EVOLUTION, SYSTEMATICS, AND VARIATION OF
PACIFIC MANGROVE MONITOR LIZARDS
ROBERT
GEORGE SPRACKLAND, Ph.D., F.L.S., F.Z.S.
Virtual Museum
of Natural History, POB 326,
Mukilteo, WA 98275, USA
and
Department of
Anatomy and Developmental Biology, University College London
Research
Associate in Zoology, The National Museums of Scotland
(31
figures, 5 tables)
Present
address: Oregon Museum of Science and Industry, 1945 SE Water Avenue, Portland,
OR 97045
©1999
Robert George Sprackland, Ph.D.
CONTENTS
PREFACE 3
INTRODUCTION
6
Taxonomy of
Mangrove Monitors, 1802-1996
20
MATERIALS
AND METHODS
31
Character
Analysis
38
Colour and
Pattern Variation
72
RESULTS
77
Populations and
Species: Analyses of Data
77
Discovery of
Species
80
Phylogenetic
Relationships Among Mangrove Monitors
81
DISCUSSION
87
Cladistic
resolution of taxa
87
Taxon
designation and recognition
90
Problems in
data interpretation
92
Morphological
limits in diagnosing cryptic taxa
93
Zoogeography
94
Classification
98
Data compared
with other evolutionary studies of Varanidae
98
Taxonomic
Accounts
103
Key to the
Species of Mangrove Monitor Lizards
131
CONCLUSIONS
133
POSTSCRIPT
137
SUMMARY
138
ACKNOWLEDGMENTS
139
REFERENCES
140
APPENDIX
I: Institutional Abbreviations
154
APPENDIX
II: Gazetteer of Locality Names
155
APPENDIX
III: Preserved
Specimens Examined
156
APPENDIX
IV: Complete Data Matrix for Mangrove Monitor Lizards
163
All
illustrations by author unless otherwise credited.
PREFACE
The facts that support the theory or organic evolution have been
accumulating at an extraordinary pace. It seems ironic that in the United
States of the late twentieth century so many people remain ignorant, if
not outright hostile, to the scientifically demonstrable concept that is
arguably the most rigorously supported theory in science. After all, what
other theory (in the scientist's sense of the word, and not the colloquial
use that equates to "hypothesis") is braced by numerous
independent theories coming from all branches of science? Physics,
chemistry, geology, biology, even meteorology form the foundation stones
upon which evolutionary theory rests. Perhaps the problem at the
large-scale lack of understanding comes from the disparate sources of
data: a fruit fly here, a microbe there, some polyploidy in an orchid, DDT
resistance in mosquitoes.
Monitor lizards are
excellent organisms for students of evolution to display as examples of
evolutionary theory. The known species number fewer than 100, making
monitors members of a manageable taxonomic group. Many of the species have
been studied in terms of ecology, zoogeography, physiology, reproduction,
paleontology, morphology, and taxonomy. The result is a vertebrate group
for which a gestalt picture exists. Biologists may point to varanids and
compare "oranges with oranges," which must simplify descriptive
biology considerably.
My own interest in
monitors began in 1969, and has held prominence over most of my other
zoological interests ever since. In 1993 I was fortunate enough to earn a
scholarship to pursue doctoral studies at University College London. My
advisor, Susan Evans, was interested in varanids from the palaeontological
perspective (looking from the beginning forward), while my interests
started with extant taxa evolutionary history perspective (looking
backwards). The species chosen for my study was "the" mangrove
monitor, following advice from Walter Auffenberg to do doctoral work on a
well-represented taxon for which lots of specimens existed. Little did any
of us know at that time that the mangrove monitors would prove to be a
complex group of cryptic species, some of which had yet to be described!
The following account
is slightly modified from my thesis. Two major projects were brought
together in this study. First, I have reconsidered the methods of
phylogenetic analysis as conventionally employed. From statistical and
logical standpoints, I found procedural assumptions that seemed at odds
with scientific standards for objectivity. My interpretation and use of
both cladistics and multiscale analysis may be less than satisfying to
many workers, but they are objectively designed, mutually testing methods
that address major complaints about black-box phylogenetics. As a test for
the hypotheses I present, I expect my phylogenetic method to be able to
sort multiple individual specimens into discrete species groups. While
much remains to be done to refine this procedure, this text provides
painstakingly researched first steps.
Second, I have
documented the range of variation among mangrove monitors, aimed at
stabilizing their taxonomy, and by so doing providing a framework for more
effective use of macromorphological characters. I trust that the character
analysis I present will clarify many incorrect interpretations from
earlier literature, and serve as a basis for additional character
inclusion.
Mangrove monitors are
neither more nor less interesting than other lizards, but their cryptic
species pattern makes them wonderful "higher" vertebrate models
for studying many aspects of evolution. The primary difficulty in studying
these species is their distribution in parts of the world that still
recall the old map-maker's words, "here be dragons." In reply, I
would respond "aye, and so many!"
ROBERT
GEORGE SPRACKLAND
Montara,
California
June,
1999
DEDICATION
For
my parents LUCILLE SMITH (1930-1996) and
JOSEPH FRANCIS SMITH (1919-1994).
Their
encouragement, support, and love will always be honored and cherished.
INTRODUCTION
The taxonomy of
species, or alpha taxonomy, is often considered well-defined and largely
complete for terrestrial vertebrates. As a consequence, many recent
systematic reviews and revisions have been concerned with higher taxonomic
categories (e.g., Baverstock et al.,
1993; Böhme, 1988; Hedges et al.,
1991; Holmes et al., 1975; King
and King, 1975; King et al.,
1991; Storr, 1980; see also Estes and Pregill, 1988). Implicit in each of
these studies is the belief that the species that comprise the higher
groupings are well-defined. In reality, this is not always the case.
Sometimes, very similar species are difficult or impossible to distinguish
based only on morphological examination (=cryptic, or sibling, species),
and have thus remained unknown to science until recently. Examples include
the Californian slender salamanders of the genus Batrachoseps,
diagnosed by skin gland proteins, and Australian hylid frogs related to Litoria ewingii, distinguished by male vocalizations. Sometimes
intraspecific variation can only be detected at the molecular level
(Sattler and Ries, 1995), or is jointly linked to sex and age-class
(Carpenter, 1995), complicating analysis. Furthermore, higher taxonomic
studies rely on the near-universal acceptance of standards for higher
taxonomic categories, though these, too, are often subjective (Dubois,
1988) or deemed archaic constructs (Gauthier et
al., 1988).
Authors such as Storr
(1980) have named new varanids without examination of type specimens of
related taxa. This approach has led, for example, to some of Storr's
(1980) names (e.g., Varanus panoptes)
being placed in junior synonomy to established names (Varanus
gouldii) (Böhme, 1991; Sprackland, 1993b, 1994a). Mertens (1951),
too, named one species, Varanus
karlschmidti, without comparing it to the types of related species,
thereby overlooking the identity of V.
karlschmidti with the earlier V.
indicus jobiensis of Ahl (1932). Nor is it unusual for a reviser to
make the a priori assumption that someone else has taken care of the alpha
taxonomy. For example, in revising the phylogeny of pythonine snakes,
Kluge states "The species I recognise are those accepted in the
majority of the recent papers on pythonines" (Kluge, 1993:5). Clearly
the scope for possible taxonomic error is compounded when the limits of
taxa considered species are not properly defined (Campbell and Frost,
1993; Good, 1994; Mayr and Ashlock, 1991). Neither is the problem
restricted to neontologists with large samples to study; the sixteen
described species of the Cretaceous dinosaur genus Triceratops
have recently been reviewed, resulting in the recognition of one species
with a normal intraspecific range of variation (Ostrom and Wellnhofer,
1990).
Mangrove monitor
lizards, so called because of their preferred habitat, represent an
important component of the
Indo-Australian carnivorous reptilian faunas, for which alpha taxonomy has
not been adequately addressed. They were first described in the early
nineteenth century (Daudin, 1802), and assigned the name Tupinambis indicus, the trivial name reflecting that the original
specimen was collected from the East Indies island of Amboin (now part of
central Indonesia). Since then, at least eleven other names have been
applied to mangrove monitors from across their huge range (see below). One
problem facing monitor systematists is that early descriptions either
lacked scientifically informative illustrations (Daudin, 1802) or any
illustrations at all (Gray, 1831, 1838; Lesson, 1830), while another
problem has been identifying type localities; Rawack, Kalum (also spelled
Ralum), and Dore (or Doreh or Dorei) are cases in point (see Appendix II).
At the time the present study began, three species were recognised among
mangrove monitors (V. indicus, V.
jobiensis, and V. spinulosus),
and another two cryptic forms, V.
doreanus doreanus and V. d.
finschi, were described (Böhme et
al., 1994) while this study was in progress.
In view of recent work with a variety of lizard families (Arnold,
1991; Auffenberg, 1994; Campbell and Frost, 1993; Crombie and Steadman,
1986; Gaulke, 1989; Good, 1994; Mouton et
al., 1992; Thorpe and Brown, 1989), it is imperative that the range of
intraspecific variation be discovered prior to using data to amend an
existing taxonomy.
The distribution of
mangrove monitors extends from the Halmahera Islands, east of Sulawesi,
Indonesia, east to the Solomon Islands, and from Guam south to northern
coastal Australia (Barbour, 1912; Böhme et al., 1994; Boulenger, 1885, 1897; Cogger, 1992; Doria, 1874;
Fisher, 1948; Günther, 1877; Hediger, 1934; Horn, 1977; Luxmore et
al., 1988; McCoy, 1980; Mertens, 1942c, 1963; O'Shea, 1991; Parker,
1970; Peters, 1876; Peters and Doria, 1878; Sprackland, 1992; Werner,
1900; Wilson and Knowles, 1988). This area includes most of lowland New
Guinea, and hundreds of tiny islands. There are questionable records for
Sulawesi (=Celebes), while records for Timor are actually misprints for
Timor Laut. Nevertheless, such a wide distribution is often a zoological
marker to indicate that more than one species is represented within that
range, because non-migratory terrestrial vertebrates are usually
restricted by ecological and geological barriers to smaller ranges. Böhme
(1991) and Sprackland (1993a) have recognized two "subspecies"
(sensu Mertens, 1942c) of Varanus "indicus" as full species, but both authors also
concede that the status of V.
"indicus" itself needs re-evaluation. Böhme et
al. (1994) have described a cryptic mangrove monitor, while several
clearly geographically linked variations may represent other new species,
or justify resurrection of older names from the literature.
Prior to World War II,
mangrove monitors (Varanus indicus,
sensu stricto, see below,
systematic accounts) were introduced to various Caroline and Marshall
Islands, largely by Japanese researchers trying to evaluate the lizards as
rat-control agents (Fritts, 1993; Rodda et
al., 1991; Uchida, 1966, 1967; Wiles et
al., 1989). This introduction greatly expanded the already tremendous
range for the group, and added confusion to any analysis of their
zoogeography. Records for the sources of the original specimens are
unavailable, but most closely resemble lizards from Halmahera, Seram, and
Ternate in both pattern and colour.
Mangrove monitor
lizards are all large, growing to 1 m or more in TL, and share similar
ecologies, generally living in forests near standing bodies of water,
though on Guam (where V. indicus
is almost certainly a human introduction; Fritts, 1993) they live in
"an uplifted limestone plateau" that has "no permanent
bodies of fresh water" (McCoid and Hensley, 1993). They are highly
adept swimmers, climbers (Barbour, 1912; Hediger, 1934; Kalken, 1994;
Neugebauer, 1976; Sprackland, 1992), and alert predators, feeding upon
smaller vertebrates and large arthropods (Hediger, 1934; Losos and Greene,
1988; pers. obs.). The study by Losos and Greene (1988) lacks a list of
specimens examined, nor is it clear how many animals they dissected (N=18,
41, 54?); it is probable that their stomach content analyses include two
or more of the cryptic taxa under consideration. I include information
about stomach contents in the comments portion of the species accounts to
follow.
Though mangrove monitors are common in some areas, it will be
difficult to determine the status of populations until the taxonomy is
clarified (Luxmore et al., 1988;
Groombridge, in prep.). Accounts of the biology of mangrove monitors in
the wild are brief and anecdotal (Barbour, 1912; Fisher, 1948; Hediger,
1934; Loveridge, 1941; Neugebauer, 1976; O'Shea, 1991; Rotter, 1963;
Werner, 1900; Wilson and Knowles, 1988), or deal with isolated aspects of
insular populations (Dryden, 1965; Hediger, 1934; McCoid, 1993; McCoid and
Hensley, 1991; McCoid and Witteman, 1993; McCoy, 1980; Wikramanayake and
Dryden, 1988). Reports concerning live animals in zoological collections
are rare (Balsai, 1993; Horn, 1977; Kalken, 1994; Kukol, 1993; Murphy,
1972; Nelling, 1995; Polleck, 1979; Sprackland, 1989, 1992, 1993a, 1997).
Mangrove monitors in captivity, in particular in collections of amateur
herpetoculturists, have begun to breed (Kok, 1993; Wesiak, 1993a, 1993b),
providing access to juveniles so that ontogenetic changes in colour and
pattern may be observed.
Little comparative
osteology has been undertaken (Mertens, 1942b), though this avenue of
inquiry may prove useful in resolving taxonomy of cryptic species. In my
preliminary examination of skeletal material, I noted that specimens of "Varanus
indicus" from the western localities had a different nasal bone
structure than those from the east. This variation was also noted by
Mertens (1942c) though he failed to ascribe it to the geographic source of
his specimens. The six skulls available to Mertens (1942b) represent
lizards of different ages, at least two species (V.
"indicus" and V.
doreanus), and some without locality data, thus obfuscating any
intrinsic pattern he might otherwise have discovered. There also appear to
be minor variations in the structure and position of the septomaxillae
which might be of taxonomic significance (Sprackland, in prep.). At this
time, no museum skull preparations exist for Varanus
jobiensis or V. spinulosus.
A specimen of V.
jobiensis from Irian Jaya, without specific locality data, was
obtained for skeletal preparation for this study. It remains one of the
banes of comparative osteology that specimens chosen for skeletonization
by museum curators are typically those without locality data and, less
frequently, possess incorrect taxonomic identification. Osteological
studies will be impeded until a better identified series of specimens is
made available for dissection.
Mangrove monitors have
become important to zoologists over the past decade for several reasons:
the increased use of monitor skins for production of native drum heads
(requiring proper identification to enforce international wildlife trade
laws); the interest in "tramp species" dispersal through western
Pacific islands (Groombridge, in prep.; Luxmore et
al., 1988); and because of the tremendous potential to learn about
captive propagation of varanids by using this widespread, hardy species (Kok,
1993; Sprackland, in press; Wesiak, 1993a, 1993b). At present, knowledge
of varanid reproductive biology is very limited (Horn and Visser, 1989;
McCoid, 1993; Sprackland, 1989, 1992; Wikramanayake and Dryden, 1988),
despite the status of most species as either rare, endangered, or
potentially threatened wildlife. The combined resources of field
ecologists (Auffenberg, 1988, 1994; Gaulke, 1989; James et
al., 1992) and herpetoculturists (e.g., Kok, 1993; Kukal, 1993; Wesiak,
1993a & b) make the Varanus
indicus complex likely candidates for producing sustainable,
captive-bred specimens for further research.
For the present study,
macromorphological characters were examined in order to: 1) record the
range of variation within and among mangrove monitors,
2) discover species-level clades in this assemblage, using
statistical and phylogenetic analyses to discover content species,
3) analyse these data to discern zoogeographical patterns that may
reflect on species relationships, and
4) establish a sound taxonomy based on Hennig86 analysis (Farris,
1986) and on an exhaustive review of literature and museum holotypes.
A general description
of varanids contains a combination of plesiomorphic (=ancestral)
characters including: deeply bifid, retractable tongue (as compared with Heloderma
and Lanthanotus); four
well-developed, pentadactyl, clawed limbs; eyelids and ear openings
present; pupils round; tail at least equal to the snout-vent length (SVL),
usually in excess of 1.5 times longer; crests, frills, and other
ornamentation lacking. Twenty-nine synapomorphies have been given (Estes et
al., 1988) for the Varanidae, but many of the characters employed are
homoplasious (e.g. forked tongue, parietal muscle attachment), or present
in only some taxa (e.g. paired nasals). I have a full review of the family
in preparation, in which the characters of Estes et al. (1988) are re-evaluated and modified. For purposes of the
present discussion, the nine unambiguous synapomorphies for varanids may
be given as:
1) Nasals and maxillae
never in contact, nasals and prefrontals in little or no contact,
producing the largest nasal foramen in extant lizards. It has been
suggested (McDowell and Bogert, 1954; Underwood, 1957) that the snout
attenuated as a derived feature, and the nostril secondarily migrated from
a posterior position near the orbit, to a more anterior position. This
view was supported by data analyzed by Sprackland (1991a, 1991b), in which
Varanus griseus was determined
to be the most ancestral extant varanid based on the analysis of 57
characters. In that species, the nostril is so posterior in position as to
contact the anterior rim of the orbit.
2) Nasal process of
maxilla rises from a posterior position on the maxilla. This placement is
subject to three major variations (Fejervary, 1935; Mertens, 1942a, 1942b)
based primarily on the angle of the anteronasal surface, and may be termed
hypsiprosopic (terminology from Fejervary, 1935) if the angle is gradual
from septomaxilla to frontals, platyprosopic if the surface is
perpendicular to the tooth-bearing surface, and mesoprosopic for an
intermediate grade.
3) Double lacrimal
foramen, seen elsewhere only in the monotypic Lanthanotus
(included in the Varanidae by Pregill et
al., 1986).
4) Well-developed
subolfactory processes of frontal in contact, or nearly so, effectively
dividing interorbital region in two (McDowell and Bogert, 1954).
5) Prearticular bone
reduced, not extending anteriorly much beyond the coronoid. Estes et
al. (1988) note that this condition is convergent in snakes.
6) Splenial and
dentary bones both move anteriorly as a function of the intramandibular
joint, providing limited jaw kinesis.
7) Nine cervical
vertebrae (seen also only in Lanthanotus).
8) Precondylar
constriction of vertebrae, giving the centrum a characteristic V-shape,
with a broad, rounded posterior condyle.
9) Caudal chevrons and
cervical hypapophyses attached only to vertebral centra (Pregill et
al., 1986), as opposed to attachment to centra and arches.
Living varanids are a
monophyletic group most closely related to Lanthanotus and Heloderma
(McDowell and Bogert, 1954; Pregill et
al., 1986; Rieppel, 1980). Together, these taxa are placed within the
Platynota (=Varanoidea), and are the sister group of the Diploglossa
(Anguidae, often including Anniellidae, and Xenosauridae and
Shinisauridae). The entire assembly is termed the Anguimorpha. Living
monitors are represented by 71 named species and subspecies, the status of
many currently under review. Though early authors (Boulenger, 1885;
Daudin, 1802; Dumeril and Bibron, 1836; Dumeril and Dumeril, 1851; Gray,
1831, 1838, 1845; Müller and Schlegel, 1845) consistently grouped
varanids together, it was not until Mertens (1942a, 1942b, 1942c, 1959,
1963) that a serious attempt was made to examine intrageneric
relationships. Mertens recognised that varanids are not as morphologically
conservative as many herpetologists claimed, and saw "natural"
groupings that he termed subgenera. It is not clear whether Mertens used
this category to sidestep the possible taxonomic confusion that erecting
several genera might produce, or to indicate that further resolution was
necessary before such a step might be valuable. In light of the numerous
subgeneric shufflings that have taken place since 1942 (Baverstock et
al., 1994; Böhme, 1988; King and King, 1975), Mertens' decision
served the interests of nomenclatural stability well at the alpha
(=species descriptions) level. Estes (1983a) extended Mertens' decision by
suggesting that either all varanids (including fossil taxa) be combined in
the genus Varanus, or all subgenera be elevated to generic rank. In the latter
case, it does not seem that taxonomic stability would be served by such a
subjective step. Though subgeneric designations are still widely used in
the literature, their status and membership remain controversial
(Sprackland, 1991b, 1992).
Mertens (1942a)
intended to follow Fejervary's (1935) taxonomic arrangement, based on
maxillary structure, but found that this feature varied within
closely-related taxa. Mertens (1942a) employed a variety of gross
morphological characters, primarily nasal bone structure, maxillary shape,
nostril position and shape, tail compression, and scutellation, to
establish eight subgenera (he would later erect two more, Philippinosaurus
and Papusaurus; the latter,
containing only V. salvadorii,
is relevant to this study). Two of these subgenera, Odatria and Varanus,
contained the overwhelming majority of species. Though the former subgenus
has held up under continued scrutiny (Baverstock et al., 1994; Böhme, 1988; King and King, 1975; King
et al., 1991; Holmes et al.,
1975; Horn et al., 1994; Sprackland, 1991a, 1991b), the diagnostic criteria
used by Mertens have been modified (e.g., not all taxa have paired nasals
or are under 1 m in length). A more detailed review of Mertens' work is in
the following section.
In contrast, the
subgenus Varanus, which Mertens
(1942a) diagnosed as having a laterally compressed tail, round, anterior
nostrils, and pointed, compressed teeth, has been demonstrated to be
polyphyletic and its taxa have been redistributed (Baverstock et
al., 1994; Böhme, 1988; King and King, 1975). Among the species
concerned is Varanus indicus,
which Mertens considered to include three subspecies (see below). He later
named Varanus karlschmidti
(Mertens, 1951), a junior synonym for V.
jobiensis (see below), and considered it closely related to V.
indicus (Mertens, 1971). The type species for Varanus
is V. varius, to which V. indicus is not particularly closely related (Böhme, 1988;
Baverstock et al., 1994; Holmes et
al., 1975). V. indicus was subsequently moved (Böhme, 1988) to Euprepiosaurus
Fitzinger (1843), for which V.
indicus is also the type species. Böhme (1988) included the New
Guinea tree monitors ( V. beccarii, V. bogerti, V. prasinus, V. telenesetes, and V.
teriae) in Euprepiosaurus. Sprackland (1991b) confirmed the close relationships
between V. indicus and the tree
monitors, but did not recommend any subgeneric assignment, pending further
study of related taxa, particularly the V. salvator complex and New Guinea endemic V. (Papusaurus) salvadorii. All these taxa, excepting V.
salvator and V. spinulosus, have centres of distribution in New Guinea, with only
V. "indicus" and V.
teriae having extralimital distributions. V.
salvator is the only taxon under consideration that is restricted to a
range west of Wallace's line (though unconfirmed reports of
V. salvator from
Halmahera, if true, would marginally extend that range). Varanus
salvadorii has been rare in collections during most of the cited
studies of varanid relationships, so its karyotype, molecular characters,
and hemipenial morphology have not been compared with hypothesized
relatives. Recent availability, however, through the exotic animal trade
has presented many specimens for morphological and behavioural
examinations (to be discussed below).
Mertens
(1942a-c) assigned living varanids to eight subgenera (later expanded to
ten), all but three of which, Varanus,
Empagusia, and Odatria, were
monotypic. The other subgenera included Polydaedalus,
Indovaranus, Psammosaurus, Empagusia, Dendrovaranus, and Tectovaranus.
Mertens (1959) would later erect Philippinosaurus
for V. grayi (=V. olivaceus), and Papusaurus
(Mertens, 1960) for V. salvadorii
(Tab. 1). At that time, he recognized 49 taxa (including subspecies), and
relationships were based upon few characters, in particular 1) sectional
shape of the tail, 2) parietal fenestration, 3) nasal bone condition, 4)
height and length of the skull, and 5) position of the nostrils.
Unfortunately for contemporary researchers, Mertens diagnosed his taxa on
the basis of many symplesiomorphous characters and virtually no
synapomorphies (Tab. 1).
Three of Mertens'
subgenera are relevant to the present study, and were erected almost
entirely on external morphological characters. Varanus has a laterally compressed tail, without annuli of
equal-sized scales; round nostrils located nearer the snout tip than the
eye; and fused nasals. Odatria
has a round tail, with caudal scales forming annuli; a round nostril; and
paired nasals. Mertens (1942c) claimed that V.
salvadorii had a compressed tail, though it is actually subtriangular
to round, as he noted when erecting the subgenus Papusaurus (Mertens, 1962). Though Mertens (1962:333) claimed he did
not generally condone monotypic subgenera, he nevertheless did so when he
made Papusaurus his seventh such
taxonomic group. Papusaurus was
diagnosed primarily on its peculiar tail shape (convergent in V.
griseus, V. spinulosus, and some Odatria);
its extraordinary tail length of 2.3-2.6 times snout-vent length (=SVL)
(convergent in Varanus (Odatria) glebopalma); and vaulted snout (convergent in V.
doreanus). While the preceding characters may in fact represent unique
evolutionary novelties (=autapomorphies) within the Varanidae, the other
diagnostic characters are certainly symplesiomorphies, and include: very
large size (of V. komodoensis, V.
salvator, V. niloticus, and V.
giganteus); nostril lateral and oval
---------------------------------------------------------------------------------------------------------------------
TABLE
1. Subgenera of Varanus and
their contents based on Mertens, 1942a, 1942c.
Subgenus
Varanus
giganteus
gouldii
indicus
indicus
kalabeck *
spinulosus **
komodoensis
salvadorii +
salvator
salvator
cumingi
marmoratus
nuchalis
scutigerulus ++
togianus
varius
Subgenus Odatria
acanthurus
acanthurus
brachyurus
primordius **
brevicauda
caudolineatus
eremius
gilleni
prasinus
prasinus +
beccarii ** +
kordensis *
semiremex
semiremex
boulengeri *
timorensis
timorensis
orientalis +++
scalaris **
tristis **
Subgenus Polydaedalus
niloticus
niloticus
ornatus
Subgenus
Indovaranus
bengalensis
bengalensis
nebulosus
Subgenus
Psammosaurus
griseus
Subgenus Empagusia
flavescens
exanthematicus
exanthematicus +
albigularis **
angolensis ***
microstictus ***
Subgenus Dendrovaranus
rudicollis
Subgenus Tectovaranus
dumerilii
dumerilii
heteropholis *
Subgenus Incertum
(later Philippinosaurus)
grayi
(=olivaceus)
*=junior
synonym, no longer in use. **=subsequently elevated to species status.
***=currently considered a subspecies of albigularis.
+=later placed in new subgenus. ++=invalid species, named in error.
+++=currently considered a subspecies of tristis.
------------------------------------------------------------------------------------------------------------------
(cf. most varanids
excluding V. griseus, V.
albigularis, V. bengalensis, and several related species); eyes
lidded, with round pupils; supraoculars undifferentiated; caudal scales
not subequal and forming whorls (vs. V.
griseus and subgenus Odatria);
hypsiprosopic maxillae (cf. most varanids); nasals paired or unpaired
(covering both character states for platynotans); external ear opening
present; and tooth structure subconical, with basal fluting
(=plicidentine). The undifferentiated supraoculars are interesting because
they clearly distinguish V.
salvadorii from all other large (2m +) varanids (excepting the
morphologically and ecologically distinct V.
giganteus), and all sympatric New Guinea varanids except the dwarf
members of the V. (Odatria) timorensis group (including V. tristis and V. similis).
Only the tooth structure is unique to V.
salvadorii, being long, triangular, and with only the anterior teeth
being slightly recurved (see below).
Mertens' works remain classics, but require updating because of
several deficiencies brought about by our rapidly increasing knowledge of
these lizards, and of new methods available to examine herpetological
specimens. Some of these deficiencies include:
1) Mertens's work was conducted in Germany during World War II,
giving him limited access to specimens and literature from other
countries. While comparing his work with actual museum specimens during
1991, I was frequently aware that some specimens, though cited by Mertens,
could not have been examined by him (e.g., it is doubtful he would really
have confused Varanus rudicollis for V.
dumerilii, or V. varius for V.
salvator), and his inclusion of such specimen numbers in his
publications may reflect examination of museum registers and earlier
literature, rather than specimens. Mertens noted the extreme variability
within "Varanus indicus"
and claimed he could not tell the locality from which specimens came
without recourse to labels (Mertens, 1942c:262). Schmidt (1932) was unable
to use Mertens' (1926) review to identify to subspecies the specimens of Varanus
indicus collected from the Solomon Islands. In 1991, using the actual
specimens, I could identify specimens to locality with greater than 90%
accuracy, leading me to believe that Schmidt was limited in the number of
specimens examined, filling in details from published accounts. Mertens
later (1958, 1959, 1962, 1963, 1971) appended his three-volume monograph,
but never incorporated the use of diagnostic apomorphies, nor did he
clarify relationships among cryptic taxa.
Compounding the
confusion is the fact that after World War II many researchers assumed
that the major herpetological collections at the Zoological Museum,
Humboldt University, Berlin were lost or destroyed (Good et al., 1993), so that few researchers consulted the considerable
and important monitor holdings of that institute until the reunification
of Germany in 1989 made access more convenient for western scientists.
2) Though Mertens's
literature review is extensive for the time, it is deficient in many
places, a fact he acknowledged (Mertens, 1942a). Additionally, cited works
are not always listed in the bibliography, or references are incomplete
(e.g., Kubary, 1872). The fact that Britain and Germany were at war
prevented Mertens from examining the specimens in the British Museum
(Natural History), so his information was restricted to published data
that did not always apply to the species, or specimens, in question. Even
when he subsequently examined Australian and British museum collections,
Mertens (1958, 1959) often did not have types available (see below for
status of types), and his examinations were hurried, resulting in further
errors and few clarifications. It is also common to find literature
accounts that represent two or more species (see Sprackland, 1991b). To
overcome this problem, none of the data in this project is based solely on
published records. Every specimen cited in this study was personally
examined by me.
3) Mertens's taxonomic
scheme is based on relatively few, predominantly plesiomorphic
(=ancestral) or ecotypic (=environmentally vs. genetically influenced)
characters. Features that may be interpreted as convergent (e.g.,
compressed or round tails) are used as both primary diagnostics and as
derived characters for assigning relationships. These characters are
useful for discriminating among and diagnosing species groups
("subgenera" of Mertens) and genera, but often fail to separate
similar species or record intraspecific variation. In particular, Mertens
failed to associate nostril position and shape, important morphological
characters, with ontogeny (Auffenberg, 1994). Additionally, though Mertens
did a major taxonomic review culminating in combining species into
subgenera, he proposed no phylogeny for relationships among those
subgenera, leaving varanid systematics unaddressed.
4) Mertens was limited
in his use of living material (about 150 specimens, Mertens, 1942a),
causing him to omit features such as colour, colour pattern, male-male
combat behaviour, and other agonistic behaviours. His work far preceded
the era of captive breeding, making the examination of ontogenetic changes
that can be done today impossible in the early 1940s. In addition, species
that have only recently become frequent imports, including Varanus
indicus and its allies, were rarely, if ever, to be seen in captivity
in the 1930s and 1940s. Of those species represented at the time in zoos,
most were single specimens in "postage-stamp" collections, and
of limited taxonomic usefulness. While extensive preserved samples showing
ontogenetic change is possible, such series are still not available in
collections for most varanid taxa. In at least one case, Mertens proposed
a subspecies of mangrove monitor (Varanus
indicus rouxi, Mertens, 1926) based on juvenile colouration. As will
be demonstrated below, juvenile colour and pattern are the most variable
intraspecific elements of mangrove monitor morphology. Additionally, even
with such series, much basic taxonomic research must still be considered
limited without access to live samples and, as the present study
demonstrates, omission of colour data can delay detection of cryptic taxa.
5) For many taxa,
there were few (N = 1-3) specimens in museum collections. In such cases,
Mertens had only literature or single specimens from which to draw his
conclusions. The situation is hardly better today for V. spinulosus (I had access to all six known specimens, three as
photographic material; Mertens had only the holotype), but far more
specimens are available for the several taxa and numerous populations of
mangrove monitors. Where Böhme et
al. (1994) had 105 specimens, and Mertens (1942c) had 177 specimens,
this study employs nearly 500 (Table 2).
Mertens examined about 1,000 lizards for his entire monograph,
while the present study employed 470 specimens from the V.
indicus group alone.
6) Mertens reviewed
the scant literature on fossil varanids, but did little to incorporate
fossil data into a larger scheme of varanid phylogeny. Much of his
terminology was taken from the Hungarian palaeontologist Fejervary (1935),
whose palaeontological work is in grammatically convoluted English, thus
confusing many points as they emerge from Hungarian to "English"
to German. Errors are more likely to occur as information goes through
multiple translations. Further compounding this problem is Fejervary's
limited use of comparative material from extant varanids, making many of
his conclusions erroneous or ambiguous. For example, as noted by Mertens
(1942a), Fejervary (1935) placed Varanus
komodoensis into a monotypic genus Placovaranus
because he believed the osteoderms of
komodoensis were unique
among "varanians". The net result for Mertens was that he
accepted Fejervary’s terminology while simultaneously dismissing most of
Fejervary's observations on extant varanids, and further omitted much of
the fairly considerable fossil data, further isolating his overview of
taxonomic relationships. This omission may have serious implications for
future varanid research, for, as noted by Estes (1983a), the differences
between living and extinct varanids are either so minor as to warrant
either including all post-Mesozoic taxa in a single genus Varanus,
or
