G. William Wolfe.
A phylogenetic investigation of Hydrovatus methlini and other plesiotypic hydropines (Coleoptera: Dytiscidae).
Psyche 95:327-344, 1988.

Full text (searchable PDF, 2384K)
Durable link: http://psyche.entclub.org/95/95-327.html

The following unprocessed text is extracted from the PDF file, and is likely to be both incomplete and full of errors. Please consult the PDF file for the complete article.

A PHYLOGENETIC INVESTIGATION OF HYDROVATUS, METHLINI AND OTHER PLESIOTYPIC HYDROPORINES (COLEOPTERA: DYTISCIDAE)*
Department of Entomology and Economic Zoology, Cook College, New Jersey Agricultural Experiment Station, Rutgers University, New Brunswick, N.J. 08903 Phylogenetic relationships of the Dytiscidae are receiving in- creased attention (e.g. Burmeister 1976, 1980; Brancucci and Ruh- nau 1985, Dettner 1985). This paper is the fifth in a series on primitive hydroporine genera (Wolfe and Matta 1981; Wolfe 1985; Roughley and Wolfe, in press; and Wolfe and Roughley, in press). The first purpose of this paper is to investigate new discoveries concerning the remarkable abdominal structure of methlines and members of Hydrovatus Motschulsky. The peculiar modifications of terminal abdominal terga of members of Hydrovatus and Meth- lini provide good evidence for monophyly of these taxa. The second purpose is to revise a previous hypothesis (Wolfe 1985) of relation- ships among plesiotypic hydroporines in light of: 1) the new evi- dence regarding hydrovatine and methline monophyly, 2) more information on Laccornellus Roughley and Wolfe, 3) different interpretations of some character systems, and 4) computer gener- ated phylogenies.
MATERIALS AND METHODS
Analyzed species are listed in Table 1. Authors of species and genera are listed the first time a name is used in text only if the name is not listed in Table 1. Table 2 lists characters used in this analysis but see Wolfe (1985) for complete details and illustrations for char- acters 1-18.
Dissecting, illustrating, and scanning electron microscope (SEM) techniques are presented in Wolfe (1985). In order to understand the *Manuscript received by the editor April 20, 1987. 327




================================================================================

328 Psyche [vo~. 95
derivation of the abdominal structure of specimens of Hydrovatini and Methlini, comparisons were made with members of other hydroporine and nonhydroporine genera listed in Table 1. Relationships between taxa were determined by phylogenetic techniques (essentially Hennigian) that are outlined in Wiley (1980), and Nelson and Platnick (1981). However, phylogenies in the analy- sis herein were generated and/or analyzed in part with computer programs developed by Dr. D. Swofford (PAUP, Phylogenetic Analysis Using Parsimony, version 2.4).
Abdominal structure in members of Hydrovatini and Methlini Three basic abdominal morphotypes (1-3) are recognized. These three types form a spectrum from the unmodified posteriorly rounded, lightly sclerotized terminal tergum found on members of L. dijjformis (Figs. 1A-B) (type l), to intermediate modifications as found in specimens of H. pustulatus Melsheimer (type 2) (Figs. 1C-D), to the most derived condition in methlines (type 3) (Figs. 2A-D).
Morphotype 1. (Fig. 1A).
In all examined specimens of Can-
thyporus Zimmermann, Deronectes Sharp, Hydroporus Clairville, Laccornis Gozis, Oreodytes Seidlitz and all nonhydroporines, the posterior edge of the eighth tergum is broadly and evenly convex in dorsal view. The eighth tergite is folded ventrally inward as a ventral flap or fold; this folded portion is hereafter referred to as the ventral fold (Fig. 1 B). The ventral fold extends anteriorly for about 25 per cent of the length of the last segment. The outwardly visible poste- rior edge of tergum-8 then, is actually the point at which the tergum folds underneath. This character state is clearly evident in many of the abdominal/ genitalic illustrations in Burmeister (1 976, 1980). Morphotype 2. (Figs. 1 C-D).
On males and females of Hydro-
vatus, only the apical tergum is distinctly modified; it is narrow, acutely pointed, and it is differentiated into a dorsal and ventral lamina (Fig. 1C). Neither lamina is distinctly sclerotized, but the ventral lamina is more membranous than the dorsal lamina (Fig. ID). Dissections were difficult but it appeared that the dorsoapical portion of the ventral lamina possessed a slight inward fold and no ventral fold was evident on the dorsal lamina.



================================================================================

19881 Wolfe - Hydrovatus methlini 329
Table 1.
List of species of Dytiscidae examined only for abdominal structure. Hydroporinae
Canthyporus hottentotus Gemminger and Harold Celina grossula J. LeConte
Celina hubbelli Young
Celina imitatrix Young
Deronectes striatellus J. LeConte
Hydroporus aulicus J. LeConte
Hydrovatus pustulatus Melsheimer
Hydrovatus sp. (from Sri Lanka)
Laccornis conoideus J. LeConte
Laccornis deltoides Fall
Laccornis etnieri Wolfe and Spangler
Laccornis lugubris (Aubk)
Laccornis copelatoides Sharp
Methles cribatellus Fairmaire
Oreodytes quadrimaculatus (Horn)
Queda compressa Sharp
Colymbetinae
Agabus spinipes Sharp
Colymbetes sculptilis Harris
Lancetes sp.
Dytiscinae
Dytiscus fasciventris Say
The terminal tergum of specimens of Q. compressa Sharp is most similar to morphotype 1.
Morphotype 3. (Figs. 2A-D). In this morphotype, the seventh and eighth terga are modified in both males and females. Tergum 8 is extremely acutely pointed posteriorly and also consists of a dorsal and ventral lamina. The ventral lamina (Fig. 2B) is a thin, flexible, triangular structure that is similar to that of Hydrovatus; the lateral edges are fringed with setae and curved dorsally thus forming a broad channel into which the dorsal lamina rests. The dorsal lamina (Fig. 2A) is quite sclerotized, rigid, and overall rather wishbone-shaped. Posteriorly, the structure is somewhat trifid with the medial portion extremely prolonged, laterally com- pressed, and apically acute; the apicolateral portion is densely setose. Anteriorly, the dorsal lamina extends as a pair of diverging thin apodemes. Each apodeme expands anteriorly into a club shaped apex and extends anteriorly underneath the seventh tergum all the way to the posterior edge of the sixth tergum (Fig. 2D).



================================================================================

330 Psyche [VOI. 95
Figure 1. Terminal abdominal structure in Laccorm and Hydrovusus, A) Dorsal view of terga 6-8 of Laccornis d'/formis (40X) B) Dissected abdominal apex of L. difformis (40X). Tergum 8 is folded back (all the way to the left) so that ventral surface is exposed revealing ventral fold (indicated by arrow); tergum 8 is not bilaminar. C) Acutely pointed eighth tergum of Hydrova~us pwtuiaius (BOX). D) Dissected abdominal apex of H. pwlulalus. Tergum 8 is folded so that dorsal lamina (left arrow) is separated from ventral lamina (right arrow) (60x1. The anterior edge of tergum 7 (Fig. 2C) is expanded anteriorly as a broad, bisinuate flange with a short anterolaterdiy extended apo- deme at each antwolateral corner. Apodemes of tergum 7 extend anteriorly for 75 per cent of the length of tergum 6. The function of the modified methline abdominal structures has not been observed. Ovipositional function is ruled out because mod- ifications are identical in males and females. Perhaps, the acutely pointed and sclerotized apex is a device for puncturing plant tissue to obtain trapped air. It is interesting that a similar behavior has



================================================================================

19881 Wove - Hydrovatus methhi 33 1
Figure 2.
Abdominal structure of C. hubelli (60X) A) Acute, pointed, spear- shaped dorsal lamina of tergite 8.
B) Ventral lamina. C) Ventral view of terga 6 and 7 showing apodemes on tergum 7. D) Composite illustration of terga 7 and 8. been postulated for larval methlines because of the peculiar poste- riorly extended lateral tracheal trunks located at the abdominal apex (SpangJer 1973).
Wolfe (1 985) presented considerable evidence that Hydroporinae is monophyletic. Among taxa examined (Table 3), Laccornis, Meth- Iini, Hydrovathi, Canthyporus, and Lioporeus were considered the



================================================================================

332 Psyche [vo~. 95
most plesiotypic lineages. It also was suggested that: 1) Hydrovatini and Methlini were sister taxa, 2) Bidessini, Oreodytes, and Hygrotus Stephens were relatively more plesiotypic than Hydroporus, Dero- nectes s.l., Vatellini, and Graptodytes Seidlitz, and 3) L. lugubris and L. copelatoides were related more closely to Canthyporus than to Laccornis.
Since that information was published, several studies have increased significantly our understanding of plesiotypic groups. First, information provided herein more definitely supports mon- ophyly of Hydrovatus and Methlini. Second, Roughley and Wolfe (1 98 7) definitively demonstrated generic status of L. copelatoides and L. lugubris and assigned those species to a new genus, Laccor- nellus; evidence supporting a close relationship between Laccornel- lus and Canthyporus was reviewed. Third, Wolfe and Roughley (in press) completely revised Laccornis and described a new tribe, Lac- cornini, for the genus.
In light of this new information, the phylogeny of plesiotypic hydroporines proposed by Wolfe (1985) is re-evaluated below. The analysis herein is facilitated by phylogenetic computer programs (PAUP) not previously available to me. I have found that use of these programs supplements interpretation of hypotheses by: 1) more accurately and repeatedly revealing the number of equally parsimonious trees derivable from a character matrix, 2) permitting rapid calculation of consensus trees so that similarities between equally parsimonious trees can be ascertained, 3) more definitively allowing assessment of assumptions used in tree construction (e.g. character weighting, character ordering, and addition and elimina- tion of taxa), and 4) allowing easier comparison of trees in terms of homoplasy and tree length.
In summary, it must be stressed that computer generated phylo- genies are not intrinsically better than mentally computed trees. However, I think that singular reliance on mental computations can be biased too easily by preconceived notions/ hypotheses concerning one or two trees that investigators often have in mind before in depth analyses even begin. Compared to purely mentally con- structed hypotheses, computerized constructions (and associated kinds of output) reveal in a more definable and consistent way the frailties of a given hypothesis(es). Although various assumptions/ limitations that are explicitly exposed through computer analysis



================================================================================

19881 Wove - Hydrovatus methlini 333
Table 2.
Synopsis of plesiotypic and apotypic characters used in analysis. Unless otherwise indicated characters are from the adult stage. See Wolfe (1985) for com- plete details and illustrations for characters 1 - 18. Character 19 illustrated herein. Character Plesiotypic state Apotypic state -
1. Mandibles ventral medial setae present ventral medial setae absent 2. Labial spines
in multiple rows, spines
in single (at most double) row,
smaller spines larger
3. Larval nasale absent present
1 -without notch
2-with notch
4. Larval galea present
absent (reduced in Methilini)
5. Prosternum not declivous declivous
6. Prosternal pore absent
present
7. Scutellum exposed
concealed
8. Elytral ridge reduced, not carinate present 1 -carinate
2-slightly posteriorly
elevated
3-distinctly posteriorly
elevated
4-ligulate
9. M4 vein contacting oblongulum not contacting oblongulum 10. Base of meta- not and cleft produced and dorsally cleft furca 1 -indistinctly developed
2-distinctly developed
1 1. Proventriculus sclerites of sulci with longi- sclerites of sulci with transverse
tudinal ridge and/ or teeth teeth
12. Protarsomere
13. sublateral row
of spines of
mesotibia
14. Proximity of
metafemur to
metacoxal
process
15. Valvifer
16. Apodeme of
pentamerous
sparser (state 0)
pseudotetramerous
1 -indistinctly pseudo-
tetramerous
2-distinctly pseudo-
tetramerous
states 1 to 3, indicate
increasing density
touching to distinctly separated
scarcely separated
present
absent
genital valves
17. Posterior apex not acuminate
of body
18. Larval short
absent
present
acuminate
long
urogomphi
19. Posterior
not apically acute 1 -tergum-8 only modified; tergite and bilaminar
acute and bilaminar
2-tergum-7 and 8 modified;
tergite-8 sclerotized, trifid
and spear shaped




================================================================================

Psyche [Vol. 95
may require that data be much more conservatively interpreted (see below), it no way implies that the only good character data is that data which produces one (or a few) easily interpreted, most parsi- monious tree(s). Furthermore, many assumptions (e.g., character- weighting, assumed monophyly of subgroups, occurrence of equally parsimonious trees) that require explicit enumeration with a compu- terized approach are used implicitly (sometimes ignored) in men- tally produced hypotheses.
Hydrovatini- Methlini monophyly
Polarizing the character states of members of Hydrovatus and Methlini is rather straight forward. Morphotype-1 occurs in members of Hydroporus, Deronectes, Oreodytes, Canthyporus, Laccornis, and all non-hydroporines (see Table 1). It is logical to postulate a morphocline that proceeds from morphotype-1 , through morphotype-2 and culminating in morphotype-3. Thus the structu- ral modifications of morphotype-2 are a synapotypy unifying Hydrovatus and Methlini and the apotypic modifications associated with morp hotype-3 phylogenetically cluster Methles and Celina. These latter facts help offset the conflict created by the exposed scutellum in members of Celina and concealed condition in Methles. There are potential synapotypies for Queda and Hydrovatus: the prosternal process is extremely broad, the metafurca is reduced in size and wishbone shaped, an elytral humeral carina is present, and the internal elytral ridge is expanded throughout its length in members of both genera. However, I had hoped that the unification of Queda and Hydrovatus in Hydrovatini (Zimmerman 1920) could be supported further through this analysis; unfortunately this is not the case. As stated above, abdominal modifications are plesiotypic in members of Queda, so much so that inclusion in Hydrovatini based on the structure of tergum 8 is not obvious. Furthermore, the meta- femoral apices are distinctly separate from the metacoxal lobes in members of Queda (apotypic) while in Hydrovatus and methlines the metafemora almost attain or do reach the metacoxal lobes (ple- siotypic). In summary, recent studies are making the phylogenetic relationship of Queda more enigmatic rather than more understand- able and this genus requires further study. Phylogenetic re-evaluation of plesiotypic Hydroporines The analysis below is based on a modified version (Table 3 herein) of the character state matrix of Wolfe (1985, Table 2, pp. 136-137).



================================================================================

19881 Wove - Hydrovatus methlini 335
Previously there were tabular errors associated with two characters. Character-2 (arrangement of labial spines) for members of Q. com- pressa should have read 0, not 1. For character-14,1 intended prox- imity of the metafemoral base to the metacoxal lobes to be coded dichotomously; therefore, each 2 in that column should have been 1 and the 1 recorded for specimens of L. lugubris and L. copelatoides should have been 0. These were tabular errors only and were not incorporated into the phylogeny proposed by Wolfe (1985). It was indicated that medial mandibular setae were absent in all hydropo- rines; however, there is a reduced row on specimens of M. mexica- nus Sharp and 0. rivalis (Gyllenhal) that is difficult to see; therefore, character 1 should have read 0, not 1, for these two spe- cies. The discovery of medial mandibular setae in M. mexicanus and 0. rivalis did not alter their phylogenetic placement; these two taxa still are regarded as rather apotypic.
Previously, I was not sure about the status or placement of L. lugubris and L. copelatoides. However, further study (Roughley and Wolfe, 1987) adequately demonstrated that those species formed a distinct unit and they were assigned to a new genus, Lac- cornellus, and Laccornellus is included in the analysis below. Before analysis with PAUP was conducted, groups of identical taxa were identified and each group was represented by one species. With these modifications, computer analysis revealed more than 100 equally parsimonious trees: however, it is interesting to note that a consensus tree of these first 100 trees showed the same basic patterns as previously proposed in Wolfe (1985).
To reduce the number of equally parsimonious trees below 100,I represented several groups considered to be monophyletic by one species. L. triangularis (Fall) was used for Lioporeus, L. kocai (Ganglebauer) for Laccornis, U. lacustris (Say) for Bidessini, and M. cribatellus Fairmaire for the clade that includes Methlini and Hydrovatini. Justification for monophyly of Lioporeus and Lac- cornis is based on information in Wolfe and Matta (1981) and Wolfe and Roughley (in press) respectively. Bidessini is considered monophyletic based on metacoxal process structure. Justification for monophyly of Methlini and Hydrovatini is not conclusive, as long as Queda is included in Hydrovatini; however, until more characters are discovered to clarify the phylogenetic position of the enigmatic Queda, I assume it shares a most recent common ancestor with Hydrovatus and that the distinct gap between metafemora and



================================================================================

Psyche
3
ia
[Vol. 95
Figure 3, above and opposite. Phylogenies discussed in text. A) Consensus tree
produced from 99 equally parsimonious trees after all characters were scaled. Species names are used for taxa actually used in analysis. Numbers 1-7 are node numbers; numbers .250, .833, etc. are branch lengths for each proximate HTU. The indicated character state changes are derived directly from computer analysis and are interpre- ted/evaluated in text. B) and C) More conservative phylogenetic hypotheses. See text for discussion. Generic and tribal names are used in place of species names. metacoxal lobes is secondarily derived in members of Q. compressa. Even with the above specified reduction in species number, more than 100 equally parsimonious trees still are produced. Rather than immediately further decrease the number of species, I next elected to scale all characters; scaling is useful because it equalizes the influ- ence of 2-state and multi-state characters by decreasing weights of character states of multi-state characters so that character states of all characters are on an interval from 0 to 1. For example, a 3-state character would be coded 0-.5-1 instead of 0-1-2 and a 4-state character recoded 0-.333-.666- 1 (see Swofford 1985). With all char- acters scaled, 99 equally parsimonious trees are produced. The consensus tree (Fig. 3A) continues to show the same basic set of relationship proposed in Wolfe (1985).




================================================================================

DERONECTES
DERONECTES
GRAPTODYTES
GRAPTODYTES
HYDROPORUS HYDROPORUS
HYDROPORUS HYDROPORUS
HYDROPORUS HETEROSTERNUTA
HYDROPORUS HE TEROSTERNUTA
HYDROPORUS NEOPOR us
HYDROPORUS NEOPORUS
HYGROTUS
OREODYTES
VATELLINI
BIDESSINI
LIOPORE US
LACCORNELLUS LACCORNELL US
CANTHYPORUS CANTHYPORUS
METHLINI-HYDROVATINI
ME THLINI- HYDRO VA TIN1
LACCORNINI LACCORNINI




================================================================================

338 Psyche [vo~. 95
Table 3.
Character states used in phylogenetic analysis (modified from Wolfe [1985]). See Table 2 for character description. Character No. 1 2 3 4 5 6 7 8 9 10111213141516171819 Colymbetinae (out-group)
Matus ovatus*
Hydroporinae
Methlini
Methles cribatellus
Celina hubbelli
C. imitatrix
C. grossula
Hydrovatini
Hydrovatus pustulatus
Queda compressa
Bidessini
Uvarus lacustris
Bidessonotus
inconspiciuus
Hydroporini sensu latu
Laccornis kocae
L. conoideus
L. etneiri
L. dehoides
Laccornellus
copelatoides
L. lugubris
Canthyporus
hottentotus
Lioporeus pilatei
L. triangularis
Hygrotus acaroides
H. nubilis
Oreodytes
quadrimaculat us
0. rivalis
0. snoqualmie
Graptodytes crux
Deronectes depressus
D. striatellus