Problems like these, as well as variation in recovery methods ... The data were evaluated to deteTmine the relative ubiq- ...... Plenum Press, New York. 1993.
13 Of Prehistoric
Bison
Hunting
and Lesser Patterns
Mammals: in the Wyoming
Basin
PatrickM. Lubinski At the margins of the Great Basin and Great Plains in the western Wyoming Basin, there appears to have been neither a Plains-like bison-focal economy, nor a Desert Archaic-Iike subsistence system, based on an analysis of 93 radiocarbon dated faunal assemblages from Paleoindian to protohistoric periods. The predominant resources throughout prehistory were bison, rabbits, rodents, and pronghorn in terms of ubiquity, proportion of NISp, and NISP rank order. Other large game, particularly elk and bighorn sheep, are very rare in the sample, which is derived primarily from lower basin compliance excavations. Although rabbits and rodents are the most common taxa in regional archaeofaunas, the role of bison cannot be overemphasized during most periods, given their size, ubiquity, and dominance of aggregate NISP. Pronghorn occur commonly as a minor assemblage constituent, but dominate few assemblages, most in the final I ,500 years B.P.An initial test of climate change as an explanation for the pattern of bison and pronghorn ubiquity indicates that, while there is good correspondence with expectations for some periods, the archaeofaunal pattern is unlikely to be explained solely as a function of climate change.
INTRODUCTION The Wyoming Basin, a seriesof intermontane basins in the middle Rocky Mountains, can be thought of as a transition zone between the Great Plains and Great Basin. Just as it shares characteristics with the bordering shortgrass prairie and desert shrublands, the basin sharessome characteristics with both areas in terms of cultural developments. For example, the Eastern Shoshone of the mid 18ooswere noted both for Plains-oriented bison hunts and the use of rodents and insects (Shimkin 1986) more typical of Great Basin peoples. Prehistorically, might a Plains-like bison-focal model or a Basin-Iike "Desert Archaic" (Jennings 1957, 1964) model better characterize this transition area, were there shifts from one to the other, or were Wyoming Basin subsistencestrategies distinctly different from either? In order to address this question and begin to understand Wyoming Basin subsistence at the regional level, I completed a comprehensive synthesis of archaeofaunal taxonomic proportions data for the study area. The study area (Figure 13,1)is the western portion of the [176]
AP 122
Wyoming Basin physiographic province (Fenneman 1931), defined here as the Green River and Great Divide hydrographic basins in southwestern Wyoming, south to the Litde Snake River hydrographic basin in northwestern Colorado. This area, referred to here as the Green River Basin, is characterized by relatively flat basins punctuated by steep escarpments, with occasional dune fields and badland topography, and surrounded by mountains in the Wyoming, Gros Ventre, Wind River, Sierra Madre, and Uinta ranges. Vegetation consists of desert shrublands at the lowest elevations, with sagebrush steppe, foothill woodlands (juniper, aspen), montane forest (conifers), and alpine tundra zones at increasing elevation (Knight 1994).Elevation ranges from about 6,000 ft in Flaming Gorge to 13,804ft on Gannett Peak in the Wind River Range. The prehistory of this area is characterized almost exclusively by huntergatherer adaptations. This is not, of course, to imply that there was a static economic foundation. For example, it has been suggested that there were a variety of significant changes from the Archaic to Late Prehistoric periods, including a significant increase in the use of weedy annual plants (C. Smith 1988)and a switch from forager-Iike to collector-Iike subsistence-settIement strategies (Sanders et al. 1982).The present study was undertaken to document patterns of prehistoric prey selection through a study of archaeological faunal assemblages.
METHODS Included in this study were all faunal assemblagesthat I could find reported for the study area, with at least 10 specimens identified to the genus level or better, and at least one radiocarbon assay.Ten was chosen as the minimum sample size becausethe use of a larger minimum value was thought potentially to eliminate many of the regional archaeofaunas, which tend to be small. (If the study were limited to assemblages with 50 or more NISp' the aggregate NISP would not be significantly reduced, but the number of assemblageswould be cut nearly in half, alld the reduction of assemblagesize bias would be offset by a corresponding loss of security about the "representativeness" of the included assemblages.)The study included only radiocarbon-dated assemblagesin order to allow for the placement of assemblages into various chronological divisions (millennia, archaeological periods, climatic intervals, etc.), to allow for re-
Figure 13.1. The Green River Basin study area in Wyoming, Colorado, and Utah. Study area is indicated by thick line. Stippled areas are over 9,000 ft in elevation. The study area is drained by the Green River and its tributaries, except for the Great Divide Basin, which has no outlet. Base map from U.S. Coast and Geodetic Survey (1956),and hydrographic basins primarily from Roberts (1989).
duction to single-point mean ages,and to avoid the difficult and contentious processof assigning age ranges to artifacts. For each assemblage,the following were recorded as possible: NISp' MNI, site type, radiocarbon age(s), season, screen size, exca\'ated volume, location, elevation, and vegetation. This information was entered into a computer databaseprogram for analysis. In most cases,the analytical units (components, occupations, etc.), dates,and faunal tabulations presented by the investigators were used in my analysis with no modification. However, in some cases,the original tabulations were modified for entry into the data ba&,e,First, analytical units with overlapping 4C age ranges were combined when individual units were too small (48SU354, 48SW6324). Secondly, an analytical unit not mentioned in the original report was defined on the suggestion of the original investigator in one case(48LN7I7). Also, radiocarbon agesnot given in the original report were used if commonly accepted (48SW5, 48SWIOI). NISP and MNI information, when not given for a particular analytical unit, was determined as possible from textual descriptions of the faunal material. Several inconsistenciesexist in the databasethat have an affect on comparability between assemblages.MNI estimates were not reported consistently, and those reported were calculated with a variety of methods.Taxonomic identifications were occasionally made by assuming unidentified specimens were attributable to previously identified taxa, while most were rI:ladeon the basis of skeletal charac-
ters alone. Some tabulations included all analyzed specimens, while others omitted unburned elements from burrowing animals, and still others omitted rodents altogether. Problems like these, as well as variation in recovery methods, analyst experience and methods, and the coarse scale of temporal associations, may compromise the comparability of these assemblages,and thus the conclusions of this study. All remains identified by the analyst were included in my tabulations as possible, including those noted as "noncultural." This practice was employed in an attempt to control for inter-analyst variability in identification of intrusi\'e bone. While this practice \\~ certainly result in the inclusion of some remains that are not cultural, the fact is that many ground nesting or burrowing mammals (namely, cottontails,jackrabbits, ground squirrels, and pocket gophers) were used by aboriginal peoples in t11eIntermountain West (e.g.,C. Fowler 1986; Shirnkin 1947;Sutton 1994),and it is often difficult to distinguish between cultural and naturally deposited remains (Grayson 199IC). In the reports examined for this study, researchers dealt with this problem in a number of ways. One researcher tabulated all rabbits but listed all rodents only as "present." This clearly is a biased approach, as it assumesrabbits are food and rodents are not, an assumption that the site inhabitants were unlikely to m akS e. ome ana1ysts d Ismlsse .. d " noncu lltura "" or mo d ern " bone and pro\ided lists only of "cultural" bone. Burning, low surface weathering, lack of bleaching, element completeness, and context were used to distinguish cultural from noncultural bone. Although these are all reasonable criteria, it is unclear whether all large mammal bones were subject to the same criteria as the small mammal bones before being pronounced cultural. It appears that the system used by some analysts is unfairly biased towards including large game as food and dismissing small animals as intruslve. All radiocarbon ages used are uncalibrated and uncorrected (RCYBP). A mean age was detennined for each assemblage using the provided radiocarbon ages and Long and Rippeteau's (1974)method of averaging. When placing assemblagesinto chronological periods, all with one.;signla age ranges that spanned period boundaries were considered "multiple," and were not included in period tabulations. However, only mean age center points were used when placing assemblagesinto millennia. The data were evaluated to deteTmine the relative ubiquity and relative dominance of genera. Ubiquity was measured simply by calculating the percent of samples (assemblages) in which a genus occurred. This is a robust measure commonly used in palaeoethnobotany (Popper 1988).Relative dominance of different genera was determined using (1)percent contribution to aggl.egateNISP per period, (2) percent contribution to aggregate NISP per assemblage, and (3) the proportion of assemblagesfor which each genus was rank order #1 in terms of NISp, referred to here as AP
122
[177]
Table 13.1. Distribution of Archaeofaunas by Millennium RCYBP Millennium
Mean Age (RCYBp)
Assemblages
Total
Total MNI
NISP
48 O 23 14 66 23 64 589 131 958
:> 4 3 2
"rank ubiquity." (Note that the first method is very sensitive to aggregation effects, and tend to over-emphasize larger assemblagesin the aggregate.) For measuresof rank order, identical NISP values were counted as ties. MNI estimates were not consistently recorded for the assemblagesincluded in this study, so MNI was used only in a limited fashion in the discussion of taxonomic distribution. All of these measuresare biased by sample size. For example, with many assemblages,larger sampleshave an increased probability for rare taxa compared to smaller samples (Grayson 1984). Several different measurements and methods of combining samples were employed in an attempt to reduce the bias of sample size in the interpretations. It should be noted that an attempt was made to record excavated volume in order to standardize taxonomic abundance values by excavation intensity, as suggestedby Bozell (1995).However, this data could be obtained in only a small number of reports (n = 5), and is not used here. In any event, taxon density values are not a panacea for problems of comparability between sites, because differences in site surface stability, occupation duration, function, and disposal patterns affect the density of archaeological deposits.
of the Green River Basin subsistence-settlement system (Creasman and Thompson 1997;Madsen et al. Chapter 2; Sanders et al. 1982; Shimkin 1947).As to the chronological distribution, the majority G4 percent of assemblagesand 65 percent of NISp) have mean agesin the second millennium before present (Table 13.1).This reflects the preponderance of excavated and dated sites of that age in the study area. This uneven distribution must be considered when comparing results by time period. What do these data indicate about which taxa are most heavily exploited? Summary data are presented in Table 13.2.A variety of measuresindicate that the dominant genera throughout prehistory were Buon, Antilocapra, SpermophiLu,SyLvilagus,and Lepus (bison, pronghorn, ground squirrel, cottontail, andjackrabbit, respectively). These five taxa dominate in terms of (1) ubiquity, (2) aggregate NISP per period, (3) aggregate MNI per period, and (4) rank ubiquity (proportion of assemblagesin which the taxon was ranked first for NISP). The majority of the following discussion will focus on the distribution of these five primary genera. The relative importance of the five major genera
RESULTS
100%
Ninety-three assemblagesfrom 58 sites in the study area, with a combined sum of 21,552NISp, were employed for this analysis (Appendix 13.1).(Two of these sites are described elsewhere in this publication [Francis and Walker, Chapter 4; McKibbin, Chapter 11].)The assemblagesare not evenly distributed in spaceor in time. The vast majority are located in sagebrush steppe (90 percent of 90 assemblages) at 6,250-7,000 ft elevation (84 percent of 89 assemblages). This distribution reflects the fact that most of the reported faunal assemblageswere excavated for mineral development projects, wQi~h tend to be located in lower elevation basins. Due to this bias, the results discussed here should be considered valid only for the lower elevation basins, excluding mountains that were almost certainly part
80%
[178] AP 122
10
9
8
7 6 5 4 millennium (RCYBP)
fl] pronghorn. bison ~ gr. squirrel D cottontail
3
2
1
.jackrabbit
Figure 13.2. Ubiquity of the five major genera, by millennium.
Table 13.2. Summary of Taxonomic Distribution Regardless of Age Percent of all NISP
Percent
Genus
Common Name
Ubiquity
No. of Times Ranked 1-3
Percent NISP per High MNI
Assemblage High
Mean (when occur)
ARTIODACTYLS:
67 58
[8 49