вход по аккаунту


Cladistics and the hominid fossil record.

код для вставкиСкачать
Cladistics and the Hominid Fossil Record
Department of Anthropology, University of New Mexico, Albuquerque,
New Mexico 87131 and U.A. 376 du C.N.R.S., Luborutoire
d’Anthropologie, Universite de Bordeaux I, 33405 Talence, France
Human paleontology, Phylogenetic reconstruction,
Cladistic methodology has become common in phylogenetic
analyses of the hominid fossil record. Even though it has correctly placed
emphasis on morphology for the primary determination of affinities between
groups and on explicit statements regarding traits and methods employed in
making phylogenetic assessments, cladistics nonetheless has limitations
when applied to the hominid fossil record. These include 1)the uncritical
assumption of parsimony, 2) uncertainties in the identification of homoplasies,
3) difficulties in the appropriate delimitation of samples for analysis, 4)failure
to account for normal patterns of variation, 5) methodological problems with
the appropriate identification of morphological traits involving issues of
biological relevance, intercorrelation, primary versus secondary characters,
and the use of continuous variables, 6) issues of polarity identification, and 7 )
problems in hypothesis testing. While cladistics has focused attention on
alternative phylogenetic reconstructions in hominid paleontology and on
explicit statements regarding their morphological and methodological underpinnings, its biological limitations are too abundant for it to be more than a
heuristic device for the preliminary ordering of complex human paleontological and neontological data.
The past decade has seen a roliferation in
the application of the princip es, techniques,
and terminolo
of cladistics in human
paleontology.’ %is has progressed to the
point that it is fre uently considered de
rigueur to include eit er a cladistic analysis
or explicit reference to the concepts of cladistics in considerations of the hominid fossil
record. Yet, even though the application of
cladistics in general has led to more explicit
statements concerning the traits and methodologies employed to arrive at a phylogeny
for the Hominidae (or subgroups thereof),
there are nonetheless implicit assumptions
and methodologies invoked in most current
‘The term “cladistics” is used here to refer generally to the
methodologies and assumptions of this form of phylogenetic
analysis, following Mayr (1974). As recently pointed out by
Schmitt (1989), this includes more general and varied procedures
than were apparently intended by Hennig (1966) by his term
“phylogeneticsystematics.” This is considered appropriate here,
since the phylogenetic analyses in hominid paleontology using
these concepts are varied and rarely adhere precisely to the tenets
as set out by Hennig and his strict followers.
0 1990 WILEY-I,ISS, INC.
cladistic analyses. These assumptions and
methodologies, if found to be ina propriate,
could place limitations on the p ylogenetic
interpretations derived from such cladistic
This presentation is a consideration of
some of-these assumptions and methodoloes, as currently applied to the hominid
ossil record. It is not intended to be a review
of either the basic tenents of cladistics or of
the (usually implicit and always important)
philosophical underpinnings of the approach. Nor is it intended to be a comprehensive review of all of the to-date cladistic
analyses of portions of the hominid fossil
record; such a review would be redundant
and serve little urpose at resent. Nor is it
intended to reso ve any oft e currently contested phylogenetic issues in hominid evolu-
Received June 12,1989; revision accepted November 30,1989.
Address reprint requests to Erik Trinkaus, Dept. ofhthropology, Univ. of New Mexico, Albuquerque, NM 87131, USA.
tion; the resolution of such ph logenetic issues, or at least the reaching o a reasonable
consensus concerning them, is probably remote given both the fragmentary and scattered nature of the hominid fossil record and
the limitations of our current approaches to
that fossil record.’
It is assumed here that the ultimate goal of
paleoanthropology is to understand the past
events and processes that led to the evolutionary emergence of the Hominidae, the
genus Homo and, eventually, modern humans from more archaic members of the
Hominoidea, Hominidae, and the genus
Homo, respectively. It is fully recognized
that, even if one’s primary goal is to understand the behavioral (or adaptive) components of the evolving hominid lineage(s), one
must have a reasonable phylo eny upon
which to hang functional morpho ogical and
adaptive trajectories through time.
In the context of this, it is apparent that
our current paleoanthropological record is
incomplete and that our interpretations of
liominid evolution will change as rurtlrier
discoveries are made and as additional research is carried out on the existing fossil
record. Furthermore, the information available at one time to researchers is an incomplete art of the otentially available data
erefore, a 1 interpretations can be
seen realistically as no more than stateof-the-art conclusions, which will inevitably
fail and be modified as the field progresses.
The relevant samples
For the purposes of this discussion, the
Hominidae will be considered as consisting
of the two genera (Australopithecus and
Homo) and seven species (A. afarensis, A.
africanus, A. robustus, A. boisei, H . habilis,
H. erectus, and H . sapiens) currently generally accepted within the Hominidae.
2No claim is being made here for originality. Many of the issues
raised are well expressed in the abundant general literature on
phylogenetic systematics, even if they are not completely referenced in this presentation (see especially Hennig, 1966; Arnold,
1981). It is primarily the failure of many human paleontologists
to take them into account that has inspired this discussion of
cladistics and its current application to the hominid fossil record.
Of the issues raised here, it is mainly the issue of homoplasy that
has received attention in the human paleontological literature
(e.g., Kimbel et al., 1986; Dean, 1988; Wood, 19881, along with
some recent attention to the issue of trait definition and significance (e.g.,Turner and Chamberlain, 1989).
It is nonetheless recognized that further
taxonomic splitting may be warranted,
the patterns of variation evident in the ominid fossil record. For example, it may be
appropriate to separate “anatomically modern” from archaic H . sapiens at the species
level (resurrecting “H.neanderthalensis” for
“archaic H . sapiens”) (Tattersall, 1986;
Trinkaus, 1987; Stringer and Andrews,
19881, H. habilis as currently described includes at least two and probably three separate species (some of which may more appropriately be placed within Australopithecus)
(Walker, 1981; Stringer, 1986; Trinkaus,
1987; Lieberman et al., 1988), some of the
geologically older specimens included within
A. boisei could be separated and placed into
their own species (“A.aethiopicus”) (Walker
et al., 1986; Walker and Leakey, 1988), and
some paleontologists feel that the “robust”
members of Australopithecus should be
placed in their own genus (resurrecting
“Paranthropus”) (Olson, 1978; Wood and
Chamberlain, 1987; Grine, 1988). Furthermore, it must be kept in mind that all post-H.
habilis members of the genus Homo were
geogra hically dispersed and that, riot surprising y, it is possible to recognize geographical variants of H. erectus and “archaic
H , sapiens”, as well as of modern H . sapiens.
Despite these caveats, the discussion here
will use the two genera and seven species
generally recognized, since the purpose is no
more to resolve taxonomic issues than it is to
sort out hominid phylogeny.
Cladistic analyses of the hominid fossil
In the application of cladistics to the hominid fossil record, there have been on1 a few
analyses that treat the whole of the yknown
Hominidae (e.g., Eldredge and Tattersall,
1975; Delson et al., 1977; Bonde, 1977,1981;
Olson, 1978). The majority have been concerned with African Pliocene and initial
Pleistocene (pre 1.0 ma B.P.) hominids (e.g.,
Delson, 1978; Johanson and White, 1979;
Olson, 1981; Skelton et al., 1986; Dean,
1986; Wood and Chamberlain, 1986, 1987;
Chamberlain and Wood, 1987; Kimbel et al.,
1988; Wood, 1988; Tobias, 1988). A few cladistic studies have been primarily concerned
withH. erectus (sensu lato)(e.g., Santa Luca,
1980; Andrews, 1984; Stringer, 1984; Wood,
1984; Bilsborough and Wood, 1986; Hublin,
1986; Turner and Chamberlain, 19891, and
recently later Pleistocene (post 300 ka B.P.)
Scenarios, while remaining important as
syntheses of our knowledge, interpretive
frameworks, and sources of hy otheses for
further analysis of the paleoant ropological
record, have played a reduced role in actual
phylogeny building.
In addition, cladistics has emphasized the
need to determine the polarities of morphological patterns and to distinguish between
ancestral (plesiomorphous) and derived
(apomorphous) traits and between shared
[symplesiomorphous (shared ancestral) and
synapomorphous (shared derived)] and
unique Lautapomorphous (uniquely derived)] traits. In this way, the relative phylogenetic utilities of various identified traits
are clearly indicated, with synapomorphous
traits having greater relevance for indicating shared ancestry than plesiomorphous
ones. In fact, in cladistics it is exclusively
synapomorphies that are capable of indicating phylogenetic closeness between groups.
As a result, once the polarities of traits at any
given branching point have been determined, the traits can be differentially included in the resulting analysis. Simple trait
lists for various groups, although of descriptive utility, are less revealing of affinities
than was once thou ht.
Practitioners of c adistics in human paleontology have frequently assumed that morphological similarities due to factors other
than common descent [homoplasies, which
include convergences, parallelisms and reThe introduction and spread of cladistics versals (Simpson, 196111 are basically noise
in human paleontology during the past in the system and should be eliminated to
decade has served primarily to make phylo- the extent possible in the final cladogram.
genetic arguments more explicit. Whereas Therefore, the clado ram with the smallest
phylogenetic reconstruction was previously number of required omoplasies is the prefrequently based on a priori views of the ferred one, and the principle of arsimony
course of hominid evolution, with strong his- is usually invoked to arrive at t e desired
torical and subjective com onents implicit in clado ram. In addition, paleontologicalunits
the assessments of mor ology and the re- (popu ations or species) with significant
sultant phylogenies, cla istics has forced hu- numbers of autapomorphous traits are conman paleontologists to be clear about their sidered to have been more likely to have
methodologies and to reassess their assump- become extinct without issue than to have
tions concerning possible (or probable) phy- experienced evolutionary reversals of previlogenetic arrangements of scattered hominid ous morphological trends.
It should be emphasized that little of the
It has therefore become increasingly rou- core of cladistics is really new. Issues of
tine for human paleontolo ists to recon- polarity, parsimony, homoplasy, the priority
struct branching patterns (c adograms) and of morphology in phylogenetic reconstrucresultant phylogenies (with the addition of tion, etc. have long been recognized in palegeological time and paleogeo aphy) start- ontology. It is primarily the weaving of these
ing with the similarities and ifferences be- principles into a coherent analytic method
tween groups in their neontologically and that has been the major contribution of clapaleontologically documented morphologies. distics in the past decade.
archaic and modern H. sapiens have come
under consideration from this perspective
(e.g., Condemi, 1988b,c; Hublin, 1988a;
Stringer and Andrews, 1988; Tillier, 1988;
Trinkaus, 1988). An additional analysis
(Stringer, 1987) has analyzed the genus
Homo as a whole.
As a result, the rinciples and methodology of cladistics, w en ripplied to the hominid fossil record, have been employed largely
at the intrageneric level and occasionally at
the intraspecific level. This should be kept in
mind when considering the possible limitations of the methodology as commonly applied to the hominid fossil record, since the
units of analysis of concern, being closely
related, are likely to have behaved more as
o ulations than as discrete, genetically
FuKy isolated species.
Furthermore, a minority of these studies
have rigorously and explicitly applied the
principles of cladistics, with ex licit definitions of the sam les and morpho o ical traits
employed and t e use of clear met odologies
to generate cladograms and, by extension,
phylogenetic arrangements of paleontological and neontological samples and assignments of polarity to traits (see below). More
often, the authors of these articles have employed select cladistic concepts (or terminology) to argue for specific phylogenetic interpretations of the samples in question.
Despite these advantages, the universal
and uncritical ap lication of cladistics has
its limitations an pitfalls. Much as the uncritical application of multivariate distance
studies during the 1970s(facilitated by package multivariate statistical programs and
the availability of mainframe computers)
generated considerable work with uninterpretable or trivial results, the widespread
and casual application of cladistics to the
human paleontological and neontological
records has considerable otential to generate work of minimal utilty to our understanding of hominid evolution. This is especially true since cladistics, like multivariate
analysis, is only a meth~dology.~
A question of parsimony
Virtually all current cladistic analyses emloy the principle of parsimony, usually to
Emit the number of character changes neces-
often parsimony is taken as a given of the
real world, rather than as a heuristic device
for the ordering of data.
Parsimony is no more than an optimization model, one which assumes that the simplest solution has the greatest robability of
accurately representing the rea world. It is a
set of algorithms and not an aspect of biology
(Felsenstein, 1983). Like all optimization
models, it is a mathematically derived theoretical ideal. It is of utility for the reliminary ordering of data, for the formu ation of
hypotheses, and for the identification of deviations in the real world from the ideal
world of the model. It can never be more than
an approximation of reality.
The identification of homoplasies
The issue of parsimony relates directly to
the identification of homoplasies, since the
solution with the fewest character state
changes is usually the one with the fewest
homoplasies. Yet, can we always be secure in
q h i s parallel with the multivariate work of the 1970s is
reinforced by the current availability of several phylogenetic
(“cladistic”) analysis software programs for microcomputers
(Fink, 19861, of which PAUP [Phy ogenetic Analysis Using Parsimony (Swofford, 198511has become the most commonly used in
human paleontology.
our identifications of homoplasies, and is
that a meaningful exercise?
Some cladistic anal ses using parsimony
[such as PAUP (Swo ord, 198511 may identify possible homoplasies mathematically,
and using comparative developmental biolo ‘cal data we may be able to resolve
w ether some of them are indeed homoplasies. Yet, most appropriate developmental
comparisons are possible only for neontolo ical material, and they are rarely availab e
through the paleontological record (patterns
of Neandertal and Australopithecus facial
and cranial base growth being one of the rare
exceptions (Broma e, 1986; Dean, 1988;
Tillier, 1988;Minug -Purvis, 1988)l.Consequently, most “identifications” of homoplasies paleontologically are the products of
mathematical com utations, not of biological evaluations oft e traits in question.
Furthermore, in closely related biological
forms, such as at least the members of Australopithecus, the various groups included
within H. habilis, and the geographical and
temporal variants of H. erectus and H. supiens, all of the populations share a substantial biological baseline upon which evolutionary processes operated. Therefore,
similar evolutionary pressures on allopatric
populations (likely in such biologically, and
by extension behaviorally, similar populations) are likely to produce similar morphologies independently. Even if A. boisei and A.
robustus are considered to have been separately derived from A. ufurensis and A. africanus respectively, would homoplasies be
unexpected among them given the morphological similarities between A. ufurensis and
A. ufricunus? Despite the geographical distance between sub-Saharan African and
eastern Asian H. erectus, is it surprising that
H. erectus samples in the two regions would
show parallel evolutionary trends given
their similar basic adaptive patterns and
common ancestry in early east African H.
Given this, can we hope to sort out consistently such expected homoplasies from symplesiomor hies or synapomorphies? If we
are not ab e to distinguish homoplasies consistently from symplesiomorphies or synapomorphies among closely related groups,
which is likely given that the methodology
operates on the adult phenotype (the paleontologically observed morpholo ) to reconstruct phylo enetic history an that the ancestral basefine is frequently unknown or
oorly known, then to what extent are we
Eiasing both our phylogenetic trees and our
interpretations of the evolutionary processes
In addition, a t the intrageneric or especially intraspecific level, many of the identified apomorphous traits are of unknown or
apparently minor functional significance.
Given the high probability of evolutionary
reversals at the genetic (nucleotide) level
and the robably low adaptive valence of
many oft e morphological traits considered,
is an apparent trait reversal likely t o be all
that unusual?
Units of analysis
Cladistic analysis requires the explicit definition of units of analysis, groups of organisms to be compared and arranged into cladograms and eventually into phylogenies.
These groups, samples or OTUs (operational
taxonomic units) are relatively easy to delimit in the modern world, since they are
usually represented by biolo ically defined
populations, species, or hig er taxonomic
units. Paleontologically, the exercise becomes more complex.
Traditionally and currently, some combination of morphology and temporal (stratigraphic) and s atial (re ’onal) criteria are
employed to efine pa eontological taxonomic groups. Yet, time and s ace are not
biological criteria for distinguis ing groups,
and using morphology alone to distin
groups can become tautological, preju gmg
the outcome of the subsequent phylogenetic
This is particularly relevant to cladistic
analvsis within maior subsets of hominids,
intrigenerically ana especially intraspecifi:
cally. Whether one uses the seven hominid
species listed above, or splits them further
into eight, nine, or more species, it is evident
that there are several groups of hominids
that are closely related and are temporally
and/or geographically adjacent. This applies
to the species of Australopithecus, various
groups usually included within H. habilis,
geographical and temporal variants of H.
erectus, and members of archaic and modern
H. sapiens.
Therefore, temporal and geographic clines
should have been common throu hout hominid evolution. In fact, wherever t e hominid
fossil record is even minimally adequate, as
among “archaic H. sapiens” in the Upper
Pleistocene of western Asia (Trinkaus, 1984,
in press), across Europe and the Near East
among “archaic H. sapiens” in the earlier
Upper Pleistocene (Condemi, 1988a; Trinkaus, 1983, in ress), in the late Middle and
early Upper P eistocene of Europe (among
“archaic H. sapiens”) (Wolpoff, 1980;
Stringer, 1985; Hublin, 1988b), through the
late Lower and Middle Pleistocene of Indonesia and China (among H. erectus) (Santa
Luca, 1980;Wu and Dong, 1985; Wu and Wu,
19851, and in the Pliocene of Africa (among
A. afarensis and A. africanus)(Tobias, 1980;
White et al., 19811,there is evidence of temporal and/or geogra hical clines across the
available samples. onsequently, the delineation of the units of analysis becomes
largely arbitrary, determined by the relatively abundant temporal and geographic
gaps in the fossil record and by a priori
mor hological decisions as to where the
brea s should occur.
The role of variation
All current modern syntheses of biological
evolution are in agreement that the existence of normal ranges of genot pic and phenotypic variation within popu ations is the
baseline of biological evolution and that the
conversion of within-population variation to
between-population variation is the core of
the evolutionary process. Whatever directional or random processes are primary in
producing a given evolutionary change or
apparent stasis, that observed pattern is the
product of the presencelabsence of shifts in
allelic (and by extension phenotypic)
fre. quencies.
These elementary Neo-Darwinian evolutionary principles bkar repeatin since most
cladistic analyses assume a sing e character
state per variable per analytical unit. Yet,
any casual observation of reasonable samples from recent populations illustrates that
there is considerable variation in many observed features. This is frequently evident
even in small paleontological samples, in
which the limited number of specimens
would tend to reduce the number of character states represented, given the high dependence of observed diversity on sample size
(Fisher et al., 1943; Pielou, 1975).
This roblem is exacerbated in paleontology by ifficulties in assigning fragmentary
s ecimens to specific or intras ecific samp es. In many cases, individua specimens
lacking highly diagnostic morphological regions have been routinely assigned to the one
species/subspecies clearly present at a site or
even to the most commonly preserved one in
the deposit in question. Yet, as the otential
itfalls of this technique are realize ,such as
for Pliocene and Lower Pleistocene deposits
in which more than one s ecies, or even
genus, may be represente or for u\per
Pleistocene deposits in which both arc aic
and “anatomically modern” H. sapiens are
present, increasingly only the more complete, “diagnostic,” specimens are being considered in ph logenetic assessments. The
result is to re uce sample sizes further and
thus narrow the perceived ranges of variation of the biological groups involved. The
ultimate result of any attempt to be absolute1 certain of the correct assignment of
fossi specimens to species or subspecies
samples would be the denial of any meaningful level of variation.
Clearly, an phylogenetic assessment of
closely relate groups must be able to assess
both the central tendencies of individual
traits within samples as well as their patterns of variation. Those central tendencies
may be significantly different between samples in a probabilistic sense, yet their ranges
of variation may overlap considerably.
Merely using a nominal characterization of
their central tendencies would deny the variations and range of overlap, whereas full
considerations of the ranges of variation
might disguise significant frequency shifts
between the samples. Consequently, both
the central tendencies and the variances of
the samples are relevant and need to be
taken into consideration.
The identification of traits
The use of cladistics, while it has made
paleontologists be more explicit about the
morphological characters used, has raised
several questions about trait definition and
First, to what extent do the traits employed need to have any biolo ical relevance,
as opposed to being merely requently preserved and occurring in easily distinguishable character states? If the distinctions being drawn are at the species or a higher
taxonomic level, then the most appropriate
traits are likely to be ones that have, or can
be presumed to have, some functional relevance (Turner and Chamberlain, 1989). At
these taxonomic levels, some degree of behavioral contrast is usually necessary for the
maintenance of species separation among
synchronic/sympatric species and some long
term adaptive shift is usually involved in our
perception of s ecies distinctions between
ancestral and escendant groups. One can
argue, however, that for intraspecific comparisons selectivelyneutral traits are preferable, since they are more likely to reflect the
minor, stochastic evolutionary processes
(e.g., genetic drift) that usually lead to distinctions between regional populations of a
species. Yet, in order to use such selectively
neutral traits effectively, one must employ
comparisons of the distributions of those
traits between samples, as is routinely done
in the analysis of metric and discrete traits
between recent human skeletal samples.
Since sufficiently large samples are rarely, if
ever, preserved in the hominid fossil record
for such analyses, the appropriate use of
these (presumably) selectively neutral traits
is difficult.
Second, to what extent might different
traits be intercorrelated? It is possible to
assess the degree of statistical correlation
between variables and their character states
(given adequate sample sizes), but it is seldom known whether such correlations are
spurious or real, size- or shape-dependent,
sample-specific,or of broader biological relevance. If they are biologically meaningful,
are they due to pleiotropy and/or functional
interrelationships? If the intercorrelations
are determined from the samples in question, how does one correct for them without
becoming tautological?
Third, to what extent are we identifying
the primary biological features or merely
some convenient secondary reflections of
them? In complex functional units, such as
the facial skeleton or the locomotor anatomy,
alteration in one feature will have consequences (both spatial and biomechanical) for
other aspects of the regional anatomy. Our
ability to identify the primary biological elements, as opposed t o secondary conseuences of their interrelationships, will in?hence the number and polarities of the
traits perceived.
Fourth, to what extent do the traits employed exist in discrete character states, as
opposed to being continuous variables? Since
most cladistic analyses require that there be
a single character state per variable per
group, the common practice is to define a
series of states based on one’s previous
knowledge of the paleontologxal record. Yet,
most variables are designed to describe re-
gional anatomical shape, and virtually all
aspects of morphology (especially skeletal
and dental morphology) exist as continuous
ranges of variation.
Some of these ranges of variation (e.g.,
dental occlusal morphology, occipitomastoid
configurations, and carpal and tarsal articular morphology) seem to fall into discrete
forms and therefore are more easily quantified as a series of character states rather
than as continuous metric variables. Yet,
large Sam les of recent skeletal remains invariably s ow continuous ranges of variation from absent, to trace, to present, to
pronounced for the feature in question. The
presence/absence o f teeth (dental agenesis)
is one of the few skeletal features that truly
exists as a discrete character, and even it is
the product of a developmental threshold
effect acting on a continuous range of size
variation (Gruneber ,1965).
Other aspects o morphology are frequentl quantified only as continuous metric
variab es (e.g., endocranial capacity, craniofacial shape, and postcranial robusticity and
roportions). In order to utilize them in cla$istic ana$ses, they must be converted to
discrete c aracter states. Therefore, researchers have either established arbitrary
metric boundaries, with character states being defined as less than or greater than a
specific value (e.g., Stringer, 1984) or comuted size-independent “codes”(e.g., Chamgerlain and Wood, 1987).Yet, how are those
boundary values and (‘codes” determined?
They are the roducts of our current knowled e of the ossil record and our a priori
definitions of paleontological samples. The
boundary values just happen to separate
(completely or largely) samples that we already “know” are distinct groups, and
‘(codes”are computed on the basis of which
existing fossil s ecimens are included within
the analysis. Tphe amount of within-group
variation obscured and the number of intermediate forms denied by these methods are
And last, to what extent can these limitations in trait definition be overridden by
simply identifying the maximum number of
traits possible on the fossils available (e.g.,
Andrews, 1984; Skelton et al., 1986; Chamberlain and Wood, 1987; Kimbel et al., 1988;
Tobias, 1988)?This approach might compensate for potential statistical biases in the
traits chosen for analysis, but it presumes
that we know little about the underlying
biologies of the traits in question. Yet, since
it is likely that many of the traits in question
are intercorrelated functionally and/or are
better seen as secondary reflections of more
primary components of morphology (see
above; also see Trinkaus, 1988; Hublin,
1989), such an approach can easily provide
erroneous results. Given appropriately biased selections of traits from large enough
001s of potential traits, it is probably possile to derive just about any cladogram for the
groups of concern, especially at the intrafamilial and frequently intrageneric levels involved in analyses of the hominid fossil
The issue of polarity
One of the major contributions of cladistics
has been to refocus attention on the hylogenetic polarities of traits and the dif erential
roles of derived versus ancestral morphologies (assuming one can identify traits appropriately) in phylogeneticreconstruction. Yet,
there are difficulties with determining the
polarities of traits.
There are two primary techniques for assessing polarity (e.g., Bonde, 1977; see also
Wiley, 1981). The first involves two closely
related techniques, which are based on prior
assessments of phylogenetic relationships
and the use of an outgroup. The second uses
ontogenetic data to sort out relative polarities.
In the first of the phylogenetic techniques,
the character states for neontological and/or
paleontological groups are analyzed, using
principles of parsimony, and the hypothetical ancestral forms, as outgroups, are reconstructed. This method contains the limitations of extreme assum tions of parsimony,
and it only identifies “typothetical” ancestors, not necessarily morpholo ’cal forms
that ever existed. The second p ylogenetic
method uses an outgroup (usually, but not
necessarily, a neontological one) to provide
information on the ancestral form. However,
if that outgroup is a neontological group
(such as the extant African apes for the
Hominidae), the degree of evolution within
the outgroup’s lineage since the common
ancestor must be assessed and the common
ancestor hypothetically reconstructed. If the
outgroup is a aleontological sample, then
assumptions a out the ancestor-descendant
relationships of the paleontological groups
must be made. However, since the ultimate
justification for determining polarity is to
provide more accurate phylogenetic reconstructions, this process can easily become
The second technique for polarity determination uses a variant of Haeckel’s “biogenetic law” and assumes that earlier ontogenetic sta es are usually representative of the
ancestra (plesiomorphous) condition. Although this technique may have some utility
for high taxonomic level comparisons [despite the numerous roblems noted over the
past century with t e “bio enetic law” (see
Churchill, 1980)],its usefu ness for intrafamilial, not to speak of intrageneric, assessments of polarity is questionable. Certainly,
ontogenetic data, both paleontological and
neontological, can be very useful in the appropriate definition of traits (see above), but
its value in indicating polarities within the
Hominidae must remain uncertain.
In addition to these considerations, the
wa s in which traits are defined can eatly
inf uence their perceived polarity. A 1 morphologically identified groups, whether neontological or paleontological, are unique in
some features. Otherwise we would be unlikely to identify them as discrete groups.
Their uni ueness can be due to their possession of tru y unique (autapomorphous) morphological patterns. However, among closely
related forms the number of truly unique
morphological patterns is likely to be small.
More robably, their unique “total morphologicafpatterns” will be produced by unique
combinations of traits, each of which is
shared with one or more closely related
groups, combined with truly unique but secondary morphological elements. Therefore,
whether a group is identified as possessing
a large number of uniquely derived traits
(autapomorphies) or some combination of
shared (symplesiomor hous and synapomorphous) traits will epend upon how the
traits are defined and to what extent they
are subdivided into biologically relevant
units. All of the issues of trait definition
discussed above apply here.
Cladograms and hypothesis testing
If one manages to deal constructively with
the various pitfalls outlined above concerning issues such as sample and trait definition, variation and aspects of polarity, one
can then presumably construct a cladogram
(or a series of cladograms of decreasing probability) using PAUP or some similar technique. The subsequent addition of time and
s ace permits the formulation of alternative
p ylogenies, and, with the addition of paleoecology, functional morphology and (for the
genus Homo) the archaeological record, scenarios of hominid evolution.
But what do such cladograms and phylogenies represent? Are they mere1 hypotheses? If so, how do we test them? ore specificall , what data sets, independent of those
useciyto formulate the cladograms, do we use
to test these hypothetical sortings of the
hominid fossil record?
Ideally, it would be ossible to generate
additional data for the ominid fossil record
to test clado am^.^ But given the finite size
of that fossi record, it is likely that most, if
not all, of the available data have already
been used to create the cladograms in the
first place. Are we then reduced to waiting
for fortuituous paleontological discoveries to
test our hypotheses?
It is therefore apparent that the application of the cladistic methodology to the hominid fossil record has some advantages and a
series of limitations. Primarily, it helps us to
be more ex licit about our morphological
criteria a n j methodological assumptions.
However, it requires us to make certain assumptions that cannot be substantiated and
forces us to place data into categories that
can considerably obscure and distort information of evolutionary relevance.
Despite these difficulties, it is evident that
cladistics is here to stay. It is too seductive in
its simplicity and a arent methodological
rigor to fade away. et, until methods are
developed that will account for the naturally
mess aspects of variation, homoplasy, sample &limitation, trait definition, and the
accurate identification of polarity, it will
remain little more than a heuristic device for
the reliminary sorting of the known hominid ossil record. It will continue to provide
interesting hypotheses concerning hominid
phylogeny. It is not, unfortunately, the panacea for our hominid phylogenetic ills.
Since this aper
tique of Haigood
discussing the ap
the analysis of t
was com leted, the cri(1989) gas appeared,
lication of cladistics to
e fossil record of the
41f, indeed, any cladistically derived phylogenetic reconstruction is fully falsifiable (Arnold, 1981; Cartmill, 1981).
Andrews (eds.): Major Topics in Primate and Human
genus Homo. In that discussion, Habgood
Evolution. Cambridge: Cambridge University Press,
primarily addresses the issues of variation,
pp. 295-316.
frequency (as opposed to categorical) differ- Bonde
N (1977) Cladistic classification as applied to
ences in trait representation between homivertebrates. In MK Hecht, PC Goody, and BM Hecht
nid sam les, and the difficulties of translat(eds.): Major Patterns in Vertebrate Evolution. New
York: Plenum Press, pp. 741-804.
ing cla istically derived arrangements of
N (1981) Problems of Species Conce ts in Palaesamples into reasonable phylogenies. I would Bonde
ontology. Inter. Symp. Concept. Paleontor Barcelona.
agree with his criticisms of cladistics, most of
pp. 19-34.
which are paralleled (without specific exam- Bromage TG (1986) A Com arative Scanning Electron
les) in the above discussion. However, I find
Microscope Study of Facial)Growth and Remodeling in
Early Hominids. Ph.D. Thesis, University of Toronto.
{is emphasis on phenetics as the answer to
M (1981) Hypothesis testing and phylogenetic
cladistics to be unfortunate. The computa- Cartmill
reconstruction. Z. Zool. Syst. Evolut. Forsch. 195‘3-96.
tion of morphological distance statistics be- Chamberlain
AT and Wood BA 119871 Early hominid
tween samples deals with the problem of
phylogeny. J . Hum. Evol. 16:119-133.
within-group variation, but it ignores the Churchill FB (1980) The modern evolutionary synthesis
and the bio enetic law In E Mayr and WB Provine
other roblems regarding sample definition,
(eds.): The &olutiona& Synthesis. Cambridge: Hartrait i entification, etc. discussed above. The
vard University Press, pp. 112-122.
best approach, if not the methodologically Condemi
S (1988a) A review and analysis of the Rissmost satisfactory one, is robably one that
Wiinn Saccopastore skulls: Can they provide evidence
regards the available fossiI;samples (or specin regard to the origin of near eastern Neanderthals?
In E Trinkaus (ed.): L’Homme de Neandertal 3:
imens) as representatives of prehistoric popL‘Anatomie. Etudes Rech. Archeol. Univ. Liege 30:
ulations or lineages acting as portions of
dynamic evolutionary units. In that respect, Condemi S (1988b)Caracteres plesiomorphes et apomorboth the cladistic and phenetic approaches,
phes de l’os temporal des neanderthaliens europeens
wiirmiens. In E Trinkaus (ed.): L’Homme de Neanalthough methodolo ’cally rigorous, are dedertal 3: L’Anatomie. Etudes Rech. Archeol. Univ.
ficient in dealing wit the biological realities
Liege 30:49-52.
of the past human groups of relevance.
Condemi S (1988~)
Reexamen des hommes fossiles Riss-
This paper has grown out of numerous
discussions with paleoanthropological colleagues concerning later Pleistocene human
phylogeny and hominid phylogeny in general and my consequent frustrations with
what appear to be overly sim listic or inapto an a Ipmittedly very
Those individuals, of
J.J. Hublin, F.H. Smith,
C.B. Stringer, A.M. Tillier, B. Vandermeersch, and M.H. Wolpoff are most prominent, have usually failed to convince me of
their positions but, in so doing, have provided me with food for thought. I am grateful
to them. In addition, J.C. Long, K.M.
Trinkaus, A.T. Chamberlain, F.H. Smith,
R.G. Franciscus, and R.L. Tompkins have
provided helpful comments on earlier versions of this paper. The liability, needless to
say, remains mine.
Andrews P (1984) An alternative interpretation of the
characters used to define Homo erectus. Cour. Forsch.
Inst. Senckenberg 69:167-175.
Arnold EN (1981) Estimating phylogenies at low taxonomic levels. Z. Zool. Svst. Evolut. Forsch. 19:l-35.
Bilsborough A and Wood BA (1986) The nature, origin
and fate ofHomo erectus. In BA Wood, L Martin, and P
Wiirm de Saccopastore (Italie): caracteres archaiques
et neanderthaliens. C. R. Acad. Sci. Paris, Serie I1
Dean MC (1986) Homo and Puranthropus: Similarities
in the cranial base and developing dentition. In BA
Wood, LMartin, and P Andrews (eds.): Major Topics in
Primate and Human Evolution. Cambridge: Cambridge University Press, pp. 249-265.
Dean MC (1988) Growth processes in the cranial base of
Hominoids and their bearing on morphological similarities that exist in the cranial base of Homo and
Purunthropus. In FE Grine (ed.):Evolutionary History
ofthe “Robust”Austra1opithecines.Chicago: Aldine de
Gruyter, pp. 107-112.
Delson E (1978) Models of early hominid hylo eny In
CJ Jolly (eds.): Early Hominids of d i c a . fondon:
Duckworth, pp. 517-541.
Delson E, Eldredge N, and Tattersall I (1977) Reconstruction of hominid phylogeny: A testable framework
based on cladistic analysis. J . Hum. Evol. 6:263-278.
Eldredge N and Tattersall I(1975)Evolutionar models,
phylo enetic reconstruction, and another loo$ a t hominid p iylogeny. Contr. Primatol. 5~218-242.
Felsenstein J (1983) Parsimony in systematics: Biological and statistical issues. Annu. Rev. Ecol. Syst.
Fink WL (1986) Microcomputers and phylogenetic analysis. Science 234:1135-1139.
Fisher RA. Corbet AS. and Williams CB (1943) The
relation between the number of species and the number of individuals in a random sample of an animal
population. J. Animal Ecol. 12:42-58.
Grine FE (1988) Evolutionary history of the “robust”
Australopithecines: A summary and historical er
5 ective. In FE Grine (ed.): Evolutionary History ofPthe
flobust” Australopithecines. Chicago: Aldine de
Gruyter, pp. 509-520.
Gruneberg H (1965) Genes and genotypes affecting the
teeth of the mouse. J. Embryol. Exp. Morphol. 14:137159.
Habgood PJ (1989) An investi ation into the usefulness
of a cladistic ap roach to t8e study of the origin of
anatomically moiern humans. Hum. Evol. 4:241-252.
Hennig W (1966) Phylogenetic Systematics. Urhana:
University of Illinois Press.
Hublin JJ (1986) Some comments on the diagnostic
features of Homo erectus. Anthropos (Brno) 23:175187.
Huhlin JJ (1988a) Caracteres derives de la region occiitomastoidienne chez les Neandertaliens. In E
krinkaus (ed.): L’Homme de Neandertal 3:
Etudes Rech. Archeol. Univ. Liege 30:
-- -~
Hublin JJ (1988b) Les plus anciens representants de la
Hublin JJ (1989) Les caracteres derives #Homo erectus:
Relation avec l’augmentation de la masse squelettique. In G Giacobini (ed.): Hominidae: Proceedings of
the 2nd International Congress of Human Paleontology. Turin: Jaca Book, pp. 199-204.
Johanson DC and White T (1979) A systematic assessment of early African hominids. Science 202:321-330.
Kimbel WH, Thite TD, and Johanson DC (1986) On the
ph lo enetic analysis of early hominids. Curr. Anthrop0[25:361-362.
Kimbel WH, White TD, and Johanson DC (1988) Implications of KNM-WT 17000 for the evolution of “robust”
Austrulo ithecus. In FE Grine (ed.): Evolutionary History of t i e “Robust” Australopithecines. Chicago: Aldine de Gruyter, pp. 259-268.
Lieberman DE, Pilbeam DR, and Wood BA (1988) A
probabilistic approach to the problem of sexual dimorphism in Homo habrlis: a com arison of KNM ER 1470
and KNM-ER 1813. J . Hum. Evol. 17:503-5k.
Mayr E (1974) Cladistic anal sis or cladistic classification? Z. Zool. Syst. Evolut. {orsch. 12:94-128.
Minugh-Purvis N (1988) Patterns of Craniofacial
Growth and Development in Upper Pleistocene Hominids. Ph.D. Thesis, University of Pennsylvania.
Olson TR (1978) Hominid phylogenetics and the existence of Homo in Member I of the Swartkrans Formation, South Africa. J. Hum. Evol. 7:159-178.
Olson TR (1981) Basicraniat morphology of the extant
hominoids and Pliocene hominids: The new material
from the Hadar Formation, Ethiopia, and its significance in early human evolution and taxonomy. In CB
Stringer (ed.): Aspects of Human Evolution. London:
Taylor and Francis, pp. 99-128.
Pielou EC (1975) Ecological Diversity. New York WileyInterscience.
Santa Luca AP (1980) The Ngandong fossil hominids: A
comparative study of a far eastern Homo erectus
group. Yale University Pub. Anthropol. 78:l-175.
Schmitt M (1989) Claims and limits of phylogenetic
systematics. Z. Zool. Syst. Evolut. Forsch. 27:181-190.
Simpson GG (1961) Principles of Animal Taxonomy.
Skelton RR, McHenry HM, and Drawhorn GM (1986)
Phylogenetic analysis of early hominids. Curr. Anthropol. 27:2143.
Stringer CB (1984) The definition of Homo erectus and
the existence of the species in Africa and Europe. Cour.
Forsch. Inst. Senckenberg 69:131-143.
Stringer CB (1985)Middle Pleistocene hominid variabil-
Stringer CB (1986) The credibility of Homo habilis. In
BA Wood, L Martin, and P Andrews (eds.): Major
Topics in Primate and Human Evolution. Cambridge
University Press, pp. 266294.
Stringer CB (1987)A numerical cladistic analysis for the
genus Homo. J. Hum. Evol. 16:135-146.
Strin er CB and Andrews P (1988) The genetic and fossil
eviience for the origin of modern humans. Science
Swofford DL (1985) PAUP: version 2.4. Champaign:
Illinois Natural History Survey.
Tattersall I (1986)Species recognition in human paleontology. J. Hum. Evol. 15:165-175.
Tillier AM (1988)A propos de sequences phylogenique et
ontogenique chez les neanderthaliens. In E Trinkaus
(ed.): L‘Homme de Neandertal3: L’Anatomie. Etudes
Rech. Archeol. Univ. Liege 30:125-135.
Tobias PV (1980) ‘Xustralopithecus afarensis” and A.
ufricanus: critique and an alternative hypothesis.
Palaeontol. Afr. 23:l-17.
Tobias PV (1988) Numerous apparently synapomorphic
features in Australopithecus robustus, Australopithecus boisei and Homo habilis: Support for the SkeltonMcHenry-Drawhorn hy othesis. In FE Grine (ed.):
Evolutionary History o?the “Robust” Australopithecines. Chicago: Aldine de Gruyter, pp. 293-308.
Trinkaus E (1983) The Shanidar Neandertals. New
York: Academic Press.
Trinkaus E (1984) Western Asia. In FH Smith and F
Spencer (eds.): The Origins of Modern Humans. New
York Alan R. Liss, pp. 251-293.
Trinkaus E (1987) Bodies, brawn, brains and noses:
Human ancestors and human predation. In MH Nitecki and DV Nitecki (eds.): The Evolution of Human
Hunting. New York Plenum Press, pp. 107-145.
Trinkaus E (1988)The evolutionary origins of the Neandertals or, why were there Neandertals? In E Trinkaus
(ed.): L’Homme de Neandertal3: L’Anatomie. Etudes
Rech. Archeol. Univ. Liege 30:11-29.
Trinkaus E (in press) Les hommes fossiles de la Grotte de
Shanidar, Irak: Evolution et continuit6 parmi les hommes archaiques tardifs du Proche Orient. L’Anthropol.
Turner A and ChamberlainA(1989) Sueciation. moruhoWalker A (1981) The Koobi Fora Hominids and their
bearing on the origins of the genus Homo. In BA
Sigmon and J S Cybulski (eds.): Homo erectus: Papers
in Honor of Davidson Black. Toronto: University of
Toronto Press, pp. 193-215.
Walker A and Leakey RE (1988) The evolution of Australopithecus boisei. In FE Grine (ed.): Evolutionary
History of the “Robust” Australopithecines. Chicago:
Aldine de Gruyter, pp. 247-258.
Walker A, Leakey RE, Harris JM, and Brown FH (1986)
2.5-Myr Australopithecus boisei from west of Lake
Turkana, Kenya. Nature 322:517-522.
Wolpoff MH (1980) Cranial remains of Middle Pleistocene European hominids. J. Hum. Evol. 9:339-358.
Wood RA (3984) The origins of Homo erectus. Cour
Forsch. Inst. Senckenberg 69:99-111.
Wood BA (1988)Are “robust”Australopithecines a monophyletic group? In FE Grine (ed.): Evolutionary History of the “Robust” Australopithecines. Chicago: Aldine de Gruyter, pp. 269-284.
Wood BA and Chamberlain AT (1986)Australopithecus:
Grade or Clade? In BA Wood, L Martin and P Andrews
(eds.): Major Topics in Primate and Human Evolution.
Cambridge: Cambridge University Press, pp. 220248.
Wood BA and Chamberlain AT (1987) The nature and
affinities of the “robust” Australopithecines: A review.
J. Hum. Evol. I6:625-641.
WuRandDongX(1985)HomoerectusinChina.In WuR
and JW Olsen (eds.): Palaeoanthropology and Palaeolithic Archaeology in the People’s Republic of China.
New York: Academic Press, pp. 79-89.
Wu X and Wu M (1985) Early Homo supiens in China. In
Wu R and JW Olsen (eds.): Palaeoanthropology and
Palaeolithic Archaeology in the People’s Republic of
China. New York Academic Press, pp. 91-106.
Без категории
Размер файла
1 441 Кб
record, cladistics, fossil, hominis
Пожаловаться на содержимое документа