larvae of Megadytes (Coleoptera: Dytiscidae: Dytiscinae): The subgenera Trifurcitus and Megadytes s. str., ground plan of chaetotaxy of the genus and phylogenetic analysis

. The three larval instars of Megadytes ( M. ) carcharias Griffini and M. ( Trifurcitus ) fallax (Aubé) are described and illustrated in detail for the first time, with an emphasis on morphometry and chaetotaxy of the cephalic capsule, head appendages, legs, last abdominal segment and urogomphi. The ground plan of chaetotaxy of the genus Megadytes Sharp is described and illustrated based on three of the four recognised subgenera. First-instar larvae of Megadytes are characterised by the presence of a large number of additional sensilla on almost every part of the body. Primary chaetotaxy of the subgenera ( Bifurcitus Brinck based on third instar) is very similar, with few differences including (1) shape of the setae on the anterior margin of the frontoclypeus; (2) presence or absence of a ring of multi-branched setae on distal third of mandible; and (3) number of setae on the urogomphus. A cladistic analysis of Dytiscidae, based on 169 larval characters and 34 taxa, indicates that: (1) Trifurcitus Brinck deserves generic status; (2) Cybistrini are not closely related to Hydroporinae; (3) the absence of a galea in Cybistrini is a secondary loss independent of that in Hydroporinae; (4) Cybistrini are well supported by many characters (including several aspects of first-instar chaetotaxy).


INTRODUCTION
A previous study (Michat, 2006) of the larval morphology of the diving-beetle genus Megadytes Sharp (Dytiscinae: Cybistrini) tested the hypothesis of a monophyletic origin of this genus and provided a key for the identification of the subgenera.At that time, however, the first instar of only one species [M.(Paramegadytes) glaucus (Brullé)] was available for study, which prevented a broad analysis of the primary chaetotaxy.In particular, any attempt to establish a ground plan for the chaetotaxy of the genus was deemed to be premature.Members of the tribe Cybistrini are particularly interesting from a chaetotaxic perspective because their larvae have many additional sensilla (Nilsson, 1988;Michat, 2006) which, to a certain degree, obscure the pattern of ancestral sensilla that is commonly observed in members of other dytiscid subfamilies and tribes of Dytiscinae.As demonstrated recently, primary chaetotaxy is an important source of characters both for phylogenetic and diagnostic purposes (Alarie et al., 2007(Alarie et al., , 2009;;Michat, 2006Michat, , 2008;;Michat & Alarie, 2009;Michat & Archangelsky, 2009;Michat & Torres, 2006, 2008, 2009;Michat et al., 2007).The development of a system of nomenclature for the primary sensilla of first-instar larvae of Dytiscidae resulted in great progress in the exploration of an extensive set of characters that is phylogenetically and diagnostically very useful.
In this contribution the first-instar larvae of species belonging to other two subgenera (Trifurcitus Brinck and Megadytes s. str.) are described, which together with the previously described first instar of Paramegadytes Trémouilles & Bachmann (Michat, 2006) increase to three the number of subgenera for which the first-instar larvae are known.A review of previous knowledge on the larvae of Megadytes can be found in Michat (2006).Also, descriptions of the mature larvae of M. giganteus (Laporte) and M. latus (Fabricius) were presented by Ferreira Jr. (2000) and Ferreira Jr. et al. (2006), respectively.Based on the detailed generic description of the larvae of Megadytes presented in Michat (2006), I present here diagnosis, descriptions and illustrations of the three larval instars of M. (M.) carcharias Griffini and M. (T.) fallax (Aubé).Also, the ground plan of chaetotaxy of the genus is described based on three of the four subgenera, and interesting features are compared and discussed.
The finding of the first-instar larvae of members of Trifurcitus and Megadytes s. str. is also interesting because it allows a reanalysis of previous results (Michat, 2006), which were based only on the mature larva, but now can also include characters of first and second instars.Though the monophyly of Dytiscinae, including Cybistrini, is well supported by both adult (Miller, 2000(Miller, , 2001(Miller, , 2003;;Miller et al., 2007) and larval (Michat, 2006) characters, recent studies based on DNA sequence data controversially indicate a position of Cybistrini outside the Dytiscinae, more closely related to Hydroporinae and Laccophilinae (Ribera et al., 2008; see also Miller, 2003).On the other hand, previous results (Michat, 2006) indicate that Megadytes is not monophyletic, with members of the subgenus Trifurcitus placed outside this genus and characterised by a subtriangular head and evenly curved mandibles, that lack a ring of multi-branched setae on the distal third, which contradicts the results of a previous study based on adult and molecular data (Miller et al., 2007).In the light of this evidence, the primary objectives of the phylogenetic analysis presented here are: (1) to re-evaluate, using a larger larval data set, if the hypotheses of monophyly for the Cybistrini but not for Megadytes are corroborated, and (2) to test, based on a broader sampling of larval characters and taxa, the phylogenetic relationships of the Cybistrini within Dytiscidae.

MATERIAL AND METHODS
Specimens were cleared in lactic acid, dissected and mounted on glass slides in polyvinyl-lacto-glycerol.Observations (at magnifications up to 1,000×) 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).

Phylogenetic analysis
The phylogenetic relationships of members of the tribe Cybistrini with other dytiscid groups were analyzed cladistically using the program TNT (Goloboff et al., 2008) and the character set provided by the larval morphology and chaetotaxy.A broad range of taxa was included, comprising members of nine of the 10 recognised dytiscid subfamilies (Nilsson, 2001) (Table 1).The subfamily Hydrodytinae was not included because their larvae are unknown.Data for most species were scored directly from observations on specimens and data for the remaining species from the literature.In order to avoid the problem of comparing different semaphoronts (see Meier & Lim, 2009), in the codification of each character the same instars were compared, based on the fact that Dytiscidae larvae pass through three instars distinguishable by the presence of egg bursters (instar I), the presence of conspicuous spiracular openings on mesothorax and abdominal segments I-VII (instar III), or the absence of both characters (instar II).The tree was rooted in Aspidytidae (another family of aquatic Adephaga) to allow the dytiscid taxa to vary freely, thus testing the relationships of Cybistrini with the other dytiscid groups.All characters were treated as unordered and equally weighted.Multistate characters were treated as non-additive.A heuristic search was implemented using "tree bisection reconnection" as an algorithm, with 200 replicates and saving 100 trees per replication (previously setting "hold 20,000").Bremer support values were calculated using the commands "hold 20,000", "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 using 2,000 replicates and P (removal probability) = 36.(2006) provides a detailed description of the larvae of the genus Megadytes.For this reason, most characters commonly recognised in all species of the genus are not included in the present paper and only diagnostic or interesting features are mentioned.

Megadytes (Megadytes) carcharias Griffini
Material.Eight specimens of instar I, three of instar II and three of instar III were used for the descriptions.The larvae were collected in association with adults at the following locality: Argentina, Chaco Province, El Cachapé refuge, Qia creek, 3.-4.xii.2008.Larvae and adults were collected among the marginal vegetation in shallow areas of the creek.The identification of the larvae is based on the following: (1) there were no other species of the subgenus Megadytes s. str. in the area where the larvae were collected [except M. laevigatus (Olivier), see below]; (2) tens of adults were taken from the creek along with the larvae, all of them belonging to M. carcharias except for one specimen of M. laevigatus (a much smaller species that can be ruled out because of its size); (3) the range of head width of the larvae described is smaller than that of M. marginithorax (Perty) (Ferreira Jr., 1995), the only other Megadytes s. str.whose distribution includes the collecting site.(Fig. 1); distal part of antennomere 1 longer than basal part (instar III) (Fig. 1); measurements and ratios that characterise the body shape are shown in Table 2.

Description of colour
Instar I. Frontoclypeus pale yellow except for some brown maculae on central portion, medial to egg bursters; parietal pale yellow ventrally, light brown dorsally except for pale yellow areas around stemmata and posterior to occipital suture, light brown area with several small brown maculae; antenna light brown except for basal half of antennomere 3 pale yellow; mandible pale yellow except for distal portion light brown; maxilla light brown except for basal half of palpomere 3 pale yellow; labium pale yellow; thoracic tergites pale yellow, protergite with one brown macula on each side of anterior half, mesoand metatergite with four brown maculae on anterior margin and one on each lateral margin; abdominal tergites I-VI pale yellow with four brown maculae on anterior margin and one light brown macula on posterolateral angle; abdominal tergites VII-VIII pale yellow except for distal third of segment VIII light brown and two brown maculae on anterior margin; membranous parts pale; legs pale yellow except trochanter, tibia and tarsus light brown; urogomphus light brown.
Instar II.As instar I except: small brown maculae on cephalic capsule more numerous; head appendages pale yellow to light brown except for distal portion of mandible and apices of antenna and maxilla darker; thoracic and abdominal tergites with numerous small brown maculae on the surface; legs evenly pale yellow.
Instar III (Fig. 1).As instar II except: darker in general; small brown maculae on cephalic capsule, thoracic and abdominal tergites more numerous; legs evenly light brown; urogomphus brown.

Megadytes (Trifurcitus) fallax (Aubé)
Material.Ten specimens of instar I, three of instar II and three of instar III were used for the descriptions.The larvae were collected in association with adults at the following locality: Argentina, Corrientes Province, Mburucuyá National Park, 5.-9.xi.1997, large permanent ponds of different sizes, with clear water, muddy bottom with organic debris and abundance of emergent and floating vegetation.The identification of the larvae is based on the following: (1) there are only two species of Trifurcitus in Argentina, M. robustus (Aubé) and M. fallax, the first is considerably larger than the second; therefore, the larvae of both species can be easily separated by size (for example head width); (2) the mature larva of M. robustus is described by Michat (2006) and differs from M. fallax in having short and stout setae on the anterior margin of frontoclypeus (in M. fallax the setae are setiform and multi-branched); (3) M. robustus was not collected at the locality where the larvae were taken.2); abdominal sclerites I and VII without an anterotransverse carina (instars II-III); meso-and metatergite without an anterotransverse carina (instar I); parts of antennomere 1 subequal in length (instar III) (Fig. 2); measurements and ratios that characterise the body shape are shown in

Description of colour
Instar I. Cephalic capsule evenly pale brown; head appendages pale brown except antennomere 4 and apices of antennomere 3, maxillary palpomere 3, and labial palpomere 2 brown; thoracic tergites pale yellow, protergite with one brown macula on each side of anterior half, meso-and metatergite with four brown maculae on anterior margin, one on each lateral margin, and one (diffuse) on posterior margin; abdominal tergites I-VI pale yellow with four brown maculae on anterior margin and one pale brown macula on posterolateral angle; abdominal tergite VII pale yellow with two brown maculae on anterior margin; abdominal segment VIII pale brown, distal half somewhat darker; membranous parts pale; legs evenly pale yellow; urogomphus brown.
Instar II.As instar I except: cephalic capsule pale brown with numerous small dark brown maculae on the frontoclypeus (covering the whole surface except the anterolateral corners) and the parietals (covering the dorsal surface and the laterals of ventral surface); thoracic tergites pale brown with numerous dark brown maculae all over the surface; abdominal terga I-VI variegated, with pale yellow and pale brown longitudinal bands; legs evenly pale brown.
Instar III (Fig. 2).As instar II except: apex of labial palpomere 2 of same colour as the rest of the segment; abdominal segment VII pale brown.

Ground plan of chaetotaxy of Megadytes Sharp
Hypotheses of homology for the setae and pores present in the genus Megadytes are presented based on the examination of members of three of the four subgenera currently recognized: Trifurcitus and Megadytes s. str.(described in the present paper), and Paramegadytes (described in Michat, 2006).On this basis, the ground plan of primary chaetotaxy for the genus is established.Setae and pores present in the larvae of Megadytes were named by comparison with those of members of other tribes of the subfamily Dytiscinae for which the primary chaetotaxy is described in detail (Michat & Torres, 2005, 2006;Michat & Alarie, 2009).Homologies were established using the criterion of similarity of position (Wiley, 1981).As stated above, larvae of Megadytes characteristically have numerous additional sensilla that obscure the establishment of homologies with the ancestral systems in other tribes of Dytiscinae.For this reason, an additional criterion for homology was implemented based on the fact that larvae of Megadytes are characterised by the presence of multi-branched setae (setae that are split into two or more branches at a certain distance from the base), and the distribution of these setae on some structures (e.g.cephalic capsule, last abdominal segment) is very similar to the distribution of the ancestral setae in other Dytiscinae.The presence of a multi-branched seta was considered as an argument for homology when more than one seta present in Megadytes is potentially homologous with a given seta in other dytiscine genera.The ancestral chaetotaxy pattern thus established for Megadytes is in good agreement with that observed in other Dytiscinae larvae (Michat & Torres, 2005, 2006;Michat & Alarie, 2009).For the urogomphus, homology of the setae and pores was not attempted because of the strongly modified shape of this structure in Megadytes.Primary chaetotaxy proves to be remarkably similar among the subgenera, with few differences in ancestral sensilla and a large number of additional sensilla on almost all the structures examined.In the following, a summary of the primary setae and pores present in Megadytes larvae is presented, and differences among the subgenera are remarked upon when appropriate.Maxilla ).Setae MX1-3 multi-branched.Setae MX8-9 and pores MXd and MXh inserted on the stipes (seta MX8 in Megadytes s. str.was not found nor was it possible to determine if it is really absent).Seta MX10 obscured by the presence of additional setae.Stipes, palpifer and palpomere 1 with several additional setae, 2-3 on palpomere 1 multi-branched.Palpomere 2 with two additional multi-branched setae apically.
Urogomphus .In Paramegadytes and Trifurcitus there are a maximum of seven setae; in some specimens only six setae could be observed, but it was not possible to determine if the seventh seta is really absent.In Megadytes s. str.there are consistently six setae, this number appears to be the maximum for this subgenus.A single pore was present in the three subgenera, but in some specimens it was not possible to find it.The extremely small size of the urogomphi in Megadytes makes the observation of the sensilla very difficult.

Character analysis
A total of 169 characters derived from the larval morphology and chaetotaxy were included, of which 125 were coded as binary and 44 as multistate (Appendix 1).The analysis of the data matrix (Appendix 2) using TNT resulted in 34 most parsimonious trees of 483 steps (CI = 0.47; RI = 0.78).The trees differed in the relative positions of the dytiscine genera and in the relative positions of some genera within Cybistrini.For this reason, the strict consensus was calculated, in which several genera collapsed in polytomies (Fig. 31).Hydroporinae, Dytiscinae and Cybistrini came out as the best supported clades in the analysis.However, the relationships within these lineages were in general poorly supported as were the basal relationships of the non-hydroporine clade (Fig. 31).The consensus revealed a basal split into two large clades, one corresponding to the subfamily Hydroporinae and the other including the remaining subfamilies.Within this last clade, Dytiscinae (including Cybistrini) was resolved as monophyletic in agreement with Miller (2000, Figs 13-16.Maxilla of first-instar larva of species of Megadytes.13-14 -M.carcharias: 13 -dorsal aspect; 14 -ventral aspect.15-16 -M.fallax: 15 -dorsal aspect; 16 -ventral aspect.Scale bars = 0.25 mm.2001,2003), Miller et al. (2007) and Michat (2006) but not with Ribera et al. (2008) (see also Miller, 2003).Dytiscinae was placed as sister to Matinae and close to Colymbetinae and Lancetinae.Cybistrini was also recovered as monophyletic in agreement with almost all previous results, but the sister group could not be clearly established and the tribe was resolved as part of a basal polytomy along with Aubehydrini, Dytiscini, Hydaticini, Hyderodini and the clade Aciliini + Eretini.No evidence was found of a close relationship of Cybistrini with Hydroporinae and Laccophilinae, as suggested recently based on molecular data (Ribera et al., 2008).On the other hand, the genus Megadytes was not recovered as monophyletic, with several species more related to a species of Cybister Curtis than to the subgenus Trifurcitus of Megadytes.This result agrees with previous results of Michat (2006) and Ribera et al. (2002Ribera et al. ( , 2008) ) but not those of Miller et al. (2007), who found Megadytes and Cybister to be both monophyletic and sisters.

DISCUSSION
Larvae of Adephaga are generally characterised by the presence of a galea, with two primary setae (MX8, MX9) and two primary pores (MXd, MXh), which are deemed to be part of the ground-plan condition of the suborder (Alarie et al., 2004;Alarie & Bilton, 2005).The galea is generally lacking within the tribe Cybistrini (Michat, 2006).However, the presence of a minute lobe-like structure in the place where the galea is commonly found, as well as the presence of the galeal sensilla in that region of the stipes in the three subgenera of Megadytes studied is of the utmost interest as it would suggest that cybistrines lost the galea secondarily.The other group of diving beetles in which most larvae lack a galea is Hydroporinae (Alarie & Michat, 2007;Michat et al., 2007).In a recent phylogenetic analysis of the Dytiscidae based on molecular data, Ribera et al. (2008) found a close relationship of Cybistrini with Hydroporinae and Laccophilinae (larvae of this last subfamily bear a well-developed galea).However, the results presented here suggest that cybistrines are not closely related to Hydroporinae and Laccophilinae, but are well placed within Dytiscinae in agreement with the previous results of Miller (2000,2001,2003) and Miller et al. (2007), based on adult morphological and molecular data.Though the larvae of both Hydroporinae and Cybistrini lack a galea, the present analysis shows that, in the context of a broad cladistic analysis including numerous characters, a close relation- ship between both groups seems unlikely.On the other hand, it was recently shown that larvae of some presumably ancestral hydroporines (Methlini, Laccornini, Hydrovatini, Canthyporus Zimmermann, Laccornellus Roughley & Wolfe) have a small galea (Alarie & Michat, 2007;Michat et al., 2007).The absence of a galea in Cybistrini thus most likely represents an independent secondary loss.
These differences may indicate that Trifurcitus deserves generic status, as suggested previously (Crespo, 1987;Michat, 2006).However, this is not formally proposed at this stage because: (1) recent studies of morphological and molecular data (Miller et al., 2007) support the monophyly of Megadytes including Trifurcitus; (2) low resolution and support within Cybistrini obtained in the present analysis does not allow the proper placement of Trifurcitus; (3) larvae of many species of Cybistrini are unknown and so a more comprehensive study is not possible.Future analyses combining adult and larval characters and molecular data would provide stronger evidence on the phylogenetic position of Trifurcitus.
Larval primary chaetotaxy of members of Trifurcitus is very similar to that of the subgenera Paramegadytes and Megadytes s. str.(see "Ground plan of chaetotaxy" above).This similarity apparently comprises also mem- bers of the genus Cybister, at least when comparing the leg chaetotaxy of three Cybister species (Nilsson, 1988) with that of the Megadytes species described here (chaetotaxy of the remaining parts of the body of Cybister is not known in detail).The resemblance in larval chaetotaxy of members of Cybistrini and the fact that the primary chaetotaxy of this tribe is highly characteristic (i.e. with a large number of additional sensilla on almost every part of the body) make them a distinctive group of diving beetles.In addition to the characters mentioned in Michat (2006), several first-instar larval characters support this tribe in the present analysis including: egg bursters rounded (character 13.1), setae MX8 and MX9 and pore MXd inserted on the stipes (characters 63.2, 64.2, 66.1), presence of natatory setae on dorsal surface of tarsus (character 126.1), presence of a single pore on the urogomphus (character 158.2), and several characters related to the presence of additional (characters 39.1, 49.1, 69.1, 86.1, 92.1, 96.1) or multibranched (characters 16.2, 60.1, 76.1, 100.1, 143.2, 144.1) setae on different parts of the body.However, as first instar larvae of several of the Cybistrini species scored in the data matrix are unknown, these characters will remain as potential apomorphies until more larvae are discovered and described.
The few differences in primary chaetotaxy among the subgenera of Megadytes are summarised as follows: (1) setae on the anterior margin of frontoclypeus are simple and spiniform in Paramegadytes, Megadytes s. str.and Bifurcitus Brinck (the latter based on instar III), whereas in Trifurcitus the setae are either setiform and multibranched (M.fallax) or short and stout (M.robustus, based on instar III) (see also Michat, 2006) In a previous paper (Michat, 2006) the presence of a median process with bifid apex on the prementum was given as a character separating the subgenus Trifurcitus from the remaining subgenera of Megadytes.However, it was mentioned that in some specimens of M. fallax the median process is not clearly bifid but rounded apically.The examination of additional material of this species indicates that the median process is rounded apically and therefore this character is not diagnostic for Trifurcitus.The presence of an apically bilobed process in M. robustus also needs corroboration as only one specimen of this species was studied (Michat, 2006).Accordingly, in the new data matrix character 73 (median process on prementum) is scored as state 1 (single) for both species of Trifurcitus.
Although Megadytes includes the largest known diving beetles their larvae have one of the smallest urogomphi in the entire family.In fact, urogomphal sensilla are difficult to evaluate due to the small size of this structure.In a previous paper (Michat, 2006) the presence of seven setae and one pore are reported on the urogomphus of M. glaucus.In M. fallax there are also seven setae and one pore, but in M. carcharias only six setae (and one pore) were found (character 157 in the present analysis).It is worth mentioning that the numbers of urogomphal sensilla reported for the three species are the maximum number observed in the several specimens analyzed.It is possible that some specimens actually bear fewer sensilla than those mentioned above.For example, the pore was not found in some specimens, and in others the number of setae is apparently reduced in one.However, these conclusions are not well founded because it was difficult to evaluate this character using a light microscope.The problem could be solved using scanning electron microscopy, but no additional specimens were available for such a study.
; (2) the mandible of Paramegadytes, Megadytes s. str.and Bifurcitus bears a ring of long multi-branched setae on distal third, which is absent in Trifurcitus; (3) Paramegadytes has four setae on the anterodorsal surface of prementum, whereas Megadytes s. str.has three and Trifurcitus has two to four.

Fig. 31 .
Fig. 31.Strict consensus cladogram, with Bremer support values indicated above branches and jackknife values above 50 indicated below branches.

TABLE 1 .
Taxa coded for cladistic analysis.

TABLE 2 .
Measurements and ratios for the three larval instars of two species of Megadytes.Three specimens of each instar were measured.