Description of the immature stages of Larinus vulpes and notes on its biology (Coleoptera, Curculionidae, Lixinae)

Abstract Mature larva and pupa of Larinus vulpes (Olivier, 1807) (Curculionidae: Lixinae: Lixini) are morphologically described for the first time and compared with known larvae and pupae of other Larinus species. Very high counts of larval body setae (pronotum with more than 25 setae and postdorsum on meso- and metathorax and also on abdominal segments I–VII with more than 12 setae) are characteristic features of the nominotypical subgenus Larinus. The biology of the species was studied in Ukraine. Echinops ruthenicus and E. sphaerocephalus were identified as host plants of both larvae and adults of this weevil based on the present research in Ukraine, which shows probably oligophagous. Overwintering beetles emerged at the end of May or earlier, then feeding and mating on the host plants. The highest level of adult activity was observed at the end of June. Larvae were endophagous within the flower heads. In July and August, the larvae pupated within inflorescences in a pupation cell. Adults exited the cells at the end of August and did not hibernate on the host plants. Sometimes, larvae and imagines of a new generation were found outside the flower heads in chambers constructed on the stems.


Introduction
The weevil genus Larinus Dejean, 1821 belongs into the tribe Lixini Schoenherr, 1823 and is represented by ca. 180 species (Csiki 1934;Ter-Minassian 1967;Gültekin 2006) of which more than 110 are known in the Palaearctic (Gültekin and Fremuth 2013;Gültekin and Shahreyary-Nejad 2015). A further 40 species are recorded from the Ethiopian region, only three species from the Oriental region, four (introduced species) from the Nearctic region (McClay 1988;Gültekin 2006), and one in New Zealand (Woodburn and Briese 1996;Gültekin 2006). The valid systematic position of this genus has been assigned for Palaearctic species in the Catalogue of Palaearctic Coleoptera (Gültekin and Fremuth 2013). The genus is divided into four subgenera: Cryphopus Petri, 1907;Larinus Dejean, 1821;Larinomesius Reitter, 1924;andPhyllonomeus Gistel, 1856 (Gültekin andFremuth 2013). Knowledge of the morphology of immature stages in Larinus is incomplete in comparison to the total number of species in this genus and to the importance of several species as potential biological control agents against weeds (Nikulina et al. 2004;Seastedt et al. 2007).
The species Larinus vulpes (Olivier, 1807) belongs to the nominotypical subgenus Larinus, which includes 35 species in the Palaearctic region (Gültekin and Fremuth 2013;Gültekin and Shahreyary-Nejad 2015), and is distributed in the western Palaearctic, east Siberia and central Asia (Gültekin and Podlussány 2012;Gültekin and Fremuth 2013). The most northern area of its range is Kungur, Russia (56-57°N) (Dedyukhin 2011). The life cycle of L. vulpes is associated only with globe thistles, the genus Echinops L. (Asteraceae). Similar to other Larinus species, adults feed on the leaves and stems. Eggs, larvae, and pupae develop in the inflorescences. Imagines of a new generation hibernate outside the host plants. Circumstantial observations of reproduction and preimaginal development of L. vulpes were reported by Fabre (1922; as Larinus maculosus Schoenherr, 1832). Next, Volovnik (1996Volovnik ( , 2016 provided sufficient details on the biology of this species. The immature stages of this species have never been morphologically described. Some species from the genus Echinops are very important invasive weeds (Czarapata 2005; Reddy et al. 2008) and also have medicinal uses (Murch et al. 2003;Eram et al. 2013;Parhat et al. 2014). The globe thistles are nectar (Wroblewska et al. 1993;Jabłoński and Kołtowski 2005) and ornamental plants (Wiersema and León 2016), a potential source of natural insecticide (Gemechu et al. 2013;Liu et al. 2013), molluscicid (Hymete et al. 2005), and energetic oil (Horn et al. 2008). The root extract of E. giganteus A. Rich. is used as a mosquito repellent (Karunamoorthi and Hailu 2014) and also as perfume (The Green Vision 2017), and these roots are used as dietary spice (Stève et al. 2016). Because of their deep roots, the globe thistles are widely used for mechanical stabilisation of banks, ravines, and slopes (Chopik et al. 1983). Therefore, there are strong arguments for a detailed investigation of the weevil L. vulpes and then promote the use of larvae of L. vulpes as potential biological control agents against this plant. The knowledge of bionomy of immature stages of Larinus species is also important for further taxonomic studies at different levels and for effective protection of endangered species. In this paper, we describe the immature stages of L. vulpes and provide details of its life history based on field observations in Ukraine.

Insect collection and laboratory breeding
The material used to describe the immature stages was collected and field observations were conducted in Ukraine in the following localities: 1) The eastern shore of Molochnyi Estuary between the two villages of Altagir (= Bogatyr) and Radyvonivka (46°38'29"N,35°16'59"E) (Kolomiychuk and Vynokurov 2016). Globe thistle Echinops ruthenicus is locally distributed between the central road and the coast of the estuary (Fig. 21).
In the above-mentioned localities, life cycle, including feeding of adults, oviposition, and early development of larvae were observed directly during the vegetation growing seasons of Echinops ruthenicus and E. sphaerocephalus L. in the time period 2012-2016.
The compound flower heads of globe thistles consist of simple capitula, each of which has only one floret. These primary capitula are aggregated in globose secondary capitula (Kadereit and Jeffrey 2007). For convenience, below, we shall name as capitula only secondary ones.
The second author collected all larvae and pupae of L. vulpes within inflorescence. Some inflorescences (n = 42) were dissected to investigate preimaginal development, and a further 250 were dissected to determine the quantity of preimaginal specimens of L. vulpes within an inflorescence. All photographs in the field were taken with digital cameras, a Nikon Coolpix 4600 and a Canon PowerShot SX500 IS.
Laboratory observations were conducted in Melitopol, Ukraine (46°50'N, 35°22'E). The dry inflorescences (n = 7) with developing mature larvae or pupae were placed into cardboard boxes. A small hole was opened in every inflorescence for possible observations of insect development. Measurements of flower head were performed with a slide caliper and ocular micrometre. The size of flower heads was determined at the greatest diameter.
Geographical distribution and phenology were studied from several entomological collections, specifically: Schmalhausen Institute of Zoology of National Academy of Sciences of Ukraine (Kyiv), TG Shevchenko Kyiv National University Zoological Museum, Zoological Institute of Russian Academy of Sciences (St. Petersburg), VN Karazin Kharkiv National University Museum of Natural History, and Igor Maltzevs' private collection (Odessa). In total, more than 130 pinned specimens were studied.

Morphological descriptions
Part of the larval and pupal material was preserved in Pampel fixation liquid (see  and used for the morphological descriptions. These specimens are deposited in the collection of Group Function of Invertebrate and Plant Biodiversity in Agro-Ecosystems of the Crop Research Institute (Prague, Czech Republic). The collectors identified the plants. To prepare slides, we followed May (1994).
The observations and measurements were conducted using a light microscope with calibrated oculars (Olympus BX 40, SZ 11, and Nikon Eclipse 80i). The following characteristics were measured for each larva: head width, length of the body (larvae fixed in a C-shape were measured in the middle of the segments in lateral view), and width of the body in the widest place (i.e., meso-and metathorax). For the pupae, the length and the width at the widest place were measured.
Drawings were created with a drawing tube on a light microscope and edited by the programs Adobe Photoshop 10, Corel Photo-Paint X7, and GIMP 2.
We used the terms and abbreviations for the setae of the mature larva and pupa found in Scherf (1964), May (1977May ( , 1994 and Marvaldi (1998Marvaldi ( , 1999. The numbers of setae of the bilateral structures are given for one side. The counts of some of the setae on the epipharynx (particularly ams and mes) have not been completely resolved. According to Marvaldi (1998Marvaldi ( , 1999, the standard status of the epipharynx in weevils is 2 ams and 3 mes; however, when the position of the distal mes is very close to the anterior margin, it appears as ams. The final decision was to add this problematic seta to the latter group (ams), and the position of this seta is similar to that in other genera, e.g., in Coniocleonus Motschulsky, 1860 or Tychius Germar, 1817. We did not follow Stejskal et al. (2014) and Skuhrovec et al. (2014) who accepted the standard status in weevils and counted the seta as mes, but we followed Trnka et al. (2015) and ), e.g., in Adosomus Faust, 1904 or Sibinia Germar, 1817. The thoracic spiracle was located on the prothorax near the boundary of the prothorax and mesothorax, as shown in the drawing (see Fig. 10), but this spiracle is of mesothoracic origin (Marvaldi et al. 2002;Marvaldi 2003).
Colouration. Head light brown or brown with a distinct pale pattern around the frontal and epicranial sutures (Fig. 7). All thoracic and abdominal segments are dark yellow with a light brown, elongate stripe on the dorsum of the pronotum (Fig. 7).
Vestiture. Setae on body thin, short, light yellow or orange (Figs 7-9). Head capsule (Fig. 1). Head suboval, slightly longer than wide, endocarinal line weak, but long as a half-length of frons. Frontal sutures distinct, wide, and extended to the antennae. Single anterior stemma (st) distinct, in the form of a slightly pigmented spot. Des 1 and des 2 located in the upper part of the central part of the epicranium, des 1 near the middle part of the epicranium and des 2 near the side of the epicranium, des 3 located anteriorly near the frontal suture, des 4 located in the central part of the epicranium, des 5 located anterolaterally; all des long, nearly subequal in length, except des 4 distinctly shorter (Fig. 1). Fs 1 , fs 2 and fs 3 placed medially, fs 4 located anteromedially, fs 4b located laterally close to fs 4 ; and fs 5 located anterolaterally, close to the epistoma; all setae very long to long, only fs 4b and fs 5 medium, distinctly shorter than very long fs 1-4 ( Fig. 1). Les 1-2 and ves 1-2 very long, as long as des 5 . Epicranial area with two sensilla, one upper des 1 and the second in upper part of posterior; and also with 3 pes in line with upper des 2 .
Antennae located at the end of the frontal suture on each side, membranous and slightly convex basal article bearing one conical sensorium, relatively long; basal membranous article with 5 sensilla, different in both shape and length (Fig. 4). Clypeus (Fig. 2) trapezoid-shaped, approximately twice as wide as long, with two relatively long cls, cls 2 slightly shorter than cls 1 , localized posterolaterally and 1 sensillum located close to cls 1 ; anterior margin concave.
Thorax. Prothorax larger than meso-and metathorax. Meso-and metathorax distinctly wider than abdominal segments I-IV. Spiracle unicameral. Cuticle densely spiculate and with distinct thorn-like cuticular processes, primarily on dorsal parts but also on pleural parts (Fig. 7). Prothorax ( Fig. 10) with ca. 30-35 relatively long to short prns unequal in length, 25 on pigmented pronotal sclerite, which is subdivided medially into two triangular plates, next 5-10 prns placed below; 20 relatively long ps also on pigmented sclerite, and 12 relatively long eus. Mesothorax ( Fig. 10) with 3 short prs; 13 relatively long to short pds; 6-7 relatively long to short as; 6 relatively long to short ss on pigmented sclerite; 6-9 relatively long to short eps on pigmented sclerite; 14 relatively long to short ps on pigmented sclerite and 12 relatively long to long eus. Chaetotaxy of meso-and metathorax ( Fig. 10) almost identical, but some specimens partly variable in the exact count of setae. Each pedal area of the thoracic segments well separated and pigmented, with 10 long pda on pigmented sclerite, unequal in length.
Abdomen. Abdominal segments I-IV of almost equal length, subsequent abdominal segments decreasing gradually to the terminal parts of the body. Abdominal segment X reduced to four anal lobes of unequal size, the dorsal being distinctly the largest, the lateral pair equal in size, and the ventral lobe very small. Anus located   terminally. Spiracles unicameral, the eight abdominal spiracles located laterally, close to the anterior margin of abdominal segments I-VIII. Cuticle also densely spiculate and with distinct thorn-like cuticular processes, primarily on dorsal parts but also on pleural parts (Figs 8-9). Abdominal segments I-VII (Figs 11-12) with 2 relatively long to short prs; 13 relatively long to short pds, 10 pds in line, and 3 pds in the below part partly anteriorly; 7 relatively long to short ss, 5 ss under pds (abdominal segment VII only with 3 setae), and 2 ss in below part of dorsal lobe; 13 (10-14) relatively long Figures 10-12. Larinus vulpes mature larva, habitus. 10 Lateral view of thoracic segments 11 Lateral view of abdominal segment I 12 Lateral view of abdominal segments VII-X. Abbreviations: prns -pronotal s., prs -prodorsal s., pds -postdorsal s., as -alar s., ss -spiracular s., eps -epipleural s., ps -pleural s., pda -pedal s., lsts -laterosternal s., eus -eusternal s., ds -dorsal s., sts -sternal s., Th1-3 -number of thoracic segments, Ab1-10 -number of abdominal seg. Scale bar 2 mm. to short eps on pigmented sclerite (only on abdominal segments I-II); 9 relatively long ps of unequal length; 2 short lsts and 2 short eus. Abdominal segment VIII ( Fig. 12) with 2 relatively long prs; 10 relatively long to long pds in line; 2 relatively long ss in below part of dorsal lobe; 13 relatively long to short eps; 9 relatively long ps of unequal length; 2 short lsts and 2 short eus. Abdominal segment IX ( Fig. 12) with 7 ds (6 long ds near posterior margin, and 1 short ds medially); 13 relatively long to long ps; and 2 relatively long and 2 short sts. Abdominal segment X (Fig. 12) with 2 very short setae (ts) on each lateral anal lobe, and 1 very short seta (ts) on dorsal anal lobe.
Colouration. All thoracic and abdominal segments light yellow or greenish-white. Cuticle smooth, except thorn-like processes on abdominal segments III-VIII.
Morphology . Body moderately slender and elongated. Rostrum long, approximately 2.5 times as long as wide, extended to mesocoxae. Antennae relatively long and slender. Pronotum 2.5 times as wide as long. Meso-and metanotum of equal length. Abdominal segments I-V of equal length, abdominal segments V-VII diminish gradually, abdominal segment VIII almost semi-circular, and abdominal segment IX distinctly smaller than other segments. Urogomphi very short, almond-shaped with acute sclerotised apexes. Spiracles placed dorso-laterally; 5 pairs functional on abdominal segments I-V and one atrophied on abdominal segment VI, on next abdominal segments spiracles invisible. Sexual dimorphism visible in the structure of abdominal segment IX: gonotheca of ♀ divided (Fig. 19), ♂ undivided (Fig. 20).
Comparison with larvae of other Larinus species. To date, larvae of 16 Larinus species have been described (Gardner 1934;Scherf 1964;Lee and Morimoto 1988;Nikulina et al. 2004;Zotov 2009aZotov , 2010Gosik and Skuhrovec 2011;Nikulina and Gültekin 2014), while detailed descriptions of the pupae are known for only 8 Larinus species (Zotov 2009a(Zotov , 2010Gosik and Skuhrovec 2011;Nikulina and Gültekin 2014). The comparison with previously described immatures of some other species, primarily of L. (Phyllonomeus) saussureae Marshall, 1924 (Gardner 1934), L. (Phyllonomeus) carlinae (Olivier, 1807) (as L. planus F.) and L. (Phyllonomeus) iaceae (Fabricius, 1775) (both in Scherf 1964), was somewhat problematic because of missing details of chaetotaxy and/or absence of quality drawings; therefore, a comparison of these three species with other known Larinus species was not possible to the level of detail required to incorporate them in the key (see Key to the immature stages of the Larinus). Lee and Morimoto (1988) provide a general larval description of the genus Larinus based on two species: L. (Phyllonomeus) latissimus Roelofs, 1873 and L. (Phyllonomeus) meleagris Petri, 1907. However, they did not present any differences between these two species (see aggregation of both species in the key at dichotomy 12).
According to May (1993), the increased number of pds on the meso-and metathorax and abdominal segments I-VII and of setae on the epipharyngeal lining (als) (i.e., more than the most frequent number of setae in weevils) are diagnostic of the mature larva of the subfamily Lixinae. The following descriptions of mature larvae from the tribe Lixini confirmed this diagnosis: genus Larinus (Scherf 1964;Lee and Morimoto 1988;Nikulina et al. 2004;Zotov 2009aZotov , 2010Gosik and Skuhrovec 2011;Nikulina and Gültekin 2014); genus Lixus (Scherf 1964;Lee and Morimoto 1988;May 1994;Nikulina 2001Nikulina , 2007Zotov 2009a, b;Nikulina and Gültekin 2011;Gosik and Wanat 2014;Skuhrovec and Volovnik 2015;Trnka et al. 2016); and Rhinocyllus conicus (May 1994), in addition to descriptions of all known species from the tribe Cleonini (Zotov 2011;Stejskal et al. 2014;Trnka et al. 2015). For a proper comparison of both tribes, including a key and detailed generic studies, further descriptions of immature stages of several Cleonini would be required. Gosik and Wanat (2014), in a precise general description of the larvae of the tribe Lixini, summarized the tribe by 16 character sets (for details, see Gosik and Wanat 2014), but some of these characters (primarily chaetotaxy on the body) do not correspond exactly with most Larinus species from the nominotypical subgenus, including the recently described L. vulpes. The species from the subgenus Larinus (except L. idoneus Gyllenhal, 1835 and L. latus (Herbst, 1783)) had very high counts of larval body setae; e.g., pronotum with more than 25 setae and postdorsum on meso-and metathorax and also on abdominal segments I-VII with more than 12 setae (see details in Key to the immature stages of the Larinus and Table 1). The pupal number of setae was identical to that of all known pupae of species from the subgenus Larinus (except L. idoneus) with a pronotum with 25 or even more setae (see details in Key to the immature stages of the Larinus and Table 2). Morphological characters of larvae and pupae distinctly separated the subgenus Larinus from the other subgenera Phyllonomeus Gistel, 1856 andLarinomesius Reitter, 1924. Only two species (L. idoneus and L. latus) from the nominotypical subgenus did not correspond with the described chaetotaxy, which could be explained considering three hypotheses: (1) the nominotypical subgenus can be divided into two distinct groups, (2) these two species do not belong in this subgenus, or (3) these species show a peculiar autapomorphy; a change in a setal number can be a mere convergence (or coincidence). To solve this problem further morphological and molecular studies would be necessary.
The (2) postlabium with 4 or 5 setae; (3) stipes with 2 long sts; (4) prodorsum on meso-and metathorax with 3 prs; and (5) dorsal part of body distinctly spiculate; and two pupal morphological characters: (6) cuticle around setae dark-pigmented, visible spots formed; and (7) rostrum with 3 pas and only 1 rs. The primary differences between L. vulpes and L. inaequalicollis were as follow (see key to the immature stages of the Larinus): postepicranial setae pes 1 -pes 2 distinct (versus L. inaequalicollis very small, indistinct); frons with 6 fs (versus with 7 fs); endocarina not distinct, its length is as half-length of frons or less (versus distinct, massive, approximately 2/3 the length of frons), and ligula with 2 very thin ligs (versus with 1 micro ligs and two sensillae). The primary differences between L. vulpes and L. capsulatus were are as follows (see key to the immature stages of the Larinus): postlabium with 4 setae (versus L. capsulatus with 5 setae); meso-and metathorax with 6-7 as, 6 ss and 6-9 eps (versus with 4 as, 4 ss and 5 eps); abdominal segments I-VII with fewer than 14 pds and more than 10 eps (versus more than 15 pds and 8 or fewer eps); and lateral lobe of abdominal segment X with 2 setae (versus 3 setae).
Moreover, detailed descriptions of immature stages of Larinus species are also important for further studies on generic and evidently also subgeneric taxonomic relationships within Lixini and to effectively protect endangered species and promote the use of larvae of Larinus species as potential biological control agents against weeds  (e.g., Carduus, Cirsium, Echinops). Species identification of larvae with morphological evidence is relatively easy, and it is generally much cheaper than identification by molecular methods (Hirsch et al. 2010). The largest problem in the identification of the immature stages is the relatively low number of available larval descriptions in comparison to the many species only known at the adult stage. However, the problem is not exclusive to Curculionidae, being common to many other beetle groups.

Biology and ecology of Larinus vulpes
Habitats. Larinus vulpes occurred in the primary and degraded steppe lands, slopes, limestone and chalk cliffs of low mountains, forest edges, man-made treelines, roadsides and other ruderal plots. This weevil preferred open, sunny areas. In Iran, the weevil was recorded as high as 2580 m a. s. l. in the mountains (Gültekin and Podlussány 2012). Adult behaviour. Adults feed on the upper surface of the leaf. As feeding was initiated, an adult raised and strongly lowered its head onto the leaf surface, which was followed by some motions of the mandibles and repeated "peck-like" motions by its rostrum. Apparently, the motion created additional pressure and helped to break through cuticle and epidermis covered with woolly hairs. Following this behaviour, an imago gnawed on mesophyll tissue, moving the head away from itself, and at one feeding, a weevil could gnaw out an irregularly shaped piece of leaf (approximately 2 × 8 mm). The translucent cuticle of the leaf downside covered with dense woolly hairs remained intact (Fig. 22). During feeding, some short pauses by the weevil were observed. After eating, the weevil cleaned the apex of its rostrum using the apexes of both tibias. Weevils moved from one flower head to another by walking; they flew very reluctantly.
Host plant. Both adults and larvae were recorded feeding exclusively on Echinops ruthenicus and E. sphaerocephalus. We never observed L. vulpes on other plant species. According to Zwölfer (1985) and Nicolas (1895), weevils feed on Echinops microcephalus Sibth. and Sm. and E. ruthenicus (as E. ritro L.). In this case, L. vulpes is oligophagous (or monophagous sensu Jolivet 1992). Imagines fed on the leaves and on the apexes of the stems, and larvae gnawed tissues in the flower head. Hoffmann (1954) noted that L. vulpes were often recorded on Cirsium ferox, but whether this plant was a host for the weevil was not resolved. Life cycle. In Ukraine, adults were recorded from the end of April (usually from the end of May) onwards. The primary peak in the population of adults was reached at the end of June and then decreased. Active imagines from the new generation were observed from the beginning of August to early September. The phenology of the weevil is closely synchronized with the phenology of its host plant (Volovnik 1996). In spring, the weevils began to feed first on the rosettes of host plants and then gnawed the young stems and leaves (Fig. 22). Larinus vulpes is a univoltine species. Mating and oviposition occurred from the second part of June to the middle of July. The common scenario of oviposition and subsequent development from egg to adult were the typical for the genus Larinus (Volovnik 2016).
Female L. vulpes preferred to lay eggs in the larger flower heads on the side stems (Volovnik 2016). In this case, the female makes a hole in the flower head. Freshly laid, eggs were oval, milky white, glossy, 1.3-1.6 mm long and 0.7-1.2 mm wide. The eggs were laid solitarily in the deepening in the receptacle, gnawed by the female, or very close to the receptacle (maybe, the rostrum of some weevils was too short for access). The process of oviposition has been described in detail previously (Volovnik 1996). Eggs were found in inflorescences from the end of June to the end of July. Very occasionally, dry fragments of the egg and also a small larva were found in the same flower head. In other instances, the fragments of eggs were recorded in the inflorescence but without any traces of weevil larvae. These eggs were likely destroyed by parasites or abiotic factors (i.e., desiccation).
Later, after hatching of larva, the oviposition site became a visible, brownish tiny spot (Fig. 23). Apparently, this spot was a result of damage to the flower head by the larva. Larvae nibbled the receptacle of the flower head and the bases of primary capitula, with the damaged capitula fixed with a sticky liquid that flowed out of receptacle, resulting in a deformed flower head (Fig. 24). The mature larva colouration was peach-orange (Figs 26,28) and that of the young pupa yellow-orange (Figs 25,27). Fabre (1922) wrote that he found "six and more" larvae in the same inflorescence and that three larvae in the same flower head "frequently happens". Our data differed greatly from the observations of Fabre (see Table 3). Apparently, two larvae could finish full development in a medium-sized flower head (Fig. 26). Larvae were recorded from the end of June to the end of August. Larval development has been described and discussed in detail by Fabre (1922). According to direct observations by Fabre, larvae feed primarily on the sap of the receptacle. Mature larvae build a pupation cell (4-5 × 7-10 mm) (Figs 29-31), with the walls significantly stronger than those of the larval cell (Figs 32-33).
After emergence, adults remained in the dry inflorescence for 5-7 days until fully sclerotised (Fig. 37), with an upper surface that was usually densely covered with a rust, pollen-like flush (Figs 34-36). Adults hibernated outside host plants, most likely in the top layer of soil or among dry plant debris.
Preimaginal development outside of flower heads. In the second part of the summer in 2015, the second author found seven chambers on the side of the stem of Echinops ruthenicus. All were located 1-2 cm below the inflorescence. Two of the seven capsules were located on the same stem and touched one another. The largest chamber was approximately 0.7 × 1.25 cm. The walls of all chambers were solid but rather brittle and easily crushed by the fingers. The material of the walls had no taste. The external surface of the capsule wall was brown to black, rough, earthy coloured, lumpy and mat; with incrustations of elements characteristic for Echinops, i.e., woollike coating of the stems, small leaves with spines and tips of spines (Figs 28, 38-39).
The inner surface of the wall was denser than that of the external item and was bright red-brown with yellowish spots and veins, glossy, and 0.8-1.0 mm thick. The inner surface was also smooth and appeared varnished and polished, similar to inner surface of the wall of the usual pupa chamber of L. vulpes, but without any incrustations. The contents of the chambers included the following: 2 chambers with dead imagines, 1 chamber with a living imago, one with only the head capsule of a larva, one chamber with a mature larva alive (Fig. 34), one with only pieces of a pupa and one with only parts of exuviae or cuticle remains. No excrement was observed but clearly the larvae incorporated their faeces into the wall during construction.
Biotic interactions. A parasitic wasp, Bracon urinator (Fabricius, 1898) (Hymenoptera: Braconidae), was reared from the pupae of L. vulpes. Sometimes the capitulum dried out, cracking the walls of larval or pupal cells, and ants, Formica imitans Ruzsky, 1902, destroyed the larvae and pupae of the weevil. Feeding on the inflorescences of Echinops, the rose beetle, Protaetia metallica (Herbst, 1782) (Coleoptera: Scarabaeidae), harmed larvae of L. vulpes developing in the same inflorescence. We found cells with dead weevil larvae together and simultaneously with living larvae or adults of carnivorous bugs, Orius sp. (Heteroptera: Anthocoridae) (Volovnik 1994).
Larinus vulpes and its host-plants. Larvae of L. vulpes living in flower heads of globe thistle consumed the ovaries and unripe seeds. The prevalence of weevils in globe thistles sometimes reached 33% of inflorescences (SV, unpublished data). Some flower heads lost all their seeds, although in an overall view, the loss of some seeds may be expected. Globe thistles have a special morphological structure that separates the compound flower head into small, single primary capitula with one seed each (Mulkijanian 1951). The durable walls of the pupal cell were very strong, which might prevent the possible separation of each capitulum. Furthermore, Mulkijanian (1951) also recorded stems breaking in the wind, primarily when larvae damaged the receptacle. Most likely, the larvae of Larinus vulpes or some similar species of Larinus damaged the receptacles. Clearly, the specialisation of globe thistles to a specific environment, such as rocky slopes, also constrained the weevils, being adapted to that same environment. The chemical composition of Echinops is also somewhat specific, and some species in this genus have insecticidal, nematicidal, antifungal, bactericidal and also antiviral properities (i.e. Fokialakis et al. 2006;Zhanget al. 2009;Abdelnabby and Abdelrahman 2012;Tekwu et al. 2012;Gemechu et al. 2013;Liu et al. 2013). Therefore, the inhabitants of Echinops flower heads might also be protected from some parasites and predators because of the chemical compounds in the plant. This aspect requires further experimental investigations.
The absence of preimaginal development outside of flower heads is a characteristic of not only L. vulpes but also for the genus Larinus in general. Only six Larinus species construct analogous capsules (pupal chambers) with a sweet taste (known as "trehala") on plant stems, namely: L. capsulatus Gültekin, 2008;L. hefenborgi Boheman, 1845;L. nidificans Guibourt, 1858;L. ruficollis Petri, 1907(Gültekin 2008; L. trehalanus Shahreyary-Nejad, 2015 (Gültekin andShahreyary-Nejad 2015); and L. onopordi (Fabricius, 1787) (Khnzoryan 1951). According to Zwölfer et al. (1971), larvae of L. vulpes sometimes construct capsules on the stems, but the authors did not provide further details for this observation. Fabre (1922) once recorded the larval chamber built in the axil of the leaf of Echinops sp. Of note, L. vulpes and the trehala building species develop on plants of the same genus, Echinops, and perhaps the construction activity outside of flower heads might be to avoid properties of the thick adhesive sap of these plants. Of the seven Larinus species that develop on Echinops spp., all construct outer pupal chambers (L. vulpes included). According to Iljin (1953), E. ruthenicus and other globe thistles contain the type of rubber that turns into a dense substance when exposed to air. Closely related to Larinus weevils, Afrolarinus moestus (Chevrolat, 1882) also develops on Echinops, but only in the flower heads (Gültekin 2013).
Two important moments in the life cycle of Larinus vulpes remain unclear: (1) why do eggs appear outside flower heads? (2) can imagines open their chambers on the stems and emerge at the proper time? According to the assumptions of Fabre (1922), an egg could fall from the inflorescence after being laid, or the female could lay an egg in an unusual place "either by inadvertence or by intention". This event is unlikely because females preferred to lay eggs in the top of flower heads. Occasionally, females laid an egg on a lateral part but never on the bottom of flower head. Thus, falling out into the axil of a leaf is doubtful. Unlikely also was "inadvertence", because this phenomenon is not that rare. On a relatively small patch of grassland (approximately 150 m²), seven chambers were found. Therefore, females laid eggs in an unusual place, for enigmatic reasons. In previous years, Volovnik conducted direct observations of L. vulpes in the field but never saw its capsules on plant stems. In the region of investigation, the growing season for vegetation during 2016 was abnormally dry and hot. Most of the globe thistles were short with small inflorescences and died earlier than usual (Fig. 40). It is possible that such an extreme situation resulted in an acute shortage of suitable places for normal oviposition; thus, a significant portion of eggs were laid in unconventional places.