Taxonomy , distribution and bionomics of Celonites tauricus Kostylev , 1935 , stat . n . ( Hymenoptera , Vespidae , Masarinae )

Male and female of Celonites abbreviatus tauricus Kostylev, 1935 are redescribed and a neotype is designated. Based on morphological characters Celonites a. tauricus is synonymized with Celonites spinosus Gusenleitner, 1966 and Celonites abbreviatus invitus Gusenleitner, 1973. The taxon is hypothesized to be reproductively isolated from Celonites abbreviatus Villers, 1789 by differences in the male genitalia and in the colour pattern of the male antennae and is therefore regarded as a separate biospecies named Celonites tauricus. Celonites tauricus is allopatrically distributed with regard to C. abbreviatus and has been recorded from the Crimea, Kos, Asia Minor and Cyprus. Within this range six intraspecific taxa can be separated by morphological characters and colour patterns. Habitat, flower association, flower visiting behaviour, mate seeking behaviour and nesting of C. tauricus are almost similar to C. abbreviatus.


Introduction
Nearly 100 years ago the first specimen of a species of the pollen wasp genus Celonites Latreille, 1802 was recorded from the Crimea. It was a single female collected by V.N. Wuczeticz on 13 June 1916 on the southern slope of Karagach in the Karadag (Kostylev 1928;original spelling "Karagatsh") which is a small mountain massive on the south-eastern coast of the Crimea. Kostylev (1928) identified the specimen as Celonites abbreviatus (Villers, 1789). Seven years later he formally described the Celonites taxon from the Crimea as a new subspecies of C. abbreviatus named Celonites abbreviatus tauricus Kostylev, 1935. He noted that specimens from the Crimea differed from specimens from the Caucasus by finely serrate margins of the second and subsequent terga and concluded that they possibly belong to a different subspecies (Kostylev 1935). Kostylev did not designate any type specimens of the subspecies either in his original description or in any further publication. Moreover, it is unknown how many specimens were available to him, when he described the taxon. In his world checklist of masarine wasps Carpenter (2001) recognized C. a. tauricus as a valid taxon. He speculated that the holotype of C. a. tauricus might be in the collection of the Zoological Institute of the Russian Academy of Sciences in Saint Petersburg (Russia), but we could not find it there, though carefully searching for it. The holotypes of all other wasp species described by Kostylev are deposited in the Zoological Museum of the Lomonosov Moscow State University in Moscow (Russia). These types are stored in a separate box apart from the main collection. There is no specimen of Celonites from the Crimea in the "type box" or in the main collection, whereas all other wasps from the Karadag listed by Kostylev (1928) are there. It can be assumed, that Kostylev (1935) based his description on the female collected by Wuczeticz. Therefore, the type locality was probably situated in the Karadag Mountains (as reported by Kostylev 1928). Evidently Wuczeticz's specimen and any further material (if it had ever existed) have been lost.
In 1966 Gusenleitner described a new species belonging to the Celonites abbreviatus-complex from western Turkey, named Celonites spinosus. He argued that this taxon could not be identical with C. a. tauricus Kostylev, based on the incorrect translation or interpretation of Kostylev's description, that antennal article A3 should not be longer than A4+A5 in C. a. tauricus. In fact, Kostylev attributed this character state explicitly to a specimen from Spain and stated that A3 is longer than A4+A5 in C. a. tauricus, which is similar to C. spinosus. Moreover, in both taxa the posterior margins of the metasomal terga II-V were described as "serrated" or "spine-like crenulated" respectively (Kostylev 1935, Gusenleitner 1966, indicating further similarity between them.
Another subspecies of Celonites abbreviatus named invitus was described by Gusenleitner in 1973 from central Turkey and Armenia. He separated this taxon from C. abbreviatus by the very shiny cuticula of vertex and mesosoma and the incomplete colour banding on the metasomal terga III-V. As a result of the investigation of further material, Gusenleitner in 1985 recognized C. a. invitus as an eastern form of C. spinosus and synonymized both taxa formally.
A specimen of Celonites abbreviatus tauricus was first collected from the Crimea in 1916, but none were recorded in the following 86 years. Then in 2002 two specimens were collected in the Karadag Nature Reserve ). Following this find intensive field research led to the discovery of nine localities of C. a. tauricus on the Crimean peninsula (Fateryga and Ivanov 2010). In addition, it was discovered that a previously unnoticed specimen of C. a. tauricus had been collected in the Crimea in 1963 , indicating the continuous existence of an endemic C. a. tauricus population in the area. Further field observations provided some bionomical data concerning flower associations and nesting , Fateryga 2010, Fateryga and Ivanov 2010. However, until now the taxonomic identity and status of C. a. tauricus with regard to C. abbreviatus and C. spinosus had not been studied and remained uncertain.
The purpose of the study presented below was to revise the taxonomic status of Celonites abbreviatus tauricus, to designate a neotype, to describe both sexes and to summarize and discuss what is known of the distribution, geographical variation and bionomical characters of the taxon.

Material and methods
The specimens of Celonites studied belong to the public collections of the Taurida  The specimens were investigated under a WILD M3 stereo microscope (maximum magnification 80 times). Measurements of the exoskeleton were made using an ocular micrometer (highest resolution 0.011 mm). The genitalia of all males were extracted after re-softening the specimens and were then studied in 80% ethanol. Drawings were made with a drawing tube (WILD Type 308700). Micro-photos were taken with a Leica IC 80 HD camera mounted on a Leica MS 5 stereomicroscope. Multifocuspictures were generated with Leica Application Suite (LAS) software. The scapus of the antenna is referred to as antennal article A1 and the pedicellus as A2, the flagellum consists of the articles A3-A12. The Kos population was studied by V. Mauss from 25 May to 4 June 2015 (for details see material studied). Observations were made with a close-up binocular (Pentax Papilio 8.5×21) and documented with a Canon EOS 70D with a 180 mm macro-lens (scale up to 1:1, resolution 20 mega pixel) and macro flash-lights. Time intervals were measured using a digital stop-watch. Specimens of flowering plants were collected and preserved dried. The material was placed in the herbarium of the Stuttgart State Museum of Natural History (Herbarium STU). The plant taxa were identified following Pils (2006) and Tutin et al. (1964Tutin et al. ( -1980. Flower preferences of imagines were studied by counting the number of sightings (= first observations) of flower visiting individuals while walking randomly across the area at six localities (total investigation time 11.30 h). Flower visiting behaviour of the imagines at Satureja thymbra and Thymus capitata was investigated at various patches of these plants at six localities for 19.45 h in total. From three localities pooled pollen samples from the crops of up to three females previously fixed in Duboscq-Brasil solution (Romeis 1989) were prepared using the method outlined by Westrich and Schmidt (1986). The different pollen types were ascertained under a light microscope at magnifications of 400× or 1000× and determined to generic level with the aid of a reference collection consisting of pollen samples of 500 mainly Mediterranean plant species.
Vicinity of Feodosiya: Karadag Nature Reserve, 44. 91667°N, 35.21667°E, 07.07.2002 1♂ (dbM No. 4704) 1♀ (dbM No. 4702) leg. S. Ivanov VM;Vicinity of Feodosiya: Lisya Bay, 44.90000°N, 35.15000°E, 22.06.20031♀ (dbM No. 4306) leg. S. Ivanov VM 1♀ leg. S. Ivanov VTNU 1♀ leg. S. Ivanov IZAN, 06.07.20051♀ leg. A. Fateryga VTNU 1♀ leg. A. Fateryga IZAN, 13.06.20072♀♀ leg. A. Fateryga VTNU, 25.06.20091♀ leg. S. Ivanov VTNU 1♀ leg. S. Ivanov ZIN, 27.06.20091♀ (dbM No. 4701) leg. A. Fateryga VM 1♀ leg. A. Fateryga VTNU, 12.06.20101♂ (dbM No. 4308) leg. A. Fateryga VM, 16.06.20101♀ (dbM No. 4305) leg. A. Fateryga VM, 09.07.2012 Diagnosis. Axilla of mesoscutellum with short blunt lateral projection that only slightly projects over adjacent posterior part of tegula. Frons and clypeus covered with pale, stiff pollen collecting setae, about as long as diameter of median ocellus. In females most of these setae with tiny spherical enlargement at tip ("knob"), in males setae with distal enlargement only present in centre of frons. Cuticula of frons and clypeus dull and densely shagreened. Males with only two oval-shaped tyloids situated ventral on articles A9 and A10 of club-shaped antennae and small spine anterior at distal end of midcoxa. Posterior margin of metasomal tergum VII divided into four lobes. Separated from Celonites abbreviatus by distinctly different colouration of club of antenna: Club has a dark tip, that is, at least distal end of A12 is blackish markedly contrasting on ventral side to adjacent light reddish brown area of antennal club. On dorsal side blackish marking extends usually over distal parts of A11 fading gradually towards proximal end. In C. abbreviatus club of individuals from Balkan populations completely orange, in some dark coloured individuals from western populations club becomes darker dorsally, while immediate tip and especially ventral side of A12 remain lighter. Male genital broader than in C. abbreviatus, in dorsal view transverse width of each stipes larger than distance between dorso-medial margins of stipites. Medial process of volsella larger than in C. abbreviatus.
Description. Female. Colour: Black. The following are weakly yellowish white ( Fig. 1a): two spots on frons; small narrow streak on occiput along occipital carina at dorso-lateral corner of head (absent in two specimens); large spot on antero-dorsal angle of pronotum (humeral spot); stripe along dorso-medial (inner) margin of pronotum, slightly enlarged anteriorly, interrupted in the middle by zone of reddish-brown colour; large spot on dorsal mesopleura; median spot on scutellum of moderate size; dorsal and ventral side of propodeal lamella; continuous posterior band on tergum I extending over lateral margin on ventral part of tergum; posterior bands interrupted on each side of middle into median and two lateral markings on terga II-V (in two specimens not completely interrupted on tergum II). Reddish-brown are: distal two third of mandible; labrum; zone in the middle of whitish stripe along dorso-medial margin of pronotum; tegula except small blackish marking at antero-medial margin; distal part of femora, tibiae and tarsi; sternum I; sternum II, becoming slightly darker posteriorly. Dark to blackish brown are: strongly sclerotized parts of labio-maxillary complex; propleura; coxae, trochanters and proximal part of femora; sterna III-V. Antenna with A1-2 black, distal margin of A2 dark brown, A3 proximally dark brown distally reddish brown, A4-11 reddish brown. A12 at least with distal tip black, on ventral side markedly contrasting to adjacent light reddish brown proximal area of antennal club, dorsally extending over distal part of A11 (except in one specimen) fading gradually towards proximal end. Wings translucent blackish-brown.
Structure: Head in front view slightly longer than broad. Clypeus a little broader than long; dull with somewhat sparsely weak macropunctation; densely shagreened; covered with pale, stiff setae arising from weak macropunctures; setae about as long as diameter of median ocellus, with tiny spherical enlargement at tip ("knob") ( Fig. 2), few somewhat shorter; some setae on ventro-lateral corners with distal ends curved towards median axis of clypeus; dorso-lateral vertical parts of clypeus more shiny not shagreened with moderately spaced micropunctures. Frons with very weak moderately spaced macropunctation; cuticula dull, densely and strongly shagreened; covered with pale outstanding knobbed setae arising from macropunctures; frontal line weak or absent, not raised to form medial carina. Vertex with close macropunctation becoming more distinct and closely reticulate behind ocelli (Fig. 3), where longitudinal interstices are more strongly raised forming lines; cuticula of interstices shiny, weakly longitudinally striated anterior to ocelli, completely smooth behind them; covered with short setae arising from macropunctures. Compound eye sparsely covered with small setae (Fig.  2). Gena narrow, preoccipital carina sharp. Antennal articles A8-12 forming ventrally flattened club (Fig. 2) slightly more than 2 times as long as broad (viewed dorsally).
Anterior margin of pronotum raised to carina distinctly present along anterior margin of pronotum, especially sharp medially (erroneously termed anterior pronotal carina by Mauss 2013). Short anterior pronotal carina (sensu Carpenter 1988) distinctly present at antero-ventral angle of pronotum running parallel to anterior margin, preceding crenate groove; distance between anterior pronotal carina and anterior margin of pronotum about width of fore metatarsus. Posterior pronotal carina forms low narrow translucent sinuate crest on humeral angle of pronotum. Pronotum with anterior side nearly vertical; dorso-medially slopes down towards mesoscutum, resulting in slight depression along the dorso-medial margin; posterior margin raised to short carina dorsally in front of upper half of tegula; cuticula shiny, with close, reticulate macropunctation, interstices smooth, rounded, without micropunctures; horizontally striated due to more strongly raised longitudinal interstices, becoming more coarsely punctured postero-ventrally with interstices raised to knife-like edges. Cutic- ula of mesoscutum shiny, reticulate with close deep macropunctation and narrow, distinctly raised interstices. Postero-medial cuticula of mesoscutum and mesoscutellum with longitudinal interstices more strongly raised leading to striated appearance; finely longitudinally wrinkled, postero-medially with, in addition, moderately spaced micropunctures especially along posterior margin. Mesoscutellum laterally with distinct carina along posterior margin, carina medially increasingly reduced so that cuticula of medial lobe continues evenly into the margin. Carina along posterior margin of metanotum medially with small tooth-like projections. Axilla with short blunt lateral projection only slightly projecting over adjacent posterior end of tegula. Tegula shiny, closely covered by macropunctures except completely smooth central convex area.
Mesepisternum with pronounced epicnemial carina deflexed backwards to run transversely in front of mid coxa; cuticula shiny, with close macropunctation; horizontally striated by raised interstices; area ventral to scrobal groove coarsely punctured with some interstices strongly raised to knife-like edges forming coarse honeycomblike sculpture. Process at mesepisternal scrobal groove of moderate size, cuticula on posterior side faintly shiny, finely but densely shagreened with irregularly moderately spaced micropunctation. Basal horizontal propodeal triangle laterally delimited by a perpendicular declivity, somewhat laterally produced at postero-lateral edge, posteriorly bordered by serrated carina; cuticula shiny, coarsely punctured, interstices almost knife-like. Posterior surface of propodeum striated by strong vertical cuticula-folds; cuticula shiny, weakly coriaceous and covered with fine pale setae. Cuticula of sides of propodeum and metepisternum shiny, densely horizontally wrinkled. Lateral lamella broad, somewhat convex; lateral margin almost straight; posterior margin straight, not crenate; medially where lamella joins central part of propodeum with a rounded emargination, ventro-medial edge of which produced to a small blunt protrusion; dorsal cuticula of lamella shiny, with moderately spaced macropunctation, interstices weakly wrinkled. Claws ventral with small tooth.
Metasomal terga with posterior two-fifth separated from anterior part by slight declivity especially laterally (Fig. 5); postero-lateral corners slightly produced; posterior margin of tergum I weakly crenulated, crenulation not produced into spines and not projecting over smooth translucent lower posterior margin of tergum; posterior margin of terga II-V strongly crenulated (Fig. 5), crenulation produced into distinct spines raised at an angle of approximately 30° projecting distinctly over the translucent lower posterior margin of terga (Fig. 5); cuticula with silken sheen, densely covered with coarse macropunctation, with about nine micropunctures along median axis of tergum III; single thin seta arises from bottom of each macropuncture, only slightly protruding over rim of puncture; interstices finely shagreened, moderately covered with very tiny, decumbent, pale setae, all setae orientated towards caudal end. Tergum VI with sides converging almost as straight lines, posterior margin with curved protrusion over central two-thirds (Fig. 5), laterally transverse, forming distinct angle to side.
Metasomal sternum I shiny, finely shagreened, with tiny setae but without punctures. Sterna II-V posteriorly with broad stripe of asetose, translucent cuticula adjacent to posterior margin of more strongly sclerotized cuticula; small sparse band of setae along posterior sclerotized margin somewhat projecting over anterior part of translucent strip of cuticula; sclerotized cuticula shiny, finely shagreened, on posterior half of sternum II-V with dense to moderate punctation of shallow micropunctures from which short pale setae arise, becoming sparser anteriorly, on sternum II antero-laterally with a few shallow macropunctures, on sternum III-V anteriorly with moderate to sparse shallow macropunctation. Sternum VI tapering towards distal end; with outer margin raised to bulged rim, posteriorly protruded into little blunt spine; cuticula with rather narrow smooth mid-line slightly raised to weak keel at distal end, at sides with strong macropunctures from which pale setae arise.
Male. Colour: Resembles female, except as follows (Fig. 1a). White (less yellowish than female): basal spot on mandible (absent in one specimen); labrum; clypeus except small reddish brown ventral margin; sinuate band at frons laterally extending towards upper inner margin of eye where it bends ventrally into ventral half of ocular sinus (shortly to broadly interrupted medially in two specimens); humeral spot posterolaterally extended to posterior margin of pronotum (except in one specimen), dorsomedially lengthened towards enlarged anterior part of stripe along dorso-medial margin of pronotum (fused with it into complete dorso-anterior pronotum band in two specimens); stripe along lateral part of dorso-medial margin of pronotum interrupted in the middle by zone of black colour; posterior bands on terga II-VI interrupted on each side of middle into median and two lateral markings; broad band on posterior four-fifth of sternum II; (in one specimen postero-lateral spot on each side of sternum III-IV). Antenna with A1-2 black; A3-6 anteriorly with whitish stripe otherwise dark brown at basal end changing continuously into reddish-brown to orange towards A6; A7 reddish-brown to orange; antennal club (A8-12) orange, ventrally with distinct blackish marking on A12 extending on distal parts of A11 well set off from orange of adjacent ventral area, orange on dorso-posterior parts of A8-A12 can be darkened to variable extend.
Structure: Resembles female, except as follows. Clypeus longer than broad, distal margin deeply emarginated; cuticula shiny, with moderately spaced, shallow macropunctation, interstices on apical third weakly shagreened becoming smooth distally; pale stiff setae arising from macropunctures without distal "knob", distal ends of setae frequently curved in distal-medial direction. Frons with moderately spaced macropunctation; interstices densely obliquely shagreened; bearing pale stiff setae, mainly with curved distal end, in centre few with distal "knob" (Fig. 4); frontal line can be raised to form small carina or protuberance in centre of frons. Antenna with two oval shaped, perhaps sensory, depressions (tyloids) on concave, ventral side of club, situated within antennal articles A9 and A10. Midcoxa with small but distinct spine at distal end on anterior side close to anterior-medial angle. Posterior margin of tergum VII with three deep emarginations, resulting in two lateral lobes that continue laterally into lateral margins of tergum VII and two lanceolate medial projections (Fig. 6). Sternum VIII with distinct little spine on each side of posterior margin; even concavely emarginated, surface not much sunk in towards emargination (Fig. 7).
Male genitalia as in Figs 7, 8. Genital broad, in dorsal view distance between dorso-medial margins of stipites smaller than transverse width of adjacent part of stipes. Dorso-medial shovel-like lobe of harpide large, densely covered ventrally with long setae. Dorso-posterior margin of stipes only slightly curved, without distinct dorsomedial concavity. Stipes with dorso-medial margin running nearly parallel to sagittal plane. Sides of stipites converging continuously anteriorly towards cupula. Volsella large and broad reaching antero-medial margin of dorso-medial lobe of harpide; dorsal area with strongly sclerotized large, dark tubercles; distances between tubercles moderate; medial process large, trapeziform due to more or less truncate apex, continuing posteriorly into posterior process at a blunt angle. Sides of thyrsoi only slightly converging towards posterior (apical) end of aedoeagus. Each thyrsos continues anteriorly into apodema thyrsos with a distinct outwardly directed curve. On ventral side cupula medially clearly protruded anteriorly, completely projecting over dorsal margin of cupula in ventral view of genital capsule.
Measurements. Measurements of the exoskeleton are listed in Table 1. Geographic distribution and variation. The geographic range of Celonites tauricus is shown in Fig. 9. Based on morphological differences six intraspecific taxa can be separated (Table 2, Figs 10-13) belonging to different geographic regions (Fig. 9). Distinct population groups exist in the Crimea, Kos, western Asia Minor, eastern Asia Minor up to the Elburs mountains, south-east of the Taurus range and in Cyprus.
Individuals from western Asia Minor are large and strongly built (Fig 1a), their metasomal terga are coarsely punctured and bear distinct pointed spines at the posterior end that project well beyond the lower posterior margin (Fig. 10). The specimens are dark with reduced whitish markings and reddish brown legs (Fig.1a). The posterior bands on the metasomal terga are laterally interrupted, usually from T2 to T5 but at least from T3 to T5. The population from the Crimea differs only slightly in that the individuals are a little less robust and the terga are somewhat less coarsely punctured. The population from Kos varies from the mainland population in that the crenulation along the posterior margin of the metasomal terga is reduced to short, weakly pointed, horizontal processes that do not project beyond the posterior margin (Fig. 12). The members of the populations to the south-east of the Taurus range are distinguished from individuals from western Asia Minor by their yellowish colour and considerably more extensive markings (Fig. 1b), including a spot on the clypeus of the females. The individuals from eastern Asia Minor are smaller and daintier than individuals from the western populations. The macropunctures on their metasomal terga are much smaller and the crenulation along the posterior margins of the terga consists of short, horizontal, truncated processes (Fig. 13). The light colour is yellowish and the markings are more extensive than in specimens from the western populations with, at least, complete posterior bands on the terga T1 to T3 (Fig. 1b). Populations with at least some specimens intergrading between the taxa from western and eastern Asia Minor exist in a zone extending from the Taurus range in the south-west to north-east Turkey and Armenia (Fig. 9). Such individuals reveal an intergrading pattern of various morphological and colour traits, but also transient characters like, for example, little spines along the posterior margins of the metasomal terga or macropunctures of intermediate size on the metasomal terga. Likewise there are populations in south-east Turkey with specimens, intergrading between the taxon from eastern Asia Minor and the taxon to the south-east of the Taurus range ( Fig. 9), that are characterized by little spines along the posterior margins of the metasomal terga and a yellowish spot on the clypeus of the females. The specimens from Cyprus are distinguished by their small size, comparatively dark coloured dorsal side of the antennae (Fig. 1b), deeper yellow colour, and the crenulation of the metasomal terga that, at least laterally, consists of short but more or less pointed processes.

Habitat
Eight out of nine localities of Celonites tauricus in the Crimea are situated in the habitat zone of submediterranean vegetation of the south coast (Fig. 14), and it can be presumed that the taxon is distributed along the whole south coast of the peninsula. A single record comes from the zone of forest steppes of the Crimean foothills. In the Crimea C. tauricus is confined to dry light forests and shrubs or rocky steppe and phrygana slopes (Fig. 15). The highest abundance of the taxon was recorded in Lisya Bay where in some years 1.5-2 foraging females per 100 m 2 were observed on average at the same time (the size of the whole sample area was 2500 m 2 ). The distribution of Celonites tauricus in the Crimea seems to be not only affected by the distribution of Teucrium chamaedrys, as its main forage plant, since this plant is very common and abundant on the entire peninsula. Probably C. tauricus does not occur both in higher altitudes of the mountains and in the plains of the Crimea for climatic reasons, since e.g. the mean winter temperatures are lower in these regions in comparison with the foothills and slopes on the south coast (Ved' 2000). In Kos Celonites tauricus was found in areas from 20 to 110 m above sea level that were covered with phrygana vegetation characterized by larger patches of flowering plants of Satureja thymbra (Fig. 20).

Flower associations
Flower visiting records of Celonites tauricus are summarized in Table 3. In all investigated populations males and females of C. tauricus were exclusively recorded from flowers of Lamiaceae, but there are distinct regional differences with regard to the species of Lamiaceae utilized. In the Crimea females of Celonites tauricus visiting flowers were primarily observed to collect pollen and nectar from flowers of Teucrium chamaedrys (Figs 16, 17). This flower association was confirmed at all localities where C. tauricus was studied and in every study period. Occasionally females were also recorded at   flowers of Thymus tauricus and Teucrium polium. However, visiting flowers of these two plant species is probably not typical for C. tauricus in the Crimea and was only observed in Lisya Bay in 2013. In that year the spring was unusually dry and flowering plants of Teucrium chamaedrys were very scarce whereas flowering Thymus tauricus were much more abundant and widespread. In five out of the eight visits to Thymus tauricus and also in the visit to Teucrium polium the females switched over to these plants from Teucrium chamaedrys and afterwards returned to visit flowers of T. chamaedrys. Females of C. tauricus were only observed to visit plants of Thymus tauricus growing in close proximity to flowering plants of Teucrium chamaedrys. Males of C. tauricus also were observed mainly at flowers of T. chamaedrys at Cape Aya in 2004 and in Lisya Bay in each year of observation. In contrast, the Kos population was mainly associated with Satureja thymbra and to minor extend also with Thymus capitatus, though in addition, at least at some localities two Teucrium species (T. divaricatum and T. polium) were in flower. However, the proportion of visits to Thymus capitatus may increase during the on-going flight season, since at the end of the investigation period Th. capitatus had just started flowering whereas S. thymbra was nearly over.
During flower visits at Satureja thymbra and Teucrium chamaedrys the females always stood on the lower lip of a flower and took up nectar and pollen simultaneously. The proboscis was protruded deeply into the corolla tube while the female performed at a high frequency slight back and forth movements of the anterior parts of her body, rubbing her head over the nototribic anthers (Figs 16, 25, 26). In this manner pollen grains were removed from the pollen sacs with the stiff knobbed setae on the frons, which form a pollen-collecting apparatus, accumulating on frons and clypeus (Fig.  27). The median duration of the visits of females to flowers of Satureja thymbra was 2.6 s (range 0.8-6.3 s, n = 43). At Thymus capitatus the flower visiting behaviour was very similar to the behaviour described above, except that the female orientated her body axis at an angle to the longitudinal plane of the zygomorphic flower both to the left and to the right so that her frons was directed towards the anthers that are situated  further laterally in this flower. Periodically flower visiting was interrupted and the pollen grains were transferred from the frons to the mouthparts by alternating brushing movements of the fore legs (Figs 17,28,29) which were brought between the mouth-parts while the pollen was being ingested (Fig. 30). This frons brushing behaviour took place on flowers (Figs 17,28,29) and also on the ground in the close vicinity of the plants (Fig. 23).
On the Crimea, individual females of C. tauricus usually visited flowers over a period of 20-40 minutes. After about 10-15 minutes they regularly interrupted flower visiting and alighted on stones or grass. They remained there for several minutes repeatedly regurgitating and withdrawing again a mass of pollen and nectar that became visible as a drop of liquid between the mouthparts. This behaviour probably served to thicken the pollen and nectar mass.
During flower visits males always inserted their proboscis into the corolla tube (Figs 22,24) indicating the uptake of nectar. At flowers of Satureja thymbra males, like the females, performed slight back and forth movements in addition rubbing with their frons over the anthers, so that they probably collected pollen as well. This rubbing behaviour was not observed at flowers of Thymus capitatus.
More than 99% of the pollen from the crops of females from Kos consisted of hexacolpat pollen grains of Lamiaceae. This pollen type occurs in several genera of Lamiaceae including Satureja and Thymus.

Nesting
Three nests of Celonites tauricus were recorded in the Crimea. The first nest was investigated on 2 July 2004 in the vicinity of the village of Veseloye near Sudak. The nest site was a rocky slope with steppe vegetation predominated by Melica taurica K. Koch and Teucrium chamaedrys along with individual trees of Celtis glabrata Steven ex Planch. The nest was located in a small cavity underneath a stone situated in the shadow of one of the Celtis trees. The nest under construction contained two brood cells. The cells were placed on the underside of the stone and were made of fine clayey soil with a few tiny stones (Fig. 18). The cells were cylindrical, rounded at the closed (basal) and truncate at the open (apical) end (Fig. 18) measuring 9 mm in length and 4 mm in width. The cell wall was 0.25 mm thick. The outer cell surface, typically for Celonites, showed a distinct "fish scale" pattern. The first cell was sealed with mud. The cell seal was positioned slightly in from the edge of the cell opening and was of the same thickness as the cell wall. The cell was provisioned with a pollen loaf and contained an egg at the basal end. Basally the loaf was situated close to the egg, while the apical end of the loaf was further away from the seal resulting in an empty space between the pollen loaf and the seal at the apical end of the cell. The pollen loaf was attached to the cell wall with numerous spike-like projections. The second cell was open and contained the female owning the nest.
The second nest was studied on 2 July 2011 in Lisya Bay. The nest site was a slope covered with phrygana vegetation predominated by Elytrigia caespitosa subsp. nodosa (Nevski) Tzvelev, Atraphaxis replicata Lam., Thymus tauricus and Teucrium chamaedrys. The nest was also situated in a small cavity on the underside of a small stone. The nest was old and contained three cells covered with an additional layer of mud forming a nest covering (Fig. 19). All cells were sealed at the apical end and had a large frontal opening that covered the apical third of the cell, which had probably been made by an emerging imago of Celonites tauricus (Fig. 19). Every cell contained an empty, thin, whitish Celonites cocoon and a meconium.
The third nest was found in Lisya Bay on 11 July 2013. It was also situated on the underside of a small stone and it was also old. The nest consisted only of a single cell covered with an additional layer of mud. The cell was sealed at the apical end and had a frontal opening at the apical part, which had probably been made by an emerging imago of C. tauricus. Inside the cell were a Celonites cocoon and meconium.
Females of C. tauricus were never observed at water collection sites. Therefore the mud used for cell construction was probably made by mixing clay particles with regurgitated nectar.

Male behaviour
Males performed patrol flights along the preferred forage plants of the females, i.e. Teucrium chamaedrys in the Crimea as well as Satureja thymbra and Thymus capitatus in Kos, and over nearby stones in a low constant flight. Patrolling was regularly interrupted by perching in the immediate vicinity of the forage plants, mainly on stones but also on the ground or on the plants themselves (Fig. 21). Moreover, patrolling was occasionally interrupted by flower visits. On one occasion a male was observed pouncing on a female resting on a stone between consecutive flower visits. The male alighted on the back of the female for a short moment and flew off again. On another occasion a male pounced on a female while it was visiting a flower of Satureja thymbra. Courtship and copulation were not observed.

Phenology
The species is univoltine. In the Crimea males were observed from 5 June to 8 July and females from 7 June to 31 July indicating slight proterandry.

Discussion
The rediscovered Celonites taxon from the Crimea, assigned to C. abbreviatus tauricus, can be consistently identified as a member of the Celonites abbreviatus-complex (sensu Mauss 2013) by the existence of a pollen collecting apparatus composed of knobbed setae on clypeus and frons, the outstanding autapomorphic character of this group (Mauss 2013). Moreover, the Crimean taxon also shares all apomorphies of the C. abbreviatus-group (Mauss 2013), that is, the males have only two oval-shaped tyloids on the antennal club, they bear a small spine at the distal end of the midcoxa, their tergum VII is posteriorly markedly lobed and the posterior margin of sternum VIII of the males is specifically emarginated.
Within the Celonites abbreviatus-complex the members of the taxon Celonites a. tauricus from the Crimea are in the main similar to individuals of Celonites spinosus from western Asia Minor in both the colouration and morphology of the exoskeleton and in the structure of the male genitalia. Because of this high degree of similarity it is hypothesized that both taxa are not reproductively isolated from each other and thus belong to the same biospecies (sensu Mayr 1967). The existence of populations of Celonites in Central and East Anatolia that include at least some individuals with intermediate characters between typical spinosus-and typical invitus-forms indicate that these eastern populations are not reproductively isolated from the western spinosus-like populations. This is in agreement with the findings of Gusenleitner (1985), who as a result formally synonymized both taxa. In the same way, the distinct yellowish populations that occur to the southeast of the Taurus range are connected to the northeastern invitus-form by populations that contain individuals with intermediate characters in southeast Anatolia. For that reason it can be hypothesized that they also are not reproductively isolated from the other taxon and therefore belong to the same biospecies. All members of this biospecies are characterized by an antennal club with dark tip and male genitalia with a broad stipes and a large medial process of the volsella. The colour pattern on the ventral side of the antennal club that is in contrast richly coloured might be associated with mate recognition, since it has been demonstrated by Mauss and Müller (2014) that in an allied Celonites species the tips of the male antennae are held above the female compound eyes for a short time during mating prior to the insertion phase. Therefore the colour pattern of the club could be part of a mating signal and therefore may act as a reproductive isolation mechanism. This is also likely for the observed differences in the proportion and structure of the male genitalia. Since the Celonites populations from Kos and from Cyprus are mainly similar in regard to these particular characters there is no evidence that they may be reproductively isolated from the mainland populations. For that reason, they are regarded as allopatric populations of a single biospecies. Due to priority the name of this biospecies has to be Celonites tauricus.
Celonites tauricus resembles Celonites abbreviatus in many characters and is therefore in all probability closely related to this species, as already established by Gusenleitner (1966). Despite some differences in specific characters of particular populations, both taxa differ mainly in the colour pattern on the ventral side of the antennal club and in the proportion of the stipes and the form of the volsella. As discussed above these characters are associated with the mating system and might act as a reproductive isolation mechanism. It is of note that neither Celonites abbreviatus nor any other species of the C. abbreviatus-complex could be demonstrated to occur within the geographic range of C. tauricus. All former records of C. abbreviatus from this region, for which voucher specimens were examined, turned out to belong to C. tauricus. Therefore C. tauricus and C. abbreviatus seem to be allopatric taxa that have evolved from a common stem species.
The areas inhabited by Celonites tauricus in the Crimea and in Kos are comparable to habitats of Celonites abbreviatus in Central Europe (Bellmann 1995, pers. obs.), southern France (pers. obs.), Spain (pers. obs.) or the Peloponnese (Mauss 2006). The flower associations of C. tauricus are very similar to those of C. abbreviatus, in that both taxa specialize in utilizing Lamiaceae with nototribic flowers that serve as the sole pollen source (Bellmann 1984, 1995, Mauss 2006, Müller 1996, Schremmer 1959. Moreover, as in C. tauricus in the Crimea, a Teucrium species seems to be of particular importance for the occurrence of C. abbreviatus in Central Europe (Bellmann 1995), and Satureja thymbra, which is the main forage plant of C. tauricus in Kos, is also the mainly visited plant of C. abbreviatus in the Peloponnese (Mauss 2006). The flower visiting behaviour of both species appears identical (cf. Bellmann 1984, 1995, Mauss 2006, Müller 1996, Schremmer 1959. The construction of the brood cell, the position and shape of the egg and the provision, as well as the frontal orientation of the emergence hole in old cells of C. tauricus are similar to C. abbreviatus (cf. Bellmann 1984Bellmann , 1995. Nests constructed on the underside of stones, as in C. tauricus, do also exist in C. abbreviatus (Bellmann 1995), though nests of the latter, at least in Central Europe, are more frequently aerial, attached to stones or twigs (Bellmann 1995). Finally, males of both taxa search for females in the vicinity of forage plants (cf. Mauss 2006). In summary, the ecology and behaviour of C. tauricus and C. abbreviatus seem to be similar in all investigated parameters and their ecological niches are probably mainly identical. This is in congruence with the perceived allopatric distribution of both taxa, as their comparable ecological requirements should prevent sympatry.
The observed distribution pattern of the intraspecific taxa of C. tauricus in Asia Minor (Fig. 9) can be biogeographically explained by the isolation of small populations in cold periods of the Pleistocene in secondary glacial refuges that have already been characterized by Lattin (1967). The ancestors of the western spinosus-like populations probably subsisted in the eastern part of the Pontomediterranean refuge, the ancestors of the populations south-east of the Taurus range in the Syrian refuge and the ancestors of the eastern invitus-like populations in the Caspian refuge sensu Lattin (1967). As a result of the post glacial warming the three populations expanded their range again and hybridized with each other where they secondarily came into contact as is indicated by individuals with intergrading characters.
It can be hypothesized that C. tauricus colonized the Crimea from the west of Asia Minor, since the Crimean population of Celonites tauricus is morphologically much more similar to the spinosus-like populations of C. tauricus from the western parts of Asia Minor than to the invitus-like populations in the east. Probably, the colonization took place in a phase of low sea-level when today's extensive western shelf of the Black Sea was situated above the level of the ancient Black Lake, resulting in a large area along the western coast of the Black Lake covered with Sub-Mediterranean vegetation that connected the Sub-Mediterranean areas on the Crimea with the north-west end of Asia Minor (Yena et al. 2005, and references therein). During the Würm glacial this western costal zone was an extensive refuge harbouring the Mediterranean flora (Yena et al. 2005). Major et al. (2006) showed that intense meltwater pulses at the end of the Würm glacial between 18 and 16 ka BP raised the level of the Black Lake, probably to its outflow, so that lower parts of the shelf were drowned. But during the following warm post glacial periods of the Bølling-Allerød (15-13 ka BP) and the Preboreal (11.6-9.4 ka BP) the complete shelf was again well above the sea level (Major et al. 2006). Consequently it seems likely that finally this entire area was inhabited by the western spinosus-like population of C. tauricus. The range was disrupted when the Black Sea shelf was flooded with water from the Marmara Sea as a result of the post glacial rise of the global sea level about 9.4 ka BP (Major et al. 2006). The flooding led to the separation of the recent relict population on the Crimea from the main spinosuslike population of C. tauricus in the western part of Asia Minor. A comparable biogeographical pattern exists, for example, in Pinus brutia Ten. (Yena et al. 2005) and some species of spiders (Kovblyuk 2014). Similarly, the population of C. tauricus in Kos probably has been isolated from the spinosus-like mainland population by the postglacial rising of the sea level of the Aegean that separated the island from Asia Minor.
In contrast, Cyprus is a primarily oceanic island of volcanic origin, that probably has been connected to the mainland only for a period of approximately 600000 years during the Messinian Salinity Crisis of the upper Miocene (5.9-5.3 Ma BP), when a land bridge existed that joined the island with continental areas of Asia Minor and Syria (Poulakakis et al. 2013, and references therein). Since the re-flooding of the Mediterranean basin in the early Pliocene Cyprus has been separated from Asia Minor as the nearest mainland by a distance of at least 30 km (Hadjisterkotis et al. 2000). Therefore, the existence of a morphologically distinct Cyprian population of C. tauricus is probably the result of dispersal, but it is uncertain whether this occurred over a land bridge or transmarine, as the divergence time from the mainland population is unknown. Based on morphological similarity the migration probably started from an eastern, more invitus-like population of C. tauricus.
Andreas Müller (Zürich) determined the pollen samples. We are especially grateful to Sarah Gess (Albany Museum, Grahamstown) for valuable comments on the manuscript and improvement of our English.