On the systematic position of the diving-beetle genus Pachydrus ( Coleoptera : Dytiscidae : Hydroporinae ) : Evidence from larval chaetotaxy and morphology

Phylogenetic relationships within the diving-beetle subfamily Hydroporinae are not well understood. Some authors include the genus Pachydrus Sharp, 1882 in the tribe Hyphydrini, whereas others are in favour of excluding Pachydrus from the Hyphydrini and placing it in its own tribe, Pachydrini. Larval characters have been underutilised in phylogenetic studies, mainly because the larvae of many taxa within the family are unknown. In this study, the phylogenetic relationships of Pachydrus are studied based on a cladistic analysis of 34 taxa and 122 morphological larval characters. For this purpose, larvae of P. obesus Sharp, 1882 are described and illustrated in detail for the first time, with particular emphasis on morphometry and chaetotaxy. First and second instars for the genus were unknown. The results support a monophyletic origin of the tribe Hyphydrini excluding Pachydrus, based on four unique character states. On the other hand, Pachydrus is resolved as the sister group of the Hydrovatini. These results suggest Pachydrus should not be placed in the Hyphydrini. Given that the Hyphydrini minus Pachydrus is a distinctive clade, based on this study, it seems useful to recognise this group as Hyphydrini. Including Pachydrus in Hyphydrini would leave the tribe with a single larval apomorphy, as most characters present in the Hyphydrini and Pachydrus are also present in the Hydrovatini. However, in the absence of larvae of Heterhydrus Fairmaire, 1869 and of a more comprehensive and inclusive analysis, we do not propose a formal exclusion of Pachydrus from Hyphydrini at this stage. Pachydrus is a highly distinctive genus within the Hydroporinae and is characterised by several larval apomorphies.


INTRODUCTION
Pachydrus Sharp, 1882 is a diving-beetle genus including nine species, all inhabiting the New World (Biström et al., 1997;Nilsson, 2001).The genus includes small-sized individuals of globous shape, and is predominantly Neotropical, with one species [P. princeps (Blatchley, 1914)] reaching the Southeast of the Nearctic Region.Pachydrus obesus Sharp, 1882, the type species of Pachydrus, is widely distributed in South America, from Venezuela and Brasil to central Argentina (Trémouilles 1995).
Pachydrus is most commonly included in the tribe Hyphydrini of the subfamily Hydroporinae.Hyphydrini includes 15 genera (Nilsson, 2001) of which only two are present in America, Desmopachria Babington, 1841 and Pachydrus.The phylogenetic position of Pachydrus has been controversial for many years.The genus was initially included (along with Heterhydrus Fairmaire, 1869) in the tribe Bidessini since the adults have the metacoxae fused to the basal abdominal sternite (Sharp, 1882).Zimmermann (1919) transferred both genera to the tribe Hyphydrini, characterised by the metatarsal claws unequal in length (Biström, 1982;Miller, 2001).Young (1980) emphasised that, even though Pachydrus is commonly included in the tribe Hyphydrini, it is not closely related to the old-world Hyphydrini genera and suggested that the genus should be placed in a new tribe, Pachydrini.Biström et al. (1997), based on a cladistic analysis including all Hyphydrini genera, found that the clade Pachydrus + Heterhydrus was resolved as the sister group of the remaining Hyphydrini.Biström et al. (1997) postulated that Hyphydrini including Pachydrus was polyphyletic, formally transferred Pachydrus and Heterhydrus to Pachydrini and proposed several adult apomorphies for the tribe.However, Miller (2001) based on a broad cladistic analysis of adult characters, found that Hyphydrini (including Pachydrus and Heterhydrus) is monophyletic and is supported by two apomorphies: the obliteration of the metacoxal lobes and metatarsal claws of unequal length.Miller (2001) criticised Biström's et al. (1997) analysis, rejected the use of Pachydrini and transferred Pachydrus and Heterhydrus back to Hyphydrini, a synonymy recognised by Nilsson (2001) and supported by Miller et al. (2006).On the other hand, studies based on larval morphology (Alarie et al., 1997;Alarie & Challet, 2006a, b) suggest a monophyletic origin of the tribe Hyphydrini including Pachydrus, and a position of Pachydrus (Heterhydrus not considered) as the sister group of the remaining Hyphydrini.However, Miller (2001) found that Pachydrus and Heterhydrus are more closely related to Desmopachria.All this reflects the differences in opinion regarding the systematic position of Pachydrus within the Hydroporinae and indicates a need for more studies, including of character sets not so far explored for the genus, such as primary larval chaetotaxy.
The poor knowledge of the larval morphology of Pachydrus and the lack of phylogenetic hypotheses about the relationships of this genus based on larval characters, make the discovery of the larvae of P. obesus of great interest.The current study had the following goals: (1) description and illustration, for the first time, of the first two larval instars of a species of Pachydrus (third instar is redescribed), including detailed morphometric and chaetotaxic analyses of selected structures; and (2) a cladistic study of the phylogenetic relationships of Pachydrus within the Hydroporinae based on larval characters.

Source of material
Three specimens of instar I, two of instar II and five of instar III of P. obesus were used for the descriptions.Larvae were collected in association with adults at the following locality: Argentina, Corrientes Province, Mburucuyá National Park, 6.-15.xi.1997 and15.i.2008, large permanent pond with irregular margins, clear water, muddy bottom with organic debris and abundant emergent and floating vegetation (Salvinia sp., Eichhornia sp.).The identification of the larvae is clear as P. obesus was the only Pachydrus species found as adults at that locality.

Methods
Specimens were cleared in lactic acid, dissected and mounted on glass slides in polyvinyl-lacto-glycerol.Observation (at magnifications up to 1000×) and drawings were made using an Olympus CX31 compound microscope equipped with a camera lucida.Drawings were scanned and digitally edited.The material is held in the larval collection of M.C.Michat (Laboratory of Entomology, Buenos Aires University, Argentina).

Morphometric analysis
We employed, with minimal modifications and additions, the terms used in previous papers dealing with the larval morphology of Hydroporinae (Alarie & Challet, 2006a, b;Alarie & Michat, 2007b;Michat et al., 2007).Paired structures of each individual were considered independently.The following measurements were taken (with abbreviations shown in parentheses).Total length (excluding urogomphi) (TL); maximum width (MW); head length (HL) (total head length including the frontoclypeus, measured medially along the epicranial stem); maximum head width (HW); length of frontoclypeus (FRL) (from apex of nasale to posterior margin of ecdysial suture); occipital foramen width (OCW) (maximum width measured along dorsal margin of occipital foramen); coronal line length (COL); length of mandible (MNL) (measured from laterobasal angle to apex); width of mandible (MNW) (maximum width measured at base).Lengths of antenna (A), maxillary (MP) and labial (LP) palpi were obtained by adding the lengths of the individual segments; each segment is denoted by the corresponding letter(s) followed by a number (e.g., A1, first antennomere).A3' is used as an abbreviation for the apical lateroventral process of the third antennomere.Length of leg (L), including the longest claw (CL), was obtained by adding the lengths of the individual segments; each leg is denoted by the letter L followed by a number (e.g., L1, prothoracic leg).The length of trochanter includes only the proximal portion, the length of distal portion is included in the femoral length.The legs of the larvae studied were considered as being composed of six segments following Lawrence (1991).Dorsal length of last abdominal segment (LAS) (measured along midline from anterior to posterior margin).Length of urogomphus (U) was derived by adding the lengths of the individual segments; each segment is denoted by the letter U followed by a number (e.g., U1, first urogomphomere).These measurements were used to calculate several ratios that characterise body shape.

Chaetotaxic analysis
Primary (present in first-instar larva) and secondary (added in later instars) setae and so-called pores were distinguished on the cephalic capsule, head appendages, legs, last abdominal segment and urogomphus.Sensilla were coded by two capital letters, in most cases corresponding to the first two letters of the name of the structure on which they are located, and a number (setae) or a lower case letter (pores).The following abbreviations were used: AB -abdominal segment VIII; AN -antenna; CO -coxa; FE -femur; FR -frontoclypeus; LA -labium; MN -mandible; MX -maxilla; PA -parietal; PT -pretarsus; TAtarsus; TI -tibia; TR -trochanter; UR -urogomphus.Setae and pores present in first-instar larva of P. obesus were labelled by comparison with the ground-plan of chaetotaxy of the subfamily Hydroporinae (Alarie & Harper, 1990;Alarie et al., 1990a;Alarie, 1991a;Alarie & Michat, 2007a).Homologies were recognised using the criterion of similarity of position (Wiley, 1981).Setae located at the apices of the maxillary and labial palpi were extremely difficult to distinguish due to their position and small size.Accordingly, they are not well represented in the drawings.

Cladistic analysis
For the study of the phylogenetic relationships of the genus Pachydrus within the subfamily Hydroporinae, P. obesus and 27 other species included in eight of the nine hydroporine tribes were analysed using the parsimony program TNT (Goloboff et al., 2003).The tribe Carabhydrini was not included because the larva of Carabhydrus Watts, 1978 is unknown.Members of six of the remaining nine dytiscid subfamilies were included as outgroups.All characters were treated as unordered and equally weighted.A heuristic search was implemented using "tree bisection reconnection" as algorithm, with 200 replicates and saving 100 trees per replication (previously setting "hold 20000").Bremer support values were calculated using the commands "hold 20000", "sub n" and "bsupport", where "n" is the number of extra steps allowed.The process was repeated increasing the length of the suboptimal cladograms by one step, until all Bremer values were obtained (Kitching et al., 1998).Jackknife values were calculated with 2000 replicates and P (removal probability) = 36.

Instar I
Colour.Cephalic capsule with dorsal colour pattern composed of a testaceous to light brown background and several brown maculae centrally on FR and on posterior two-third of PA; head appendages testaceous to light brown except for MN light brown; thoracic and abdominal sclerites I-V with colour pattern composed of   Body.Subcylindrical, narrowing towards abdominal apex (Fig. 1).Measurements and ratios that characterise the body shape are shown in Table 1.

Instar II
As first-instar larva except for the following features.Colour.Distal portion of segment VIII of similar colour as the rest of the segment.Body.Measurements and ratios that characterise the body shape are shown in Table 1.7-8 spiniform secondary setae on each lateral margin of PA; MN with 1 setiform secondary seta on basoexternal margin; prementum with one secondary seta on each lateral margin; thoracic tergites with numerous setiform secondary setae; secondary leg setation detailed in Table 2; FE with a row of natatory setae on anteroventral margin; abdominal sclerites I-VIII with numerous setiform secondary setae; ventral surface of siphon with several setiform secondary setae.

Comparative notes
The third-instar larva of P. obesus described here is very similar morphometrically to that described by Crespo (1996).The seta TR2 and the pore FEa, reported as absent by Crespo (1996) are also absent in our material.However, Crespo (1996) reported the absence of pore URc.This pore is present in our larvae, located terminally on the dorsal surface of the first urogomphomere.Due to the terminal location of URc, and the presence of setae on that region of the urogomphus, this pore is difficult to see and may be easily overlooked.
According to Alarie & Megna (2006) third-instar larvae of P. globosus and P. obniger are very similar morphologically, and no differences were found to separate them.The third instar of P. obesus is also very similar to those of P. globosus and P. obniger, in morphometry as well as chaetotaxy, suggesting a marked structural homogeneity within the genus.The seta identified as LA1 in P. globosus (Alarie et al., 1997) is secondary in P. obesus.Alarie et al. (1997) and Alarie & Megna (2006) report the absence of secondary setae on the ventral surface of the siphon in P. globosus and P. obniger.These setae are present in P. obesus, which may constitute a diagnostic difference.However, we have not seen material of P. globosus and P. obniger, so these setae may have been overlooked.

Character analysis
One hundred and twenty-two characters (100 binary and 22 multistate) were coded for larvae of 28 species of Hydroporinae and six outgroups (Table 3).The characters used and their states are listed in Appendix 1.The analysis of the data matrix (Appendix 2) using TNT resulted in 13 most parsimonious cladograms of length 350.In all trees, Pachydrus was resolved as the sister group of Hydrovatus Motschulsky, 1853.The trees differed largely in outgroup topology and/or in the relative positions of several Hydroporini genera.For this reason, the strict consensus was calculated, in which several taxa collapsed in polytomies (Fig. 20).In the consensus, the clade Hyphydrini minus Pachydrus was recovered as monophyletic and well supported, as part of a polytomy along with Vatellini and the genus Antiporus Sharp, 1882 (Hydroporini), whereas Pachydrus was recovered as sister to Hydrovatus, and more closely related to Canthyporus Zimmermann, 1919, Laccornellus Roughley & Wolfe, 1987and Laccornis Gozis, 1914 than to other Hyphydrini genera.Characters of interest were mapped (using ACCTRAN optimization) in one of the most parsimonious cladograms (Fig. 21).The support obtained was  variable throughout the tree, with some clades well supported and others showing lower values.

DISCUSSION
The results of the cladistic analysis are interesting with regard to the phylogenetic position of the genus Pachydrus within Hydroporinae.The analysis supports a polyphyletic origin of the tribe Hyphydrini as long as Pachydrus is included.In fact, whereas Pachydrus appears among the ancestral groups of Hydroporinae, more closely related to Hydrovatini, the clade formed by the remaining Hyphydrini genera appears as monophyletic and well supported (Fig. 20).Though the identity of the sister group of Hyphydrini remains obscure based on the results of this analysis, the tribe apears to be more closely related to Vatellini and the genus Antiporus.The results obtained here are in agreement with previous studies that suggest that Pachydrus is a strange element within the Hyphydrini and may be improperly placed in that tribe (Young, 1980;Biström et al., 1997).
Larval morphology of members of Hyphydrini is in need of further study.The larvae of most genera within the tribe (e.g., Agnoshydrus Biström, Nilsson & Wewalka, 1997, Allopachria Zimmermann, 1924, Coelhydrus Sharp, 1882, Darwinhydrus Sharp, 1882, Dimitshydrus Uéno, 1996, Hovahydrus Biström, 1982, Hydropeplus Sharp, 1882, Hyphovatus Wewalka & Biström, 1994) are still unknown, and few species of other, speciose genera (e.g., Desmopachria, Hyphydrus Illiger, 1802, Microdytes J. Balfour-Browne, 1946) are known in detail.In particular, detailed studies (including chaetotaxy) of the larvae of Heterhydrus would be of great interest.With the very deficient knowledge of the larval morphology of Heterhydrus that we have at present (Bertrand, 1972) little can be said about whether larval characters support a close relationship of this genus with Pachydrus or the placement of Heterhydrus within Hyphydrini.

Fig. 20 .
Fig. 20.Strict consensus cladogram of 34 terminal taxa of Dytiscidae.Bremer support values are indicated above branches; jackknife values above 50 are indicated below branches.

Fig. 21 .
Fig. 21.One of the most parsimonious cladograms of 34 terminal taxa of Dytiscidae, with character changes mapped for the clades of Hydroporinae.Solid rectangles indicate unique character state transformations; open rectangles indicate homoplastic character state transformations.

TABLE 3 .
Taxa coded for parsimony analysis.