Arthropod Cladistics

49 Pages·2000·1.49 MB·English

Cladistics16,155–203 (2000)
doi:10.1006/clad.1999.0125,availableonlineathttp://www.idealibrary.comon
Arthropod Cladistics: Combined Analysis of Histone
H3 and U2 snRNA Sequences and Morphology
Gregory D. Edgecombe, George D. F. Wilson, Donald J. Colgan,
Michael R. Gray, and Gerasimos Cassis
CentreforEvolutionaryResearch,AustralianMuseum,6CollegeStreet,Sydney,NewSouthWales2010,Australia
AcceptedNovember15,1999
Morphological,developmental,ultrastructural,andgene INTRODUCTION
ordercharactersarecataloguedforthesamesetofarthro-
pod terminals as we have scored in a recent study of
histone H3 and U2 snRNA sequences (D. J. Colgan et Despite a flurry of research activity in recent years,
al., 1998, Aust. J. Zool. 46, 419–437). We examine the the interrelationships between and within the major
implications of separate and simultaneous analyses of clades of arthropods remain a contentious issue. The
sequenceandnon-sequencedata forarthropodrelation- monophyletic status of the Arthropoda is one of the
ships. The most parsimonious trees based on 211 non- few points of widespread consensus, although a few
sequence characters (273 apomorphic states) support workers (e.g., Fryer, 1996) still endorse the view (An-
derson,1973;Manton,1977)thatarthropodsareapoly-
traditionalhighertaxaasclades,includingMandibulata,
phyleticgroup.Reviewsofmajorcompetinghypothe-
Crustacea,Atelocerata,Myriapoda,andHexapoda.Com-
ses for the relationships between chelicerates,
bined analysis of morphology with histone H3 and U2
crustaceans, myriapods, and hexapods, as well as the
sequenceswithequalcharacterweightsdiffersfromthe
status of Onychophora and Tardigrada relative to the
morphologicalresultsaloneinsupportingProgoneata1
euarthropods, have been outlined by Wheeler et al.
Hexapoda (5 Labiophora) in favor of a monophyletic
(1993),Willsetal.(1995,1998),RegierandShultz(1997),
Myriapoda, resolves the entognathous hexapods as a
andZrzavy´ etal.(1997),amongothers.Tobrieflysum-
grade, and supports pycnogonids as sister group to
marizetheseissues,ongoingcontroversyconcernsthe
Euchelicerata(ratherthanasbasaleuarthropods).Mono-
status of
phylyofChelicerata(includingpycnogonids),Mandibu- x a clade composed of Crustacea, Myriapoda, and
lata, Crustacea, Progoneata, Chilopoda, and Hexapoda Hexapoda (the Mandibulata hypothesis) or crusta-
is maintained under a range of transition/transversion ceansalternativelygroupingwiththeChelicerata(the
andthirdcodonweights,whereasAtelocerataandMyria- TCC or Schizoramia hypothesis);
poda/Labiophora do not withstand all sensitivity analy- x Crustaceaasamonophyleticgroup,aparaphyletic
ses. q2000TheWilliHennigSociety grade to other mandibulates, or a paraphyletic grade
to hexapods (the Pancrustacea hypothesis; Zrzavy´ et
al., 1997);
0748-3007/00$35.00 155
Copyrightq2000byTheWilliHennigSociety
Allrightsofreproductioninanyformreserved
156 Edgecombeetal.
x Myriapodaaseitheramonophyleticoraparaphy- etal.,1993),elongationfactor-1a,andRNApolymerase
letic group; II(RegierandShultz,1997,1998)havealsobeenconsid-
x myriapodsasasisterorbasalparaphylumtoHexa- ered. In a forthcoming study in collaboration with W.
poda(theTracheataorAteloceratahypothesis),asister C. Wheeler and G. Giribet, we will integrate histone
grouptoChelicerata,orabasalgroupofeuarthropods; H3 and U2 sequence data with other available se-
x pycnogonids as either sister group to Chelicerata quencesandincludefossilsformorphologicalcodings.
or basal within the Euarthropoda; and Thepresentsynthesislimitsthedatapooltomorpho-
x OnychophoraorTardigradaasmostcloselyrelated logicalcharactersandextanttaxainordertominimize
to other arthropods. theeffectsofmissingdata(morphologybeingtheonly
data set for which all of the terminals sequenced for
Ambiguityordisagreementisalsofoundwhencon- H3andU2arecurrentlyscored).Forsomecharacters,
sidering recent ideas concerning relationships within alternative interpretations based on fossils are noted.
the major groups. Crustacean phylogeny has proven
especiallyrecalcitrant,withsignificantdiscordancebe-
tween recent cladograms (Spears and Abele, 1997;
MATERIALS AND METHODS
Wills, 1997;Schram and Hof,1998). In additionto the
pycnogonid problem, chelicerate phylogeny is most
complicatedbycompetingschemesofrelationshipbe-
Taxonomic Sampling
tweenthearachnidorders(WeygoldtandPaulus,1979,
versus Shultz, 1989, 1990; see Wheeler and Hayashi, Thedeficienciesoftaxonomicsamplinginearlymo-
1998;Weygoldt,1998).Amajorcontroversyinhexapod lecularanalyseswerewellsummarizedbyWheeleret
phylogeny is the mono-, para-, or polyphyly of the al. (1993), and some of these flaws have persisted in
Entognatha, i.e., whether the Diplura, if themselves a subsequent work. The essential problem is that too
clade,aremorecloselyrelatedtoCollembolaandPro- fewexemplarshavebeenexaminedinmoststudiesto
turaortotheInsecta(Kukalova´-Peck,1991;Sˇtysetal., adequatelysampletheenormoustaxonomicdiversity
1993; Kristensen, 1997; Bitsch and Bitsch, 1998). oftheArthropoda.Anempiricaldemonstrationofthe
The objectiveof thisstudy is toevaluate competing need for rigorous sampling is shown by work on 18S
hypotheses of arthropod relationships based upon rDNA,themostexhaustivelysampledgeneforarthro-
morphological and molecular evidence. We employ a pods. Previous indications of a chelicerate–myriapod
broadertaxonomicsamplethanhasbeenusedinmost grouping(FriedrichandTautz,1995;Giribetetal.,1996;
prior molecular work and consider genes that have SpearsandAbele,1997)arerejectedinfavorofChelic-
notpreviouslybeenexaminedinrelationtoarthropod erata as sister to Mandibulata with the inclusion of
phylogeny.SequencedataarederivedfromhistoneH3 additionaltaxa,suchaspycnogonidsandmoreeuchel-
andthesmallnuclearRNAU2(Colganetal.,1998).In icerates (Giribet and Ribera, 1998).
order to subject these sequence data to simultaneous Wherepossible,wehaveavoidedassumingancestral
analysis withmorphological evidence, thesame taxo- (groundplan)characterstatesforthetaxathatwehave
nomicsampleisscoredforanatomical,developmental, scored,usingamoreexplicitexemplarmethod(Yeates,
ultrastructural,andgeneordercharacters.Thisserves 1995).Wehaveselectedseveralrepresentativeswithin
tocompilemuchoftheclassicalevidenceandputsthis each putative major clade of the Arthropoda-and in
evidence in a form by which it can be evaluated for particularfocusedongroupsthathavebeenregarded
its most parsimonious cladograms. as nearly basal in previous investigations. Such taxa
PreviousDNAsequencingstudiesofarthropodphy- aremostlikelytoprovideanestimateofplesiomorphic
logeny have concentrated on ribosomal RNA cis- characterswithintheclade.Somesuchexemplarsmay,
trons—particularly 12S (Ballard et al., 1992), 18S infact,behighlyautapomorphic,butsamplingseveral
(Wheeleretal.,1993;FriedrichandTautz,1995;Giribet putatively basally derived taxa within each clade
etal.,1996;SpearsandAbele,1997;GiribetandRibera, wouldbeexpectedtocounterthisproblem.Sampling
1998), and 28S (Friedrich and Tautz, 1995; Wheeler, has been notably deficient for myriapods, a situation
1998), although the nuclear genes ubiquitin (Wheeler we attempt to address by including all five extant
Copyrightq2000byTheWilliHennigSociety
Allrightsofreproductioninanyformreserved
ArthropodCladistics 157
chilopod orders, diplopod representatives from the that can be more generally applied (presence or ab-
twomostbasallyderivedlineages(fideEnghoff,1984), sence of limb on the eight metamere). Segmental ho-
a pauropod and a symphylan. mologies for euarthropods as outlined by Schram
(1978,Table2),basedontheneuralinnervationscheme
of Bullock and Horridge (1965), are followed in this
workwiththeexceptionofTrilobita(which,whilenot
Coding coded as a terminal, figure in some character discus-
sions).Thetrilobiteantennaisregardedashomologous
For gross anatomical characters codings are based with the (deutocerebral) antennule of crustaceans
on the particular species that was sequenced in the (Mu¨ller and Walossek, 1986), the first biramous limb
molecularanalyses.Werethesameprocedureapplied pair of trilobites with the antenna, the second pair
to most of the developmental, histological, and ultra- withthemandible,andsoon.Analternativehomology
structural (e.g., sperm) characters, very few of these schemeforchelicerateshasbeenadvancedbyDamen
couldbescoredandalargeamountofpreviouswork et al. (1998) and Telford and Thomas (1998) based on
would go unused in this analysis. To use a broader expressiondomainsofHoxgenes.Theseworkerscon-
range of developmental and histological characters, siderthechelicerahomologouswiththeantennule(see
someassumptionsofmonophylyhavebeenmade.For Wheeleretal.,2000,forcommentaryonthisreinterpre-
example, embryological data available for any tation).
chilopod species have been coded for the representa- Homologies of appendage podomere characters
tiveofthesameorder(Scutigeromorpha,Lithobiomor- across the arthropods are an ongoing problem. Al-
pha, Scolopendromorpha, Geophilomorpha) used in thoughmany authorshave attemptedto demonstrate
our analysis. The same practice has been applied pan-arthropodan podomere homologies (e.g., Sharov,
withinthefollowinggroups:Peripatidae,Peripatopsi- 1996; Brusca and Brusca, 1990), these attempts have
dae, Pauropoda, Symphyla, Penicillata and Penta- beenflawedbecauseinappropriatetaxawereusedfor
zonia.DevelopmentalandspermcharactersforTricho- comparison. The fundamental crustacean limb bears
lepidion have been scored based on other Zygentoma. scantresemblancetothatofatelocerates.Thispointis
Some characters (e.g., those dealing with gene ex- clearlyshownbytheallegedstem-lineagecrustaceans
pression and mitochondrial gene arrangements) have (WalossekandMu¨ller,1990)—orstem-lineagemandib-
been examined in few arthropod taxa. We have in- ulates (Lauterbach, 1988; Wa¨gele, 1993; Moura and
cluded these characters because they have featured Christoferssen, 1996)—in which the paucisegmented
prominentlyinrecentdebates,suchasthecrustacean– endopod defies podomere homologies with atelocer-
hexapod (5 Pancrustacea) hypothesis, and should ates or chelicerates. In characters for which some ho-
therefore be used in the analysis if these hypotheses mologies are dubious, we have scored states only for
aretoappraisedviacharactercongruence.Thesechar- those taxa in which we are confident of homologies
acters must be scored as missing data for most termi- (e.g.,withintheAtelocerataorChelicerata).An“uncer-
nals. Some minimal assumptions of monophyly were tain” coding has been used for other taxa.
made to code at least a three-taxon statement (e.g., Except where indicated or where larval or embryo-
codingthesolepterygote,anephemeropteran,fordata logical characters are specifically involved, we have
fromotherpterygotes;codingthehoplocaridKempina restricted decisions of homology to adults. For exam-
for data involving eumalacostracans; and coding Epi- ple, although the nauplii of the branchiopods scored
cyliosoma for chilognathan millipedes). inthisanalysis(BranchinellaandTriops)havebiramous
Character definition obviously requires strict appli- antennae,theadultsdonot;therefore,wehavescored
cationofahomologyschemebetweenmetameresthat this character state as absent for these taxa.
transcends differences in tagmosis between major
groups.Whereacharactermaybeinformativewithin
Outgroups
aparticulargroup(e.g.,presenceorabsenceofalimb
on the first opisthosomal somite in Chelicerata) we The selection of outgroups for rooting arthropod
havemadeattemptstopresentanalternativedefinition trees highlights ongoing controversies in protostome
Copyrightq2000byTheWilliHennigSociety
Allrightsofreproductioninanyformreserved
158 Edgecombeetal.
phylogeny. The classical hypothesis that Annelida is ignored. Tree-space searches (cf. Swofford, 1993:104,
the closest relative of Arthropoda (the Articulata hy- as in Reid, 1996) were performed using 100 random
pothesis)hasfoundsupportinsomecladisticanalyses addition sequencereplicates with threetrees sampled
basedonmorphologicalanddevelopmentalcharacters per iteration (nchuck 5 3, chuckscore 5 1). All trees
(Rouse and Fauchald, 1995; Nielsen et al., 1996). An found by this procedure were then branch-swapped
alternative to this is the Eutrochozoa hypothesis, in using tree bisection–reconnection to check for shorter
which annelids are allied with molluscs and arthro- resolutions and to fill out tree space. Bremer support
podsaremorecloselyrelatedtokinorhynchsandnem- (Bremer,1994)wascalculatedusingMacCladetestver-
atodes(Eernisse etal. 1992).Thelatter grouping,with sion4.0b7(MaddisonandMaddison,1999)toautomat-
the addition of the priapulids and nematomorphs, icallygeneratethePAUP*commandfilewithnegative
findssupportin18SrDNAsequencedataandhasbeen constraints.Theauthorsholddifferingopinionsonthe
namedEcdysozoa (Aguinaldoet al.,1997; Giribetand merits of bootstrapping as a measure of support; we
Ribera,1998).Asourprincipalgoalisexaminingrela- providebootstrapvaluesforcomparisonwithBremer
tionships within the Euarthropoda,we have included support.Bootstrappingused100randomized(withre-
several onychophorans (including representatives of placement) character samples, with each bootstrap
bothfamilies,thePeripatidaeandthePeripatopsidae) sample using the heuristic search options addseq 5
and a tardigrade. All recent work identifies these random, nreps 5 10, nchuck 5 10, chuckscore 5 1.
groupsasappropriateoutgroupsfortheeuarthropods. Eachbootstrapsamplecouldcontainnomorethan100
We have also coded annelids, which accommodates trees, thus reducing iteration bias.
theArticulatahypothesis,butcautionthattheinterpre- LaboratoryandanalyticaltechniquesforhistoneH3
tationofsomecharactersinonychophoransandtardi- and snRNA U2 sequences are described elsewhere
gradeswilldifferfromtreesinwhichaschelminthEc- (Colganetal.,1998).Inthepresentstudy,wesubjected
dysozoa were used as outgroups. Given that 18S these data to a range of weighting schemes to assess
phylogenies agree with results obtained here with re- theeffectsoftheseparametersoncongruencebetween
spect to tardigrades and onychophorans as parts of data sets as well as cladogram sensitivity (Wheeler,
the first and second outgroup branches, respectively, 1995; Wheeler and Hayashi, 1998). Congruence be-
to the Euarthropoda (Fig. 1 in Giribet and Ribera, tween data partitions is measured by incongruence
1998), resolution and character interpretations within length difference (ILD; Mickevich and Farris, 1981),
Euarthropoda should be unaffected by more distant based on individual partition and combined analyses
outgroups. runusingtheheuristicparametersaddseq5random,
nreps5100,nchuck53,chuckscore51.Significance
Analytical Methods
ofpartitionincongruence(Farrisetal.,1994)wastested
Themorphologicaldatasetconsistsof211characters with the partition homogeneity test as implemented
with273apomorphicstatesintotal,withoneconstant inPAUP*;homparheuristicoptionsusedwerenbest5
character.Thelatter(character33,crustaceancardioac- 3, allswap 5 yes, addseq 5 asis.
tive peptide) was included in our data because it has ForH3,thearthropodshave151variablesitesin327
beenaccordedsomesignificanceinotherstudies(Wa¨- aligned bases, of which 139 are informative. Thirty-
gele,1993),despitebeinguninformativeatthelevelof oneof109first-basepositionsarevariable,with4that
oursampling.Twocharactersinvolvingmitochondrial are uninformative. The third-base positions have 106
gene order were included. Multistate characters were variable sites (105 informative). Only 14 second-base
treatedasunorderedexceptforcharacters5,17,25,54, positionsarevariableandonly7oftheseareinforma-
87, 108, 144, and 155. In each of these, evidence for tive.Therefore,mostoftheinformativesitesareinthe
homology of state 1 and state 2 is compelling, and a H3 third position, although these sites may be satu-
transformationseries(0–1–2)canbeposited.Thedata rated (Colgan et al., 1998). The aligned U2 data, 133
wereanalyzedwithPortablePAUPforWindows,ver- bases total, have 53 informative characters, with 64
sion 4.0b2 (Swofford, 1999). Heuristic multiple parsi- constantand16variablecharactersthatareuninforma-
monysearchesemployed10iterationsofrandomstep- tive. Taxa for which only one of the two sequences
wise addition of the taxa, uninformative characters wereobtainedareasfollows:Lithobius,Mecistocephalus,
Copyrightq2000byTheWilliHennigSociety
Allrightsofreproductioninanyformreserved
ArthropodCladistics 159
Campodea, Petrobiinae, Hutchinsoniella (histone H3 se- appropriate starting point for simultaneous analysis.
quenceonly;U2missing),Derocheilocaris,Limulus,and To examine the sensitivity of the combined results to
Macrobiotus(U2sequenceonly;H3missing).Colganet differentanalyticalparametersforthemolecularchar-
al. (1998) tested sensitivity to alternative methods of acters,wevariedtheweightsoftransitionsandofthe
data combination, analyzing with all terminals for third-codon position in the H3 sequences. Higher ho-
which either sequence was available as well as with moplasy rates for third codon positions in protein
only those taxa with both sequences (“spliced” and genes and for transitions are well known (Philippe et
“merged”analysesofNixonandCarpenter,1996).The al., 1996; Yoder et al., 1996). Many systematists have
former approach, of more inclusive taxonomic sam- advocated the downweighting or elimination of the
pling, yielded much more explicit results (Fig. 1) and third codon to decrease putatively negative effects of
is employed in the present study. this homoplasy (see citations in Wenzel and Siddall,
As argued by Allard and Carpenter (1996), equal 1999).Fractionalweightings,implementedbyPAUP*,
weighting (one morphological state change equal to wereusedtovarytheH3third-codonweightingsand
onebasechange,transitionsequaltotransversions,all thetransitionsintransversion/transitionstepmatrices.
codon positionsequally weighted)is theobvious and Tranversionsthereforealwayshadaweightof1,while
transitionsand theH3 thirdcodonhad weightsvary-
ingbetween1and0.Thisprocedureallowedthemor-
phological characters and putatively less homoplastic
DNAcharacterstoretainequalweightsinalltests.We
didnotexploredifferentpartitionweights.Noreason
exists to suspect “swamping” of one partition by an-
other.ThemorphologyandDNAdatasetswerenearly
equalintheirstrength, morphologyhaving210parsi-
monyinformativecharacterswhileH3andU2together
have 192 informative sites.
MORPHOLOGICAL CHARACTERS
1. Non-migratorygastrulation:0,absent;1,present.
Anderson (1973) described the unique pattern of gas-
trulation in onychophorans, and this character has
been accepted as an autapomorphy for Onychophora
(Monge-Na´jera, 1995).
2. Engrailed expressed in mesoderm patterning: 0,
absent; 1, present. Zrzavy´ and Sˇtys (1995) surveyed
“compartment-likepatterning”inthemesodermofan-
nelidsandarthropods,asmarkedbyengrailedexpres-
sion.Limiteddataareavailabletoindicatetheabsence
of such mesodermal patterning in some insects and
crustaceans versus its presence in at least some
chilopods, onychophorans, and annelids. Despite the
preponderance of missing data, this character is in-
cluded as a target for future investigation.
3. Early cleavage: 0, spiral; 1, total cleavage with
FIG.1. Strictconsensusof14shortestcladogramsbasedonhistone
radiallyorientedpositionofcleavageproducts;2,intra-
H3 and U2 sequences (Colgan et al., 1998). Length 1376 steps, CI
0.27,RI0.40.Bremersupportisshownatnodes. lecithalcleavage.Awiderangeofeuarthropodsshare
Copyrightq2000byTheWilliHennigSociety
Allrightsofreproductioninanyformreserved
160 Edgecombeetal.
early total cleavage without oblique spindles, which of the tardigrade brain indicate that “head lobes”
Scholtz (1997) suggested is an autapomorphy of Eu- would form during neurogenesis (R. Dewel, pers.
arthropoda.Wehavecodedforthemostcloselyrelated comm., 1998).
proxy in groups identified by Scholtz as possessing 9. Fatbody:0,absent;1,fatbodycellsdevelopfrom
this derived cleavage pattern. vitellophages in yolk; 2, fat body cells develop from
4. Annelidcrosscleavagepattern:0,absent;1,pres- walls of mesodermal somites. The presence of a ce-
ent. Annelida and Echiura share a distinctive blasto- phalicstorage organ,the fatbody,has beenidentified
merepattern, witha crossformed byblastomere cells as an atelocerate synapomorphy (Boudreaux, 1979).
1a112–1d112 (Rouse and Fauchald, 1995). Anderson (1973) made a distinction between vitello-
5. Blastokinesis:0,absent;1,amnioticcavityopen;2, phagalfatbodycells(Symphyla1Pauropoda1Diplo-
amnioticcavityclosed.Insectembryologyisuniquely poda) and an origin of the fat body in the mesoderm
characterized by the division of the dorsal extra-em- (Chilopoda 1 Hexapoda). Dohle (1980) upheld this
bryonicectodermintoanamnionandaserosa,termed distinctionandemployedtheformerconditionasevi-
blastokinesis (Anderson, 1973). We follow Whiting et denceformonophylyoftheprogoneatemyriapods.A
al. (1997) in regarding the closed amniotic cavity of partial uncertainty coding (states 1 or 2, but not 0) is
Dicondyliaasamodificationoftheopen(Larink,1983) employedforseveraltaxainwhichafatbodyispresent
amniotic cavity of Archaeognatha (i.e., the character but its embryological origin is unknown.
is ordered). 10. Fate map ordering of embryonic tissues: 0, pre-
6. Blastodermcuticle:0,absent;1,present.Anderson sumptivemesodermposteriortopresumptivemidgut;
(1973) identified a thin, highly resistant blastoderm 1,presumptivemesodermanteriortomidgut;2,meso-
cuticle beneath the chorion as shared by Progoneata derm midventral, cells sink and proliferate, midgut
and lacking in Chilopoda. Blastoderm cuticle is also internalizes during cleavage; 3, mesoderm diffuse
present in Collembola, Diplura, Archaeognatha, and through the ectoderm; 4, midgut develops from ante-
Zygentoma(Anderson,1973:180),sothecharacterisa riorandposteriorrudimentsateachendofmidventral
potential synapomorphy for Labiophora, though dis- mesoderm band. Fate map patterns follow Anderson
tinctionfromblastodermcuticleinXiphosura(Ander- (1973,1979).Availabledataforpycnogonidsincludea
son, 1973:370) would be required. diversity of patterns (Schram, 1978) but some species
7. Ectoteloblasts: 0, absent; 1, present, arranged in conform most closely to the chelicerate pattern and
ring.Ectoteloblastsarespecializedstemcellsthatgive have been coded as such.
rise to the ectoderm of most postnaupliar segments 11. Epimorphic development: 0, absent; 1, present.
in Cirripedia and in Malacostraca (Gerberding, 1997). Severalarthropodgroupshavebeendiagnosedbyepi-
They are absent in branchiopods (Gerberding, 1997) morphosis,hatchingwiththecompletecomplementof
andarelackingonlyinAmphipodaamongtheMalac- segments(e.g.,EpimorphawithinChilopoda;Diplura
ostraca (Dohle and Scholtz, 1988; present in Leptos- 1 Insecta fide Kraus, 1997). This character exhibits
traca and Stomatopoda coded here fide Weygoldt, considerablehomoplasywithinArthropodabutserves
1994). Malacostracan ectoteloblasts are characterized as a synapomorphy for several clades.
by their circular or semi-circular arrangement at the 12. Naupliuslarvaoreggnauplius:0,absent;1,pres-
anterior border of the blastopore (Weygoldt, 1994). ent. The nauplius is a “short head” (Walossek and
8. Head lobes (paired semicircular lobes that give Mu¨ller,1997)swimminglarvathathasonlythreepairs
risetothelateraleyesandlateralpartsoftheprotocere- oflimbs(antennule,antenna,mandible),eachofwhich
brum): 0, absent; 1, present. Scholtz (1997) regarded has a generalized morphology. The naupliar antenna
headlobesasasynapomorphyofonychophoransand has a proximal enditic process (“naupliar process”)
euarthropods, with no corresponding structure in an- that acts as a food-pushing device in the absence of a
nelids. Although Scholtz (1997) did not address this gnathal lobe on the mandible. Some authors (e.g.,
characterintardigrades,totalcleavageandthelackof Fryer, 1992, 1996) have suggested that the nauplius is
agermbandstageintardigradesmeanthatstructures the primordial crustacean form, although even Cam-
comparable to the head lobes of germ band embryos brian fossils well known to be crustaceans (Walossek,
are absent, although the prominent dorsolateral lobes 1993)showclearlythatthenaupliusisonlyatransient
Copyrightq2000byTheWilliHennigSociety
Allrightsofreproductioninanyformreserved
ArthropodCladistics 161
larvalstage.Absenceofanaupliuscannotbemeaning- excludes annelids. Ecdysone-like steroidal hormones
fully interpreted for terrestrial arthropods, and we induceandcontrolmoltinginnon-arthropodEcdyso-
have opted for an inapplicable coding (cf. Wa¨gele, zoa (see nematode citations in Schmidt-Rhaesa et al.,
1993:278). 1998) as well as in Arthropoda.
13. Pupoid stage: 0, direct hatching; 1, motionless 20. Ecdysialsuturepattern:0,transverserupturebe-
stageafterhatching,pupoidremainsencasedinembry- tween headand trunk; 1, dorsallongitudinal ecdysial
oniccuticle.Anderson(1973)summarizedevidencefor suture;2,marginalecdysialsuture;3,ruptureatstylus
a pupoid stage in Chilopoda, Diplopoda, and Pauro- apparatus.States0–2wereusedbyZrzavy´ etal.(1997).
poda.Dohle(1997),however,identifiedapupoidstage Boudreaux(1979) regardedecdysis atthe head–trunk
asconfinedtodiplopodsandpauropods.Werecognize contact (state 0) to be diagnostic of Myriapoda and
the peripatoid and fetoid stages of Epimorpha dorsal longitudinal ecdysis (state 1) to be diagnostic
(Chilopoda) as character 181. for Hexapoda. Snodgrass (1952:269) specified that the
14. Sclerotization of cuticle into hard, articulated latterpertainedtoInsectainparticular,whereasColle-
exoskeleton: 0, absent; 1, present. mbola and Protura have a head–trunk ecdysial split
15. Cuticlecontaininga-chitinandprotein:0,absent; (Kaufman,1967:16).R.Dewel(pers.comm.,1998)indi-
1, present (Weygoldt, 1986). The composition of the cates that tardigrades molt at the stylus apparatus.
chitinintardigradecuticleisnotknownwithcertainty 21. Resilin protein: 0, absent; 1, present. Weygoldt
(DewelandDewel,1997).a-chitinis,however,shared (1986)indicatedthatthespiralproteinresilinisknown
by euarthropods (but not pentastomids), onychopho- only from euarthropods and onychophorans. Nielsen
rans, and non-arthropod Ecdysozoa (Priapulida; see (1995),however, mappedresilin ontothe treeas aeu-
Schmidt-Rhaesa et al., 1998, for citations). arthropodsynapomorphy,indicatingitsabsenceintar-
16. Exocuticular cones: 0, absent or moderately de- digrades and onychophorans.
veloped; 1, extensively developed in cuticle of head. 22. Molting gland: 0, absent; 1, present. Wa¨gele
Manton (1965) identified cuticular specializations for (1993)citedamoltingglandasadiagnosticcharacterof
flexibilityandstrengthassynapomorphiesoftheGeo- Mandibulata.Thiswasbasedonaproposedhomology
philomorpha, linking these to the burrowing habits between the Y-organ of Malacostraca and the protho-
of the clade. Exocuticular cones are especially well- racicglandofinsects.Wa¨gelenotedthatsuchmolting
developedintheheadandmaxilliped.Theyareabsent glandsininsectsandcrustaceansarehypodermalderi-
in“anamorphic”chilopodsandmuchlessextensively vationsofthesecondmaxillaandareabsentinchelic-
developed in scolopendromorphs than in geophilo- erates. An alleged ecdysial gland in some chilopods
morphs (Manton, 1965). (Lithobiomorpha, Seifert and Rosenberg, 1974;
17. Ectodermal cilia: 0, present in many tissues; 1, glandulacapitisinScutigeromorpha,Seifert,1979)may
present in photoreceptors and sperm; 2, present in be homologous. Evidence for an ecdysial gland has
spermonly.WefollowWheeleretal.(1993,theircharac- not been found in other myriapods (Tombes, 1979)
ter65)inscoringthisasanorderedmultistatecharacter. except for polyxenid millipedes (glandula perioeso-
18. Tendoncellswithtonofilamentspenetratingepi- phagealis,Seifert,1979).TherestrictionoftheY-organ
dermis: 0, absent; 1, present. Boudreaux (1979) and to Malacostraca within the Crustacea (Fingerman,
Wa¨gele (1993) acknowledged tonofilaments as a eu- 1987)isproblematicforthehomologyoftheseglands.
arthropod synapomorphy, and Dewel and Dewel Studies of branchiopods have not discovered similar
(1997) confirmed their absence in onychophorans moltingglandsalthoughmoltinghormonesappearto
and tardigrades. be present (Martin, 1992).
19. Molting with ecdysone: 0, absent; 1, present. 23. BismuthstainingofGolgicomplexbeads:0,not
Molting is frequently evoked as a synapomorphy of staining; 1, staining. Locke and Huie (1977) observed
Panarthropoda (e.g., Weygoldt, 1986), although it has GolgibeadstostainwithBismuthinvariouseuarthro-
alternativelybeensuggestedtobeaplesiomorphyfor podsandtardigrades,butnotinOnychophora,Anne-
abroadergroupthatalsoincludesnematodes,nemato- lida,Mollusca,Nematoda,orPlatyhelminthes.Dueto
morphs, priapulids, kinorhynchs, and loriciferans thedepauperatetaxonomicsamplingforthischaracter,
(Aguinaldoetal.,1997;Schmidt-Rhaesaetal.,1998)but we have coded with the following proxies examined
Copyrightq2000byTheWilliHennigSociety
Allrightsofreproductioninanyformreserved
162 Edgecombeetal.
by Locke and Huie (1977): undetermined isopod and with To¨mo¨sva´ry organs, whereas Tuxen (1959) re-
OrconectesforKempina,undeterminedpolydesmoidfor gardedthemasantennalvestigesonthebasisofmus-
Epicyliosoma, and representatives of four pterygote culation. The postantennal organs of Collembola may
orders for Atalophlebia. alsobehomologous(Haupt,1979).Wehavescoredthe
24. Metanephridiawithsacculuswithpodocytes:0, temporalorgansofEllipuraashomologouswiththose
absent; 1, present. While metanephridia are probably ofMyriapoda,followingHaupt(1979)andBitschand
plesiomorphic for arthropods (Fauchald and Rouse, Bitsch(1998).ThehomologueoftheTo¨mo¨sva´ryorgan
1997),thesacculusandpodocytesarenovelnephridial in Craterostigmus is a ringed organ set on a triangular
structuresforonychophoransandeuarthropods,lack- sclerite lateral to the clypeus (Fig. 50 in Shear and
ingintardigrades(Nielsen,1997;Schmidt-Rhaesaetal., Bonamo, 1988; Fig. 2 in Dohle, 1990). The form and
1998). Absence of nephridia in pycnogonids is coded positioningofthisorganarecomparabletotheTo¨mo¨s-
following King (1973) and Clarke (1979). va´ry organ in Lithobiomorpha (Henicopidae).
25. Distribution of segmental glands: 0, on many 27. Salivary gland reservoir: 0, absent; 1, present.
segments;1,inatmostthelastfourcephalicsegments Monge-Na´jera(1995)identifiedasalivaryglandreser-
andfirsttwopost-cephalicsegments;2,onsecondan- voir as an onychophoran autapomorphy.
tennal and maxillary segments only. Definition of the 28. Malpighian tubules formed as endodermal ex-
basic euarthropod distribution of segmental glands, a tensions of the midgut: 0, absent, 1, present. Shultz
reductionfromthatinOnychophora,followsWeygoldt (1990) claimed that endodermal Malpighian tubules
(1986). We have not attempted to define all variants areuniquetoArachnida and,despitetheirabsencein
ofsegmentalglanddistributionwithinEuarthropoda, someingrouptaxa(suchasOpiliones),resolvedthem
and state 1 above is an artificial grouping. A more asanarachnidautapomorphy.AtelocerateMalpighian
advanced reductionin Crustacea,restricted tothe an- tubules, in contrast, are extensions of the ectodermal
tennal and maxillary segments, has been regarded as hindgut(seecharacter29).Thenon-homologyofthese
a crustacean synapomorphy (Lauterbach, 1983, 1986). structuresisgenerallyrecognized,andwehaveaccord-
WalossekandMu¨ller(1990:410)consideredremipedes ingly coded them as separate characters.
(Schram and Lewis, 1989) and anostracans to deviate 29. Malpighiantubulesformedasectodermalexten-
from this state in possessing additional cephalic seg- sions of the hindgut: 0, absent; 1, single pair of Mal-
mental glands, but Wa¨gele (1993) dismissed these as pighian tubulesat junctureof midgut andhindgut; 2,
integumental glands and embryonic mesodermal multiple pairs of tubules at anterior end of hindgut.
cells, respectively. The presence of Malpighian tubules in Collembola is
26. To¨mo¨sva´ry organ (“temporal organs” at side of dubious (Clarke, 1979; Bitsch and Bitsch, 1998), while
headbehindinsertionofantennule):0,absent;1,pres- Proturahaveseveralpairsofpapillaebehindthemid-
ent. Homology of To¨mo¨sva´ry organs (Fig. 2A) across gut–hindgutjunction(seecharacter30).Distinctcondi-
the Myriapoda has been widely accepted (Snodgrass, tionscanberecognizedwithinthemyriapodsandthe
1952), but relationships to similarly positioned struc- ectognathhexapodsandserveasthebasisforstates1
turesinhexapodsarecontentious.Franc¸ois(1969),for and 2 above. The origin of insect Malpighian tubules
example, homologized the pseudocellus of Protura (whether ectodermal or entodermal) is controversial
FIG. 2. (A–C)Hanseniella n.sp. (Symphyla), Mt.Colah, NSW,Australia. (A)Lateral view ofhead, showingmandibular base plate(Md),
To¨mo¨sva´ryorgan(TO),andspiracle (SP).(B)Ventralviewoftrunk,showingstyli(St) andeversiblevesicles/coxalsacs(Vs).(C)Pretarsal
claws.(D)Schizoribautian.sp.(Chilopoda,Geophilomorpha),Sydney,NSW,Australia.Lateralviewoftrunk,showingventralconfluenceof
procoxa (PrCx) and metacoxa (MtCx) and elaboration of pleurites, including scutellum (Sc), stigmatopleurite (StPL), and small pleurites
(SmPL)betweentergite(Tg)andsternite(St).(E)Allomachilisfroggatti(Archaeognatha),Kiama,NSW,Australia.Ventrolateralviewofhead,
showingdivisionoflabialglossae(GL)andparaglossae(Pg).(F)Pauropodinae,3kmNofWeldborough,Tasmania,Australia.Ventralview
ofposteriorpartofhead,collumsegment,andanteriorpartoffirstleg-bearingtrunksegment,showingcollumorgans(CO)variablyinterpreted
asvesicles(Tiegs,1947)orappendagevestiges(KrausandKraus,1994).Scalebars:AandE,200m;BandD,100m;F,20m;C,10m.
Copyrightq2000byTheWilliHennigSociety
Allrightsofreproductioninanyformreserved
ArthropodCladistics 163
Copyrightq2000byTheWilliHennigSociety
Allrightsofreproductioninanyformreserved
164 Edgecombeetal.
(Dohle, 1997). One or a small pair of supernumerary derivedcharacterofonychophorans,tardigrades,and
Malpighian tubules is present in some chilopods euarthropods (Weygoldt, 1986).
(PrunescuandPrunescu,1996).Theso-calledMalpigh- 36. Dorsalheartwithsegmentalostiaandpericardial
ian tubules of eutardigrades are not in contact with sinus:0,absent;1,present.Thedorsal,ostiateheartand
cuticleand assuchdonot appeartobe ectodermalin pericardialsinus/septumaresharedbyOnychophora
origin (Møbjerg and Dahl, 1996). andEuarthropoda,butabsentintardigradesandpaur-
30. FormofectodermalMalpighiantubules:0,elon- opods. We have not coded with the assumption that
gate;1,papillate.TheMalpighiantubulesareelongate these absences are due to miniaturization.
in myriapods and most hexapods, Bitsch and Bitsch 37. Arterialbloodsupplytolimbs:0,absent;1,limbs
(1998, theircharacter 14) interpreting thisas the basal receive blood from a supraneural artery; 2, limbs re-
stateforAtelocerata.BitschandBitschcodedthevery ceivebloodfromasubneuralartery.Mostatelocerates
similarpapillaeofProturaandCampodeinaashomol- lackanarterialbloodsupplytothelegs.Clarke(1979)
ogous, a procedure adopted here. identified different patterns of arterial branching in
malacostracan crustaceans and in many chelicerates.
31. Neck organ: 0, absent; 1, present. Martin and
38. Slitsensilla:0,absent;1,present.Slitsensillaare
Laverack (1992) have reviewed the so-called dorsal
small clefts or slits in the cuticle, used in detecting
organ or neck organ (Walossek, 1993) of crustaceans.
compressionalforcesactingontheexoskeleton(Shultz,
The term “neck organ” is preferred for this structure
1990).Theyhavebeenrecognizedasasynapomorphy
so as to avoid confusion with the region of extra-em-
for Arachnida, but are (doubtfully) present in the ex-
bryonic ectoderm that is commonly called a “dorsal
tinct Eurypterida (Dunlop and Braddy, 1997).
organ” in many groups of arthropods (Fioroni, 1980).
39. Neuroblasts: 0, absent; 1, present. The identity
32. Hemocyanin: 0, absent; 1, present. Codings for
and relative positions of cell types in the central ner-
the presence of hemocyanin follow Gupta (1979). In
vous system exhibit impressive similarities between
pycnogonids, hemocyanin is only found dissolved in
insectsandsomemalacostracancrustaceans(forwhich
theplasma(ArnaudandBamber,1987),withoutcyano-
Kempina is coded as a proxy). Certain chilopods have
cytessensuGupta.Someremipedeshavelargehemocy-
differentpatternsofsegmentalneurons(Whitingtonet
anincrystalsscatteredthroughouttheheadandswim-
al., 1993), and neuroblasts, the precursor cells of the
ming appendage tissue (J. Yager, pers. comm., 1998).
embryonic ganglia, are variably described as lacking
WithinChilopoda,Scutigeromorphahavehemocyanin
(Osorio and Bacon, 1994; Zrzavy´ and Sˇtys, 1994) or
astheoxygentransportmolecule(Mangumetal.,1985)
present(Scutigera:Knoll,1974).Sincerecenttreatments
versusgaseousexchangebetweenthetracheaeandthe
(e.g., Gerberding, 1997) consider myriapod ganglia to
tissuesinPleurostigmophora(Hilken,1997).Hemocy-
developwithoutneuroblasts,wearereluctanttoaccept
anin is lacking in hexapods (Beintema et al., 1994).
Knoll’s(1974) interpretationof Scutigera.Scutigera ap-
33. Crustacean cardioactive peptide in neurosecre-
pearstohavesomelargercellsintheneurogenicarea
torycellsofnervoussystem:0,absent;1,present.Wa¨-
but the asymmetric division that characterizes neuro-
gele(1993)documentedthesimilarityinthesequence
blasts is not shown (G. Scholtz, pers. comm., 1999).
of this nonapeptide in insects and eumalacostracans
Dohle(1997)indicatedthatonychophoranandchelic-
and proposed it as a mandibulate synapomorphy, al- erateganglionformationresemblesthatofcentipedes
though no evidence was presented to confirm its ab- and millipedes.Weygoldt (1998:68)likewise observed
senceincheliceratesoritspresenceinmyriapods.Miss- thatthenervoussystemincheliceratesandmyriapods
ingdatarenderthischaracterentirelyambiguous,but arises by invagination. We have coded Euperipatoides
we include it to encourage further investigation. and Epicyliosoma as proxies because details of neuro-
34. Subcutaneous hemal channels in body wall: 0, genesis are unknown for most taxa considered in this
absent;1,present.PresenceisuniquetoOnychophora analysis. Gerberding (1997) showed that Cladocera
(Brusca and Brusca, 1990; Monge-Na´jera, 1995). have cells with the characteristics of neuroblasts and
35. Hemocoel: 0, absent; 1, present. Disintegration that their morphology is identical to that in Malacos-
of the coelomic cavities, resulting in the body cavity traca;branchiopodsinthepresentstudyarethuscoded
being used as a hemocoel or mixocoele, is a shared ashavingneuroblasts.Neuroblastshavebeenreported
Copyrightq2000byTheWilliHennigSociety
Allrightsofreproductioninanyformreserved