Description of the larval stages of Gymnochthebius jensenhaarupi and phylogenetic analysis of the relationships with other species of the subfamily Ochthebiinae ( Coleoptera : Hydraenidae )

The three larval instars of Gymnochthebius jensenhaarupi (Knisch, 1924) are described and illustrated, including a detailed analysis of their chaetotaxy and porotaxy. The specimens used in this study were collected with adults of G. jensenhaarupi and have been identified as such by association. Comparative notes on the morphology of these larvae with other species of the subfamily Ochthebiinae are given. A hypothesis of phylogenetic relationships between G. jensenhaarupi and other members of Ochthebiinae with thoroughly described larvae is presented. The monophyly of Ochthebiinae is supported by additional larval features. On the other hand Ochthebius, as currently composed, seems to by paraphyletic. Gymnochthebius Orchymont, 1943 is confirmed as the sister group of Aulacochthebius Kuwert, 1887.


INTRODUCTION
Gymnochthebius Orchymont, 1943 is a genus of Hydraenidae composed by thirty-one species.Most of them are endemic to the Neotropical (20) or Australian (6) regions and a few (3) are known exclusively from the Nearctic region.The two remaining species have a wide American distribution and they are found both in the Neotropical and Nearctic realms (Hansen, 1998;Perkins & Archangelsky, 2002).As some other ochthebiine genera Gymnochthebius was originally proposed as a subgenus of Ochthebius Leach, 1815 but was convincingly considered as a distinct genus by Perkins (1980).Hansen (1991) pointed out morphological similarities between adults of Gymnochthebius and Aulacochthebius Kuwert, 1887, but it was Perkins (1997) who explicitly postulated a close phylogenetic relationship between both genera (along with Gymnanthelinus Perkins, 1997) in a lineage comprising also the genera Ochthebius, Hughleechia Perkins, 1981 andMicragasma Sahlberg, 1900 within the subfamily Ochthebiinae.Unfortunately, this close phylogenetic relationships could not be tested in Beutel et al. (2003) since these authors did not include Aulacochthebius in their cladistic study.
As in many other groups of Hydraenidae, the immature of Gymnochthebius are poorly known, thus these stages have been of little help until now both in taxonomic and in phylogenetic studies.Richmond (1920) studied the larva of G. fossatus (LeConte, 1855) (under Ochthebius tuberculatus LeConte, 1878) but, even though lengthy, this description lacks good illustrations and does not mention key morphological characters of hydraenid larvae.Recently, several adults of Gymnochthebius jensenhaarupi (Knisch, 1924) were collected in Argentina associ-ated with a number of larvae that can be assigned to this species.This finding offers a good opportunity to describe and figure the larval morphology of Gymnochthebius comparing its morphology with those of other known larvae of Ochthebiinae in order to evaluate their relative position within this subfamily.

First instar
Elongate, slender.Head and body sclerites moderately sclerotized.Colour of sclerites uniformly yellowish.Total body length about 1.2 mm.
epistomal ridge moderately developed near the dorsal mandibular articulations; posterior arms of tentorium widely separated by the gula; dorsal arms short; tentorial bridge moderately narrow; anterior arms well developed.

Abdomen
Abdomen subcylindrical in cross-section, straight to slightly curved ventrally and tapering posteriorly.Abdominal terga and sterna I-VIII each with a well sclerotized plate; abdominal pleura with two small sclerites: a dorsopleural and a ventropleural sclerite on each side.Tergites I-VIII as in Fig. 13.Number of setae and campaniform sensilla reduced: setae A1, Db1, Dc1, P2 and sensilla C1, C2, C4 absent.Remaining setae are arranged in a transverse row.Dorsopleural sclerites of abdominal segment I, with a circular annular spiracle and one seta (DP1).Dorsopleural sclerites of segments II to VIII (Fig. 13), each with a spiracle (SP) of the same kind to that of segment I and two setae (DP1, DP2).Ventropleural sclerites of segment I each with one seta: VP1.Ventropleural sclerites of segments II to VIII (Fig. 15) each with two setae (VP1, VP2).Sternite I (see Fig. 20) with five setae on each side: one small presternal seta (PS1), one discal seta (D1), and three posterior setae (P1 to P3).Sternites II-VIII as in Fig. 15, similar to sternite I but with two additional setae (D2 and P4).Segment IX bearing a pair of urogomphi, without dorsopleural nor ventropleural sclerites and with reduced chaetotaxy.Urogomphi (Fig. 22) two-segmented: segment I (URI) mobile, not fused to tergite IX, with six setae (U1-U6) and four campaniform sensilla (C1-C4), seta U6 far from base of urogomphus; segment II (URII) 0.25× as long as segment I, with an apical seta (AS).Abdominal segment X as that shown in Fig. 21, but smaller since this figure refers to a third instar larva, forming a completely sclerotized ring, with three long ventrolateral setae of unknown homology and three spine-like ventral setae (SV) on each side.Anal lobes (Fig. 21) moderately developed, each with one anal hook (AH), frequently retracted.

Second instar
Total body length about 1.6 mm.

Thorax
Pronotum as in first instar, with one additional subprimary seta on each side (Fig. 10); according to their position these setae should be considered as Db'.Subprimary setae Da', Da", Db'', Dc' and Dc'' absent.Mesoand metanotum similar in shape to those of first instar, with three pairs of additional subprimary setae: Db', Dc' and Dc''.

Abdomen
Abdominal dorsopleural sclerites I to VIII with a pair of additional subprimary setae (DP').Abdominal ventropleural sclerites and sternites (Fig. 16) without additional subprimary setae.Dorsopleural sclerites not fused to their respective tergites.Shape and chaetotaxy of segments IX and X as in first instar larvae.

Third instar
Total body length about 2.4 mm.

Thorax
Pronotum and mesonotum as in Figs 11 and 18, respectively.Chaetotaxy as that of second instar larvae.

Abdomen
Besides an obvious size increase, the single morphological distinction of third instar larvae is in segments I to IV in which the dorso-and ventropleural sclerites are fused to their respective tergites and sternites (Figs 19 and 20).Consequentially, the chaetotaxy of the dorsolateral plates is the addition of the setae from the tergites plus those from the dorsopleural sclerites, while the chaetotaxy of the ventrolateral plates is the addition of the setae from the sternites plus those from the ventropleural sclerites.Setae DP1, DP2, DP' and spiracle placed on lateral margins of dorsolateral plate; setae VP1 (abdominal segment I) and VP2 (all abdominal segments) placed on lateral margins of lateroventral plate (Figs 17 and 20).Shape and chaetotaxy of segments IX and X (Fig. 21) as that of first and second instars.

Habitat
The larvae were found associated to filamentous algae growing on the margins of small montane creeks, in areas where the water flow was slow.

Morphological remarks and comparative notes
The larva of G. jensenhaarupi resembles those ochthebiine larvae with well developed anal hooks, namely all the described larvae of Ochthebius s. str.as well as the larva of Aulacochthebius exaratus (Mulsant, 1844).Compared with these larvae, G. jensenhaarupi can be recognized by the elongation of some structures and appendages such as the solenidium IIS1 of antennomere II, maxillary palpomeres, legs (especially distal segments) and urogomphi.However, these general features are not useful enough to clearly distinguish among these larvae with certainty.More informative is the shape of the cephalic egg-bursters in first instar larvae.Gymnochthebius jensenhaarupi differs from all described ochthebiine larvae by the strong reduction of these cuticular structures.Richmond (1920) reared and described first instar larvae of G. fossatus, and in his description there is no reference to the presence of egg-bursters.We believe that these structures may have been overlooked by Richmond (1920) due to their minute size, and that this trait is probably a synapomorphy with other larvae of the genus Gymnochthebius.
Besides this feature, only useful in first instar larvae, G. jensenhaarupi can only be distinguished with certainty from the other described Ochthebiinae larvae using pore and chaetotaxic traits.Gymnochthebius jensenhaarupi differs by the strong reduction of prothoracic discal setae Db1 and Dc1  and by the presence of abdominal sternal setae P4 on segments II to VIII .The minute prothoracic setae Db1 and Dc1 are the most evident diagnostic characters for this species.Gymnochthebius fossatus has well developed prothoracic discal setae in rows Da, Db and Dc, described as "median transverse row of setae" by Richmond (1920: see text and Plate III, Fig. 12).
The character matrix analyzed (Table 1) includes 23 binary characters and one multistate character (24), which was treated as non-additive (unordered).The absence of information is represented by a question mark (?) in Table 1.The larva of Limnebius cordobanus d'Orchymont, 1938 (Delgado & Soler, 1997c) was used as the outgroup.Since the monophyly of the ingroup is not in doubt (Beutel et al., 2003, Perkins, 1997) we used only one outgroup.However, several character comparisons were made with the larvae of Hydraena hernandoi Fresneda &
Character 10.Cephalic ventral setae V2.Two states.(0) present; (1) absent.A pair of ventral setae (V1 and V2) is present on each side of the cephalic capsule of Limnebius larvae (Delgado & Soler, 1997c) and in several larvae of Ochthebiinae.The larvae of O. quadricollis, O. subinteger and A. exaratus have only a single ventral seta on each side of the head capsule.CI = 0.5.
Character 12.Additional prothoracic campaniform sensilla C'.Two states.(0) absent; (1) present.The presence of two additional campaniform sensilla on the pronotum in the three known species of Meropathus is a unique trait that should be considered an autapomorphy for this genus (Delgado & Palma, 1998).CI = 1.

RESULTS OF THE CLADISTIC ANALYSIS
Twenty-four morphological characters were used in this phylogenetic analysis as is indicated in Table 1.Several of the characters used are autapomorphies, therefore not informative, but they are included because of their importance as diagnostic characters.The cladistic analysis yielded a single shortest tree (length 29; CI = 0.87; CI excluding uninformative characters = 0.76; RI = 0.76; RC = 0.659), shown in Fig. 23.Monophyletic groups and apomorphies are commented in the following paragraphs.
The subfamily Ochthebiinae is clearly monophyletic, characterized by five autapomorphies: i) presence of cephalic egg-bursters in first instar larvae (character 2); ii) a single solenidium at the distal end of the antennomere II (character 3); iii) short antennal solenidium IIIS3 (character 7); iv) a single pair of thoracic glands (character 13); and v) urogomphi inserted very close to each other (character 20).
Finally, the clade formed by G. jensenhaarupi and A. exaratus is weakly supported by a single transformation: presence of abdominal sternal setae P3 (character 19); jackknife values for this clade range between 38-49.

DISCUSSION
As stated previously, this analysis represents an attempt to solve the phylogenetic relationships of G. jensenhaarupi, as a representative of the genus Gymnochthebius with other species of Ochthebiinae based on larval characters.However, in addition to the main subject of this work, several other points that we find of interest deserve some comments.
Our work provides additional support to the monophyly of the subfamily Ochthebiinae.In this sense, larval morphology is consistent with recent studies based on adult morphology (Beutel et al., 2003;Perkins, 1997).
The close phylogenetic relationship between O. quadricollis and O. subinteger, two members of the questioned subgenera Calobius Wollaston, 1854 and Cobalius Rey, 1886 of Ochthebius respectively, is another interesting result of this study.The present analysis suggests that members of this clade have accumulated a high number of apomorphic character states during the evolution of the subfamily Ochthebiinae (five larval characters).Both this close relation and their derived placement within Ochthebiinae, has been suggested previously in other studies dealing with both adult (Perkins, 1997;Beutel et al., 2003) and larval morphology (Delgado & Soler, 1996a).Our results reinforce these ideas, but conflict with the taxonomic status given by Perkins (1997) to these subgenera.Perkins (1997) regarded the clade Calobius + Cobalius as the quadricollis species group, synonymizing Calobius and Cobalius with Ochthebius s. str. in the light of an extensive adult morphological analysis.Hansen (1998) followed this suggestion in the case of Cobalius, but maintained Calobius as a valid subgenus.Recently Beutel et al. (2003) have suggested a separate generic status for Calobius based on several adult head traits.Larval morphology points in this same direction since Ochthebius, as currently composed, seems to be paraphyletic.The five species of Ochthebius studied by us are nested separately within the two main clades of our cladogram.In fact, the non-monophyly of Ochthebius is hardly a surprise since for a long time there has been little compelling evidence for the monophyly of this genus and virtually nothing is known about the relationships among its different subgenera (Hansen, 1991).Our results suggest that at least some species of Calobius and Cobalius could represent a different lineage within Ochthebiinae.Interestingly no synapomorphies for the members of Ochthebius s. str.were found in our analysis.However, in the present study we do not suggest any changes to the current classification of Ochthebiinae since our results are based on the reduced number of taxa presently at hand and this could lead to a spurious conclusion.
The genus Meropathus is depicted in our cladogram as a well supported clade.Larvae of this genus are very distinctive by the absence of campaniform sensilla FC1 and setae DP', as well as by the presence of additional sensilla C'.It is noteworthy that two other characters observed in these larvae: presence of a globose ligula and a strong reduction of anal hooks (characters 5 and 24), seem to have evolved at least twice during the evolution of the Ochthebiinae (Fig. 23), once in Meropathus, and once in the clade O. quadricollis + O. subinteger.
As we have pointed out in the introduction, Gymnochthebius was proposed to be closely related to Aulacochthebius by Hansen (1991) and, especially by Perkins (1997) who based this phylogenetic relationship on the similar shape of some aedeagal and abdominal structures of the adults.The cladogram produced in the present work supports this close relationship with larval characters, adding a new putative synapomorphy: the presence of abdominal sternal seta P3 in the larvae of these two genera.Unfortunately, this study failed to discover more synapomorphic features between these larvae.However, our results suggest that larval chaetotaxic traits offer valuable information, which can be used to critically appraise the current classification and phylogenetic relationships of Hydraenidae.Evidently, future studies should include more taxa, improving progressively our knowledge about the phylogenetic relationships of this interesting group of water beetles.

Fig. 23 .
Fig. 23.Provisional phylogeny of known Ochthebiinae larvae.Character state changes are superimposed onto the cladogram; black circles = synapomorphies; white squares = homoplasies (independent gains).The upper bold numbers on some branches represent the range of jackknife values.