590.5

FI

V. 70:1-4

1976-77

cop. 3

590 .i'-^' ?^ nJ-M^i^.

C T - X.

FI f(?
V.70 t
no.l

Ccf,3

FIELDIANA
Zoology

Published by Field Museum of Natural History

Volume 70, No. 4 AprU 29, 1977

A Phylogeny of the Sea Snakes (Hydrophiidae)

Harold K. Voris

Assistant Curator, Division of Amphibians & Reptiles
Field Museum Of Natural History

ABSTRACT

The purpose of this study has been to derive in an explicit fashion a phylogeny of
the sea snakes (Hydrophiidae). In all, 632 specimens representing 50 nominal species
of sea snakes, were examined. A total of 153 quantitative and qualitative characters
were collected from specimens representing most of the species. The size of the overall
data set was reduced on the basis of redundancy among the characters and missing
data among some species, resulting in a data matrix of 40 species by 43 characters.
Character state trees were constructed for the 43 characters.

The combinatorial method of Sharrock and Felsenstein (1976) was used in a unique
way to organize the species into groups sharing the same set of character states.
Character states were grouped together according to their degree of derivativeness
(i.e., character state tree layer) and the 40 species were analyzed one character state
tree layer at a time. The combinations generated at each character state tree layer
were surveyed using several operational criteria and certain combinations were se-
lected for further consideration. The selected combinations were related in a flow-
chart diagram from which the relationships of the snakes were extracted. A phenetic
analysis using the simple matching coefficient and single linkage is included for
purposes of orientation and comparison.

The sea snakes consist of three major stocks: 1) the Laticauda, 2) the Aipysurus and
Emydocephalus, and 3) all other species. The data suggest that these groups may
have independent origins among the elapids or a single origin with a very early
separation. The relationships within the three stocks are discussed in detail.

INTRODUCTION

The Hydrophiidae is a group of marine snakes which have an-
terior non-rotatable fangs, neurotoxic venom, and flattened oar-like
tails. They occur from the east coast of Africa to the western coast of
Central America with most species found in tropical and subtropical
Asian seas (Minton, 1968; see also fig. 7). They are chiefly neritic

Library of Congress Catalog Card No.: 76-53446

US ISSN 0015-0754 -.TMf.ii Hm«uiu Mmun.

NATORAL HISTORY SURVEY

Publication 1258 79

AUG 1 1 1977
LIBRARY

80 FIELDIANA: ZOOLOGY, VOLUME 70

and frequent mangrove swamps, estuaries, bays, river mouths, coral
reefs, and coastal water situations (Voris, 1972).

The Hydrophiidae were recognized as a cohesive group of 40 spe-
cies of snakes by Boulenger in 1896. Since that time new species
have been discovered but no species have been moved in or out of the
family.^ Malcolm Smith published two major taxonomic works on
this family. In 1920 he described 23 species of sea snakes in a paper
which included a complete list of the specimens he examined and
the data he collected from them. A brief introduction was concerned
with miscellaneous notes on natural history, zoogeography and
methods of character collection. Smith's Monograph of the Sea
Snakes (Hydrophiidae), published in 1926, included descriptions of
nearly every species currently known, about 50. Many skulls and
some heads were figured. Individual specimens examined were
listed with some of the data collected from them. Smith also in-
cluded introductory remarks on the evolution of the sea snakes and
on the "Cranial characters," "External characters," and the "Hemi-
penis," and a section on the habitat and zoogeography. Keys to both
the genera and species were given.

Since Smith's work of 1926, studies on the sea snakes have been
limited in scope. There have been cursory studies on the ecology
(e.g.. Saint Girons, 1964; Kropach, 1971) or on the anatomy and
physiology of a few selected species (e.g., Bal and Nawathe, 1949;
Dunson and Ehlert, 1971). Taxonomic studies have also been lim-
ited, either to geographic areas (e.g., Vols0e, 1939; Wang, 1962) or to
small sample sizes and/or only a relatively few consistently applied
taxonomic characters (e.g., Underwood, 1967; McDowell, 1972).
None of the classifications has been based on more than a few se-
lected "key" characters and taxonomic procedures have remained
obscure.

The purpose of this study has been to derive in an explicit fashion,
a highly tenable phylogeny of the Hydrophiidae. My aim is not to
defend exhaustively the methods used to derive the phylogeny, but
rather to assure that the methods utilized are procedurally and logi-
cally defined, repeatable and therefore open to testing and criticism.
This work is an offshoot of a much larger study (Voris, 1969) which
included the comparison and evaluation of several numerical and
orthodox phylogenetic methods. The methods used here were in part
selected from the larger study.

^McDowell (1972), however, has put all sea snakes within the Elapidae. See p. 123
for comments on classification.

VORIS: PHYLOGENY OF SEA SNAKES 81

MATERIALS AND DATA COLLECTION

Data on external anatomy were collected from 632 preserved sea
snakes representing 50 of the 52 nominal species. The average
number of specimens examined per species was 10.9; the median
was 7. In addition, 134 skulls representing all 16 genera and 42 of
the species, 88 tail x-rays representing all the genera and 46 species,
and 90 whole body x-rays representing all the genera and 44 species
were studied. A list of species and a summary of specimens exam-
ined is given in Table 1. A total of 153 characters was collected from
the above material. Table 2 lists the characters and arranges them
into several categories related to the nature of the characters and
their location on the snakes.

The characters selected for study were chosen on the basis of one
of four criteria. First, some characters were chosen because they
were previously in wide use in taxonomic literature. Second, charac-
ters were chosen simply on the basis of accessibility. That is, if a
character showed variation within the family and was easily ob-
served, it was included. Third, characters were chosen for special
attention because they were related to specific aspects of the biology
(such as behavior or development) of the organisms. Fourth, an ef-
fort was made to obtain a sample of characters that would reflect as
large a proportion of the phenotype as practical (see table 2).

Traditional techniques were usually employed in the examination
of the various types of characters. Modifications of these techniques,
problems encountered, and further details concerning the materials
examined (including a complete list of specimens) and character
selection are presented elsewhere (Voris, 1969).

Additional Sources of Data

Occasionally, when samples were small and when Smith had seen
specimens not available to me, it seemed desirable to pool his and
my data to make the delineation of some species more reliable. Prior
to doing this, comparisons were made between my observations and
Smith's (1920) on several series of the same specimens. Compari-
sons of statistics on Smith's and my data were also made. When
Smith and I had examined the same specimen, I included the data
from that specimen in my data set and excluded it from his when
making the calculations. In the majority of samples. Smith's and my
observations appeared to be fully comparable and pooling of data in
these species was judged to be permissible (see Voris, 1969, pp. 16-
17 for statistical comparisons).

82 FIELDIANA: ZOOLOGY, VOLUME 70

DESIGNATION OF CHARACTER STATES

Several of the numerical approaches utilized in this study re-
quired discrete character states for each Operational Taxonomic
Unit (OTU). In this study, all OTU's were species. Also, after states
were initially designated, some pooling of states occurred because
the computer programs used accommodated only characters with
nine or fewer states.

Character states were designated as follows. For qualitative char-
acters, states were defined and designated as the different states
were encountered. In cases in which a species showed two or more
discrete states or a gradation between states, intermediate states
were designated.

Quantitative data were treated more extensively. Although
sample sizes were small, several basic statistics (range, mean, and
standard deviation) were calculated wherever possible. For each
character the mean value of the specimens comprising the sample
was chosen to represent that species.

The boundaries of quantitative character states were defined in
two ways. First, the species' means were plotted in ascending order
and the resulting curve examined. The following criteria were used
to decide where character state boundaries would be designated:

1. State boundaries were set at sharp changes in the slope.

2. Boundaries were set at gaps in the distribution of points
along the curve.

3. When the values over all species spanned a narrow range,
and there were no sharp changes in the slope, nor breaks in
the curve, states of about equal size range were established.
(Those characters with wide ranging values were treated
using arithmetic incrementation (see below)).

4. Additional states were designated for uncertain cases. Thus,
similar character states would be oversplit but rarely would
distinct states or groups be consolidated.

A second method of defining states using successive incrementa-
tion was applied to the characters whose states spanned a wide
range (referred to in number 3 above). There are several different
methods of incrementation that are designed to take into account
the observation that with larger measurements an increase in vari-
ation is frequently noted (Kendrick, 1964). In many such situations
a specific method of incrementation is justifiable, however, all cases

VORIS: PHYLOGENY OF SEA SNAKES 83

do not fit a single scheme. Also, there are a few species in which
higher values are not accompanied by increases in variation. For
example, in Kolphophis annandalei the number of ventrals is great
yet variation is low. In this study arithmetic (fixed) incrementation
was employed. In arithmetic incrementation, the number of data
points included in a state is increased by a constant value as succes-
sive states are designated. For example, if one is incrementing by
two, as was done in this study, and has data values running from 1
to 25, the following five states would result:

Data values 1; 2,3,4; 5,6,7,8,9; 10,11,12,13,14,15,16; 17,18.. .24,25

Number of data
points included
in each state 13 5 7 9

The character states designated for all 153 characters are de-
scribed in Voris, 1969, Table 2.

PHYLOGENETIC ANALYSIS

Phylogenetic studies are composed of three general processes.
Primary data collection is the first, consisting of the collection of
data from the organisms and the designation of character states.
The second process is the manipulation of these data, including all
operations performed on the data through the formation of a phylo-
geny. The third process is the construction of a classification consis-
tent with the phylogeny and preferably based directly on the data
analysis.

This study is mainly concerned with the second of these three
processes. It deals with the recognition of species and procedures for
determining relationships among them.

Species Determination and Recognition

The first task after data collection was to designate species. As a
theoretical basis for delimiting species, the biological species defini-
tion of Mayr (1942) and Dobzhansky (1951) was accepted. However,
in this study, everything concerning reproductive isolation and
community gene pools was inferred from data other than breeding
tests. That is, two sympatric populations represented by adults of
both sexes collected over a several month period and having phe-
netic differences that could not be attributed to polymorphism (i.e.,
there were two or more characters which were complex, and showed

84 FIELDIANA: ZOOLOGY, VOLUME 70

no intergradation) were interpreted as having genetic differences
that have resulted from reproductive isolation of the populations.

In the case of populations that are not sympatric, or for which
certain samples were unavailable, i.e., certain age classes, one sex,
or some ecotypes were missing, I estimated whether or not the popu-
lations were reproductively isolated from the degree of difference
observed between them as compared to the differences observed
between well established sympatric species.

Specimens were first roughly grouped according to the above cri-
teria. Next, these groups were compared to Smith's (1926) species
and usually were found to correspond with them. For example.
Smith's and my data show that bisexual adult samples of the sym-
patric species Aipysurus eydouxi (Smith, 19 specimens; Voris, 6
specimens) and A. laevis (Smith, 15 specimens; Voris, 4 specimens)
differ by non-overlapping states in characters 4, 73, 77, 82, 84-88
(see table 2). Microcephalophis gracilis (Smith, 10 specimens; Voris,
13 specimens) and M. cantoris (Smith, 4 specimens; Voris, 5 speci-
mens), also sympatric species, differ in the number of ventral body
scales and in the relationship of the prefrontal to the third suprala-
bial scale.

In three pairs of species my data, and Smith's data where applica-
ble, indicated that the character states of two species overlapped
broadly. The pairs were: Laticauda laticaudata (1) and L. crockeri
(5); Emydocephalus annulatus (13) andE. ijimae (14); and. Lapemis
curtus (26) andL. hardwickii (25). The first two pairs have been left
separate because the available sample was very small in one species
of each pair. The third pair is thought to represent conspecific popu-
lations (Voris, 1969, Appendix B). However, these nominal species
were retained as individual units in this study for practical purposes
and it will become evident that doing so did not significantly affect
the results. In sum, the absence of intergradation between sympat-
ric forms was the operational basis for recognition of the 50 species
of sea snakes included in this work (table 1).

Reduction of the Data Set

For several compelling reasons, the entire data set was not used
in all aspects of this study, i.e., a subset of the 153 characters was
prepared for certain analyses beyond the species level. Thus, it is
appropriate to consider the logic and procedures used to reduce the
size of the data set.

A number of characters are obviously repetitious and introduce

8a

V^^/

8b

8c

8d

8e

Fig. 1. Outlines of the major types of premaxillary bones found in the Hydrophi-
idae (character 137). Numbers refer to designated character states (table 3).

85

86 FIELDIANA: ZOOLOGY, VOLUME 70

undue weight to one or another aspect of the phenotype. The task of
detecting and documenting character redundancy was approached
systematically through an analysis of character association. Char-
acters may be correlated for many reasons, one of which is redun-
dancy. Correlations were measured using the chi-square statistic
(Voris, 1971) and in the event that two characters showed high cor-
relation the following possibilities of redundancy were investigated.

Redundancy due to physical factors of measurement. — ^This tj^^e of
redundancy occurs when two or more measurements or character
states describe the same feature (Davis and Heywood, 1963, p. 130).
It is a simple matter to detect characters which are 100 per cent
redundant for this reason. For example, it is redundant to use both
the measurements of surangular (character number 119) and den-
tary length (120) and total jaw length, because in sea snakes, the
jaw length equals the surangular length plus dentary length. This
relationship may not necessarily hold true for newly discovered spe-
cies and thus character relationships must always be re-evaluated
when new organisms are added to a study. That is, 100 per cent
correlation between characters in a small set of data does not justify
assuming the same correlation in a larger data set.

Redundancy due to character repetition. — These associations are a
result of characters being descriptions of the same tjrpe of feature.
There are examples of complete and partial repetition of characters
in the sea snakes. For example, in most species of sea snakes the
number of scale rows around the body varies along the length of the
body. However, in some species of Laticauda the scale row count
remains nearly constant throughout the length of the snake so that
all five counts are identical and effectively 100 per cent repetitious
after character coding. When changes in number of scale rows occur
they generally take place gradually. Smith (1926) made a scale-row
count at the neck and in several places around the mid-body to find
the maximum row, and was able to separate some species on this
basis. After making survey counts I found that if counts were made
at the neck, one-fourth, one-half, and three-fourths the number of
ventral scales from the neck, and at 10 ventrals anterior to the vent
(character numbers 84-88), a few additional bases for species sepa-
ration were found. Thus it is conceivable that one might want to
make scale row counts at every tenth ventral to study the scale row
increases more thoroughly or even count every row. However, the
latter procedure would add 300 to 400 characters to the study and
effectively "swamp" the other characters.

VORIS: PHYLOGENY OF SEA SNAKES

87

Fig. 2. Outlines of the three major types of nasal bones (character 138) found in
the Hydrophiidae. Numbers refer to designated character states (table 3).

The choice as to how many characters of this type and which ones
are retained remains for the moment with the judgement of the indi-
vidual taxonomist. In this study redundancy of this type was mini-
mized in the reduced data set.

Redundancy due to mechanical relationships. — This type of re-
dundancy has been called necessary correlation (Cain and Harrison,
1960) and refers to characters that are mechanically related to one
another. An obvious example in the sea snakes would be the two
characters, number of pairs of enlarged chin shields (81), and the
number of enlarged chin shields touching (82). In character 81, zero
pairs of chin shields is necessarily 100 per cent correlated with zero
pairs touching in character 82. One pair of shields will also be asso-
ciated with one or zero pairs touching. An example in this study of
character association where redundancy is not certain but a possible
factor would be the association between the position of the nostril (2)
and the presence or absence of the internasal scale (1).

In instances in which a group of characters was found to be redun-
dant for one of the above reasons all but one of the characters were
removed from the data set. The question of which character was
retained in each case is not crucial but the decision was based on an
evaluation of the reliability and completeness of the data on the
characters. Character redundancy was the principal and first crite-
rion applied to reduce the character set.

Two state characters which have one state almost universally dis-
tributed and the other state highly restricted in distribution, are not
helpful in a study of most species' relationships. This is true irre-

Fig. 3. Four major types of rostral grooves found among the Hydrophiidae. State 1
as in Laticauda colubrina (FMNH 13817); State 2 as in Laticauda semifasciata
(FMNH 75169); State 3 as in Aipysurus duboisii (British Museum 1926.5.28.26);
State 4 as inLapemis curtus (Universitetets Zoologiske Museum R 66149).

88

VORIS: PHYLOGENY OF SEA SNAKES 89

gardless of which state, the common one or the rare one, is deriva-
tive. Character number 6, gular azygous scale present or absent,
and number 7, anterior prefrontal azygous scale present or absent,
are examples of these kinds of characters in that only one species of
sea snake exhibits the presence state of each of the two characters.
These characters were removed from the data set because they lend
no relational information about the taxa.

Some characters and some species were removed from the data set
in order to obtain a nearly complete data matrix. This was necessary
because it has been learned in the process of these studies, that even
an apparently small amount of missing data could have a signifi-
cant impact on the results. Thus it was decided on the basis of exper-
ience that no species should have more than one missing data point.
Species for which skull preparations were not possible could not be
retained without eliminating all skull characters. In addition some
characters, particularly hemipenis characters and measurement
characters, were uncollected in some species. Thus 10 species were
removed from further consideration in the study (species nos. 5, 16,
29, 32, 35, 37, 38, 39, 43, 52; recall also that species 8 and 42 were
not initially included, see table 1).

Practical reasons such as computer program restrictions might
necessitate further reduction of the data set in some studies, but in
this one the reductions made on the other grounds were sufficient to
avoid making reductions on this basis alone.

The data set which remained after applying these various criteria
consisted of 43 characters on 40 species (table 3). It is worth noting
that all characters not eliminated for a specific reason were re-
tained. Thus the process of reduction of the data set was one of elimi-
nation and not selection. Some characters from every area and por-
tion of the snakes' morphology were retained and are represented in
the reduced data matrix (table 4).

Relationships Among the Sea Snakes

All methods employed to deduce relationships of organisms use
phenetic data. However, the methods which have been developed to
manipulate these data vary drastically procedurally and philosophi-
cally.

Phenetic and phylogenetic methods can most easily be distin-
guished on the basis of objectives. Phenetic methods are largely con-
cerned with estimating "overall similarity" (Sokal and Sneath,

90 FIELDIANA: ZOOLOGY, VOLUME 70

1963, p. 3). Phylogenetic methods on the other hand, are concerned
with estimating geneological relationships (Throckmorton, 1968).
Generally, methods associated with phylogenetic studies involve
many more implicit and/or explicit assumptions.

A Phenetic Treatment as a Point of Reference

Today the taxonomist has a vast number of phenetic methods
from which he can choose (see Sneath and Sokal, 1973; or Jardine
and Sibson, 1971). Numerous comparative studies have been con-
ducted (e.g., Sokal and Michener, 1967; Boyce, 1969; Voris, 1969),
and from them one generality has seemed to emerge: there is no
single phenetic method that has general enough properties to sat-
isfy more than a few of the practitioners. Rather, the tendency has
been to recognize that each method supplies a distinct "window"
into the nature of the relationships. Sneath and Sokal (1973, p. 147)
after a lengthy discussion of estimations of taxonomic distances,
came up with only one recommendation regarding the choice of
methods, "of each type of coefficient considered, the simplest one
should be chosen out of consideration for ease of interpretation."

The results of a phenetic analysis, using the simple matching
coefficient (Sokal and Michener, 1958) with single linkage or
nearest neighbor clustering (Sneath, 1957) are presented here to
give the reader a relatively assumption-free view of the phenetic
relationships of the sea snakes.

The simple matching coefficient (S.C.) is calculated for each pair
of species as follows:

g (-J _ Number of character states shared
Total number of characters

Descriptions of single linkage clustering are given elsewhere (Voris,
1969, Appendix D; or Sneath and Sokal, 1973).

These procedures were selected for two reasons. First, both the
simple matching coefficient and the single linkage clustering
method have been used widely since the late 1950's (see Sneath and
Sokal, 1973, p. 132). Thus, from an intuitive standpoint, few
methods have been more widely tested and evaluated. Second, these
methods are among the simplest of all numerical taxonomic
methods and thus the interpretation of the results is relatively
straightforward.

The data matrix presented in Table 4 was used in the phenetic
analysis. The relationships which resulted are presented in the form

SIMILARITY LEVEL

_4:

1.0

I

1

Lat lat

2

Let col

3

Lat sem

4

Lat sch

6

Aip eyd

13

Emy ann

14

Emy ijj

7

Aip fus

9

Aip lae

10

Aip dub

11

Aip fol

12

Aip apr

31

Eph mer

15

Hyl dar

20

Kol ann

41

Hyd maj

30

Hydkin

21

Ast sto

24

Ker jer

19

Ths ano

18

Aca per

26

Lap cur

25

Lap har

22

Pel pla

23

Enh sch

28

Mic can

27

Mic gra

48

Hyd mam

49

Hyd cae

47

Hyd lap

46

Hyd ino

45

Hyd orn

44

Hyd tor

36

Hyd cya

50

Hyd fas

51

Hyd bro

40

Hyd klo

34

Hyd bel

33

Hyd mel

17

Thn vip

Fig. 4. Dendrogram of the phenetic relationships among 40 species of Hydrophi-
idae, based on 43 characters (table 4), using the simple matching coefficient and
single linkage clustering.

91

92 FIELDIANA: ZOOLOGY, VOLUME 70

of a dendrogram (fig. 4). The Laticauda separate from all others at a
relatively low level of similarity and have high similarity among
themselves. The Aipysurus and E my docephalus also separate early.
The similarity among the species of Aipysurus and E my docephalus
is less than that observed among the Laticauda species and, al-
though the two Emydocephalus are depicted as nearest neighbors to
each other, they are among the Aipysurus species. The remaining
species form almost a continuum on the similarity scale. Within this
group, Ephalophis mertoni and Hydrelaps darwiniensis are the first
to separate. Several monotypic genera and two Australian Hydro-
phis branch off in rapid succession, until finally the majority of the
genus Hydrophis joins at relatively high similarity levels. In sum-
mary, three major groups are indicated: Laticauda, Aipysurus-
Emydocephalus, and all others.

A Phylogenetic Treatment

Both logically and procedurally the derivation of a phylogeny is
complex and problematic. It necessitates making inferences con-
cerning connectedness, direction of evolution, and evolutionary dis-
tance. The method which has been developed in this study to accom-
plish this and which is discussed in the following paragraphs, is not
set forth as the method of deriving phylogeny, but as one attempt to
utilize objectively defined and repeatable procedures to this end.

Character Selection: A Phylogenetic Perspective

Characters that are phylogenetically useful are those whose
states are associated with one another for reasons of common evolu-
tionary history (Davis and Heywood, 1963; Blackwelder, 1967;
Voris, 1971). The task is first to detect those characters that are
highly associated with other characters and then to evaluate the
basis for the observed associations. Although historically important
characters show high association, the converse is not necessarily
true. Characters may show a high degree of association with one
another for reasons other than common history, e.g., the various
types of redundancy described on p. 86. High associations due to func-
tional relationships (adaptive or convergent association) of char-
acters are also possible, but difficult to detect a priori.

By removing from the data set those characters which could be
shown to be highly associated due to various types of redundancy, a
process which constituted residual weighting, I arrived at a data set
which represented many aspects of the phenotype and which I think
has a high probability of containing valid historical information.

VORIS: PHYLOGENY OF SEA SNAKES 93

Character State Trees

The methods used in this study to determine the evolutionary
connectedness of OTU's, direction of evolution among the OTU's,
and the evolutionary distances between the OTU's, utilized char-
acter state trees. Character state trees depict the primitive state
and the sequence and direction of change for the character states.
The character state tree concept is not a new one. The terms used in
this study — "character state tree," "primitive state," "sequence and
direction of change," — correspond to Maslin's (1952, p. 51) terms
"morpho-cline" and "polarity"; Hennig's (1966, p. 95) "character
phylogeny," "transformation series," and "plesimorphous" (primi-
tive) and "apomorphous" (derived) conditions; Throckmorton's
(1962, p. 309) "stepwise sequence," "primitive characteristics," and
"derivative characteristics"; and Camin and Sokal's (1965, p. 312)
"presumed evolutionary sequences."

Primitive state information is obviously critical to the formation
of character state trees. In this study, the primitive state informa-
tion was derived from the contemporary group most closely related
to the hydrophiids, which in the view of the majority of herpetolo-
gists is the Elapidae (Bellairs and Underwood, 1951; McDowell,
1972). The Hydrophiidae is considered to be the most recent group to
be derived from the elapid stock, and thus the Elapidae is in a "pre-
group" position. The evidence can be summarized as follows: be-
cause of certain derived character states common to the Elapidae
and Hydrophiidae (e.g., anterior nonrotatable fang and neurotoxic
venom) the Hydrophiidae are considered to be phylogenetically
more closely related to the elapids than to any other contemporary
group of snakes; and because the Hydrophiidae are highly special-
ized in their way of life, they are considered the more derivative of
the two groups. The elapids have remained terrestrial for the most
part as were their presumed ancestors, a stock common to the Xeno-
peltidae, Boidae, and Pythonidae (Bellairs and Underwood, 1951).

Returning to the specific problem of the determination of primi-
tive states, I determined which of the hydrophiid states were primi-
tive on the basis of their distribution among the elapids and the
number of times a hydrophiid state appeared in the elapids. My
primary criterion was that a hydrophiid state was designated as
primitive if it appeared throughout the elapids, i.e., in many diverse
groups. The reasoning behind this was that it is more logical to pos-
tulate that a widely distributed state is primitive and has been re-
tained throughout the various lineages, than to postulate numerous

94 FIELDIANA: ZOOLOGY, VOLUME 70

independent origins of it. A secondary criterion was invoked when
the appropriate data were available. If the alternative states were
rare and limited in their distribution to highly specialized elapids, I
took this as an indication of derivativeness and a confirmation of the
designation of the primitive state. If a hydrophiid state did not ap-
pear at all among the elapids, there was no basis for considering it
anything but derivative among the hydrophiids. A qualification of
this would be needed if the sea snakes were known to have sepa-
rated from the elapid stock very early in their history, and in that
sense would be as primitive as the elapids themselves. However, the
possibility of this having occurred is not supported by evidence cur-
rently available. It should be noted that assumptions regarding the
absolute occurrence or absence of character state reversal, the
unique derivation of states, or multiple origins of states are not re-
quired here. Rather the logic is based on Occam's razor. Although
multiple origins and reversals of character states as well as unique
origins do occur, most character states have not originated many
times, reversed many times or originated only once, and we thus
assume that when a state is widely distributed throughout various
linages of a group the simplest explanation is that it is a retained
primitive state rather than a derivitive state with many inde-
pendent origins.

Once the primitive state was designated, the rest of the character
states were ordered in a logical numerical or phenetic sequence. For
example, for character 95, number of maxillary teeth, the states in
the sea snakes were designated as: state 1, 0-1 teeth; state 2, 2-4
teeth; state 3, 5-7 teeth; state 4, 8-10 teeth. State 2 was designated
as primitive. Therefore the logical sequence is as follows:
1<^^^3^'4. Character 41, sulcus shape, describes the shape of the
sulcus and has four designated states.

State 1 was designated as primitive and the phenetic sequence des-
ignated as(l^2^3^4. Clearly, these sequences are not the only
possible ones, but sequences are necessary and these examples dem-
onstrate how I proceeded with this task.

The elapid data used came from three sources: Marx and Rabb,
1972; Marx and Rabb, personal communications; and a survey of 24
skull characters on 28 species representing 22 genera of elapids

VORIS: PHYLOGENY OF SEA SNAKES

95

(Voris, 1969, Appendix C). The character state trees for the 43 char-
acters used in this study are given in Table 3.

Arriving at a Consensus of Character State Trees

At this point we have 43 characters, represented by 146 states.
Each of the 40 species of sea snakes has been assigned one state for
each character. For each character a primitive state, connectedness
of states, and direction of evolution, i.e., a character state tree, has
been designated. Thus each character designates its own phylogeny
of the 40 species. For example, character number 4 (parietal frag-
mentation) designates the following relationships (see tables 3 and
4 for the character state tree and the states for each species respec-
tively).

1,2,3,4,5, 13,15,
17,19,21,24,27,
Species: 28,30,31,33,34,

36,40,41,44,45,
46,47,49,50,51

12, 14,20,22,
23,25,26

7,9,10,
11,18

State:

(1)

©

(2)

(3)

Since the character state trees represent the only phylogenetic
information available, it may be argued that a consensus of char-
acter state trees would provide the closest approximation to the ac-
tual phylogeny. However, all character state trees do not designate
the same or even compatible arrangements of species, as may be
seen from a comparison of just two characters, the tree above and
the one given below (Character 46, relation of frontal bone to orbit).
The problem of integrating character state trees to derive a con-
census thus presents itself.

96

FIELDIANA: ZOOLOGY, VOLUME 70

Species:

1,2,3,4,5,6,7,10,
11, 12, 13, 14,22

9,36,50 15,17,18,19,20,21,

23,24,25,26,27,28,
30,31,33,34,40,41,
44,46,47,48,49,51

State:

1)

(2)

0

Several diverse approaches have been applied to use character
state trees to derive phylogenies (e.g., Camin and Sokal, 1965;
Throckmorton, 1965; Inger, 1972; Heyer, 1974). Some of these
studies have dealt with the integration of character state trees
through the manipulation of character-state reversals and the mul-
tiple origin of character states. This study utilizes the Sharrock and
Felsenstein (1976) combinatorial method in a novel way as part of
an attempt to build a consensus phylogeny. This method allows data
on which the phylogeny is based, to be retrieved in their original
form. The combinatorial method has been described and/or used
previously (Liem, 1970; Inger, 1972; Heyer, 1974; Zehren, 1974;
Sharrock and Felsenstein, 1976). However, because the original
manuscript describes the method in somewhat abstract terms and
because other workers have simply cited the later paper a synopsis
of the method is included here.

Sharrock and Felsenstein Combinatorial Method

The computer programmed combinatorial method, as it has most
commonly been used, operates on a binary data matrix of species by
character states. Each species is coded as having (1) or not having
(0) each character state. This binary matrix can be constructed in
numerous ways. For example, all primitive states can be eliminated
and/or species can be coded as possessing all states which they ac-
tually possess plus all those states which are designated by the char-

VORIS: PHYLOGENY OF SEA SNAKES 97

acter state trees as primitive to those states. In phylogenetic studies
assumptions are routinely incorporated when the binary matrix is
prepared and when the combinations generated are analyzed. A
manuscript on the variety of options available and their application
is in preparation (Marx et al., MS in prep.).

The criterion of similarity in the Sharrock and Felsenstein combi-
natorial method is not a similarity coefficient, but rather it is based
on the actual number of the same character states held in common
by all species in a group. These groups of species are called the non-
redundant, monothetic combinations. Specifically, a non-redundant
monothetic combination is the largest group of species sharing a
given set of character states.

An example will help clarify the procedure. Below, the presence
(1) or absence (0) of eight character states^ is presented for three
hjTDothetical species.

Character States

Species

1

2

3

4

5

6

7

8

A

0

1

1

1

1

1

1

1

B

1

0

0

0

0

1

1

1

C

0

0

0

0

1

0

1

1

The computer output for this data set would give the following
information:

Members of

Group Size

Combination

Combination

Number of

(Number of

or Group

Shared

Number

Shared States

Species)

(Species)

Character States

1

7

1

A

2,3,4,5,6,7,8

2

4

1

B

1,6,7,8

3

3

2

AB

6,7,8

4

3

2

AC

5,7,8

5

2

3

ABC

7,8

In the output the nonredundant monothetic combinations are listed
in descending order based on the number of shared character states.
For each combination the number of members within the combina-
tion is listed, the members are listed and the character states shared

'This might represent data from one to four characters depending on the number of
states per character.

98 FIELDIANA: ZOOLOGY, VOLUME 70

by all the members are given. The first combinations usually (as in
this example) are simply single species with unique combinations of
character states. In this example, the third combination reveals that
species A and B share three characters (numbers 6, 7, and 8). The
fourth combination shows that species A and C also share three
states (numbers 5, 7, and 8). Note that the latter suite of characters
is distinct from the suite of characters shared by species A and B. All
three species form a combination sharing character states 7 and 8.

It should be emphasized that the combinatorial method is a tool to
derive groups of taxa with specified characteristics and the method
itself assumes nothing. Its striking advantage is that the actual
characters designating a group of species remain known, since they
are listed with the combination they form, and are not obscured as
in the calculation of a similarity coefficient. This is of particular
significance in that the combinatorial method has the potential to
satisfy one of the criteria set forth earlier: that the data on which
the final phylogeny is based be known. How the combinatorial
method is used to arrive at a consensus of character state trees fol-
lows.

Approach to Combinatorial Method Computer Output

In this study the combinatorial method has been applied to the 40
species of sea snakes six separate times, once at each character state
tree layer, each time independent of the states at other layers in the
character state trees. The character state tree layers are defined as
on p. 99, with two character state trees used as examples.

The most common way of coding data for use with the combina-
torial method was briefly mentioned earlier, namely coding the data
to imply that all states primitive to the state actually exhibited by a
species were also possessed by it. While this allows the character
state tree to be incorporated into the coded data to some extent, it
has a significant disadvantage in that in effect it differentially
weights those characters which have the greatest number of states,
giving them more influence on the results on that basis alone
(Heyer, 1974; Zehren, 1974). Another possibility, investigated at
one point by me, is to code each state as present or absent in a
straightforward way, but this is purely phenetic and effectively
eliminates the phyletic information provided by the character state
tree. The approach used in this study overcomes both of these objec-
tions in that each state is used once and only once, and it also allows
incorporation of tree information in that the character state tree

VORIS: PHYLOGENY OF SEA SNAKES 99

Character State Trees (2 different characters)
Layer

;

;

;

/

Sixth or sexternary (comprised of those states
derived directly from the fifth layer
states)

Fifth or quinquenary (comprised of those
states derived directly from the fourth
layer states)

Fourth or quaternary (comprised of those
states derived directly from the third
layer states)

Third or tertiary (comprised of those states
derived directly from the second layer
states)

3 2 2 3 Second or secondary (comprised of those

♦ "^ 'T "^ states derived directly from the first

\ / \ / layer states)

®/7\ First or primary (comprised of primitive

vi/ states)

layers are considered sequentially. It may also be argued, that for a
given species, the least advanced character state tree layer(s) (most
primitive) will tend to designate largely phenetic relationships
while more advanced layers (more derivative) tend to reveal its phy-
letic relationships (Hennig, 1966).

The first character state tree layer, composed of all primitive
states\ consisted of a total of 43 states, that is one state for each
character. The second layer consisted of 65 states. This value ex-
ceeds the number of characters because bifurcating and trifurcating
character state trees contribute two and three states respectively to
this layer (as in the example given above). The third, fourth, fifth,
and sixth layers contained 23, 10, 3, and 1 character states respec-
tively. Since each species is represented by only one state per char-
acter, the maximum number of states which any single species could

'This study differs from many previous ones in that primitive states are included
and used in the data analysis rather than eliminated. The argument for including the
primitive states is that the concept of primitive states is a relative one. This approach
is in part an outgrowth of Hennig's (1966) discussion of the relative nature of char-
acter states to each other, and Throckmorton's (1968) "operational primitive" con-
cept. The full argument for the use of primitive states is presented in much greater
detail in a forthcoming study (Marx et al., MS in prep.).

100 FIELDIANA: ZOOLOGY, VOLUME 70

possess is 43 (less if the species is missing data on a character). The
number and percentage of states each species possesses at the var-
ious character state tree layers is of some interest and value in later
aspects of the analysis (table 5).

Applying the combinatorial method to each layer of the character
state trees resulted in the generation of 688 combinations of species
at the first layer, 1,661 combinations at the second layer, 182 combi-
nations at the third layer, 17 combinations at the fourth layer, 4
combinations at the fifth layer, and 1 combination at the sixth layer.

When the combinations generated by the combinatorial method
are surveyed, two types of combinations may be observed: first,
there are combinations of species whose members are not parts of
any other combination at the same number of shared states. In this
data set, a second type of combination is also often observed: these
are combinations in which some members are found in one or more
other combinations formed at the same level of shared character
states. The existence of these two types of combinations suggests
that there are some groups of species which are quite distinct from
the others at a given character state tree layer, and other groups
which are diffuse and whose species form more or less a continuum
of phenotypes. Groups of species which are distinct in this sense
from other groups and formed with high numbers of shared char-
acter states, are likely valid phyletic units. The "contested" combi-
nations are also important because they illustrate the degree and
way in which various groups are inter-related.

A set of systematic procedures has been developed to be applied to
the computer output of the combinatorial method for each character
state tree layer to detect this kind of structure in the data set. The
result is a "flow-chart" of species relationships at each character
state tree layer. The procedures were developed in accord with two
general principles: they must be consistent and repeatable, and they
must expose the relationships of the species as represented by the
sometimes large number of possible combinations, as completely as
possible without incorporating obfuscating redundancy. The proce-
dures used to accomplish this are described below:

1. At each character state tree layer the combinations of spe-
cies generated by the combinatorial method are scanned begin-
ning at the highest number of shared states and proceeding
level by level down through the lowest level of shared states,
detecting and diagrammatically representing structure among
the species according to the following criteria:

VORIS: PHYLOGENY OF SEA SNAKES 101

2. A single species or a combination of species is considered to
have formed an uncontested group at a given level of shared
states, if none of those species are found in any other group at
that same level of shared states.

3. A group or combination is considered to be contested if one or
more of the species in that group are also found in one or more
other groups at that same level of shared states.

4. As groupings at lower levels of shared characters are consid-
ered, groups once formed are never broken up, whether they
were first formed as contested or uncontested groups (although
the contested-uncontested status itself may change, see below).
That is, as lower levels of shared states in common are consid-
ered, groups designated at higher levels may enlarge to form
new groups by joining one another, or new species may appear
and/or be added on to an existing group, but once a species has
been placed in a group or groups, it is never linked to another
species or group of species to form a new group unless all mem-
bers of at least one of its groups are linked to that new species
or group of species. Thus, contested groups may resolve into a
single, larger, uncontested group at a lower level of shared
states if all members of the two or more contested groups are
found within a single group at that level. Also, uncontested
groups may become parts of contested groups if a whole uncon-
tested group is found in two or more combinations containing
different species as their additional group members at a lower
level.

5. A previously formed group or groups (which may consist of
one or more species and be contested or uncontested) may be
detected at a lower level of shared states within a combination
which also contains some but not all species from one or more
other previously formed groups. According to number 4 above,
the whole combination cannot constitute a new group without
breaking up a previously formed group. However, the informa-
tion provided by such a relationship is an important aspect of
the data set, and so those extraneous species which are parts of
previously formed groups, are listed as neighbors to whatever
whole group exists within the larger combination at that level,
or whatever new group is formed within that combination. The
first time, i.e., the highest level, at which this occurs for each
group, the neighbors are referred to as near or nearest neigh-
bors.

102 FIELDIANA: ZOOLOGY, VOLUME 70

As mentioned previously, the structure detected by the procedure
outhned above was originally recorded in the form of a flow-chart.
Although highly informative, the flow-charts are extensive and
complex and cannot be reduced to the size of the standard printed
page. Thus each flow-chart was converted to a punctuation repre-
sentation. A sample flow-chart for the first few levels of shared
states for the first character state tree layer is presented in Figure 5.
The punctuation representation of the data (explained in the table
heading) is given in Table 6 for the first character state tree layer,
in Table 7 for the second layer, in Table 8 for the third layer, and in
Table 9 for the fourth, fifth, and sixth layers. Comparison of Figure
5 with the beginning of Table 6 will clarify how each representation
may be converted to the other.

Since one of the goals of these procedures was to present the spe-
cies relationships as completely as possible without redundancy, it
is of interest to consider what proportion of the total combinations
generated by the combinatorial method at each character state tree
layer, were included (either as uncontested or contested groups or
near neighbors to groups) in the final representation of the struc-
ture at each level. Of the 688 first-layer combinations, 287 (42 per
cent) are represented in Table 6. Of 1,661 second-layer combina-
tions, 575 (35 per cent) are represented in Table 7. Of 182 third-
layer combinations, 158 (87 per cent) are represented in Table 8. Of
24 fourth-layer combinations, all 24 (100 per cent) are depicted in
Table 9, and similarly 4 out of 4 (100 per cent) fifth-layer combina-
tions, and 1 out of 1 (100 per cent) sixth-layer combinations are
included in Table 9.

The next task, the integration of the structure from each of the
character state tree layers (Tables 6-9) into a final consensus phylo-
geny, was approached by first placing species into one of several
groups based on the relative percentage of character states exhi-
bited at each of the character state tree layers (see table 5 and
below). These groups were formulated only to facilitate discussion,
and whether or not manageable groups of species would be desig-
nated by this procedure in another study is dependent on the data

Opposite:

Fig. 5. Flow-chart representation of selected combinations of Hydrophiidae species
to shared state level 13. For all levels see Table 6. Selection was from those combi-
nations generated by the Felsenstein and Sharrock combinatorial method at the first
character state tree layer.

103

104 FIELDIANA: ZOOLOGY, VOLUME 70

set. For example, a data set with all two-state characters would ob-
viously require a different scheme or criteria in order to separate
the data set into groups small enough for efficient discussion. How-
ever, in this study, in addition to providing small enough groups for
efficient discussion, this basis for grouping species had two further
advantages: it separated species into groups with similar distribu-
tions of amounts of information (character states), and, it directed
attention to the character state tree layer(s) at which each species'
relationships should be best resolved and most reliable.

The following groups have been designated in this study, with
respect to the relative percentages of states at each character state
tree level. ^

L 1st. layer > 2nd. layer > 3rd. layer > 4th., 5th., or 6th. layer

1) Laticauda laticaudata

2) Laticauda colubrina

3) Laticauda semifasciata

A) Laticauda schistorhynchus

n. 1st. layer = 2nd. layer > 3rd. layer > 4th., 5th., or 6th. layer

1 1 ) Aipysurus foliosquama

in. 2nd. layer > 1st. layer > 3rd. > 4th., 5th. or 6th. layer.

6) Aipysurus eydouxi

7 ) A ipys urus fuscus
9) Aipysurus laeuis

10) Aipysurus duboisii

12) A ipys urus apraefron talis
13) Emydocephalus annulatus
14) Emydocephalus ijimae
15) Hydrelaps darwiniensis
19) Thalassophis anomalous
24) Keriliajerdoni

25) Lapemis hardwickii
26) Lapemis curtus

30) Hydrophis kingi

31) Ephalophis mertoni

IV. 2nd. layer > 1st. = 3rd. layer > 4th., 5th. or 6th. layer
18) Acalyptophis peronii
21) Astrotia stokesii
23) Enhydrina schistosa
21) Microcephalophis gracilis
28) Microcephalophis cantoris
33) Hydrophis melanosoma

^Percentage values were considered to be about equal ( = ) when they differed by five
or fewer percentage points.

VORIS: PHYLOGENY OF SEA SNAKES 105

34) Hydrophis belcheri
36) Hydrophis cyanocinctus
40) Hydrophis klossi
4 1 ) Hydroph is major
45) Hydrophis ornatus
46) Hydrophis inornatus
50) Hydrophis fasciatus
51)Hydrophis brookii

V. 2nd. layer > 3rd. layer > 1st. layer > 4th., 5th., or 6th. layer.
17) Thalassophina viperina
20) Kolpophis annandalei
22) Pelamis platurus
44) Hydrophis torquatus
41) Hydrophis lapemoides
48) Hydrophis mamillaris
49) Hydrophis caerulescens

The structure at each of the character state tree layers is dis-
cussed below for each group of species.

Results of Analysis of Combinatorial Method Computer Output

Group I Species
Four species all of the genus Laticauda ilaticaudata (1), colubrina
(2), semifasciata (3), and schistorhynchus (4)) have over 75 per cent
of their states at the primary or first character state tree layer (table
5). Of the 34 or 35 states which each of these species possesses, 30(88
per cent) are shared by all four species (table 6; fig. 5). There is a gap
of 13 character states between this group and the character-state
level at which the next species, E my docephalus annulatus (13), joins
the group. On the basis of the bulk of the phenotype measured, the
genus Laticauda is a tight cluster of species very distinct from all
other taxa.

These species of Laticauda have eight or nine states (about 20 per
cent) at the second character state tree layer (table 5). The relation-
ships designated by these more derived states are compatible with
the relationship designated by the first layer (table 7). However, at
the second character state tree layer, the combination including all
four species of Laticauda does not occur, until the level of three
shared states; and, at the level of three and four shared states, both
species pairs (1, 2 and 3, 4) have near neighbors from the genera
Aipysurus and Emydocephalus. The Laticauda have very little in-
formation above the second character state tree layer (table 5).

Conclusions on Group I Species. — The genus Laticauda is a phe-
netically tight group of closely related species. On the basis of a few

106 FIELDIANA: ZOOLOGY, VOLUME 70

derived states, it appears to have weak phyletic relationships with
several Aipysurus and Emydocephalus species, leaving open the
possibility that these groups are monophyletic. However, at the
same time, because of the weakness of the phyletic relationship, a
very early separation of the groups or even polj^jhyly is implicated.

Group II and Group III Species

Because only one species fell into the group II category, group II
and group III species were considered together. At the first char-
acter state tree layer these species have from 21 states (49 per cent)
in Aipysurus foliosquama (11) to eight states (19 per cent) in La-
pemis hardwickii (25) (table 5). With two exceptions (Lapemis hard-
wickii (25) and Lapemis curtus (26)), at this character state tree
layer the neighbors to each of the group II and group III species are
one or more species of the genus Laticauda (fig. 5; table 6). This
phenomenon is a function of the fact that the Laticauda possess a
very high proportion of primitive states and the group II and group
III species possess the same primitive states, although not as many.
The Lapemis species do not share these states and in this respect
they are more like species in groups IV and V. At this layer the
relationships among the group II and group III species are, for the
most part, complex. Looking beyond the fact that neighbors are pre-
dominantly among the Laticauda, it is evident that some of these
species are very similar to each other. Aipysurus foliosquama (11)
and A. apraefrontalis (12) have 45-49 per cent of their states at this
level and they share most of them, namely 18. Aipysurus fuscus (7)
is similar to both these forms at the level of 16 shared states, but it
is a bit closer to A. foliosquama (11) which is a near neighbor at 17
shared states. It is crucial to note that both A. duboisii (10) and A.
laevis (9) do not have as many states at this level as the previously
mentioned species of Aipysurus, but when they first appear they are
in contested combinations with A. foliosquama (11) and/or A. fuscus
(7). That is, all the primitive states that A. laevis (9) andA. duboisii
(10) have are also possessed hy A. foliosquama (11) and/or A. fuscus
(7). Aipysurus eydouxi (6) stands slightly away from all other Aipy-
surus at this level and the fact that it forms an uncontested combi-
nation with A. duboisii (10) at the level of 12 shared states is clearly
an artifact of the combination selection procedures. That is, both
these species are clearly closest individually (see shared states
levels 16, 15, 14) and as a group (see shared states level 12) to A.
foliosquama (11), A. fuscus (7), andA. apraefrontalis (12). Emydoce-
phalus ijimae (14) has relatively high affinities with the Laticauda ,

VORIS: PHYLOGENY OF SEA SNAKES 107

the various Aipysurus, and^. annulatus (13). Because of these affin-
ities and because E. annulatus (13) shares 17 states with all four
Laticauda forming an uncontested group with them, several con-
tested groups are generated at and below the level of 14 shared
states. Without exception, contested groups and their neighbors are
species of Aipysurus, Emydocephalus, or Laticauda down to the
level of six shared states.

Hydrelaps darwiniensis (15) and Ephalophis mertoni (31) form an
uncontested group with 14 shared states and have neighbors with
only the Laticauda until the level of 10 shared states. Kerilia jerdoni
(24) and Hydrophis kingi (30) do not share more than 12 primitive
states (about 30 per cent) with any species, but at 11, 10, and 9
shared states they are linked with species oi Laticauda (1, 2, 3, 4),
Aipysurus (11), Emydocephalus (13, 14), Hydrelaps darwiniensis
(15), Thalassophis anomalus (19) and Hydrophis major (41). Both
species ofLapemis (25, 26) have only eight or nine primitive states,
of which they share seven with Hydrophis belcher i (34). At six
shared states, Lapemis (25, 26) clusters with several species found
in groups IV and V.

At the second character state tree layer, most group II and group
III species have the bulk of their states, with the range from 18 (42
per cent) in Hydrophis kingi (30) to 29 states (67 per cent) in La-
pemis hardwickii (25) (table 5). In terms of secondary character
states, the genus Aipysurus (species 6, 7, 9, 10, 11, 12) is relatively
compact (table 7). These species have between 26 and 22 secondary
states. Although two uncontested combinations occur, Aipysurus
fuscus (7) with A. laeuis (9) at 21 shared states, and A. duboisii (10)
with A. foliosquama (11) at 20 shared states, the near neighbor rela-
tionships indicate the compactness of the genus. For example, al-
though A. fuscus (7) and A. laevis (9) form an uncontested combina-
tion at 21 shared states, both these species and A. foliosquama (11)
are near neighbors to A. duboisii (10) at the level of 20 shared
states. Although A. eydouxii (6) does not share as many states with
the other Aipysurus (its first near neighbors occur at 18 shared
states) it demonstrates the same general pattern — numerous combi-
nations with other Aipysurus, with no species of other genera
joining until the level at which it joins all other Aipysurus, namely
14 shared states. At this level it also shares overlapping sets of char-
acter states with both species of Emydocephalus (13, 14). The two
species of Emydocephalus possess 21 and 22 secondary states and
they share 18 of them. No near neighbors to this combination occur

108 FIELDIANA: ZOOLOGY, VOLUME 70

until level 14 where it has a group of four species ofAipysurus (9, 10,
11, 12) as its nearest neighbor. All the Aipysurus and Emydoce-
phalus share 12 secondary states and form an uncontested combina-
tion at this level (species 6, 7, 9, 10, 11, 12, 13, 14).

Hydrelaps darwiniensis (15) has 22 secondary states. It has only
one neighbor, Lapemis hardwickii (25) at 17 shared states prior to
forming an uncontested combination with Ephalophis mertoni (31)
at the same level. Ephalophis mertoni (31) has 20 secondary states
and no near neighbors prior to combining with//, darwiniensis (15)
at level 17. At 14 shared states,//, darwiniensis (15) and£. mertoni
(31) have the two Lapemis (25, 26) as near neighbors and at 13
shared states, Hydrophis torquatus (44) (see group V) is their
neighbor. At 12 shared states, many neighbors are designated from
several genera \T\c\\i6.m.g Aipysurus.

The two species of Lapemis (25, 26) share 26 secondary states and
have Thalassophina viperina (17) as a near neighbor at 20 shared
states. Thalassophis anomalus (19) has as its nearest neighbors at
19 shared states, Thalassophina viperina (17) and Hydrophis inor-
natus (46). At this same level of shared states Thalassophis anom-
alus (19) and K. jerdoni (24) form a contested combination. Hydro-
phis kingi (30) initially comes in at 18 states and its nearest
neighbor is another Australian Hydrophis, H. major (41), at level
17. Relationships of these species below 18 or 19 shared states are
complex and involve many of the group IV and group V species.
Species of Laticauda, Aipysurus, and Emydocephalus do not com-
bine with these species here.

The group-II-and-III species have relatively few tertiary states.
The range is from one state (2 per cent) in Aipysurus eydouxii (16)
and A. apraefrontalis (12) to seven states (16 per cent) in Hydrophis
kingi (30) (table 5). Aipysurus duboisii (10) first comes in with four
tertiary states, and at the level of two shared states is a member of
two contested combinations, one of which contains three other Aipy-
surus (species 7, 9, and 11), and the other of which contains the two
Emydocephalus species (13, 14) (table 8). These are all the Aipy-
surus and Emydocephalus which have come in prior to or at this
level of shared states. At this same level of two shared states, the
two states shared in the combination containing the Emydocephalus
are also present in Hydrelaps darwiniensis (15) and Kolpophis an-
nandalei (20), that is, species 15 and 20 are near neighbors to the
group A. duboisii (10), E. annulatus (13),£. ijimae (14) at this level.
At one shared state, two contested combinations of Aipysurus and

VORIS: PHYLOGENY OF SEA SNAKES 109

Emydocephalus (3, 7, 9, 10, 11) and (6, 7, 9, 10, 11, 12, 13, 14) have
neighbors from numerous other combinations. At five shared states
Hydrelaps darwiniensis (15) and Ephalophis mertoni (31) are sepa-
rate but have near neighbors. Hydrelaps darwiniensis (15) is closest
to Kolpophis annandalei (20), while Ephalophis mertoni (31) shares
five states with three species o^Hydrophis (44, 45, 46).

Thalassophis anomalus ( 19) has six tertiary states and shares five
of them with several group IV and group V species. Kerilia jerdoni
(24) has only four tertiary states and shares these with six species
from group IV and group V. The Lapemis (25, 26) share five states
with each other, and L. curtus (26) has a near neighbor of three
other species at the same level. Hydrophis kingi (30) shares all
seven of its tertiary states with // . major (41). Several of the above
relationships will be encountered and discussed in more detail
under the section on group IV and group V species.

Of the group II and group III species, only Kerilia jerdoni (24),
H. kingi (30), andE. mertoni (31) have more than 1 per cent of their
character states at the fourth, fifth, or sixth character state tree
layer. These species will be discussed with the group IV and group V
species with respect to their relationships at these layers.

Conclusions on Group II and Group III Species. — Analysis of the
first layer states shows that the Emydocephalus and Aipysurus
share a very large proportion of their primitive states with the La^i-
cauda. This is not surprising because Laticauda has such a great
proportion of primitive states. On the other hand, Lapemis fails to
overlap the Laticauda in primitive states to nearly the same extent.
The phenomenon of a very large proportion of shared primitive
states is considered to be weak evidence of monophyly for the Lati-
cauda, Emydocephalus, and Aipysurus. Further examination of the
primary state data show the Aipysurus species to be close to each
other; Hydrelaps darwiniensis and Ephalophis mertoni closely
linked; and the Emydocephalus separated and in a sense bridging a
gap between the Laticauda and Aipysurus. Thalassophis anomalus,
Kerilia jerdoni, and Hydrophis kingi show ties with the Laticauda,
Aipysurus, Emydocephalus, among themselves, and to some group
IV and group V species. The Lapemis are clearly associated with the
group IV and group V species.

The secondary and tertiary state data reinforce and clarify the
above relationships. Aipysurus is a compact, monophyletic group of
species. Aipysurus eydouxi, a slightly more generalized species, is on
the periphery. The Emydocephalus species are also monophyletic

110 FIELDIANA: ZOOLOGY, VOLUME 70

and arose out of the early Aipysurus stock. Hydrelaps darwiniensis
and Ephalophis mertoni are separate from all others and have their
closest relationships with the two Lapemis species. Thalassophis
anomalus, Keriliajerdoni, and Hydrophis kingi are related to group
IV and group V species through Thalasophina viperina and the two
Lapemis species. These relationships will be more fully discussed in
the following pages.

Group IV and Group V Species

The species in groups IV and V are discussed together because
both groups lack a large proportion of primary states and because
they show many relationships with each other. These species are the
most derivative, with the proportion of primary states being small,
ranging from 12 states (28 per cent) in Microcephalophis gracilis
(27) to five states (12 per cent) in Astrotia stokesi (21) (table 5).

Microcephalophis gracilis (27) with 12 primary or first layer
states, has two Laticauda (3, 4) as near neighbors at the level of 10
shared states, and all four Laticauda (1, 2, 3, 4) at level 9 (table 6).
Hydrophis major (41) has 10 primary states which it shares with//.
kingi (30), and nine of which it shares with Kerilia jerdoni (24).
These three species share eight states with Hydrelaps darwiniensis
(15) and Thalassophis anomalus (19), and at the level of seven
shared states, Ephalophis mertoni (31) joins this group to form an
uncontested combination. All these species except Hydrophis major
(41) are group III species and have a relatively large proportion of
primary states compared to most group IV and group V species. It is
interesting to note that at the level of six shared states, Laticauda
colubrina (2) is the only near neighbor to this group, and all four
species oi^ Laticauda (1, 2, 3, 4) among some others, are near neigh-
bors to this group at the level of five shared states.

At the level of eight shared states, two species of Microcephal-
ophis (26, 27) independently form combinations with Hydrophis bel-
cheri (34), and the latter is in another combination with Hydrophis
melanosoma (33) and Hydrophis fasciatus (50) at the same level of
shared states. At seven shared states, Acalyptophis peroni (18)
forms separate groups with two Laticauda (3, 4); Hydrelaps darwin-
iensis (15); Kerilia jerdoni (24); and Microcephalophis gracilis (27).
At this level of shared states and below, the relationship of the
group IV and group V species becomes very complex. The number of
contested groups directly reflects the degree to which the species are
inter-related with one another in terms of primary states. For exam-

VORIS: PHYLOGENY OF SEA SNAKES 111

p\e,Hydrophis melanosoma (33) and Hydrophis belcheri (34) share a
suite of seven characters with M. gracilis (27) and H. fasciatus (50),
and a different suite of characters with Hydrophis klossi (40) and
Hydrophis brookii (51) at the same level; and a third combination
contains two species which occur for the first time at level seven:
Hydrophis ornatus (45) and Hydrophis lapemoides (47), as well as
Hydrophis fasciatus (50). It is worth noting that most of the combi-
nations containing group IV and group V species, and sharing six or
fewer states, do not have Laticauda species as near neighbors as did
many of the earlier combinations.

Most of the group IV and group V species have about 50 per cent of
their states at the second character state tree layer (table 5). Micro-
cephalophis gracilis (27) has the fewest with 20 states (47 per cent)
and Astrotia stokesi (21) has the most with 30 states (71 per cent) at
the secondary layer.

Within the six levels between the levels of 27 shared states and 22
shared states, 14 of the group IV and group V species appear for the
first time (table 7). The relationships among these species become
complex almost immediately. Thalassophina viperina (17) with 27
secondary states, joins Hydrophis inornatus (46) at 24 shared states.
Hydrophis ornatus (45) and Hydrophis lapemoides (47) each with 24
secondary states, form an uncontested combination with 23 shared
states. However, at this same level, Hydrophis ornatus (45) has a
second near neighbor, Hydrophis inornatus (46). Thus, although in
terms of uncontested combinations,//, ornatus (45) and//, inornatus
(46) do not appear close, in fact//, inornatus (46) is a near neighbor
to H. ornatus (45). At the level of 22 shared states, the inter-rela-
tionship between these two uncontested groups (species 17, 47, and
45, 47) is further substantiated by the fact that each group has as its
nearest neighbor, a species from the other group. Hydrophis tor-
quatus (44) and Hydrophis fasciatus (50) each with 23 secondary
states, have nearest neighbors to Hydrophis inornatus (46) and
Hydrophis lapemoides (47) of these two groups at the level of 22
shared states. At 21 shared states, four contested combinations exist
involving the above species, and including As^ro^ta stokesi (21). In
addition, Hydrophis cyanocinctus (36), Hydrophis brookii (51), and
Hydrophis klossi (40) form two contested combinations at level 23.
Hydrophis melanosoma (33) and Hydrophis belcheri (34) both come
in for the first time at 22 shared secondary states in a combination
including H. cyanocinctus (36) and //. brookii (51). At 21 shared
states, //. klossi (40) becomes part of the latter group and forms an

112 FIELDIANA: ZOOLOGY, VOLUME 70

uncontested combination. Although this is unquestionably a close
knit group of species, it is also crucial to point out that the combina-
tion of if. melanosoma (33), H. belcheri (34), H. cyanocinctus (36),
and H. brookii (51) has two separate near neighbors at the same
level that it combines with//, klossi (40), namely T. viperina (17)
and//, lapemoides (47). In fact, the nearest neighbors to this entire
group of five Hydrophis (species 33, 34, 36, 40, 51) occur at level 20
and are T. viperina (17) and//, lapemoides (47). It is also true that
one of//, fasciatus (50) near neighbors is //. cyanocinctus (36) at
level 22, and at level 20, three of the four combinations containing
the species T. viperina (17), //. torquatus (44), H. ornatus (45), H.
inornatus (46), H. lapemoides (47), and H. fasciatus (50) have the
near neighbor H. cyanocinctus (36). Thus it is clear that these two
clusters of species which appear in separate series of combinations
are inter-related with one another.

It is critical to note the near neighbors of the various groups of
species at the levels of 21, 20, and 19 shared states. Because groups
of species have formed relatively large combinations of four to five
species by level 21, a species such as Microcephalophis gracilis (27),
M. cantoris (28), Hydrophis mamillaris (48), or Hy drop his caerules-
cens (49) which have only 20, 22, 21, and 21 secondary states respec-
tively, are unlikely to share all their states with a large series of
species and, in fact, do not. Thus, H. caerulescens (49) at level 20,
has as its nearest neighbor H. torquatus (44), and at level 19 has
neighbors of Astrotia stokesi {21); Lapemis hardwickii (25); H. tor-
quatus (44) and//, inornatus (46); and//, cyanocinctus (36),//. klossi
(40), and//, brookii (51). Eventually at a level of 18 shared states//.
caerulescens (49) joins the two Lapemis (25, 26) to form a contested
combination. At the same level it also joins the group H. melano-
soma (33), H. belcheri (34), //. cyanocinctus (36), H. klossi (40), and
H. brookii (51).

At 18 shared states Microcephalophis gracilis (27) andM. cantoris
(28) form a contested combination, with M. cantoris (28) also ap-
pearing in a combination with//, melanosoma (33),//. belcheri (34),
//. cyanocinctus (36),//. klossi (40), and//, brookii (51). At this same
level M. gracilis (27) has a near neighbor, //. brookii (51); and M.
cantoris (28) has three rather large groups of species as neighbors.
At the level of 17 shared states, the group of the two Microcepha-
lophis (27, 28) has a near neighbor of H. melanosoma (33), H. bel-
cheri (34),//. cyanocinctus (36), and//, brookii (51).

Several genera, Acalyptophis (18), Enhydrina (23), and Pelamis

VORIS: PHYLOGENY OF SEA SNAKES 113

(22), do not enter combinations including the other species men-
tioned above, but have near neighbors at relatively low levels with
species within those combinations. For example, at level 20 Enhy-
drina schistosa (23) which forms an uncontested group v/ithAcalyp-
tophis peroni (18), has the near-neighbor As^ro^ia stokesi (21); and
Pelamis (22) has the near-neighbor Thalassophina viperina (17).
The genera Acalyptophis (18), Enhydrina (23), Astrotia (21), Pe-
lamis (22), and Lapemis (25, 26) {the Lapemis being from group III)
can be characterized as having their nearest neighbors several
levels below the maximum level of 24 states which they possess.
Although these monotypic genera do tend to stand off, it is also true
that they are linked to the central groups of IV and V species
through Thalassophina viperina (17).

To illustrate the complexity of the inter-relationships that exist at
or below level 19, we can look specifically at the four contested com-
binations including Thalassophina viperina (17) at 19 shared states.
As mentioned earlier, two of these combinations actually include//.
cyanocinctus (36) as a near neighbor at this same level. The parent
combination {T. viperina (17),//. ornatus [45), H. inornatus (46), and
H. lapemoides (47)) which was formed at the level of 20 shared
states, has near neighbors at level 19 which include Lapemis hard-
wicki (25); H. melanosoma (33), H. belcheri (34), H. cyanocinctus
(36),//. brooki (51); and separately,//, klossi (40). These near neigh-
bors are parts of two uncontested groups (L. hardwicki (25), L.
curtus (26); and //. melanosoma (33), H. belcheri (34), H. cyano-
cinctus (36),//. klossi (40),//. brookii (51)), which occur at level 19.
Both of these uncontested groups have as near neighbors at the
same level, T. viperina (17) and other species from the four contested
groups.

Unlike the group II and III species, the group IV and V species
have a significant number of their states at the tertiary character
state tree level. Astrotia stokesi (21) and Kerilia jerdoni (24) have
the fewest tertiary states, four, while Kolpophis annandalei (20) and
Hydrophis mamilaris (48) each have 11 or 26-27 per cent of their
total states at the tertiary level (table 5).

The tertiary states depict relationships of group IV and V species
that are effectively identical to those depicted by the secondary
states (table 8). Between the levels of 11 and 9 shared states, 16 of
the group IV and group V species occur for the first time. At the
level of eight shared states, there are already 10 contested combina-
tions. Again, as for the secondary states, the near neighbors to spe-

114 FIELDIANA: ZOOLOGY, VOLUME 70

cies and groups of species are crucial to a clear picture of the rela-
tionships. For example, Microcephalis gracilis (27) and M. cantoris
(28) each contain 10 tertiary states. M. cantoris (28) shares nine
states forming a contested combination with H. klossi (40) which
has only these nine states; M. gracilis (27) has no near neighbors at
this level. However, at the level of eight shared states, the nearest
neighbor to M. gracilis (27) is M. cantoris (28) (which it can't com-
bine with since, as indicated, M. gracilis (27) already formed two
groups at a higher level), and if. belcheri (34) with which it forms a
contested group. At this same level of eight shared states, M. can-
toris (28) forms a contested combination with H. belcheri (34), H.
klossi (40), and//, lapemoides (47). Also, at the level of eight shared
states, the two Microcephalophis (27, 28) and//, belcheri (34) are
involved in four contested combinations, andH. torquatus (44) and
H. ornatus (45), and H. inornatus (46) are involved in three con-
tested combinations. These series of combinations are not mutually
exclusive, for example, the contested combination at eight shared
states, of//, belcheri (34) and//, caerulescens (49) overlaps the two
groups of combinations (27, 34) (28, 38, 40, 47) (28, 40, 48) and (44,
45, 46) (44, 46, 47) (45, 47) (47, 49) and, in addition, each of these
species individually has H. torquatus (44) as a near neighbor at the
same level. At the level of seven shared states these groups of combi-
nations overlap broadly in contested combinations (e.g., the combi-
nation M. cantoris (28),//. belcheri (34),//. klossi (40),//. lapemoides
(41), H. mamilaris (48), and//, caerulescens (49)).

At the tertiary level the monotypic genera Acalyptophis peroni
(18), Kolpophis annandalei (20), Pelamis platurus (22), and Enhy-
drina schistosa (23) as at the secondary level, are somewhat sepa-
rated from the central core of species which have been discussed
above. Thalassophina uiperina (17) and K. annandalei (20) share
eight states with each other and they share seven states with sev-
eral of the group IV and group V species involved in contested com-
binations at that level. Although//, fasciatus (50) does not occur in a
contested or uncontested combination until level six, at level seven
it has three near neighbors which designate its affinity to other
group IV and group V species. Hydrophis kingi (30) and//, major
(41) form an uncontested combination at seven shared states, but
the affinity of these two species to the group IV and group V species
is clear. For example, at this same level of seven shared states, H.
major (41) is found with/C. annandalei (20) and//, lapemoides (47),
and both species//, kingi (30) and//, major (41) share a set of six
characters with/C. annandalei (20) and another set of six characters

VORIS: PHYLOGENY OF SEA SNAKES 115

with H. mamilaris (48). An examination of the combinations and
neighbors of the combinations occurring at level six and below, reit-
erates the pattern of inter-relationships described above.

Group IV and group V species have between 0 and 7 per cent of
their character states at the fourth character state tree layer. Thus
the maximum number of states present is three, as in Enhydrina
schistosa (23), i/. kingi (30), and//, melanosoma (33). At the level of
two shared states, eight species come in by themselves for the first
time and six come in for the first time in combination with other
species. With so few character states involved, the relationships des-
ignated are of limited value. It is possibly worth noting that//, kingi
(30) and H. major (41) form an uncontested combination at two
shared states. This repeats the relationships observed between
these two species at previous character state tree layers. In addition,
at this level there are combinations of other Hydrophis species with
each other and with three of the monotypic genera. At the level of
one shared state, nine contested combinations occur. These con-
tested combinations are a product of the fact that every species or
combination of species which occurs at the level of two shared states
(with the exception of one, K. annandalei (20)) occurs in two sepa-
rate combinations at the level of one shared state. At the level of one
shared state, some of the combinations contain species from groups I
and III. The six largest combinations at this level depict relation-
ships among the species similar to those of previous layers in that
they contain overlapping sets of species, several of which are species
ofHydrophis. Astrotia stokesi (21) and Pelamis platurus (22) share
one state at this level and share 18 states in common at the second
character state tree layer. However, at neither the second nor the
third layers are these two species near neighbors.

Three species,//, klossi (40) , H . fasciatus (50), and//, brookii (51),
have two character states at the fifth character state tree layer
which they all share. These three species are all microcephalic Hy-
drophis species and at the level of one shared state they share a
state with M. cantoris (28), another microcephalic species. In addi-
tion, at the level of one shared state, Hydrophis species form a con-
tested combination which also contains the three microcephalic
Hydrophis species 40, 50, and 51.

At the sixth character state tree layer, no species contains more
than one character state and most species contain none at all.

Conclusions on Group IV and Group V Species. — Several conclu-
sions seem to emerge from an analysis of the group IV and group V

116 FIELDIANA: ZOOLOGY, VOLUME 70

species. First, these species are related to Thalassophis anomalus
and Lapemis of group III. The bulk of the species in group IV and
group V appears to represent the result of a rapid period of specia-
tion in that the species show a pattern of complex inter-relation-
ships.

Within this group there are some consistent patterns which con-
firm certain perviously recognized genera or species clusters, (e.g.,
Lapemis, Microcephalophis, and some clusters of Hydrophis spe-
cies), but these species as well as the majority of the others show
complex relationships among one another. A few monotypic genera,
for example, Pelamis, Enhydrina, and Acalyptophis, have clearly
diverged from the central stock of group IV and group V species but
this does not necessarily make them significantly earlier lineages.
They show no greater affinity to the early lineage of this group, e.g.,
Hydrelaps, than do other species in the group.

In summary, the complex inter-relationships among group IV and
group V species designated by the character states at all levels re-
flect a rapid radiation of species which is not resolvable by the char-
acter states under consideration.

Summary Statement of Phytogeny

Figure 6 is a diagram summarizing the relationships of the sea
snakes. The relationships depicted in this figure are based on the
interpretations of the combinations of species at each character
state tree layer. Several species which are omitted from the data
matrix and the subsequent analysis due to missing information are
placed in the summary phylogeny where they are thought to belong.

The Laticauda are clearly a group of very closely related species.
They are very distinct from all other sea snakes and either repre-
sent an independent evolutionary line or a very early separation
from all other sea snakes. They are by far the most primitive stock
of sea snakes and they possess many elapid character states.

The Aipysurus are a group of closely related species. Aipysurus
eydouxi is the most generalized of the Aipysurus species and it is
phenetically on the periphery of the central group of five. However,
its eventual combining with several Aipysurus and Emydocephalus

Opposite:

Fig. 6. Diagram of the major phylogenetic relationships among the Hydrophiidae
drawn from the detailed account starting on p. 105.

117

118 FIELDIANA: ZOOLOGY, VOLUME 70

seems to indicate that it is, in a phylogenetic sense, a stem species.
The two Emydocephalus species are very closely related to each
other and emerged hora Aipysurus stock. The Aipysurus stock (in-
cluding Emydocephalus) shows weak affinity with both the Lati-
cauda and the other sea snakes and like the Laticauda has either an
independent origin among the elapids or a very early separation
from ancestral sea snakes.

Hydrelaps darwiniensis, Ephalophis greyi, and Ephalophis mer-
toni represent early lines from the stock of remaining sea snakes
which includes the Hydrophis and several other genera. Although
these three species are relatively primitive, they have clear ties to
this stock.

Hydrophis kingi and H. major along with the monotypic genera,
Keriliajerdoni and Thalassophis anomalus are relatively primitive
and may have diverged from the main Hydrophis stock just prior to
the radiation which produced the majority of sea snake species. In
Figure 6 these species with few exceptions are given in numerical
order since branching details within the group are not resolved and
since the relationships of the species are too complex to be repre-
sented in a two or three dimensional diagram.

An Amplification of the Phylogeny

The previous sections of this paper are an attempt to explore the
phylogeny of the sea snakes with explicit methods. Here a few in-
terpretive comments are added.

Evolutionary trends within the Hydrophiidae: Within the sea
snakes I have found no evidence for a single unifying adaptive trend
reflected by a sequence of related morphological conditions. Grade
levels do exist. The Laticauda are amphibious (feed largely on eels,
lay eggs on land, and locomote terrestrially) while all other sea
snakes are fully aquatic (bear live young at sea and have poor ter-
restrial locomotion). However, the Laticauda do not stand on a
character-by-character basis between the terrestrial elapids and the
other sea snakes. In fact, the Laticauda share very few character
states with the other sea snakes which do not clearly show by their
distribution within the family a tendency toward multiple origin
and/or character state reversal.

The Aipysurus and Emydocephalus share many characters which
reflect their overall phenetic similarity and the common adaptive
zone which these species have exploited. Most of these species are

VORIS: PHYLOGENY OF SEA SNAKES 119

associated with coral reefs. In addition, the habit of eating demersal
fish eggs is wide spread in this group and does not occur among
other sea snakes (Voris, 1972). Trends in skull morphology asso-
ciated with egg eating are a dominant theme among these species,
with the most derived conditions occurring in the Emydocephalus.
Like the Laticauda , the Aipysurus and Emydocephalus do not have
many character states which are intermediate between states found
in elapids and the advanced sea snakes.

Within the other sea snakes, including several monotypic or bi-
typic genera and the genus Hydrophis, trends in body form have
occurred but there is some evidence which indicates that these
trends are recurring themes. For example, the reduction of head and
neck size occurs to different degrees in several species and is an
adaptation to nook and cranny feeding behavior and the consump-
tion of eels (Voris, 1972). It reaches a rather extreme condition in
the two species of Microcephalophis, Hydrophis brookii, and H.
klossi (two very similar species), as well as in Hydrophis torquatus,
H. caerulescens , and H. mamilaris. There are no other morpholog-
ical or geographical data which support a hypothesis of monophyly
for this assemblage and thus the microcephalic condition appears to
be a recurring adaptation.

A complex of characters including the number and size of ventral
scales, the number of vertebrae, the costo-cutaneous muscle system
and the overall body form have been directly involved in the shift
from a terrestrial to an aquatic existence (Voris, 1975). The adapta-
tions involved are complex and have not followed the same path of
change in all lineages of sea snakes. However, the overall trends are
clear. For example, there is a general tendency toward a prolifera-
tion of ventral scutes, and to a lesser extent a proliferation of verte-
brae with the consequent loss of correspondence between the verte-
brae and ventrals. Ventral scales have tended either to become
sharply keeled medially or reduced in size. In addition, within the
framework of each of the various body forms the posterior part of the
body has tended to become laterally flattened.

Uniquely derived states: Only a very few of the character states
studied here are likely to be uniquely derived within the sea snakes.
Several character states which are restricted in distribution to
single species are possible examples of uniquely derived states.
Characters 6 (gular azygous scale) and 7 (anterior prefrontal azy-
gous scale) are examples. If one assumes that these are uniquely
derived states they necessarily become states of little importance to

120 FIELDIANA: ZOOLOGY, VOLUME 70

interspecies relationships because they are states which arose after
all branching involving the species possessing the trait had taken
place.

An even smaller number of characters have states which are
widely distributed and are likely to be uniquely derived. A possible
example of such a state for which much data is available, is the loss
of correspondence between the number of ventral scales and the
number of body vertebrae. However, although a single origin inter-
pretation is compatible with the phylogeny in Figure 6, a detailed
analysis of this character (Voris, 1975) has revealed intraspecific
variation in one species and patterns of variation in the relationship
of ventrals and vertebrae along the length of the body which suggest
some experimentation and more than one mode of change for this
character.

Age and distribution of the Hydrophiidae: The age of the Hydro-
phiidae is indefinite since no fossil record of the group has been
discovered. During the Mesozoic and up through the Miocene of the
Cenozoic, the Tethys Sea formed a large variable marine environ-
ment from Southeast Asia to Eastern Europe (Darlington, 1957).
Since Hydrophiidae are not found in the Mediterranean Sea nor the
tropical Atlantic Ocean, it is possible that the snakes are more re-
cent than the Miocene (Tethys Sea). On the other hand, the current
distribution of the Hydrophiidae might also indicate that sea snakes
cannot live in the Mediterranean Sea or the Atlantic Ocean for eco-
logical reasons. One might argue that the former hypothesis is more
likely because the Persian Gulf (with 11 species of sea snakes) has
habitats very similar to those found in parts of the Mediterranean.
Concurring with this hypothesis is the observation that of the spe-
cies that appear in the Persian Gulf, and their close relatives Pe-
lamis platurus ranges as far south as the Cape of Good Hope, South
Africa, and Hydrophis cyanocinctus ranges as far east and north as
the Sea of Japan, and Hydrophis semperi (very similar toH. cyano-
cinctus) is found in a fresh- water lake. Lake Taal, in the Philip-
pines. It may be argued that if the Persian Gulf species are so wide
ranging, it is unlikely that they are excluded from the Mediterra-
nean for ecological reasons. This argument fails to be completely
convincing, however, because ecological data on the Hydrophiidae
are so lacking that even the most tentative hypothesis on this sub-
ject seems premature.

Zoogeographic data have been collected from the literature and
from the specimens. The major literature sources were the fol-

o%** l/> ©

* Ti

=. •*'

-Z> 2

i"

C

3 ^
3 CO

S s

o .2

[2h <V

121

122 FIELDIANA: ZOOLOGY, VOLUME 70

lowing: Wall, 1909; Smith, 1920, 1926, 1935; Taylor, 1922; Bourret,
1936; Vols0e, 1939; De Haas, 1950; Suvatti, 1950; Wang, 1962; and
Worrell, 1963. (The detailed distribution data are available and dis-
cussed in Voris, 1969.) The species frequency distributions are given
in Figure 7 and are subject to considerable error. There is a decrease
in the numbers of species along the coasts (the habitat of most spe-
cies) going from the Straits of Malacca to the Persian Gulf. This
decrease is rather erratic and is probably in large part due to a sam-
pling error. Generally, the areas with large numbers of species, the
Straits of Malacca and the Gulf of Siam (Smith, 1920), Ceylon (Wall,
1909), and the Persian Gulf (Vols0e, 1939), are areas which have
been collected and studied. The other areas, mouths of the Irrawady,
Ganges, and Indus Rivers, have been given less attention and thus
appear to be depauperate.

Taking into account the limitations of the available data, I con-
sider the center of distribution in terms of both number of species
and of genera to be the in shore areas of the Straits of Malacca, the
Gulf of Siam, and the southern end of the South China Sea. As one
travels outward from this region, fewer species are found with in-
creasing distance.

Genera from every major lineage depicted in Figure 6 are found in
and/or between the Gulf of Siam and Northern Australia. In fact,
several Aipysurus, Hydrelaps, and Ephalophis are endemic to this
region. Thus it is likely that the large number oiHydrophis species
(23) and associated genera (10) is the product of a relatively recent
radiation in northern Australian and Southeast Asian waters. The
fluctuating borders of the South China Sea during the Pleistocene
and possibly up to recent times (Darlington, 1957) placed the Sunda
Shelf (Straits of Malacca, the Gulf of Siam, the south end of the
South China Sea, and the Java Sea) above sea level, making Suma-
tra, Java, and Borneo part of the Asian continent. One can easily
imagine that at the time this barrier was forming, early representa-
tives oi Aipysurus eydouxi, Emydocephalus ijimae, and numerous
populations of Hydrophis were separated to the north of the main
populations and isolated there for various lengths of time. There is
some evidence that the sea level rose and fell more than once (Dar-
lington, 1957) and this alone may have caused the splitting of popu-
lations which gave rise to the numerous species ofHydrophis.

COMMENTARY

Certain procedural and philosophical aspects of this study require
further clarification or comment.

VORIS: PHYLOGENY OF SEA SNAKES 123

Nomenclature. — The nomenclature used here is, with few excep-
tions, that of Malcolm Smith (1925). Laticauda crockeri is from
Sleven (1934) and Hydrophis parviceps is from Smith (1935). The
designation for Ephalophis mertoni is from McDowell (1969). Mc-
Dowell (1972) and Burger and Natsuno (1974) have suggested ex-
tensive nomenclature changes among the sea snakes, some of which
in my opinion, are unwarranted and will require modification when
substantial samples are evaluated. The Linnean system of nomen-
clature can carry only a limited amount of phyletic information and
when relationships are largely speculation a conservative position
on the nomenclature has been taken and nomenclature changes
have not been proposed in this study. Thus, because Smith's (1926)
work has served as the standard reference for many years and is by
far the best known in the field, I have used its nomenclature for the
most part.

Character Analysis and Weighting, and Equivalency of States. —
The process of character analysis is a rapidly developing area (see
issues of Systematic Zoology, 1971, 1973), but its role and mode of
application in the process of derivation of phylogeny is not yet clear.
In this study it has been assumed or implied that the process of
character selection and elimination has resulted in a group of char-
acters containing a substantial amount of historical information.
Thus, character weighting was considered unnecessary. However, it
is generally agreed that characters are not historically equivalent
in any exact sense. This point extends to the character states also
and the procedure of classifying states into a few character state
tree layers is not justifiable in a purely theoretical sense in that the
states of different characters have, in general, evolved independ-
ently. The grading of states as was done in this study, is in the
process of further exploration and refinement. Also underway is an
extensive, in-depth analysis of ways of handling all data simulta-
neously to derive a consensus of the character state trees using the
combinatorial method (Marx et al., MS in prep.).

Resolution and Precision. — The data matrix of 40 species by 43
characters on which most of the conclusions are based is limited in
the following sense. Most characters have relatively few states (two
or three) and thus their "partitioning power" is limited. This study
deals with 50 species. Many of the characters designate the same or
similar partitions or major groups and do not vary within these
groups. Thus there is not much information bearing on the "micro-
relationships" of species. In short, a detailed study using many more

124 FIELDIANA: ZOOLOGY, VOLUME 70

characters (including some of those removed from this data set due
to restricted distribution) would definitely add resolution to the
intra-group relationships of some of the Aipysurus and Hydrophis.

Phenetic and Phyletic Results. — In my opinion it is not surprising
that the dendrogram (fig. 4) and the phylogeny (fig. 6) presented do
not differ from each other greatly or for that matter from Malcolm
Smith's (1926) overall view as reflected by his classification. This
has likely occurred because data used in the different approaches
overlap extensively, and because in many groups of organisms, esti-
mates of overall similarity (phenetics) are highly congruent with
the history of the group.

ACKNOWLEDGEMENTS

I am indebted to numerous individuals and institutions for assis-
tance in this study. I am grateful to the following individuals and
institutions for the use of specimens: C. M. Bogert and R. C. Zweifel,
American Museum of Natural History; H. G. Cogger, Australian
Museum; G. M. Storr, Western Australian Museum; A.G.C. Grandi-
son, British Museum (Natural History); A. E. Leviton, California
Academy of Sciences; Chulalongkorn University; E. H. Taylor, E. H.
Taylor Collection; R. F. Inger and H. Marx, Field Museum of Nat-
ural History; J. R. Hendrickson, J. R. Hendrickson Collection, Uni-
versity of Hawaii; E. R. Alfred, Singapore National Museum;
E. Williams, Museum of Comparative Zoology, Harvard; J. Eiselt,
Naturhistorisches Museum, Austria; R. Etheridge, San Diego State
College; W. Auffenberg, University of Florida; W. E. Duellman,
University of Kansas; C. F. Walker, University of Michigan, Mu-
seum of Zoology; D. M. Cochran and the late J. A. Peters, United
States National Museum; and F. W. Braestrup, Universitetets Zoo-
logiske Museum. I would like to add my thanks to Field Museum for
the use of its laboratories and other facilities while I was engaged in
this work in the 1960's before I joined the staff and to the University
of Michigan Museum of Zoology and the Chicago Zoological Society
for the use of their x-ray equipment. Joseph Felsenstein and Glenn
Sharrock contributed computer programs which were used in the
study.

Much of this work is a result of thought-provoking conversations
with George Rabb and Hymen Marx on the evolution of snakes and
applications of the combinatorial method. Discussions with Joseph
Felsenstein, Glenn Sharrock, Robert Inger, and Lynn Throckmorton
also contributed directly to the development of this work. All of

VORIS: PHYLOGENY OF SEA SNAKES 125

these sources of stimulation and assistance have been very much
appreciated.

The painstaking task of t5T)ing and proofreading numerous drafts
of the tables and text was performed very ably and co-operatively by
Marilyn Belka. Final drafts of most of the illustrations were prepared
by Zbigniew Jastrzebski, Field Museum of Natural History.

Especially, I gratefully acknowledge the help and encouragement
of my wife Helen in nearly every aspect of this work. She made
significant intellectual contributions to my understanding of the
methods used in this study, in addition to aiding in the preparation
of the manuscript.

REFERENCES

Bal, D. V. and K. V. Nawathe

1949. The circulatory system of Hydrophis caerulescens. Jour. Univ. Bombay, NS
15(5), pp. 1-4.

Bellairs, a. D. a. and G. Underwood

1951. The origin of snakes. Biol. Rev. Cambridge Phil. Soc, 26(2), pp. 193-237.

Blackwelder, R. E.

1967. Taxonomy, a text and reference book. John Wiley & Sons, New York, N.Y.

Boulenger, G. a.

1896. Catalogue of snakes in the British Museum (Natural History), 3, pp. 1-265.

Boyce, a. J.

1969. Mapping diversity: a comparative study of some numerical methods, pp. 1-
31. In Cole, A. J., ed.. Numerical taxonomy. Proc. CoUoq. Numerical Taxonomy.
Academic Press, London.

Bourret, R.

1936. Les serpents marins de I'lndochine Frangaise. fidite par L'institute Oceano-
graphique de L'Indochine.

Burger, W. L. and T. Natsuno

1974. A new genus for the Arafiira Smooth Seasnake and redefinitions of other
seasnake genera. Snake, 6, pp. 61-75.

Cain, A. J. and G. A. Harrison

1960. Phyletic weighting. Proc. Zool. Soc. London, 135, pp. 1-31.

Camin, J. H. and R. R. Sokal

1965. A method for deducing branching sequences in phylogeny. Evolution, 19(3),
pp. 311-326.

Darlington, P. J.

1957. Zoogeography: the geographical distribution of animals. John Wiley &
Sons, New York, N.Y. 675 pp.

Davis, P. H. and V. H. Heywood

1963. Principles of angiosperm taxonomy. D. Van Nostrand Co., Princeton, N.J.

126 FIELDIANA: ZOOLOGY, VOLUME 70

DeHaas, C. p. J.

1950. Checklist of the snakes of the Indo-Australian Archipelago. Treubia, 20(3),
pp. 511-625.

DOBZHANSKY, Th.

1951. Genetics and the origin of species. Columbia Univ. Press, New York, N.Y.

DuNSON, W. A. and G. W. Ehlert

1971. Effects of temperature, salinity and surface water flow on distribution of
the sea snake Pelamis, Limnol. Oceanogr., 16(6), pp. 845-853.

Hennig, W.

1966. Phylogenetic systematics. Univ. of 111. Press, Urbana, 111.

Heyer, W. R.

1974. Relationships of the Marmoratus species group (Amphibia, Leptodactylidae)
within the subfamily Leptodactylinae. Contr. Sci., no. 253, pp. 1-46.

Inger, R. F.

1972. Bufo of Eurasia. In Blair, W. P., ed.. Evolution in the genus Bufo, Univ.
Texas Press.

Jardine, N. and R. Sibson

1971. Mathematical taxonomy. John Wiley & Sons, N.Y.

Kendrick, W. B.

1964. Quantitative characters in computer taxonomy, pp. 105-114. In Heywood,
V. H. and J. McNeill, eds., Phenetic and phylogenetic taxonomy. Systematic
Assn., Publ., 6.

Kropach, C.

1971. Sea snake iPelamis platurus) aggregations on slicks in Panama. Herpetolo-
gica, 27(2), pp. 131-135.

LlEM, S. S.

1970. The morphology, systematics and evolution of the old world tree frogs
(Rhacophoridae and Hyperoliidae). Fieldiana, Zool., 57, pp. 1-145.

Marx, H. and G. B. Rabb

1972. Phyletic analysis of fifty characters of advanced snakes. Fieldiana: Zool.,
63, pp. 1-321.

Maslin, T. Paul

1952. Morphological criteria of phyletic relationships. Syst. Zool., 1, pp. 49-70.

Mayr, E.

1942. Systematics and the origin of the species. Columbia Univ. Press, New York,

N.Y.

McDowell, S. B.

1969. Notes on the Australian sea snake Ephalophis greyi M. Smith and the

origin and classification of sea snakes. Zool. Jour. Linn. Soc, 48, pp. 333-349.
1972. The genera of sea snakes of the Hydrophis group (Serpentes: Elapidae).

Trans. Zool. Soc. London, 32, pp. 189-247.

MiNTON, S. A.

1968. Distribution and identification of poisonous sea snakes. In Poisonous
snakes of the world, Dept. of Navy Publ., U.S. Govt. Printing Office.

VORIS: PHYLOGENY OF SEA SNAKES 127

Saint Girons, H.

1964. Notes sur I'ecologie et la structure des populations des Laticaudinae en
Nouvelle Caledonia. Terre Vie Rev. d'ecol. Appliquee, 2, pp. 185-214.

Sharrock, G. and J. Felsenstein

1976. Finding all monothetic subsets of a taxonomic group. Syst. Zool., 24(3), pp.
373-377.

Sleven, J. R.

1934. Notes on the reptiles and amphibians with the description of a new species
of sea snake. (The Templeton Crocker Expedition to western Polynesian and
Melanesian Islands, 1933.) Proc. Calif. Acad. Sci., 4th ser., 21(15), pp. 183-188.

Smith, M.

1920. On the sea snakes from the coasts of the Malay peninsula, Siam, and

Cochin-China. Jour. Fed. Malay States Mus., 10(1), pp. 1-65.
1926. Monograph of the sea-snakes. Oxford Univ. Press, London.

1935. The sea snakes. Dana Rept. No. 8.

Sneath, p. H. A.

1957. The application of computers to taxonomy. Jour. Gen. Microbiol., 17, pp.
201-226.

Sneath, P. H. A. and R. R. Sokal

1973. Numerical taxonomy. W. H. Freeman & Co., San Francisco.

Sokal, R. R. and C. D. Michener

1958. A statistical method for evaluating systematic relationships. Univ. Kansas
Sci. Bull., 38, pp. 1409-1438.

1967. The effects of different numerical techniques on the phenetic classification
of bees of the Hoplitis complex (Magachilidae). Proc. Linn. Soc. London, 178, pp.
59-74.

Sokal, R. R. and P. H. A. Sneath

1963. Principles of numerical taxonomy. W. H. Freeman and Co., San Francisco,
Calif.

SUVATTI, C.

1950. Fauna of Thailand. Dept. of Fisheries, pp. 512-518.

Taylor, E. H.

1922. The snakes of the Philippine Islands. Bur. Sci. Manila, Pub. no. 16, pp. 1-
312.

Throckmorton, L. H.

1962. The problem of phylogeny in the genus Drosophila. Stud. Genetics, 2, pp.
207-343.

1965. Similarity versus relationship in Drosophila. Syst. Zool., 14(3), pp. 221-236.

1968. Concordance and discordance of taxonomic characters in Drosophila classifi-
cation. Syst. Zool., 17(4), pp. 355-387.

Underwood, G.

1967. A contribution to the classification of snakes. Brit. Mus. (Nat. Hist.), Lon-
don, 179 pp.

128 FIELDIANA: ZOOLOGY, VOLUME 70

VOLS0E, H.

1939. The sea snakes of the Iranian Gulf and the Gulf of Oman. Danish Sci. Inves.
Iran, 1, pp. 9-45.

VoRis, H. K.

1969. The evolution of the Hydrophiidae with a critique on methods of approach.
Unpub. Univ. of Chicago thesis, 429 pp.

1971. New approaches to character analysis applied to the sea snakes (Hydrophi-
idae). Syst. Zool., 20(4), pp. 442-458.

1972. The role of sea snakes (Hydrophiidae) in the trophic structure of coastal
ocean communities. Jour. Marine Biol. Assn. India, 13(4), pp. 1-14.

1975. Dermal Scale-Vertebra Relationships in Sea Snakes (Hydrophiidae). Cop-
eia, (4), pp. 746-757.

Wall, C. F.

1909. A monograph of the sea snakes. Mem. Asiatic Soc. Bengal, 2(8), pp. 169-
251.

Wang, C.

1962. The reptiles of Botel-Tobago. Quart. Jour. Taiwan Mus., 15(3,4).

Worrell, E.

1963. Reptiles of Australia. Angus and Robertson, Sydney.

Zehren, S. J.

1974. The comparative osteology and phylogeny of the Beryciformes. Unpub.
Univ. of Chicago thesis, 458 pp.

Table 1 -- List of Hydrophiidae materials utilized in this study. The
species are those recognized by Smith (1926, 1935) and
Slevin (1934) . The number preceeding each species name is
the code for that species. The species abbreviations follow
each name in parentheses.

No,

Species

Whole
Spec- Tail Body

imens Skulls X-rays X-rays

1 Laticauda laticaudata (Lat lat) 7

2 Laticauda colubrina (Lat col) 35

3 Laticauda semifasciata (Lat sem) 4

4 Laticauda schistorLynchus (Lat sch) 19

5 Laticauda crockeri (Lat cro) 1

6 Aipysurus eydouxi (Aip eyd) 12

7 Aipysurus fuscus (Aip fus) 9

8 Aipysurus tenuis (Aip ten) 0

9 Aipysurus laevis (Aip lae) 4

10 Aipysurus duboisii (Aip dub) 4

11 Aipysurus foliosquama (Aip fol) H

12 Aipysurus apraefrontalis (Aip apr) 3

13 Emydocephalus annulatus (Emy ann) 5

14 Emydocephalus ijimae (Emy iji) 6

15 Hydrelaps darwiniensis (Hy 1 dar) 8

16 Ephalophis greyi (Eph gre) 2

17 Thalassophina viperina (Thn vip) 13

18 Acalyptophis peronii (Aca per) 6

19 Thalassophis anomalous (Ths ano) 13

20 Kolpophis annandalei (Kol ann) 6

21 Astrotia stokesii (Ast sto) 6

22 Pelamis platurus (Pel pla) 17

23 Enhydrina schistosa (Enh sch) 27

4
6
5
2
0
1
1
0
2
1
1
1
2
2
1
1
4
1
1
1
2
15
10

129

TABLE 1 -- Continued

Whole
Spec- Tail Body

No. Species imens Skulls X-rays X-rays

24 Kerilia jerdoni (Ker jer) 7 12 1

25 Lapemis hardwickii (Lap har) 37 14 4 9

26 Lapemis curtus (Lap cur) 9 12 0

27 Microcephalophis gracilis (Mic gra) 13 114

28 Microcephalophis cantoris (Mic can) 5 12 0

29 Hydrophis nigrocinctus (Hyd nig) 3 0 12

30 Hydrophis kingi (Hyd kin) 1 111

31 Ephalophis mertoni (Eph mer) 1 111

32 Hydrophis spiralis (Hyd spi) 6 Oil

33 Hydrophis melanosoma (Hyd mel) 7 111

34 Hydrophis belcheri (Hyd bel) 14 3 16

35 Hydrophis elegans (Hyd ele) 1 Oil

36 Hydrophis cyanocinctus (Hyd cya) 45 11 5 6

37 Hydrophis semperi (Hyd sem) 1 0 0 0

38 Hydrophis melanocephalus (Hyd me c) 8 10 1

39 Hydrophis obscurus (Hyd obs) 2 0 12

40 Hydrophis klossi (Hyd klo) 17 12 3

41 Hydrophis major (Hyd maj) 1 111

42 Hydrophis bituberculatus (Hyd bit) 0 0 0 0

43 Hydrophis s tr icticollis (Hyd str) 5 Oil

44 Hydrophis torquatus (Hyd tor) 23 7 2 6

45 Hydrophis ornatus (Hyd orn) 14 2 13

46 Hydrophis inornatus (Hyd ino) 16 2 11

47 Hydrophis lapemoides (Hyd lap) 15 3 2 2

48 Hydrophis mamillaris (Hyd mam) 1 111

130

Table 1 -- Continued

Whole
Spec- Tail Body

No. Species imens Skulls X-rays X-rays

49 Hydrophis caerulescens (Hyd cae)

50 Hydrophis fasciatus (Hyd fas)

51 Hydrophis brookii (Hyd bro)

52 Hydrophis parviceps (Hyd par)

Total 546 134 88 90

26

5

2

4

36

6

3

5

13

6

4

7

1

0

0

0

Table 2 is located on the back cover.

131

^,

^.

0 0 0 0

(0 ;0 ;0

e

o

o

to

4_i

u

(0

nj

4-1

j:

C/2

M 0)

5b

oj w

CO

u ^

u

a to

U-l

W 0)

CO

i-H 1-1

■r-l

rt «

>

tn

w en

0)

W CO

J-1

4-J

1—1

c c

c

CO

CO

r-l

tJ iJ

OJ

T— 1

tj

CO

Q) 01

U3

a.

OJ

cn

4-1 -U

OJ

4-1

ij

CO

a; 0)

01

.Q XI

•H

3 3

>.

4.)

4-1 4-1

4-1

•i-l

•H

to

X) T3

to

o

OJ 0)

o

1-1

(-1 V4

u

to

01

•H ■r^

(U

<D

Xi

CO to

XI

to

•r-l

3

D. a.

3

(0

,-t

4-1

4-1

4-1

r-(

Q)

•r-l

OJ CM

41

0)

10

•o

u

c

0)

^

O

0)

0 c

CO

(J >-i E CO ••-I

01

CO

0)

x:

>,

>>

01

TD

CO

QJ

TD

CO

X

to

IJ

4-1

OJ

I-H

•r-l

g,

4-1

r— 1

1— 1

x:

(U

o;

1-1

01

OJ

1-1

3

Q-

O

e

a.

4-J

O

4-1

4J

M

.^

Vj

4J

^

u

4-1

4-1

CU

!-i

to

0

to

to

3

•r-l

c

•r-l

c

O

o

V4

<u

o

CO

>-l

e

0

o

O

0

to

OJ

o

CO

OJ

O

?

?

o

c

o

CM

pL.

C/3

X

s

OS

z

pj

u

z

Pm

u

z

H

H

S

o

I-l c^j m <)•

1-1

0)

4-1

ij

U

0)

to

X

1-1

E

to

3

x;

z

u

132

^e

^0

B ^e

^0 0

0

rt CO

g

J3

(0

0)

tfl

'-'

■-I

^

^

(B

«

tfl

^J

M

^J

+J

4-)

u

C

c

c

01

0)

<u

>

>

>

J=

J=

x;

■u

■u

■u

o

o

O

•H >,

tn

w

o

cn

m

o

w

CO

01

W

)-i ^

w

, — 1

•rH

■ — 1

, — 1

•H

, — 1

T3 4-1

OJ

«

4-1

nJ

CO

4J

ra

T3

^

OJ

c j:

^

!-i

«

u

U

ra

S-i

0)

tfl

■H 60

a

4-1

•r-(

4-1

4-1

•H

4-)

Jj

■r-(

M

I— I •(-(

B

d

)^

d

d

>J

d

■H

U

d

>^ ^

o

0)

ffl

0)

OJ

ra

QJ

qj

ttf

•H

U M

o

>

>

>

>

>

>

PM

>

M

O

0)

to

4-1

S-i

(B

CO

4-1

J=

CO

•r4 x:

4-1 4J

to 60

■-I d

OJ (U

^4 i-H

d >

IS

d

(U

0)

o

>

CO

r-l

■•")

4-1

T3

CO

CO

CO CO

J-l <-<
01 CO

3 >

J= u

1

m OJ

<U CO

d J3

>

O g

■r-i m

•H 3

4_t ^^

4-1 d

CO CN

CO

l-l

0)

4-1

>-i

o

OJ

CO

43

1-1

g

CO

3

s:

Z

CJ

133

c -S

ro CNI oj CM

e ^e ;e B ^ b ^e b

(U Q)

<u

c c

c

•r^ -H

•--1

■U 4-1

4-1

rt CO

CO

.— 1 ^

1—1

to CO

CO

a a

a.

II A

V

i-H

4-1

tn w

en

CO

CO

4-J

c

T3 TD

•a

•r-l

•rH

c

OJ

C C

c

C

a

OJ

w

0) 0)

0)

(U

0)

en

(U

■u -u

4-1

1-1

, — 1

^

^

X X

X

a.

a.

CO

a.

CD 0)

<u

tfl

m

C

c

>. >.

>.

II

/\

o

o

^4 1-1

S-i

•r-l

■r-l

(0 CO

CO

u

^4

4-1

4J

^ 1-H

CO

CO

O

o

r-H i-H

1-1

i-H

1—1

0)

(U

•H -1-1

•H

3

3

X X

X

00

00

o

o

CO CO

CO

C

c

)-l

1-4

s: s

s:

<!

<

Pu

Ph

4-11-1 -i-l "-I

o

(U

CO

4-1

1-1

CO

CO

4-1

J=

!/)

CN 1-1 CM m

1— I cvj CO -J- in vD

1-1

(U

4-1

^-1

O

01

CO

x>

1-1

g

CO

3

JS

z

o

X 00
CO (U

c c

0) -r-l

I-H

w t3

1-1 -H
CO

a -u

CO

O 00

134

B

"e

B

'0 ^e

o m o o o in

cN r^ VO i-H o CN

\D in r— 1 1— < .— 1 r— I

r^ ro CN] vo I— I ^
I— I .— I CN rvi CO m

CN fO <j- m \X>

CO

iri

0^

-d-

in

n <}■ m vo

i-i (Tsi n <f m

CNj m <t "~i o

u c
^ >

p

0)

iJ

u

o

0)

tfl

XJ

1-1

CO

3

£

z:

o

135

:e

e

:e

:e B

u

dJ

4-1

u

o

0)

CO

J3

1-1

B

rt

d

X.

z

o

r-i <j- u-i ^ r^ 00

n) j-i

4J 3
C J3

CO CO CO

(A (A CO

CO CO CO

Z Z 2

0)

o

c

J3

o

XI

1—1

CO

,—4

4-)

CO

c

U)

o

CO

u

Z

^-t

136

(3 'Q

'o h

en Qj
P-i <;

U

+

o

<D

XI

>^

c

c

3

D.

c

dJ

OJ

OJ

OJ

0)

D"

«

nJ

o

■d

o

o

T3

W

>-i

B

c

C

c

c

c

4-1

<u

o

(P

0)

o

■^

w

■a

a

TD

•a

a-

i^

CJ

nj .^^ ^1 sj >j M

a. > <u OJ 0) <D

• HO iJ -U 4-1 4J

> > c C (0 C

O O < < J <

tN ^ osi ro

)-l

OJ

4-J

Vj

o

QJ

to

^

u

E

to

3

j:

z

o

137

Table 4. Matrix of the character states of 43 characters for 40 species of sea
snake. See Table 3 for the names of the characters and character states.

character Number

Spec.

No.

Species

01

02

03

04

05

12

14

17

22

23

28

30

34

35

36

46

48

52

57

58

59

60

77

1

Lat

lat

1

1

1

2

1

1

1

1

1

2

2

2

Lat

col

1

1

2

2

1

1

1

1

1

2

2

3

Lat

sen

1

1

2

2

1

1

1

1

1

2

1

4

Lat

sch

1

2

2

3

1

1

1

1

1

2

1

6

Aip

eyd

2

2

2

3

2

2

1

3

2

1

2

7

Aip

fus

2

2

1

3

2

2

1

3

1

2

2

2

9

Aip

lae

2

2

1

2

2

1

3

2

2

2

2

10

Aip

dub

2

2

3

2

2

1

3

1

2

2

11

Aip

fol

2

2

1

2

2

1

3

1

2

12

Aip

apr

2

2

1

2

1

1

3

1

9

13

Emy

ann

2

2

2

2

2

1

3

1

2

14

Emy

iji

2

2

1

2

1

1

3

2

1

15

Hyl

dar

2

2

1

3

1

1

3

2

3

2

2

17

Thn

vip

2

2

3

3

2

2

2

2

2

2

3

2

2

2

18

Aca

per

3

2

2

4

2

3

2

2

2

2

2

3

2

2

2

19

Ths

ano

1

2

2

4

2

3

1

1

1

2

2

3

2

2

2

20

Kol

ann

1

2

2

1

3

3

2

2

2

2

3

2

1

2

2

2

21

Ast

sto

2

2

2

2

2

3

1

2

2

2

2

2

3

2

2

2

22

Pel

pla

2

2

2

1

3

3

2

2

3

3

2

2

3

2

2

2

23

Enh

sch

3

2

1

2

4

3

4

2

3

2

2

2

2

2

3

2

2

2

24

Ker

jer

2

2

1

1

2

2

4

2

1

1

1

1

2

3

2

1

2

2

2

138

Table 4. continued.

Spec .

No.

1
2
3

4

6

7

9

10

11

12

13

14

15

17

18

19

20
21
22
23
24

Character Number

Species 82 83 84 87 90 92 93 95 98 137 138 139 142 143 144 145 146 148 149 153

Lat lat 5 3 12 112 13 1

Lat col 6 3 12 112 111

Lat sen 3 112 112 111

Lat sch 3 112 112 111

Aip eyd 3211112443

Alp fus 3211212215

Aip lae 2211212212

Aip dub 4 2 11112 2 15

Aip fol 12 11112 2 15

Aip apr 3211112235

Emy ann 4211112126

Emy iji 4211212126

Hyldar 4222212237

Thn vip 4326323238

Aca per 5 113 2 13 3 4 9

Ths ano 3 3 2 3 3 13 13 4

Kol ann 4 5 5 6 3 4 4 2 19

Ast sto 2 3 4 6 4 2 3 2 4 8

Pel pla 2 5 4 6 3 3 3 2 3 9

Enh sch 2 4 4 6 2 3 3 118

Ker jer

3 3

12 2 113 4 4 3

1

1

1

1

1

1

2

1

1

1

1

1

1

1

2

1

1

1

1

1

2

1

2

1

1

1

1

2

1

2

2

2

2

2

3

2

1

2

2

2

2

3

2

2

2

2

2

2

3

2

2

2

2

2

2

3

3

2

2

2

2

2

3

2

2

2

2

2

2

3

2

7

2

2

9

2

2

4

2

2

2

2

2

2

4

2

3

2

2

2

2

2

4

2

4

1

3

2

2

2

3

2

4

2

4

3

2

2

2

2

3

2

4

2

4

3

3

2

9

2

2

2

4

2

4

3

2

2

2

2

3

2

4

2

4

3

2

3

9

2

3

2

4

2

4

2

3

2

2

2

3

2

4

2

4

3

2

9

2

2

3

2

4

2

4

2

3

2

2

2

2

2

4

2

4

3

139

Table 4. continued.

Character Number

Spec.

No.

Species

01

02

03 04

05 12 14

25

Lap

har

2

2

2 2

2 1 1

26

Lap

cur

2

2

2 2

1 1 1

27

Mic

gra

2

2

1 1

2 1 1

28

Mic

can

2

2

2 1

1 1 1

30

Hyd

kin

2

2

1 1

1 1 1

31

Eph

mer

2

2

1 1

1 1 1

33

Hyd

mel

2

2

1 1

4 11

34

Hyd

bel

2

2

1 1

1 1 1

36

Hyd

cya

2

2

2 1

4 1 1

40

Hyd

klo

2

2

2 1

2 1 1

41

Hyd

ma j

2

2

1 1

2 1 1

44

Hyd

tor

2

2

1 1

4 11

45

Hyd

orn

2

2

1 1

2 1 1

46

Hyd

ino

2

2

1 1

1 1 1

47

Hyd

lap

2

2

1 1

2 1 1

48

Hyd

mam

2

2

1 9

1 1 1

49

Hyd

cae

2

2

2 1

4 1 1

50

Hyd

fas

2

2

1 1

2 1 1

51

Hyd

bro

2

2

2 1

2 1 1

2432222232112221
2332222232112221
3322123232221221
4323123232212211
2311311332121123
1213111332312221
3423222232112221
3323222232112221
3323222222112221
2323212232112221
2311311332312223
3423222232312221
2423222232112221
2423222232312221
2323222232112221
2333323232312221
3323212232312221
3423222222112221
3423222232112221

140

Table 4. continued.

Character Number

Species 82 83 84 87 90 92 93 95 98 137 138 139 142 143 144 145 146 148 149 153

12 2 3 2 13 2 18

2223213118

4414313218

4614333218

3 5 2 4 4 13 13 8

5234212331
4425433218
4425313218
4525323238
4625333218
3324413248
4435323338
4336423348
4336323338
4525323348
5525333338
14 3 5 3 2 3 4 4 8
4625334118
5 6 2 5 3 3 3 2 18

Spec.

No.

Species

25

Lap har

26

Lap cur

27

Mic gra

28

Mic can

30

Hyd kin

31

Eph mer

33

Hyd mel

34

Hyd bel

36

Hyd cya

40

Hyd klo

41

Hyd maj

44

Hyd tor

45

Hyd orn

46

Hyd ino

47

Hyd lap

48

Hyd mam

49

Hyd cae

50

Hyd fas

51

Hyd bro

2

2

2

2

2

2

4

2

4

3

2

2

2

9

2

2

4

2

4

3

3

2

2

2

3

2

4

2

4

3

3

2

2

2

3

2

4

2

4

3

3

2

2

2

3

2

4

2

4

3

2

2

2

1

2

2

4

2

4

1

3

2

2

2

3

2

4

2

4

3

3

2

2

2

3

2

4

2

4

3

3

2

2

2

3

2

4

2

4

3

3

2

2

2

3

2

4

2

4

3

3

2

2

2

3

2

4

2

4

3

2

2

2

2

3

2

4

2

4

3

3

2

2

2

3

2

4

2

4

3

3

2

2

2

3

2

4

2

4

3

3

2

2

2

3

2

4

2

4

3

3

2

2

2

3

2

4

2

4

3

2

2

2

2

3

2

4

2

4

3

3

2

2

2

3

2

4

2

4

3

3

2

2

2

3

2

4

2

4

3

141

Table 5 -- For each species the number and percentage of character states at each
character state tree layer is given. The first layer corresponds to
the primitive states. See the text for a further explanation.
Character State Tree Layers

Species

Second Third Fourth Fifth Sixth

7„ No. 7„ No. 7o No. 7o No. 7,

1

Lat

lat

34

79

8

19

0

00

1

02

0

00

0

00

2

Lat

col

34

79

9

21

0

00

0

00

0

00

0

00

3

Lat

sem

35

81

8

19

0

00

0

00

0

00

0

00

4

Lat

sch

34

76

9

20

1

02

1

02

0

00

0

45

6

Aip

eyd

16

39

24

58

1

02

0

00

0

00

0

41

7

Aip

fus

18

41

23

54

2

05

0

00

0

00

0

00

9

Aip

lae

15

35

26

61

2

05

0

00

0

00

0

00

10

Aip

dub

16

37

22

51

4

09

1

02

0

00

0

00

11

Aip

fol

21

49

20

47

2

05

0

00

0

00

0

00

12

Aip

apr

19

45

22

52

1

02

0

00

0

00

0

00

13

Emy

ann

19

45

21

50

2

05

0

00

0

00

0

00

14

Emy

iji

19

44

22

51

2

05

0

00

0

00

0

00

15

Hyl

dar

16

37

22

51

5

12

0

00

0

00

0

00

17

Thn

vip

6

14

27

63

9

21

0

00

0

00

1

02

18

Aca

per

9

21

22

52

9

21

2

05

0

00

0

00

19

Ths

ano

12

29

24

57

6

14

1

02

0

00

0

00

20

Kol

ann

8

19

20

48

11

26

2

05

1

02

1

02

21

Ast

sto

5

12

30

71

4

10

2

05

0

00

1

02

22

Pel

pla

6

14

23

55

10

24

2

05

0

00

1

02

23

E.ih

sch

7

17

24

57

8

19

3

07

0

00

1

02

24

Ker

jer

14

33

22

51

4

09

2

05

0

00

0

00

25

Lap

har

8

19

29

67

5

12

1

02

0

00

0

00

26

Lap

cur

9

21

27

64

6

14

0

00

0

00

0

00

27

Mic

gra

12

28

20

47

10

23

1

02

0

00

0

00

28

Mic

can

8

19

22

51

10

23

2

05

1

02

0

00

30

Hyd

kin

16

37

18

42

7

16

3

07

0

00

0

00

31

Eph

mer

15

35

20

47

6

14

2

05

0

00

0

00

33

Hyd

mel

8

19

22

51

8

19

3

07

1

02

0

00

142

Table 5

--

Species

Charact(

2r State Tr

ee Layers

First

Second

Third

Four

th

Fifth

Six

th

No.

7o

No,

%

No.

7,

No.

\

No.

7o

No.

°L

34

Hyd

bel

10

24

22

52

9

21

0

00

1

02

0

00

36

Hyd

cya

6

14

26

61

7

16

2

05

1

02

0

00

40

Hyd

klo

8

19

24

56

9

21

0

00

2

05

0

00

41

Hyd

maj

10

23

23

52

9

21

2

05

0

00

0

44

44

Hyd

tor

6

14

23

54

10

23

2

05

1

02

0

00

45

Hyd

orn

7

16

24

56

9

21

2

05

0

00

1

02

46

Hyd

ino

7

16

25

58

9

21

1

02

0

00

1

02

47

Hyd

lap

7

16

24

56

10

23

1

02

1

02

0

00

48

Hyd

mam

6

15

21

51

11

27

2

05

1

02

0

41

49

Hyd

cae

7

17

21

51

10

24

2

05

1

02

0

41

50

Hyd

fas

9

20

23

52

9

20

1

02

2

05

0

44

51

Hyd

bro

7

16

24

56

7

16

2

05

2

05

0

00

143

Table 6 -- Combinations of species selected from those generated by the

combinatorial method from 40 species and the 43 character states
at the first character state tree layer (see text for selection
procedures). The following punctuation system is used in Tables
6,7,8, and 9 (examples are taken from Table 6): A species is
circled at the level of shared states at which it first appears
(£.£., species 3 has 35 character states at the first character
state tree layer, see also Fig. 5). An asterisk denotes the
formation of a new combination (£•£., species 3 and 4 form a
combination at the level of 34 shared character states). A
combination is solidly underlined if it is uncontested, that is
none of its members are found in another group at that level of
shared states (£.£., the group 1 , 2 is uncontested at tie level
of 33 shared states, and at level 14, the species 1,2,3,4,13
within the group i^j 2^,^)4 >i3j ^^ ^'^^ underlined individually be-
cause they formed an uncontested group at level 17). The neigh-
bors of species or combinations of species are ^iven in paren-
theses (£•£., species 1] is a near neighbor to species 10 at
level 16, and species 2,3,4,11 and 11,12 are neighbors to species
10 at the level of 15 shared states). Brackets denote the
neighbors of combinations which merge to form a new combination
at that level of shared states (£•£., the combination 7,11,12
formed at the level of 16 shared states. At the same level
species 7 has neighbors 2,3,11 and 3,4; and the combination
11,12 has the neighbors 1,2,3,4).

144

SHARED
STATES

35

SELECTED

(D

COMBINATIONS

SHARED
STATES

19

SELECTED COMBINATIONS

1,2,3,4

11(3)

@

0

@

34

0

®

-1©

* 111

3.4

33

18

1,2,3,4

(7)

32

no change

s

* 11.12 [11(2.3) (3,4) 12(1,2,3)]
13(3,4)

31

1j1(3)

14

1.2.3.4

30

* 1^

^3,4

29

no

changes

28

no

changes

27

no

changes

26

no

changes

25

no

changes

24

no

changes

23

no

changes

22

no

changes

1.2.3.4.13 [1,2,3,4(12)1

7(3) (11)

11.12(1,2,3)

14(1,2,3)

1.2.3.4.13

®
' 7,11,12 [7(2,3,11) (3,4) 11,12

(1,2,3,4)1
@(11)
14(1,2,3,4) (12)

15 1.2.3.4.13

7.11.12(1,2,3)
0X7,11)
10(2,3,4,11) (11,12)

, 14(1,2,3,12) (11, 12)(13)

no changes ^ > , > / ^ ,

15

30

145

Table 6 -- continued

SHARED

STATES SELECTED COMBINATIONS

14 * 1,2,3,4, n^, 14 [1,2,3,4,13(11,12) 14 (1 ,2 ,3 ,4, 12) (1 ,2 ,3, 11 , 12) ]

6(1,2)

* 7_,9,n_,12 [7,11,12(1,2,3,4) 9(2,3,7,11)1

* 7,n,12_,14 [7,11,12(1,2,3,4) 14 (1 ,2 ,3 ,4 , 12) ( 1 ,2 ,3 , 11 , 12) ]
10(1,2,3,4,11,12) (7,11) (11,12,13)

* 15,31 [15(1) 31(1)1

@
30

13 i.2,3_,i.i3,14

6(1,2,3,4) (11,12)

* 7,9,n_,12_, 14 [7,9,11,12(1,2,3) 7,11,12,14(1,2,3)1
10(1,2,3,4,11,12,13) (2,3,4,7,11) (7,11,12) (11,12,14)
15,31(1)

24
30

12 l,2,2,4,13_.l'i(ll,12)

* 6.10(11,12) [6(1,2,3,4,11,12) (7,11,12) (13) (14) 10(1 ,2 ,3 ,4 , 7, 11 , 12) (1 ,2 ,

3,4,11,12,14) (7,9,11) (7,9,12,13) (7,11,12,14) (11,12,13,14)1
2,9,11,12,14(1,2,3)
15,31(1,2)
0
24

o

30(1) (3)

11 j,,2_,3,4,12,l^(7,ll,12) (10,11,12)

6,10(1,2.3,4,11,12) (7,11,12) (11,12,13)

2,9,11,12,14(1,2,3,4) (10) (13)

15,31(1.2.3)

19

24(1) (3,4)

27

30(1,2,3) (3,4,13) (11) (14)

146

Table 6 -- continued

SHARED

STATES SELECTED COMBINATIONS

"■■ 1.2.3. i'Z. 9, 11,12, 12,14 [1,2,3,4,13,14(7,10,11,12) 7,9,11,12,14
(1,2,3,4,10) (10,13)1

* b_,l,9,10_,U,U,U [6,10(1,2,3,4,7,11,12,14) (1,2,3,4,11,12,13) (7,11,

12,13,14) 7,9,11,12,14(1,2,3,4,10) (10,13) 1
15.31(1,2,3,4) (1,2,3,12)

* 19,24(15) [19(1)(3,4) 24(1,2,3,4)1

* 24,30 [24(1,2,3,4) 30(1 ,2 ,3 ,4 , 13) (1 ,2 , 3 , 14) (3 , 11) (15) 1
27(3,4)

* 30,@ (30(1,2,3,4,13) (1,2,3,4,14) (3,11) (15) ]
@

1.1.1.^,1.9,11,12., 11, 14(10)
6,7,9,10,1]_,12_, 14(1,2,3,4) (11)

* 15,19,24,21 [ 15,31(1,2,3,4,7,9,11,12) (1,2, 3,4, 12)(1, 2, 3, 12, 14)

(30)]

@

24,30,41 [30,41(15)1

@

27(1) (2,3,4)

34

* 1.2.3.4.6,7,9.10,11,12.13.14
15,19,24,31(1)

18

®

24,30,41(15,19)

* 26,34 (34(15) (24) (30) 1

* 27,34 [27(1,2,3,4) (3,4,11) (24) 34( 15) (24) (30) ;
@

* @,34,50 [34(15) (24) (30) ]

147

SHAEIED

STATES SELECTED COMBINATIONS

7 1.2.3.4.6.7.9.10.11.12,13,14

* 15.19.24.30.31,41 [15,19,24,31(1,2,3,4) 30 , 31 ,41(1) (3 ,4 , 13) )
18(3,4,) (15) (24,27)

20
0(50)

* 25,26,34

* 2 7,33,34.50
28(27)

■" 33,34,40,(3)

* 33, 34, @,©, 50
@(15.34)

'4J

6 1.2.3.4,6,7,9.10.11,12,13,14(30)

15,19,24,30.31.41(1)

* (0),27,33,34,@),45,46,47,50(15,19,24) [33,34,45,47,50(24,30) 46(15,30,34)

(15,31,34)1
18(1) (3,4,11) (15,19,24,2 7,34) (15,31)
20(26,34)

@

* 23,27,33,34,50 [ 23 (19 ,50) (26 ,50) |

'■•- 25,26,33,34,40,51(50) [ 25 ,26 , 34(24) (27) 1
■'■■- 27,33,34,40,50,51
28(1) (2,3) (34)

* 33, 34, ©.'iO, 45, 47, 50, 51(24) [33,34,45,47,50(24,30)1

* 46,@(15,34) [46(15,30,34) (15,31,34)1
49(15,19,24,40) (30)

5 1.2.3.4.6.7.9,10.11.12.13.14(24) (30,41)

15,19,24,30,31,41(1,2,3,4,13) (2 7,34)

* 17,18,2 3,2 7,33,34,44,45,46,47,48,50(15,19,24) [17,27,33,34,44,45,46,

47,50(15,19,24,30,41) (15,19,24,31) 18(1,2,3,4,7,9,11,12) (1,15,31)
(1,24,2 7) (3,4,11,13) (3,4,11,15) (3,4,11,24,27) (15,19,24,25,26,2 7,34)
(15,19,24,2 7,30,34,41) (15,19,24,2 7,34) 46,48 (15 , 30, 34) (15 , 31 , 34) ]

* 17,2 7,33,34,36,40,44,45,46,47,49,50,51(15,19,24) [17,27,33,34,44,45,

46,47,50(15,19,24,30,41) (15,19,24,31) 33,34,36,40,45,47,50,51(24,

30) 49(15,19,24,30,40,41) (15,19,24,31,40) (25) 1

* 20,25,26,33,34,40,51(50) [ 20(3) (15) (2 6 , 30 , 34) 1
^(24,30,41)

* 22,46,48(15,26,34) (46 ,48 (15 , 30 , 34) (15 ,31 , 34) ]

* 23,25,26,27,33,34,40,50,51

* 25,26,33,34,36,40,45,47,50,51(24) [33,34,36,40,45,47,50,51(24,30) ]

148

Table 6

-- continued

SHARED

STATES

SELECTED COMBINATIONS

-V

27,28,33,34,40,50,51

(26,34) 1

[28(1,2,3) (1,2 7) (2,3,4,2 7) (15,31,34,46) (24,2 7)

1,2,3.4.6,7.9.10.11.12,13,14(15.30,31.41) (24,30,41)
15,19,24,30,31,41(1,2,3,4,6,13) (1 ,2 , 3 ,4 , 7 ,9, 10 , 11 , 12 , 13 , 14) ( 1 ,27 , 34)

(17,27,33,34,44,45,46,47,50) (18,2 7,34) (40,49)
-■■■17,18,22,23,25,26,27,33,34,36,40,44,45,46,47,48,49,50,51(15,19,24) [17,

18,2 3,2 7,33,34,44,45,46,47,48,50(15,19,24,30,41) (15,19,24,31) 22,

46,48(15,20,26,30,34) (15 ,26,28,31 ,34) 1

* 17,21,2 7,33,34,36,40,44,45,46,47,49,50,51(15,19,24,30,41)

* 17,2 7,28,33,34,36,40,44,45,46,47,49,50,51(15,19,24,31)

* 20,23,25,26,27,33,34,40,50,51

* 20,25,26,33,34,36,40,45,47,50,51(24,30)
^■••23,2 5,26,2 7,2 8,3 3,34,40,50,51

" 1, 2, 3, 4, 6, 7, 9, 1£,JJ., 12, 11,14, 15, 19, 24, 30, 31, 41 [1,2,3,4,6,7,9,10,11,12,
13,14(2 7) 15,19,24,30,31,41(1,2,3,4,13,2 7,34) (1,17.2 7.3 3,34,44,45,
46,47,50) (1,18,2 7,34) (1,40,49) (17,18,23,27,33,34,44,45,46,47,48,50)
(18,25,26,27,34)1

"" 15,12,19,21,24,2 7,28,30,31,33,34,36,40,41,44,45,46,47,49,50,51 [15,19,24,
30,31,41(1,2,3,4,13,2 7,34) (1,17,2 7,33,34,44,45,46,47,50) (1,18,27,34) (1,
40,49) (17,18,23,2 7,33,34,44,45,46,47,48,50) (18,25,26,2 7,34) 17,27,
28,33,34,36,40,44,45,46,47,49,50,51(1,15,19,24,31) I

* 17,18,20,21,22,23,25,26,27,33,34,36,40,44,45,46,47,48,49,50,51(15,19,24,

30,41)

* 17,18,22,23,25,26,27,28,33,34,36,40,44,45,46,47,48,49,50,51(15,19,24,31)

[17,27,28,33,34,36,40,44,45,46,47,49,50,51(1,15,19,24,31) I
* 20,23,25,26,2 7,28,33,34,40,50,51

2,2,2,4,6,2,9,l^,U,li,i3,14,15_,19,24,30,3]_,41
* 15., 1 7, 18, 19, 20, 2 1,22, 23, 24, 25, 26, 2 7, 28, 30, 3J., 33, 34, 36, 40, 41, 44, 45, 46,
47,48,49,50,51 [15,17,19,21,24,27,28,30,31,33,34,36,40,41,44,45,
46,47,49,50,51(1) 17,18,22,23,25,26,27,28,33,34,36,40,44,45,46,
47,48,49,50,51(1,15,19,24,31) 1

1, 2, 2, 4. i. Z. 1.1°. 11. 11. 11. li,15.,li, 24, 30, 11, 41(18, 2 5, 2 6, 2 7, 34) (40,49)
15_, 17, 18, 12, 20, 21, 22, 23, 24, 25, 26, 2 7, 2 8, 30, 11, 33,34,36,40,41,44,45,46,47,
48,49,50,51(1)

149

Table 7 -- Combinations of species selected from those generated by the com-
binatorial method from 40 species and the 65 character states at
the second character stale tree layer (see text for selection pro-
cedures, and see heading for Table 6 for an explanation of the

punctuation system used) .

SHARED SHARED

STATES SELECTED CO>fBINATIONS STATES SELECTED COMBINATIONS

25

2i 23 6

■^^ ©

21

9
'^ 17.46

21
25

17
21

25.26
36

%

24 (6)

9

17.46
21

25,26

'^

19
21

© 23

21 25.26

25 * 36,51

^ * 40,51

®
17

%
■ 25.26

@ 22 6

7

45.47 [45(46)

17.46(45)
@
19

21(17) (25)

22

23

@

25.26

@
" ©,©,36,51
* 36,40,51

41

44(46)

45.47(46)

50f36) (47)

150

Table 7 -- continued

SHARED

STATES SELECTED COMBINATIONS

21 6

7,9
10
12
©
14
15

* 17,21,46 [17,46(36) 21(45) ]

* 17,44,46 [17,46(36) 44(36,46)]

* 17,45_,46_,47 [17,46(36) 45 ,47(36,46) (40) (51) 1
18

19
22
23
24
25,26

28

* 33,34.36.40.51 | 33 , 34, 36,51 (17) (47) ]
41

* 45_, 47,50 [45,47(36,46) (40) (51) 50( 36,47) (47,51)
@(17)

6

7.9

10 .0 [10(7,9);

12

13

14

15

17_.21,46(45)

12,44,46(36)

17,45,46,47(36)

18.23 (23(21)1

19

@

22(17)

24

25.26(17)

©

28(36,51)

151

Table 7 -- continued

SHARED

STATES SELECTED COMBINATIONS

33.34,36.40,51(17) (4 7)
41(17)

45,47,50(36,46) (51)
48(17,46)
49 (44)

19 6

7.9(10)

10.11(7) (9)

12(7) (10)

13

14

15

* 17. 21, -^4, 46

* 17,21,45,46,42 [17,45,46,47(25) (33,34,36,51) (40) )

* 17>'*^'^>^'^(36) [17,45,46,47(25) (33,34,36,51) (40) ]

* ^,^>^,^,50(36) (17,45,46,47(25) (33,34,36,51) (40) 45,47,50(33,34,36,

46,51) (36,44,46) (40,51)]
18.23

* 19,24 [19(17) (46)1

* 19,41 [19(17)(46) 41(21)(40)1
20(25)

22(25) (36)

25,26(17.47) (21) (36,51) (40)

27

28(33,34,36,51) (36,40,51)

31

33.34.36.40.51(17.25) (17,47) (45,46,47) (47,50)

48(17,36) (17,47)

49(21) (25) (36,40,51) (44,46)

18 6(7) (9)

7,9,10,11_

* 10,11,12 [12(7,10) (9,10) 1

* 13.14
15

* 17_,21,44,45,46,47(36) [17,21,45,46,47(25) 17,44,45,46,47(25,33,34,36,51)]

* 17,44,45,46,47,50 [17,44,45,46,47(25,33,34,36,51) 17.45,46,47,50(33,34,

36,51)1
18.23(44)

152

Table 7 -- continued

SHARED

STATES SELECTED COMBINATIONS

19,24

19,41

20

22(17,36) (17,46) (21) (26) (36,40,51)

* 25,26,49 [25,26(17,33,34,36,47,51) (17,40,47) (17,45,46,47) (23) (36,40,51) (44)

49(17,44,46) (18,44) (21,25) (21,44) (23,44) (25,36,40,51) (25,44) )

* 27,28 [27(51) 28(17, 33 ,34,36,51) (25 , 36 ,51) (33 , 34, 36 ,45 ,46,47,51) ]

* 28,32,34,36,40,51 ( 28(17,33 , 34, 36,51) (25 ,36,51) (33 , 34 , 36,45 ,46,47,51) 33,

34,36,40,51(17,25,47) (17,45,46,47) (17,47,50) (45,46,47,50) ]
@
31

* 11,34,36,40,49,51(44,45,46,47) [33,34,36,40,51(17,25,47) (17,45,46,47) (17,

47.50) (45,46,47,50) 49(17,44,46) (18,44) (21,25) (21,44) (23,44) (25,36,

40.51) (25,44) I

48(17,19) (17,36,46) (17,40,47) (17,41) (17,44,46) (17,45,46,47)

17 6(7,9)

1,9,10,11

10,11,12

13,14

* 15.31 [15(25) I

* 17.21,44.45.46.47.50(36) [17,21,44,45,46,47(25,33.34,36,51) 17.44,

45,46,47,50(25,33,34,36,51)1
18.23(21,44) (25) (44,46) (44,49)
19,24(45)
19,41(17)
20(21) (25,26)

* 22.48(17.46) [22(17,25) (17,33,34,36,40,51) (17,36,46) (17,44,46) (25,26) (25,

36,40,51) (28,36,40,51) 48(17,19,46) (17,21,46) (17,25,26,47) (17,25,
33,34,36,47,51) (17,25,40,47) (17,25,45,46) (17,28,33,34,36,45,46,47,51)
(17,33,34,36,40,47,51) (17,33,34,36,47,50,51) (17,36,44,46) (17,40,41,
47) (17,40,45,46,47) (17,41,46) ]

25,26,49(21) (36,40,51) (44)

27,28(33,34,36,51)

28,31,34,36,40,51(17) (45,46.47)

30(41)

31,34,36,40,49,51(17,25,44,45,46,47) (44,45,46,47.50)

153

Table 7

-- continued

SHARED

STATES

SELECTED COMBINATIONS

16

6(7,9,10) (7,12)

*

7,9,10,11,12

15,31

17,21,44,45,46,47,50(25.33,34,36.51)

18,23(17.21,44.46) (21,25,44) (21,44,49) (25,26) (33,34,36,44,45,46,47,51) (44,

46,49)
19,24(40,45,47) (45,46) (45)
19,41(17,46) (17,48) (40,47)
20(17,25) (21,25) (22,25) (25,36,51)
22.48(17,19.46) (17,36,46) (17,44,46)

* 25,26,33,34,36,40,49,51(17,44,45,46,47) [25,26,49(21,36,40,51) (21,44)

33,34,36,40,49,51(17,21,25,44,45,46,47) (17,25,44,45,46,47,50) (18,
44,45,46,47) (23,44,45,46,47)

* 27,28,22,34,26,40,51 [ 27,28(17,33 , 34,36,51) (33 , 34,36,45 ,46,47 ,51) 28,

33,34,36,40,51(17,22) (17,25) (17,41) (17,45,46,47,48) (45,46,47,50)]

* 28, 33, 24, 2i, 40, 49, 51(44, 45, 46, 47) 1 28,33 , 34,36,40 ,51 (17,22) (17 ,25) (17,41)

(17,45,46,47,48) (45,46,47,50) 33,34,36,40,49,51(17,21,25,44,45,46,
47) (17,25,44,45,46,47,50) (18,44,45,46,47) (2 3,44,45,46,47)]
30(17,48)

15 6(7,9,10,11) (7,9,10,12)

7,9,10,11.12
13,14
15,31

17.21,44.45.46.47.50(2 3,25,33.34,36,51) (25,26,33,34,36,51) (25,33,34,36,

40,49,51)
18, 23 ri7, 2 1,2 5, 33, 34, 36, 44. 45, 46, 47, 51) (17,21,44,46,49) (21,25,26,44) (21,

25,44,49) (33,34,36,40,44,45,46,47,49,51) (33,34,36,44,45.46.47,50,51)
19,24.41(45) [19,24(17,45,46) (25) (40,45,46,47) (40,45,47,50,51) 19,41(17,

40,47,48) (17,45,46) (17,46,48) (21))
20(17,25,33,34,36,51) (18,25,44) (21,25,36,51) (23,25) (25,49) (25,36,40,51)
22,48(17.19,36,46) (17,19,44,46) (17,28,33,34,36,40,45,46,47,51) (17,36,44,

46) (17,41,46) (17,44,46,49)
25, 2j6, 22, 2i. 16, ^.^9, 51(1 7, 21, 44, 45, 46, 47^(17, 44, 45, 46, 47, 50)
27,28,22,34,36.^,51(17) (45,46,47)

28,23,34,26,40,49,51^(17,25,44,45,46,47,48) (44,45,46,4 7)
30(17,36,48) (17,40,41,47,48) (17,46,48) (19)

154

Table 7 -- continued

SHARED

STATES SELECTED COMBINATIONS

* 6.7.9.10.11.12 [6(9,14)(13)1
13.14(9,10.11,12)
15.31(25,26)

* 17,18,21, 23_,44, 45, 46, 42, 50(25, 33, 34, 36, 51) [17,21,^4,45,46,47.50(23,25

26,33,34,36,51) (17,21,33,34,36,40,49,51) (25,28,33,34,36,48,51) 18,
23(17,21,22,44,46,49) (17,21 ,25 ,26,33 ,34,36,44,45 ,46 ,47 ,51) (17 ,21 ,
25,33,34,36,40,44,45,46,47,49,51) (17,21,44,46,48) (19,44,46,49) (21,25,
26, 44, 49) (28, 33, 34, 36, 44, 45, 46, 47, 51) (33,34,36,40,41,44,45,46,47,49,
50,51)1

* iZ. li. 21.26, 32, 34, 36,40, 44, 45., 46, ^,'i'?. 50, 51 r 17,21 ,44,45 ,46,47 50(23,

25,26,33,34,36,51) (23,25,33,34,36,40,49,51) (25,28,33,34,36,48,51) 25,
26,33,34,36,40,49,51(17,18,44,45,46,47) (17,21,23,44,45,46,47) (17,22,
44,45,46,47)1

* 18,20,23(21,25,44) [ 18,23(17 ,21 ,22 ,44,46 ,49) (17 ,21 ,25 ,26, 33 ,34,36 ,44,45 ,

46,47,51) (17,21,33,34,36,40,44,45,46,47,49,51) (17,21,44,46,48) (19,44,

46.49) (21,25,26,44,49) (28,33,34,36,44,45,46,47,51) (33,34,36,40,41,44,
45,46,47,49,50,51) 20(17,18.25,33,34,36,44,45,46,47,50,51) (17,21,25,33,
34, 36, 51) (17, 22, 25) (17,25,26) (17,25,33,34,36,40,51) (18,25,26,44) (18,
25,44,49) (21,25,26) (21,25,49) (21,25,36,40,51) (22,25,26) (22,25,36,40,

51) (2 3,25,26) (25,26,36,51) (25,26,49) (25,28,36,51) (25,36,40,49,51)
19,24,41(17,45,46) (40,45,46)

* 22_,28,33,34,36,4£,48,49,51(17,25,44,45,46,47) (22,48(17,18,19,44,46) (17,

19,28,33,34,36,40,45,46,47,51) (17,19,36,44,46) (17,19,41,46) (17,21,44,
46,49) (17,24,28,33,34,36,40,45,46,47,50,51) (17,28,33,34,36,40,41,45,
46,47,51) (17,41,44,46,49) 28,33,34,36,40,49,51(17,21,25,44,45,46,47,
48) (17,25,41,44,45,46,47,48) (17,25,44,45,46,47,48,50) (18,19,44,45,46,
47) (24,44,45,46,47,50) 1
"" 25., 26, 28, 33, 34, 36, 40, 49, 51(1 7, 44, 45, 46, 47, 48) [25,26,33,34,36,40,49,51(17,
18,44,45,46,47) (17,21,23,44,45,46,47) (17,22,44,45,46,47) 28,33,34,36,
40,49,51(17,21,25,44,45,46,47,48) (17,25,41,44,45,46,47,48) (17,25,44,
45 ,46, 47, 48, 50H18, 19, 44, 45, 46, 47) (24,44,45,46,47,50) 1

* 27,28,33,34,36,40,49,51(44,45,46,47) (45,46,47,50) [27,28,33,34,36,40,51(17,

22) (17,25) (17,41) (17,45,46,47,48) (45,46,47,50) 28,33,34,36,40,49,51
(17,21,25,44,45,46,47,48) (17,25,41,44,45,46,47,48) (17,25,44,45,46,47,

48.50) (18,19,44,45,46,47) (24,44,45,46,47,50)]

30(17,19,48) (17,25,40,41,47.48) (17,33,34,36,40,41,47,48,51) (17,36,46,48) (17,
40,41,45,46,47,48) (19,41)

155

Table 7 -- continued

SHARED

STATES SELECTED COMBINATIONS

13 6.7.9,10.11.12

13.14.(7.9.10.12)
15.31(44)
"" iZ. 11, 20, n, 23, 44, 45, 46, 47.50 (17,18,21,23,44,45,46.47,50(25,26,33,34,
36,51) (25,33,34,36,40,49,51) 18,20,23(21,25,26,44) (21,25,44,49)1
12, 21,25., 26, 33_, 34, 36.,40, 44, 45, 46, 42,'i9, 50, 5J_(23)
19.24.41(17.40.45.46.47.48) (21,45) (25,40,45,47) (40,45,47,50,51)

* 22,25,26,28,33,34,36,40,48,49,51(17,44,45,46,47) [22,28,33,34,36,40,48,

49,51(17,18,19,25,44,45,46,47) (17,21,25,44,45,46,47) (17,24,25,44,45,
46,47,50) (17,25,41.44,45,46,47) 25,26,28,33,34, 36,40,49,51(17,21,
44,45,46,47,48) (17,41,44,45,46,47,48) (17,44,45,46,47,48,50)
27,28,33_,34_,36_,40_,49,51(17,25.44,45,46.47,48) (44.45,46.47.50)
30(17,19,36,48) (17,19,40,41,47,48) (17,22,36,46,48) (17,25,26,40,41,47,48)

(17,25,33,34,36,40,41,47,48,51) (17,25,40,41,45,46,47,48) (17,28,33,34,
36,40,41,45,46,47.51) (17.33,34,36,40,41,47,48,50,51) (17,36,44.46.48)
(21,41)

12 ■'' 6,7.9.10.11.12.13.14

15.31(9) a4) (17,22,44,46,48) (18,23) (19) (21,44,49) (24,25,26)
17,18,20,21,22,44,45,46,47,50(25.26.33,34,36.51) (25,33,34,36,40,49,51)
IZ-Zl, 25, 26_, 33, 34, 36, 40, 44, 45, 46, 47, 49, 50, 51(18, 2 3) (22) (28,48)
19,24,41(17.21.45.46) (17 ,22,28,33 , 34,36,40 ,45 ,46 ,47 ,48, 50 ,51) (17 , 25 ,40 ,

45,46,47,48) (18,25,40,45,47,50,51) (21,25,40,45,47)
22, 25_, 26, 28, 32, 34, 36, 40, 48, 49, 51(17, 18, 19, 44, 45, 46, 47) (17,21,44,45,46,47)

(17,24.44,45,46,47,50) (17,41,44,45,46,47)
27,28,32,34,36,40,49,51(17,21,25,44,45,46,47,48) (17,22,25,44,45,46,47,48)
(17,25,26,44,45,46,47,48) (17,25,41,44,45,46,47,48) (17,25,44,45,46,47,
48) (18,19,44,45,46,47) (23,44,45,46,47) (24,44,45,46,47,50)
30(17,19,22,36,46,48) (17,19,25,40,41,47,48) (17,19,33,34,36,40,41,47,48,51)
(17,19,40,41,45,46,47,48) (17,21,25,40,41,45,46,47,48) (17,22,28,33,34,
36,40,41,45,46,47,48,51) (17,22,36,44,46,48) (17,24,40,41,45,46,47,48)
(17,25,26,33,34,36,40,41,47,48,51) (17,25,26,40,41,45,46,47,48) (17,
25,28,33,34,36,40,41,44,45,46,47,48.49,51) (17,25,33,34,36.40,41,47,
48,50,51) (17,2 7,28,33,34,36,40,41.45,46,47,48,51) (17,28.33.34.36.40.
41,45.46,47,48.50,51)

11 6,7,9,10,11.12.13.14

15.31(6.9) (7,9) (9,14) (9,25,26) (14,25,26) (17,21,22,41,44,46,48,49) (17,22,36,
44,46,48) (18,21,23,44,49) (18,23,25,26) (21,25,26.44.49)

* 11, 18, 20, 21, 23, 25, 26, 22,34, 36_, 40, 44, 45., 46, 47, 52, 51 (17,18,20,21,23,44,

156

Table 7 -- continued

SHARED

STATES SELECTED COMBINATIONS

45,46,47,50(22,25,33,34,36,40,49,51) (25,28,33,34,36,48,51) 17,21,
25,26,33,34,36,40,44,45,46,47,49,50,51(18,22,23) (2 3,48) (28,41,48) 1

* 12, 11, 12,22,26, 28, 33, 34, 26, 40, 44, 45, 46, 47, 48, 49^, 5J,(24) (17,21,25,26,33,

34,36,40,44,45,46,47,49,50,51(18,22,23) (23,48) (28,41,48) 22,25,26,
28,33,34,36,40,48,49,51(17,18,19,21,23,44,45,46,47) (17,18,19,24,44,
45,46,47,50) (17,18,19,41,44,45,46,47) (17,21,41,44,45,46,47) (17,24,

31.41.44.45.46.47.50) ]

* 19,24,30,41(40,45,46,47,48 [19,24,41(17,18,20,22,25,28,33,34,36,40,44,

45,46,47,48,49,50,51) (17, 21, 25, 40, 45, 46, 47, 48) (17, 25, 26, 40, 45, 46, 47,
48) (18,21,25,40,45,47,50,51) (2 7,40,45,47,50,51) 30(17,19,22,28,33,
34,36,40,41,45,46,47,48,51) (17,19,22,44,46,48) (17,19,25,26,40,41,47,
48) (17,19,25,33,34,36,40,41,47,48,51) (17,19,25,40,41,45,46,47,48) (17,
19,33,34,36,40,41,47,48,50,51) (17,21,25,26,40,41,45,46,47,48) (17,21,
25,28,33,34,36,40,41,44,45,46,47,48,49,51) (17,22,24,28,33,34,36,40,

41.45.46.47.48.50.51) (17,22,25,28,33,34,36,40,41,45,46,47,48,49,51)
(17,22,2 7,28,33,34,36,40,41,45,46,47,48,51) (17,24,25,40,41,45,46,4 7,
48) (17,25,26,28,33,34,36,40,41,44,45,46,47,48,49,51) (17,25,26,33,34,
36,40,41,47,48,50,51) (17,25,28,33,34,36,40,41,44,45,46,47,48,49,50,
51) (17,27,28,33,34,36,40,41,45,46,47,48,50,51)1

■•■' 21, 25., 16, 2 7, 28, 32,34, 16, 40, 48, 49, 51(17, 44, 45, 46, 47) [22,25,26,28,33,34,
36,40,48,49,51(17,18,19,21,2 3,44,45,46,47) (17,18,19,24,44,45,46,47,
50) (17,18,19,41,44,45,46,47) (17,21,41,44,45,46,47) (17,24,31,41,44,45,
46,47,50) 27,28,33,34,36,40,49,51(17,18,19,22,25,44,45,46,47,48)
(17,21,22,25,44,45,46,47,48) (17,21,25,26,44,45,46,47,48) (17,21,25,
41,44,45,46,47,48) (17,21,25,44,45,46,47,48) (17,21,25,44,45,46,47,50)
(17,22,24,25,44,45,46,47,48,50) (17,22,25,41,44,45,46,47,48) (17,25,
26,41,44,45,46,47,48) (17,25,26,44,45,46,47,48,50) (17,25,41,44,45,46,
47,48,50) (18,19,23,44,45,46,47) (18,19,24,44,45,46,47,50^(23,44,45,
46,47,50) )

* 27,28,30,21,34,36,40,49,51(17,25,41,44,45,46,47,48) [27,28,33,34,36,

40,49,51(17,18,19,22,25,44,45,46,47,48) (17,21,22,25,44,45,46,47,48)
(17,21,25,26,44,45,46,47,48) (17,21,25,41,44,45,46,47,48) (17,21,25,
44,45,46,47,48) (17,21,25,44,45,46,47,50) (17,22,24,25,44,45,46,47,48,
50) (17, 22, 25, 41, 44, 45, 46, 47, 48) (17,25,26,41,44,45,46,47,48) (17,25,
26,44,45,46,47,48,50) (17,25,41,44,45,46,47,48,50) (18,19,23,44,45,
46,47) (18,19,24,44,45,46,47,50) (23,44,45,46,47,50) 30(17,19,22,28,
33,34,36,40,41,45,46,47,48,51) (17,19,22,44,46,48) (17,19,25,26,40,
41,47,48) (17,19,25,33,34,36,40,41,47,48,51) (17,19,25,40,41,45,46,47,
48) (17,19,33,34,36,40,41,47,48,50,51) (17,21,25,26,40,41,45,46,47,48)
(17,21,25,28,33,34,36,40,41,44,45,46,47,48,49,51) (17,22,24,28,33,34,
36,40,41,45,46,47,48,50,51) (17,22,25,28,33,34,36,40,41,45,46,47,48,

157

Table 7 -- continued

SHARED

STATES SELECTED COMBINATIONS

49,51) (17,22,27,28,33,34,36,40,41,45,46,47,48,51) (17,24,25,40,41,45,
46,47,48) (17,25,26,28,33,34,36,40,41,44,45,46,47,48,49,51) (17,25,26,
33,34,36,40,41,47,48,50,51) (17,25,28,33,34,36,40,41,44,45,46,47,48,
49,50,51) (17,2 7,28,33,34,36,40,41,45,46,47,48,50,51) I

10 6.7.9.10.11.12,13.14(31)

* l^.IL.2i,22_,25_,26_,28^n'li>i^>2i'itO.M..i5,46.iZ.iL8.'i9.50,ll(24,41) (15,31

(6,7,9) (6,9,14)(6,9,25,26) (7,9,14) (7,9,25,26) (9.14,25,26) (12) (13,14)
(17,19,22,36,44,46,48) (18,20,21,2^,25,26,44,49) (18,23,24,25,26) 17,
21,22,25,26,28,33,34,36,40,44,45,46,47,48,49,50,51(18,19,23,24) ]
17,18,20,21,23,25,26,33,34,36,40,44,45,46,47,50.51(22,49)

19, 24, 30, 4j^( 17, 22, 28, 33, 34, 36, 40, 45, 46, 47, 48, 50, 51) (17,25,40,45,46,47,48)
22, 25., 16, 2 7, 28, 33, 34, 36, 48, 49, 51(17,18, 19, 20, 31, 41, 44, 45, 46 .47, 50) (17,18,

19,44,45,46,47) (17,21,44,45,46,47) (17,24,44,45,46,47,51) (17,41,44,45,

46,47)
27,28,30,33,34,36,40,49,51(17,21,25,41,44,45,46,47,48) (17,22,25,41,44,45,

46,47,48) (17,25,26,41,44,45,46,47,48) a7, 25 ,41 ,44,45 ,46,47,48,50)
_ _

©
6.7.9.10.11.12.13.14(15.31)

* 15, 17, ]J_, 20, n,22_, 22, 25, 26, 28, 21, 33, 34, 36, 40, 4^, 4^, 4^, 47, 41, 49, 50, 5^
19,24,30,41(17,18,20,22,25,28,33.34.36,40,44,45,46,47,48,49.50,51) (17,

21,25,40,45,46,47,48) (17,22,27,28,33,34,36,40,45,46,47,48,50,51) (17,
25,26,40,45,46,47,48)

* 22, 25, 26, 2 7, 28, 30, 33_. 34. 36^.40, 48. 49, 5^(1 7, 41, 44, 45, 46, 47) [22,25,26,27,28,

33,34,36,40,48.49.51(17,18,19,21,23,44,45,46,47) (17,18,19,24,44,45,46,
47,50) (17.18,19,41,44,45,46.47) (17,21,24,44,45,46,47,50) (17,21,41,44,
45,46,47) (17,24,31,41,44,45,46,47,50) 27,28,30,33,34,36,40,49,
51(17,18,19,22,25,41,44,45,46,47,48) (17,21,22,25,41,44,45,46,47,48)
(17,21,2 3,25,41,44,45,46,47,48) (17 ,21 ,25 ,2 6,41 ,44,45 ,46,47 ,48) (17 ,
21,25,41,44,45,46,47,48,50) (17,22,24,25,41,44,45,46,47,48,50(17,2 5,26,
41,44,45 ,46,47,48,50) ]

158

Table 7 -- continued

SHARED

STATES SELECTED COMBINATIONS

8 ®
2

0
4
i'l.i.iO 'ii'li-il'liL^lS .25 ,26,31)

19., 24, 30, 41 (17, 18, 20, 2 1,22, 2 3, 25, 28, 33, 34, 36, 40, 44, 45, 46, 47, 48, 49, 50, 51)
(17,18,20,22,25,2 7,28,33,34,36,40,44,45,46,47,48,49,50,51) (17,21,25,
26,40,45,46,47,48)

22_, 25, 26, 27, 28, 30, 33, 34, 36, 40, 48, 49, 51_( 17, 18, 19, 41, 44, 45, 46, 47) (17,21,41,
44,45,46,47^(17,24,31,41,44,45,46,47,50)

7 * 111

* 3.4
6,7.9,10.11.12,13,14(15.25,26,31)

* li,lZ.ii.ii.20,2i,21,22.24,25_,16,2 8,30,n,23.34,36,40,4i,44.«,4^,^

49,50,5J. [15,17,18.20,21,22.2 3.25,26,2 8.31.33,34.36,40.44,45,46.47,
48,49,50.51(19,24.2 7.41) 19,24.30.41(17.18.20,21.22,23,25.2 7,28,33,
34.36.40.44.45.46,47,48.49,50.51)1

* ii. 12, 24, 25., 26., 27, 28, 30. 22. 34, 36., 40. 41, 48_.49, 51^(17, 18, 20, 31, 44, 45, 46. 47,

50) (19,24,30,41(17,18,20,21,22.23.25,2 7,28,33,34,36,40,44.45,46,
47,48,49,50,51) 22,25,26,27,28,30,33,34,36,40,48,49,51(14,15.17.21.
24,31,41,44,45,46,47,50) (17, 18,19,21 . 23 ,41,44.45 .46 ,47)|

6 1.2
3.4
6.7.9.10.11.12.13.14(15.21.25,26,31.44.49) (15,24,25,26,31)

* 15,17,18,19,20,21,22,23,24,25,26,2 7,28,30,31,33,34,36.40.41,44,45,46,47.

48,49,50,51(14) [15,17,18,19,20,21.22,23,24,25.26.28,30.31.33.34,
36,40,41,44,45,46,47,48,49,50,51(6,9) ]

5 1,2
3.4
6.7.9.10.11.12.13.14(15,17,21,22,24,25,26,2 7,28,30.31,33,34,36,40,41,44,45,

46,47,48,49,50,51) (15,18,20,21,23,25,26,31.44.49) (15.19.24.25.26.31)
15.17.18.19.20.21.22.23.24.25.26.27.28.30.31.33.34.36.40.41.44.45.46.47,

48.49.50.51(6.9,14) (13,14)

4 1.2(3)

3.4(7) (2)

* 6.7.9.10.11.12.13.14.15.17.18.19.20.21.22.2 3.24,25.26.2 7.28,30,31.33,34.

36.40.41.44.45.46.47.48.49.50.51

159

Table 7 -- continued

SHARED

STATES SELECTED COMBINATIONS

1,2,3,4 (1,2(6) (7,9,10) 3,4(7,9,10,11,12,13)1

6,7,9,10,11,12.13.14.15.17,18,19.20,21.22,23.24.25.26.2 7.28,30.31,33.34,
36,40,41.44.45.46.47.48.49.50.51

1.2.3.4(6) (7,9,10,11,12,13H15)

6.7.9.10.11.12.13.14.15.17.18.19,20.21,22,2 3,24,25,26.2 7.28.30.31.3 3.34,
36.40.41.44.45,46,47,48,49,50,51

1.2.3.4(6.7.9.10.11.12.13.14.15.31) (7,9, 10,11 , 13 ,27,28) (15 ,24)
6.7.9.10.11.12.13,14,15.17.18.19.20.21.22.23.24.25.26,2 7.28.30.31.33.34.
36.40.41.44.45,46,47,48.49.50.51

160

Combinations of species selected from those generated by the com-
binatorial method from 40 species and the 23 character states at
the third character state tree layer (see text for selection pro-
cedures, and see heading of Table 5 for an explanation of the

punctuation system used).

SHARED

STATES SELECTED COMBINATIONS

10

@

o

20
22

27

28,@

28,48

47
49

* 17,20
18

* 20,47
22 ''48)
©

* 27,34 [27(28) 34(44) I

* 28,34,40,47 [34(44) ]

* 28,40,48
©

-f 34,49 [34(44) 49(44) |
41

* 44,45,46

* 44,46,47

161

Table 8 -- continued

SHARED

STATES SELECTED COMBINATIONS

■■■• 45,47

47,49 [49(44)
50

* 17,20,27,28,34,40,47 (17,20,(48) 27,34(44) (49) )
■•■' 18,45,47

* 20,47,49
22(27) (28,48)
23

* 28,34,40,47,48,49 [34,49(44)1

* 28, 34, @, 40, 47

* 28,34,40,44,46,47

* 28,40,48,(0)

* (3d. 41 [41(20) (47)1
33(28,40) (34,44)

* 44,45,46,47
50(20) (28,40) (47)

17,20,27,28,34,40,47(36) (44,46) (48,49)

* 18,44,45,46,47 [18,45,47(41)]

20,47,49(41)

22a7,20,48) (27,28,48) (28,40,48)

* 23,33 [23(49) 33 (27 , 34,44) (34 ,44 ,^9) ]
@

* 28,33,40,48,51 [ 33(27,34,44) (34,44 ,49) ]

* 28,33,34,40,44,45,46,47 [ 33 (27, 34 ,44) (34,44 ,49) ]

* 28,34,36,40,47,48,49

* 28,34,36,40,44,46,47,50 [ 50 (20 ,47) (45 ,47) (47,49) 1

* 28,34,40,44,46,47,48,49,51
28,40,48,50,51 [50(20,47) (45,47) (47,49) |
30,41(20) (48)

@

* 33,50 (33(27,34,44) (34,44,49) 50(20 ,47) (45 ,47) (47,49)

162

SHARED

STATES SELECTED COMBINATIONS

5 @(20)

* 17,20,22,27,28,34,36,40,47,48,49 [ 20 ,47,49(18 ,41 ,45) (50) 1

* 17,20,27,28,30,34,40,41^,47,49(48) [ 20 ,47,49 (18,41 ,45) (50) ]

* 17,20,27,28,33,34,40,44,45,46,47

* 17,20,27,28,34,36,40,44,46,47,50

* 17,20,27,28,34,40,44,46,47,48,49,51 ( 20 ,47,49''18 ,41 ,45) (50) ]

* 18,28,33,34,40,44,45,46,47,48,49,51 [ 18,44,45 ,46 ,47''41) ]

* 19,28,34,40,44,46,47,48,49,51 [19(18)]

* 22,28,40,48,50,51

* 23, 28, 33, 40, 4;-., 51

■•■^ ©,26 126(17,20,48) J

'■' 28,33,34,36,40,44,45,46,47,50

* 28,33,40,48,50,51

* 28,34,36,40,44,46,47,48,49,50,51
31(44,45,46)

33,50

4 @)

15(18,20,2 7,28,34,40,47) (20,30,41)

* 17,18,20,27,28,30,33,34,40,41,44,45,46,47,48,49,51 [17,20,27,28,30,34,40,

41,47,49(26,48) ]

* 17,19,20,2 7,28,34,40,44,46,47,48,49,51

* 17,20,22,2 7,28,30,34,36,40,41,47,48,49 [17,20,27,28,30,34,40,41,46,49(26,

48)1

* 17,20,22,2 7,28,34,36,40,44,46,47,48,49,50,51

* 17,20,2 7,28,33,34,36,40,44,45,46,47,50

* 18,19,28,33,34,40,44,45,46,47,48,49,51

* 18,2 3,28,33,34,40,44,45,46,47,48,49,51

* 18,28,33,34,36,40,44,45,46,47,48,49,50,51

* 19,28,34,36,40,44,46,47,48,49,50,51
(Q)(17,20,48)

* 22,28,33,40,48,50,51

* 23,28,33,40,48,50,51
@(18,41,44,45,46,47)
25.26(17,20.48) (18,19)
31(18,44,45,46,47) (41) (44,45,46,49)

163

Table 8 -- continued

SHARED

STATES SELECTED COMBINATIONS

3 10

* 15,31(20,30,41) [15(17,20,26,2 7,28,30,34,40,41,47,48,49) (17,20,27,28,33,

34,40,44,45,46,47) (17,20,2 7,28,34,36,40,47) 31(18,24,41,44,45,46,47)]

* 17,18,19,20,24,25,26,2 7,28,30,33,34,40,41,44,45,46,47,48,49,51 [25,26(17,

20,22,48)1

* 17,18,20,21,23,2 7,28,30,33,34,40,41,44,45,46,47,48,49,51 [21(17,20,22,48)]

* 17,18,20,22,2 7,28,30,33,34,36,40,41,44,45,46,47,48,49,50,51 [17,20,22,27,

28,30,34,36,40,41,47,48,49(26) )

* 17,19,20,22,27,28,34,36,40,44,46,47,48,49,50,51

* 18,19,23,28,31,33,34,40,44,45,46,47,48,49,51 [31(18,24,41,44,45,46,47)]

* 18,19,28,33,34,36,40,44,45,46,47,48,49,50,51
" 18,2 3,28,33,34,36,40,44,45,46,47,48,49,50,51
-21,25,26(17,20,48) [21(17,20,22,48) 25,26(17,20,22,48)]

* 22,23,28,33,40,48,50,51

2 *(3@10,(0) [10(1)]

* 10 ,©,©(15, 20) [10(17)]

" 15, 17, 18, 19, 20, 21, 23, 24, 25., 26, 2 7, 28, 30, 31, 33, 34, 40, 41, 44, 45, 46, 47, 48, 49,
51 [21,25,26(17,20,22,48)]

* 17, 18, 19, 20, 22, 24, 25, 26., 2 7, 28, 30, 33, 34, 36, 40, 41, 44, 45 ,46, 47, 48, 49, 50, 51

* 17,18,20,21,22,23,2 7,28,30,33,34,36,40,41,44,45,46,47,48,49,50,51

* 18,19,2 3,28,31,33,34,36,40,44,45,46,47,48,49,50,51

1 (4X18,20,23,41,45,47,49,50)

*(6) 7, 9, 10, 11, (g), 13, 14 (15, 20, 30, 31, 41) [10,13,14(15,17,20,27,28,33,34,36,
40,44,45,46,47,50)1
7,9,10,11(18)

* 15,17,18,19.20.21,22.23.24,25.26.2 7.28,30,31.33.34.36,40,41,44.45,46,47.

48.49.50.51

164

Table 9 -- Combinations of species selected from those generated by the com-
binatorial method from 40 species and the 10, 3, and 1 character
states at the fourth, fifth and sixth character state tree layers
respectively (see text for selection procedures, and see heading
of Table 6 for an explanation of the punctuation system used).

SHARED

STATES SELECTED COMBINATIONS

FOURTH LAYER

©

* ^,(^.(51)
@

@
©

* 23,33,@

@

* 30.^

@

* 33, @

*X<. 18, 24, 3 1,48, 51
*@49

* @,2 8

* 18,©,23,24,@,33,44,45,@,|^,51

* 20,22,36,@),48(30)

* 21,22(23)

* 21,30,33,41,45

* 23,33,36,44,49

* 0,28,30,31,41

165

Table 9 -- continued

SHARED

STATES SELECTED COMBINATIONS

FIFTH LAYER

2 *(^MMi

1 (^

* @), 40, 50, 51

* @),@,(r^,40,@,O,@,@,50,5i

SIXTH LAYER

1 ^^Q^iMMMMM

166

li

Is

■1
1

:

i
II

i

;■

i

?

:■

;

;

:•

S

'

;

:■

;

I 1

• :

2

5

■

;

:•

;

:;

;

1

s : -s^
■St - :
; s :; ;

! 2 ! E i
s » " ss

1? leJii !i ft! ? ' s ^i- 1 fi-s =

11 lllll-l^s lsl|i|s|||||;|:|;!:f|P l

Ills

||l ii|Lj

llllllllll

jIi nil

slltl i

|S ij ■ 1

s 1

1 1 If .

I 3 si 2„l

1 Ififs ;;|
llliylllll

^ 1 » 1
1 f i :

I I s S g
"IS Si'?:

II ill III

III s ill! 1'°

s K a si s' c 2 s s : s

i

: i

s i 4 i

ssss

i

If
II

Jill
S 3 s s

1
i

is

Is
fi-S
III

i

t-

Ls if If.

■ ? -'1

I2I2II

■ s • s • •

1 y III

= : ■ • = s s -
III U ! 1 l^ts

IlliJil ililsllll

a.,»

'"•

-

1V,1».

mi

MO,

1,.,

on~

,m.

,1

,n» '

a™

,.0.

IIYI

™,

»»r

iao.

,1,1

»-'"™-

jy,«.«„

IJ3S

IVUHMS

--

-»"-

iviaiixs

J

M«1».V lV(.,3IilI

™.JL-,.

™»„«i

5

s„.».

S^M.i»«»

3»u«mv*

UNIVERSITY OF ILLINOIS-URBANA

3 0112 033017044