﻿Mordellistenaplatypoda, a new species of tumbling flower beetle from the island of Ischia in Italy (Coleoptera, Mordellidae)

﻿Abstract Mordellistena A. Costa, 1854, the most species-rich genus of tumbling flower beetles comprises more than 800 species worldwide and more than 150 reported from Europe. Here, a new species Mordellistena(s. str.)platypoda is described from the island of Ischia in Italy. The species hypothesis is based primarily on morphological characters which are visualised using scanning electron microscopy images, high-resolution photographs, and drawings. The species hypothesis is supported by analysis of a 658 bp fragment of cytochrome c oxidase subunit I (COI). Divergences in the COI gene are evaluated using maximum likelihood and Bayesian inference analyses. The species delimitation is assessed using Assemble Species by Automatic Partitioning (ASAP) and Poisson Tree Processes (PTP) methods. Genetic distances are visualised using multidimensional scaling. Mordellistenaplatypoda Selnekovič, Goffová & Kodada, sp. nov. is recovered as a well-separated species by both molecular and morphological analyses. Our results show that M.platypoda Selnekovič, Goffová & Kodada, sp. nov. is most closely related to M.tarsata Mulsant, 1856, although the two species differ significantly in vestiture colouration, presence of lateral ctenidia on the third metatarsomere, and presence of sexual dimorphism on the protibia. The results indicate that such morphological differences, which were traditionally used to distinguish between species groups, may in fact be present between closely related species. Interestingly, examination of the numerous museum material did not reveal additional specimens of the new species, and therefore M.platypoda Selnekovič, Goffová & Kodada, sp. nov. is currently known only from the Italian island of Ischia.


Introduction
More than 800 species are currently classified within Mordellistena A. Costa, 1854, making it the most species-rich genus of tumbling flower beetles. It is reported from every continent except Antarctica. However, the generic placement of many species, especially Indomalayan and Neotropical, needs to be reassessed according to the current generic classification. In Europe, the genus is represented by more than 150 species (Horák 2020), which are pollinivorous in the adult stage and are among the most frequently encountered flower-visiting beetles. Larval stages are documented in approximately 18 European species (ca. 12%) and develop in the stems of herbaceous plants, feeding on the plant tissue (e.g., Borowiec 1996;Odnosum 2010;Zemoglyadchuk and Buialska 2020). The genus was generally characterised in terms of morphology by Ermisch (1950) and Franciscolo (1957). Identification of the European species is based on the concept of the species groups defined by combinations of morphological characters (Ermisch 1956(Ermisch , 1977. Due to the homogeneity in the external morphology, species identification is often challenging and is possible only by examination of the genitalia and comparison with the type specimens. Essential works covering the European representatives of the genus, including identification keys and figures of diagnostic characters, were provided by Costa (1854), Mulsant (1856), Schilsky (1894Schilsky ( , 1895Schilsky ( , 1898, Ermisch (1956Ermisch ( , 1963Ermisch ( , 1969Ermisch ( , 1977, Batten (1977), and Horák (1983Horák ( , 1990Horák ( , 1996. The comprehensive catalogue of Palaearctic fauna was provided by Horák (2008Horák ( , 2020. The phylogeny of the genus has not yet been studied. A recent collecting trip to the island of Ischia in Italy in June 2019 yielded more than 1,000 Mordellidae specimens representing 12 species. Within this material, we recognised a series of 52 specimens belonging to the new species described herein, Mordellistena platypoda Selnekovič, Goffová & Kodada, sp. nov. (Fig. 1). The aim of the present study was to describe the new species and delimit the species boundaries. To achieve this goal through an integrative approach, we tested the morphology-based species hypothesis with the results of single-gene DNA analyses. We documented the morphological characters using scanning electron microscopy, high-resolution photographs, and line drawings, and compared the new species with morphologically similar congeners. In order to analyse divergences in the mitochondrial gene that encodes cytochrome c oxidase subunit I (COI), we applied two probabilistic statistical methods (maximum likelihood and Bayesian inference) and compared the results with distancebased (Assemble Species by Automatic Partitioning) and tree-based (Poison Tree Processes) species delimitation analyses.

Materials and methods
This study is based on the examination of more than 400 specimens of the genus Mordellistena deposited at the following institutions: The specimens selected for DNA isolation were killed in 96.3% ethanol and stored at -20 °C. The remaining specimens were killed with the fumes of ethyl acetate. After DNA isolation, the specimens were soaked in 5% acetic acid, dissected, and mounted on cards. The dissected genitalia were cleared in lactic acid for several days, or in 10% KOH overnight, then dehydrated in 96.3% ethanol and mounted on slides in Euparal (Paradox Co., Cracow, Poland). After examination, genitalia were mounted on the card with the respective specimen using dimethyl hydantoin formaldehyde (Entomopraxis, Barcelona, Spain). Specimens were observed under an MZ16 stereomicroscope (Leica, Wetzlar, Germany) with magnification up to 120× with diffused LED light (6400 K). The drawings were prepared using a drawing tube attached to a DM1000 compound microscope (Leica, Wetzlar, Germany) and subsequently inked with the Isograph technical pens (Rotring, Hamburg, Germany). Photographs of habitus were taken with an EOS 5D mark II camera (Canon, Tokyo, Japan) attached to an Axio Zoom V16 stereoscope (Zeiss, Oberkochen, Germany); photographs of genitalia were taken with an Axio Imager 2 (Zeiss, Oberkochen, Germany). The images were stacked in Zerene Stacker 1.4 software (https://zerenesystems.com/cms/stacker) and edited in Adobe Photoshop CC (https://www.adobe.com/products/photoshop.html) and DxO Photolab 5 (https://www.dxo.com/dxo-photolab/). For scanning electron microscopy, the body parts were disarticulated, cleaned in lactic acid for several days, dehydrated in 96.3% ethanol, coated with 25 nm thick gold layer, and examined using Quanta 250 FEG (FEI Europe B.V., Eindhoven, The Netherlands). Measurements were made with an ocular micrometre in a MZ16 stereomicroscope (Leica, Wetzlar, Germany) and are given in the text as the range followed by the arithmetic mean and standard deviation enclosed in parentheses. The measured characters are abbreviated in the text as follows: EL elytral length from scutellar apex to elytral apices along suture; EW maximum elytral width; HL head length from anterior clypeal margin to occipital carina along midline; HW maximum head width; LPrL maximum left paramere length; PL pronotal length along midline; PW maximum pronotal width; PygL maximum pygidial length; RPrL maximum right paramere length; TL combination of head, pronotal and elytral lengths.
The species description follows the conventional terminology as used in Franciscolo (1957), Lu et al. (1997), and Lawrence and Ślipiński (2010). Terminology for sensilla types follows Snodgrass (1935). All nomenclatorial acts follow the regulations of the International Code of Zoological Nomenclature (International Trust of Zoological Nomenclature 1999).
In total, 56 specimens were used for DNA isolation. Details on the voucher specimens, including sampling localities and GenBank accession numbers, are presented in Table 1. DNA was isolated from the entire specimens using the E.Z.N.A. Tissue DNA kit (OMEGA Biotek Inc., Norcross, GA, USA) according to the manufacturer's protocol. The COI gene was amplified by PCR using standard primers LCO1490 and HCO2198 (Folmer et al. 1994). Individual PCR reactions were carried out in a total volume of 20 μl and included 10 μl of GoTaq Green master mix (Promega, Fitchburg, WI, USA), 0.52 μl of each primer (10 pmol/μl), 5 μl of extracted DNA, and 3.96 μl of nuclease-free water. The PCR thermocycler program was as follows: 94 °C for 120 sec, 40 cycles of 94 °C for 40 sec, 52 °C for 40 sec, 72 °C for 60 sec, and 72 °C for 10 min. PCR products were purified using EPPiC Fast (A&A Biotechnology, Gdansk, Poland) and sequenced from both sides in Macrogen Europe B.V. (Amsterdam, The Netherlands).
The consensus sequences, alignment, and final matrix were produced in Unipro UGENE 44.0 software (http://ugene.net/). Pairwise p-distances were calculated using MEGA X (Kumar et al. 2018). Maximum likelihood analysis (ML) was performed using IQ-TREE (Nguyen et al. 2015) on the web server (http://iqtree.cibiv.univie. ac.at) with the best substitution model (TIM2+F+I+G4) identified by the in-built ModelFinder according to BIC criterion. The node support values were obtained from 10,000 ultrafast bootstrap replicates (Hoang et al. 2017) and tested by the SH-aLRT branch test (Guindon et al. 2010). Bayesian inference was carried out in MrBayes 3.2.7a (Ronquist et al. 2012) on XSEDE available in CIPRES Science Gateway (htt-ps://www.phylo.org/index.php/). The number of substitution types was set to six, the nucleotide substitution model set to 4 × 4, and the among-site variation rate set to invgamma. Markov Chain Monte Carlo (MCMC) simulations included two independent runs each with four simultaneous chains, ten million generations, a sampling frequency of trees and parameters set to 1,000, and a burn-in fraction of 25%. The convergence of the MCMC analyses and an adequate sample size from the posterior distribution were confirmed using the in-built MrBayes diagnostics. Both trees were rooted subsequently in FigTree 1.4 (http://tree.bio.ed.ac.uk/software/figtree/), with Mordellochroa abdominalis (Fabricius, 1775) selected as outgroup. A distance-based species delimitation analysis, Assemble Species by Automatic Partitioning (ASAP) (Puillandre et al. 2020), was run on the ASAP web server (https://bioinfo.mnhn.fr/abi/ public/asap/) using uncorrected p-distances. A tree-based species delimitation analysis, Poisson Tree Processes (PTP) (Zhang et al. 2013), was run on the bPTP web server (https://species.h-its.org) with the ML tree used as an input, in 100,000 MCMC generations with 25% burn-in fraction. The distribution of intraspecific, interspecific, and intergeneric distances was calculated from uncorrected p-distances using an Automatic Barcode Gap Discovery (ABGD) analysis (Puillandre et al. 2012) carried out on the ABGD web interface (https://bioinfo.mnhn.fr/abi/public/abgd/) in 20 steps, with P min = 0.001, P max = 0.1, and relative gap width = 1.5. Multidimensional scaling (MDS) of uncorrected p-distances was performed in IBM SPSS Statistics software (https://www. ibm.com/products/spss-statistics).

COI gene analyses
We generated and analysed a set of 56 sequences of the 658 bp COI gene fragment. The dataset comprised 237 parsimony-informative sites (36%) and 22 singleton sites (3.34%). We recovered almost identical topologies with both maximum likelihood (ML) and Bayesian inference (BI) methods, and therefore only the ML tree is presented with ML bootstrap values and BI posterior probability values (Fig. 2). ML and BI analyses revealed ten well-separated monophyletic clades (excluding outgroup) with high bootstrap (> 93) and posterior probability (> 0.99) values. Each of these clades represents a single species, initially recognised based on morphological characters. Mordellistena platypoda sp. nov. and M. tarsata Mulsant, 1856 were grouped in one clade with high statistical support (99.6/1.00). Another well-supported clade with high statistical support (97.6/1.00) consists of M. purpurascens and M. hirtipes. The M. micans species group which is defined by combination of morphological characters (represented here by M. austriaca, M. hirtipes, M. minima A. Costa, 1854, M. platypoda, M. pseudorhenana Ermisch, 1977, andM. purpurascens) was recovered as polyphyletic, with members of the M. tarsata species group (M. tarsata) and the M. pumila species group (M. purpureonigrans) placed within. ASAP and PTP delimitation methods recovered ten species (excluding outgroup) in congruence with morphology-based delimitation, ML, and BI analyses (Fig. 2). The lowest ASAP score was 1.5.
The uncorrected p-distances within and between species are summarised in Appendix 1. The lowest interspecific distance within the entire data set is between M. platypoda sp. nov. (represented by a single haplotype) and M. tarsata with a range of 8.21% to 8.97%. The highest interspecific distance within the ingroup is observed between M. lindbergi and M. tarsata (19.15-19.76%). The highest intraspecific distance (1.06%) is observed in M. tarsata. The distribution of p-distances (Fig. 2) shows a distinct gap between the highest intraspecific distance (1.06%) and the lowest interspecific distance (8.21%). The multidimensional scaling of p-distances shows separate clusters for each species (Fig. 2). Differential diagnosis. Mordellistena platypoda is included in the M. micans species group as defined by Batten (1977) on the basis of the following characters: the four first antennomeres are narrower and shorter than the following ones (Fig. 3D); hind tibia besides the subapical ctenidium with four lateral ctenidia that are more or less perpendicular to the dorsal edge of the tibia (Fig. 4A); only the first two metatarsomeres with lateral ctenidia (Fig. 4A); punctation of elytra not conspicuously coarse; metatibial spurs black. A rather unique character that separates the species from most of its congeners is the shape of pro-and mesotarsomeres, which are flattened and expanded. The entire male mesotarsus and the female pro-and mesotarsus are dilated apically (Figs 4D, E, 5A, B). Similarly formed tarsi are present in M. rugipennis Schilsky, 1895 andM. latitarsis Batten, 1983, both    ♂♂ 3.85-5.37 mm, ♀♀ 3.67-5.33 mm). The latter species is separated from M. platypoda by almost square antennomeres 5-10, the lateral pronotal sides straight in the lateral aspect, the posterolateral pronotal angels obtuse, the elytra shorter (EL/EW ratio ≤ 2.2), and the genitalia differently shaped.
Mordellistena platypoda most closely resembles a sympatric species M. purpurascens but differs in paler vestiture, weakly expanded male second maxillary palpomere, with few long setae (Fig. 3A) compared to strongly expanded, with numerous long setae in M. purpurascens (Fig. 3C), weakly expanded male protibia, with few inconspicuous extended setae (Fig. 4B) compared to strongly expanded, with conspicuous long setae in M. purpurascens (Fig. 4C), protarsus expanded in both sexes and distinctly dilated apically in females (Fig. 5A, B) compared to weakly narrowed apically in M. purpurascens (Fig. 5C fig. 7). Mordellistena tenuicornis Schilsky, 1899 is separated from M. platypoda by the presence of two or three lateral ctenidia on the third metatarsomere, antennomeres 5-10 ca. 2.0× longer than wide, and pygidium distinctly longer and more slender.
The results of the COI gene analyses show the close relationship of M. platypoda with M. tarsata Mulsant, 1856, with p-distances of 8.21-8.97%. The two species are easily distinguished by the colouration of the dorsal vestiture, which is black with a greenish lustre in M. tarsata compared to yellowish in M. platypoda (Fig. 1), the presence of two lateral ctenidia on the third metatarsomere in M. tarsata compared to the absence of lateral ctenidia on the segment in M. platypoda, the presence of a conspicuous cluster of long setae in the proximal portion of the male protibia in M. tarsata compared to few elongated setae in in M. platypoda, and the different shape of the parameres. Body elongated, wedge-shaped, widest in anterior half of elytra (Fig. 1). Dorsum slightly convex, venter strongly so. Entire body surface uniformly black, except for reddish brown anteclypeus. Vestiture consisting of decumbent lanceolate setae (Fig. 4F); yellowish, darkened before elytral apices and along posterior margin of ventrites 4 and 5.
DNA sequences. Partial COI gene sequences of holotype and eight paratypes were submitted to GenBank (https://www.ncbi.nlm.nih.gov/genbank/). The accession numbers are listed in Table 1.
Etymology. The specific epithet is derived from the Greek words πλατύς (platýs), meaning wide, broad, and πόδι (pódi) meaning foot. It refers to the expanded proand mesotarsi, an unusual condition that separates M. platypoda from morphologically similar congeners.
Distribution. The species is known only from the island of Ischia in Italy.

Discussion
Mordellistena platypoda is morphologically well defined. It is included in the M. micans species group defined by Batten (1977). It should be noted, however, that the results of the molecular analyses presented in this study as well as other preliminary and yet unpublished results show that, despite the similarity in morphological characters shared by its members, the M. micans species group is polyphyletic, and therefore we mention it only marginally in this study to facilitate species identification. The recognition of the new species and the species hypothesis were based primarily on the unique combination of morphological characteristics, e.g., remarkably large body, expanded and dilated pro-and mesotarsus, length of antennal segments, inconspicuous secondary sexual dimorphism in the protibia, and unique shape of the genitalia. Expanded and dilated fore and middle tarsi are present in both sexes, but the condition is more pronounced in females. It is an uncommon character state present, among the European species, only in M. rugipennis Schilsky, 1895(Horák 1990) and M. latitarsis Batten, 1983. Both species differ from M. platypoda by the black vestiture pubescence on the dorsal surfaces. Another notable characteristic of M. platypoda is its large body that reaches up to 5.6 mm in males and 6.2 mm in females (TL). Comparably large species within the M. micans group are M. purpurascens A. Costa, 1854 andM. hirtipes Schilsky, 1895, which are sympatric with M. platypoda. Both species can be easily distinguished from M. platypoda by different shapes of the protarsus and mesotarsus, shorter antennal segments, distinct sexual dimorphism in the protibia, and different shapes of genitalia (Selnekovič and Kodada 2019).
The morphology-based species hypothesis was evaluated by analysing a 658 bp fragment of the COI gene. This standard DNA barcoding marker has frequently been used in the taxonomy of beetles not only to identify species as originally proposed (Hebert et al. 2003; DeSalle and Goldstein 2019), but also to explore species boundaries and refine morphology-based hypotheses (Pentinsaari et al. 2016;Fossen et al. 2016;Bergsten et al. 2017;Maddison and Sproul 2020). Despite the frequent use of this marker in beetle taxonomy, it has only been used once in the study of Mordellidae to identify species and interpret morphological variability (Selnekovič et al. 2021). Here, we recovered a phylogenetic tree using two probabilistic methods, maximum likelihood (ML) and Bayesian inference (BI). To assess the accuracy of the results, we compared them with those of distance-based (ASAP) and tree-based (PTP) species delimitation methods. ML and BI analyses show well-separated homogenous clades with strong statistical support for each presumed species including M. platypoda (ML bootstrap values higher than 97 and BI posterior probabilities 1.00) (Fig. 2). Mordellistena platypoda forms a clade with M. tarsata Mulsant, 1856 with strong statistical support (ML 97, BI 1.00). A close relationship between the two species is also supported by the interspecific divergence of 8.21%-8.87%, which is the lowest within the entire data set. Mordellistena platypoda and M. tarsata also have several morphological similarities, e.g., overall body shape, length of antennal segments, and shape of genitalia. However, despite the close genetic relatedness, M. platypoda and M. tarsata can be distinguished easily by the different colouration of the dorsal vestiture, which is black in M. tarsata and yellowish in M. platypoda, and by the absence of lateral ctenidia on the third metatarsomere in M. platypoda. Such characters were traditionally used to differentiate species groups among European representatives of the genus Mordellistena, for example, the M. micans group vs. M. tarsata and M. pumila group. Interestingly, our results show that such differences may, in fact, also be present among closely related species. It should be mentioned that our molecular analyses did not include species such as M. tenuicornis, M. latitarsis, and M. rugipennis, which share several morphological similarities with M. platypoda; therefore, future studies may provide new insights into the phylogenetic relationships between the aforementioned species.
The distribution of pairwise genetic distances (p-distances) shows a distinct gap between the highest intraspecific divergence (1.06%) and the lowest interspecific divergence (8.21%) (Fig. 2). Genetic distances between the analysed Mordellistena species range from 8.21% to 19.76%, similar to the previous analyses (Selnekovič et al. 2021). The multidimensional scaling of the p-distances shows separate clusters for each species (Fig. 2).
Mordellistena platypoda was collected in a relatively large series of 52 specimens during one collecting event on the island of Ischia in Italy. Revision of the museum specimens identified as M. micans (Germar, 1817), M. stenidea Mulsant, 1856, andM. grisea Mulsant, 1856 in the Franciscolo collection in MSNG, a series of M. micans in SDEI, and a series of M. micans in HNHM did not reveal any additional specimens of the new species. Therefore, M. platypoda can now be considered endemic to the island of Ischia. However, given the island's proximity to the shore (ca. 30 km) and the similarity of its fauna to that of the mainland, it is possible that M. platypoda may also occur on the Italian mainland or surrounding islands. Table A1.