Evolutionary trends and patterns in centipede segment number based on a cladistic analysis of...

Evolutionary trends and patterns in centipedesegment number based on a cladistic analysis ofMecistocephalidae (Chilopoda: Geophilomorpha)

LUC IO BONATO , DONATELLA FODDAI andALESSANDRO MINELL IDepartment of Biology, University of Padova, Padova, Italy

Abstract. Evolutionary changes in segment number during the radiation ofMecistocephalidae, a group of geophilomorph centipedes with segment numberusually invariant at the species level, were explored based on a cladistic analysis offorty-six mecistocephalid species, representative of the extant diversity in segmentnumber. The data matrix included 118 morphological characters. Trends wererecognized in the evolution of segment number and discussed in relation to theunderlying ontogenetic mechanisms of segmentation. The basic trend was towardsan increasingly higher number of leg-bearing segments, from (most probably)forty-one to sixty-five (101 in one exceptional case). Changes always involvedeven sets of segments. Additions of two, four or eight segments usually occurred,but a case of overall duplication of the whole number was also documented. Mostchanges occurred starting from values belonging to the arithmetical series forty-one, forty-five, forty-nine, whereas the intermediate values forty-three, forty-seven, fifty-one were often evolutionary dead-ends. This evidence suggests amultiplicative mechanism of segmentation involving one or more final run ofduplication, as well as a precise control of the final number of segments whichproduces absolute number stability, except for a single, highly derived species withan exceptionally high number of segments. These ideas contribute to a moregeneral model of arthropod segmentation recently developed by Minelli. A taxo-nomic revision of mecistocephalids is presented: three subfamilies are proposed(Arrupinae, Dicellophilinae and Mecistocephalinae) and Sundarrup is recognizedas a junior synonym of Anarrup.

Introduction

Phylogenetics vs developmental genetics in the study of

centipede segmentation

The origin and evolution of segmentation features as one

of the most fashionable topics in evolutionary developmental

biology (Gerhart & Kirschner, 1997; Hall, 1998; Wolpert,

1998; Carroll et al., 2000). Fundamental contributions have

been provided by the study of the embryonic pattern of

expression of the so-called segmentation genes. A wealth

of data has been gathered, in particular, for Drosophila,

where a dozen segmentation genes have been characterized

in terms of nucleotide sequence and of spatial and temporal

patterns of expression. Since the emergence of this new

awareness of the complexity of the genetic control of seg-

mentation processes in the fruitfly embryo, a major problem

emerged, i.e. to what extent segmentation in Drosophila can

be safely regarded as representative of segmentation in

insects, or in arthropods at large. More recently, a sequence

of putative orthologues of Drosophila segmentation genes

was obtained from a variety of metazoans and the embryonic

expression patterns of some of these genes were described in

other arthropods and in nonarthropod metazoans. With the

progress in the knowledge of these and other genes involved

in establishing basic features of body architecture, however,

it became increasingly clear that the presence of conserved

Correspondence: Alessandro Minelli, Department of Biology,

University of Padova, via U. Bassi 58B, I-35131 Padova, Italy.

E-mail: [email protected]

Systematic Entomology (2003) 28, 539–579

# 2003 The Royal Entomological Society 539

orthologues of these genes does not guarantee, per se, the

conservation of their expression patterns. Furthermore, the

conservation of their expression pattern does not guarantee

the conservation of the role of these genes in the specification

or patterning of body features, due to possible changes in

the control cascades of which these genes are part. This may

be due, in particular, to changes in promoter sequences.

These circumstances cause difficulties in generalizing from

one or a few model systems to a whole phylum or, still

worse, to interphylum comparisons. This explains the

disparity of opinions recently expressed as to the origin

and evolution of segmentation in metazoans. On the one

hand, the presence of orthologues of Drosophila segmenta-

tion genes in nonarthropod animals, including vertebrates,

and the similarities in the pattern of expression of at least

one of these genes between arthropods and vertebrates, has

been used to support the concept of a very ancient origin of

segmentation, in the so-called Urbilateria, the putative

ancestor of all triploblastic animals (Kimmel, 1996;

Balavoine, 1997; DeRobertis, 1997; Holland et al., 1997).

On the other hand, the differences between arthropods and

annelids in the way segmentation is achieved have been

construed as proof that these two clades evolved segmenta-

tion independently from one another (Minelli & Bortoletto,

1988; Conway Morris, 1994; Valentine, 1994, 1995; Budd,

1996; Minelli, 1998, 2000), thus bringing fundamental

support to the dismantling of the traditional concept of

Articulata, a superphylum of segmented protostomes,

mainly ArthropodaþAnnelida, a view also supported by

molecular evidence (Eernisse et al., 1992; Aguinaldo et al.,

1997; Eernisse, 1997; Giribet et al., 2000).

We believe that it is now time to study evolutionary

developmental problems, such as the origin and evolution

of segmentation, using an integrated approach, where the

new evidence stemming from developmental genetics and

molecular biology of a few model animals is supplemented

by a comparative study of those groups where the trait we

are interested in (in this case, segmentation) exhibits an

extensive pattern of variation.

From this point of view, myriapods and, especially,

centipedes offer a unique opportunity among extant

arthropods. The diversity of intra- and interspecific vari-

ation in segment number, coupled with the diversity of post-

embryonic schedules in the expression of the final

complement of trunk segments, is obviously attractive as a

potential source of information about the way arthropod

segmentation may have evolved. No fewer than three

research groups are currently working on myriapod devel-

opmental genetics, but the results are still too fragmentary

to be used in a comparative context. Unfortunately, due to

their very long life cycle and to many aspects of their

reproductive biology, including the obligate parental care

provided by scolopendromorph and geophilomorph centi-

pedes and some millipedes to their broods, no millipede or

centipede species seems likely to become a convenient model

animal for experimental study like Drosophila, Schistocerca

and Artemia.

On the other hand, an extensive dataset is available on

the diversity and variation of trunk segment number in

millipedes and centipedes. In this paper, we demonstrate

that plotting these data on to a cladistic analysis of the

relevant taxa may provide a valuable contribution to the

study of evolutionary trends in arthropod segmentation.

The group focused on in this study wasMecistocephalidae.

The reasons for this choice are quite obvious, in the light

of the phylogenetic position of this taxon. Morphological

(Prunesco, 1967; Foddai, 1998; Foddai & Minelli, 2000)

as well as molecular evidence (Edgecombe et al., 1999;

Giribet et al., 1999; Edgecombe & Giribet, 2002) agree

in placing mecistocephalids in a basal position within the

geophilomorph centipedes (Fig. 1). This is reflected in two

features of their segmentation, both of them of crucial

Scutigeromorpha Craterostigmomorpha

Lithobiomorpha

Scolopendromorpha Adesmata

MecistocephalidaeDevonobiomorpha

Fig. 1. Phylogenetic relationships of the major centipede groups [according to Dohle (1985), Shear & Bonamo (1988), Borucki (1996),

Edgecombe et al. (1999, 2000), Giribet et al. (1999, 2001), Foddai & Minelli (2000), Kraus (2001) and Edgecombe & Giribet (2002)]. For a

different view, however, see Shultz & Regier (1997) and Regier & Shultz (2001). Habitus figures modified after Manton (1965), Shear &

Bonamo (1988) and Shinohara (1999).

540 L. Bonato et al.

# 2003 The Royal Entomological Society, Systematic Entomology, 28, 539–579

importance in the context of this paper. On the one hand,

the number of leg-bearing segments is invariant within

each species (but see below for an exception) and identical

in the two sexes, as it is in the most basal centipede clades

(scutigeromorphs, lithobiomorphs, craterostigmomorphs,

scolopendromorphs), whereas in all species of higher

geophilomorphs (Adesmata) there is a more or less large

amount of intraspecific variability in the number of trunk

segments and the females have a segment number higher

than the conspecific males. On the other hand, the number

of leg-bearing segments in mecistocephalids (forty-one to

101) is much higher than in the more basal centipedes

(fifteen in scutigeromorphs, lithobiomorphs and cratero-

stigmomorphs, twenty-one or twenty-three in scolopendro-

morphs) and well in the range of the other geophilomorphs

(twenty-seven to 191); still more relevant, the number is

different in different species.

The regular spacing of the most frequent segment numbers

in this family (forty-one, forty-five and forty-nine) has

already been remarked upon (Minelli & Bortoletto, 1988)

and is obviously suggestive of numerical constraints deriving

from the very mechanism of segmentation. Many questions,

however, have never been addressed before. First, which is

the plesiomorphic segment number within mecistocepha-

lids? Second, whether this clade exhibits a consistent trend

towards higher (or lower) segment numbers or not. Third,

whether the evolutionary changes which occurred in the

segment number followed some regularity or not. Finally,

the recent discovery (Bonato et al., 2001) that the mecisto-

cephalid species with the highest number of trunk segments

shows an intraspecific variability in this character, at variance

with all other species in the family, prompted us to

investigate the phyletic position of this unique species

within mecistocephalids. We tried to address these questions

by reading the evolution of the segment number upon a

phylogeny of mecistocephalids to recognize possible trends,

constraints and patterns.

Centipede segmentation: pattern and process

The view of centipede segmentation providing the back-

ground to our analysis of mecistocephalids is the double

segmentation model first suggested by Maynard Smith

(1960) and subsequently developed, with special regard to

the centipedes, by Minelli & Bortoletto (1988), Minelli

(2000, 2001) and Minelli et al. (2000). Mecistocephalids

are quite typical of the apparently counterintuitive behav-

iour observed by Maynard Smith, of animals with a very

high number of serial elements (here, trunk segments) not

showing any evidence of intraspecific variability. It seems

improbable, indeed, that these elements are serially gener-

ated from a growth zone by an error-free process. A lack of

intraspecific variability, instead, could be expected if these

serial elements were produced in two steps: first, production

of a small number of first-order units; second, subdivision

of each first-order unit into a fixed number of second-order

units. According to Minelli (2001), the trunk segments of a

centipede (including the forcipular segment, the leg-bearing

segments and the terminal segments) possibly originates

through the secondary subdivision of nine primary units.

The number of primary units is supposed to be the same in

all centipedes, the different centipede clades differing

instead in the number of secondary units derived from

each primary unit (e.g. two from each primary unit in

lithobiomorphs, three from most primary units in scolopen-

dromorphs, four from most primary units in geophilo-

morphs with thirty-one pairs of legs). In this context, a

comparison with the expression patterns of the pair-rule

genes in Drosophila (Lawrence, 1992; Carroll et al., 2000)

is interesting in that a first seven-stripe expression is

followed by a secondary fourteen-stripe expression, with

sets of two secondary stripes corresponding, more or less

closely, to each primary stripe. It is therefore possible that a

‘double segmentation’ mechanism represents a generalized

feature of arthropod segmentation (Minelli, 2001).

Mecistocephalidae

Within geophilomorph centipedes, Mecistocephalidae is

a monophyletic basal clade well characterized by several

peculiar morphological features of the cephalic capsule,

the mouthparts, the forcipular segment and the trunk

sterna. In all mecistocephalids, the number of segments

does not increase during the postembryonic development:

the centipede hatches with the full complement of segments

because segmentation is completed during the embryonic

life.

Mecistocephalids are characterized by the intraspecific

invariance in segment number, a plesiomorphic trait which

is lost in the other geophilomorphs, i.e. Adesmata (see

Minelli & Bortoletto, 1988). A single case of intraspecific

variability in segment number recently found in Mecistoce-

phalus microporus (see Bonato et al., 2001), does not detract

from the phylogenetic relevance of this trait.

Different species are characterized by different numbers

of segments. Fifteen different numbers are known: almost

all the odd numbers from forty-one to sixty-five (except for

fifty-five and sixty-one) and the odd numbers from ninety-

three to 101 (except for ninety-nine).

The taxonomy of the family has not been revised since

Attems’s (1929) standard account, which is by now largely

unsatisfactory. Taxonomic work on this group has mostly

overlooked the standardization of morphological descrip-

tions and the checking of the diagnostic value of traditional

characters, particularly with respect to developmental

changes and intraspecific variability. Useful exceptions are

three of Crabill’s papers (Crabill, 1959, 1964, 1970), whose

contribution, however, has been substantially ignored by

subsequent authors.

About 170 mecistocephalid species in a dozen genera are

currently recognized. Most of the species are included in the

large genus Mecistocephalus, especially in the nominotypical

subgenus.

Evolution of segment number in Mecistocephalidae 541


Materials and methods

A cladistic analysis of the whole group of mecistocephalids

was performed. The monophyly of this group is supported

by reliable synapomorphies (see below under Taxonomic

implications and Appendix 3). As operational units, species

were preferred to taxa belonging to higher taxonomic levels.

Data were thus collected referring to individual species

rather than to hypothetical ancestors, according to the

exemplar approach (Yeates, 1995; Prendini, 2001).

Ingroup

Forty-six species of mecistocephalids were considered

(Table 1), representatives of the morphological diversity

and the geographical distribution of the group, as well as

all different numbers of segments on record. Species belong-

ing to all current genera and subgenera were included,

except for the following minor taxa, too incompletely

known and for which no specimens were available: Fusichila

Chamberlin (monotypic), Mecistocephalus (Ectoptyx)

Chamberlin (five species), Megalacrus Attems (monotypic),

Partygarrupius Verhoeff (monotypic). As far as possible,

type species were preferred over other species. For Agno-

strup Foddai, Bonato, Pereira & Minelli, Mecistocephalus

(Dasyptyx) Chamberlin, Mecistocephalus (Pauroptyx)

Chamberlin and Tygarrup Chamberlin, for which

the type species are too poorly known, other species were

chosen.

Outgroup

Outgroup species were chosen on the basis of the phylo-

genetic scenario most generally accepted for the centipedes

(Fig. 1). Two species were selected from Adesmata, one

from Scolopendromorpha, one from Lithobiomorpha

(Table 1). Adesmata are very diverse and most of them are

characterized by specialized traits, thus they could be mis-

leading as outgroup taxa. Scolopendromorphs appear more

homogeneous and conservative in morphological features;

however, the anatomy of their forcipular segment is highly

derived and some similarities between geophilomorphs and

cryptopid scolopendromorphs may be due to convergent

adaptation to underground life. The extinct devono-

biomorphs were not considered because their morphology

is only partially known. Craterostigmomorphs were not

included because some derivative characters, the anogenital

capsule in particular, are hard to interpret comparatively.

Lithobiomorphs, instead, seemed to be very suitable as an

outgroup, because their anatomy is very close to the

hypothetical ground-plan of the centipedes (Dohle, 1985;

Borucki, 1996).

In the analysis, only the lithobiomorph species was used

to root the trees. The other species were not bound.

Characters

In total, 118 characters were considered, referring to

morphological and anatomical features (Appendix 1).

Characters representing autapomorphies of single species

were excluded.

To define the characters properly, preliminary investiga-

tions were performed on large series of specimens belonging

to representative species. All characters were checked for

their stability during postembryonic development, for

sexual dimorphism and for interindividual variability.

Characters traditionally used in the taxonomy of this

group were checked for their diagnostic value. Additional

characters were recognized as informative and, thus,

considered in the analysis.

Collection of data

As far as possible, series of adult specimens, both males

and females, were studied for each species. All four out-

group species and thirty-one of the forty-six ingroup species

could be studied directly (Table 1). For the remaining

species, data were collected from the primary literature.

For light microscopy investigations, specimens were

clarified with a lactophenol solution and mounted on

temporary glass slides. For some specimens, dissection of

the mouthparts was needed. Standard anatomical parts

were drawn for each species, by means of a camera lucida.

Phylogenetic analysis

Two different analyses were performed, one considering

only those characters not referring to the number of seg-

ments (116 characters, chs 1–116 in Appendix 1), the other

considering all the 118 characters. By excluding the char-

acters which we wanted to optimize a posteriori, the first

analysis allowed us to avoid any possibility of circular

reasoning (see, e.g. Brooks & McLennan, 1991). Conver-

sely, the second analysis allowed us to exploit completely all

potentially informative data to infer phylogeny, not exclud-

ing any character a priori (see, e.g. Miller & Wenzel, 1995).

The inclusive data matrix is shown in Appendix 2. Binary

coding was applied as far as possible. When multistate

coding was required, the different states were treated as

unordered. All characters were originally equally weighted.

PAUP* 4.0 b10 (Swofford, 2002) was used to find the most

parsimonious trees by means of heuristic search. The search

strategy actually consisted of 1000 replicates of a stepwise

addition procedure with random sequence, followed by tree

bisection and reconnection (TBR) branch swapping of a

maximum of ten trees (PAUP* command: HSearch

AddSeq¼ random NReps¼ 1000Hold¼ 10). Both AccTran

and DelTran character optimization methods were

performed. An iterative procedure of successive weighting

and search was then applied (Farris, 1969): at each subse-

quent run, each character was re-weighted according to its



Table 1. Species considered in the analysis. Within Mecistocephalidae, species are entered with their current name and are listed in

alphabetical order. For each species, the number of leg-bearing segments (n) and the overall distribution are given. Species which are types in

their genera are marked with an asterisk.

Species n Distribution Direct study

Lithobiomorpha

Lithobius forficatus (Linnaeus, 1758) 15 Europe, Mediterranean Basin þScolopendromorpha

Cryptops anomalans Newport, 1844 21 Europe, Mediterranean Basin þAdesmata

Geophilus insculptus Attems, 1895 43–47 Europe þStigmatogaster gracilis Meinert, 1870 83–111 Mediterranean Basin þ

Mecistocephalidae

Agnostrup paucipes (Miyosi, 1955) 41 Hondo –

Anarrup nesiotes Chamberlin, 1920* 41 Sulawesi –

Arrup dentatus (Takakuwa, 1934) 41 Hokkaido þArrup holstii (Pocock, 1895) 41 Eastern Asia þArrup pylorus Chamberlin, 1912* 41 California –

Dicellophilus anomalus (Chamberlin, 1904) 41 California þDicellophilus carniolensis (C. L. Koch, 1847) 43 Central Europe þDicellophilus latifrons Takakuwa, 1934 41 Hondo þDicellophilus limatus (Wood, 1862)* 45 California þKrateraspis meinerti (Sseliwanoff, 1881)* 45 Central Asia –

Krateraspis sselivanovi Titova, 1975 53 Central Asia –

Mecistocephalus (Mecistocephalus) angusticeps (Ribaut, 1914) 47 Eastern Africa –

Mecistocephalus benoiti Dobroruka, 1958 49 Eastern Africa þMecistocephalus (Mecistocephalus) conspicuus Attems, 1938 49 Indochinese Peninsula, Java –

Mecistocephalus (Mecistocephalus) diversisternus (Silvestri, 1919) 57 Hondo, Hainan þMecistocephalus (Mecistocephalus) guildingii Newport, 1843 49 Central America –

Mecistocephalus (Mecistocephalus) itayai Takakuwa, 1939 49 Caroline Islands –

Mecistocephalus japonicus Meinert, 1886 63 Hondo, Taiwan þMecistocephalus (Mecistocephalus) lifuensis Pocock, 1899 51 New Caledonia, Loyalty Islands –

Mecistocephalus longiceps Lawrence, 1960 49 Madagascar –

Mecistocephalus (Formosocephalus) longichilatus Takakuwa, 1936 49 Taiwan –

Mecistocephalus microporus Haase, 1887 93–101 Philippines þMecistocephalus (Mecistocephalus) mikado Attems, 1928 49 Eastern Asia þMecistocephalus (Brachyptyx) mirandus Pocock, 1895 65 Hondo, Taiwan þMecistocephalus (Mecistocephalus) modestus (Silvestri, 1919) 49 Southeastern Asia þMecistocephalus (Mecistocephalus) multidentatus Takakuwa, 1936 49 Hondo, Taiwan þMecistocephalus (Mecistocephalus) nannocornis Chamberlin, 1920 45 Southeastern Asia þMecistocephalus (Mecistocephalus) punctifrons Newport, 1843* 49 Indian Peninsula þMecistocephalus (Mecistocephalus) spissus Wood, 1862 45 Hawaii þMecistocephalus (Dasyptyx) subgigas (Silvestri, 1919) 49 New Guinea þMecistocephalus (Pauroptyx) superior (Silvestri, 1919) 49 Indian Peninsula –

Mecistocephalus (Mecistocephalus) tahitiensis Wood, 1862 47 Southeastern Asia, Oceania þMecistocephalus (Mecistocephalus) takakuwai Verhoeff, 1934 59 Hondo, Taiwan þMecistocephalus sp. A 47 Marquesas Islands þMecistocephalus sp. B 51 Sulawesi þMecistocephalus sp. C 53 Hainan þNannarrup hoffmani Foddai, Bonato, Pereira & Minelli, 2003* 41 Unknown þProterotaiwanella sculptulata (Takakuwa, 1936) 49 Taiwan –

Proterotaiwanella tanabei Bonato, Foddai & Minelli, 2002* 45 Ryukyu Islands þSundarrup flavipes Attems, 1930* 41 Lesser Sunda þTakashimaia ramungula Miyosi, 1955* 45 Hondo þTygarrup anepipe Verhoeff, 1939 45 Mascarene Islands –

Tygarrup javanicus Attems, 1929 45 Southeastern Asia þTygarrup muminabadicus Titova, 1965 45 Himalayas þTygarrup takarazimensis Miyosi, 1957 45 Tokara Islands þTygarrup sp. A 43 Andaman Islands þ



maximum rescaled consistency index (CI), which was calcu-

lated on the trees obtained during the preceding run (PAUP*

command: Reweight); at each run, a search was performed

according to the same strategy as the first one.

A bootstrap analysis was performed to evaluate the

statistical support of the clades: subsequent to the applica-

tion of the character weights obtained by the iterative

procedure described above, 100 resamplings of characters

were performed, with an equal probability among charac-

ters; for each resampling, 100 replicates of the original

heuristic procedure were performed (PAUP* command:

Bootstrap NReps¼ 100 Wts¼ simple / AddSeq¼ random

NReps¼ 100 Hold¼ 10).

The evolution of the number of leg-bearing segments in

mecistocephalids was then analysed by optimizing this char-

acter (ch. 118 in Appendix 1) on the phylogenetic trees,

according to a parsimony assumption (PAUP* command:

Reconstruct 118). Both AccTran and DelTran character

optimization methods were performed.

Results and discussion

Phylogeny of Mecistocephalidae

The analysis performed on the 116 characters not refer-

ring to the number of segments, produced one most parsi-

monious tree [109.25 steps after successive weighting; CI

(excluding uninformative characters)¼ 0.55; retention

index (RI)¼ 0.85; Fig. 2], whereas the analysis performed

on all 118 characters produced three equally most parsimo-

nious trees [120.99 steps after successive weighting; CI

(excluding uninformative characters)¼ 0.57; RI¼ 0.85;

Fig. 3]. The same trees were obtained using both AccTran

and DelTran character optimization methods.

The trees differ only for minor details. The three trees

obtained from the second analysis differ in the position of

Mecistocephalus modestus, which emerges close to Mecisto-

cephalus longichilatus in all the trees. Apart from this, these

trees differ from the tree obtained from the first analysis in

the position ofMecistocephalus lifuensis, a species with fifty-

one leg-bearing segments, which emerges either within a

Mecistocephalus group with forty-nine segments or as sister

species of another species with fifty-one segments.

In all trees obtained, the scolopendromorph Cryptops

anomalans is basal to all geophilomorphs, following the

fixed outgroup Lithobius forficatus. The two representatives

of the adesmate geophilomorphs, i.e. Stigmatogaster gracilis

and Geophilus insculptus, branch together and opposite to

all mecistocephalids. This is in full agreement with the

currently recognized phyletic assessment of the major

groups of centipedes (Dohle, 1985; Borucki, 1996;

Edgecombe et al., 1999, 2000; Giribet et al., 1999, 2001;

Foddai & Minelli, 2000; Kraus, 2001; Edgecombe &

Giribet, 2002; Fig. 1).

The monophyletic condition of mecistocephalids is con-

firmed and strongly supported in all trees. The internal

topology of the group is nearly resolved and many, but

not all, nodes are supported by reliable synapomorphies

(see Appendix 3) and high bootstrap values.

For some characters both analyses suggest some evolu-

tionary trends. In particular, the head becomes more and

more elongate, as already hypothesized by Crabill (1970)

without the support of an adequate phylogenetic analysis.

The elongation involves the cephalic plate, buccae and

maxillary coxosterna (ch. 50), but not the clypeus, labrum

or mandibles. The ratio of the length to the width of the

cephalic plate is as low as 1.2–1.4 in some Arrup and Dicel-

lophilus, 1.4–1.5 in other close genera, 1.7 in Tygarrup,

Takashimaia, Krateraspis and the basal Mecistocephalus,

and reaches even higher values, up to 2.1, in some derived

species of Mecistocephalus.

A clear trend also marks the evolution of the forcipular

segment: the tergum becomes increasingly narrow (ch. 74),

so that the pleura are broadly visible from above; the tro-

chanteropraefemur becomes relatively longer and narrower

(chs 86 and 87) and its proximal tooth becomes evident

(ch. 88). A primitive condition is retained in Arrup and

other related genera, whereas the most derived condition

is seen in some Mecistocephalus.

A clear trend also appears in the evolution of the cerrus

(chs 84 and 85). Whereas the cerrus is completely lacking in

Arrup and other basal genera, a pair of lateral groups of

setae is present in Dicellophilus, in other related genera and

in some basal Mecistocephalus. In some derived Mecistoce-

phalus, two additional paramedian rows of setae are also

developed. A peculiar pattern is found in Mecistocephalus

punctifrons and a few other species, where the rows are

enlarged in bands coalescent with the lateral groups.

The position of the metameric pores, the direction of the

grooves starting from these pores and the relative elonga-

tion of the foraminal processes probably underwent parallel

evolution in different mecistocephalid clades (ch. 69).

Crabill (1964) speculated on a trend towards a lateral

displacement of the metameric pores, a lateral opening of

the groove and a reduction of the foraminal process, in

relation to the elongation of the whole head. Our phylo-

genetic analysis confirms this hypothesis in part only.

Taxonomic implications

On the basis of this cladistic analysis, as well as a critical

evaluation of current taxonomy, a revised phylogenetic sys-

tem of Mecistocephalidae is proposed (Table 2). Recent

taxonomic contributions on Proterotaiwanella and Arrupi-

nae (Bonato et al., 2002; Foddai et al., 2003) are included.

The taxonomic arrangement and nomenclature introduced

in those papers are adopted in the following discussion.

We recognize three subfamilies, fundamentally corres-

ponding to three main well-supported clades within the

mecistocephalids but only partially equivalent to the trad-

itional subfamilies. These three taxa are Arrupinae, Dicello-

philinae and Mecistocephalinae (see below).

Sundarrup Attems, 1930 is recognized as a junior synonym

ofAnarrup Chamberlin, 1920 (syn.n.). The original diagnoses



of these two nominal genera are overlapping and their type

species, Sundarrup flavipes Attems, 1930 and Anarrup

nesiotes Chamberlin, 1920, respectively, are sister species in

our phylogenetic analyses. Worth noticing is that Sundarrup

and Anarrup were previously placed in two different subfam-

ilies, Mecistocephalinae and Arrupinae, respectively.

The internal phylogeny of Mecistocephalus does not sup-

port the current partition of the genus into subgenera.

Lithobius forficatus

Cryptops anomalans

Stigmatogaster gracilis

Geophilus insculptus

Arrup pylorus

Arrup dentatus

Arrup holstiiNannarrup hoffmani

Agnostrup paucipes

Dicellophilus limatus

Dicellophilus anomalus

Dicellophilus latifrons

Dicellophilus carniolensis

Anarrup nesiotes

Sundarrup flavipes

Proterotaiwanella sculptulata

Tygarrup anepipe

Tygarrup javanicus

Tygarrup muminabadicus

Tygarrup takarazimensis

Krateraspis meinerti

Krateraspis sselivanovi

Takashimaia ramungula

Mecistocephalus M. spissus( )

Mecistocephalus benoiti

Mecistocephalus M. nannocornis( )

Mecistocephalus M. angusticeps( )

Mecistocephalus M. tahitiensis( )

Mecistocephalus F. longichilatus( )

Mecistocephalus M. modestus( )

Mecistocephalus M. conspicuus( )

Mecistocephalus M. lifuensis( )

Mecistocephalus M. guildingii( )

Mecistocephalus longiceps

Mecistocephalus M. itayai( )

Mecistocephalus M. diversisternus( )

Mecistocephalus M. takakuwai( )

Mecistocephalus japonicus

Mecistocephalus M. multidentatus

Mecistocephalus M. punctifrons( )

( )

Mecistocephalus M. mikado( )

Mecistocephalus P. superior( )

Mecistocephalus D. subgigas( )

Mecistocephalus microporus

Mecistocephalus B. mirandus( )

Mecistocephalus sp. A

Mecistocephalus sp. B

Mecistocephalus sp. C

Proterotaiwanella tanabei

Tygarrup sp. A

5251

50

49

42

4140

39

3837

36

35

34

31

30

29

2624

23

22

21

11

13

16

8

3

7

6

5

15

12

10

14

20

1918

17

9

2

1

4

2528

27

4846

43

47

6353

41

59

10058

98

8431

20

10

36

29

17

39

37

3577

92

63

44

86

98

90

61

79

82

74

77

98

84

41

100

3032

62

27

39

99

100

100

9623

26

3669

27

21

Fig. 2. The most parsimonious phylogenetic tree obtained from the cladistic analysis excluding characters related to the number of segments

[116 characters; 109.25 steps after successive weighting; consistency index (CI; excluding uninformative characters)¼ 0.55; retention index

(RI)¼ 0.85]. For each node, a conventional number (above) and the bootstrap value (below) are indicated. Species are entered with their

current names (Table 1).



Whereas Brachyptyx Chamberlin, 1920, Pauroptyx

Chamberlin, 1920, Dasyptyx Chamberlin, 1920, Ectoptyx

Chamberlin, 1920 and Formosocephalus Verhoeff, 1937

are probably monophyletic, the nominotypical subgenus

Mecistocephalus Newport, 1843 is clearly paraphyletic. The

present arrangement is thus not satisfactory, but we cannot

yet confidently suggest an alternative one. For the moment,

we propose to ignore the traditional subgenera within

Mecistocephalus. In the same vein, we have already proposed

to abandon Megethmus Cook, 1896, a nominal genus intro-

duced for Mecistocephalus microporus, which clearly falls

within the radiation ofMecistocephalus (Bonato et al., 2001).

Fusichila waipaheenas Chamberlin, 1953 and Megalacrus

obscuratus Attems, 1953 are provisionally maintained with


Cryptops anomalans


Geophilus insculptus

Arrup pylorus

Arrup dentatus

Arrup holstii

Nannarrup hoffmani

Agnostrup paucipes





Anarrup nesiotes

Sundarrup flavipes


Tygarrup anepipe

Tygarrup javanicus






Mecistocephalus (M.) spissus


Mecistocephalus (M.) nannocornis

Mecistocephalus (M.) angusticeps

Mecistocephalus (M.) tahitiensis

Mecistocephalus (F.) longichilatus

Mecistocephalus (M.) modestus

Mecistocephalus (M.) conspicuus

Mecistocephalus (M.) lifuensi

Mecistocephalus (M.) guildingii


Mecistocephalus (M.) itayai

Mecistocephalus (M.) diversisternus

Mecistocephalus (M.) takakuwai


Mecistocephalus (M.) multidentatus

Mecistocephalus (M.) punctifrons

Mecistocephalus (M.) mikado

Mecistocephalus (P.) superior

Mecistocephalus (D.) subgigas


Mecistocephalus (B.) mirandus





Tygarrup sp. A

5251

50

4948

47

46

45

4241

40

3938

37

29

26

23

21

11

13

12

1415

16

67

8

3

5

10

2019

18

28

34

31

30

27

25

24

22

17

9

4

2

1

35

44

6646

28

6034

17

56

48

9760

96

82

29

35

56

41

93

44

86

100

87

9892

84

8089

61

86

86

30

3940

64

47

29

11

41

49

99

71

59

28

43

100

100

100

33

21

A

Fig. 3. Most parsimonious phylogenetic trees (A, B, C) obtained from the cladistic analysis including characters related to the number of

segments [118 characters; 120.99 steps after successive weighting; consistency index (CI; excluding uninformative characters)¼ 0.57; retention

index (RI)¼ 0.85]. For each node, a conventional number (above) and the bootstrap value (below) are indicated. Species are entered with their

current names (Table 1).



their current names, although their taxonomic status needs

to be clarified. Both species were insufficiently described

to determine their actual relationships to other mecistoce-

phalids.

Mecistocephalidae Bollman, 1893

Synapomorphies. Anterior part of clypeus and buccae

areolate, posterior part virtually not areolate; internal

margin of each bucca thickened into a stilus; each labrum

side-piece divided into alae by a transverse thickened line;

sterna of anterior part of trunk provided with apodemes

and mid-longitudinal sulci.

Diagnosis. Body slightly depressed, uniformly wide in its

anterior three quarters but tapering backwards. Adult body

length from c. 2 to 14 cm. Colour from pale yellow to

red-brown, head and forcipular segment darker. Antennae

2–3� longer than head, distally attenuated. Antennal setae

increasing in density from basal article to tip of appendage.

Cephalic plate subrectangular (length to width ratio 1.2–2.1),

frontal line usually present. Clypeus subdivided into an

anterior areolate part and a posterior part which is

virtually not areolate (plagula/ae). Buccae areolate in the

anterior part only, internal margin thickened (stilus). Labrum

composed of a mid-piece and 2 side-pieces, each side-piece

divided by a transverse thickened line into an anterior and

posterior ala. Paralabial sclerites not recognizable. Mandible

only provided with a series of pectinate lamellae. First

lamella similar to other lamellae but smaller, last lamellae

rudimentary. Mandibular basal tooth conical. Coxal pro-

jection of first maxillae similar in shape and extension to the

corresponding telopodite, both of them being uniarticulate

and composed of a sclerotized base and a hyaline distal

part, without any additional lobe. Coxosternum of second

maxillae usually undivided, areolate in the median part and

always provided with a pair of metameric pores. Telopodite





Mecistocephalus (M.) lifuensis










Mecistocephalus (P.) superiorMecistocephalus (M.) subgigas


Mecistocephalus (B.) mirandusMecistocephalus sp. B


5251

50

4948

47

46

45

4241

40

3938

37

34

32

30

35

44

6646

28

6034

17

56

48

9760

96

8229

35

29

29

41

33

21

B





Mecistocephalus (M.) lifuensis










Mecistocephalus (P.) superior

Mecistocephalus (D.) subgigas


Mecistocephalus (B.) mirandus



5251

50

4948

47

46

45

4241

40

3938

37

34

35

44

6646

28

6034

17

56

48

9760

96

82

29

35

29

33

21

3041

3324

C

Fig. 3. Continued.

Table 2. Revised taxonomic system of Mecistocephalidae, based on the phylogenetic analysis presented here (see Taxonomic implications;

also Bonato et al., 2001, 2002; Foddai et al., 2003). The enigmatic Megalacrus and Fusichila, both marked with an asterisk, are provisionally

conserved as valid genera, but their taxonomic status requires further study.

Family Subfamily Genus Species

Mecistocephalidae Arrupinae Chamberlin, 1912 Agnostrup Foddai, Bonato, Pereira & Minelli, 2003 3

Bollman, 1893 Nannarrup Foddai, Bonato, Pereira & Minelli, 2003 1

Arrup Chamberlin, 1912 11

Partygarrupius Verhoeff, 1939 1

Dicellophilinae Cook, 1896 Proterotaiwanella Bonato, Foddai & Minelli, 2002 2

Anarrup Chamberlin, 1920 2

Dicellophilus Cook, 1896 4

Mecistocephalinae Bollman, 1893 Tygarrup Chamberlin, 1914 15

Krateraspis Lignau, 1929 2

Takashimaia Miyosi, 1955 1

Mecistocephalus Newport, 1843 130þ*Megalacrus Attems, 1953 1

*Fusichila Chamberlin, 1953 1



of second maxillae triarticulate, usually provided with a

reduced claw. Forcipular tergum narrower than cephalic

plate, partially covered by the latter and by tergum of first

leg-bearing segment. Forcipular pleura widely visible from

above, each provided with a dorsal setigerous ridge and

ending anteriorly in a pointed scapula. Forcipular coxoster-

num wider than cephalic plate, its antero-external parts

visible from above. A pair of tiny teeth on anterior margin

of forcipular coxosternum. No chitin-lines. Forcipular

telopodites rather large, clearly visible from above beyond

lateral margins of cephalic plate, usually also in front of

same. Forcipular tarsungulum relatively long. Forcipular

trochanteropraefemur with a distal, sometimes also a

proximal tooth; intermediate articles often with a tooth

each. Poison calyx elongate, usually reaching distal part of

trochanteropraefemur. Terga of leg-bearing trunk with 2

paramedian sulci, fading in most posterior segments. Sterna

of leg-bearing trunk with an internal apodema and a mid-

longitudinal sulcus, fading in posterior segments. Anterior

sterna with a posterior endosternal process, gradually

reduced in size in posterior segments. First pair of legs

shorter than following pairs. Tergum and sternum of last

leg-bearing segment rather elongate. Coxopleura covered

by 10s of circular pores. Telopodite of last legs of 6 articles,

thin (although sometimes slightly swollen in males)

and longer than telopodite of remaining legs. Praetarsus of

last leg extremely reduced. Posterior part of last sternum

and ventral internal margins of coxopleura covered by

dense pilosity. Gonopods biarticulate. Anal pores usually

present.

Nomenclature. Bollman (1893) introduced the name

Mecistocephalinae as a subfamily of Geophilidae to group

all the mecistocephalid species hitherto known. The name

was first used for a family by Verhoeff in 1908 (Verhoeff,

1902–1925).

Arrupinae Chamberlin, 1912

Type genus. Arrup Chamberlin, 1912.

Included genera. Arrup Chamberlin, 1912 (¼ Prolam-

nonyx Silvestri, 1919; Nodocephalus Attems, 1928) (eleven

species); Partygarrupius Verhoeff, 1939 (one species);

Agnostrup Foddai, Bonato, Pereira & Minelli, 2003 (three

species); Nannarrup Foddai, Bonato, Pereira & Minelli,

2003 (one species).

Remarks on monophyly. The clade composed by Arrup,

Agnostrup and Nannarrup is well supported by both reliable

synapomorphies and good bootstrap values. Partygarrupius

was not sufficiently known to be included in this analysis,

but some morphological features suggest it is most closely

related to the former genera than to other basal mecisto-

cephalids, as also suggested by a more restricted phylogenetic

analysis (Foddai et al., 2003).

Synapomorphies. Telopodites of second maxillae quite

short, not overreaching those of first maxillae.

Diagnosis. Body inconspicuously tapering backwards.

Leg-bearing trunk uniform in colour, without dark patches.

Cephalic plate only slightly longer than wide. Usually 2

clypeal plagulae divided by a mid-longitudinal stripe, not

covering more than posterior half of clypeus. Clypeal setae

a few to 10s, mainly placed in 2 lateral quite long areas.

Buccae without setae. Spiculum absent. Internal margin of

labral anterior ala reduced to a pointed end. Posterior alae

without longitudinal stripes. Posterior margin of labral side-

piece sinuous, not fringed. Coxosternum of first maxillae

either divided and nonareolate or undivided and areolate;

anterolateral corners virtually absent. Coxosternum of sec-

ond maxillae undivided or coxae connected by a membran-

ous isthmus. Groove from metameric pore and foraminal

process reaching postero-external corner of coxosternum.

Telopodites of second maxillae not overreaching those of

first maxillae. Forcipular tergum evidently wider than long,

without a mid-longitudinal sulcus. Cerrus absent. Forci-

pular trochanteropraefemur stout, with a distal tooth only.

Sternal mid-longitudinal sulci not furcate. Number of pairs

of legs 41. Sternum of last leg-bearing segment without a

pillowlike process.

Distribution. Eastern Asia from Hokkaido to Taiwan

(Partygarrupius, Agnostrup and Arrup), central Asia

(Arrup), California (Arrup). The true homeland of

Nannarrup, whose only species was described on specimens

collected in New York (U.S.A.), is unknown.

Nomenclature. Chamberlin (1912) introduced the name

Arrupidae for the single genus Arrup, later shifting the name

to subfamily level under Mecistocephalidae (Chamberlin,

1920a). Attems (1929) followed this arrangement, which

was later critically discussed only by Crabill (1964). The

delimitation proposed herewith is different from all

previous ones in the inclusion of Partygarrupius and the

exclusion of Anarrup.

Dicellophilinae Cook, 1896

Type genus. Dicellophilus Cook, 1896.

Included genera. Dicellophilus Cook, 1896 (four species);

Anarrup Chamberlin, 1920 (¼Sundarrup Attems, 1930)

(two species); Proterotaiwanella Bonato, Foddai & Minelli,

2002 (two species).

Remarks on monophyly. The clade comprising Dicello-

philus and Anarrup is well supported by both reliable syna-

pomorphies and good bootstrap values. Proterotaiwanella is

basal to this clade in all the trees and some reliable synapo-

morphies support this position, despite low bootstrap

values.



Synapomorphies. A spinous tubercle on tip of last pair

of legs.

Diagnosis. Body evidently tapering backwards. Leg-

bearing trunk uniform in colour, without dark patches.

Cephalic plate evidently longer than wide. Usually an entire

clypeal plagula, covering more than posterior half of

clypeus. Tens of clypeal setae. Spiculum absent. Internal

margin of labral anterior ala reduced to a pointed end;

posterior alae with notches or longitudinal stripes. Coxo-

sternum of first maxillae divided, nonareolate; anterolateral

corners usually absent. Coxosternum of second maxillae

either undivided or divided by a mid-longitudinal suture.

Groove from metameric pore and foraminal process reach-

ing either postero-external corner or lateral margin of

coxosternum. Telopodites of second maxillae usually well

developed and overreaching those of first maxillae, terminal

article often swollen and homogeneously covered with

setae. Forcipular tergum slightly wider than long, with a

mid-longitudinal sulcus. Cerrus composed of 2 lateral

groups of setae only. Forcipular trochanteropraefemur quite

stout, with a distal tooth only. Sternal mid-longitudinal

sulci not furcate. Number of pairs of legs 41, 43, 45 or

49. Sternum of last leg-bearing segment with a pillowlike

process. Legs of last pair with several additional short setae

and a tuberclelike praetarsus covered with tiny spines.

Distribution. Eastern Asia from Hondo to Malay Archi-

pelago (Dicellophilus, Proterotaiwanella, Anarrup), central

Europe (Dicellophilus), California (Dicellophilus).

Nomenclature. The name was originally proposed by

Cook (1896) as family Dicellophilidae, to include all the

mecistocephalid species described to that time. Hence,

equivalent to Mecistocephalidae Bollman, 1893. Cook did

not base the family group name on Mecistocephalus New-

port, 1843 because he designated as type of this genus the

geophilid Geophilus attenuatus Say, 1821, a nominal species

he regarded as a senior synonym of Geophilus ferrugineus

C.L. Koch, 1835, which is currently ascribed to Pachymer-

ium in family Geophilidae. As a consequence, Cook

regarded Mecistocephalus as not belonging to the family

under discussion and proposed three new generic names,

Dicellophilus, Lamnonyx and Megethmus, to include the

mecistocephalid species known to his date. This matter

was clarified by Crabill (1957). The name Dicellophilinae

is used here in a more restricted sense.

Mecistocephalinae Bollman, 1893

Type genus. Mecistocephalus Newport, 1843.

Included genera. Mecistocephalus Newport, 1843

(¼ LamnonyxCook, 1896;MegethmusCook, 1896. Including

also: Brachyptyx Chamberlin, 1920; Dasyptyx Chamberlin,

1920; Ectoptyx Chamberlin, 1920; Pauroptyx Chamberlin,

1920; Formosocephalus Verhoeff, 1937) (c. 130 species);

Tygarrup Chamberlin, 1914 (¼ Brahmaputrus Verhoeff,

1942) (fifteen species); Krateraspis Lignau, 1929 (two

species); Takashimaia Miyosi, 1955 (one species). Addition-

ally, Megalacrus Attems, 1953 (one species) and Fusichila

Chamberlin, 1953 (one species) are provisionally assigned to

this subfamily.

Remarks on monophyly. The clade composed by Krater-

aspis, Takashimaia and Mecistocephalus is well supported

by both reliable synapomorphies and good bootstrap

values. Tygarrup is basal to this clade in all our trees and

some reliable synapomorphies support this position, despite

low bootstrap values.

Synapomorphies. Clypeal setae limited to a short trans-

verse band; thickened transverse line of labrum side-piece

straight.

Diagnosis. Body evidently tapering backwards. Leg-

bearing trunk often provided with dark patches. Cephalic

plate evidently longer than wide. Clypeal plagula/ae cover-

ing more than posterior half of clypeus. Clypeal setae

usually a few, limited to a short transverse band and to

anterolateral corners. Posterior alae without longitudinal

stripes. Posterior margin of labrum sinuous. Coxosternum

of first maxillae divided, nonareolate. Coxosternum of sec-

ond maxillae undivided, medial part areolate; groove from

metameric pore and foraminal process reaching lateral mar-

gin. Telopodites of second maxillae well developed, over-

reaching those of first maxillae; terminal article usually

covered with setae, mainly on internal side, and bearing a

reduced claw. Forcipular tergum slightly wider than long,

with a mid-longitudinal sulcus. Forcipular trochanteroprae-

femur usually elongate, sometimes provided with a proxi-

mal tooth. Sternal mid-longitudinal sulci either furcate or

not. Number of pairs of legs 45, 47, 49, 51, 53, 57, 59, 63 or

65; in only one species, odd numbers between 93 and 101.

Sternum of last leg-bearing segment often with a pillowlike

process. Legs of last pair usually provided with an apical

spine.

Distribution. Mainly tropical regions, in particular

Southeast Asia, but also eastern Asia, Pacific islands,

Australia, Indian Peninsula, Africa and the Americas.

Nomenclature. Chamberlin (1920a) was the first to

distinguish two subfamilies within Mecistocephalidae, thus

reserving the name Mecistocephalinae for the nominotyp-

ical subfamily. Here we use Mecistocephalinae in a more

restricted sense.

Evolution of segment numbers

Optimizing the number of leg-bearing segments (ch. 118)

on the phylogenetic trees, we could estimate the most prob-

able number at each node, under a parsimony hypothesis

(Figs 4, 5). Analyses with both AccTran and DelTran



optimization methods were performed, but the DelTran

option generated some clearly unreliable hypotheses

(e.g. fifteen leg-bearing segments at the base of the

mecistocephalids; cf. e.g. Minelli et al., 2000) and thus

was not considered further.

Following AccTran optimization, the number of

leg-bearing segments of the common ancestor of mecisto-

cephalids was probably forty-one, which is also the

most common number in both Arrupinae and Dicello-

philinae. In Arrupinae, the ancestral number was strictly

conserved in all species, irrespective of the overall morpho-

logical diversity in this subfamily. In Dicellophilinae,

conversely, relevant changes occurred, in particular in

Dicellophilus (from forty-one to forty-three and, indepen-

dently, from forty-one to forty-five) and in Proterotaiwa-

nella (from forty-five to forty-nine). In Mecistocephalinae,

the basal number forty-five changed independently to

forty-three in Tygarrup and to fifty-three in Krateraspis.

Changes from forty-five to either forty-seven or forty-nine

happened at the base of Mecistocephalus, but the topology

is too weakly supported to recognize them unambiguously.

Some further changes from forty-nine to higher numbers

occurred within different derived groups of Mecisto-

cephalus.


Cryptops anomalans


Geophilus insculptusArrup pylorusArrup dentatus

Arrup holstii

Nannarrup hoffmani

Agnostrup paucipes

Dicellophilus limatusDicellophilus anomalusDicellophilus latifrons

Dicellophilus carniolensisAnarrup nesiotes

Anarrup flavipes


Tygarrup anepipe

Tygarrup javanicus



Krateraspis meinertiKrateraspis sselivanovi


Mecistocephalus spissus


Mecistocephalus nannocornis

Mecistocephalus angusticeps

Mecistocephalus tahitiensis

Mecistocephalus longichilatus

Mecistocephalus modestus

Mecistocephalus conspicuus

Mecistocephalus lifuensis

Mecistocephalus guildingiiMecistocephalus longicepsMecistocephalus itayai

Mecistocephalus diversisternus

Mecistocephalus takakuwai


Mecistocephalus multidentatus

Mecistocephalus punctifronsMecistocephalus mikado

Mecistocephalus superiorMecistocephalus subgigas


Mecistocephalus mirandus





Tygarrup sp. A

41

45

45

45

45

45

47

49

49

49

49

49

4949

49

4949

6359

5749

4949

49

49 49

49

49

49

49

53

49

51

6563

59

57

494949

51

4949

49

47474745

45

45

5345

4545

45

45

4549

4141

43

41

4541

4141

41

4143 – 4783 – 111

21

15

41

43

93 – 101

49

45

45

4545

4545

45

41

41

41

45

4141

4141

4141

?

4747

?

+4

+2

+2

+4

+ 2+ 2

+2

+2

+8

– 2

–4

+ 4

+2

+4

+8

+4

×2

?

Fig. 4. Evolution of the number of trunk segments in mecistocephalids, based on the AccTran optimization of the number of leg-bearing

segments (ch. 118) on the most parsimonious tree obtained excluding characters related to this number (Fig. 2). For each species, the number

of leg-bearing segments is reported. For each node, the hypothetical ancestral number is reported below the node. The number of segments

probably added or lost is indicated above the branches. Species names follow the revised taxonomy presented here.



Increases by two, four or eight segments most often

occurred in the evolution of mecistocephalids. Two-segment

increases most probably happened with the origin of

Dicellophilus carniolensis (from forty-one to forty-three),

with the differentiation of a basal group of Mecistocephalus

species (from forty-five to forty-seven) and in the radiation

of the most derived Mecistocephalus (for instance from

forty-nine to fifty-one). The analysis of characters not

related with segmentation suggested that this latter change

happened twice independently, whereas the analysis of all

characters suggested that it happened only once. Although

the apparent additions of more than two segments could be

due to the iterated additions of two segments, direct four-

segment increases most probably occurred with the origin of

Proterotaiwanella sculptulata (from forty-five to forty-nine),

in the differentiation ofDicellophilus limatus (from forty-one

Mecistocephalus angusticeps


Cryptops anomalans

Arrup pylorus

Arrup dentatus

Arrup holstii

Nannarrup hoffmani

Agnostrup paucipes





Anarrup nesiotes

Anarrup flavipes


Tygarrup anepipe

Tygarrup javanicus






Mecistocephalus spissus


Mecistocephalus nannocornis

Mecistocephalus tahitiensis





Mecistocephalus guildingii


Mecistocephalus itayai





Mecistocephalus punctifrons

Mecistocephalus mikado

Mecistocephalus superior

Mecistocephalus subgigas






Tygarrup sp. A

41

4141

41

?

45

41

41

41

45

4141

–4

+4

+2

+4

45

45

45

4545

+8

–2

45

4747

6359

57

49

49

49

+2

+8

+4

49

49

49

49

4949

4949

+4

+2

×2

41

45

45

45

45

45

47

49

49

49

49

?

+4

+2

+2

?

41

41

43

41

45

41

4141

41

41

43–47 Geophilus insculptus

83–111 Stigmatogaster gracilis

21

15

41

49

47

47

47

45

45

45

53

45

45

45

45

45

45

49

43

49

53

49

51

51

65

63

59

57

49

49

49

49

49

93–101 Mecistocephalus microporus

49

49

49

49

+2

51

A

Fig. 5. Evolution of the number of trunk segments in mecistocephalids, based on the AccTran optimization of the number of leg-bearing

segments (ch. 118) on the three most parsimonious trees (A, B, C) obtained including characters related to this number (Fig. 3). For each

species, the number of leg-bearing segments is reported. For each node, the hypothetical ancestral number is reported below the node. The

number of segments probably added or lost is indicated above the branches. Species names follow the revised taxonomy presented here.



to forty-five) and within the radiation of the most

derived Mecistocephalus (from forty-nine to fifty-three). In

the same way, eight-segment increases most probably

occurred at the origin of Krateraspis sselivanovi (from

forty-five to fifty-three) at least. Additions of six segments,

conversely, seem never to have occurred, as suggested by

both AccTran and DelTran optimization. The same seems

true for additions of more than eight segments, except for

the anomalous case of Mecistocephalus microporus (see

below). Transitions internal to the Mecistocephalus group

with fifty-seven to sixty-five leg-bearing segments are hard

to determine confidently, but additions of two, four and

eight segments may fully explain the observed values.

Evolutionary decreases of segment number seem to have

occurred only rarely. The best supported case is a two-segment

decrease at the origin of aTygarrup species (from forty-five to

forty-three segments). A transition from forty-five to forty-

one segments at the base of Dicellophilinae is more dubious.

A peculiar transition occurred with the origin of Mecis-

tocephalus microporus, characterized by a variable number

of ninety-three to 101 leg-bearing segments, from a group of

Mecistocephalus species with an invariant number of forty-

nine. Such a dramatic transition appears as a direct overall

duplication of the original number of segments (see also

Bonato et al., 2001).

As a result, the evolution of mecistocephalids followed a

general trend towards higher numbers of segments: the

addition of two, four or eight segments occurred often, as

well as one overall duplication of the total number of seg-

ments (or something like that). Conversely, a decrease in

segment number was a rare occurrence and involved low

numbers of segments only (probably, just two).

The very derived position of Mecistocephalus microporus

within Mecistocephalus and, by implication, within the

whole family, shows that the intraspecific variability in the

number of segments evolved at least twice in geophilo-

morph centipedes, i.e. in this species and in the common

ancestor of Adesmata. It is also possible that some addi-

tional, albeit less conspicuous, case of intraspecific variation

in segment number is present in other mecistocephalid spe-

cies, but was overlooked because of preconceived views on

segment stability in this group.

Biogeography and segment numbers

Most mecistocephalid species occur in tropical and sub-

tropical regions (Fig. 6). The group is widespread from the

Pacific islands, through southern Asia, to most of Africa

and also reaches tropical America. Some species, however,

live in temperate regions such as the Japanese Archipelago

and some restricted zones of Europe, the Asiatic mainland

and North America. Maximum morphological diversity is

recognizable in the Japanese area, more precisely in the

Hondo Archipelago. Here, the mecistocephalid fauna com-

prises some tens of species belonging to eight genera in all

three subfamilies and eight different numbers of segments

are represented. Less diverse but quite rich faunas are found

in Southeast Asia and the whole oriental region generally.

Several tens of species were described from these regions

and many other species still await description.

A detailed biogeographical analysis was outside the aims

of this work. Nevertheless, the geographical distribution of

the main mecistocephalid groups was revised and the occur-

rence of different numbers of segments throughout the

world was analysed (Fig. 6). Some biogeographical evidence

suggests evolutionary patterns broadly consistent with the

phylogenetic hypotheses obtained from the morphological

analysis, thus bringing further support to them.

In Arrupinae, Dicellophilinae and basal Mecistocephali-

nae (Tygarrup, Krateraspis and Takashimaia), most species

ranges are quite restricted and separate, suggesting a vicar-

iance pattern. Conversely, in the most derived group, i.e.

Mecistocephalus, the ranges are significantly wider and

often overlapping, suggesting dispersal patterns.

Dicellophilus is characterized by a relatively low morpho-

logical diversity (four species, very similar to each other),

contrasting with their diversity in segment number (three















Mecistocephalus superiorMecistocephalus subgigas

Mecistocephalus mirandusMecistocephalus sp. B


49

53


49

49

49

49

49

51

5165

63

59

57

49

4949

49

4949

49

49

49

49

4949

49

5759

49

49

4949

49

4949

51

63

×2

+4

+2

+8+2

+4

49

+2

B















Mecistocephalus superior

Mecistocephalus subgigas




49

53


49

49

49

4949

49

4949

4949

+4

×2

49

51

51

65

63

59

57

49

49

49

49

4949

49

49

49

49

+251

6359

57

4949

49

+4+2

+2

49

49

+8

C

Fig. 5. Continued.



different values). The genus is limited to temperate

regions and shows a strongly disjoint distribution, each

species living in a quite restricted range: D. carniolensis

occurs in central Europe (eastern Alps, Dinarids and

Carpathians), D. latifrons in Honshu, D. anomalus and

D. limatus in California. All these features indicate the

relic condition of this group and agree with the phylogeny

in suggesting a consistent biogeographical history: the

two Californian species are sister species and more

closely related to the Japanese species than to the European

one.

Anarrup and Dicellophilus were reliably resolved as sister

groups, but their contrasting distribution (tropical vs tem-

perate) is a bit puzzling and possibly relic. The phylogeny of

Tygarrup, although supported by low bootstrap values,

suggests a consistent evolutionary trend in geographical

colonization and ecological adaptation. The basal Tygarrup

species, i.e. T. takarazimensis and most probably the related

T. quelpartensis, live on some minor islands south of Hondo

(Tokara and Quelpart Islands, respectively), within the

range of the most basal mecistocephalid groups, very far

from the other Tygarrup species and north of all of them.

B

Mec 49Mec 51

Mec 51

Mec 51

Mec 47

Mec 57

Mec 59Mec 63Mec 65

Mec 47Mec 57

Mec 49

C

Mec 93–101Mec 45

Mec 45

Mec 47

Mec 45

Mec 53Mec 57

Arr 41

Arr 41Dic 43

Agn 41 Dic 41Dic 41Dic 45

Arr 41Par 41

A

Nan 41

Tyg 45

Tak 45

Pro 45

Ana 41

Kra 45Kra 53

Tyg 43

Tyg 45

Pro 49

Fig. 6. Distribution of the mecistocephalid genera, according to the revised taxonomic system proposed in this paper. Separate ranges are

drawn for different numbers of segments. The first three letters of the generic name are followed by the number of leg-bearing segments in the

species belonging to the same genus. A, Partygarrupius, Agnostrup, Nannarrup, Arrup and Dicellophilus; B, Proterotaiwanella, Anarrup,

Tygarrup, Krateraspis and Takashimaia; C, Mecistocephalus.



Both species are characterized by medium size (adults are

25–30mm long) and are evidently adapted to temperate

climates. Tygarrup muminabadicus is representative of a

first group of species which emerged from the basal Tygar-

rup. These species live on the mountains of south-central

Asia, from the Indochinese mountains to the Himalayas.

These species are apparently adapted to cold alpine climates

(up to 4000m a.s.l) and are characterized by larger size

(adults are often 40–50mm long). The most derived group

of Tygarrup species is composed of T. javanicus, T. anepipe,

an undescribed species with forty-three leg-bearing seg-

ments and most probably some other related species. All

these species live on tropical islands, usually at low altitude,

in particular on some large Indonesian islands (Java,

Sumatra, Halmahera) and on some small archipelagos in

the Indian Ocean (Andaman, Seychelles and Mascarene

Islands). They are all characterized by very small size

(adults at most 20mm long). Thus, the evolutionary history

of Tygarrup was probably characterized by a progressive

colonization of southern regions, from Japan to the south-

ern continental Asia, then through Indochina to the Great

Sunda Islands and finally to the Indian Ocean archipelagos

by sea dispersal.

Mecistocephalus occupies a very large geographical

range. Maximum diversity, in terms of species as well as

segment numbers, is in the Japanese region and Southeast

Asia. A gradual decrease can be observed along the tropical

regions both eastwards, through the Pacific islands, and

westwards, through the Indian Ocean to Africa and

America. Eastwards, different species with forty-seven,

forty-nine or fifty-one leg-bearing segments are found on

the Melanesian islands; fewer species with either forty-seven

or forty-nine segments are found on the archipelagos of the

central Pacific; on Clipperton Island, facing the American

coast, only one species with forty-nine segments is present.

Westwards, different species with forty-seven, forty-nine

and fifty-one segments (possibly also fifty-seven) inhabit

the Seychelles and Mascarene Islands; fewer species exist

in Africa and Madagascar, most of them with forty-nine

segments but also one species with forty-seven segments is

recorded; only one species (or a very small number of

closely related species) with forty-nine segments is found

in America.

Mecistocephalus species with fifty-one leg-bearing

segments occur within the overall range of those with

forty-nine segments, but with a discontinuous distribution,

mainly from Sulawesi through the Melanesian islands to the

Fiji Islands. There are some records also from the Seychelles

and the Middle East. This pattern agrees with the hypoth-

esis that Mecistocephalus species with fifty-one leg-bearing

segments originated from species with forty-nine segments,

but at different occasions and places, as suggested by our

cladistic analysiswhere segmentation characterswere excluded.

Mecistocephalus species with fifty-seven, fifty-nine, sixty-

three and sixty-five leg-bearing segments show quite coin-

cident distributions, well within the overall range of the

congeneric species with forty-nine segments: all four num-

bers were recorded from Hondo and all but fifty-seven from

Taiwan. These biogeographical elements support the mono-

phyly of this species group and its evolution from a Mecis-

tocephalus with forty-nine segments.

Mecistocephalus microporus, characterized by a high and

variable number of leg-bearing segments (ninety-three to

101), is only known from two major islands of the Philip-

pines (Bonato et al., 2001). The presence of this species on a

young archipelago, close to the centre of specific diversity of

the Mecistocephalus with forty-nine pairs of legs, confirms

its relatively recent origin from an ancestor with forty-nine

pairs of legs, as suggested by the phylogenetic analysis.

Segmentation: evolution and development

The phylogenetic analysis of mecistocephalids revealed

interesting trends and patterns in the evolution of the seg-

mental arrangement of these centipedes. Some of them have

been suggested by previous comparative studies (see Intro-

duction), others are new and somehow unexpected.

Evolutionary changes always involved an even number of

segments. In the evolution of the mecistocephalids, the

number of segments changed sixteen times at least. At

each event, the difference between the plesiomorphic and

the derived number was an even number, usually two, four

or eight. As a consequence, the number of leg-bearing seg-

ments remained invariantly odd. This confirms the well-

known rule, common to all the centipedes, according to

which only odd numbers of leg-bearing segments are pre-

sent in the adult (Minelli & Bortoletto, 1988; Arthur &

Farrow, 1999; Minelli et al., 2000). This rule suggests a

very strong developmental constraint. The minimum struc-

tural unit in both development and evolution is thus a pair

of contiguous segments rather than a single segment. This is

also suggested by a bulk of other evidence from previous

investigations on centipedes, as well as on other arthropods,

by means of comparative morphology and developmental

genetics, e.g. the invariably odd pairs of legs in all centi-

pedes, the diplosegments in millipedes and the pair-rule

genes in Drosophila.

Evolutionary changes usually implied an increase in the

number of segments. The evolutionary history of mecisto-

cephalids was characterized by a trend towards an increas-

ingly higher number of segments, from a putative original

number of forty-one leg-bearing segments to sixty-five and

in one exceptional case to 101. Number increases occurred

often, whereas only very few instances of decrease have

been detected. Moreover, additions sometimes involved

large sets of segments (up to eight, but also c. fifty segments

in the exceptional case of Mecistocephalus microporus),

whereas decreases were always by no more than four seg-

ments. The same trend is clear in the evolution of the

centipedes as a whole (Minelli et al., 2000): the ancestral

number of fifteen leg-bearing segments (conserved in scuti-

geromorphs, lithobiomorphs and craterostigmomorphs)

increased both in scolopendromorphs (probably to



twenty-one and then to twenty-three) and in geophilo-

morphs (to a basal unknown value). Within the geophilo-

morphs, the number increased further many times, up to

191, whereas it decreased less often and less conspicuously,

down to twenty-seven in a few Adesmata (Minelli et al.,

2000). The same general trend has been recognized in the

evolution of millipedes. The hypothetical ancestor of the

diplopods probably had seventeen pairs of legs and the

ancestor of the chilognathans c. thirty-six pairs (Enghoff,

1990; Enghoff et al., 1993). The extant millipedes, instead,

have from fourteen to more than 300 pairs. This trend,

common to most lineages of myriapods, evidently refutes

a previously fashionable hypothesis on the evolution of

segmented bodies: according to this hypothesis (known as

Williston’s rule), primitive multisegmented and little pat-

terned organisms evolved into shorter and more extensively

patterned forms (for a critical discussion, see Dohle, 1985;

Berto et al., 1997; Fusco & Minelli, 2000a; Minelli et al.,

2000).

Evolutionary changes involved a number of segments

apparently belonging to the geometrical series two, four,

eight. Whereas the addition of two, four and eight segments

surely happened many times during the evolutionary his-

tory of mecistocephalids, no case of the addition of six

segments is suggested by the phylogenetic analysis. This

rule is mirrored by the patterns recognizable in the fre-

quency distributions of the number of segments in other

centipedes. Comparing the species within each geophilo-

morph family, the most frequent numbers usually differ by

four, eight or sixteen segments (Minelli & Bortoletto, 1988).

Comparing individuals of the geophilomorph Himantarium

gabrielis, the most frequent numbers differ by sixteen seg-

ments (Minelli et al., 1984). Within Adesmata, the number

of additional segments found in females compared with

conspecific males are often two, four, eight or sixteen

(Minelli & Bortoletto, 1988; Minelli, 2000; Minelli et al.,

2000). A highly probable developmental constraint thus

emerges. It could be explained by admitting that at least

three runs of overall duplication of a small number of

primary trunk segments occur in the late stages of segmen-

tation. Under this hypothesis, a developmental fault produ-

cing one additional segment just before one of these runs

could determine, as a final outcome, an addition of two,

four or eight segments, depending on the stage affected by

the fault. This hypothesis is consistent with the recent model

of segmentation proposed by Minelli (2000, 2001) and

briefly summarized in the Introduction: the three runs of

duplication hypothesized here could be considered as the

last part of the hierarchical pattern of the secondary seg-

mentation of each primary segment.

Evolutionary changes most often occurred from values of

the segment number belonging to the arithmetic series forty-

one, forty-five, forty-nine. During the evolution of mecisto-

cephalids, the possible numbers of segments seem to have

followed two alternative behaviours. Lineages with forty-

one, forty-five or forty-nine leg-bearing segments under-

went abundant speciation as well as further changes in the

number of segments. Conversely, lineages with forty-three,

forty-seven or fifty-one leg-bearing segments showed a very

low evolvability, often featuring as evolutionary ‘dead

branches’, poorly able to change further and even to speci-

ate. This evidence is somehow congruent with the frequency

distribution of the number of leg-bearing segments in Ades-

mata, where the most frequent numbers are 39þ 4k, with k

an integer (Minelli, 2000). The morphogenetic processes

underlying this puzzling pattern, however, are still unclear.

One evolutionary change at least implied an overall

duplication of the total (or almost total) number of

segments. The origin of Mecistocephalus microporus was

evidently accompanied by a conspicuous increase in the

number of leg-bearing segments, to a number about double

that of the hypothetical ancestor (ninety-three to 101 vs

forty-nine). The fact that Mecistocephalus microporus is

most probably the only species of mecistocephalids that is

certainly variable in segment number could be enlightening.

We could hypothesize that an additional run of duplication,

involving all (or most) of the trunk segments, occurred

during the morphogenetic process of segmentation and

was accompanied by a loss of the precise morphogenetic

control of the definitive number of segments (Bonato et al.,

2001).

In conclusion, the phylogenetic analysis of mecistocepha-

lids contributed consistently and sometimes unexpectedly to

the evidence that the comparative re-examination of centi-

pede morphology is gathering, towards an understanding of

the evolution of the trunk structure of these animals. In

particular, the morphological patterns and constraints

shown by the present analysis are consistent with a model

of arthropod segmentation (Minelli, 2001), the predictions

of which now need to be tested experimentally.

Acknowledgements

We are sincerely grateful to numerous colleagues and

friends for the loan of specimens, and in particular to

J. Beccaloni (The Natural History Museum, London,

U.K.), P. Beron (Bulgarian Academy of Sciences, Sofia,

Bulgaria), H. W. Chang (National Sun Yat-Sen University,

Kaohsiung, Taiwan), M. Daccordi (formerly, Museo Civico

di Storia Naturale, Verona, Italy), L. Deharveng (Universite

Paul Sabatier, Toulouse, France), J. Dunlop (Museum fur

Naturkunde, Berlin, Germany), H. Enghoff (Zoologisk

Museum,Copenhagen,Denmark), C. E.Griswold (California

Academy of Science, San Francisco, California, U.S.A.),

R. L. Hoffman (Virginia Museum of Natural History,

Martinsville, Virginia, U.S.A.), K. Ishii (Dokkyo University

School of Medicine, Mibu, Tochigi, Japan), J. Ledford

(California Academy of Science, San Francisco, California,

U.S.A.), P. Lehtinen (University of Turku, Turku, Finland),

L. Leibensperger (Museum of Comparative Zoology,

Cambridge, Massachusetts, U.S.A.), J. Martens (Johannes

Gutenberg Universitat, Mainz, Germany), J. P. Mauries



(Museum National d’Histoire Naturelle, Paris, France),

G. B. Osella (formerly, Museo Civico di Storia Naturale,

Verona, Italy), N. Platnick (American Museum of Natural

History, New York, U.S.A.), R. Shelley (North Carolina

State Museum of Natural Sciences, Raleigh, North Carolina

U.S.A.), A. A. Schileyko (Zoological Museum, Moscow,

Russia), T. Tanabe (Tokushima Prefectural Museum,

Hachiman-cho, Tokushima, Japan) and N. Tsurusaki

(Tottori University, Tottori, Japan). This work was sup-

ported by grants from the Italian MURST to A. M. and

from the University of Padova to L. B., and by financial

help to D. F. from the Zoological Museum, Copenhagen

(through COBICE facilities) and The Natural History

Museum, London (through Sys-Resource facilities). We

are very grateful to Wallace Arthur, Greg Edgecombe,

Henrik Enghoff and John G. E. Lewis for their sugges-

tions on a previous draft of this paper, and to Enrico

Negrisolo for his comments on the cladistic analyses.

Three referees helped us to improve the paper.

References

Aguinaldo, A.M.A., Turbeville, J.M., Linford, L.S., Rivera, M.C.,

Garey, J.R., Raff, R.A. & Lake, J.A. (1997) Evidence for a clade

of nematodes, arthropods and other moulting animals. Nature,

387, 489–493.Archey, G. (1937) Revision of the Chilopoda of New Zealand.

Records of the Auckland Institute and Museum, 2, 71–100.Arthur, W. & Farrow, M. (1999) The pattern of variation in

centipede segment number as an example of developmental

constraint in evolution. Journal of Theoretical Biology, 200,

183–191.Attems, C.G. (1907) Javanische Myriopoden gesammelt von

Direktor K. Kraepelin im Jahre 1903. Mitteilungen aus dem

Naturhistorischen Museum in Hamburg, 24, 77–142.Attems, C.G. (1926) Vierter Unterstamm der Arthropoda:

Progoneata. Handbuch der Zoologie, Vol. 4 (ed. by W. Kukenthal

and K. Krumbach), pp. 7–238. De Gruyter, Berlin.Attems, C.G. (1928) Eine neue Gattung und eine neue Art der

Mecistocephalidae (Chilopoden). Zoologischer Anzeiger, 75,

115–120.Attems, C.G. (1929) Myriapoda I: Geophilomorpha. Das Tierreich,

Vol. 52. De Gruyter, Berlin.Attems, C.G. (1947) Neue Geophilomorpha des Wiener Museums.

Annalen des Naturhistorischen Museums in Wien, 55, 50–149.Attems, C.G. (1953) Myriopoden von Indochina. Expedition von

Dr C. Dawydoff (1938–39). Memoires du Museum National

d’Histoire Naturelle, Paris, Serie A, Zoologie, 5, 133–230.Balavoine, G. (1997) The early emergence of platyhelminths is

contradicted by the agreement between 18S rRNA and Hox

genes data. Comptes Rendus de l’Academie des Sciences, Sciences

de la Vie, 320, 83–94.Berto, D., Fusco, G. & Minelli, A. (1997) Segmental units and

shape control in Chilopoda. Entomologica Scandinavica Supple-

ment, 51, 61–70.Bollman, C.H. (1893) Classification of the Syngnatha. Bulletin of

the United States National Museum, 46, 163–167.Bonato, L., Foddai, D. & Minelli, A. (2001) Increase by

duplication and loss of invariance of segment number in the

centipede Mecistocephalus microporus Haase, 1887 (Chilopoda,

Geophilomorpha, Mecistocephalidae). Italian Journal of Zoology,

68, 345–352.Bonato, L., Foddai, D. &Minelli, A. (2002) A new mecistocephalid

centipede from Ryukyu Islands and a revisitation of ‘Taiwanella’

(Chilopoda: Geophilomorpha: Mecistocephalidae). Zootaxa, 86,

1–12.Bonato, L. & Minelli, A. (2002) Parental care in Dicellophilus

carniolensis (C. L. Koch, 1847): new behavioural evidence with

implications for the higher phylogeny of centipedes (Chilopoda).

Zoologischer Anzeiger, 241, 193–198.Borucki, H. (1996) Evolution und phylogenetisches System der

Chilopoda (Mandibulata, Tracheata). Verhandlungen des

Naturwissenschaftlichen Vereins in Hamburg, 35, 95–226.Brolemann, H.-W. (1909) A propos d’un systeme des Geophilo-

morphes. Archives de Zoologie Experimentale et Generale, Paris,

3, 303–340.Brolemann, H.-W. (1930) Elements d’une Faune des Myriapodes de

France: Chilopodes. Imprimerie Toulosaine, Toulouse.Brooks, D.R. & McLennan, D.A. (1991) Phylogeny, Ecology and

Behavior. Chicago University Press, Chicago.Budd, G. (1996) The morphology of Opabinia regalis and the

reconstruction of the arthropod stem group. Lethaia, 29, 1–14.Carroll, S., Grenier, J. & Weatherbee, S. (2000) From DNA to

Diversity. Blackwell Science, Oxford.Chamberlin, R.V. (1912) The Chilopoda of California III. Pomona

Journal of Entomology, 4, 651–672.Chamberlin, R.V. (1914) The Stanford Expedition to Brazil, 1911.

JohnC. Branner, Director. The Chilopoda of Brazil.Bulletin of the

Museum of Comparative Zoology at Harvard College, 58, 151–221.Chamberlin, R.V. (1920a) The Myriopoda of the Australian region.

Bulletin of the Museum of Comparative Zoology at Harvard College,

64, 1–269.Chamberlin, R.V. (1920b) On chilopods of the family Mecistoce-

phalidae. Canadian Entomologist, 52, 184–189.Chamberlin, R.V. (1920c) A new genus in the chilopod family

Mecistocephalidae. Psyche, 27, 144–146.Conway Morris, S. (1994) Why molecular biology needs palaeon-

tology. Development Supplement, 1994, 1–13.Cook, O.F. (1896) An arrangement of the Geophilidae, a family of

Chilopoda. Proceedings of the United States National Museum,

18, 63–75.Crabill, R.E. (1957) On the Newport chilopod genera. Journal of

the Washington Academy of Sciences, 47, 343–345.Crabill, R.E. (1959) Notes on Mecistocephalus in the Americas,

with a redescription of Mecistocephalus guildingii Newport

(Chilopoda: Geophilomorpha: Mecistocephalidae). Journal of

the Washington Academy of Sciences, 49, 188–192.Crabill, R.E. (1964) A revised interpretation of the primitive

centipede genus Arrup, with redescription of its type-species and

list of known species. Proceedings of the Biology Society of

Washington, 77, 161–170.Crabill, R.E. (1968)Abizarre case of sexual dimorphism in a centipede,

with consequent submergence of a genus (Chilopoda: Geophilo-

morpha: Mecistocephalidae). Entomological News, 79, 286–287.Crabill, R.E. (1969) Revisionary conspectus of Neogeophilidae

with thoughts on a phylogeny. Entomological News, 80, 38–43.Crabill, R.E. (1970) Concerning mecistocephalid morphology and

the true identity of the type-species of Mecistocephalus. Journal

of Natural History, 4, 231–237.Demange, J.-M. (1963) La segmentation dorsale des myriapodes

chilopodes au niveau de la zone des 7e et 8e segments. Comptes

Rendus de l’Academie des Sciences, Paris, 257, 514–518.Demange, J.-M. (1981) Les Mille-Pattes: Myriapodes. Societe

Nouvelle des Editions Boubee, Paris.



DeRobertis, E.M. (1997) The ancestry of segmentation. Nature,

387, 25–26.Dohle, W. (1985) Phylogenetic pathways in the Chilopoda.

Bijdragen tot de Dierkunde, 55, 55–66.Eason, E.H. (1964)Centipedes of the British Isles. F.Warne, London.Edgecombe, G.D. (2001) Revision of Paralamyctes (Chilopoda:

Lithobiomorpha: Henicopidae), with six new species from Eastern

Australia. Records of the Australian Museum, 55, 201–241.Edgecombe, G.D. & Giribet, G. (2002) Myriapod phylogeny and

the relationships of Chilopoda. Biodiversidad, Taxonomıa y

Biogeografia de Artropodos de Mexico: Hacia Una Sıntesis de Su

Conocimiento (ed. by J. Llorente Bousquets and J. J. Morrone),

pp. 143–168. Universidad Nacional Autonoma de Mexico,

Mexico D.F.Edgecombe, G.D., Giribet, G. & Wheeler, W.C. (1999) Filogenıa

de Chilopoda: combinando sequencias de los genes ribosomicos

18S y 28S y morfologıa. Boletın de la Sociedad Entomologica

Aragonesa, 26, 293–331.Edgecombe, G.D., Wilson, G.D.F., Colgan, D.J., Gray, M.R. &

Cassis, G. (2000) Arthropod cladistics: combined analysis of

histone H3 and U2 snRNA sequences and morphology.

Cladistics, 16, 155–203.Eernisse, D.J. (1997) Arthropod and annelid relationships

re-examined. Arthropod Relationships (ed. by R. A. Fortey and

R. H. Thomas), pp. 43–56. Chapman & Hall, London.Eernisse, D.J., Albert, J.S. & Anderson, F.E. (1992) Annelida and

Arthropoda are not sister taxa: a phylogenetic analysis of

spiralian metazoan morphology. Systematic Biology, 41, 305–330.Enghoff, H. (1990) The ground-plan of chilognathan millipedes

(external morphology). Proceedings of the 7th International

Congress of Myriapodology (ed. by A. Minelli), pp. 1–21. Brill,

Leiden.Enghoff, H., Dohle, W. & Blower, J.G. (1993) Anamorphosis in

millipedes (Diplopoda): the present state of knowledge with

some developmental and phylogenetic considerations. Zoologi-

cal Journal of the Linnean Society, 109, 103–234.Ernst, A. (1981) Die Ultrastruktur der Sinneshaare auf den

Antennen von Geophilus longicornis Leach (Myriapoda, Chilo-

poda). III. Die Sensilla brachyconica. Zoologische Jahrbucher,

Abteilung fur Anatomie, 106, 375–399.Ernst, A. (1983) Die Ultrastruktur der Sinneshaare auf den

Antennen von Geophilus longicornis Leach (Myriapoda, Chilo-

poda). IV. Die Sensilla microtrichoidea. Zoologische Jahrbucher,

Abteilung fur Anatomie, 109, 521–546.Ernst, A. (1995) Die Ultrastruktur der Sensilla coeloconica auf den

Maxillipeden des Chilopoden Geophilus longicornis Leach.

Verhandlungen der Deutschen Zoologischen Gesellschaft, 88, 160.Ernst, A. (1997) Sensilla microtrichoidea – mutmaßliche ‘Stel-

lungsrezeptoren’ an der Basis der Antennenglieder des Chilopo-

den Geophilus longicornis Leach. Verhandlungen der Deutschen

Zoologischen Gesellschaft, 90, 274.Ernst, A. (2000) Structure and function of different cuticular

sensilla in the centipede Geophilus longicornis Leach. Fragmenta

Faunistica, Warszawa, 43(Suppl.), 113–129.Farris, J.S. (1969) A successive approximation approach to

character weighting. Systematic Zoology, 18, 374–385.Foddai, D. (1998) Phylogenetic relationships within geophilo-

morph centipedes based on morphological characters: a

preliminary report. Memorie del Museo Civico di Storia Naturale

di Verona, 13, 67–68.Foddai, D., Bonato, L., Pereira, L.A. & Minelli, A. (2003)

Phylogeny and systematics of the Arrupinae (Chilopoda

Geophilomorpha Mecistocephalidae) with the description of a

new dwarfed species. Journal of Natural History, 37, 1247–1267.

Foddai, D. & Minelli, A. (2000) Phylogeny of geophilomorph

centipedes: old wisdom and new insights from morphology.

Fragmenta Faunistica, Warszawa, 43(Suppl.), 61–71.Fuller, H. (1960) Uber die Chiasmen des Tracheensystems der

Geophilomorphen. Zoologischer Anzeiger, 165, 289–297.Fusco, G. & Minelli, A. (2000a) Measuring morphological

complexity of segmented animals: centipedes as model systems.

Journal of Evolutionary Biology, 13, 38–46.Fusco, G. & Minelli, A. (2000b) Developmental stability in

geophilomorph centipedes. Fragmenta Faunistica, Warszawa,

43(Suppl.), 73–82.Gerhart, J.C. & Kirschner, M.W. (1997) Cells, Embryos and

Evolution. Blackwell Science, Oxford.Giribet, G., Carranza, S., Riutort, M., Baguna, J. & Ribera, C.

(1999) Internal phylogeny of the Chilopoda (Myriapoda,

Arthropoda) using complete 18S rDNA and partial 28S rDNA

sequences. Philosophical Transactions of the Royal Society of

London B, 354, 215–222.Giribet, G., Distel, D.L., Polz,M., Sterrer,W. &Wheeler, W.C. (2000)

Triploblastic relationshipswith emphasis on the acoelomates and the

position of Gnathostomulida, Cycliophora, Plathelminthes, and

Chaetognatha: a combined approach of 18S rDNA sequences and

morphology. Systematic Biology, 49, 539–562.Giribet, G., Edgecombe, G.D. & Wheeler, W.C. (2001) Arthropod

phylogeny based on eight molecular loci and morphology.

Nature, 413, 157–161.Haase, E. (1887) Die Indisch-Australischen Myriopoden: 1

Chilopoden. Abhandlungen und Berichte des Koniglichen Zoolo-

gischen und Anthropologisch-Ethnographischen Museums in

Dresden, 1(5), 1–118.Hall, B.K. (1998) Evolutionary Developmental Biology, 2nd edn.

Chapman & Hall, London.Hoffman, R.L. (1982) Chilopoda. Synopsis and Classification

of Living Organisms (ed. by S. P. Parker), pp. 681–688.

McGraw-Hill, New York.Holland, L.Z., Kene, M., Williams, N.A. & Holland, N.D. (1997)

Sequence and embryonic expression of the amphioxus engrailed

gene (AmphiEn): the metameric pattern of transcription

resembles that of its segment-polarity homolog in Drosophila.

Development, 124, 1723–1732.Hopkin, S.P. & Anger, H.S. (1992) On the structure and function

of the glue-secreting glands of Henia vesuviana (Newport, 1845)

(Chilopoda: Geophilomorpha). Berichte des Naturwissenschaftlich-

Medizinischen Vereins in Innsbruck, 10(Suppl.), 71–79.Hopkin, S.P. & Gaywood, M.J. (1987) Encounters between the

geophilid centipede Henia (Chaetechelyne) vesuviana (Newport)

and the Devil’s Coach Horse beetle Staphylinus olens (Mueller).

Bulletin of the British Myriapod Group, 4, 22–26.Hopkin, S.P., Gaywood,M.J., Vincent, J.F.V. &MayesHarris, E.L.V.

(1990) Defensive secretion of proteinaceous glues by

Henia (Chaetechelyne) vesuviana Newport (Chilopoda, Geophi-

lomorpha). Proceedings of the 7th International Congress of

Myriapodology (ed. byA.Minelli), pp. 175–181. J. E. Brill, Leiden.Kimmel, C.B. (1996) Was Urbilateria segmented? Trends in

Genetics, 12, 329–331.Kraus, O. (2001) ‘Myriapoda’ and the ancestry of the Hexapoda.

Annales de la Societe Entomologique de France, 37, 105–127.Lawrence, P.A. (1992) The Making of a Fly. Blackwell Science,

Oxford.Lewis, J.G.E. (1981) The Biology of Centipedes. Cambridge

University Press, Cambridge.Lewis, J.G.E. (1991) Scolopendromorph and geophilomorph

centipedes from the Krakatau Islands and adjacent regions,

Indonesia. Memories of the Museum of Victoria, 52, 337–353.



Lewis, J.G.E. (2000) Centipede antennal characters in taxonomy

with particular reference to scolopendromorphs and antennal

development in pleurostigmomorphs (Myriapoda, Chilopoda).

Fragmenta Faunistica, Warszawa, 43(Suppl.), 87–96.Lewis, J.G.E. & Rundle, A.J. (1988) Tygarrup javanicus (Attems) a

geophilomorph centipede new to the British Isles. Bulletin of the

British Myriapod Group, 5, 3–5.Littlewood, P.M.H. (1983) Fine structure and function of the coxal

glands of Lithobiomorpha centipedes: Lithobius forficatus and

L. crassipes (Chilopoda: Lithobiidae). Journal of Morphology,

177, 157–179.Manton, S.M. (1965) The evolution of arthropodan locomotory

mechanisms. Part 8. Functional requirements and body design in

Chilopoda, together with a comparative account of their skeleto-

muscular system and appendix on a comparison between

burrowing forces of annelids and chilopods and its bearing

upon the evolution of the arthropodan haemocoel. Journal of the

Linnean Society (Zoology), 46, 251–484.Maynard Smith, J. (1960) Continuous, quantised and modal

variation. Proceedings of the Royal Society of London B, 152,

397–409.Miller, J.S. & Wenzel, J.W. (1995) Ecological characters and

phylogeny. Annual Review of Entomology, 40, 389–415.Minelli, A. (1985) Post-embryonic development and the phylogeny

of geophilomorph centipedes (Chilopoda). Bijdragen tot de

Dierkunde, 55, 143–148.Minelli, A. (1998) Segmented animals: origins, relationships, and

functions. Italian Journal of Zoology, 65, 1–4.Minelli, A. (2000) Holomeric vs. meromeric segmentation: a tail of

centipedes, leeches, and rhombomeres. Evolution and Develop-

ment, 2, 35–48.Minelli, A. (2001) A three-phase model of arthropod segmentation.

Development, Genes and Evolution, 211, 509–521.Minelli, A. & Bortoletto, S. (1988) Myriapod metamerism and

arthropod segmentation. Biological Journal of the Linnean

Society, 33, 323–343.Minelli, A., Foddai, D., Pereira, L.A. & Lewis, J.G.E. (2000) The

evolution of segmentation of centipede trunk and appendages.

Journal of Zoological Systematics and Evolutionary Research, 38,

103–117.Minelli, A., Pasqual, C. & Etonti, G. (1984) I Chilopodi

Geofilomorfi del genereHimantarium C.L. Koch con particolare

riferimento alle popolazioni italiane. Lavori della Societa

Veneziana di Scienze Naturali, 9, 73–84.Miyosi, Y. (1957) Beitrage zur Kenntnis japanischer Myriopoden.

20. Aufsatz. Uber eine neue Gattung von Diplopoda, eine neue

Art und eine neue Unterart von Chilopoda. Zoological

Magazine, Tokyo, 66, 264–268.Pereira, L.A. (1999) Un nouveau cas de dimorphisme sexuel chez

les Schendylidae: Schendylops virgingordae (Crabill, 1960),

espece halophile nouvelle pour la Martinique (Myriapoda,

Chilopoda, Geophilomorpha). Zoosystema, 21, 525–533.Pereira, L.A., Foddai, D. & Minelli, A. (2002) A new Brazilian

schendylid centipede (Chilopoda: Geophilomorpha)

with unusually structured antennae. Zoologischer Anzeiger, 241,

57–66.Pocock, R.I. (1891) Viaggio di Leonardo Fea in Birmania e regioni

vicine: 31. On the Myriapoda of Burma, part II. Report upon

the Chilopoda collected by Sig. L. Fea and Mr. E.W. Oates.

Annali del Museo Civico di Storia Naturale di Genova, 10,

401–432.Prendini, L. (2001) Species or supraspecific taxa as terminals in

cladistic analysis? Groundplans versus exemplars revisited.

Systematic Biology, 50, 290–300.

Prunesco, C. (1967) Le systeme genital femelle de l’ordre

Geophilomorpha. Revue Roumaine de Biologie, Serie de Zoolo-

gie, 12, 251–256.Regier, J.C. & Shultz, J.W. (2001) A phylogenetic analysis of

Myriapoda (Arthropoda) using two nuclear protein-encoding

genes. Zoological Journal of the Linnean Society, 132, 469–486.Ribaut, H. (1914) Chilopoda. Voyage de Ch. Alluaud et R. Jeannel

en Afrique Orientale (1911–12). Resultats Scientifiques.

Myriapoda, pp. 1–35. A. Shulz, Paris.Rosenberg, J. (1982) Coxal organs in Geophilomorpha (Chilo-

poda): organization and fine structure of the transporting

epithelium. Zoomorphology, 100, 107–120.Rosenberg, J. (1983a) Coxal organs in Lithobius forficatus

(Myriapoda, Chilopoda). Fine structural investigation with

special reference to the transport epithelium. Cell and Tissue

Research, 230, 421–430.Rosenberg, J. (1983b) Coxal organs in Scolopendromorpha

(Chilopoda): topography, organization, fine structure and

signification in centipedes. Zoologische Jahrbucher, Abteilung

fur Anatomie, 110, 383–393.Rosenberg, J. & Bajorat, K.H. (1984) Einfluß der Coxalorgane bei

Lithobius forficatus L. (Chilopoda) auf die Sorption von

Wasserdampf. Zoologische Jahrbucher, Abteilung fur Physiolo-

gie, 88, 337–344.Shear, W.A. & Bonamo, P.M. (1988) Devonobiomorpha, a new

order of centipeds (Chilopoda) from the Middle Devonian of

Gilboa, New York State, USA, and the phylogeny of centiped

orders. American Museum Novitates, 2927, 1–30.Shinohara, K. (1999) Chilopoda. Pictorial Keys to Soil Animals of

Japan (ed. by J. Aoki), pp. 693–709. Tokai University Press,

Tokyo.Shultz, J.W. & Regier, J.C. (1997) Progress towards a molecular

phylogeny of the centipede orders (Chilopoda). Entomologica

Scandinavica Supplement, 51, 25–32.Silvestri, F. (1895) I Chilopodi e i Diplopodi di Sumatra e delle

isole Nias, Engano e Mentavei. Annali del Museo Civico di Storia

Naturale di Genova, 34, 707–760.Silvestri, F. (1919) Contributions to a knowledge of the Chilopoda

Geophilomorpha of India. Records of the Indian Museum, 16,

45–107.Swofford, D.L. (2002) PAUP* Phylogenetic Analysis Using Parsi-

mony (*and Other Methods) Beta V 4 0. Sinauer Associates,

Sunderland, Massachusetts.Takakuwa, Y. (1934) Japanese Mecistocephalidae II. Shokubutsu-

Oyobi-Dobutsu, 2, 878–884.Takakuwa, Y. (1936) Eine neue interessante Mecistocephalus-Art

aus Formosa. Transactions of the Natural History Society of

Formosa, 26, 215–216.Takakuwa, Y. (1937) Uber die Mecistocephalus-Arten mit 65 und

63 Beinpaaren. Transactions of the Natural History Society of

Formosa, 27, 49–51.Takakuwa, Y. (1938) Uber eine weitere 45 Beinpaare neue

Mecistocephalus-Art aus Japan. Transactions of the Natural

History Society of Formosa, 28, 281–283.Takakuwa, Y. (1940) Geophilomorpha. Fauna Nipponica, Vol. 9.

Sanseido, Tokyo.Takakuwa, Y. (1942) Myriapoda of the Micronesia. Kagaku-

Nanyo, 5, 14–44.Takashima, H. & Shinohara, K. (1952) The centipede-fauna of the

Tokyo District. Acta Arachnologica, 13, 3–17.Titova, L.P. (1965) A new chilopod (Tygarrup muminabadicus

Titova sp. n., Mecistocephalidae, Chilopoda) from South

Tajikistan. Zoologicheskii Zhurnal SSSR, 44, 871–876.

(in Russian)



Titova, L.P. (1975) Geophilids of the family Mecistocephalidae

(Chilopoda) in the fauna of the USSR. Zoologicheskii Zhurnal

SSSR, 54, 39–48. (in Russian)Titova, L.P. (1983) Two new Tygarrup Chamb. (Chilopoda,

Geophilida, Mecistocephalidae) from Indochina. Annalen des

Naturhistorischen Museums in Wien, 85B, 147–156.Turcato, A., Fusco, G. & Minelli, A. (1995) The sternal pore areas

of geophilomorph centipedes (Chilopoda: Geophilomorpha).

Zoological Journal of the Linnean Society, 115, 185–209.Valentine, J.W. (1994) Late Precambrian bilaterians: grades and

clades. Proceedings of the National Academy of Sciences, 92,

6751–6757.Valentine, J.W. (1995) Late Precambrian bilaterians: grades and

clades. Tempo and Mode in Evolution: Genetics and Paleontology

50 Years after Simpson (ed. by W. M. Fitch and F. J. Ayala), pp.

87–107. National Academy Press, Washington.Verhoeff, K.W. (1902–1925) Chilopoda. Bronn’s Klassen und

Ordnungen des Tierreiches, Vol. 5. C.F. Winter, Leipzig.Verhoeff, K.W. (1934) Beitrage zur Systematik und Geographie der

Chilopoden. Zoologische Jahrbucher, Abteilung fur Systematik,

66, 1–112.

Verhoeff, K.W. (1937) Chilopoden aus Malacca, nach den

Objecten des Raffles Museum in Singapore. Bulletin of the

Raffles Museum, 13, 198–270.Verhoeff, K.W. (1939) Chilopoden der Insel Mauritius. Zoolo-

gische Jahrbucher, Abteilung fur Systematik, 72, 71–98.Verhoeff, K.W. (1942) Chilopoden aus innerasiatischen Hochge-

birgen. Zoologischer Anzeiger, 137, 35–52.Wang, Y.M. (1951) The Myriopoda of Philippine Islands. Serica,

1, i–vi, 1–80.Wolpert,L. (1998)PrinciplesofDevelopment.CurrentBiology,London.Wurmli, M. (1972) Chilopoda von Sumba und Flores. II.

Geophilomorpha, Lithobiomorpha, Scutigeromorpha. Verhand-

lungen der Naturforschenden Gesellschaft in Basel, 82, 205–214.Yeates, D.K. (1995) Groundplans and exemplars: paths to the tree

of life. Cladistics, 11, 343–357.

Accepted 22 January 2003

Appendix 1. Characters used in the cladistic analysis

Characters are numbered in anteroposterior anatomical

order. Plesiomorphic states are coded as ‘0’. Apomorphic

states are usually coded as ‘1’, ‘2’, ‘3’, ‘4’, in arbitrary order,

but letters from ‘a’ to ‘o’ are used for ch. 118. Anatomical

terminology basically follows Crabill (1959, 1964, 1970).

Colour

1. Leg-bearing trunk, dark patches on the ground colour:

(0) absent; (1) present, in some individuals at least.

In mecistocephalids, the extent of dark patches on the

body surface is largely variable within a species. In some

species, however, all individuals are invariably uniform,

whereas in other species most individuals are covered with

dark patches. Similar patches also occur in some Adesmata,

but are generally absent in lithobiomorphs and scolopen-

dromorphs.

Antenna

2. Antenna, number of articles: (0) more than 14, often

variable within a species; (1) 14, invariable within a

species.

In all geophilomorphs, the antenna is invariably com-

posed of fourteen articles, without any difference related

to age or sex (Eason, 1964; Lewis, 1981). However, abnor-

mal antennae composed of fewer than fourteen articles are

quite frequent, even within mecistocephalids; this anomaly

is often unilateral (Minelli et al., 2000). In both lithobio-

morphs and scolopendromorphs, the number of antennal

articles is usually larger than fourteen (but only fourteen in

some adults of the lithobiomorph Anopsobius neozelanicus

Silvestri, 1909; see Archey, 1937). In these groups, the num-

ber may be different in individuals of the same population

and between the left and the right antennae of the same

specimen (Eason, 1964; Lewis, 1981; Minelli et al., 2000).

3. Antenna, number of articles: (0) increasing during

postembryonic development; (1) not increasing during

postembryonic development.

In all geophilomorphs, the number of antennal articles

does not increase during postembryonic life. This is also the

case in most scolopendromorphs, in which only minor

changes may occur (Minelli et al., 2000). In lithobiomorphs,

conversely, the number of articles increases with growth.

4. Antenna, intermediate articles, ratio of length to width:

(1) 1.4–3.0; (2) 0.7–1.3.

For geophilomorphs, the antennal article VII was chosen

as representative of the intermediate articles of the append-

age. The ratio of maximum length to maximum width of this

article was taken as a measure of the degree of elongation of

the intermediate part of the antenna. Within mecistocepha-

lids, ratios seem to aggregate into two main intervals, as

distinguished here. This ratio may be affected by age and sex

and is slightly variable among conspecific specimens; thus,

for each species, the ratio has been estimated as far as

possible from a series of adult females. We did not apply

this character to lithobiomorphs and scolopendromorphs

because antennal article VII in these groups is not strictly

homologous to article VII in geophilomorphs. More gener-

ally, individual elements in the series of numerous and

variable antennal articles of lithobiomorphs and



scolopendromorphs are difficult to homologize with those

of the series of invariably fourteen articles in geophilo-

morphs. The developmental mechanisms determining the

segmentation of centipede antennae are still unknown

(Lewis, 2000; Minelli et al., 2000).

5. Antenna, ordinary setae, size and density: (0) not

changing along the antennal axis; (1) changing gradually

along the antennal axis; (2) changing abruptly along the

antennal axis.

In geophilomorphs, the size and density of the antennal

setae change gradually along the appendage. However, in a

few species of Adesmata the distribution of setae on the

antennae is sexually dimorphic (Pereira, 1999; Pereira et al.,

2002; we have also observed a case in an undescribed species

of Mecistocephalus from New Guinea). In scolopendro-

morphs, the change is generally more abrupt than in geo-

philomorphs (Lewis, 2000). In lithobiomorphs, conversely,

the size and density of the setae are virtually homogeneous

along the antennae.

6. Antennae and last pair of legs, rows of spinelike sensilla:

(0) absent; (1) present.

The spinelike sensilla [‘sensilla microtrichoidea’ of Ernst

(1983, 1997, 2000); ‘sensilla for telescopic movements’ of

Foddai & Minelli (2000)] are conical processes, similar in

shape to the setae but relatively shorter, a few micrometres

long, articulated at the base with the cuticular surface. They

are distributed in rows (up to six sensilla in each row), close

to the arthrodial membranes between sclerites. These sen-

silla are supposed to be mechanoreceptors, involved in con-

trolling the relative position of articles of appendages and

segments (Ernst, 2000; Foddai & Minelli, 2000). In geophi-

lomorphs, rows of spinelike sensilla are always present both

on the antennal articles (in three series: dorsal, ventro-

internal and ventro-external) and on the articles of the telopo-

dites of the last pair of legs (in two series: internal and

external). These sensilla are present in all individuals, but

the number of sensilla in each row increases during growth

and shows some interindividual variability. The same pat-

tern of rows of sensilla is present in some scolopendromorphs,

whereas it is not known in lithobiomorphs. In the latter,

however, groups of spinelike sensilla, not precisely aligned,

are present on some antennal articles, at least inLithobius and

Henicopidae (Edgecombe, 2001).

7. Antenna, buttonlike sensilla: (0) present; (1) virtually

absent.

The buttonlike sensilla [‘sensilla coeloconica’ of Ernst

(1995, 2000)] are stout subconical processes, very few micro-

metres long and wide, ringed by a shallow relief. Their

sensorial function is not known. Most mecistocephalid

species are characterized by buttonlike sensilla on the anten-

nae. It is not clear whether buttonlike sensilla, or at least

different sensilla in a homologous position, occur on the

antennae of lithobiomorphs and scolopendromorphs

(Lewis, 2000). However, preliminary observations suggest

that a few sensilla of this kind also occur in Lithobius and

Cryptops. The presence of these sensilla in mecistocephalids

was largely ignored in the taxonomic literature: only a few

authors have reported their occurrence, but they often mis-

interpreted them as either glandular pores (Verhoeff, 1934,

1939; Wurmli, 1972) or punctuate depressions (Haase, 1887;

Silvestri, 1919).

8. Antenna, spearlike sensilla: (0) absent; (1) present.

The spearlike sensilla [Fig. 7; a type of ‘sensilla brachy-

conica’ of Ernst (1981, 2000)] are thin processes, usually

5–10 mm long, bearing a crown of stout processes at mid-

length which are coalescent to different degrees. They were

supposed to be thermo- and hygroreceptors (Ernst, 1981,

2000). Within mecistocephalids, a few spearlike sensilla are

usually found on the antennal tip and their presence is not

affected by either age or sex. In some species, spearlike

sensilla are apparently lacking in all individuals. Similar

sensilla are also known in many Adesmata, either limited

to the antennal tip or also occurring on other antennal

articles. Conversely, they apparently never occur in litho-

biomorphs and scolopendromorphs. Interestingly, how-

ever, apparently homologous elements are present on the

antennal tips of craterostigmomorphs, a group probably

basal to all Epimorpha (Fig. 1). Furthermore, preliminary

investigations by light and electron microscopy revealed

that the antennal tip of many lithobiomorphs bears a

group of sensilla, apparently different from the spearlike

sensilla but possibly homologous in position (personal

observations).

Cephalic plate

9. Cephalic plate, lateral margins: (0) parallel; (1) slightly

convergent backwards; (2) strongly convergent backwards.

In some mecistocephalid species, the lateral margins of

the cephalic plate are slightly and uniformly convergent

backwards (Fig. 8A), in other species the margins are sub-

parallel at head mid-length (Fig. 8D) and in a few other

species they are strongly convergent backwards (Fig. 8C).

ab

c

Fig. 7. Spearlike sensilla (a), ordinary setae (b) and clublike

sensilla (c) on the antennal tip in a representative mecistocephalid

(Tygarrup muminabadicus, /, subadult, Kashmir) (simplified

drawing after a scanning electron micrograph). Scale¼ 0.01mm.



The lateral margins are generally subparallel in all other

geophilomorphs, more diverse in the other orders.

10. Cephalic plate, posterior margin: (0) straight;

(1) rounded.

In mecistocephalids, the posterior margin of the cephalic

plate is usually straight or truncate (Fig. 8D), but curved in

a few species (Fig. 8A). A straight posterior margin also

characterizes the head of most Adesmata, lithobiomorphs

and scolopendromorphs.

11. Cephalic plate, frontal line: (0) present; (1) absent.

A transverse frontal line occurs on the cephalic plate of

most geophilomorphs (Eason, 1964; Foddai & Minelli, 2000;

Fig. 8). A probably homologous line is present in lithobio-

morphs (see, e.g. Eason, 1964; Edgecombe et al., 1999;

Edgecombe & Giribet, 2002), but not in scolopendromorphs.

12. Cephalic plate, frontal line: (0) uniformly rounded;

(1) with a vertex pointing backwards; (2) with a vertex

pointing forwards.

This character obviously only applies to species with the

frontal line (see ch. 11). Within mecistocephalids, the

frontal line is generally uniformly rounded, the convexity

bending backwards (Fig. 8D). In some species, however, it

forms a backward directed angle (Fig. 8C), whereas in other

species it forms a forward directed angle (Fig. 8B). But the

angle is not always recognizable in all individuals of a given

species. Therefore, this character has been evaluated, when-

ever possible, on a series of conspecific individuals. In

Adesmata, scolopendromorphs and lithobiomorphs, the

frontal line is usually rounded.

13. Cephalic plate, cephalic sulci: (0) present; (1) absent.

The cephalic sulci are a pair of shallow grooves, running

longitudinally on the posterior part of the cephalic plate, in

correspondencewith the paramedian rows of setae and sensilla.

Some species of mecistocephalids have cephalic sulci, whereas

others lack them. No intraspecific variability has been noticed

for this character, but in the smallest individuals the sulci are

difficult to observe. Cephalic sulci are not recognizable in

Adesmata and scolopendromorphs. In the latter, however,

lines usually regarded as ‘sutures’ (Eason, 1964) are present in

the corresponding position but are actually different in struc-

ture. Possibly homologous sulci are present in lithobiomorphs,

in some species at least.

14. Eyes: (0) present; (1) absent.

All geophilomorphs are blind, whereas most lithobio-

morphs and many scolopendromorphs have ocelli, up to

about thirty per side in the former and up to four per side in

the latter. The complete regression of the eyes evolved

independently in different groups of centipedes, i.e.

in some lithobiomorph lineages, in the ancestor of

scolopendromorph family Cryptopidae and in the ancestor

of geophilomorphs.

15. Organ of Tomosvary: (0) present; (1) absent.

Absence of the organs of Tomosvary (also called the

postantennal organs) has been recognized as a synapomor-

phy of Epimorpha (see, e.g. Dohle, 1985).

Clypeus

16. Clypeus: (0) convex, projecting; (1) flat.

In all geophilomorphs, the clypeus is quite flat and devoid

of processes. In both lithobiomorphs and scolopendro-

morphs, the central part of the clypeal surface is swollen

and protruding ventrally.

17. Paraclypeal sutures: (0) parallel; (1) convergent poster-

iorly.

The paraclypeal sutures are at the lateral margins of the

clypeus, dividing the latter from the buccae. In most mecisto-

cephalid species, these sutures are convergent posteriorly

(Fig. 9A). In a few species, the lateral margins are subparallel,

A

DC

B

Fig. 8. Head (dorsal view) of some mecistocephalids. One antenna

as well as all antennal setae and sensilla are omitted. A, Anarrup

flavipes,/, subadult, Lombok; B, Dicellophilus anomalus, /, adult,

California; C, Tygarrup muminabadicus, /, subadult, Kashmir; D,

Mecistocephalus microporus, /, subadult, Cebu. Scales¼ 1mm.



at least at mid-length (Fig. 9F). In most Adesmata, scolopen-

dromorphs and lithobiomorphs, the paraclypeal sutures are

not convergent.

18. Clypeus, areolation: (0) homogeneous; (1) marked on

the anterior part, virtually absent on the posterior part.

In all mecistocephalids, the anterior part of the clypeus

(areolate clypeus) is characterized by a marked areolation,

well visible even before handling the specimen with a clarifying

medium. The posterior part (clypeal plagula or plagulae) is

strongly sclerotized and not evidently areolate, the pattern of

scutes not being visible even after clarification. This condition

is common to all species, with the possible exception of the

enigmatic Megalacrus obscuratus, the clypeus of which has

been described as homogeneous (Attems, 1953). The distinc-

tion between an anterior and a posterior part of the clypeus has

already been recognized as a peculiar feature of mecistocepha-

lids (Hoffman, 1982). In the literature, however, different

authors recognized different parts in the clypeus, or at least

referred to the same parts by means of different terms. In

Adesmata, scolopendromorphs and lithobiomorphs, the cly-

peus is homogeneous and the areolation is variable in evidence.

19. Areolate clypeus, clypeal insulae: (1) absent; (2) present.

In some mecistocephalid species, one or more poorly areo-

late areas (clypeal insulae) are present inside the areolate

A B

F

GH

CD

E

Fig. 9. Clypeus, labrum and anterior part of buccae (ventral view) of some mecistocephalids. The areolation is omitted, but the margins of

areolate areas are indicated as dashed lines. A, Arrup holstii, /, subadult, China; B, Proterotaiwanella tanabei, /, subadult, Ryukyu Islands;

C, Anarrup flavipes, /, subadult, Lombok; D, Dicellophilus anomalus, /, adult, California; E, Tygarrup muminabadicus, /, subadult,

Kashmir; F, Takashimaia ramungula, ?, adult, Honshu; G, Mecistocephalus punctifrons, /, subadult, India; H, Mecistocephalus microporus,

/, subadult, Cebu. Scales¼ 0.5mm.



surface of the anterior part of the clypeus (Fig. 9G). These

insulae are usually grouped on the posteromedian part

of the areolate clypeus. Conspecific individuals may be

very different in number, pattern, total extent and gradual

vs abrupt delimitation of these insulae. In a population

characterized by clypeal insulae, a few individuals may

completely lack them. Insulae are more frequent and

numerous in the oldest and largest individuals. Thus, as

far as possible, the occurrence of these areas in a species

has been checked in a series of specimens. This character

does not apply to the outgroup taxa, because their clypeus is

uniform and only slightly areolate. One or two finely areo-

late areas (called the ‘clypeal areas’) are known in some

Adesmata but they are most probably not homologous to

the clypeal insulae of mecistocephalids.

20. Clypeus, areolation on the central part: (0) virtually

absent; (1) present, as strong as on the remaining

areolate clypeus; (2) present, but weaker than on the

remaining areolate clypeus.

In mecistocephalids, a wide rhomboidal area in the mid-

dle of the clypeus may be areolate, with different degrees of

evidence. In some species, this area is virtually nonareolate,

like the posterior part of the clypeus, and it is thus con-

sidered part of the clypeal plagula (Fig. 9E). In other species,

it is areolate; thus, it is considered part of the areolate clypeus

(Fig. 9H). In this latter case, however, the areolation may be

as evident as in the remaining areolate clypeus or less

marked. Intraspecific variation is low; however, in some

species in which the central clypeus is areolate, the difference

between this central area and the remaining areolate

clypeus becomes more evident with growth. In lithobio-

morphs and scolopendromorphs, a clypeal part homo-

logous to the central part of mecistocephalid clypeus is

hardly recognizable. Despite this, the central part of the

clypeus may be considered virtually nonareolate, because

this is the condition of the whole clypeus.

21. Clypeus, mid-longitudinal areolate stripe: (0) absent;

(1) present.

A mid-longitudinal areolate stripe in the clypeus is exclu-

sive to some mecistocephalid species (Fig. 9A), whereas it is

lacking in the other mecistocephalids as well as in all other

centipede groups.

22. Clypeus, forward extension of the lateral part of the

clypeal plagula along the paraclypeal suture: (1) absent;

(2) at least half the length of the paraclypeal suture,

without a marginal areolate band; (3) at least half the

length of the paraclypeal suture, with a marginal

areolate band.

The relative extension of the clypeal plagula/ae is very

diverse among mecistocephalid species. The minimum

extension is in some Arrup, where the maximum length of

the two plagulae is 0.1–0.2 times the total length of the

clypeus. The maximum extension is in Anarrup, where the

only plagula covers the clypeus almost totally. The relative

extension of the lateral margins of the plagula/ae along the

paraclypeal margins is diverse too. In most mecistocephalid

species, the clypeal plagula/ae are largely in touch with the

paraclypeal margins (Fig. 9D). In some species, however, two

areolate bands extend backwards from the areolate clypeus

to separate the plagulae from the buccae (Fig. 9A). In other

species, the plagulae themselves seem to gradually become

areolate along the margins (Fig. 9F). This character does not

apply to the outgroup taxa, because the clypeal plagulae are

not defined in centipedes other than mecistocephalids.

23. Clypeus, setae at each antero-external corner: (0) none;

(1) one; (2) more than one.

In some mecistocephalid species, one to a few strong setae

develop at each anterolateral corner of the clypeus, quite

separate from the main array of clypeal setae (Fig. 9D,E).

In other species, conversely, no setae at all develop in this

position (Fig. 9F). The occurrence of these setae, however, is

quite variable among conspecific individuals. In the species

characterized by single setae, e.g. inTygarrup, these setae may

be undetectable in some of the youngest specimens; in the

species characterized by tufts of setae, e.g. in Dicellophilus

and in some Mecistocephalus, the number usually increases

during growth, possibly starting with only one seta at each

corner. As a consequence, this character was evaluated, as far

as possible, in series of individuals. No homologous setae

seem to occur in lithobiomorphs and scolopendromorphs.

In Adesmata, a high interspecific diversity is known.

24. Clypeus, setae on the lateral parts: (0) none; (1) on a

narrow transverse band; (2) on a wide surface.

Within mecistocephalids, the pattern of clypeal setae is

widely diverse. The number of setae ranges from only six in

some Mecistocephalus to about 100 in Dicellophilus. When

few in number, the setae are patterned and symmetrical;

when numerous they are scattered and irregularly distribu-

ted. In some species, the clypeal setae are limited to an

anteromedian area of the clypeus (Fig. 9F), whereas in

other species they evidently extend to the lateral parts of

the same. In this latter case, two conditions may be recog-

nized: one to two dozen setae, covering a wide but short

transverse band, as in some Tygarrup and Mecistocephalus

(Fig. 9E); more numerous setae, extending on a longer area,

as in Arrup and Dicellophilus (Fig. 9A). These basic patterns

are poorly affected by both the developmental increase in

the total number of setae and the interindividual variability

in number, position and symmetry. The other centipede

groups show a wide diversity in the pattern of setae, but the

setae usually do not extend to the lateral parts of the clypeus.

25. Clypeus, posteromedian pair of setae: (1) absent; (2)

present, close to the posterior margin of the clypeus; (3)

present, close to the centre of the clypeus.

In most mecistocephalids, a pair of strong setae may be

easily distinguished from the other clypeal setae. This pair is

placed posterior to all other setae, either at mid-length on

the clypeus (in most species; Fig. 9E) or close to the poster-

ior margin (in someDicellophilus; Fig. 9D). This character has

not been coded for the outgroup taxa, because homologies

among different patterns of clypeal setae are very dubious.

26. Clypeus, lateral groups of buttonlike sensilla: (0) absent;

(1) present.

In mecistocephalids, buttonlike sensilla are usually

spread among the clypeal setae. Two additional groups of

similar sensilla, however, are present in some species. These



sensilla are evident on the lateral parts of the clypeus, often

at the antero-external corners of the plagula/ae (Fig. 9H).

The number of these sensilla increases during growth, but

their occurrence seems invariant among conspecifics. In the

literature, the emergence sockets of these sensilla were erro-

neously interpreted as glandular pores (e.g. Verhoeff, 1934;

Takakuwa, 1937).

27. Clypeus, micropores: (0) absent; (1) present.

In some mecistocephalid species, the clypeal plagula/ae

are pierced with several scattered micropores [so called by

Crabill (1970); probably the same as the ‘micropores’ of

Turcato et al. (1995)]. The occurrence of similar pores in

other centipede groups is largely ignored, but our prelimin-

ary observations detected them in some scolopendro-

morphs.

Bucca

28. Bucca, spiculum: (0) absent; (1) present.

The spiculum (Fig. 9F) is a strongly sclerotized process,

pointed forwards and present on the most anterior part of

each bucca, in some mecistocephalid species only.

29. Bucca, stilus: (0) absent; (1) present.

In all mecistocephalids, the internal margin of each bucca

is shaped as a thickened stick (stilus), which is characterized

by an anterior pointed tip and an internal notch on the

anterior part (Fig. 9A). All other centipede groups lack

these structures.

30. Bucca, areolation on the anterior part: (0) absent;

(1) present.

In all mecistocephalids, the anterior part of each bucca is

evidently areolate, whereas the remaining part is not

(Fig. 9A). In all other centipede groups, the buccae are

usually homogeneously nonareolate.

31. Bucca, areolation along the paraclypeal margin:

(0) absent; (1) present, very narrow, as in Fig. 9(C);

(2) present, as wide as in Fig. 9(E); (3) present, very

wide, as in Fig. 9(H).

In all mecistocephalids, the areolation of the anterior part

of each bucca extends as a band along the paraclypeal

suture. The relative width of this band is different in differ-

ent species, and three conditions are recognized here

(Fig. 9C,E,H). A similar areolate band is lacking in all

other centipede groups.

32. Bucca, setae on the anterior half: (0) absent or very tiny;

(1) present, long.

In a few mecistocephalid species, a group of strong and

long setae develops on the anterior half of each bucca

(Fig. 9C). In the other species, only short processes, similar

to buttonlike sensilla, may be present. Anterior buccal setae

are also present in some Adesmata.

33. Bucca, setae on the posterior half: (0) absent; (1) present.

In some mecistocephalid species, the posterior half of

each bucca is covered with a group of strong and long

setae, completely lacking in other species. Posterior buccal

setae are also present in some Adesmata.

Labrum

34. Labrum, side-piece, transverse thickened line: (0) absent;

(1) present.

In all mecistocephalids, the labrum is composed of three

sclerites, a mid-piece and a pair of side-pieces. Each

side-piece is divided into two parts (alae) by a thickened

transverse line (Fig. 9A). This line, however, is weak or

discontinuous in miniaturized species such as Nannarrup

hoffmani (see Foddai et al., 2003) and in juveniles of

other species. This line was sometimes considered a

true suture between sclerites (Hoffman, 1982; Edgecombe

et al., 1999). In all other centipede groups, the side-pieces

seem to be undivided. In some Adesmata, however, the

shape and structure of the labrum are so modified that

homologous relationships are difficult to recognize. The

labrum of the schendylids, for instance, was interpreted as

divided into alae like those of mecistocephalids (Crabill,

1970).

35. Labrum, side-piece, transverse thickened line:

(1) straight; (2) curved, convex forwards; (3) sinuous,

convex forwards on the internal part, backwards on

the external part; (4) curved, convex backwards.

The shape of the transverse thickened line of the labrum

side-piece is different in different mecistocephalid species.

Four different conditions may be recognized, i.e. straight

(Fig. 9E), convex forwards (Fig. 9D), sinuous (Fig. 9B) and

convex backwards (Fig. 9F). This character does not apply

to the outgroup taxa because the thickened lines are present

in mecistocephalids only.

36. Labrum, anterior ala, internal margin: (1) reduced to a

pointed end; (2) long.

In some mecistocephalid species, the internal margin of

the anterior ala is very reduced in length, with respect to the

internal margin of the posterior ala (Fig. 9C). In other

species, it is significantly elongate, its length comparable,

although shorter, with the length of the internal margin of

the posterior ala (Fig. 9G). This character does not apply to

the outgroup taxa because the thickened lines are present in

mecistocephalids only.

37. Labrum, posterior margin, backwards convexity:

(0) present; (1) absent.

In most mecistocephalid species, the posterior margin of

the labrum extends backwards in a weak convexity on its

internal half (Fig. 9A). In a few species only, the entire margin

is homogeneously concave (Fig. 9D). In lithobiomorphs,

scolopendromorphs and Adesmata, a convexity is usually

present, sometimes even more pronounced than in mecisto-

cephalids.

38. Labrum, side-piece, posterior margin, medial process:


In a few mecistocephalid species, the postero-internal

corner of each labral side-piece extends to a short but

evident subrectangular process. In the other species, this

corner is either not pronounced or extends into a tiny

tooth only.

39. Labrum, side-piece, longitudinal stripes: (0) absent;

(1) present.



Longitudinal stripes, alternatively lighter and darker,

typically appear on the posterior alae in a few species of

mecistocephalids (Fig. 9C).

40. Labrum, side-piece, posterior margin, medial notches:


In a few mecistocephalid species only, the most internal

part of the posterior margin of each labrum side-piece is

marked by a short series of distinct notches (Fig. 9B). In

other mecistocephalids, the entire posterior margin is either

partially crenate or not crenate.

41. Labrum, hairlike dorsal processes: (0) present;

(1) absent.

Within mecistocephalids, either papillae or short hairlike

processes cover a pair of peculiar sclerites placed dorsal

to the labral side-pieces. The occurrence of these processes

was checked by observing the labrum from below, by

light transmission. The occurrence, size and extension of

these processes are highly variable among conspecific

individuals. Accordingly, they were evaluated, as far as

possible, in series of specimens and were scored as

present when observed in at least some conspecifics. Tenta-

tively, we regard similar processes known in some groups

of Adesmata as homologous to those in mecisto-

cephalids. This character has often been dramatically

misinterpreted. In some mecistocephalid species, ‘hairs’

were erroneously described as present along the

posterior margin of the labrum only, as in some Mecisto-

cephalus traditionally assigned to the subgenera Brachyptyx

and Dasyptyx (see Silvestri, 1919; Chamberlin, 1920a, b;

Attems, 1929; Takakuwa, 1937, 1940, 1942); in other

species, different authors disagree as to whether the

labrum is ‘hairy’ or not, as in Tygarrup javanicus (see

Attems, 1907, 1926, 1928, 1929; Demange, 1981; Lewis &

Rundle, 1988).

42. Labrum, fringe along the posterior margin: (0) absent;

(1) present.

A fringe of hairlike projections is present along the pos-

terior rim of the labrum side-pieces in some mecisto-

cephalid species (Fig. 9D). These projections are

distinguished from the shorter hairlike processes covering

the additional dorsal sclerites (see ch. 41). Elements of

the fringe are longer and less flexible than the dorsal

hairlike processes and their occurrence may be independent

of the latter. They were described in the literature

under different names, because they were sometimes not

distinguished from the dorsal hairlike processes or other

projections. In mecistocephalids, only a few species actually

have a fringe and no intraspecific variability occurs. In

other species, however, some dorsal hairlike processes

may also be visible along the labral margin, but they

appear structurally different from the elements of the

fringe and their occurrence is largely variable. Different

conditions may be recognized in Adesmata. In scolopendro-

morphs, branched projections along the labrum margin

are tentatively recognized here as homologous to the

fringe. In lithobiomorphs, different projections (branching

bristles) develop on the dorsal surface of the labrum rather

than along the margin.

43. Paralabial sclerites: (0) present; (1) absent.

Two paralabial sclerites are found beside the labrum side-

pieces in most centipede groups, but are virtually absent in

most geophilomorphs, including mecistocephalids.

Mandible

44. Mandible, basal tooth: (0) rounded; (1) subconic.

A basal tooth is borne on the dorsal margin of the man-

dible, at the base of the first lamella (Fig. 10A). It is subconic

in all mecistocephalid species, except for the enigmatic

Megalacrus obscuratus, which is described as lacking a

mandibular tooth (Attems, 1953). This tooth is rounded in

all lithobiomorphs and scolopendromorphs, whereas Ades-

mata show diverse conditions.

45. Mandible, dentate lamellae: (0) present; (1) absent.

In geophilomorphs, two types of mandibular lamella are

traditionally recognized, i.e. the dentate lamellae (one or

none on each mandible) and the pectinate lamellae (one to

several on each mandible). In mecistocephalids, all mandib-

ular lamellae are quite similar to each other and are uni-

versally considered homologous to the pectinate lamellae of

A

B

Fig. 10. A, Mandible of a mecistocephalid (Tygarrup muminaba-

dicus, /, subadult, Kashmir); B, hyaline scales on the medial

projection of the first maxillae (dorsal view) in a mecistocephalid

(Arrup dentatus, /, adult, Hokkaido). Scales¼ 0.1mm.



the other geophilomorphs, whereas no dentate lamellae are

thought to be present (Fig. 10A). In Adesmata, very differ-

ent conditions evolved in different groups, but an array of

one dentate lamella followed by some pectinate lamellae is

recognizable as the ancestral condition. In both lithobio-

morphs and scolopendromorphs, at least one dentate

lamella is always present.

46. Mandible, number of fully developed pectinate lamellae:

(0) 1; (1) 4–5; (2) 6–7; (3) 8–12; (4) 15–25.

The number of pectinate lamellae (Fig. 10A) is very vari-

able among mecistocephalid species, ranging from very few

(four in most Arrup species, but only two recorded in sub-

adult Arrup pylorus; see Chamberlin, 1920c) to twenty-seven

(in adult Mecistocephalus gigas Haase, 1887). Because the

number is quite variable among conspecific individuals and

increases appreciably during growth, the average number in

the adult stage was checked for each species on a series of

specimens. Because on each mandible the series of lamellae

fades ventrally into some final rudimentary elements, only

the fully developed lamellae were counted. The mandible of

lithobiomorphs and scolopendromorphs, although well

known and very conservative, is hard to compare with

that of geophilomorphs, particularly with that of mecisto-

cephalids. In lithobiomorphs and scolopendromorphs,

numerous elongate projections (‘aciculae’) are inserted on

each mandible close to the dentate lamella. These elements

are probably homologous to the pectinate lamellae of geo-

philomorphs (Cook, 1896; Edgecombe et al., 1999;

Edgecombe & Giribet, 2002), but the actual number of

homologous elements is difficult to assess. Most authors

(e.g. Eason, 1964; Lewis, 1981) tentatively recognized only

one element in lithobiomorphs and a few elements in scolo-

pendromorphs. Because uncertainty persists, we preferred

to code this character as unknown in the outgroup species

Lithobius forficatus and Cryptops anomalans. In all Ades-

mata, at least one pectinate lamella is recognizable, but its

shape is highly different in different families.

47. Mandible, pectinate lamella, size of teeth: (0) changing

gradually along the lamella; (1) changing abruptly

along the lamella.

In each mandibular pectinate lamella, the size and length

of the teeth increase from the most basal teeth to the tip of

the lamella. The transition is gradual in most mecistocepha-

lids, but it is evidently abrupt in a few species. We have

evaluated this character in one of the intermediate lamellae,

because in the first lamella all teeth appear quite similar and

in the rudimentary lamellae teeth are not well developed.

The state of this character has been coded as unknown in

both lithobiomorphs and scolopendromorphs because of

the difficulty in recognizing mandibular elements homo-

logous to geophilomorph lamellae and teeth (see ch. 46).

48. Mandible, ‘hairs’ on the ventral surface: (0) absent;

(1) present.

In a few mecistocephalid species, the ventral surface of

the mandible is covered with thick short ‘hairs’ at the base

of the rudimentary lamellae. In all lithobiomorphs, in scolo-

pendromorphs and at least in the Adesmata considered here,

this mandibular region is not hairy. Descriptions available in

the literature for some species are ambiguous because ‘hairs’

and lamellar rudiments were sometimes confused.

49. Mandibular fulcrum: (0) well developed; (1) reduced

and displaced.

The fulcra are a pair of articulatory sclerites between the

mandibles and the cephalic shield. They are very reduced in

all mecistocephalids, in contrast to all other centipede groups.

First maxillae

50. First maxillae, coxosternum, ratio of width to length:

(0) 4.0–4.8; (1) 3.3–3.6; (2) 2.3–2.8; (3) 1.8–2.2.

Within mecistocephalids, the degree of elongation of the

coxosternum of the first maxillae (Fig. 11) is variable among

species. The ratio of maximum width to medial length was

evaluated after dissecting the maxillary complex and flat-

tening the coxosternum completely. Ratios were generally

clustered around three modal values, thus allowing us to

recognize tentatively as many intervals. A fourth interval

was defined for the outgroup taxa. Allometric changes

occur during growth as well as slight interindividual vari-

ability, thus the ratio was calculated, as far as possible, for a

series of conspecific adults or subadults. In lithobiomorphs,

scolopendromorphs and most Adesmata, the coxosternum

is less elongate than in mecistocephalids.

51. First maxillae, coxosternum, antero-external corners:


Within mecistocephalids, the first maxillary coxosternum

takes either of two alternative shapes. In some species, the

lateral margins are convergent forwards and about aligned

with the external margins of the telopodites (Fig. 11A). In

the other species, the coxosternum is expanded in antero-

external corners, either rounded or pointed (Fig. 11B,F).

52. First maxillae, coxosternum, pointed process on each

antero-external corner: (0) absent; (1) present.

In some mecistocephalid species, the coxosternum of the

first maxillae is characterized by a pair of anterolateral

pointed corners (Fig. 11F). Similar pointed processes do

not occur in the other centipede groups.

53. First maxillae, coxosternum: (0) divided mid-

longitudinally, ventral surface nonareolate; (1) undivided

mid-longitudinally, ventral surface areolate.

In most mecistocephalid species, the first maxillary coxo-

sternum is divided into lateral halves by a sutural line,

which is evident in all individuals (Fig. 11B). In a few spe-

cies, however, the coxosternum is undivided, at least on the

ventral surface, as also confirmed by the presence of a

continuous areolation on the intermediate part (Fig. 11A).

Postembryonic development does not change the structure

of the coxosternum, and no interindividual differences

occur. Within Adesmata, the coxosternum is undivided in

most lineages, but it is evidently divided in neogeophilids

(Crabill, 1969), which are regarded as basal Adesmata by

Foddai & Minelli (2000). In centipedes other than geophi-

lomorphs, the coxosternum is usually divided mid-

longitudinally. The actual structure of the coxosternum in

some species of Arrup has often been misunderstood, thus



supporting erroneous genus-level taxonomic arrangements.

Some authors described the coxosternum as divided mid-

longitudinally, as in the other mecistocephalids (see

Brolemann, 1909; Chamberlin, 1912; Takakuwa, 1934, 1940;

Takashima & Shinohara, 1952). Indeed, a weak medial line is

visible on the coxosternum of some specimens, but this is not

actually a suture, rather it probably marks internal muscular

attachments [as recognized by Crabill (1964)].

54. First maxillae, coxosternum, areolation along the ante-

rior margin: (0) absent; (1) present.

In all mecistocephalid species, areolation occurs along the

anterior margin of the first maxillary coxosternum, at the

base of the medial projections and the telopodites (Fig. 11).

This areolation is not present in Adesmata, at least in the

taxa considered here. It has been observed in scolopendro-

morphs, but not in lithobiomorphs.

55. First maxillae, coxosternum, setae: (0) present; (1) absent.

In some mecistocephalid species, one to a few setae

develop on the antero-internal corner of each of the two

halves of the first maxillary coxosternum (Fig. 11B). In

these species, the number and size of the setae may differ

among conspecifics, but some setae are always present in all

individuals. In other species, the coxosternum is invariably

devoid of setae (Fig. 11A). The condition is variable within

Adesmata, whereas in both lithobiomorphs and scolopen-

dromorphs, setae are always lacking.

56. First maxillae, medial projection, ratio of length to

width: (0) 0.7–1.2; (1) 1.3–1.9; (2) 2.0–2.6; (3)>4.0.

In the firstmaxillae ofmecistocephalids, twopairs of append-

ages are articulated on the anterior margin of the coxosternum

(Fig. 11). The internal and external appendages are traditionally

referred to as the medial projections (also called the coxal

projections) and the telopodites, respectively. The relative

elongation of themedial projections is very different in different

species. Itwas evaluated here as the ratio between the length and

the maximum width. This ratio ranged from <1 to about 3,

exceptionally to>4. InMecistocephalus longichilatus, the distal

hyaline parts of the projections are exceptionally elongate, over-

reaching the anterior margin of the cephalic shield (Takakuwa,

1936). Because the ratio may vary among conspecific individ-

uals, it was calculated, as far as possible, on series of

specimens for each species. The elongation is variable within

Adesmata, scolopendromorphs and lithobiomorphs.

57. First maxillae, medial projection: (0) evenly sclerotized;

(1) basal part sclerotized, distal part hyaline.

In all mecistocephalids, the medial projections appear as

independent sclerites articulated to the coxosternum and

each bears a more sclerotized basal part and a hyaline distal

part. In Adesmata, scolopendromorphs and lithobio-

morphs, the medial projections are structurally continuous

with the coxosternum and evenly sclerotized.

58. First maxillae, medial projection, distal part: (0) not

enlarged; (1) enlarged, subtriangular; (2) enlarged, sub-

elliptical.

Within mecistocephalids, the shape of the distal hyaline

part of each medial projection is very different in different

species. In some species it is homogeneous in width

(Fig. 11A); in others it is enlarged, either in a subelliptical

(Fig. 11E) or a subtriangular shape (Fig. 11C). In the

exceptional case of Mecistocephalus longichilatus (see ch.

56), the hyaline parts are evenly narrowed to the tip. In

Adesmata, scolopendromorphs and lithobiomorphs, the dis-

tal parts of the medial projections are generally not enlarged.

59. First maxillae, medial projection, hyaline scales:


In some mecistocephalid species, the medial projections

of the first maxillae have peculiar hyaline scales (Fig. 10B).

These scales are arranged in a narrow band, longitudinally

aligned on the dorsal surface and close to the internal

margin of the projection. The number of scales increases

during growth, but their presence is invariant within a

species. These scales have never been described or illu-

strated for mecistocephalids. Thus, this character has

been coded as unknown for all the species not directly

studied by us. Very little information is available for a few

species of Adesmata recently illustrated in taxonomic papers.

60. First maxillae, telopodite, length to width ratio:

(0) 1.0–1.6; (1) 2.3–2.8; (2) 3.0–3.8; (3)>4.0.

Within mecistocephalids, the elongation of the first max-

illary telopodites is different in different species. The ratio of

the maximum length to the maximum width was evaluated.

This ratio ranged mostly from about 2 to about 4, falling into

two separate intervals. Mecistocephalus longichilatus, how-

ever, is exceptional in having the hyaline parts of the telopo-

dites so elongate as to overreach the anterior margin of the

cephalic shield (Takakuwa, 1936). The ratio may be variable

among conspecific individuals. Thus, it was calculated, as far

as possible, on a series of individuals for each species. In

Adesmata, scolopendromorphs and lithobiomorphs, telopo-

dites are relatively stout.

61. First maxillae, telopodite, articles: (0) 2; (1) 1.

In all mecistocephalids, each telopodite of the first

maxillae is invariably composed of only one sclerite,

even though a more sclerotized basal part contrasts with

a hyaline distal part, like the medial projections (see ch.

57). In lithobiomorphs and scolopendromorphs, the

telopodites are composed of two articles, whereas within

Adesmata they are composed of either one or two articles.

Second maxillae

62. Second maxillae, coxosternum: (0) undivided; (1)

divided mid-longitudinally.

In most mecistocephalid species, the coxosternum of the

second maxillae appears as an undivided sclerite, although

the median part is quite short (Fig. 11D). Only in two

species is the coxosternum divided into halves by a complete

mid-longitudinal sutural line, clearly visible on the ventral

surface (Fig. 11B). In scolopendromorphs and lithobio-

morphs, the coxosternum is also divided. In many Adesmata,

it is undivided; in the other ones a sutural line is present.

63. Second maxillae, coxosternum, ratio of maximum length

to medial length: (0) 3.3–5.0; (1) 2.2–3.0; (2) 1.6–2.2.



We evaluated the ratio between the maximum length

(from the telopodite-bearing processes to the posterolateral

corners) and the medial length (between the anterior and

the posterior concave margins of the median part) as a

proxy for the general shape of the second maxillary coxos-

ternum. Within mecistocephalids, three intervals are tenta-

tively recognized. Because of allometric growth and

interindividual variability, measures were taken, as far as

possible, from series of conspecific adults.

64. Second maxillae, coxosternum, median part: (0) hyaline,

different from the lateral parts; (1) sclerotized, con-

tinuous with the lateral parts.

In most mecistocephalid species, the second maxillary cox-

osternum is evenly sclerotized in the intermediate part as well

as in the lateral parts, the former thus appearing continuous

with the latter (Fig. 11E). In some species, the two sclerotized

halves are connected by a diaphanous membranous part,

distinct from the lateral parts (Fig. 11A). In Adesmata, the

A B

D

C

E F

Fig. 11. First and second maxillae (ventral view) of some mecistocephalids. Only the right half of the whole maxillary complex is illustrated.

The areolation is omitted, but the margins of areolate areas are indicated as dashed lines. A, Arrup holstii, /, subadult, China; B, Anarrup

flavipes, /, subadult, Lombok; C, Dicellophilus anomalus, /, adult, California; D, Tygarrup javanicus, /, subadult, Halmahera; E,

Mecistocephalus tahitiensis, /, adult, Palawan; F, Mecistocephalus microporus, /, subadult, Cebu. Scales¼ 0.2mm.



median part is also usually sclerotized. In both lithobiomorphs

and scolopendromorphs, there is a hyaline connection.

65. Second maxillae, coxosternum, areolation on the median

part: (0) virtually absent; (1) present, evident.

In most mecistocephalid species, the intermediate region of

the second maxillary coxosternum is areolate (Fig. 11C). Even

in the species with a hyaline connection (see ch. 64), the inter-

nal margins of the sclerotized lateral halves are areolate

(Fig. 11A). In some species, however, areolation is completely

lacking on the median part of the coxosternum (Fig. 11B). In

Adesmata, scolopendromorphs and lithobiomorphs, the entire

intermediate part of the coxosternum is virtually nonareolate.

66. Second maxillae, coxosternum, smooth insulae on the

median band: (1) absent; (2) present.

The areolation covering the intermediate part of the sec-

ond maxillary coxosternum may fade in subcircular areas,

which appear nonareolate (Fig. 11E). These areas, called

smooth insulae, are variable in number, diverse in extent

and irregularly placed. When particularly wide, they may be

coalescent, to form an entire mid-longitudinal nonareolate

band (Fig. 11F). Smooth insulae occur in some species only.

The number and extent are largely variable among individ-

uals, possibly related to age, and insulae may be lacking in

some individuals. Thus, species were coded as lacking smooth

insulae after checking more than one specimen. This character

does not apply to the species lacking areolation on the median

part of the coxosternum, all outgroup taxa in particular.

67. Second maxillae, coxosternum, areolation on a posterior

band: (0) virtually absent; (1) present, evident.

In some mecistocephalid species, the areolation of the

second maxillary coxosternum extends to a wide transverse

band, which runs along the posterior margin between the

two postero-external corners (Fig. 11E).

68. Second maxillae, coxosternum, areolation on the lateral

parts: (0) virtually absent; (1) present, evident.

In a few mecistocephalid species, the areolation covers the

lateral parts of the second maxillary coxosternum

extensively, whereas in other species it is limited to the

intermediate and posterior parts (Fig. 11E). Differences

in the degree of areolation may be observed

among conspecific individuals. Therefore, more than

one specimen was checked for each species, whenever pos-

sible.

69. Second maxillae, metameric pore and foraminal process:

(0) groove from the metameric pore running towards

the posterior margin of the coxosternum, foraminal

process not detectable; (1) groove from the metameric

pore running towards the postero-external corner

of the coxosternum, foraminal process reaching

the postero-external corner of the coxosternum;

(2) groove from the metameric pore running towards

the lateral margin of the coxosternum, foraminal

process not reaching the postero-external corner of

the coxosternum.

Within mecistocephalids, the postero-external regions of

the second maxillary coxosternum may exhibit either of two

conditions (Fig. 11A,B) differing in the position of the

metameric pore, the length and the direction of the groove

running from the pore and the elongation of the foraminal

process, i.e. the narrow projection lateral to the pore. In

lithobiomorphs and Adesmata, the arrangement is different

from those found in mecistocephalids. The metameric pores

are not detectable in scolopendromorphs. Thus, the char-

acter does not apply to the relevant outgroup species. Until

now, the taxonomic value of this character was largely

underestimated. Only Crabill (1964) discussed it critically,

hypothesizing an evolutionary trend from metameric pores

opening backwards to pores opening laterally, but only on a

speculative basis.

70. Second maxillae, telopodite, first article, ratio of length

to width: (0) 2.4–4.0; (1) 1.0–2.0.

In most mecistocephalid species, the first article of the

second maxillary telopodite is long and slender (Fig. 11B),

whereas in other species it is stout (Fig. 11A). This differ-

ence was evaluated as the ratio of the maximum length to

the maximum width.

71. Second maxillae, telopodite, distal article, ratio of length

to width: (0) 1.8–3.5; (1) 1.3–1.6.

In a few mecistocephalid species, the distal article of the

second maxillary telopodite is quite swollen, wider than

the second article (Fig. 11B); in most species it is

slender (Fig. 11D); in Takashimaia ramungula it is typically

constricted. This difference was evaluated by the ratio of

the maximum length to the maximum width of the

distal article. Only two intervals were recognized because the

ratio is affected by allometric changes during growth, by

interindividual differences and (rarely) by asymmetry.

72. Second maxillae, telopodite, praetarsus: (0) present;

(1) absent.

In most mecistocephalid species, the second maxillary

telopodite bears a tiny apical claw, the praetarsus

(Fig. 11E), usually a subconic rigid process, slightly curved

inwards. Its shape is highly variable, even between the two

telopodites of the same specimen, possibly because of devel-

opmental problems or secondary damage due to feeding

activity. It often bears short accessory spines at the base

and, in rare instances, it is forked. Sometimes it is not

pointed but in the shape of a stout tubercle crowned with

short spines. In all these species, however, the praetarsus is

present in all individuals, independent of age or sex. In

other species, the praetarsus is regularly lacking

(Fig. 11A). In both lithobiomorphs and scolopendro-

morphs, the praetarsus is always present, in the shape of a

well-developed claw, more evident than in mecistocepha-

lids. Accessory processes are also typically present. In Ades-

mata, the praetarsus is highly variable among species, from

well developed to completely absent (see Borucki, 1996).

73. Second maxillae, telopodite, distal article, setae on the

external side: (0) less numerous than on the internal

side; (1) as numerous as on the internal side.

In most mecistocephalid species, the setae covering the

third article of the second maxillary telopodite are not evenly

placed on the entire surface. They are usually aggregated on

the internal distal part of the article, but virtually absent on

the proximal external side of the same article (Fig. 11E). Only

in a few species, the setae are dispersed almost evenly on the



entire surface of the article (Fig. 11B). In Adesmata, litho-

biomorphs and scolopendromorphs, at least in the outgroup

species considered here, the pattern of setae is heterogeneous,

as in most mecistocephalids.

Forcipular segment

74. Forcipular segment, tergum, length to width ratio: (0)

0.2; (1) 0.4–0.5; (2) 0.6–0.8.

All mecistocephalids are characterized by the reduction

of the width of the forcipular tergum and by the corres-

ponding dorsal overgrowth of the forcipular pleura. Two

different degrees of reduction occur in different species, as

expressed by the ratio between the medial length and the

maximum width of the tergum (Fig. 12A,B). In some spe-

cies, e.g. in Anarrup flavipes, an additional sclerite, tiny and

weakly sclerotized, is visible just anterior to the forcipular

tergum. This sclerite, possibly recognizable as a praetergum,

was disregarded when measuring the tergum. This character

does not apply to scolopendromorphs because in this group

an independent forcipular tergum is not recognizable: in all

scolopendromorphs, a single large tergum covers the for-

cipular segment and the first leg-bearing segment; this scler-

ite was often considered to be derived from the ‘fusion’ of

the forcipular tergum and the following tergum (Eason,

1964; Lewis, 1981), but its developmental and evolutionary

origin is actually unknown. In all lithobiomorphs and in

most Adesmata, the forcipular tergum is about as wide as

the whole forcipular segment. Thus, the ratio considered

A B

C D

Fig. 12. Forcipular segment (dorsal view; head detached) of some mecistocephalids. Only the left forcipule is illustrated. The areolation of the

tergum is omitted, but the margins of areolate areas are indicated as dashed lines. Setae on pleurae and forcipules are omitted. A, Arrup

dentatus, /, adult, Hokkaido; B, Takashimaia ramungula, ?, adult, Honshu; C, Mecistocephalus tahitiensis, /, adult, Palawan; D,

Mecistocephalus microporus, /, subadult, Cebu. Scales¼ 1mm.



here is very low. In some geophilids, the tergum is strongly

reduced, as in mecistocephalids.

75. Forcipular segment, tergum, areolation: (0) uniform; (1)

evident along the anterior and the lateral margins,

virtually absent on the remaining surface; (2) evident

along the anterior and the lateral margins and along 2

paramedian stripes, virtually absent on the remaining

surface.

In most mecistocephalid species, the areolation on the

forcipular tergum is virtually limited to a broad band run-

ning along the anterior and lateral margins (Fig. 12B). In a

few species, it also extends along two paramedian stripes. In

a few other species, the entire surface of the tergum is

uniformly areolate. Because of interindividual variability,

both in the extension and the appearance of the areolation,

each species was coded according to the most frequent

pattern. This character does not apply to scolopendro-

morphs because in this group an independent forcipular

tergum is not recognizable (see ch. 74). In lithobiomorphs

and Adesmata, the areolation is so weak that the entire

tergum appears to be uniformly areolate.

76. Forcipular segment, tergum, mid-longitudinal sulcus: (0)

virtually absent; (1) present.

In mecistocephalids, a mid-longitudinal furrow may be

present on the anterior part of the forcipular tergum

(Fig. 12B). Its occurrence was checked under incident

light. Its presence is quite variable among conspecific indi-

viduals, but the species with a distinct sulcus may be distin-

guished from those in which no more than a trace is

detectable. This character does not apply to scolopendro-

morphs because in this group an independent forcipular

tergum is not recognizable (see ch. 74). In both lithobio-

morphs and Adesmata, no furrow is visible.

77. Forcipular segment, pleuron, dorsal ridge: (0) virtually

absent, not sclerotized; (1) present, well sclerotized.

In some mecistocephalid species, a shallow dorsal ridge

runs longitudinally on each forcipular pleuron (Fig. 12C).

Species bearing distinct sclerotized ridges were distin-

guished from those in which no more than a trace was

detectable under incident light. In all other centipede

groups, the forcipular pleura do not have ridges.

78. Forcipular segment, pleuron, scapular point: (0) reaching

the anterior margin of the coxosternum; (1) not

reaching the anterior margin of the coxosternum.

The position of the anterior tips of the forcipular pleura

(scapular points) in relation to the forcipular coxosternum

was observed from above after removing the head. In some

mecistocephalid species, the pleura extend forward to reach

the anterior margin of the coxosternum and in some cases

they overreach the margin (Fig. 12C). In other species, the

scapular points are far from the coxosternal margin

(Fig. 12A).

79. Forcipular segment, pleuron, scapular point: (0) not

elongate; (1) elongate.

In a few mecistocephalid species, the anterior tip of each

forcipular pleuron (scapular point) is evidently elongate as a

slender projection (Fig. 12B). In the other species, as well as

in the outgroup taxa, it appears stout (Fig. 12A).

80. Forcipular segment, coxopleural sutures: (0) anterior

part dorsal, posterior part ventral; (1) entirely ventral.

In all mecistocephalids, the coxopleural sutures, i.e. the

sutures between the forcipular coxosternum and the pleura

of the same segment, run ventrally in their posterior part,

turning dorsally in their anterior part (Fig. 12). The condi-

tion is the same in all lithobiomorphs and in a few Ades-

mata. In scolopendromorphs and most Adesmata, these

sutures run on the ventral side for their whole length.

81. Forcipular segment, coxosternum, chitin-lines: (0) absent;

(1) present.

Chitin-lines are a pair of sclerotized paramedian lines

present on the ventral surface of the forcipular coxoster-

num. They are characteristic of some Adesmata, e.g. geo-

philids, but they are always lacking in mecistocephalids.

82. Forcipular segment, coxosternum, condylar processes:


In a few mecistocephalid species, a pair of elongate for-

ward pointing projections emerge from the anterior margin

of the forcipular coxosternum, close to the dorsal condyli of

the articulations between the coxosternum and the trochan-

teropraefemur (Fig. 12B). These projections are here called

the condylar processes. They may be well sclerotized, slen-

der and sinuous. Some interindividual variability in the

degree of elongation occurs. Similar projections are com-

pletely absent in most mecistocephalid species (Fig. 12A), as

well as in the other centipede groups, at least in the out-

group species considered here.

83. Forcipular segment, coxosternum, anterior marginal

teeth: (0) absent; (1) present.

In most mecistocephalid species, a pair of short teeth is

present on the anterior margin of the forcipular coxoster-

num (Fig. 12A), but these teeth are variably distinct among

individuals. In a few species, the teeth can be considered as

virtually absent because only two shallow processes are

present. In lithobiomorphs, scolopendromorphs and Ades-

mata, no teeth of this kind are present. Anterior plates, each

with two to eleven tiny teeth, are inserted on the anterior

margin of the coxosternum in all lithobiomorphs and in

some scolopendromorphs. These plates, however, are not

considered homologous to the teeth of mecistocephalids

because their anatomical position is not the same.

84. Forcipular segment, coxosternum, cerrus, lateral groups

of setae: (0) absent; (1) present.

In mecistocephalids, a system of short setae (cerrus) is

usually present on the dorsal surface of the forcipular coxo-

sternum (Crabill, 1970). It appears quite variable in extent and

pattern, but it can easily be described as composed of a lateral

group and a paramedian row on each side (Fig. 12D), usually

well separated but sometimes coalescent. The cerrus is often

virtually absent in juveniles: the setae appear and increase in

number during growth. The cerrus is regularly present in some

species, but completely absent in others. The occurrence and

pattern of the cerrus were observed after detaching the head

from the trunk. The occurrence of the lateral groups is quite

independent from that of the paramedian rows (Fig. 12B). In

the outgroup taxa, at least in the species considered, no setae

are present on the dorsal surface of the coxosternum. In



lithobiomorphs, a few small setae are present on the anterior

plates only, but these are not interpreted as a cerrus.

85. Forcipular segment, coxosternum, cerrus, paramedian

rows of setae: (0) absent; (1) present.

In some mecistocephalid species, the cerrus (see ch. 84)

includes, in addition to the lateral groups of setae, two

paramedian rows of setae. These rows are typically conver-

gent forwards, either straight or widely curved (Fig. 12D).

Usually the setae are quite precisely aligned, but in Mecis-

tocephalus punctifrons and other species they are scattered

inside two wide bands.

86. Forcipular segment, trochanteropraefemur, length to

width ratio: (0) 1.7–1.9; (1) 0.6–1.2; (2) 1.3–1.6;

(3) 2.0–2.1.

The elongation of the forcipular trochanteropraefemur

was expressed as the ratio between the length and the max-

imum width of the article. The length was conventionally

measured in dorsal view, after detaching the head, as the

distance between the two condyli (Fig. 12). The width was

measured at mid-length, perpendicular to the lateral sides.

To cope with possible allometric changes and interindividual

variability, the ratio was calculated, whenever possible, from

series of conspecific adults. Within mecistocephalids, four

intervals were tentatively recognized. In lithobiomorphs and

scolopendromorphs, the trochanteropraefemur is not very

elongated, whereas in Adesmata different conditions occur.

87. Forcipular segment, ratio of trochanteropraefemur length

to coxosternum width: (0) 0.2–0.6; (1) 0.6–0.7; (2) 0.7–0.8.

The ratio between the length of the forcipular trochanter-

opraefemur and the width of the corresponding coxoster-

num was used to describe the general elongation of the

forcipules with respect to the body. Both measures were

taken in dorsal view after detaching the head (Fig. 12): for

the trochanteropraefemur, the distance between the dorsal

condyli; for the coxosternum, the maximum transverse

width between the lateral sides. To cope with possible allo-

metric growth and interindividual variability, measures

were taken, whenever possible, from series of conspecific

specimens. Within mecistocephalids, three intervals were

tentatively recognized. In all lithobiomorphs, scolopendro-

morphs and Adesmata, forcipules are typically stout with

respect to body width.

88. Forcipular segment, trochanteropraefemur, proximal

tooth: (0) virtually absent; (1) present.

In all mecistocephalids, a weak incomplete sutural line

runs on the internal side of the trochanteropraefemur, prob-

ably marking the ancestral joint between two separate arti-

cles. In some species, a sclerotized tooth (here called the

proximal tooth) develops just proximal to the sutural line

(Fig. 12C). The proximal tooth is usually reduced compared

with the distal tooth (see ch. 89), although it exhibits con-

spicuous intraspecific variability. In some species, however,

it is virtually absent and the internal surface is only slightly

swollen (Fig. 12B). Lithobiomorphs, scolopendromorphs

and Adesmata do not have proximal teeth.

89. Forcipular segment, trochanteropraefemur, distal tooth:

(0) virtually absent; (1) present.

In all mecistocephalid species, the trochanteropraefemur

bears a distal tooth, which emerges at the end of the internal

side of the corresponding article (Fig. 12A). It is usually the

most conspicuous tooth of the forcipule. In lithobiomorphs,

the trochanteropraefemur always lacks teeth, whereas in

scolopendromorphs a distal tooth is sometimes present.

The pattern is more diverse in Adesmata.

90. Forcipular segment, articulation between trochantero-

praefemur and tarsungulum: (0) absent; (1) present.

In most Epimorpha (except for some Cryptops; Borucki,

1996), the trochanteropraefemur and the tarsungulum of

each forcipule are directly articulated by means of a scler-

otized hinge on the external side (Fig. 12). The intermediate

articles, i.e. the femur and the tibia, are thus incomplete. In

all other centipede groups, the trochanteropraefemur and

the tarsungulum are never directly in touch. This feature

has long been recognized as a synapomorphy of Epimorpha

(see, e.g. Dohle, 1985; Borucki, 1996).

91. Forcipular segment, tibia, tooth: (0) virtually absent;

(1) present.

In most mecistocephalid species, a sclerotized tooth is

present on the internal side of the third article of the for-

cipule (Fig. 12B). It is usually small and variable in size.

Thus, its occurrence was checked, whenever possible, in

series of conspecific specimens. In lithobiomorphs, scolo-

pendromorphs and Adesmata, the forcipular tibia is always

devoid of teeth.

92. Forcipular segment, tarsungulum, teeth: (0) none; (1) 1;

(2) 2, dorsal and ventral, respectively.

In all mecistocephalids, the forcipular tarsungulum is

typically swollen at the base of the internal side. In some

species, one or two sclerotized teeth emerge at this level

(Fig. 12C). These teeth are usually tiny compared with

other forcipular teeth, but intraspecific differences occur.

No teeth are present on the forcipular tarsungulum of

lithobiomorphs and scolopendromorphs. In Adesmata, the

condition is diverse.

93. Forcipular segment, calyx: (0) reaching the intermediate

articles only; (1) reaching the distal part of the

trochanteropraefemur; (2) reaching the trochantero-

praefemur mid-length.

In all centipedes, a channel runs inside the forcipule from

the poison gland to the tarsungulum, where it opens on the

external side close to the tip. The innermost part of this

channel, bearing short secondary channels, is called the poi-

son calyx. In mecistocephalids, the calyx is very elongate,

uniformly wide but not perfectly straight (Fig. 12). In most

species, it ends at the level of the distal part of the trochan-

teropraefemur. In a few species it is shorter, reaching the

intermediate articles only. In other species, it deepens to the

mid-length of the trochanteropraefemur. Slight variation

also occurs among conspecifics. This character was evaluated

by observing the forcipules from above in adequately cleared

specimens. In lithobiomorphs, the calyx is quite deep,

whereas in scolopendromorphs it is similar to most mecisto-

cephalids. In most Adesmata, it does not overreach the

intermediate articles (Foddai & Minelli, 2000).



Leg-bearing trunk

94. Leg-bearing trunk, segments: (0) heteronomous;

(1) homonomous.

All geophilomorphs are characterized by a homonomous

leg-bearing trunk: segments are very similar to one another,

with the exception of the forcipular segment and the ter-

minal ones. Morphological changes along the body axis are

gradual and a mid-length transition or anomaly is detect-

able in some groups only (Demange, 1963; Minelli et al.,

2000). In all lithobiomorphs and scolopendromorphs, a

homonomous pattern is present only on the ventral side of

the trunk, whereas lateral and dorsal sides are characterized by

a heteronomous pattern involving tracheal spiracles and terga.

95. Leg-bearing trunk, praetergum: (0) absent; (1) present.

In all geophilomorphs, each segment is covered by two

dorsal sclerites and the anterior one (praetergum) is shorter

than the posterior one (tergum). In both lithobiomorphs

and scolopendromorphs, each segment is covered by only

one sclerite (tergum); transverse lines actually separate a

shorter anterior part from a longer posterior part, but this

pattern is probably not homologous to that of geophilo-

morphs (see Eason, 1964 vs Lewis, 1981).

96. Leg-bearing trunk, praesternum: (0) absent; (1) present.

In all Epimorpha, the main ventral sclerite (sternum) of

each leg-bearing segment is accompanied by an anterior

sclerite (praesternum), which is shorter than the sternum

and is often reduced to two separate paired elements. Only

the first leg-bearing segment lacks a praesternum. Praes-

terna and sterna are traditionally considered different scler-

ites, but their actual separation has been questioned (Fusco

& Minelli, 2000b). Mecistocephalus microporus has long

been considered as lacking the praesternum of the last leg-

bearing segment (Attems, 1928, 1947; Wang, 1951; Crabill,

1964), but our observations proved this is not true (Bonato

et al., 2001). In lithobiomorphs, praesterna are not recognizable.

97. Leg-bearing trunk, tracheae, longitudinal anastomoses:


In all Epimorpha, the tracheae of each leg-bearing seg-

ment are fused to those of other segments, producing inter-

segmental anastomoses or, at least, intersegmental bridges

between adjacent stigmata of the same side (Fuller, 1960;

Manton, 1965). In lithobiomorphs, tracheae of different

segments do not match. This feature was checked by obser-

ving cleared specimens from the dorsal side.

98. Leg-bearing trunk, tracheae, transverse anastomoses:


In all geophilomorphs, right and left tracheae anasto-

mose within each leg-bearing segment (Fuller, 1960;

Manton, 1965; Minelli, 1985). Similar anastomoses are not

present in lithobiomorphs and are rare in scolopendro-

morphs. This feature was checked by observing cleared

specimens from the dorsal side.

99. Leg-bearing trunk, sternum, endosternal process:


In all mecistocephalids, each sternum of the trunk has a

tongue-shaped posterior process, completely covered by the

following sternum and usually called the endosternal pro-

cess (Brolemann, 1930). The endosternal processes are more

extended in the anterior part of the trunk, gradually

decrease in length towards the posterior part and com-

pletely fade at about mid-length of the trunk. These pro-

cesses were observed through the integument in adequately

cleared specimens. Endosternal processes are absent in

lithobiomorphs, most scolopendromorphs and all Adesmata.

In cryptopids, similar processes are possibly homologous to

the endosternal processes of mecistocephalids. Worth notice

are similar processes in the extinct devonobiomorphs, basal

to scolopendromorphs (Shear & Bonamo, 1988).

100. Leg-bearing trunk, sternum, sternal apodema and mid-

longitudinal sulcus: (0) absent; (1) present.

In all mecistocephalids, the sternum of each leg-bearing

segment is thickened along a mid-longitudinal line, corres-

ponding to an internal projection which extends from about

the mid-length of the sternum to the end of the endosternal

process. This structure is called the sternal apodema or

rhachis (Crabill, 1959, 1964, 1970). The external surface of

the sternum is enfolded in correspondence to the apodema,

forming a mid-longitudinal sulcus, visible under incident

light. No such feature is known in the other centipede

groups, although the longitudinal thickenings or furrows

found on the sternal surface in some Strigamia species

(Geophilomorpha: Linotaeniidae), would be worth closer

attention, as would the possibly homologous sulci found in

Cryptops, which are, however, very weak and do not reach

the posterior margins of the sterna.

101. Leg-bearing trunk, sternum, mid-longitudinal sulcus:

(1) not furcate; (2) furcate.

In some mecistocephalid species, the mid-longitudinal

sulcus of each sternum divides forwards into two short

diverging branches. This bifurcation is more evident in the

most anterior segments, fading gradually towards the pos-

terior part of the trunk. The width of the angle between the

two branches gradually changes along the trunk. The occur-

rence of this bifurcation was checked under incident light.

This character does not apply to the outgroup taxa, because

they lack a mid-longitudinal sulcus. The presence vs

absence of the bifurcation has long been used as an

important taxonomic character in mecistocephalids, but

some inconsistencies appeared in the literature. For

instance, all Tygarrup species have a weak bifurcation, but

different authors describe it as either completely absent

(in most species), only shortly branched (in Tygarrup

takarazimensis; seeMiyosi, 1957; Titova, 1965, 1975) or evident

(in Tygarrup intermedius Chamberlin, 1914; see Chamberlin,

1914). Furthermore, some Tygarrup specimens were erro-

neously identified as belonging to Mecistocephalus (see, e.g.

Pocock, 1891; Silvestri, 1895, 1919; Attems, 1907).

102. Leg-bearing trunk, sternum, pore field: (0) absent;

(1) present.

In most mecistocephalids, the sterna of the trunk are not

pierced by the usual sternal pores found in most geophilo-

morphs. Only males (but not females) of some species of

Tygarrup have pores, grouped into pairs of lateral elliptical

fields which decrease in area towards the posterior part of

the trunk (Verhoeff, 1942; Titova, 1983). No sternal pores



occur in lithobiomorphs, scolopendromorphs and neogeo-

philid geophilomorphs. Most of the other groups of Ades-

mata are characterized by peculiar pore fields, without any

sexual difference. These pores are the openings of glands

which produce sticky secretions and cyanide derivatives

(Hopkin & Gaywood, 1987; Hopkin et al., 1990; Hopkin

& Anger, 1992; Turcato et al., 1995). These products are

thought to repulse predators and parasitic micro-organisms.

The homology between the male pore fields of Tygarrup and

those of most geophilomorphs is dubious (Foddai & Minelli,

2000; Bonato & Minelli, 2002). Until now, pores were

recorded in only two species of Tygarrup, i.e. T. poriger

from the Himalayas and T. javanicus originally described

from Java (Verhoeff, 1942; Crabill, 1968).

103. Leg-bearing trunk, leg, tarsus: (0) divided into 2

articles; (1) undivided.

In all mecistocephalids, the telopodite of all the legs,

except the last pair, is invariably composed of five articles

(trochanter, praefemur, femur, tibia and tarsus) and ends

with a sclerotized claw. The same condition is found in all

other geophilomorphs. In many lithobiomorphs (but not in

Monotarsobius and various Henicopidae), the telopodite

comprises six articles, with two articles (tarsus I and tarsus

II) corresponding to the single tarsus of geophilomorphs.

Scolopendromorphs are characterized by a pattern of arti-

cles similar to that of geophilomorphs, but an independent

trochanter is not detectable and the legs of the penultimate

pair have two tarsal segments, in Cryptops at least.

104. Leg-bearing trunk, ratio of leg I to leg II length:

(0) 0.7–0.9; (1) 0.3–0.6.

In all mecistocephalid species, the first pair of legs is

shorter and more slender than the following legs, but two

degrees of size reduction can be distinguished. This was

estimated as the ratio between the length of the telopodite

of the first pair and that of the second pair. This ratio

ranges from 0.3, e.g. in Mecistocephalus angusticeps, to

0.9, e.g. in Mecistocephalus satumensis Takakuwa, 1938

[according to Ribaut (1914) and Takakuwa (1938), respect-

ively]. In lithobiomorphs, scolopendromorphs and Ades-

mata, the first pair of legs is always smaller than the

following pair, but the difference is not so consistent as in

most mecistocephalids.

Last leg-bearing segment

105. Last leg-bearing segment, sternum, lateral notches:


In some mecistocephalid species, the sternum of the last

leg-bearing segment is characterized by a pair of weak con-

strictions, at about mid-length of the lateral margins

(Fig. 13C). These notches are variably marked among con-

specific individuals and anomalous specimens occur. In other

mecistocephalids, as well as in the other centipede groups, the

last sternum is usually shaped as a simple shield, e.g. tri-

angular or trapeziform, without any lateral notch (Fig. 13B).

106. Last leg-bearing segment, sternum, pillowlike posterior

process: (0) absent; (1) present.

In some mecistocephalid species, the sternum of the last

leg-bearing segment appears as a simple shield, with a regu-

larly curved posterior margin (Fig. 13B). In other species,

the sternum has a short pillowlike posterior projection

(Fig. 13A). The relative size of this process is affected by

both inter- and intraspecific variation and anomalous speci-

mens occur. Thus, this character was checked, whenever

possible, in series of conspecific specimens. In lithobio-

morphs, scolopendromorphs and Adesmata, the last ster-

num is usually simple, without a posterior process.

107. Last leg-bearing segment, sternum: (0) not constricted

by the coxopleura; (1) constricted by the coxopleura.

In a few mecistocephalid species, the sternum of the last

leg-bearing segment is peculiarly elongate and constricted

between the coxopleura, so that the lateral margins are

concave rather than convex. In all other centipede groups,

no similar shape occurs. Different authors described this

shape in different ways, thus failing to recognize the actual

similarity among some species [cf. the illustrations ofMecis-

tocephalus diversisternus, M. mirandus, M. takakuwai and

M. smithii by Silvestri (1919), Verhoeff (1934), Takakuwa

(1940) and Takashima & Shinohara (1952)].

108. Last leg-bearing segment, posterior part of the sternum

and internal part of the coxopleura, thick pilosity:


In all mecistocephalids, short and dense setae cover the

posterior part of the sternum and the postero-internal parts

of the coxopleura, on the ventral side of the last leg-bearing

segment (Fig. 13). These setae become gradually longer and

less dense towards the anterior part of the sternum and

towards the lateral parts of the coxopleura. The condition

is usually similar in males and females. Although the

setae increase in number during growth, the thick pilosity

is already visible in juveniles. In lithobiomorphs, scolo-

pendromorphs and Adesmata, no differentiated area is

recognizable.

109. Last leg-bearing segment, coxopleuron, coxal pores on

the dorsal surface: (0) absent; (1) present.

The coxal pores are the external openings of peculiar

organs placed inside the coxopleura (Fig. 13). The anatomy

and function of these organs were not investigated in mecis-

tocephalids, but they most probably play a function in

water balance, as in lithobiomorphs, scolopendromorphs

and Adesmata (Rosenberg, 1982, 1983a,b; Littlewood,

1983; Rosenberg & Bajorat, 1984). In most mecistocephalid

species, the coxal organs open on the ventral and lateral

sides of each coxopleuron. Only in a few species do some

pores open on the dorsal side as well. The number and area

covered by coxal pores increase during growth. Thus, the

pattern was evaluated in adult specimens. In lithobio-

morphs and scolopendromorphs, the coxal pores are ven-

tral and lateral only, whereas in Adesmata the distribution

patterns are very diverse.

110. Last leg-bearing segment, coxopleuron, macropore:


Mecistocephalid species differ in the adult pattern of

coxal pores, in terms of number, size and position. These

differences correspond to differences in the developmental



schedule of the pores. In some species, juveniles hatch with

no pores, but they soon develop a set of similar pores on

each coxopleuron. In a few species, however, one large pore

(macropore) develops before all other pores and remains

recognizable as the largest one throughout growth

(Fig. 13A). This pattern was well studied in Dicellophilus

carniolensis (Verhoeff, 1902–1925). There are no sexual dif-

ferences. The occurrence of this developmental pattern was

evaluated by observing either juveniles (in which the macro-

pore is more conspicuous) or grown individuals (in which

the macropore may be distinguished from the other pores).

A similar developmental pattern is unknown in lithobio-

morphs, scolopendromorphs and Adesmata.

111. Length ratio of last leg to penultimate leg: (0) 1.2–2.0;

(1) 2.2–3.0.

In all mecistocephalid species, the telopodites of the last

pair of legs are longer than those of all other legs, but to a

different degree. The length ratio between the last and the

penultimate legs was taken as a measure of relative elonga-

tion. Sexual dimorphism in the last pair of legs, when pre-

sent, does not affect this ratio.

112. Last leg: (0) as slender in males as in females;

(1) swollen in males, slender in females.

The legs of the last pair are similar in both sexes in some

species of mecistocephalids, but are evidently swollen in the

males of other species. This sexual dimorphism is even

A B C

Fig. 13. Last leg-bearing segment and terminal segments (ventral view) of some mecistocephalids. Only the left telopodite is illustrated. Setae

on the right coxopleuron and on the terminal segments are omitted. A, Dicellophilus latifrons, /, subadult, Honshu; B, Tygarrup

muminabadicus, /, subadult, Kashmir; C, Mecistocephalus punctifrons, /, subadult, India. Scales¼ 0.5mm.



stronger in many Adesmata, the last legs differing some-

times in both width and structure.

113. Last leg, short setae: (0) absent; (1) present.

In some mecistocephalid species, some short but strong

setae are interspersed among the common setae on the

telopodites of the last pair of legs, particularly on the ven-

tral side (Fig. 13A). In these species, short setae are present

in all individuals, but their density may be affected by age

and sex. Similar setae are also present in scolopendro-

morphs and Adesmata, at least in the species considered

here as outgroups, but are absent in lithobiomorphs, even

though short sensilla occur in the same position.

114. Last leg, short setae: (0) as dense in males as in

females; (1) more dense in males than in females.

In mecistocephalid species bearing short setae on the last

pair of legs (see ch. 113), the pattern of these setae may be

either similar or different in the two sexes. When sexual

dimorphism occurs, the short setae are more numerous

and dense in males than in females. This character does

not apply to species lacking short setae. The pattern of

setae is often sexually dimorphic in Adesmata, but not in

scolopendromorphs.

115. Last leg, praetarsus: (0) a well-developed claw; (1) a

group of short spines; (2) 1 short slender spine;

(3) none.

In most mecistocephalid species, the legs of the last pair

bear a short and slender spine on the tip. This spine is

usually interpreted as homologous with the claws of the

other legs and the well-developed praetarsus of the last

pair of legs in other centipedes. The spine may not be

detected in all specimens of a given species. In other mecis-

tocephalids, the praetarsus is represented by a stout

rounded tubercle, crowned by small spines. In some other

species, no spine or tubercle is present. Because of the large

interindividual variation, we examined this feature, when-

ever possible, in many specimens of each species. In all

lithobiomorphs and most scolopendromorphs and Ades-

mata, the praetarsus of the last legs is a well-developed

claw. In some scolopendromorphs and Adesmata, however,

it is either reduced or absent.

Terminal segments

116. Genital segment, female gonopods: (0) of 3 articles,

with claws; (1) of 2 articles, without claws.

In all adult mecistocephalids, gonopods are short two-

article appendages (Fig. 13). Male gonopods are slender and

well separated, whereas female gonopods are subtriangular

and usually in touch. Gonopods of females, in particular,

are undeveloped in the first juvenile stages, and have only

one article when they first appear. Female gonopods are

variable in Adesmata: for instance, a pair of two-article

gonopods similar to those of mecistocephalids are present

in the himantariids, whereas only a very rudimentary struc-

ture is present in geophilids. In all lithobiomorphs, the

female gonopods are well-developed three-article appen-

dages with accessory spurs. In scolopendromorphs, these

appendages are virtually absent. From the literature, it is

apparent that developmental changes were often over-

looked, so that the gonopods of mecistocephalid females

were erroneously assumed to be separated in some species

and in contact in other species, and either of one or two

articles (see, e.g., Haase, 1887; Attems, 1929; Verhoeff,

1937, 1939; Lewis, 1991).

Number of segments

117. Number of leg-bearing segments, interindividual vari-

ation at adult stage: (0) absent; (1) present.

In most mecistocephalid species, the number of leg-bear-

ing segments is the same in all specimens, without any age-

or sex-related variation. The number, however, may be

different in different species (see Introduction). To date,

interindividual variation is known only in Mecistocephalus

microporus, where the number of leg-bearing segments

ranges from ninety-three to 101 at least (Bonato et al.,

2001). However, this character has been coded as unknown

for species known from one specimen only, either directly

observed by ourselves or recorded in the literature. Intra-

specific variation in the adult stage is unknown in lithobio-

morphs and scolopendromorphs, whereas it is documented

in most Adesmata and it is supposed to be a common

features of this latter group.

118. Number of leg-bearing segments, number at adult stage:

(0) 15; (a) 21; (b) 41; (c) 43; (d) 45; (e) 47; (f) 49; (g) 51;

(h) 53; (i) 57; (j) 59; (k) 63; (l) 65; (m) 93–101;

(n) 43–47; (o) 83–111.

Different codes have been assigned to all different numbers

of leg-bearing segments recorded in the species considered.

For species with intraspecific variation (see ch. 117), different

codes have been assigned to different ranges of variation.



Appendix

2.Data

matrix

forthecladisticanalysis.

Specieswereenteredin

random

order.Speciesnames

correspondto

those

inTable

1.Character

numberscorrespondto

those

inAppendix

1.Plesiomorphic

statesare

coded

as’0’;

unknownstatesare

indicatedbyaquestionmark

(?).Whereacharacter

wasim

possible

todefineordoes

notapply

toaspecies,itisindicatedbyadash

(-).Characters117and118are

relatedto

segmentationandwereusedasoptionalin

ouranalyses(see

Materialsandmethods).

0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001111111111111111111

0000000001111111111222222222233333333334444444444555555555566666666667777777777888888888899999999990000000000111111111

1234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678

Lithobiusforficatus

000–00000000000000–00–00?000000000––000000000??000000000000000000–0000000000000000000000000000000000–000000000000–0000

Krateraspismeinerti

011?11??1000?1111112130030?01120011100001011130?120001111??3101111102000?2??10?00?1??200111?11111111101?000100?????10d

Arrupdentatus

011211111000111111111122300011300131000010111100110011111011100011001100011000000010020011111111111110100001000010210b

Mecistocephalussp.A

0112110110000111111112003001113011120000101111001310010112021011111020000211100000100011111211111111101?0101000?0–210e

Cryptopsanomalans

001–2100001–111000–00–00?010000000––000001000??000000102000000000–00–0000–––000100000200010010011010–01000000000100–0a

Mecistocephalussp.B

?11111010000011111111200300111301112000010111200131101011001102111102000021111000011131111121111111110110101000?1??10g

Dicellophilusanomalus

011111011002111111100222201011200121101001111201120001021113100111001010121010000111010011121111111110100101010111110b

Tygarrupsp.A

01121111100001111110021130001120011100001011120012000101100210111100200002101100001002001110111111111111000100000–210c

Mec.(M

.)angusticeps

111?11?1100001111111120030?1113011120000101113001310010212?2101111102000021110?00?1??011111??111111110110101000?0–210e

Krateraspissselivanovi

011?11??1000?1111112130030?01120011100001011130?120001021??310111110200002??1??00?1??200111?11111111101?000100?0???1?h

Mec.(M

.)guildingii

0112110?000001111111120030011130111?0000?01112001311010210?210111110200002111??00?1??321111?011111112010010110010–210f

Dicellophiluslatifrons

011111011000111111100222101011200121101001111201110001021112100111001010121010000111010011121111111110100001010111110b

Mecistocephalussp.C

0111110100000111112112003011113011120000001114001311010111021021121020001221110000111011111111111111201?0101101?1?21?h

Geophilusinsculptus

01111111101–111100–00–00?000000000––0000111?100000101000000000010–0001000100000110000100010101111100–11000000001110–1n

Mec.(P.)superior

111?11??10000111111112003??111??111201000?1114001211010010?210211210200002??10?00?1??011111??1111111201?01011010???10f

Tygarruptakarazimensis

011111112000011111100211300011200111000010111300130001011002101111002000021011000010020011101111111110110001000?0–210d

Mec.(M

.)conspicuus

1111110?100001111111120030?1113011120000101112001311010111?210111110200002?110?00?1??211111??11111111011010100?????10f

Arrupholstii

011211011000111111111122300011300131000010111300120011111011101011011101011000000010020011112111111110110001100011210b

Mec.(M

.)diversisternus

01111101000101111112122131011130011200000011130013110112110210211110200002211110011103211111111111111011001100001?21?i

Mec.(M

.)itayai

111111??100101111112120030?1113011121000101113001311011211?210211110200002??10?00?1????11112?1111111201?00?1?0?????10f

Anarrupnesiotes

011111??1100111111100221?0?0111??1?100?0??1113101210010110?211110–002000121010000?1??100111??1111111201?0001?001???10b

Mecistocephalusjaponicus

01111101000101110112122131011130011100000111130013110112111310211110200002111010011100211110111111111010001110001031?k

Mec.(B.)mirandus

01111101000101110112122131011130011200000111130013110112111310211110200002211010011100211100111111111010001110001?31?l

Proterotaiwanella

sculptulata

011111??10001111111111023100112001?10001101113011?0001??1???10?1110?2?0002?0??00001????01110?111111110100001000????10f



Appendix

2.Continued.

0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001111111111111111111

0000000001111111111222222222233333333334444444444555555555566666666667777777777888888888899999999990000000000111111111

1234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678

Mec.(F.)longichilatus

011111??10??01111111120010?11100111?000010111100131?011310?410111110200002??1??00??????1111??1111111101???01000????10f

Mec.(M

.)modestus

111111111000011111111200300111301112000010111100131101011002101111102000021111000010000111120111111110110101000011210f

Mec.(M

.)multidentatus

011111010000011111211200300111301112000000111300131101021003102112102000021111000001101111111111111120111101101010210f

Mec.(M

.)nannocornis

011111010000011111111200310111301112000010111300131001011202100111102000021111000010020111121111111110110101000010210d

Mecistocephalusbenoiti

111111000000011111211200301111201112000000111300131101011201102112102010021111000011101111122111111120110101000010210f

Agnostruppaucipes

011?11??00001111111112023000112001410000101112011300011212?21021111111010??0??00001??2001101?1111111101?000100?????1?b

Mec.(M

.)tahitiensis

01111101100001111111120031011130111200001011110013100101120210111210200002111100001000011112111111111011010100000–210e

Dicellophiluscarniolensis

011111011100111101100222101011200121101001111101110001001002101111001000121010000011010011111111111110100001010111110c

Mec.(M

.)takakuwai

01111101000101111112120131011130011200000011130013100112110210211110200002111010011103211112111111111011001100101031?j

Mec.(M

.)punctifrons

111111000000011111211200301111311112000000111300121101021202102112102000021110000011101111121111111120110101100010210f

Tygarrupmuminabadicus

11111111210101111110021130001120011100001011130012000101100210111100200002101100001002001110111111111110000100000–210d

Mec.(M

.)mikado

11111101100001111121120030111121111200000011130012110101100210111210200002111000001110011112111111111011110110100–210f

Mecistocephalusmicroporus

01111100000001111111120031111130111200001011130012110102100310211210200002111110001110211111111111112011100110100–211m

Nannarruphoffmani

01121101001–1111111111023000113001?10000??11110?1100011110?110111101110101?00000001002001101111111111010000100001?210b

Stigmatogaster

gracilis

111111110000111100–00–2??000000110––0000101103000000101100?000210–0001100000000110000100110011111100–1100000100111211o

Sundarrupflavipes

011111011100111111100221?0101111112100101011131012100101100211110–002000121010000001010011122111111110110001100?1?110b

Proterotaiwanella

tanabei

01111101101–111111111102310011300131000110111301120001011002102111012000020110000111001011100111111110100001000?1?11?d

Mec.(D

.)subgigas

1111110110000111112112003101112111220100011114101311010111021021121020001211110000111011111??11111112010010111001?210f

Takashim

aia

ramungula

0111110120001111011002003111113001?1000010111300120001011202101111102000021110100111020011101111111110110001000011210d

Dicellophiluslimatus

011111011002111111100222201011200121101001111101120001021112100111001010121010000111010011121111111110100001010111110d

Tygarrupjavanicus

01121111200001111110020131001120012100001011130012000101100210111100200002100100001002001110111111111111000100000–210d

Mecistocephaluslongiceps

011?11??00000111111112003??111?011?20000101113001?1?01??1???10?11?10200002??1??00?1??3211112?11111112011??01?0?0??21?f

Mec.(M

.)lifuensis

011?11??000001111111120030?1112011120000101112001311010210?210111110200002??1??00?1??011111??111111120110?01?0?0??210g

Arruppylorus

011211??100011111111112230?0113001?10000101111001200111110?110101100110101?0???00?1??2001111?1111111101000010000???10b

Mec.(M

.)spissus

011211010000011111111200301111301112000010111200131001011202101112102000021111000?1002011111?1111111101101010000??210d

Tygarrupanepipe

011211??10000111111002?13??0112001?1000010111?001200010?10??10?11100200002????000?1????0110??11111111?1?00010000??210d



Appendix 3. Main synapomorphies supporting the clades of the phylogenetic trees.

Relevant characters and state transitions are indicated for each clade. Clade numbers correspond to those plotted on the trees

(Figs 2 and 3). Character number and state codes correspond to those in Appendix 1.

Clade Character: state changes

1 3: 0!1 15: 0!1 90: 0!1 96: 0!1 103: 0!1

2 2: 0!1 16: 0!1 94: 0!1 95: 0!1

3 102: 0!1 117: 0!1

4 18: 0!1 29: 0!1 30: 0!1 34: 0!1 49: 0!1 69: 0!1 100: 0!1

5 70: 0!1 72: 0!1

6 22: 2!1 31: 2!3

7 53: 0!1 64: 1!0

8 68: 1!0

9 74: 1!2

10 115: 2!1

11 22: 2!1 40: 0!1

12 23: 0!2 27: 0!1 35: 3!2 39: 0!1 73: 0!1

13 24: 2!1 31: 2!1 47: 0!1 48: 1!0 51: 0!1 62: 0!1 65: 1!0

14 37: 0!1 41: 1!0 42: 0!1 110: 0!1

15 58: 0!1 59: 0!1 71: 0!1 82: 0!1

16 12: 0!2 25: 1!2

17 13: 1!0 24: 2!1 35: 3!1 48: 1!0

18 7: 0!1 9: 1!2 23: 0!1 78: 0!1

19 102: 0!1

20 4: 1!2

21 9: 2!1

22 67: 0!1 76: 0!1

23 22: 2!3 60: 2!3

24 28: 0!1 31: 2!3

25 20: 0!1 33: 0!1 36: 1!2 50: 2!3 51: 0!1 88: 0!1 106: 0!1

26 9: 1!0

27 86: 2!0

28 113: 1!0

29 –

30 –

31 1: 0!1

32 –

33 – -

34 84: 0!1

35 9: 1!0

36 56: 1!2

37 –

38 46: 2!3

39 12: 0!1 20: 1!2

40 23: 0!2 24: 0!1 26: 0!1 33: 1!0

41 115: 2!3

42 17: 1!0 42: 0!1 59: 0!1 60: 2!3 92: 2!0 109: 0!1

43 78: 0!1 85: 0!1

44 –

45 –

46 19: 1!2 27: 0!1 41: 1!0 66: 1!2

47 109: 0!1

48 1: 1!0

49 60: 2!3 105: 0!1

50 32: 0!1

51 –

52 38: 0!1 46: 3!4



Evolutionary trends and patterns in centipede segment number based on a cladistic analysis of...

Documents

Transcript of Evolutionary trends and patterns in centipede segment number based on a cladistic analysis of...