Molecular and morphological systematics of soil-inhabiting Cryptorhynchinae of the genus Acallorneuma and the tribe Torneumatini (Coleoptera: Curculionidae), with description of two new species

Starting from an ecological classification of the morphotypes of apterous western Palaearctic Cryptorhynchinae, molecular systematic and morphological results for the monophyletic weevil genus Acallorneuma Mainardi, 1906 and the tribe Torneumatini Bedel, 1884 are presented. Based on the mitochondrial CO1 barcoding region, we discuss the limits of comparative morphology in the uniform Acallorneuma species. A catalogue and a pictorial key of all 8 species of Acallorneuma are provided. In a second step we compare morphology-based systematics of the genus Acallorneuma with our molecular reconstruction. Finally, we focus on the related blind, equally wingless and uniform, currently 71 species of the tribe Torneumatini living deep in the soil. This overview of the present state of research shows that molecular intrageneric resolution is highly dependent on the number of sampled species, especially in those cases with particularly long edges in the dendrogram. But although Torneumatini sampling was not complete due to the elusiveness of these subterranean species, some taxonomic changes could still be implemented: Torneuma s. str. with the type species Torneuma caecum Wollaston, 1860 occurs only on the Madeira archipelago. The species of the subgenus Paratyphloporus Solari, 1937 stat. nov. - only from the western Canary Islands(!) - must be transferred into the genus subgenus Paratorneuma Roudier, 1956 stat. nov. For all other species of the Mediterranean area and the eastern Canary Islands, the systematic classification needs to be remade (incertae sedis, see also appendix 2). Torneuma deplanatum deplanatum (Hampe, 1864) is the type species of the subgenus Typhloporus that includes some, but not all Mediterranean species with a constantly deep and wide pectoral canal, which - as it now seems likely – was developed several times. Two new species are described: Torneuma (s. str.) isambertoi Stüben spec. nov. from Madeira and Torneuma (s.l.) cadizensis Stüben spec. nov. from the south of Spain. In both cases keys are given to differentiate from the closely related species. nur die von den Kanarischen Inseln – gehören in das Subgenus Paratorneuma Roudier, 1956 stat. nov. Für alle anderen Arten aus dem mediterranen Gebiet und den östlichen kanarischen Inseln ist eine endgültige Klassifikation zur Zeit noch nicht möglich (incertae sedis), auch wenn erste Gruppen - eingeteilt vor allem nach der Innensackstruktur des Aedoeagus - hier bereits vorgestellt werden (siehe Anhang 2). Torneuma deplanatum deplanatum (Hampe, 1864) ist die Typusart des Subgenus Typhloporus und schließt einige, aber eben nicht alle mediterranen Arten mit einem konstant tiefen Rüsselkanal ein, der – das zeigen unsere vergleichenden Studien – offensichtlich mehrere Male in der Evolution ausgebildet wurde. Zwei neue Arten werden abschließenden beschrieben: Torneuma (s. str.) isambertoi spec. nov. von Madeira and Torneuma (s.l.) cadizensis Stüben spec. nov. aus dem Süden Spaniens. Für beide Arten werden Schlüssel mit den nächst verwandten Arten präsentiert.


Ecological-morphological considerations
Cryptorhynchinae occur in almost all terrestrial habitats. Within the deciduous and the evergreen western Palaearctic forests these habitats reach from canopies over trunks, shrub layer and herbaceous layer down to the leaf litter zone (Fig. 1). Cryptorhynchinae also live in the humusrich topsoil or among calcareous or volcanic rocks of the subsoil. Their larvae feed on dying twigs, root crowns or roots of lignified (also poisonous) stressed plants that have previously been damaged e.g. by wind or rockfall.
It is noticeable that the morphological variability within genera is highest in the canopy and upper zones of the understory. Morphological similarity among congerenic species conspicuously increases closer to the soil level. Thus, leaf litter-dwelling species of the genus Echinodera show only limited exoskeletal differences, and an unambiguous identification by external morphology is highly difficult in Acallorneuma species living in the topsoil. Without consideration of the male genital, morphological identification becomes virtually impossible in the subterranean, blind and extremely uniform-looking species of Torneumatini.
Species of Acalles can be collected by sieving from the top layer of leaf litter in Central European forests or the detritus of low shrubs in Central and Southern Europe. Acalles specimens are brown and low in visual contrast, without conspicuous elytral marks or bands; in most cases also without bristle tufts or elytral protuberances. Without knowledge on aedeagi, species identification can often be very challenging, in some cases even impossible (molecular studies in: . Species of Echinodera occur mainly in the Mediterranean area and on the Macaronesian islands, often as endemics, and can be collected in considerable numbers from the bottommost, often moist layer of forest leaf litter. Other than the above species, these Cryptorhynchinae can usually not be collected by beating of shrubs or branches. Legs are very short, eyes narrow and partly reduced; overall body shape is oval. Species are hard to tell apart and punctuation of the pronotum, number of elytral bristles and shape of the aedeagus all need to be carefully scrutinized (molecular studies in: Astrin & Stüben 2010).
Species of Acallorneuma inhabit the uppermost thin soil layer among calcareous rocks in the southwestern Mediterranean zone, with numerous endemics. The inclusion of Acallorneuma within Cryptorhynchinae has yet to be validated (doubts arise e.g. from Stüben et al. 2015, similarly for Acallocrates and Torneumatini). Acallorneuma species are characterized by uncontrasting brown, flattened species, nearly 'bald' , without protruding bristles. Eyes very small, strongly reduced. Species very difficult or impossible to tell apart without inspection of the aedeagus (morphological studies in : Stüben 2006a).
Torneumatini encompass Mediterranean and Macaronesian, almost exclusively endemic species that always follow a subterraneous way of life and are associated with roots in often calcareous or volcanic rocks. All characters mentioned above are strongly reduced in this group: eyeless, extremely flattened, short-legged, nearly 'bald' , light brown species; virtually indistinguishable based on outward exoskeletal characters. For morphological species identification, inspection of the aedeagus and its internal sac is indispensable (morphological studies in: Stüben 2007).
The increasingly complex habitat structures illustrated in figure 1 are interestingly mirrored -to some degree -by the phylogenetic reconstruction for western Palaearctic genera of Cryptorhynchinae based on the mitochondrial COI + 16S and nuclear 28S genes (see Stüben et al. 2013: fig. 1). Indeed and generalizing, it can be said, that -for example -the Macaronesian tree climbers, the species of Dendro-and Silvacalles ), appear as highly derived, younger taxa, whereas the Acallorneuma and Torneumatini living near or within the soil are certainly older. It could be shown for the first species of Acallorneuma that they are around 30 million years old, wheras the species of Dendroacalles formed 7 millions of years ago, and some species of Silvacalles only separated just 600,000 years ago. The Kyklioacalles of the lower strata can be classified between these evolutionary tendencies: they probably have devoleped around 17 million years ago (Stüben & Astrin 2010b: fig. 1B).
It is questionable and not always helpful if such distinct genera and ecological classifications are partly 'leveled out' by molecular analyses based on mostly conservative (nuclear) genes. This low 'resolution' at genus level leaves something to be desired in the light of worldwide thousands of described Cryptorhynchinae species. And these sometimes lead to far-fetched proposals suggesting e.g. that a "splitting" in several Western Palaearctic genera could not be helpful and "Torneumatini (represented solely by [one specimen of] Torneuma in [the] analysis) appear to be part of the Acalles group" (Riedel et al. 2016: 9) -without having to preoccupy oneself with the substantial morphological and mitochondrial differences among the only 6 species-rich genera of the ca. 400 Cryptorhynchinae species in the Western Palearctic; overlooking the fact that the habitat and host requirements of these species are completely different! However, the different morphotypes (body plans) described above show a marked correlation with the ecological stratification in a tree/shrub community: in western Palaearctic Cryptorhynchinae, intrageneric morphological variability is highest in the canopy-dwelling taxa, decreases in the species living closer to the soil ('epigaion') and reaches almost complete uniformity in exoskeletal characters in the soil-dwelling taxa ('edaphon'). In other words: the higher and brighter the inhabited layer, the larger the morphological differences among congeneric species, enabling easier and faster phenotypic identifications. This also has substantial consequences for phylogenetic approaches in western Palaearctic Cryptorhynchinae. In this group, classification at genus level is still straightforward based on clear and practically useful morphological, ecological and molecular differences (a fact not recognized by Riedel et al. 2016, who muse about western Palaearctic cryptorhynchine genera while ignoring 98.7 % of the group's diversity). However at species level, the situation changes gradually: Recognizing evolutionary novelties on a morphological basis, without associated molecular analyses (like DNA sequencing), is still very well possible in species of Dendroacalles , Dichromacalles (Stüben & Behne 1998) or Kyklioacalles (Stüben 1999(Stüben , 2003, but becomes much more challenging for species of Acalles s. str. . Ultimately, for the strongly uniform epigeous or edaphic species, phylogenetic reconstructions based exclusively on morphological characters would be impractical. It is already difficult to discern, without additional infor-DOI: 10.21248/contrib.entomol.66.2.  mation, morphological differences among the scarce, reduced exoskeletal characters of species belonging to the genus Echinodera, or even more so in the subterraneously living species of Acallorneuma and Torneumatini. Within a cladistics analysis, interpretation of character polarity and character weighting based on morphology alone becomes impossible in such a scenario. The minimal changes and displacements in lengths, distances, length-width ratios of elytral setae or the minimal differences in puncture size and distance on the elytral intervals or on the pronotum are possibly sufficient for a simple dichotomous differential diagnosis of a species description. They do not, however, offer approriate guidance for establishing a character matrix and reconstructing the underlying evolution. These realizations from practical work in the mentioned species match the conclusions from purely morphological studies by the first author (Stüben 2006a(Stüben , 2007: species delimitation and plain dichotomous differential diagnosis require studies on morphology, ecology and reproductive biology; these however do not suffice to adequately meet the challenges of reconstructing a phylogeny in subterranean Acallorneuma and Torneumatini species. In such morphologically cryptic cases, molecular analysis offers the most promising choice (Stüben 2006a(Stüben , 2007, as many more characters are available than from comparative morphology (cf. Tautz 2006).
The present work aims mostly at answering three distinct questions: 1. What do we have to take into account when describing mostly uniform (not cryptic) species of Acallorneuma, and where do we reach the limits of comparative morphology? 2. What can we learn from the molecular reconstruction of Acallorneuma phylogeny and where are the differences from purely morphological reconstruction (for example by ? 3. What can be said about genus-level systematics and taxonomy of the soil-dwelling higher taxa of Torneumatini?

Material and methods
A molecular phylogeny of the western Palearctic weevil genera Acallorneuma Mainardi, 1906 and Torneuma Wollaston, 1860 is presented here in a Bayesian analysis (Fig. 3). In total the dataset contains 7 Acallorneuma species, 11 Torneuma species, 2 Paratorneuma species and 1 Paratyphloporus species. Furthermore, we included 4 Kyklioacalles (Cryptorhynchinae) sequences as outgroup taxa. Collecting and vouchering information as well as GenBank accession numbers are given in Appendix 1. We sequenced the (5') barcoding section of the CO1 gene.
Altogether 22 sequences were generated specifically for this study, the remaining 20 sequences were taken from previous studies of the ZFMK and Curculio Institute. Sequence length was 658 bp for CO1. Total genomic DNA vouchers and voucher specimens are deposited at the Zoological Research Museum Alexander Koenig (ZFMK). The laboratory routine followed Schütte et al. 2013, except samples marked with a star* in Appendix 1, which were processed as described in . In both cases the same pimers were used , based on the typical barcoding primers (Folmer et al. 1994): LCO1490-JJ 5'-CHACWAAYCATAAAGATATYGG-3' and HCO2198-JJ 5'-AWACTTCVGGRTGVCCAAARA ATCA -3'. DNA sequence alignment has been performed with Biomatters Geneious 6.  (Posada 2008), implementing the Bayesian information criterion (BIC, Schwarz 1978), identified the HKY+I+G model of nucleotide substitution (Hasegawa et al. 1985) as the best-fit model for the CO1 alignment provided. The sequence data of the mitochondrial CO1 gene was used in parallel Bayesian Markov chain Monte Carlo (MCMC) analyses, as implemented in MrBayes ver. 3.2.0 (Ronquist & Huelsenbeck 2003). We applied the model of sequence evolution diagnosed by the BIC (nst=2 rates=invgamma). Parameters were unlinked between the 3rd versus 1st plus 2nd codon positions. Analyses were run for 40 million generations using the default chain number and temperatures, sampling 40,000 trees (average standard deviation of split frequencies: 0.000816). Every 1,000th tree was sampled. Negative log-likelihood score stabilisation was determined graphically. Accordingly, we retained 39,000 trees. These were used for building a 50 %-majority rule consensus dendrogram with posterior probability values. Geneious was used to display the tree and also to calculate uncorrected (p-)distances provided in Fig. 7 and within the text.  Stüben, 2006) The morphological characters mentioned below enable differential diagnosis and offer a rough orientation among the 8 valid species of Acallorneuma. Focus characters are 1) the median lobe of the aedeagus, 2) elytral punctuation or form and arrangement of bristles on the elytral intervals and 3) species-specific ventral indentation of front femora (see pictorial key). However these characters do not suffice for a phylogenetic reconstruction as shown below.

Key to the species of the genus Acallorneuma
Even with such a pictorial key ( Fig. 2), it is difficult to identify unambiguously the often uniform species of Acallorneuma. Furthermore, there remains the question whether accurate species assignments can also be performed on females. Osella & Zuppa seem to believe this is feasible, as they depict the spiculum ventrale and the spermatheca in comparative tables of their revision of Acallorneuma , and even base their species descriptions mainly on the female genital, sometimes considering only a single female (e.g. in Acallorneuma sabellai syn. or -based on three females -in Acallorneuma poggii syn.). The first author has demonstrated that, with the exception of very few characters of the spiculum ventrale (apodeme length, form of bracchia), the female genital is unfit to characterize species of Acallorneuma adequately (Stüben 2006a). When larger series of specimens are available for investigation, the high variability of the spiculum ventrale (ibid. Tab. 454.14) and also (to a lesser extent) of the spermatheca become apparent.  failed to prove the stability of these in fact unspecific characters, especially with regard to the species Acallorneuma reitteri, where many specimens were available to them and where they acknowledged the high variability   fig. 6a-c).
The high variability of the spiculum ventrale is typical not only for species of the genus Acallorneuma, but for all Cryptorhynchinae of the Western Palaearctic. Unambiguous (re-) description of species belonging to the genus Acallorneuma -or in fact to any other genus of Cryptorhynchinae -should therefore never be based solely on female specimens, especially not exclusively on the spiculum ventrale (in more detail: Stüben 2006a).
Concluding, we hold that the few (mostly male) morphological characters used here in compiling the Acallorneuma key can serve as a first orientation aid within the genus, but do not -as with the uniform Torneumatini (see below) -constitute a sufficiently solid basis as to induce hypotheses on species relationship.

Morphological and molecular systematics of the species of Acallorneuma Mainardi
Our molecular analysis is based on 15 individuals belonging to 7 of 8 valid Acallorneuma species (sequence of Acallorneuma peyerimhoffi is not yet available). The resulting phylogeny shows marked differences from the first reconstruction for the genus obtained by Osella and Zuppa (2002) in their 'analisi filogenetica del genere Acallorneuma Mainardi, 1906' . We have reproduced their 'morphological' (i.e. morphology-derived) tree in Fig. 3 (top right). Using the same weevil genus as a model system, we can thus contrast two different character systems typically used in phylogenetics against each other. Tautz (2006) holds that morphological and molecular trees have to be considered independently from one another. However, striking inconsistencies should prompt us to search for erroneous assumptions or glitches in either method. It has to be noted that Acallorneuma ibericum Stüben from southern Spain could not yet be included in the analyses of , as it was described at a later point . Regarding the other taxa, phylogenetic placement varies considerably between the morphological tree and the molecular dendrogram presented here. 1. First of all we have to consider the imbalance between number of species vs. number of characters within the morphological reconstruction, an aspect often underestimated in morphology-based phylogenetics.
It is difficult to imagine that, like in the present case, 15 characters should suffice to establish a meaningful cladogram (Hennig 1966) for 10 ingroup species (cf. Osella & Zuppa 2002: 464). In cases of morphological uniformity such as the present, DNA sequence analysis seems to be the better option as it offers many more (quantifiable) characters at species level.
Within an integrative taxonomic framework, it has been frequently shown that DNA-based phylogenies prompted the search for new morphological characters, led to modified character weighting or helped in unmasking plesiomorphic morphological characters. Molecular data have long since become an important corrective for morphologically-oriented taxonomists.
2. A purely morphological phylogenetic reconstruction does not gain additional precision or meaningfulness by subdividing already analyzed characters into several subcharacters. Such a practice merely leads to multiplication of the existing evidence. Over-splitting of characters has to be critically kept in mind when considering the characters connected to the spiculum ventrale (female genitalia) or the median lobe of the aedeagus (male genitialia) (see : Table II). Splitting complex characters into various components would be justified, however, if a gradual evolutionary increase in character complexity had been proven. Otherwise, complex characters are weighted too strongly just because they 'strike the eye' , or because they fit the picture that already exists in the experienced taxonomist's / systematist's mind, a common danger in purely morphological phylogenies. It is not our intention to criticize such an intuitive phylogenetic approach by experienced taxonomists -on the contrary, the resulting concepts often show striking concordance with 'proper' , quantitative molecular phylogenies without room for interpretation. If this was not the case, we would have to redefine the largest part of insect systematics.
Thus, our critique is mostly directed at a-posteriori pseudo-legitimations of such intuitive phylogenetic concepts, which -in morphologically highly similar taxa -expose themselves to the suspicion of manipulative selection of characters. The same applies to the species of Glaberacalles. But these adaptational (better: eco-functional) similarities in body plan cannot prompt the conclusion that both taxa are necessarily closely reated. In fact, Acallorneuma species always cluster very 'basally' (often in conjunction with Acallocrates) in the various molecular western Palaearctic Cryptorhynchinae trees , and it cannot even be taken for granted that their inclusion within the subfamily Cryptorhynchinae is justified.

Intergeneric classification of the Torneumatini Bedel, 1884
Even more uniform than Acallorneuma are the species of the tribe Torneumatini. If not for the partly complex internal sac structrures of the aedeagus (endophallus), one would have to speak of cryptic species in most Torneumatini. The currently 71 valid species and subspecies in the tribe have been assigned, according to length and form of the pectoral canal, into initially three, later four genera (Stüben 2007, see Fig. 4): -Pseudotorneuma Solari, 1937: lacks a pectoral canal or a mesosternal receptaculum between the mid-coxae; -Paratyphloporus Solari, 1937: with pectoral canal that forms only a flat depression in front of the forecoxae (Paratyphloporus I) or that ascends right before or between the fore-coxae, to drop steeply from here towards the mesosternal receptaculum (Paratyphloporus II); -Torneuma Wollaston, 1860: with a constantly deep, tube-formed pectoral canal, reaching from the anterior margin of the prosternum to the mid-coxae and ending between the mid-coxae in an equally deep mesosternal receptaculum.
This preliminary subdivision (see Fig. 4), oriented at morphologically easily evaluable characters as the pectoral canal and mesosternal receptaculum, was meant as a useful provisional solution from early on (see Stüben 2007). Such heuristic instruments cannot convey any meaning on character polarity. Therefore the question could not be addressed whether Torneumatini originally completely lacked any pectoral canal (this seems to be the notion of Roudier 1956), or if instead the lack of a canal constitutes a secondary reduction. If we were dealing with a gradual mode of evolution, we would always find transitional forms (as intermediate character states) and would never be able to define (sub)genus-specific characters with certainty or conclusively.
In fact, recent molecular phylogenetic reconstructions (e.g. Stüben & Astrin 2010a) have shown that the pectoral canal has been acquired and reduced several times in Torneumatini evolution. As a first approximation, we currently reach the following conclusions:  (Formánek, 1912) from the Dalmatian coast, because also these species evolved the pectoral canal in form of a more or less shallow depression in front of the fore-coxae (Stüben 2007(Stüben , 2008. However, it was overlooked, that the rhomboid-like structures of the internal sac of the aedeagus of the Paratorneuma species from the Canary Islands are very different from the complex structures of the Paratyphloporus species from the Mediterranean area (see the figures in Stüben 2007). Furthermore, the latter also have a deeper depression in front of the fore-coxae and/or these fore-coxae are far away from each other. In other words: although the more or less deep canal of the rostrum in front of the fore-coxae is present in both genera -reaching its highest point between the coxae and dropping steeply from here towards the mesosternal receptaculum -, it seems to be clear that both genera are not closely related as we had stated previously. It should therefore be presumed that we are here confronted with a convergent evolution of a reduction of the rostrum-canal, which leads de facto to the total loss of the rostrum-canal in the case of the few North African Pseudotorneuma species.
What can we say conclusively about the three species from Gran Canaria, T. viti, T. solarii and T. canariense (all decribed by G. & M. Osella, 1984)? At this time, the molecular basis is too little and the high molecular p-distances of the mitochondrial CO1 gene do not allow to include the species into the subgenus Paratorneuma. Furthermore, the rostrum-canal of these species from the eastern Canary Islands is clearly deeper, the fore-coxae are wider compared to the Paratorneuma species, the elytrae are longer and the internal sac of the aedeagus, consisting of two parallel bars or lines, is different from the species of the western Canary Islands. Therefore these last three species cannot be directly allocated currently to a subgenus.
However and for the time being, they are considered incertae sedis with regard to a subgenus until more molecular data of more Mediterranean species are available.

What does this mean in practice? Even though most
Torneumatini cannot be told apart by exoskeletal characters, using the partly complex structure of the internal sac of the aedeagus usually allows easy assignment of specimens to individual groups (cf. structure of the internal sac of the aedeagus and the depth of the rostrum channel) will hold in the light of a molecular phylogenetic reconstruction by using more genes. But the fact that 'not more' than 21 out of 71 species of Torneumatini could be analyzed molecularly in the last eight years exemplifies how much patience we will have to muster until a conclusive systematic classification will be reached: the search for these subterranean species in the Mediterranean area and on the Macaronesian islands is among the most difficult, but also among the most urgent tasks in present-day Cryptorhynchinae research. (Neither sieving nor floating techniques seem to have predictable success -even in places where Torneumatini species have previously been found). Therefore, and keeping in mind the conspicuously long branches of the dendrogram (CO1 p-distances among species mostly 15 % or above), it would be premature (see Fig. 3) to formally consider a homogeneous Typhloporus subclade.

Taxonomy
Preliminary remark: A quick and unambiguous determination of Torneumatini species should always begin with a meticulous examination of the structure of the male endophallus. An identification of females is not promis-ing considering the high number of species and also in the light of the taxonomic 'meaninglessness' of female genitalia in the tribe. Description of new species based on a single female should be avoided and make morphological revision or simple species description impossible in whole or in part! A molecular analysis should be obligatory in any such case. 1. Elytra short-oval, broad: < 1.68 x as long as wide: (Fig. 8-9 3. Elytra longer, ogival-shaped in front of the apex (more clearly in females) (Fig. 6); median lobe of the aedeagus 9x-10x as long as wide (Fig. 13). 3*. Elytra shorter, ovally rounded towards the apex (without flat dents on each side immediately before the apex) (Fig. 5, 10-11); median lobe of the aedeagus at most 2.2x as long as wide (Fig. 12, 16-17 4. Lower part of the interior sides of fore-tibiae () 'sickle-shaped'; aedeagus: median lobe in the middle part convex and more acuminate: (Fig. 16). Habitus: (Fig. 10) Interior sides of the fore-tibiae () nearly straight; aedeagus: median lobe parallel or narrowing rectilinearly and less acuminate: (Fig. 12, 17)
Venter: With a constantly deep and wide pectoral canal, from the fore-margin of the prosternum to the midcoxae, terminating between the mid-coxae in an equally deep mesosternal receptaculum (Fig. 5); fore-margin of the prosternum low-cut, forming an arc of a circle; the ground between the prae-coxae slightly lifted, dropping down towards the mesosternal receptaculum. The distance between the prae-coxae is large, as is their diameter. The brink of the mesosternal receptaculum is semicircular and sharp-edged; base of the receptaculum with a step. 1 st and 2 nd abdominal segment of the male with a wide and flat hollow.
Head & Rostrum: Without eyes; T. isambertoi belongs to the 'long-nosed' species: rostrum brown, 3.50x () and 3.70x () as long as wide between the insertions of the antennae; with a fine median edge and on each side with a conspicuous and elongated edge; insertions of the antennae near the apex, approximately at the end of the first quarter () or first third ().
Pronotum: 1.18-1.20x as long as wide; widest at the end of the basal third of the pronotum; narrowing rectilinearly laterally towards the fore-margin, clearly less rounded than towards base. Disc of pronotum flattened, with fine, not deep punctures, covered with numerous round scales. The fore-margin of the pronotum with a slight indentation in the middle. Elytra: Brown-russet, strongly flattened and elongate, 1.80x () -1.90x () as long as wide; elytra in both sexes approximately parallel-sided ('cylindrical') in the middle, ovally rounded directly in front of the apex; base line of elytra curved slightly S-shaped. The puncture stripes clearly more slender than the slightly arched intervals. These are covered with a single row of very fine, short and hardly discernible bristles.  Aedeagus: Median lobe 1.9x as long as wide, viewed laterally it forms approximately a right angle (Fig. 12).

Etymology:
The new species is dedicated to Isamberto Silva (Madeira, Funchal), who has supported the first author with his excellent expert knowledge to collect the Curculionidae on the Desertas Islands. As a "Vigilante da Natureza" he is doing outstanding conservation work on the Madeira archipelago. His collection of insects and molluscs is unique.

Ecology:
The new species was sieved out of dead and broken branches (detritus) by the first author in the ground under Ficus carica and Euphorbia piscatoria always together with Torneuma korwitzi (Fig. 7, bottom right) in the proportion 1:3 from the banana plantations near Paul do Mar (Madeira).
Distribution: So far this species is only known from the southwest of Madeira (Portugal). An overview of the species, collecting localities and the populations of all Torneuma species known from the Madeira Archipelago as well as the p-distances of the mitochondrial COI gene between Torneuma isambertoi and the related species are presented here in map: (Fig. 7).   Description of the holotype (Fig. 18, 20): Length: 3.25 mm (without rostrum).
Venter: With a constantly deep and wide pectoral canal, from the fore-margin of the prosternum to the midcoxae, terminating between the mid-coxae in an equally deep mesosternal receptaculum (Fig. 18); the distance between the prae-coxae is large, as is their diameter. The brink of the mesosternal receptaculum is semicircular and sharp-edged; base of the receptaculum with a step. 1 st and 2 nd abdominal segment of the male with a wide and flat hollow.  Head & Rostrum: Without eyes; T. cadizensis belongs to the 'long-nosed' species: rostrum brown, 3.8x () as long as wide between the insertions of the antennae, without a mid-edge, but on each side with a short edge; insertions of the antennae near the apex, approximately at the end of the first third ().
Pronotum: 1.16x as long as wide; widest at the end of the basal second fifths of the pronotum; narrowing rectilinearly laterally towards the fore-margin, clearly less rounded than towards base. Disc of pronotum flattened, with fine, deep and extensively placed punctures, covered with numerous round scales and isolated, scarce bristles. The fore-margin of the pronotum with a slight indentation in the middle.
Elytra: Brown-russet, flattened (much flatter and broader than those of T. morandae, see Fig. 19), 1.72x () as    long as wide; elytra approximately parallel-sided in the middle, ovally rounded directly in front of the apex; base line of elytra curved slightly S-shaped. Puncture stripes on the front half almost as broad as the arched intervals, on the posterior half clearly more slender. These intervals are covered with a single row of very fine, but easily discernable bristles.
Aedeagus: Median lobe 1.45x as long as wide, with a simply V-shaped structure of internal sac (Fig. 20) characteristic for the species of the T. robustum group (see key below and also Fig. 21).

Etymology:
The species name refers to the province of Cádiz (Spain), in which the new species was found near La Línea de la Concepción (Fig. 22).

Ecology:
The single specimen of the new species was collected by the first author's colleague J.L. Torres (Spain, La Línea) under Quercus coccifera.
Distribution: So far this species is only known from the Sierra Cabonera near La Línea de la Concepción in the south of Spain.