Ecological Dynamics on Yellowstone's Northern Range (2002)

Mammal records from the late Pleistocene are sparse, but sites have been studied in several caves surrounding Yellowstone National Park (YNP). Usually only one species is found within one site, but these sites place the Greater Yellowstone ecosystem (GYE) in a regional perspective for the late Quaternary, especially for taxa that occur throughout the surrounding area.

Mammal faunas from these sites date from the latest Pleistocene (20,000 to 10,000 years before present [YBP]) through the Holocene (10,000 to 500 YBP). Because species responded individualistically to climate warming at the end of the Pleistocene, many of these communities do not have modern analogs. For example, most of the sites contain the remains of tundra species (pika [Ochotona princeps], collared lemming [Dicrostonyx spp.], caribou [Rangifer tarandus]) in association with forest mammals (porcupine [Erethizon dorsatum], marten [Martes americana], deer [Odocoileus spp.]) and plains-dwelling forms (bison, ground squirrels, pronghorn). A heterogeneous parkland/savanna is suggested as the best environment for supporting these types of diverse faunas (Faunmap Working Group 1996).

Extinct ungulates that appear to have inhabited the GYE include the Columbian mammoth (Mammuthus columbi), horse (probably several species of Equus), camel (Camelops hesternus), woodland muskox (Bootheerium

bombifrons=Symbos cavifrons), and mountain deer (Navahoceros fricki) (Faunmap Working Group 1996). In addition, several large extinct carnivores—the short-faced bear (Arctodus simus), American lion (Panthera atrox), and American cheetah (Miracinonyx trumani)—were common to this area (Martin and Gilbert 1978, Chomko and Gilbert 1987, Walker 1987). Other ungulates (e.g., flat-headed peccary [Platygonus compressus]) and carnivores (e.g., saber-tooth cat [Smilodon floridanus]) may also have been present, but their remains are sparse and not easily extrapolated to the GYE.

These large mammals as well as at least 24 other genera became extinct in North America at the end of the Pleistocene (11,000 YBP). The two main hypotheses about this extinction involve overexploitation by human hunters and climatically driven environmental changes (Martin and Klein 1984). Whatever the cause of the extinction, it must have had broad ramifications, including alterations in biological interactions such as predation and competition; vegetational structure and composition created by seed dispersal, browsing, and grazing; and nutrient recycling. The disappearance of species assemblages for which there is no present-day representative is coincident with the extinction event (Graham and Lundelius 1984).

As the climate began to warm about 14,000 years ago and glaciers receded northward and higher, so did some of the boreal mammalian fauna. The collared lemming that today lives on the tundra in Alaska and Canada was one of the first mammals to be extirpated from the surrounding areas and probably the GYE, having vacated the contiguous northwestern United States by at least 10,000 YBP (Faunmap Working Group 1994). Other boreal species like the pika (O.princeps) and the heather vole (Phenacomys intermedium) remained at elevations lower than their current distribution until the middle Holocene, a time of maximum warmth and dryness (Grayson 1977, 1981; Mead et al. 1982; Mead 1987). As these species dispersed to higher elevations, species like the pygmy rabbit (Oryctolagus idahoensis) decreased in abundance and others, which were adapted to drier habitats (e.g., Lepus spp.), first appeared or increased in abundance (Grayson 1987). Little is known of the Holocene history of the Great Basin ungulates (Grayson 1982, 1993).

Similar changes took place in Wyoming east of Yellowstone as summarized by Walker (1987). At 10,300 to 9,300 YBP, some boreal mammal species were still present in the basin areas, but steppe forms were starting to be found in association with boreal habitat types. However, by 5,060 to 2,760 YBP, modern mammalian distributions are believed to have been established.

Lamar Cave has yielded an extensive fossil mammal record from the

Lamar Valley for the past 3,200 years (Hadly 1996). The late Holocene fauna of the Lamar Cave is nearly identical to the modern fauna of the area. Only one species, the prairie vole (Microtus ochrogaster), does not occur in the modern fauna but it is found today in environments 100 km north of that site (Barnosky 1994, Hadly 1996). The sporadic presence of this species in the cave record may indicate the occurrence of more tall grass habitat in the vicinity of the cave at certain periods in the past.

Although general faunal composition has been relatively stable, relative frequencies of environmentally sensitive species fluctuated through time and appeared to correspond with climatic events in the late Holocene. High ratios of voles (Microtus spp.) and ground squirrels (Spermophilus spp.) indicate moister environments. Based on such analyses, Hadly (1996) reconstructed the environmental sequence as follows: from 2,860 to 1,370 YBP the environment was moister than it is today; from 1,500 to 560 YBP the environment became drier; cooler and moister conditions have prevailed for the last 700 years, including the Little Ice Age.

Pocket gophers (Thomomys talpoides) were most abundant at Lamar Cave during times with moister climates (Hadly-Barnosky 1994, Hadly 1996). Pocket gopher-body size also appears to have responded to climate change (Hadly 1997).

In other areas of the United States, similar body-size changes have been demonstrated (Purdue 1989) for ungulates such as white-tailed deer (Odocoileus virginianus) where body size decreased during the warmest and driest climates of the middle Holocene. Bison also became steadily smaller from the Pleistocene through the Holocene (Kurten and Anderson 1980). Although similar studies have not been conducted for the GYE, GYE ungulates probably responded similarly to middle Holocene climates.

Hadly (1996) concluded that the bones accumulated in Lamar Cave were the result of the activities of owls, small to medium-sized carnivores, and woodrats. Therefore, the faunal record from Lamar Cave is dominated by small mammals (Hadly-Barnosky 1994). Large ungulate remains (elk, deer, bison, and bighorn sheep) occur in the cave deposits but their frequencies probably do not reflect the actual abundance of those animals. Other archaeological sites in YNP (Cannon 1997) have yielded the remains of large ungulates. They document the occurrence of elk, deer, bighorn sheep, and bison in the park by about 1,200 years ago; however, they cannot be used to estimate population sizes.

Some scholars have proposed that the environments of the middle Holo-

cene caused reductions in the size of Bison herds (Dillehay 1974, Frison 1978). These assumptions were based on the reduced number of Bison sites and their apparent absence in areas that supported Bison in the late and early Holocene. However, this hypothesis has not been quantitatively assessed. The variables involved must be evaluated before estimates of relative abundance over time (e.g., Cannon [1992]) can be evaluated quantitatively.

Factors that influence which species are likely to be preserved in particular sites are listed in Table A-1. For example, caves are frequently used as dens by predators, and consequently, the remains of predators and their prey are often preserved. At open sites, like a water hole, the remains of ungulates are far more common than the remains of predators. Also, an assemblage of bones from a pit cave, which acts as a natural trap, is quite different from that of a cave with a horizontal entrance, which permits both entrance and egress (Brain 1981).

The behavior of animals is also important in the formation of bone accumulations. Some animals may be preferentially attracted to different site types. Bats use caves as hibernacula; their remains often are very abundant in cave deposits. Predator-prey relationships may also be a critical factor in determining the composition of a bone assemblage. Owl roosts may have massive accumulations of small mammal bones but do not contain any large mammal remains. Conversely, a wolf den would primarily consist of ungulate remains.

The location of a site and its catchment area also are important. The catchment is the area from which fossils can be derived. For rodents and insectivores falling into a pit cave, the catchment may be only a few meters up to perhaps a few hundred meters across. However, the catchment of a large river system can be hundreds of square kilometers. Sites located in upland (plateaus, interfluves, etc.) and bottom land (floodplains) environments will frequently have different types of faunal assemblages. Owls may forage over a few kilometers, whereas large mammal carnivores may bring prey from tens to more than 100 kilometers. Also, distance between the site of accumulation and the site of the kill can influence what types of bones, if any, are brought to the site. This situation, which is especially important in archaeological bone assemblages, has been referred to as the “schlep effect.” If an animal is killed near the site, the predator may bring the entire carcass back to the site of

accumulation. On the other hand, if the kill is made some distance from the site of accumulation, the predator may strip the carcass of meat and not return any bones to the site of accumulation or may selectively pick bones of different nutritional value (i.e., marrow content).

Time averaging can inflate bone counts. Bones may accumulate on a surface over an extended period of time. If bones accumulate at the same rate but with greater rates of sedimentation, then bone density will be lower. The only way to correct for time averaging is to have a series of radiocarbon dates that allow sedimentation rates to be calculated.

Differential preservation of bone can also alter counts. In sediments deposited in fast currents, small and light bones can be swept away, leaving the larger and denser bones. Also, in sites in which bones are trampled or crushed by other means, foot bones are preferentially preserved, and more fragile bones like skulls and scapulas are destroyed. In an archaeological site where marrow was processed, limb bones can be reduced to small unidentifiable splinters. Kay (1990) describes these types of assemblages in his samples. Finally, bones deposited in alkaline environments like cave sediments may be preserved, but bones in acid soils (e.g., some floodplains) may dissolve over time.

The ability to identify skeletal elements to species or even generic levels varies with the taxon and the experience of the faunal analyst. Most of the skeletal elements of the ungulates should be readily identifiable, but there is a

great deal of variability in the amount of bone that is identified from various sites. Also, the amount of bone from a site varies with the amount of the site excavated. Excavations to determine the potential of a site generally are limited to a relatively small volume. On other occasions, hundreds of cubic meters of a site may be removed.

Methods of collection can also bias bone samples. Semken and Graham (1996) used ordination techniques to show how species composition of archaeological sites can vary on the basis of whether the site was screened or the size of the mesh used in the screening. Sites that are not screened generally lack the bones of smaller animals and smaller bone fragments of larger species.

In most cases, bone accumulations result from multiple pathways. Also, a single site may have different stratigraphic levels that accumulated bones by different pathways. Therefore, if these levels are combined, the different pathways are mixed. To have meaningful comparisons, it is essential to compare assemblages (sites or levels within sites) with similar histories of accumulation.

A series of late Quaternary sites from the Pryor Mountains of Montana clearly illustrate the problems of various accumulation histories and how they can influence faunal samples and bone counts. The Pryor Mountains are composed of two fault-lifted blocks (East Pryor and Big Pryor) of Paleozoic sedimentary rocks. Caves are developed in the Madison Limestone on both East and Big Pryor. Two cave sites have been excavated on East Pryor that are only 300 m apart. Therefore, the caves have sampled the same environment. Also, both caves contain relatively complete sequences of Holocene deposits so they have both sampled the same time interval. The primary differences between the caves are site type and agents of bone accumulation.

One site is False Cougar Cave, which is located on East Pryor Mountain at about 8,600 ft (2,621 m). This is a small cave developed in a small outcrop of Madison Limestone. It has a horizontal entrance and contains multiple sedimentary layers that date from the late Holocene to the late Pleistocene. The cave was also used by humans, as evidenced by stone artifacts and hearths found in the cave sediments. Vegetation around the cave is a mixture of open meadow and coniferous forest.

The other cave, Shield Trap, is about 300 m west of False Cougar Cave. It is a pit cave with a vertical shaft of about 10m. The opening to the cave is about 3 m in diameter. The cave is situated on a relatively flat ridge surrounded by open vegetation, primarily grasses. It is developed in the same limestone as False Cougar Cave.

Both caves have been excavated by the same method of 10-cm levels within natural stratigraphic units. Large specimens were piece-plotted with respect to their horizontal and vertical coordinates. In addition, the orientation and plunge of bones with long axes were measured with a Brunton compass. All sediments were collected and water screened through fine mesh (1/16-inch [0.16-cm] mesh) screens. Also, both caves contain a relatively complete sequence of Holocene sediments. Depositional environments are similar, but Shield Trap generally has more breakdown.

Therefore, the primary difference between the two sites is how each cave sampled the living fauna. Shield Trap was a pit into which animals randomly fell and were trapped. On the other hand, the horizontal entrance of False Cougar Cave did not serve as a trap. Instead, faunal remains were brought into the cave by agents, primarily predators (owls, carnivores, and humans). In some cases, bones may have accumulated as the result of an animal’s dying in the cave as it was used for shelter, but these remains are an extremely minor component of the fauna.

The main difference between the bone assemblages from the two caves is in the abundance of large mammal remains. The assemblage from Shield Trap is primarily composed of bison bones, whereas False Cougar Cave contains small mammal remains (rabbit to marmot size and smaller). Shield Trap also contains small mammal remains, but they are not nearly as abundant as those from False Cougar Cave. The remains of large mammals are sparse in False Cougar Cave. Therefore, if the bone frequencies are taken at face value, Shield Trap suggests that bison were abundant throughout the Holocene. However, False Cougar Cave, which has sampled the same environment at the same time, suggests that bison, specifically, and ungulates in general were rare. Typical of most pit caves, Shield Trap also has a more complete complement of carnivores than does False Cougar Cave.

Big Lip is a rockshelter located at a lower elevation of 6,000 to 7,000 ft (1,829 to 2,134 m) on East Pryor Mountain. It contains sediments similar to those of both Shield Trap and False Cougar Cave, and it is located less than 10 miles (16 km) from these other two sites. The vegetation around the cave consists of Douglas-fir and lodgepole-pine forest. Big Lip also contains human artifacts of middle Holocene age. The primary difference between Big Lip and the other two sites is its lower elevation and vegetational surroundings. However, the fauna from Big Lip is quite different from either Shield Trap or False Cougar Cave. Big Lip rockshelter does not contain abundant microfauna, and the dominant ungulate is Ovis canadensis. Based on the artifact

assemblage and bone breakage patterns, it appears that Big Lip served as a hunting and butchering site for bighorn sheep. Again, Big Lip would provide a very different picture of ungulate abundance if faunal remains were interpreted at face value without considering accumulation pathways.

These problems apply to the sites analyzed and interpreted by Kay (1990). For the Myers-Hindman site, Kay (1990) lumped faunal remains from seven cultural levels and eight settlement units dating from 9,000 to 800 YBP and calculated total Minimum Number of Individuals (MNI) for the site. He then compared taxa. This technique averages and smears any fluctuations in abundance due to environmental fluctuations throughout the entire Holocene. Also, the percentage comparisons between taxa are not independent. Kay (1990) concluded that the proportions of ungulates did not correspond to today’s relative abundance of ungulate species in the vicinity of the site.

For Mummy Cave, Kay (1990) again lumped MNI and Number of Individual Specimens (NISP) values for 38 distinct layers that date between 9,000 and 300 YBP. Also, he noted that the faunal remains from the site have never been completely identified. As indicated by Kay (1990), the materials reported by Harris (1978) represent only a sample of the entire site. Again, by comparing the lumped MNI values for the entire site for the various ungulates, Kay (1990) concluded that these proportions were quite different from the proportions of the ungulate species in modern populations.

For the Dead Indian Creek site, Kay did not explicitly explain how he treated the sample, but again it appears that he calculated MNIs for the various ungulate species for the entire site and then compared proportions. These proportions of ungulates did not match those of modern populations.

The Bugas-Holding site is a single-component and probably single-occupation site, which is an ideal sample. However, again the proportions of ungulates in the faunal sample did not match those of the modern populations. However, Kay (1990) did not consider factors other than actual abundances that could have caused this difference. The sample is dominated by bison. One possibility for this bias is in the methods of procurement of bison and elk. Frison (1974) notes that there are many close parallels between handling bison and domestic cattle. He further suggests that there may be a critical size for a manageable herd that would then result in the slaughter of more animals (Frison 1974). In comparison, there is no known evidence for artificial elk traps or communal procurement as with bison (Frison 1978). However, if the mature female leader is killed the remainder of the elk herd often mill in circles, and the entire herd, or a good portion of it, can be killed. On the other

hand, the herd can disperse for several miles and be hard to find (Frison 1978). These differences, as well as others, could easily account for the faunal differences at Bugas-Holding and hence the abundance of remains at the site is not necessarily related to the abundance of the animals.

The Joe Miller site (48AB18) is the only archaeological excavation in Wyoming where elk are the most common ungulate (Kay 1990). Creasman et al. (1982) concluded that the upper component of the Joe Miller site represents an area for processing elk.

Kay has shown that elk bones are generally not as abundant as the bones of other ungulates such as bison, deer, sheep, and pronghorn. The paucity of elk bones could reflect low population levels as postulated by Kay (1990) and others (Keigley and Wagner 1998), but it is just as likely, and perhaps more probable, that differences in the abundances of bones are an artifact of processes by which bones were accumulated. Also, lumping of quantitative data from different stratigraphic levels, as done by Kay (1990), has created samples time-averaged over as much as 10,000 years, virtually the span of the entire Holocene. Also, grouping of data, as done by Kay (1990) by combining different stratigraphic units within sites, probably mixes various accumulation pathways for these stratigraphic levels, which again would significantly bias the frequency distributions.