Methodology of Comparison in Evolutionar
Methodology of Comparison in Evolutionar
in Evolutionary Archaeology
HectorNeff
University ofMissouri, Columbia
Daniel O. Larson
California State University, Long Beach
ABSTRACT
Evolutionary archaeology needs methods for recognizing how selective retention of cultural variation
shaped the archaeological record. One approach that has not yet been exploited fully is to derive adap-
tive hypotheses from evolutionary design arguments, then to test the hypotheses' predictions by com-
paring two or more segments of the archaeological record. In order to encourage further development
of comparative methods, we discuss aspects of their application within evolutionary archaeology. We
also present an example in which comparative observations are used to test a hypothesis about the
evolution of local productive specialization and circulation of materials through space.
75
76 Hector Neff * Daniel O. Larson
general discussion of methods of evolutionary investiga- highly debatable because they were compiled by repre-
tion, then to a more focused consideration of how com- sentatives of an expanding European culture at a time
parative data of use to evolutionists can be collected from when most of the subject groups were becoming extinct
the archaeological record. In the final section, we present (Dunnell 1980: 46). Since (a) cultural evolution fails to
an example illustrating the role of comparative methods provide explanatory mechanisms and (b) the empirical
in evolutionary archaeology. generalizations that it does provide are highly suspect
outside of the very limited historical contexts in which
ethnographic data were collected by European travelers
ORIGINS OF AMBIVALENCE REGARDING and ethnographers, Dunnell concluded that cultural evo-
COMPARISON IN EVOLUTIONARY lution is not a suitable basis on which to build a scien-
ARCHAEOLOGY tific archaeology.
Comparative methods emerge from Dunnell's dis-
Although Darwinian evolution had had some in- cussion of cultural evolution in a bad light because they
fluence in archaeology before 1980 (e.g., Meggers, Evans, are central to the nonscientific version of evolution that
& Estrada 1965, Meggers 1983; also see Lyman & O'Brien, had to be rejected decisively in order to clear the decks
this volume), the persuasive essays of Dunnell (esp. 1980) for construction of a scientific evolutioa Dunnell (1980:
sparked a wave of interest that sustained a concerted ef- 50) is explicit regarding the necessity to discard every-
fort to explore the theory's implications for archaeol- thing associated with cultural evolution: "the flaws in
ogy. In light of their seminal role in evolutionary ar- cultural evolution are fundamental and characterize all
chaeology, Dunnell's essays demand special attention in aspects of the approach; no simple refinement of its con-
a search for understanding of methodological trends and cepts will materially alter its structure."
biases. Dunnell does not explicitly reject comparative
methods, but neither does he mention any specific role Cultural Ecology and Adaptation
for them. More to the point, he argues that several ap-
proaches which do incorporate comparative methods are The association of comparative methods with cul-
incompatible with evolutionary archaeology. tural ecology and the concept of adaptation may also be
seen as a liability in the context of Dunnell's initial pro-
Cultural Evolution grammatic statements on evolutionary archaeology. Cul-
tural ecology, the methodological arm of Steward's
Dunnell's (1980) point of departure in advocating multilinear cultural evolution (see above), was conceived
development of a scientific evolutionary theory for ar- as "a method for recognizing the ways in which culture
1
chaeology was a categorical rejection of "cultural evolu- change is induced by adaptation to environment* (Stew-
tion." Cultural evolution was integral to the new archae- ard 1955: 5). It is an explicitly comparative method de-
ology of the 1960s (e.g., Binford 1962,1965), which drew signed to document the occurrence of similar processes
from White (1949), Sahlins and Service (1960), Steward of adaptation in similar environments (Steward 1955:39-
(1955), and, more remotely, the early anthropological 42). The concept of adaptation was discussed extensively
evolutionists, Morgan and Tylor. Comparative ethno- in both anthropology (e.g., Alland 1975) and archaeol-
graphic analysis is a crucial methodological component ogy (e.g., Binford 1962, 1964; Kirch 1980) during the
of all cultural evolution: for unilinear evolutionists (e.g., 1960s through 1980s.
White 1949), comparisons among unrelated groups lead Several interrelated suspicions about the role of
to recognition of universal stages, whereas for multilinear adaptation in evolution arose around the time that the
evolutionists (e.g., Steward 1955), comparisons reveal first proposals for a selectionist archaeology were being
developmental regularities. advanced. For one thing, Dunnell (1980: 49-50; also see
Dunnell (1980) argued that cultural evolution has Dunnell 1987) suggested that cultural evolutionists like
little in common with scientific evolution but a great Steward perhaps had embraced "adaptation" rather than
deal in common with the nonscientific, progressive evo- selection because adaptation kept causality internal to
lutionary philosophy of Darwin's contemporary, Herbert culture, within the realm of human problem solving and
Spencer. Unlike the scientific theory of evolution, cul- intention. In a slightly different but related vein, Rindos
tural evolution provides no mechanisms with which to (1989: 35-37) has argued that cultural ecology embraces
explain history (Dunnell 1980:44), but instead yields plau- a non-Darwinian view of evolution as a process driven
sible empirical generalizations that merely summarize the by directed variation that arises at the level of "culture"
comparative ethnographic data. Further, the extent to in order to bring about more perfect adaptation of hu-
which ethnographic observations can be generalized is man societies to ecological conditions. Linking cause to
Methodology of Comparison in Evolutionary Archaeology 77
the intentions of the actors (or to "culture") whose be- ber of places (O'Brien 1996, O'Brien & Holland 1990,
havior needs to be explained contrasts with science, in 1992,1995a, 1995b; O'Brien et al. 1994) is that consider-
which theoretical propositions about mechanisms furnish ation of adaptation within archaeology is misplaced with-
causes (Dunnell 1992, Neiman 1995). In a scientific sense, out a focus on the only mechanism capable of generat-
then, "adaptation" explains nothing but rather itself needs ing adaptation, natural selection. We discuss this point
to be explained (e.g., O'Brien & Holland 1992). Once in more detail later on.
"adaptation" is shown to be devoid of explanatory con-
tent, the rationale for using comparative methods to docu- Reconstruction
ment similar adaptations becomes subject to debate.
Unlike anthropologists and archaeologists, biolo- A third theme in Dunnell's writings that works
gists have long conceived of adaptation as a product of against serious consideration of comparative methods is
natural selection (e.g. Mayr 1982). Around 1980, how- his opposition to reconstruction of behavior from ar-
ever, the "adaptationist program" began drawingfirebe- chaeological data. The inference of human behavior from
cause of the propensity of its practitioners to advance material traces is addressed most systematically by be-
untestable "just-so" stories to account for any observed havioral archaeology (e.g., Reid et al. 1975, Schiffer 1996,
pattern of phenotypic variation (Lewontin 1979, Gould Walker et al. 1995). This approach is manifestly com-
& Lewontin 1979).1 Gould and Lewontin (1979: 584) re- parative, with observation of material—behavior relation-
fer to this as the "Panglossian paradigm." They claim ships in the present (i.e., experimental archaeology and
that its adherents tend to analyze individual traits of or- ethnoarchaeology) making up a key component. Dunnell
ganisms in isolation and to explain each as an optimal (1978a, 1978b, 1980, 1982, 1989), however, has long re-
design for some function. Suboptimality is accommo- jected a role for behavioral inference in evolutionary ar-
dated by arguing that optimal design for the whole or- chaeology on the grounds that the theory must be writ-
ganism may require suboptimality of some parts, and ten in terms that are observable in the archaeological
"the notion that suboptimality might represent anything record, not in terms of behaviors that are connected to
other than the immediate work of natural selection is the record only by lengthy, ad hoc inferential chains.
not entertained" (Gould & Lewontin 1979: 585). Socio- Furthermore, since behavior-material relations must be
biological explanations of human behavior are singled invariant in order for behavior to be inferred from mate-
out as particularly egregious examples of this Panglossian rial, reconstruction of behavior requires the essentialist
paradigm (Lewontin 1979). The critique of adaptation and obviously incorrect assumption that human behav-
ior never changes (Dunnell 1989: 43-44). Dunnell's re-
within biology provided additional grounds for Dunnell
jection of behavioral inference does not necessarily close
to minimize its importance for evolutionary archaeology:
the door on all use of comparative data in evolutionary
.. . showing how something is adaptive is not archaeology, but it does rule out the use of comparisons
equivalent to demonstrating why it exists in the between the archaeological and ethnoarchaeological
first place, and it can lead to the ad hoc Dar- records (e.g., Walker et al. 1995: 8).
winian explanations criticized by Lewontin In summary, comparative methods are central to
(1977) in modern biology. Functional explana- cultural evolution, cultural ecology, the adaptationist
tions and adaptation are valuable elements of program, and behavioral reconstruction, all of which were
evolution to be sure, but they are not identical singled out for criticism by Dunnell in his initial theo-
with evolutionary explanations founded in natu- retical and methodological statements regarding evolu-
ral selection (Dunnell 1980: 50). tionary archaeology. In light of these criticisms, it is per-
With so little room for the concept of "adaptation" in haps understandable that comparative methods have not
evolutionary archaeology, it is no surprise that compara- loomed large within the programmatic literature of evo-
tive methods, which are basic to the study of adaptation lutionary archaeology. This is not to say that compari-
(see below), did not figure in Dunnell's initial method- son has been neglected entirely; despite the implications
ological suggestions. of Dunnell's programmatic essays, evolutionary archae-
Discomfort with cultural ecology and the ologists have looked to comparisons as means for test-
adaptationist program has continued to color discussions ing hypotheses generated from evolutionary theory. Later
of how to fashion an evolutionary approach to the ar- on, we return to the topic of how comparative archaeo-
chaeological record. O'Brien (1996:26), for instance, syn- logical data can be brought to bear in the testing of evo-
opsizes his views by noting that <f[s]illy adaptationist sto- lutionary hypotheses. First, however, we need to back-
ries are as much a problem in archaeology "as they are in track and say a few words about the derivation of the
biology." O'Brien's point, which is developed in a num- hypotheses that are the focus of evolutionary science.
78 Hector Neff * Daniel O. Larson
METHODOLOGICAL ISSUES IN have shown how neutral trait frequency distributions can
EVOLUTIONARY INVESTIGATION be used to infer the spatial structure of past populations
(see Chapters 10,15).
Evolutionary archaeology is based on the idea that An alternative to the investigation of evolution
artifacts in the archaeological record are parts of past based on the equations of population genetics is one that
human phenotypes (Dunnell 1989, Leonard & Jones 1987, focuses on the evolutionary entities that Hull (1980,1988:
Neff 1992,1993; O'Brien & Holland 1992,1995a, 1995b; 408; also see Williams 1992) calls 'interactors.' These are
O'Brien et al. 1994). As with other phenotypic traits, the the cohesive units that interact with the environment in
frequencies of traits recorded in assemblages of artifacts such a way that replicators may persist differentially.
wax and wane through history, giving rise to what Dar- Whereas the population genetics approach originated with
win called "descent with modification." For cultural traits, the neo-Darwinian synthesis of genetics and natural se-
this process is generated by (1) communication among lection, the focus on interactors dates back to The Origin
individuals of information on how to make and use ofSpecies. The basic assumption underlying this approach
implements, i.e., cultural inheritance, and (2) variation is that the observed adaptations of plants and animals in
in the success individuals achieve at communicating the present were created by natural selection based on
knowledge about how to make and use their implements. differential adaptedness (fitness) within populations in
Despite the apparent simplicity of the above con- past environments. This approach entails "optimality
ception and increasing acceptance of its usefulness among modeling," in which an optimality criterion and alterna-
archaeologists, methods for constructing and testing evo- tive strategy sets are specified, after which an optimal
lutionary accounts of the archaeological record cannot morphology or behavior is deduced based on the pre-
yet be said to have materialized. What might such meth- dicted fitness payoffs from each strategy (Harvey & Pagel
ods look like? For a historical science like archaeology, 1991, Maynard Smith 1994(1978], Sober 1993, Williams
whose subject matter consists of a finite, unique material 1992). Some assumption about heredity is also required,
record of the human past, the conclusion seems inescap- but Maynard Smith(1994[1978]) has suggested that a
able that testing of evolutionary hypotheses will have to broad "like begets like" assumption (explicitly including
be based not on experimentation but on observation and an assumption of cultural inheritance) is sufficient for
comparison. 2 Before explicating a methodology of com- most evolutionary optimization models. As mentioned
parison, however, some preliminary theoretical issues above, this basic approach to evolutionary investigations
related to the investigation of evolutionary phenomena has been severely criticized as the "Panglossian paradigm"
need to be addressed. The literature of evolutionary biol- by Gould and Lewontin (1979).
ogy and philosophy of evolutionary science treats these Inasmuch as archaeologists have little or no access
issues at length, so, inasmuch as artifactual evolution is to transmission mechanisms (replicators) but a great deal
produced by the same mechanisms as biological evolu- of access to interactors in the form of material remains
tion, the theoretical contributions of evolutionary bi- in the archaeological record, it would seem that an ap-
ologists and philosophers are directly relevant to the proach that focuses on adaptation and optimality mod-
methodology of evolutionary archaeology; evolutionary eling would have much to offer.3 First, however, it is nec-
archaeologists (e.g., Dunnell 1980,1989; O'Brien & Hol- essary to salvage the concept of adaptation from the criti-
land 1990, 1992) have, of course, recognized this before. cisms leveled within both evolutionary biology and ar-
Biologists tend to distinguish two different ap- chaeology.
proaches to the investigation of evolution (Harvey & Pagel
1991, Lewontin 1979,Maynard Smith 1994(1978]). On The Adaptationist Program
one hand, population geneticists use formal mathemati-
cal descriptions of migration and innovation rates, spa- Defenders of the adaptationist program in biol-
tial structure, sampling error, and selection coefficients ogy note that its purpose is not to test the general con-
to model changes in gene frequencies (e.g., Crow & cept of adaptation, as suggested by critics (e.g. Gould &
Kimura 1970). In Hull's (1980; also see Dawkins 1976) Lewontin 1979, Lewontin 1979), but rather to test spe-
terminology, this approach emphasizes replicators as cific hypotheses about the diversity of life that can be
opposed to interactors. In archaeology, such an approach deduced from the theory of natural selection (Maynard
is endorsed by Bettinger (1991, Bettinger et al. 1996), who Smith 1994(1978], Mayr 1983). In essence, "adaptation"
adapts Boyd and Richerson's (1985) cultural transmis- is a key component of a strategy for understanding the
sion models for archaeological ends; by Neiman (1995), living world as the product of natural selection coupled
who has modeled the temporal dynamics of stylistic (se- with chance, historical constraints, and developmental
lectively neutral) traits; and by Lipo et al. (n.d.), who constraints.
Methodology of Comparison in Evolutionary Archaeology 79
Comparisons may even straddle the gap between introduced through mechanisms such as accident, im-
genetic and cultural inheritance. Rindos' (1984) r e v o - perfect communication, and intentional problem solv-
lutionary theory of plant domestication and the origins ing (e.g., NefF 1992; Rindos 1980, 1989). New cultural
of agriculture is largely an argument about the general variants also tend to diffuse rapidly between populations,
conditions that favor evolution of mutualistic relation- thus further enhancing the available cultural variation
ships between organisms. Specific examples include mu- on which selection may act. To paraphrase Sober, it is
tualism between organisms whose phenotypes are deter- reasonable to expect existing cultural variation to be richer
mined primarily by genes, such as ants and acacias, but even than existing genetic variation, so natural selection
also mutualism between various plants and organisms has even greater latitude to shape adaptations out of ex-
whose phenotypes are determined to an important ex- isting cultural variation. Moreover, because of the rich-
tent by cultural inheritance (humans). In a further ex- ness of cultural variation, natural selection can be ex-
tension of this same basic design argument, Cackler and pected to create phenotypic convergence in response to
Neff (n.d.) have suggested recently that the history of similar constraints and opportunities even more fre-
Maya society can be viewed as a history of fluctuating quently for traits inherited culturally than for traits in-
mutualistic interaction between elites and commoners. herited genetically. In short, because cultural variation is
In summary, comparisons offer a means for test- so abundant, historical constraints on optimal design may
ing adaptive hypotheses derived from optimality (design) be less important than they are for genetically determined
arguments. The method of testing is to derive from the traits. This in turn suggests that comparison and other
design argument predictions about how a cultural trait methods for testing evolutionary design arguments may
or constellation of traits should vary, and then to check have an even more obvious role to play in archaeology
the predictions against two or more segments of the ar- than they do in biology.
chaeological record. The use of comparative methods in Parenthetically, the forgoing observations provide
evolutionary archaeology is illustrated in more detail in a counter-argument to the often-expressed view that evo-
the final section of this paper. Before proceeding, how- lutionary approaches in archaeology are flawed because
ever, four issues regarding the implementation of com- they deny the importance of consciousness and inten-
parative methods in evolutionary archaeology need to tionality in human behavior (e.g., Trigger 1989: 306). In
be clarified. fact, these human characteristics play a central role in
evolutionary theory as applied to people. That role, how-
Cultural Variation ever, is not to direct the course of evolution but to pro-
vide a constant supply of novel variation on which selec-
One of Gould and Lewontin's (1979) criticisms of tion may act in the future (Rindos 1989). And, to repeat,
the adaptationist program is that there is no reason to the abundance of cultural variation created by problem
assume that the requisite genetic variation will be avail- solving and other expressions of human intentionality
able in a population at the precise time when it is needed gives reason to expect adaptation and convergence to be
for natural selection to fashion an optimal solution to frequent outcomes of selection acting on cultural aspects
some set of current environmental conditions and op- of the human phenotype.
portunities. In another discussion of how history con-
strains optimal design, Gould and Vrba (1982) point out Observational Language
that traits designed by natural selection for one purpose
(or evolved for no purpose at all) are often co-opted for In the two examples of biological 'fitness propen-
other purposes altogether, thus becoming "exaptations." sity* mentioned above (running speed in zebras and color
Maynard Smith (1994(1978]) takes a different view of in moths), the organisms are described in terms of mea-
historical constraints, arguing that genetic variance of surement units (speed and color) that are independent of
some appropriate kind does usually exist in populations; the specific phenomena on which they are observed. Simi-
existing variation may not be that required for selection larly, Braun's (1983) argument about the design of Wood-
to fashion the absolute optimal solution, but it will be land ceramics from the Midwest US concerns measur-
adequate to fashion a better solution. Similarly, Sober able attributes of any ceramic vessel (vessel wall thick-
(1993:121) has observed that "natural selection will op- ness), and Rindos' (1984) revolutionary model of plant
timize with respect to existing variation, and it is reason- domestication invokes measures, such as per-capita yield
able to expect existing variation to be rich" (emphasis in and total yield, that are independent of any specific em-
original). pirical entities.
One of the distinguishing characteristics of cul- While seemingly obvious, these observations high-
tural inheritance is the ease with which new variation is light a fundamental distinction that is crucial not just to
82 Hector Neff * Daniel O. Larson
comparative methodology but to the overall goal of de- ational*1 in that any ceramic assemblage could be described
veloping scientific evolutionary archaeology. The distinc- in the same terms. The use of theoretical units that are
tion is between what Dunnell (1971, 1986, 1995) calls measurable in any ceramic collection makes it possible
"ideational," or "theoretical," units on the one hand and to compare the Plumbate industry, which is known only
"phenomenological," or "empirical," units on the other from archaeological collections, with modern ceramic
hand. Ideational units, such as classes, define phenom- industries, which are documented through the much
ena and are space-time invariant, whereas phenomeno- richer medium of ethnography or ethnoarchaeology. It
logical units, such as groups, are manifest by the phe- is clear, then, that many measurements are space-time
nomena that they contain and thus are defined only for invariant or "ideational" and thus suitable for compar-
a unique set of coordinates in space and time. According ing two or more evolutionary sequences.
to Dunnell (1995), this distinction resolves the "materi-
alist paradox:" ideational units make up the terms in Empirical Sufficiency, Ethnoarchaeology
which laws of history (e.g., evolutionary theory) must be and Experimentation
written, whereas empirical units subsume the individu-
als on which the historical laws act. This issue would Perhaps the most explicit use of optimality mod-
probably not even be worth mentioning were it not for eling in anthropology and archaeology has been the ap-
the fact that archaeologists still tend to view units such plication of evolutionary ecology (Smith & Winterhalder
as ceramic "types" as if they measured some theoretical 1992), in the form of optimal foraging theory, to hunter-
dimension of variation.6 If we wish to make compari- gatherer studies (e.g., Bettinger 1991, Hawkes & O'Connell
sons between evolutionary sequences observed in differ- 1992, O'Connell 1987, 1995). As already mentioned,
ent places and times, it is clear that the evolutionary selectionists (e.g., O'Brien 1996) have sometimes charac-
phenomena of interest (archaeological remains, which terized optimal foraging as Panglossian, adaptationist
are empirical) must be described in terms of units that story telling, a characterization which would surely be
are ideational (or theoretical). debated by the other side (e.g., Bird, Chapter 16). A more
The importance of describing archaeological phe- serious problem with optimal foraging applications in
nomena in terms of theoretical units rather than empiri- archaeology has been pointed out by one of its own prac-
cal units has not been lost on those seeking to develop titioners: "[models from behavioral ecology] were de-
methods for evolutionary archaeology. The engineering signed to analyze the behavior of individual organisms
design analyses favored by O'Brien and Holland (e.g., directly observed in ecological time, not the behavior of
1990,1992,1995b; O'Brien et al. 1994) are cases in point: aggregates indirectly reflected in material evidence accu-
the variables they and others (e.g., Feathers 1989, Schiffer mulated and incompletely preserved in archaeological
& Skibo 1987) use to describe ceramics, such as temper time" (O'Connell 1995:237). The observational units of
type and amount, ceramic strength and toughness, ves- evolutionary ecology (e.g., travel time, processing time,
sel-wall thickness, and others, are ideational and related patch choice, etc) are ideational, as required for scien-
to one another by an explicitly articulated model. If evo- tific evolutionary archaeology, but in practice these units
lutionary archaeology is to escape the "materialist para- are unobservable to archaeologists and instead must be
dox," such theoretical units must be adopted as the ob- "reconstructed" or inferred from the archaeological re-
servational language used to describe and explain change mains. The problem here is with what Dunnell (1980)
(Dunnell 1995: 35). calls "empirical sufficiency."
Ideational units suitable for construction of evo- As mentioned previously, opposition to the "re-
lutionary archaeology can be derived from any kind of construction" or "behavioral correlates" approach to ar-
design analysis, not just the engineering analyses advo- chaeology is a consistent theme running through
cated by O'Brien and colleagues. For instance, in previ- Dunnell's work (e.g., Dunnell 1978a, 1978b, 1980,1982,
ous work, Neff (1989, 1990, 1992, 1995b) has suggested 1989). In DunnelKs view, not only must the data lan-
that miniaturization and decorative elaboration associ- guage of evolutionary archaeology incorporate ideational
ated with the emergence of a highly distinctive Meso- (theoretical) rather than phenomenological (empirical)
american ceramic ware known as Tohil Plumbate are con- units, but the theoretical units must themselves be mea-
vergent with a number of modern instances of miniatur- surable directly in the record. If, as proposed by evolu-
ization and decorative elaboration among traditional tionary ecologists, evolutionary archaeology were phrased
potters faced with growing external demand for their in transactional (behavioral) terms that had to be inferred
products. Although the observational language in this from the material record, then it would continue,to pro-
case consists of simple measures of vessel size and extent duce untestable "just-so" stories (Dunnell 1982, O'Brien
of decorative modification, the language is clearly "ide- 1996). As noted above, at least one archaeologist work-
Methodology of Comparison in Evolutionary Archaeology 83
ing from an evolutionary ecological perspective whether it requires extensive inferential chains that lead
(O'Connell 1995) explicitly acknowledges this liability; by a circuitous path from measurable material proper-
unfortunately, he provides no advice as to how it might ties (whether theoretically justified or not) to reconstruc-
be circumvented. tions of gender, kinship, political structure, and other
Schiffer (1996: 650-652) has recently argued that nonobservables. It may be true that "every selectionist
Dunnell's opposition to behavioral inference is itself in- model can be shown to rest, implicitly or explicitly, on a
compatible with evolutionary archaeology. Schiffer ar- network of behavioral inferences" (Schiffer 1996: 650),
gues, first, that measurements of interest in scientific in- but the requirement of empirical sufficiency forces us to
vestigation can only rarely be made directly in the phe- admit only the shortest, strongest possible inferential
nomenological world, as Dunnell would require of an chains into our research programs.
empirically sufficient archaeology. Second, he argues that The major implication that Schiffer (1996: 652)
evolutionary biology depends on paleoenvironmental draws from his discussion of Dunnell's views on recon-
reconstruction as well as inferences about the behavior struction and inference is that comparative strategies of
of extinct organisms, so the opposition to reconstruc- actualistic and ethnoarchaeological research would en-
tion has no parallel in other evolutionary sciences. Third, hance evolutionary archaeology if they were taken seri-
he suggests that close inspection of evolutionary models ously. This is a point with which a number of evolution-
in archaeology, even Dunnell's, reveals that they are ac- ists (eg., Neff 1990,1992: 179,1995; O'Brien & Holland
tually expressed in behavioral terms. Fourth, Schiffer ar- 1990: 48, 1992: 4) would agree. But this does not mean
gues that behavioral inference does not require stasis, as that we need to adopt an observational language filled
Dunnell asserts, but rather establishes the boundary con- with behavioral terms. Rather, actualistic and ethno-
ditions under which certain behavioral processes will lead archaeological studies must be carried out using terms
to certain material results. that are both time-space invariant (ideational) and mea-
To some extent, Dunnell and Schiffer are talking surable in the archaeological record. Experimental stud-
past one another. Certainly Dunnell would not assert ies cited by Schiffer (Dunnell & Feathers 1991, O'Brien
that "empirical in the record" means that human senses et al. 1994) as well, arguably, as Schiffer's own work (e.g.,
can provide the only access to archaeological phenom- Schiffer & Skibo 1997), exemplify these principles.
ena. Almost any scientific measurement entails theoreti- The Plumbate case introduced above exemplifies
cal arguments that connect what is actually measured with the importance of appropriate measurement scales and
the dimensions of interest. In chemistry-based archaeo- empirical sufficiency when ethnoarchaeological data are
logical provenance research, for instance, the goal is to combined with archaeological data in a comparative
monitor geographic space (the location of a "source"), study. Ceramic size and decorative coverage can be ob-
but the actual measurements made in the laboratory are served both archaeologically and ethnographically, and
parts per million or weight-percent of the determined therefore a hypothesis of convergence can be formulated
oxides or elements; the archaeological units of interest to account for the trends toward miniaturization and
(geographic coordinates) are connected to the analytical decorative elaboration observed both in the Plumbate
measurements by a theoretical warrant labeled the "prov- industry and several ethnographically known ceramic
enance postulate" (Neff n.d., Weigand et al. 1977). As industries. This hypothesis gains additional credence from
Schiffer (1996: 650) points out, Dunnell himself utilizes the observations that, like the modern industries, which
such inferences, as in his reference to "costly imports" in have experienced growing tourist demand during the 20th
the Woodland mortuary cult of the Midwest U.S. century, Plumbate potters were tied into an interregional
(Dunnell 1989: 48). Schiffer is right to characterize this exchange system that expanded rapidly at the beginning
as an inference: "import" is an inference based on knowl- of the Early Postclassic period, at roughly the same time
edge of the natural distribution of raw materials, and that elaborately decorated and miniaturized vessels be-
"costly" is an inference based on the distance to source gan to appear (Neff 1989b, 1990, 1995).
areas of the imports; both inferences are based on mea-
surements given theoretical warrant by the provenance Homology and History
postulate, and the compound characterization "costly
import" then reflects only a deduction about what must We noted previously that abundant novel varia-
have been necessary (long-distance movement by humans) tion tends to reduce historical constraints on adaptive
to give the artifacts their observed archaeological distri- design of cultural traits. We wish to disavow emphati-
butions. In accord with O'Brien and Holland (1995b: cally, however, any claim that cultural aspects of human
159-160), we would suggest that the issue is not whether phenotypes are liberated from the constraints of history.
evolutionary archaeology requires inference but instead As stressed by one of us elsewhere (Neff 1991), evolu-
84 Hector Neff & Daniel O. Larson
tionary archaeology is as much about history as it is about appeared before the Late Postclassic period, is hypoth-
ecology and economics. To paraphrase Stephen J. Gould esized to have undergone convergent evolution with sev-
(1986: 74), the material reality of history flows through eral 20th century Mesoamerican ceramic traditions whose
the design standards that are external to history, forming roots lie in late prehistoric or post-conquest ceramic tra-
a set of contrasts that define evolutionary archaeology: ditions of their respective regions. Moreover, parallels in
homology and analogy, history and immanence, move- the selective pressures favoring miniaturization and deco-
ment and stability. rative elaboration (increasing external demand) are rooted
History matters in the present discussion because in entirely different historical circumstances, one associ-
use of comparative data to test hypotheses about cul- ated with growing interregional interaction in Early
tural adaptations requires independence of the cases be- Postclassic Mesoamerica and the other associated with
ing compared (Mace & Pagel 1994; also see Bettinger et growing tourism during the 20th century.
al. 1996: 153). The problem is parallel to that encoun- The continuous historical character of the archaeo-
tered in comparative biology (e.g., Harvey & Pagel 1991), logical record offers additional potential for the testing
in which comparisons are relevant to testing adaptive of adaptive hypotheses, as O'Brien and Holland (1990,
hypotheses only if a common origin for the trait(s) in 1992) have recognized. The argument rests on Dunnell's
question can be ruled out. In anthropology, the problem (1978a) distinction between stylistic (adaptively neutral)
of distinguishing historical (homologous) from func- and functional (non-neutral) traits. As corroborated re-
tional (analogous) associations is known as Galton's prob- cently by Neiman's (1995) simulations, neutral traits show
lem after Francis Galton, who first described it to the smooth, temporal distributions approximating the battle-
Royal Anthropological Institute in 1889 (Naroll 1961). ship curves on which convincing chronologies are based.
Mace and Pagel (1994) suggest that the key to resolving Traits under the control of selection, in contrast, are ex-
Galton's problem lies in reconstructing cultural phylog- pected to increase rapidly to fixation once they come
enies on the basis of genetic, linguistic, historical, or ar- under the control of selection. O'Brien and Holland
chaeological evidence. Based on such phylogenies, the (1990, 1992) view the rapidity with which shell temper
number of independent occurrences of some trait asso- replaced grit temper in the Late Woodland period of the
ciation can then be determined. Midwest as convincing evidence that several aspects of
Verifying the independent origins of presumed ceramic resource selection and pyrotechnology were
adaptive traits is no less important in evolutionary ar- shaped by directional selection into a new adaptation
chaeology than in evolutionary anthropology or evolu- around 1000 A.D.7 Similarly, Neff and Arroyo (1996)
tionary biology (Neff 1990; also see Kirch & Green 1987: argue that the rapidity with which ceramics appeared soon
443). The methods are clearly distinct, however. Whereas after 2000 B.C. on the southern Mesoamerican coast in-
anthropologists may have access to linguistic, genetic, dicates that their adoption was a selection-driven pro-
and historical data, archaeologists most often have only cess. These examples illustrate how the high-resolution
the material record on which to base conclusions about ratio-scale measurement of time possible with archaeo-
phylogeny. Where evolutionary archaeologists have the logical data permits comparison not only of the final
advantage is in their access to a more complete, better- outcomes of adaptive processes but of their temporal tra-
resolved historical record of the traits of interest: com- jectories as well. Access both to the direction of evolu-
parative biologists and anthropologists look only (or tionary change and to its pace gives archaeology a de-
primarily) at the tips of a phylogenetic tree and must cided advantage over most other evolutionary sciences:
infer relationships based on parsimony and the pattern in archaeology, comparative methods can be applied not
of contemporary trait occurrence; archaeologists, in con- just to adaptive traits (the outcomes of selection) but to
trast, can look at the histories of traits in one or more the selection-driven adaptive trends themselves.
sequences. In the archaeological case, independent ori-
gin and the direction of evolutionary change are moni-
tored by establishing the timing of introduction, fixa- EXAMPLE OF A DESIGN ARGUMENT
tion, and disappearance of the traits in the sequences AND ITS EVALUATION BY COMPARATIVE
being compared and by considering the evidence pertain- METHODS
ing to common descent (Neff 1990, 1993).
Decorative elaboration and miniaturization in the Ultimately, methodological discussions like that
Plumbate tradition again serves as a convenient example presented in the previous sections should pay off in the
of how independent occurrence can be established in ar- form of enhanced understanding of how evolutionary
chaeological investigations that employ the comparative mechanisms shaped specific segments of the archaeologi-
method. The Plumbate tradition, which apparently dis- cal record. Therefore, in this section we consider a model
Methodology of Comparison in Evolutionary Archaeology 85
1
Predictable Environment Unpredictable Environment
Figure 5.1 A model of how selection shapes the patterning of formal and compositional diversity in the archaeological record.
86 Hector Ncff & Daniel O. Larson
of how human movement of goods through space is of the goods that are easiest to produce locally and then
shaped by selection associated with environmental pre- seek to obtain the needed goods through exchange. The
dictability. The empirical details alluded to here are docu- second strategy, economic interaction, will be favored if
mented mote extensively in other publications (Larson its fitness costs are lower than those associated with steal-
et al. 1996, Neff 1995a, Neff& Hodge n.d.,Neff & Larson ing (which may include, for example, the cost of plan-
n.d., Neffetal. 1997). ning and executing a raid, the cost of guarding one's own
resources, the increased possibility of suffering a violent
The Model death, etc.). Thus, given the right balance of fitness costs
Movement of objects through space is a universal and benefits, the selective pressures on populations dis-
human propensity that entails significant investment of tributed across a landscape of consistent differentials in
human energy and therefore must confer some substan- comparative advantage will be complementary and will
tial fitness benefit in order to persist. Because of the en- favor the growth of local specialization coupled with re-
ergy investment, one would also expect the specific ways gional economic interaction. Increasing regional eco-
in which humans move materials to be controlled by nomic interaction permits individuals to capitalize on
selection. We therefore propose that selection designs the potential for local specialization due to differential
different kinds of mechanisms for moving objects through comparative advantage, and this reinforcing feedback
space in predictable vs. unpredictable environments: lo- favors increasingly complex patterns of local specializa-
cal specialization and economic interaction are selected tion, such as specialization in specific kinds of ceramic
for simultaneously in predictable environments, but ex- vessels. In short, local specialization and regional eco-
change and other mechanisms of ceramic movement are nomic interaction are mutually reinforcing aspects of a
favored without local specialization in unpredictable en- single selective process arising from consistent differen-
vironments. This is an adaptive hypothesis that we would tials in comparative advantage. Arnold (1975,1978) and
like to be able to test by evaluating how well its predic- Stark (1991) enumerate ethnographic examples of part-time
tions hold under various historical circumstances. For- specialist pottery making communities that apparently
tuitously, the history of how humans have moved non- evolved through this process of niche diversificatioa
perishable objects (such as ceramics) through space in Archaeological expectations for predictable envi-
various times and places can be monitored with some ronments are shown at the bottom left in Figure 5.1.
confidence using methods of archaeological provenance First, the selection-driven trend toward community spe-
investigation (Neff n.d.). As a result, we can connect the cialization within a region is reflected in declining com-
design argument of our model to specific predictions positional diversity of specific forms or decorative classes
about the patterning of ceramic remains in the archaeo- of pottery, as some communities stop producing vessels
logical record (Figure 5.1; also see Neff & Larson n.d.). that they increasingly obtain through economic interac-
Let us first consider an ideal predictable environ- tion. Second, ceramic assemblages are expected to show
ment (the left half of Figure 5.1). Geographic differences a strong association between ceramic form/decoration
in return from subsistence and other productive activi- on the one hand and raw material source on the other
ties are consistent from year to year, and these consistent hand. One extreme possibility is that pottery production
differentials in comparative advantage favor differentia- becomes the exclusive specialty of one community in a
tion of productive strategies. Consider two locations, one region, in which case the ceramics in the region's assem-
in which biotic productivity is consistently high but ce- blages manifest a restricted range of compositions char-
ramic resources are scarce or of low quality, and another acteristic of raw materials near the specialized commu-
in which biotic productivity is consistently low but ce- nity. In more complex regional economic systems, sev-
ramic resources are high-quality and plentiful. Other eral ceramic producing communities may specialize in
things being equal, everyone's fitness is maximized by distinctive shapes and/or decorative techniques, in which
obtaining sufficient food and sufficient pots for the low- case the accumulated ceramic assemblages should show
est possible expenditure of time and energy: for people strong association between composition and formal/deco-
situated favorably with respect to ceramic resources, fit- rative modes.
ness is enhanced by any practice that reduces the costs of Now consider the mechanisms that would move
obtaining food; conversely, for people situated favorably ceramics through space in an extremely unpredictable
with respect to subsistence resources, fitness is enhanced environment (the right side of Figure 5.1). In such an
by any practice that lowers the costs of obtaining suffi- environment, differentials in comparative advantage are
cient pots. Two alternative strategies for lowering fitness inconsistent from one year to the next, so selection for
costs in each location are (1) steal the needed goods (pots local specialization is relaxed or non-existent Without
or food) from someone else or (2) produce a slight excess local specialization, exchange is favored as but one of
Methodology of Comparison in Evolutionary Archaeology 87
several mechanisms that permit differential allocation of to do with the way that attributes of form and decora-
subsistence effort across space and through time. Other tion pattern with respect to compositional group (source).
mechanisms may include migration, short-term popula- In some cases, the formal attributes that we record may
tion movements, and maintenance of widespread net- coincide with attributes used to identify "wares," "types,"
works of social interaction (Braun & Plog 1982, Leonard or similar empirical units that appear in culture-histori-
1989, Leonard & Reed 1993, Plog 1983, 1984). Any one cal frameworks. But this does not mean that the formal
or combination of these behaviors confers a fitness ben- attributes we are concerned with should be regarded as
efit by increasing the likelihood that an adequate food empirical rather than ideational, even if we use "type" or
supply will be available, despite the unpredictability of "ware" names to refer to a set of consistently associated
biotic productivity. Goods such as ceramics move across formal attributes. What transforms the latter (which are
the landscape as a result of any or all of these behaviors, ideational) into the empirical units of culture history is
but the movement of ceramics can be seen as an inciden- the hypothesis that common descent connects em-
tal byproduct of selection for mobility and/or social in- pirical entities on which certain formal attributes can
teraction among people who happen to have made and be observed (Neff 1993, Tschauner 1994; also see foot-
used ceramic pots. Pots may be imported to a location as note 3).8
gifts in anticipation of reciprocity during a future short- Another important observation about the model
fall, as containers used to transport foodstuffs during an diagrammed in Figure 5.1 is that it does not explicitly
actual shortfall, by an immigrant relocating in an attempt differentiate between ceramic traits that are functional
to enhance return from his/her agricultural labor, or by (under the control of selection) and traits that are stylis-
an immigrant relocating as a result of marriage. These tic (selectively neutral). Instead, it concerns the control
behaviors are equivalent phenotypic expressions within that selection exerts on the distribution of productive
the framework of the model. That is, the model predicts activities among populations; it contrasts the pattern of
only that ceramics will move through space and makes production and movement of goods expected in predict-
no specific prediction about conditions that would fa- able vs. unpredictable environments. Thus, for example,
vor one behavior over another. It connects observations given a sufficiently predictable environment, compara-
that can be made on the archaeological record (the pres- tive advantage may favor local specialization in ceramic
ence of a nonlocal pot) to selective pressures in an un- production in some community, even though the spe-
predictable environment cific formal/decorative attributes that characterize pot-
Expectations for ceramic assemblages accumulated tery from that community reached high frequencies en-
in unpredictable environments are shown on the bot- tirely by chance processes (drift); alternatively, their high
tom right in Figure 5.1. First, because selection does not frequencies may be a result of a selective process that
favor local specialization, regional compositional diver- shaped the technological and decorative practices of pot-
sity is expected to be high (people living at many loca- ters in the community. Our model specifies only that
tions produce ceramics out of local raw materials). Sec- selection favors a pattern in which pottery-making prac-
ond, because selection favors various activities that move tices (including whether or not pottery is made at all) are
vessels across the landscape, evolution is expected to bring distributed heterogeneously among populations and there-
similarceramicsfromdifferent sources together in the same fore that a strong association will be detectable between
assemblage. That is, in contrast to the expectation for a formal/decorative attributes and composition (source).
predictable environment, there should be weak or incon- What we wish to evaluate, then, is an adaptive de-
sistent association of decorative and formal attributes sign argument that predicts a contrast between the pat-
with composition. Whether ceramics moved through terning of ceramic diversity in assemblages accumulated
migration, residential mobility, or exchange, accumula- in predictable vs. unpredictable environments (Figure 5.1).
tion of a diverse yet weakly patterned ceramic assem- In keeping with the principles discussed previously, the
blage is favored by selection associated with fluctuating en- expectations about ceramic diversity are written in an
vironmental conditions and the attendant subsistence risk. ideational data language, and they are connected to the
It is important to emphasize that the model dia- adaptive model by short, explicit inferential chains, as
grammed in Figure 5.1 is written in an ideational data dictated by the requirement of empirical sufficiency.
language, as required for an evolutionary design argu-
ment (Dunnell 1995; see above). Resource procurement Evolution of Ceramic Diversity in Predictable
practices are monitored via provenance determination, Environments
and forming and decorative practices of potters are moni-
tored via attributes of form and decoration observed on Humid tropical regions tend to have high and pre-
sherds. The specific predictions that we are testing have dictable available moisture, which translates into predict-
88 Hector Neff 8c Daniel O. Larson
able variation in primary biotic productivity and, accord- border (Neff 1995a; Neff, Bove & Arroyo n.d.).
ing to the model presented above, into strong selection Other specialized ceramic industries evolved else-
in favor of coupled local specialization and exchange. where on the Guatemalan Pacific slope during the Clas-
Miriam Stark's (1991) survey of specialized pottery-mak- sic period. Of particular interest is the history of Flesh
ing communities described in the ethnographic litera- ware, which originates as Esperanza Flesh during the Early
ture bears out the prediction that they should be most Classic period. There are at least four compositional vari-
frequent in low-latitude regions. In Sober's (1984) terms, ants of Esperanza Flesh, some linked to ceramic resources
Stark's observations provide a warrant via "boot- on the coast and some linked to resources in the high-
strapping" for the part of the model depicted on the left lands (Neff et al. 1994). Flesh ware occurs archaeologically
side of Figure 5.1. Based on the theoretical arguments throughout the central highlands and on the coastal plain
outlined above and Stark's ethnographic data, we predict as far east as the province of Santa Rosa, near the Salva-
that the archaeological records for humid tropical regions doran border. Amatle ware, the Late Classic successor to
should show trends over time toward increasing local spe- Esperanza Flesh, was made in only a single compositional
cialization and regional economic interaction. variant, presumably within a much more restricted zone
than the earlier Esperanza. Apparently, then, Flesh ware
potters were exploiting opportunities presented by inter-
Pacific Coastal Guatemala
regional demand for their products, but over the course
One region where economic interaction has been of the Classic period, some potters (those utilizing the
examined in some detail via archaeological ceramic sourc- Amatle source) came to dominate this particular special-
ing research is Pacific coastal Guatemala. With a mean ized pottery production niche. As predicted by our model,
of 200+ cm of rainfall annually and a coefficient of varia- stable differentials in comparative advantage selected for
tion in rainfall of less than 15% in recent times (Mosino increasing local specialization, and this is reflected in the
Aleman & Garcia 1974, Vivo Escoto 1964), the region trend toward decreased compositional diversity in Flesh
epitomizes a predictable environment for agricultural ware during the Classic period.
production. Furthermore, the coastal plain itself receives
runoff from the volcanic mountain chain to the north, The Southern Basin of Mexico
where rainfall is even higher, so it seems unlikely that
severe drought or flooding would have posed a severe The history of ceramic production at the Aztec
threat to primary biotic productivity. Assuming that city of Chalco, in the southern Valley of Mexico (Hodge
moisture in this region also was plentiful and predictable n.d.) also documents the evolutionary effect of consis-
in the past, such things as soil type differences and drain- tent differentials in comparative advantage. The south-
age would have created consistent, long-term differen- ern Valley of Mexico consistently receives more rainfall
tials in agricultural productivity that would have selected than the north (Tolstoy 1975), and as a result, the south-
for both local specialization and regional exchange. In ern Valley was better able to produce an agricultural sur-
the absence of historical discontinuity, these features plus. This is reflected, for instance, in the fact that the
should become more prominent over time. Tepaneca attempted raids on Chalco around A.D. 1400
Ceramics first appeared on the southern Meso- in an attempt to acquire grain reserves (Chimalpahin 1965).
american coast shortly after 2000 B.C. The initial ce- Chalco was probably founded in the 6th or 7th cen-
ramic assemblages are very similar across a wide area, tury A.D., during the period of turmoil that surrounded
and, while there is some slight evidence for vessel move- the fall of Teotihuacan (Hodge n.d.). At first, differen-
ment, there is no indication of local specialization (Neff tials in comparative advantage among the various
& Arroyo 1996). On the central Guatemalan coast, the Epiclassic communities established around this time
Late Formative through Early Classic periods (500 B.C. - would not have been manifest. Not surprisingly, similar
500 A.D.) saw the emergence of a tendency for vessels to Red-on-Cream and Red-on-Buff vessels were produced
flow west to east, toward sites located on the well-drained, at Chalco and elsewhere in Central Mexico during the
extremely productive agricultural land of central Escuintla Epiclassic period (Neff & Hodge n.d.). As time passed,
(Neff 1995a; Neff, Bove & Arroyo n.d.). The trend to- differentials in returns from various productive activities
ward local specialization and regional exchange on the would have arisen gradually as a result of natural envi-
coastal plain culminates in the Late Classic period (A.D. ronmental differences (e.g., rainfall differences, differen-
700 - 900), when potters exploiting a distinctive, volca- tial access to ceramic resources, etc.) and chance histori-
nic ash-derived clay source located on poorly-drained land cal differences (e.g., differential proficiency of local pot-
in northwestern Escuintla supplied serving vessels to ters). At Chalco, a local serving vessel tradition contin-
coastal plain sites perhaps as far east as the Salvadoran ued into Early Aztec times, but, while other Black-on-
Methodology of Comparison in Evolutionary Archaeology 89
Orange production centers in the Basin of Mexico had Pueblo IV sites in the Grasshopper region demonstrates
begun to supply serving vessels through a growing re- that post-1300 assemblages contain both local and non-
gional economic system, Chalco supplied vessels prima- local Fourmile Polychrome. Migration, which may ac-
rily to a local, southern Basin market In Late Aztec times, count for some of the non-local White Mountain
Chalco was incorporated into the Aztec empire centered Redware in east-central Arizona (Tridan 1994), is a par-
at Tenochtitlan, an event that can be expected to have ticularly salient aspect of the later prehistory of the South-
amplified disparities in returns from different produc- west, especially after the Great Drought of the late 1200s
tive activities, thus favoring even more strongly the ac- (Larson et al. 1996).
tivities (e.g., agricultural production) for which local re- The case for vessel movement in the absence of
turns were relatively high. As would be expected under local specialization may be even stronger in areas to the
our model, Aztec Black-on-Orange production ceased at north of the Mogollon Rim. In the Mesa Verde region,
Chalco in Late Aztec times, and the city imported Black- Glowacki (1995, Glowacki et al. 1995) has found that
on-Orange serving vessels from other centers, such as Black-on-White pottery in Pueblo-III assemblages in-
Tenochtitlan and Texcoco. In sum, then, the 1000-year cludes locally made examples together with examples
history of serving vessel production at Chalco follows a derived from non-local production zones. On Black Mesa,
course that is highly predictable given the city's consis- Deutchman (1980) found several compositional groups
tently advantageous situation with respect to agricultural represented in Black-on-White assemblages and no sta-
production. tistical association between "typ e " and compositional
group. Based on "low-tech" analyses, Kojo (1996) con-
Evolution of Ceramic Diversity curs with Deutchman's conclusion regarding widespread
in Unpredictable Environments: the Southwest production of Tusayan White and Gray ware, but he sug-
gests that Deutchman's estimate of the incidence of ves-
Arid and temperate regions contrast with humid sel movement may be somewhat inflated.
tropical regions in that they are marked by high tempo- Our own analysis of a ceramic assemblage from a
ral variability in the availability of moisture, which trans- Pueblo-III site in south central Utah has revealed sur-
lates into high variability in primary biotic productivity. prising compositional diversity that shows only slight
Floods and, especially, droughts can dramatically reduce patterning with formal diversity (Larson et al. 1996, Neff
supplies of wild or domestic food supplies, as Larson et al. n.d.). A two-level structure is evident in the chemi-
(1994) has demonstrated for the historic period in Cali- cal data for 206 analyzed sherds, with two broad but very
fornia. High-resolution dendroclimatic records available distinct compositional groups represented and a total of
for several parts of the Southwest (Larson et al. 1996, seven compositional subgroups present within the two
Larson & Michaelson 1990) provide grounds for recog- main groups. All five decorative (ware) categories present
nizing specific periods of extreme subsistence risk in sev- in the assemblage are represented in both of the main
eral parts of this region. Selection for exchange in the compositional groups, and all are present in more than
absence of local specialization should be strongest dur- one of the subgroups of the main groups. There seems
ing periods of extreme climatic variability (Larson et al. little question that all decorative categories originate in
1996). more than one location. The best-sampled decorative
Numerous ceramic sourcing studies in the South- class, Black-on-White, is represented almost evenly in the
west have yielded evidence consistent with the view that two major compositional groups and occurs in all but
exchange or other mechanisms of vessel movement one of the seven subgroups.
evolved in the absence of local specialization. Several The patterning of compositional and decorative
provenance investigations have recently been undertaken diversity in the assemblage just described cannot possi-
in the Grasshopper region of east central Arizona. At bly have been produced by exchange coupled with local
Chodistaas, a small Pueblo-III site abandoned around specialization based on comparative advantage (Neff et
A.D. 1300, whole vessels of Corrugated, Plain, and Cibola al. n.d.). Although the pottery in this assemblage clearly
White ware excavated from room floors were found to originated in several distinct locations, there was a high
include local compositions along with several nonlocal degree of formal and decorative overlap in the output of
compositions (Zedeno 1994). In accord with our model, the spatially distinct workshops, and most or all decora-
Zedeno (1994:102) suggests that environmental margin- tive classes may have been produced locally. Since move-
ality favored maintenance of wide social networks, which ment of pottery cannot have been based on consider-
in turn produced accumulations of non-local ceramics ations of direct economic benefit, it must have been a
at sites such as Chodistaas. Triadan's (1994) study of byproduct of activities like small- and large-scale resi-
White Mountain Redware at Grasshopper and other dential mobility and non-economic reciprocal exchanges
90 Hector Neff * Daniel O. Larson
(gift-giving). As specified by the model presented here things contained in the archaeological record, and theo-
(Figure 5.1), such activities enhance individual fitness in retical (ideational) units in terms of which the record is
an unpredictable environment like the northern South- described and comparisons between different segments
west; incidentally, they move ceramics through space in of it are made. Another problem is unevaluated behav-
the absence of local specialization and regional economic ioral inference: while many statements about the archaeo-
interaction. logical record may involve inference, the requirement of
empirical sufficiency demands that inferential chains be
short and explicit Finally, again on the positive side, the
CONCLUSION continuous historical character of the archaeological
record ameliorates the problem of establishing indepen-
In order to persist, energetically expensive behav- dent origins of adaptive traits and extends the potential
iors must confer offsetting fitness benefits, i.e., they must for comparison beyond adaptive traits, to include the
be shaped and maintained by selection. If this is a valid direction and pace of the adaptive trends themselves.
assumption, then the signal of past selective processes We do not believe that comparative methods
must be strong in the material record accessible to ar- should be abandoned because of past associations with
chaeologists; every lithic flake, potsherd, and animal bone approaches now considered untenable. Instead, their util-
in an archaeological deposit records energy and time spent ity for evolutionary archaeology should be judged by
by someone making his/her way through life in the past. whether or not they improve our ability to understand
Although we can be quite confident that selection the archaeological record as a product of selective reten-
has shaped the material record of the human past, so far tion of cultural variation. The example presented in the
we have made little progress toward perceiving the na- final section of this paper provides one piece of evidence
ture of its effect. Part of the solution, as argued by O'Brien pertinent to such a judgment: we believe our compara-
and Holland (1992), is to adopt archaeological defini- tive observations establish the importance of environ-
tions of adaptation and adaptedness that parallel the defi- mental predictability as a source of selective pressures
nitions used in evolutionary biology: adaptations are traits that shape patterns of local specialization and movement
that were shaped by selection in a series of past environ- of materials through space; by so doing, they help estab-
ments, and adaptedness (or fitness) is the propensity lish the viability of comparative methods within evolu-
(which is assessed by design analysis) of a trait or set of tionary archaeology.
traits to increase in frequency. Attention to these defini-
tions can promote the formulation of plausible adaptive
ACKNOWLEDGMENTS
explanations. In archaeology as in biology, however, even
the most ingenious adaptive design arguments by them- We are extremely grateful to R. Lee Lyrnan and
selves represent wasted effort ("silly just-so stories") un- Michael J. O'Brien, both of whom read and commented
less explicit methods of hypothesis testing are brought extensively on an earlier draft of this paper. Their astute
to bear. Comparative methods, which are widely advo- criticisms helped us clarify our thinking on several key
cated as key to the testing of adaptive hypotheses in evo- issues. We hasten to add that we alone are responsible for
lutionary biology (e.g., Harvey & Pagel 1991, Maynard any remaining blunders or misinterpretations. We also
Smith 1994[1978], Mayr 1983; Sober 1984,1993), would thank C. Michael Barton and Geoffrey A. Clark for en-
also seem to hold considerable potential for testing adap- couraging us to contribute a paper to this volume.
tive hypotheses in archaeology.
The unique properties of the archaeological record
present both unique opportunities and special demands END NOTES
for archaeologists seeking to employ comparative meth-
1
ods. First, it must count as an opportunity that cultural The Panglossian paradigm appears also to have been recognized and
named earlier in anthropology than it was in evolutionary biology.
variation is abundant and easily generated, and adapta- Harris (1997:414) has noted recently that his own use of "Panglossian"
tion and convergence are therefore expected to arise com- to characterize naive functionalist arguments in anthropology dates
monly as the result of selective processes acting on cul- back to 1967 (Harris 1967), which is long before it was used by Gould
tural aspects of human phenotypes. On the other hand, and Lewontin in reference to adaptationist research in biology.