Keywords

1 Introduction

The prey choice model (also known as the diet breadth model) grounded in optimal foraging theory has proven to be a valuable tool for formulating predictions regarding animal exploitation across space and time, especially among small scale societies. The underlying logic of the model is quite simple: individual hunters make decisions when foraging that favor resources that provide the highest net return based on a currency. Such high ranked resources should always be pursued when encountered by the hunter, while the inclusion of lower ranked resources into the diet is dependent on the encounter rates of higher net return resources. Since high ranked prey are typically large bodied animals with low recruitment rates (Broughton et al. 2011), the long-term effect is resource depression when predation exceeds reproduction and in-migration. As encounter rates for higher ranked resources decline, foragers are expected to incorporate greater numbers of lower ranked prey items into their diet.

The artiodactyl index (AI ) is commonly used for testing the predictions of the prey choice model in western North America due to its simple computation and intuitive meaning. As originally conceived (Bayham 1979), the AI is a ratio-based measure that compares the relative abundance of artiodactyls (e.g., deer, bighorn sheep, pronghorn) to leporids (e.g., jackrabbits, cottontail rabbits) to measure the trade-offs predicted by the prey choice model within a typical terrestrial patch. AI values are expected to decline over time if large game populations were depressed, and such changes may be tested using chi-square analysis and similar operations (Cannon 2001). The measure has since been modified into a variety of abundance indices that compare any set of high ranked and low ranked resources within a patch (e.g., Broughton 1994).

While abundance indices are simple to use, researchers must carefully review the data being inputted in such measures. For example, the practice of aggregating faunal data from multiple assemblages masks fundamental differences among the assemblages related to variation in habitats, seasonality, site function, settlement patterns, recovery strategies, and analytical decisions (e.g., Fisher 2015; Lyman 2003). Here, I focus on one particular kind of variation: differing taphonomic trajectories that may impact the relative taxonomic abundances. Researchers most commonly use the total number of identified specimens (NISP) of each set of taxa when calculating the abundance index, but NISP values may be affected by differences in selective transportation, butchering and processing methods, and post-depositional attrition (Lyman 2008, pp. 29–30).

Taphonomic forces begin with an animal’s death and include decisions made by the hunter about which portions of the carcass to transport. Transportation decisions depend on a wide range of situational variables, such as the size of the animal, the number of individuals in a hunting party, and the distance to the residential base (e.g., Bartram 1993; Lupo 2006; Metcalfe and Jones 1988; O’Connell et al. 1990). As such, a single artiodactyl that has undergone some field processing will be represented by fewer skeletal parts than smaller game at the residential site. All else being equal, the AI is expected to decrease as large game are transported from greater distances even if the actual number of individual animals captured has remained the same. Subsequently, skeletal parts are fragmented through cultural and natural processes that affect the NISP and abundance measures (Cannon 2013), adding a second layer of complexity. Moderate fragmentation rates may result in an increase in NISP if each resulting portion can still be identified to taxon and element, while heavier fragmentation will lead to a decrease in NISP as fragments can no longer be identified (Grayson and Delpech 1998). For example, fracturing long bones to access marrow from the medullary cavity may increase the NISP, while heavy grease extraction activities that require pot-sized portions may result in a decrease in NISP.

Further fragmentation and deletion of skeletal parts may occur after discard through a number of post-depositional processes. Bone waste that is left on the surface is more likely to suffer from weathering (Behrensmeyer 1978; Phoca-Cosmetatou 2005), trampling (Behrensmeyer et al. 1986; Olsen and Shipman 1988), and scavenging (e.g., Marean and Spencer 1991; Marean et al. 1992; Munson and Garniewicz 2003) processes that systematically damage bone. Bone survivorship may be reviewed by comparing volume density values for specific skeletal portions against the frequency in which these portions are present in an archaeological assemblage (see Lam et al. 2003; Lyman 1994 for review). Density-mediated destruction studies have largely centered on disentangling the equifinality between bone survivorship and cultural practices , such as selective transportation and access to animal resources (Grayson 1989; Lam and Pearson 2005; Lyman 1985, 1994; Marean et al. 1992; see also Faith and Thompson, Chap. 6). This is reflected by the greater number of studies on bone density of large bodied mammals (Brain 1981; Elkin 1995; Kreutzer 1992; Lam et al. 1998; Lyman 1984, 1985; Stahl 1999; Symmons 2005) compared to those on small mammals (Lyman et al. 1992; Pavao and Stahl 1999).

Importantly, cultural practices of converting raw animal resources into consumable products may have direct consequences on the subsequent preservation of skeletal parts and thus NISP values. Scavengers systematically damage and delete skeletal parts from the assemblage in ways that correspond to the nutrition value and bone densities of the portion (e.g., Marean and Spencer 1991; Marean et al. 1992; Munson and Garniewicz 2003). Bone grease is frequently located in the portions of bone that are the least dense (e.g., cancellous bone of long bone epiphyses and the axial skeleton), and scavengers are more likely to destroy these parts in proportion of their fat content if it was not extracted prior to discard (Hudson 1993; Kent 1993; Lupo 1995; Lupo and Schmitt 1997; Speth 2000; Ugan 2005, 2010). As such, the attractiveness of discarded bone debris to scavengers is partly dependent on how animal resources were processed, which in turn is expected to vary according to the nutritional content of meat and skeletal components of the animal resource, the degree of nutritional stress for both the prey and consumer, and other factors (Church and Lyman 2003; Fisher and Johnson 2014; Outram 2002; Ugan 2005; Wandsnider 1997). For example, Speth (2000) applies this logic, comparing skeletal part representation with bone density, marrow index, and grease index values, to argue that jackrabbit and cottontail rabbits at the Henderson Site in New Mexico were mostly likely stewed, while prairie dogs (Cynomys ludovicianus) were roasted.

Notably, processing methods for a single resource may vary across time according to the overall foraging efficiency of the consumer. The marrow and grease component of the skeleton is a critical resource to foragers during times of resource stress (Speth and Spielmann 1983), and the extent of fragmentation and bone survivorship may be a function of dietary stress (Outram 2002). Marrow may be accessed with relative ease by breaking open bones to access the medullary cavity, but foragers may ignore low yielding parts (e.g., phalanges) during more plentiful times. Grease rendering has comparatively high extractive costs (Binford 1978, Lupo and Schmitt 1997, Outram 2002), although less intensive processes such as stewing may sufficiently extract grease as well. Outram (2002) uses the marginal value theorem to show that foragers should continue to process a carcass despite declining marginal net returns when the costs of finding, killing, and transporting a second resource are high. Intensified processing is expected when encounter rates with high ranking resources are low due to resource depression caused by overpredation or less favorable climatic conditions. Thus, the NISP of the high ranked resource may increase due to greater processing when in fact hunting efficiency has actually decreased and greater numbers of lower ranked resources are being acquired.

Ugan (2005, 2010) previously evaluated the relationship between culinary processing, density-mediated destruction, and taxonomic abundances using Parowan Valley assemblages from the eastern Great Basin. He found that density-mediated destruction varied among the artiodactyl assemblages from occupational units; when artiodactyl remains were abundant, there was lower skeletal part diversity and a strong statistical relationship between bone density and skeletal part representation (2005, p. 238). Just as Outram (2002) predicts using the marginal value theorem , Ugan hypothesized that during drier climatic periods, lower overall return rates for artiodactyls would have led to more intensive grease processing, which in turn would have promoted higher bone survivorship as such discarded remains are less attractive to carnivores. During more favorable climatic conditions, hunting success of such large game would have increased due to an increase in encounter rates corresponding with higher game population densities; consequently, less intensive processing during plentiful times would have resulted in a decreased representation of artiodactyl remains as scavengers ravage discarded remains with nutritional content. Thus, the abundance index values were ultimately a reflection of climate rather than encounter rates with high ranking taxa, the decline of which is frequently taken to reflect resource intensification related to increases in human population densities. Fisher and Johnson (2014) similarly found high preservation levels for leporids remains at Antelope Cave, a Virgin Ancestral Pueblo site, despite evidence that domesticate dogs were present. They attribute the high level of preservation to intensive processing and limited access to higher ranked prey in a marginal arid environment.

Bone survivorship is expected to vary across contemporaneous deposits within a site due to variation in foraging efficiency differences among households, as well as differences in disposal. Bone that is rapidly buried is less likely to be ravaged by scavengers or exposed to weathering and other destructive forces. These forces may result in certain site contexts having strongly positive relationships between volume density and element representation, such as abandoned use surfaces that are left exposed, compared to contexts formed through the rapid accumulation of bone waste , such as in designated trash disposal area. Similarly, rapid accumulation of faunal remains due to feasting or hunting strategies (e.g., large communal hunts of pronghorn or jackrabbits) may result in greater survivorship but are not a reflection of day-to-day foraging activities.

In summary, the rate of bone attrition is expected to vary among taxa and assemblages in ways that influence our ability to evaluate the predictions of the prey choice model. If large game become less abundant on the landscape due to overhunting or less favorable environmental conditions, the prey choice model predicts that diet breadth should incorporate greater quantities of higher cost and typically smaller game resources. Yet, large game that are successfully captured are expected to be processed to a greater extent, which may counterintuitively lead to increased fragmentation and increased survivorship, both of which may result in an increase in NISP for large game when in fact fewer individuals were captured. The effects of differential survivorship may be investigated by identifying the variance in attrition among different animal resources and correlating this with an abundance index. Here, I draw upon data from a single archaeological site to illustrate such variation in density-mediated destruction among artiodactyl and leporid taxa across various contexts.

2 Five Finger Ridge Site Background

Five Finger Ridge is a large Fremont Period village site located in Clear Creek Canyon of central Utah (Fig. 7.1). The site was occupied at the end of the Fremont Period (ca. AD 400–1300), an archaeological culture of the eastern Great Basin and northern Colorado Plateau. This period correlates with increased summer temperatures and moisture that allowed for maize horticulture and the appearance of bison (Bison bison) in the region (Grayson 2006; Madsen 1989; Madsen and Simms 1998; Rhode 2000; Talbot and Wilde 1989; Wigand and Rhode 2002). When summer monsoonal precipitation weakened ca. AD 1300 archaeological traces of bison, maize, and the Fremont disappear.

Fig. 7.1
figure 1

Map of United States with location of Five Finger Ridge site (star) in the state of Utah

Full size image

The site was excavated in 1984 by the Office of Public Archaeology at Brigham Young University in preparation for the construction of an interstate highway (Janetski et al. 2000; Talbot et al. 1998). Backhoes were initially used to define features, followed by hand excavations using a 1 × 1 m grid. Structure fill was excavated to within 20 cm of the floor as a single unit. The upper 10 cm of the lower fill was sifted through 1/4 inch (6.4 mm) screens, switching to 1/8 inch (3.2 mm) screens for the lower 10 cm of fill immediately above floor. All floor sediments, subfloor pits, hearths, and other features were screened using 1/8 inch sieves.

Excavated structures consist of subterranean, subrectangular pithouses (n = 37), circular to oval subterranean secondary pit structures (n = 23), rectangular surface structures (n = 19), a single square surface structure, and a single jacal (thatched wattle-and-daub) surface structure. Activity areas include use surfaces (n = 21), borrow areas where earth has been moved to construct other features (n = 7), and open features (n = 6). Pithouse structures vary in size from 5.5 to 31.6 m2 with a mean of 12.9 m2; however, the largest structure is an outlier and may be associated with village leaders. Secondary pit structures vary from 1.2 to 9.9 m2 with a mean of 3.9 m2, and rectangular storage structures range from 3.6 to 11.5 m2 with a mean of 6.8 m2 (Talbot et al. 2000). Feature floor deposits were assigned to temporal period based on a battery of dating methods, including radiocarbon, dendrochronology, archaeomagnetic dates, and obsidian hydration (Talbot et al. 2000). Three temporal periods were established: Period 1 (dates older than AD 1200), Period 2 (AD 1200–1300), and Period 3 (younger than AD 1300). Five Finger Ridge was most intensively occupied during Period 2, which is further subdivided into Period 2A (AD 1200–1250) and Period 2B (AD 1250–1300).

A greater abundance of faunal remains from pithouse floors compared to the other structures reflects heavier consumption and food preparation within these areas (Talbot et al. 2000). Variation in the quantity of bone on structure floors may represent differences in floor maintenance or seasonality of abandonment (Talbot and Janetski 2000). Food preparation appears to have been focused in the front of the pithouse and the area surrounding a central hearth (Talbot and Janetski 2000). Further, the distribution of faunal remains differs significantly between the floor and lower fill contexts, indicating that they were deposited as separate events (Talbot and Janetski 2000). As seen in Table 7.1, approximately 90% of the total NISP consists of unassigned medium-bodied artiodactyls, deer (Odocoileus hemionus), bighorn sheep (Ovis canadensis), unassigned leporids, jackrabbits (Lepus sp.), and two species of cottontail rabbits (Sylvilagus audubonii and S. nuttallii). Also noteworthy is the presence of canid remains, as domestic dogs and coyotes likely would have been attracted to disposed animal remains.

Table 7.1 Mammalian fauna identified at Five Finger Ridge
Full size table

3 Methods

Measures of density-mediated destruction, abundance indices, and related analyses were computed among structural and activity area contexts, as well as across temporal units. Density-mediated destruction was evaluated for the artiodactyl taxa and cottontail rabbits . These taxa were chosen based on their relatively high abundances at the site, the availability of bone density data, and their frequent use in constructing abundance indices for evaluating resource intensification in western North America (e.g., Broughton 1994; Janetski 1997; Szuter and Bayham 1989; Ugan 2005). Medium-bodied artiodactyls (deer, sheep, and pronghorn) were aggregated into a single category since some elements cannot be identified to species and intertaxonomic variability in density values between artiodactyl taxa appear to be relatively minimal (Lam et al. 1999).

The density values for sheep provided by Lyman (1984) are used for all artiodactyl skeletal parts with the exception of selected long bone scan sites; I use the corrected values for limb bones provided by Lam et al. (1998). Since some skeletal parts of sheep were not included in these two analyses, values for deer (Lyman 1984) are used to supplement the missing values. Bone tool fragments, tool manufacturing debris, and neonatal specimens were removed from analysis. Neonatal (late fetal or newborn individuals less than a few weeks of age) artiodactyl remains were recovered frequently (NISP = 437). Juvenile skeletal parts have bone density values that are typically lower than adults, and high density portions are not distributed across the skeleton in the same rank order (Symmons 2005). Values for Sylvilagus floridanus (Pavao and Stahl 1999) are used as a proxy for S. audubonii and S. nuttallii. The caudal vertebrae, sternebrae, ribs, astragalus, metapodials, and phalanges of Sylvilagus were removed from this analysis to control for potential screening biases resulting from the use of 1/4 inch screens (Shaffer 1992; Shaffer and Sanchez 1994). Recovery of adult artiodactyls is not expected to be significantly impacted by screen-size recovery biases since small fragments (<1/4 inch) of large mammal bone cannot generally be identified to taxon and skeletal part.

Standardized number of identified specimens (NNISP) is used as the measure of skeletal abundance for comparison against density values. This measure is computed by dividing the number of identified specimens (NISP) containing a particular scan site on a skeletal part by the number of times the element is represented in a body (e.g., the NISP of paired appendicular elements and the mandible are divided by two; proximal, middle, and distal phalanges by eight, lumbar vertebrae by seven, etc.). NNISP has been shown to be a strong predictor of the minimum number of elements (MNE) measure due to statistical sampling (Grayson and Frey 2004). This relationship is present for both Sylvilagus and the artiodactyl taxa from the Five Finger Ridge assemblage (Odocoileus r 2 = 0.79, p < 0.001; Ovis r 2 = 0.88, p < 0.001; Sylvilagus r 2 = 0.95, p < 0.001). With the significantly positive relationship between MNE and NNISP, density-mediated destruction is evaluated using NNISP values instead of the derived MNE measure.

Null NNISP values are removed from analysis as it is unknown whether they represent real absences in the original population or are the result of destructive forces (Lam and Pearson 2004). Including the null values assumes that the faunal assemblage originally contained the complete skeleton, which is especially unlikely to be the case for large game that was selectively transported to reduce travel costs from the kill locality. The absence of high density skeletal parts would be suggestive of selective transportation, especially when such parts have low economic utility (e.g., tarsals and carpals). However, the absence of low density skeletal parts, such as much of the axial skeleton, could be due to subtraction associated with post-depositional forces or disposal at a primary processing site to reduce travel costs. It is noted that density-mediated destruction analyses were conducted with the null values, but the more conservative approach of removing the null values is presented here since the results are not substantially different. When evaluating variability in attrition among site contexts, I restricted the analysis to proveniences with artiodactyl sample sizes (NISP) of greater than 30.

In evaluating the impact of differential survivorship on relative taxonomic abundances, I use an abundance index (AI) formed by dividing the frequency of artiodactyls with the total frequency of artiodactyls and Sylvilagus:

$$ AI=\frac{\varSigma NIS{P}_{Artiodactyl}}{\varSigma NIS{P}_{Artiodactyl}+\varSigma NIS{P}_{Sylvilagus}} $$

I select this measure because it has become standard in addressing issues related to resource depression and diet breadth (e.g., Bird and O’Connell 2006; Broughton 2002; Lupo 2007), including those in the Fremont area (e.g., Janetski 1997; Janetski et al. 2000; Ugan 2005). Neonatal specimens are not calculated into this measure since such young individuals represent sessile resources with low pursuit costs and lower caloric returns. As discussed above, NISP may be impacted by a number of factors, such as differential survivorship, selective transportation, the number of skeletal parts among species (Lyman 2008), and there are reasons to believe that these influences are not equally distributed across taxa. Using the minimum number of individuals (MNI) when computing the abundance index may circumvent these issues since each individual should be represented by dense skeletal portions. MNI was computed based on the most redundant skeletal part within each context for spatial comparisons and the aggregated dated contexts for temporal periods.

4 Results

4.1 Site Wide Evaluation

The relationship between artiodactyl NNISP for the site-wide assemblage and volume density is significantly positive for artiodactyls (r 2 = 0.42, p < 0.001; Fig. 7.2). The relationship for Sylvilagus is marginally insignificant (r 2 = 0.08, p = 0.07; Fig. 7.3). These data suggest that bone density partly explains the skeletal part representation for artiodactyls, but not for cottontail rabbits, although some other factors are likely influencing skeletal representation as well (e.g., variation in food processing, use of high density parts for bone tools, etc.).

Fig. 7.2
figure 2

Relationship between bone density values and normed number of identified artiodactyl specimens (NNISP) for the complete Five Finger Ridge assemblage (r 2 = 0.42, p < 0.001)

Full size image
Fig. 7.3
figure 3

Relationship between bone density values and normed number of identified Sylvilagus specimens (NNISP) for the complete Five Finger Ridge assemblage (r 2 = 0.08, p = 0.07)

Full size image

Much of the attrition at Five Finger Ridge is best explained by carnivore ravaging based on the relative abundance of carnivore markers (e.g., digestive polishing, gnaw marks, punctures, etc.) across the site and among skeletal parts of varying nutritional quality. Carnivore marks on artiodactyl bones were not uncommon (9.6% of total NISP) and mostly consist of flaking (1%), pits (18%), punctures (3%), scores (13%), digestive polishing (16%), or a combination of these traits (46%). These markers are non-randomly distributed among artiodactyl skeletal parts (χ2 = 92.03, p < 0.001). When the chi-square adjusted residuals for the presence of carnivore markers are compared with Binford’s (1978) logged-transformed grease index values to correct for a curvilinear relationship, there is a significant relationship (r 2 = 0.29, p = 0.004; Fig. 7.4). Carnivore markers, primarily in the form of digestive polishing (95% of markers), were found on approximately 10.2% of the identified Sylvilagus specimens. These markers are non-randomly distributed among Sylvilagus skeletal parts (χ2 = 231.24, p < 0.001). Following Speth (2000), the adjusted residuals for carnivore markers on leporids are compared against Binford’s utility index values for caribou since comparable data on grease content that takes into account bone density and volume is not available for the former; as there is no reason to believe that the absolute values for caribou correspond tightly with those for Sylvilagus, the relationship is examined using Spearman’s rho. As with the artiodactyl remains, there is a significant relationship between the frequency of carnivore markers and fat volume among Sylvilagus remains (Spearman’s rho: r s  = 0.41, p = 0.04; Fig. 7.5).

Fig. 7.4
figure 4

Relationship between the chi-square adjusted residuals for the presence of carnivore marks and the grease index for each artiodactyl skeletal part (r 2 = 0.29, p = 0.004)

Full size image
Fig. 7.5
figure 5

Rank-ordered relationship between the chi-square adjusted residuals for the presence of carnivore marks and the grease index for each Sylvilagus skeletal part (Spearman’s rho: r s  = 0.41, p = 0.04)

Full size image

Burned artiodactyl specimens are rare (2.5% of NISP). Burning is randomly distributed across the artiodactyl skeleton (χ2 = 7.30, p = 0.40). As such, heat-altered surfaces cannot be conclusively related to roasting activities and may have resulted from being exposed to fire after disposal. Filleting cutmarks (Binford 1981) are common in the artiodactyl assemblage (41% of total cutmarks), and it may be that meat was deboned prior to cooking. The presence of impact marks on 10.5% of the artiodactyl specimens indicates that marrow was accessed from bone cavities. Intensive grease extraction would have occurred after the meat and marrow were removed, consisting of boiling bone parts that have been reduced in size for long durations of time. The reduction of bone portions may be identified by the presence of chop marks, which are rare in the Five Finger Ridge assemblage (n = 12) and likely represent disarticulation of the carcass as they are generally located near joints. “Pot polish” on the edges of bone surfaces may occur after boiling bones for long durations in ceramic pots, but polishing may also result from trampling, alluvial actions, and other post-depositional forces (e.g., Hurlbut 2000; Turner and Turner 1999; White 1992). Pot polishing was not noted in the assemblage. While grease may still have been extracted less intensively via wet cooking, the relationship between carnivore marks and bone grease values suggests that secondary consumers were attracted to discarded bones with remaining nutritional value.

Evidence of culinary processing is limited to burning for the Sylvilagus assemblage. While the distribution of burning is more strongly patterned among the Sylvilagus remains compared to the artiodactyls (χ2 = 27.16, p = 0.007), it is so infrequent that roasting does not appear to have been a common preparation method at the site. Although it cannot be conclusively identified, it is likely that leporids were stewed but for a relatively brief duration since it is clear that carnivores were still scavenging skeletal parts with the highest grease content.

4.2 Spatial Comparisons

There is considerable variability in the degree to which density-mediated destruction explains skeletal part frequencies for both artiodactyls and Sylvilagus within individual contexts (Table 7.2). Some contexts follow the site-wide pattern where attrition is relatively high for artiodactyls but not Sylvilagus, such as Structures 30 and 38; comparatively low AI values from these contexts may be due to relatively lower rates of survivorship for artiodactyl remains. In contrast, bone density values from Structure 29 and Activity Area 9 are a stronger predictor for Sylvilagus skeletal part representation compared to artiodactyls, and these two contexts have relatively high AI values. This suggests that the varying degrees of bone attrition has an impact on this measure used to evaluate the prey choice model.

Table 7.2 Artiodactyl and Sylvilagus skeletal part attrition, number of identified specimens (NISP), minimum number of individuals (MNI), and abundance indices (AI) in Five Finger Ridge structure and activity area fill contexts
Full size table

AI values calculated using NISP and MNI values vary considerably among proveniences, ranging from 0.10 (high Sylvilagus abundance) to 0.73 (high artiodactyl abundance). This variation is not dependent on sample size (r 2 = 0.04, p = 0.23). To evaluate the relationship between attrition and the abundance index, the log-transformed abundance index values were compared against significant (p < 0.05) coefficient of determination (r 2) values for artiodactyl attrition. There is a significant negative relationship (r 2 = 0.47, p = 0.004; Fig. 7.6) that demonstrates that as density-mediated destruction increases, artiodactyl relative abundance decreases. Since this comparison does not take into account the rate of attrition for Sylvilagus, a similar comparison was made using coefficient of determination values for Sylvilagus; there is no significant relationship (r 2 = 0.04, p = 0.62). It appears that the relative survivorship of artiodactyl remains is a strong predictor for the abundance index.Footnote 1

Fig. 7.6
figure 6

Relationship between the abundance index based on NISP values and significant coefficient of determination values for artiodactyl density-mediated destruction (r 2 = 0.47, p = 0.004)

Full size image

The relationship between MNI and NISP among site contexts is significant for both artiodactyls (r 2 = 0.43, p < 0.001; Fig. 7.7) and Sylvilagus (r 2 = 0.85, p < 0.001; Fig. 7.8). It is noteworthy that NISP is a much weaker predictor of MNI for artiodactyls, possibly reflecting issues associated with density-mediated destruction that results in the survivorship of fewer but denser specimens per individual. When the abundance index is computed using MNI values and compared against the coefficient of determination for artiodactyl attrition, there is no relationship (r 2 = 0.06, p = 0.39; Fig. 7.9). However, this does not necessarily mean that the abundance index based on MNI values is a stronger reflection of foraging behaviors across site contexts, as it is highly probable that large game were shared broadly across the site and secondary consumers were an active depositional agent. For both taxa, there is significantly higher frequency of carnivore markers in structural fill than floor surfaces (artiodactyl χ2 = 6.89, p = 0.009; Sylvilagus χ2 = 47.87, p < 0.001), indicating that secondary consumers were likely depositing bones in abandoned site contexts.

Fig. 7.7
figure 7

Relationship between artiodactyl MNI and NISP values for individual site contexts (r 2 = 0.43, p < 0.001)

Full size image
Fig. 7.8
figure 8

Relationship between Sylvilagus MNI and NISP values for individual site contexts (r 2 = 0.85, p < 0.001)

Full size image
Fig. 7.9
figure 9

Relationship between the abundance index based on MNI values and significant coefficient of determination values for artiodactyl density-mediated destruction

Full size image

4.3 Temporal Comparisons

Density-mediated destruction was also evaluated across temporal units and compared against the abundance index values. As seen in Table 7.3, there is no significant relationship between volume density and skeletal part representation for Sylvilagus for all four temporal periods. In contrast, the relationship for artiodactyl remains is significant for three of the four temporal periods; the marginally insignificant relationship for Period 3 is likely a reflection of the small sample size. The abundance index values show relatively little variation among temporal units regardless of whether NISP or MNI values are used (Fig. 7.10); the relative abundance of artiodactyls to cottontail rabbits does not significantly differ among temporal units (NISP: χ2 = 4.15, p = 0.25; MNI: χ2 = 0.80, p = 0.85).

Table 7.3 Artiodactyl and Sylvilagus skeletal part attrition, number of identified specimens (NISP), minimum number of individuals (MNI), and abundance indices (AI) among Five Finger Ridge temporal periods
Full size table
Fig. 7.10
figure 10

Relationship between abundance index values based on NISP and MNI, and temporal designations

Full size image

5 Discussion

Since measures of relative taxonomic abundances are ultimately based on the number of identified specimens, it is critical that we evaluate the influence of selective transportation, survivorship, fragmentation and similar factors vary across taxa and time prior to testing hypotheses regarding human behavior and exploitation of faunal resources. It is expected that the rate at which secondary consumers ravage discarded bone will correspond with the methods used to transform raw animal products into consumables; methods that effectively remove lipid content from bone are likely to result in greater survivorship of the discarded remains, and such processing methods are expected to correspond with individual animal, the degree of resource stress experienced by the foragers, and other factors.

Corroborating previous investigations (e.g., Ugan 2005), this case study demonstrates that the number of identified specimens in an assemblage is ultimately influenced by complex taphonomic histories that may increase or decrease the rate of identification for one taxon over another. The relationship between attrition and taxonomic abundances from Five Finger Ridge is different from that detected by Ugan (2005), who found that density-mediated destruction for artiodactyls and lagomorphs were strongly correlated. At Five Finger Ridge, density-mediated destruction of artiodactyls and Sylvilagus do not correspond with one another. Nonetheless, the fact that both studies have demonstrated relationships between attrition and relative taxonomic abundances is the critical lesson here.

Regardless of how the differences in density-mediated destruction patterns are explained, differing rates of survivorship can have an impact on relative taxonomic abundances. This finding demonstrates the importance of accounting for variation in density-mediated destruction among multiple species before any human behavioral inferences are formed from taxonomic diversity measures. My original research goals at Five Finger Ridge (Fisher 2010) were to identify real spatial differences in taxonomic representation that represent differences in individual foraging decisions and spatial organization. Spatial differences are expected to vary between the sexes (e.g., Bird 1999; Hawkes 1996), between individuals within a sex (e.g., Lupo and Schmitt 2004; Smith et al. 2003), and within a single individual’s life-history (e.g., Bird and Bliege Bird 2000; Kaplan et al. 2000). However, relative skeletal part abundances and taxonomic abundances within each context at Five Finger Ridge are not an accurate reflection of real spatial differences in taxonomic representation, but are instead the product of varying taphonomic processes.

The intertaxonomic and intrasite variation in skeletal part attrition may relate to depositional processes. Carnivores frequently remove faunal remains from their original context and redeposit them elsewhere (e.g., Kent 1981; Marean et al. 1992; Ugan 2010), and this may alter the distribution of skeletal parts in a way that corresponds with their nutritional utility and density values. Skeletal remains with high grease content are expected to be removed by scavengers from their original location of discard and deposited elsewhere (within or outside the archaeological site), leaving skeletal parts with low nutritional value behind. Based on ethnoarchaeological work in the American Southwest, Kent (1981) found that that the spatial distribution of faunal remains is determined primarily by domestic dogs that frequently take bones to specific locations to avoid competition with other dogs. However, identifying secondary deposition by carnivores may be difficult, as Kent also found an absence of carnivore markers on bones that were broiled or boiled. As such, cultural or behavioral interpretations based on spatial distribution of faunal remains must be demonstrated rather than assumed in regions where secondary consumers such as coyotes and domestic dogs are present.

The lack of demonstrable temporal change in the abundance index regardless of whether NISP or MNI measures are used is in contrast to other data from Five Finger Ridge that show changes in taxonomic exploitation when analyses are restricted to animals of similar body size. Fisher (2012) argues that there was an expansion of pinyon-juniper woodlands in the vicinity of Five Finger Ridge based on an increase in the relative abundances of Sylvilagus nuttallii over S. audubonii, as well as a decrease in the relative abundance of jackrabbits to cottontails. There is also a significant decrease in the relative abundance of bighorn sheep to deer in Period 2A, corresponding with significantly different carbon and strontium isotope values that collectively suggest that fewer sheep were acquired from higher, more distant elevations at this time (Fisher and Valentine 2013). Further, when relative body part representation is evaluated using Stiner’s (2002) anatomical units that contain an even distribution of high density parts to control for density-mediated attrition, the mean food utility index values (Broughton 1994; Metcalfe and Jones 1988) increases incrementally through time. This indicates that low utility parts were increasingly discarded prior to transportation to Five Finger Ridge, most likely to reduce travel costs as local abundances of artiodactyls decreased (Broughton 1994; Nagaoka 2005). Collectively, these data demonstrate that real temporal trends associated with foraging efficiency and local environmental changes can be identified even when the abundance index measure does not demonstrate such a trend.

6 Conclusion

Abundance indices are frequently used to evaluate the diet breadth model based on optimal foraging theory (for examples, see Bird and O’Connell 2006; Lupo 2007; Morgan 2015). This research area has significantly contributed to our understanding of the relationship between humans and their surrounding environments, such as the impact of climate-induced resource fluctuations and overhunting by prehistoric peoples. The potential relationship between conversion of raw animal resources into food products and bone attrition rates demand the evaluation of density-mediated destruction. Variation in differential survivorship among taxa across sites and time should be expected by researchers. Since these influences cannot easily be controlled, caution must be exercised prior to testing hypotheses derived from optimal foraging theory, especially when multiple assemblages are combined into a spatial or temporally averaged dataset (Lyman 2003). One cannot simply remove from analysis assemblages that demonstrate high levels of attrition as it is also critical to understand why some assemblages appear to be unaffected by density-mediated destruction. The relative survivorship of zooarchaeological remains captured in these variable assemblages may actually be a product of behaviors related to foraging efficiency. While using derived measures (e.g., MNI) may bypass some of the issues with differential transportation, survivorship, and fragmentation , these introduce a range of additional problems (see Lyman 2008 for review). Instead, researchers must rely on a number of independent measures in conjunction with relative taxonomic abundances for testing foraging theory predictions.