QUANTIFYING BUTCHERY MARKS IN A SHELL MIDDEN Laura

QUANTIFYING BUTCHERY MARKS IN A SHELL MIDDEN Laura MAMELI Juan A. BARCELO UNIVERSITAT AUTONOMA DE BARCELONA

The problem During 19 th. Century european hunters produced a heavy impact on local populations of sea mammals. Is it possible to identify in an archaeological site of the same period, an increase in bird carcass processing as a consequence of the continuing decrease of the most important resource affected to indigenous populations (sea mammals)?

The problem We are looking for: • changes through time in the average of butchery marks present in bird bones • changes through time in total biomass exploitable from birds • changes through time in taxonomic diversity of hunted birds

The Site Túnel VII is an archaeological site located in the Beagle Channel (Tierra del Fuego. Argentina) produced by the garbage accumulated by a group of people over a series of occupations at the same place.

Different stratigraphic subunits have been identified during fieldwork. Each stratigraphic unit corresponds to a single deposition of residues.

Stratigraphically related subunits have been integrated into occupation phases (temporal episodes). Nine episodes have been identified. We have limited our analysis to a seriation of bird bones in 8 temporal stages: A-B, C, D, E, F, G, H, J (from the oldest to the latest).

Bird Bones More than 5000 bird bone sherds have been studied Raptors Cormorant Penguins Small Terrestrial Birds Coastal Birds Big Sea Birds Small Sea Birds

QUANTIFYING BUTCHERY MARKS IN BIRD BONES Bird bone remains have been investigated for the presence/absence of butchery marks. Using low resolution microscopes (x 20), different types of marks have been described and quantified.

Mapping Traces on bones

CORRESPONDENCE ANALYSIS. Taxa (Nisp) Dimension 1 explains 83% of total variance. The variable with maximal inertia is the existence of Big Sea Birds. There is a strong serial correlation from the oldest levels with only a 4% of total NISP until 28% of total NISP at the latest levels.

CORRESPONDENCE ANALYSIS. Marks A single dimension explains 100% of diversity. This Dimension discriminates episode H, because of the greater average of bones with butchery marks. It is interesting to remark that if we delete episode H because of its specificity, there is a relative linear order from Episode B (the oldest) with the lowest average of bones with marks (3%) until the last (episode J), with the very high scores (16. 2%).

Historical Trend Can we conclude the existence of an historical trend based on these multivariate results of the similarities between stratigraphic layers?

Historical Trend The problem with these results is the fact that direct comparison of stratigraphic ordering and frequencies can be misleading, because it does not render correctly temporal evolution. It is impossible to render time just with the stratigraphic order, without taking into consideration the existence of linear trend and/or the different temporal duration of episodes.

FREQUENCY SERIATION TOTAL NISP TOTAL BIOMASS (based on MNI)

FREQUENCY SERIATION TAXONOMIC DIVERSITY. ENTROPY AVERAGE OF MARKED BONES

Temporal Duration • The amount of residues accumulated in a single deposition event (stratigraphic subunit) depends on the number of people generating garbage, the time during which they have been producing continuos garbage, and the social way of disposing garbage. • We can approximate temporal duration of each temporal phase or episode counting the number of single deposition events associated to the episode and the total volume of excavated sediment from all single

TEMPORAL DURATION Volume of sediment for all subunits of each occupational phase Older episodes have less subunits but with higher volumes, whereas the last episodes have more subunits with smaller volumes.

TEMPORAL DURATION The sum of volumes for all subunits within an episode does not show any serial correlation

TEMPORAL DURATION The longuer the ocupation, the more NISP Linear Regression between temporal duration (number of subunits within a phase) and total NISP. R-square is relevant (0. 48), and increases if we delete outliers (too many subunits in layer F). [0. 89].

TEMPORAL DURATION The average of marks does not depend on temporal duration Linear Regression between temporal duration (number of subunits within a layer) and average of marked bones. Rsquare is irrelevant (0. 202). F and H, the longuest episodes accoring to the number of subunits have the higher averages of marked bones, but episode H has a higher average although it has less duration than episode F.

TEMPORAL DURATION The older the ocupation, the lower total biomass and average of bones with butchery marks. Weighted data have been normalised.

Historical Trend THE GENERAL STRUCTURE OF BIRD HUNTING AND PROCESSING AT THE TUNEL VII SITE SHOWS THAT: • The only consistent temporal trend is the general increase in bird hunting. • There is strong linear correlation between the quantity of data and stratigraphic ordering. • This fact does not coincide with a general increase in taxonomic diversity, and it is a consequence of the increase in the capture of Big Sea Birds. We can suggest an historical trend characterized by the growing of Biomass by hunting a profitable bird (in terms of biomass) and not increasing the number of individual preys from other taxa.

Historical Trend • There is not a consistent temporal trend in the average of bones with butchery marks. The total number of marked bones increases because bird hunting has increased through time, but preys are not being processed differentially. • The longer the temporal duration of an episode, the more single deposition events, and the more biomass from bird will be used and the more bones will show butchery marks.

Historical Trend The fact that the site was continually reoccupied by the same population implies that something about the past of the series is likely to remain fairly stable. We have seen that successive episodes have more in common than episodes not temporally contiguous. That means that the remote past is not as good a predictor as the recent past, and that our ability to predict an event is best nearer the event, and worsens as we recede from it.

Statistics: The Easier the Better Correspondence Analysis detects some of these trends, but in most cases it is overdetermined by the existence of an underlying linear trend. The big difference in the number of bone fragments between the oldest and the newest layer influences the results. Historical trends are far more complex that suspected after Multivariate Statistics. Paradoxically, easier statistical methods (robust exploratory tools) are sometimes better adapted for discovering complexity.