Revision of taxonomic status of Anthrenus pimpinellae isabellinus (Coleoptera: Dermestidae)

. For 160 years, Anthrenus pimpinellae isabellinus Küster, 1848 has been considered a subspecies of A. pimpinellae Fa-bricius, 1775. However, habitus shape differs between the subspecies with A. p. isabellinus being broader than A. p. pimpinellae and resembling more closely A. dorsatus Mulsant & Rey, 1868. Here A. p. pimpinellae and A. p. isabellinus , are examined to look for evidence that they comprise a single taxonomic unit. Habitus and genital structures are considered, and the universal animal barcode region of the mitochondrial cytochrome oxidase I gene is sequenced. The results of the morphological, morphometric, and genetic analyses mirror each other perfectly and suggest that A. p. isabellinus is the same species as A. dorsatus rather than being a subspecies of A. pimpinellae . The very small intraspeci ﬁ c DNA sequence variation supports the view that A. dorsatus and A. p. isabellinus belong to a single species that diverges considerably from A. p. pimpinellae . Morphology, including genital structure, is congruent with the genetic data and provides a powerful way of resolving species organisation in these widespread beetles. In view of these ﬁ ndings, Anthrenus isabellinus Küster, 1848 is restored to full species status and Anthrenus dorsatus Mulsant & Rey, 1868 becomes its new junior subjective synonym.


INTRODUCTION
The Dermestidae is a moderately large family of beetles with the number of species currently claimed to lie between 1600 and 1700 (Háva, 2015(Háva, , 2020, but the taxonomy of parts of the family is poorly understood. Anthrenus is a relatively large genus within the Dermestidae numbering about 260 species (Háva, 2020). Anthrenus provides an example of complex, unresolved, taxonomy and is split into 10 subgenera. Kadej (2018) carried out an examination of the genus Anthrenus using larval characteristics and concluded that only the species within the subgenus Anthrenus were monophyletic, all other subgenera forming a polyphyletic assemblage.
Even though the subgenus Anthrenus appears to be monophyletic, it is not without its diffi culties. Notably, it gene) to establish phylogenetic relationships among the species. The nomenclature and zoogeography follow Háva (2015), and the conventional nomenclature, including of the taxa under study, is used.

Study insects
Material was collected from around Thessaloniki (Greece), Mallorca (Spain), and Maryland (the United States of America), supplemented with preserved specimens from the Natural History Museum (NHM), London. From the fi eld, specimens were almost exclusively collected from white fl owers such as Hoary Cress (Lepidium draba L., Brassicaceae) and Hemlock (C. maculatum) (see Holloway et al., in press). Dermestidae were knocked from the fl owers into a plastic tray to facilitate aspiration using a pooter. All fi eld collected specimens along with those from the NHM were used for the morphological analysis.
Only in Greece were A. p. pimpinellae, A. p. isabellinus and A. dorsatus found together in the fi eld, along with Anthrenus scrophulariae albidus Brullé, 1832. In the laboratory, the Greek insects collected from Sindos, near Thessaloniki (40. 673368N, 22.806583E) were separated by species/subspecies and retained on a mixture of dead insects, feathers, and bone and blood meal. When the adult insects died, they were removed from the breeding medium and stored in 2% acetic acid until dissection. F1 offspring of each species/subspecies were reared through to produce insects for genetic analysis. When insects emerged from their pupal cases, they were fl ash frozen at -30°C then stored in 99% ethanol to preserve the DNA. Prior to homogenisation to extract the DNA, some specimens were quickly dissected (see morphometric analysis) under a Brunel BMSL zoom stereo LED microscope to confi rm species identity.
(1862) concluded that A. p. pimpinellae and A. p. isabellinus were conspecifi c. No examination of the genitalia appears to have been carried out, the colour patterns differ, and the distributions overlap. Anthrenus p. pimpinellae is claimed to have a cosmopolitan distribution (Háva, 2020), whereas A. p. isabellinus is distributed around the western Mediterranean (Háva, 2020). In other words, the subspecies are sympatric in the western Mediterranean and, according to theory (Mallet, 1995(Mallet, , 2008, should not be able to retain integrity.
In previous studies, A. dorsatus Mulsant & Rey, 1868, another species in the A. pimpinellae complex, has been noted both in Greece  and in USA (Holloway et al., in press

MATERIALS AND METHODS
A three-pronged approach is taken: (1) to examine the genital structure, supplemented by other features such as antennal structure, (2) to carry out a morphometric examination, and (3) to sequence a fragment of the mitochondrial COI gene (the universal animal barcode region of the mitochondrial cytochrome oxidase I Table 1. List of DNA sequenced Anthrenus specimens whose parents were collected at Sindos (Greece). "% ID to nominal species" is the GenBank % DNA sequence identity to the colour-based species identity, except in A. dorsatus (values asterisked) whose % sequence ID is to Anthrenus pimpinellae (the highest GenBank sequence ID to all Anthrenus dorsatus sequences) because GenBank does not contain any Anthrenus dorsatus sequences. ND -sex not determined. Reference specimens (parents and additional F1 offspring from the same breeding vial) for sequenced individuals are located in the private collection of G.J. Holloway. The following specimens were used in the morphometric analysis: Anthrenus pimpinellae pimpinellae: Greece 9 males (fi eld collected, Thessaloniki), France 1 male (NHM collection); Anthrenus pimpinellae isabellinus: Mallorca 2 males (fi eld collected, Pollensa), Greece 1 male (fi eld collected, Thessaloniki), US 1 male (fi eld collected, Maryland), Spain 1 male (NHM collection), 1 male labelled 'Europe' (NHM collection), Algeria 1 male (NHM collection), Morocco 1 male (NHM collection); A. dorsatus: Greece 9 males (fi eld collected, Thessaloniki), Mallorca 9 males (fi eld collected, Pollensa), Spain 1 male (NHM collection).

Label on
Identifi cation was based on genital and antennal structure, along with habitus body plan and colouration. Dissection involved detaching the abdomen from the rest of the insect using two entomological micropins. The soft tergites were then peeled off the harder ventrites to expose the aedeagus. Sternite IX was also detached from the aedeagus. Images were taken using a Canon EOS 1300D and stacked by Helicon Focus 6-Pro focus stacking software. Habitus images were captured at ×20 magnification and images of the antennae were recorded at ×63 magnification. Images of the genitalia were captured at ×100 magnifi cation using a Brunel monocular SP28 microscope. Morphometric measurements were taken using DsCap.Ink software. Identifi cation was confi rmed using Kadej et al. (2007),  and . All statistical analyses were carried out using Minitab (version 19) software. To increase stringency in statistical analyses between the same pair of taxa, Bonferroni correction was applied to P values (Rosenthal & Rubin, 1987). Following correction, signifi cance levels p < 0.05, p < 0.01, and p < 0.001, are indicated as *, **, and ***, respectively. Non-signifi cant results are indicated ns. Means and standard errors are provided.
The following measurements were taken: 1. Body length (BL) (front edge of pronotum to tip of elytra) 2. Body width (BW) (maximum width across the elytra) 3. Paramere length (PL) (from the posterior tip of the paramere to the to the anterior end where the parameres meet)

Genetic analysis
DNA was extracted using a high salt protocol (Paxton et al., 1996) from 25 adult beetles, bred in captivity from insects col-lected from Sindos, near Thessaloniki (Greece), comprising A. dorsatus (n = 5), A. p. isabellinus (n = 7), A. p. pimpinellae (n = 8) and A. scrophulariae albidus (n = 5; see Table 1). Sequenced individuals comprised a mix of males and females that were determined to species based on colour and morphology (Table 1). Entire insects were crushed for DNA extraction; pinned reference specimens relating to these samples are the parental generation and other F1 offspring of the same breeding vial (G.J. Holloway, private collection). The 25 samples were DNA barcoded at COI, using standard protocols recommended by BOLD (http:// www.barcodinglife.org) with the animal barcode oligonucleotide PCR primers LCO/HCO (Folmer et al., 1994). DNA sequences were used to interrogate NCBI's database using BLAST (https:// blast.ncbi.nlm.nih.gov/Blast.cgi) and the BOLD COI database (http://www.barcodinglife.org). Beetles bred for sequencing were uniquely numbered (Table 1) and therefore genetic analysis was undertaken blind to species or subspecies identity.
Phylogenetic and evolutionary analyses of the 25 sequences plus 7 additional reference sequences, including conspecifi cs [A.
(Anthrenus) festivus Erichson, 1846 from France, A. (A) pimpinellae from Germany, A. (A) scrophulariae (Linnaeus, 1758) from Germany, and two species of different subgenera (two sequences each of Anthrenus (Nathrenus) verbasci (Linnaeus, 1767) from Germany and Canada, and Anthrenus (Florilinus) museorum (Linnaeus, 1761) both from Germany], retrieved from the NCBI (GenBank) database, were conducted using MEGA version X (Kumar et al., 2018;Stecher et al., 2020). Sequences were aligned using ClustalW which, after removal of gaps, revealed a single open reading frame in all sequences of 575 bases (191 amino acids), suggesting sequence quality was good. Substitution model selection by Maximum Likelihood showed the best model based on AICc and BIC to be the TN93+G+I (Tamura-Nei model with a discrete Gamma distribution to model evolutionary rate differences among sites, allowing for some sites to be evolutionarily invariable), which was then employed to generate a phylogenetic tree by Maximum Likelihood, with 500 bootstrap replicates to estimate support for branches. The 25 new sequences generated in this study are publicly available in BOLD (Table 1).  The colour patterns of A. p. pimpinellae (Fig. 1a) and A. dorsatus (Fig. 2a) consist of black, white and orange scales in relatively similar distributions across the elytra. Differences include the width of the trans-elytral white band (narrower in A. p. pimpinellae than A. dorsatus), tightness of packing of scales of the white band (spaced apart in A. p. pimpinellae, overlapping in A. dorsatus), and the fi nger of white scales joining the posterior edge of the white band to the middle lateral elytral white spot (broken or rudimentary in A. p. pimpinellae, complete in A. dorsatus). The elytral colour pattern of A. p. isabellinus (Fig. 3a) is distinctive, and much of the patterns shown in Figs 1a and 2a are covered in creamy coloured scales, tightly packed at the base of the elytra and becoming more scattered towards the elytral apices.

Figs 1, 2 and 3 show typical examples of
The ventrites of A. p. pimpinellae (Fig. 1b) are off-white as a result of the mixing of white with pale brown scales. At the lateral margins of the sternites are large patches of black scales. The patch on sternite I is very large, meets the lateral margin, and has no white scales on its anterior margin. The ventrites of A. p. isabellinus (Fig. 2b) and A. dorsatus (Fig. 3b) are very similar in appearance; the bright white scales are closely packed together (in A. p. pimpinellae the scales are more spaced), the patches of black scales along the lateral margins of each sternite are smaller than those of A. p. pimpinellae, and the patches of black scales on sternite I are sub-lateral, small and surrounded by white scales.
The aedeagus of A. p. pimpinellae (Fig. 1d) expands from the base to the apex ending in broad, heavily hooked parameres that are covered in shaggy hairs on the dorsal surface. The aedeagi of A. dorsatus (Fig. 2d) and A. p. pimpinellae (Fig. 3d) are very similar to each other. They are narrower than A. p. pimpinellae, do not expand much from the base to the apex, are not as broad and hooked as A. p. pimpinellae, and have shorter, sparser hairs on the dorsal surface.
The sternite IX of A. dorsatus (Fig. 2e) and A. p. isabellinus (Fig. 3e) are very similar to each other and show fl aps between the anterior horns. Sternite IX of A. p. pimpinellae (Fig. 1e) does not show these fl aps (Kadej et al., 2007).  Fig. 4 shows a plot of PL on BW/BL for all specimens of all taxa together. There is an obvious split in the data, with the values for A. p. pimpinellae occupying a different area within the plot than A. dorsatus. The values for the A. p. isabellinus specimens nestle comfortably within the A. dorsatus data points and are clearly removed from those of A. p. pimpinellae.

Genetic analysis
All 25 sequences gave high GenBank query coverage (> 97%) and closest sequence identity (> 85%) to Anthrenus species (Table 1). The closest BLAST hits in GenBank and the closest BOLD match of A. p. pimpinellae and A. s. scrophulariae were to their respective species, consistently with high sequence identity of > 98% (Table 1). GenBank and BOLD databases lack reference barcode sequences for A. dorsatus and A. p. isabellinus. Both A. dorsatus and A. p. isabellinus had GenBank closest hits to A. pimpinellae, but both with a relatively low sequence identity of < 87% (Table 1). BOLD returned 'no match' for both taxa due to lack of close identity to any reference sequence in the BOLD database. Phylogenetic analysis of our CO1 barcode sequences (Fig. 5)

DISCUSSION
Here we examined two forms that have been accepted as subspecies for nearly 160 years, but one subspecies (A. p. isabellinus) was believed to exist wholly within the spatial distribution of the nominotypical subspecies (A. p. pimpinellae). We found that A. p. isabellinus specimens from a variety of geographical locations bear little resemblance to A. p. pimpinellae. They are, without question, different species. Küster (1848) originally described A. isabellinus as a full species and Beal (1998) speculated that A. p. pimpinellae and A. p. isabellinus were different species. This study has demonstrated that both Küster and Beal were correct.
We examined the specimens using three techniques: morphological comparison, morphometrics, and genetic analysis. Gratifyingly, all three approaches arrived at the same conclusion, that A. p. isabellinus is not closely related to A. pimpinellae and resembles A. dorsatus in all respects. This fi nding is signifi cant since not all workers are able to make genetic comparisons, but morphological and morphometric comparisons are more widely available. Genital structure is one of the most important characters used by taxonomists to differentiate among species (Arnqvist, 1998), and we demonstrate here that careful examination of the genitalia is important in Anthrenus species identification. Moreover, morphological and morphometric comparisons can produce results as sound as those based on DNA sequence data.
Beal (1998) suggested that we probably have a poor understanding of the distribution of the nominotypical A. p.  (Tamura & Nei, 1993). The tree with the highest log likelihood is shown, and the percentage of trees (> 80%) in which the associated taxa clustered together is shown next to the branches (500 bootstraps). The tree is drawn to scale, with branch lengths measured in number of substitutions per site (scale bar as %) and is midpoint rooted. pimpinellae due to the degree to which workers confused different species from the A. pimpinellae complex. Anthrenus p. pimpinellae is claimed to be almost cosmopolitan. Specimens held in the NHM would suggest that this claim might not be true (G.J. Holloway, pers. obs.). Holloway & Bakaloudis (2020) showed that there is little intra-specifi c variation in colour pattern in A. p. pimpinellae. The current study demonstrated that A. p. pimpinellae is narrow in shape relative to A. isabellinus, with brownish versus clean white ventrites, respectively. These features, body shape and colour pattern, can be quite easily assessed under fi eld conditions. It is hoped that fi eld workers will be able to utilise the information provided here to help to understand the true distribution of A. p. pimpinellae.
The majority of A. isabellinus both from the fi eld (a sample of over 500 insects from across Europe, GJH unpubl. data) and others reared through in the laboratory have a colour pattern resembling Fig. 2a (80% of individuals). The remaining A. isabellinus possess more whitish elytral scales (20% of individuals) along a continuous gradient of increasing numbers of paler elytral scales to the palest specimens as shown in Fig. 3a. Furthermore, insects with the colour patterns shown in Fig. 2a produce some offspring resembling Fig. 3a, whilst parental insect resembling Fig.  3a produce some offspring resembling Fig. 2a. There is no evidence that the different colour patterns are the result of genetic polymorphism. A more parsimonious explanation for the colour pattern gradient is phenotypic plasticity. Colour pattern plasticity is very common across a range of insect taxa. Many Lepidoptera display wing scale colour plasticity in response to developmental period (Holloway et al., 1993;Kemp & Jones, 2001), with paler wing scales produced by individuals with the shortest developmental periods. In fact, colour pattern plasticity in response to developmental period is common across many insect groups, including Coleoptera (Holloway et al., 1995;Michie et al., 2010), Diptera (Marriott & Holloway, 1998;Gibert et al., 2007), and Hemiptera (Sorokor et al., 2013;Sibilia et al., 2018). In many cases it has been shown that colour pattern plasticity in insects has adaptive signifi cance (Brakefi eld & Reitsma, 1991;Ottenheim et al., 1999;Sibilia et al., 2018). More work is required to establish why and how the variation in colour pattern in A. isabellinus is produced.