140–143 Paraplatynychus mixtus 140, 141 male 142 proximal sclerite 143 aedeagus 144–146 Triplonychus acuminatus, Lectotyp. 144, 145 male 146 aedeagus 147–150 Cardiodontulus brandti, male paratype 147 aedeagus 148, 149 adult 150 tarsal claw 151–153 Craspedostethus rufiventris 151 female Lectotype 152 lateral view of pronotal hind angle of female Lectotype 153 aedeagus 154 Craspedostethus culcarius , bursal sclerites of female labeled as type. Scale bars: 1 mm for habiti, 0.5 mm for detail photos, 0.25 for 142, 143. Captions reflect the revised classification.

Figures 155–162. 

155–157 Austrocardiophorus humeralis. 155 adult 156 proximal sclerite 157 aedeagus 158–160 Buckelater argutus 158 male 159 anterior view of head capsule of male 160 aedeagus 161–162 Ryukyucardiophorus loochooensis 161 female 162 proximal sclerite. Scale bars: 1 mm for habiti, 0.5 mm for detail photo. Captions reflect the revised classification.