Estimating Mean Shell Size

Contrasts in the average size of marine shells belonging to particular species and recovered from archaeological middens may reveal information about spatial or chronological variations in prehistoric harvesting strategies or in the condition of the local shellfish population. A systematic decrease in shell size might indicate that a higher-cost harvesting strategy had been adopted by the collectors, that the shellfish population was in a natural decline (perhaps due to the evolving physical condition of the coast), or that this resource was being overexploited by humans. An increase in mean shell size might reflect the presence of an older or more vigorous population, perhaps in response to more favorable environmental conditions or to lowered predation pressure from humans.

A methodological issue concerns the best way to assess contrasts in average shell size:

• Direct measuring of the lengths or the weights of whole shells that were recovered from a midden may be practical in particular for species whose shells are small, numerous, and frequently found in unbroken condition, such as the bean clam Donax spp. (Byrd 1997:82; Serr and Byrd 1996:225-229). However, this method may produce imprecise or biased results in cases where breakage was common. The proportions of broken shells probably differed between large and small shells of the same species. The extent of breakage also varied between different sites and between different components within a single site.
• An alternative way to assess contrasts in shell size is to calculate the mean shell weight by dividing the total weight of the whole and fragmentary shells from a species by an estimate of the total number of shells represented in the sample, as inferred from a count of pelecypod hinges or of gastropod apices (Laylander 2006; Laylander and Becker 2004). This should produce a reliable and efficient index for shell size, assuming that there was no depositional segregation of the different parts of shell fragments and that relatively little shell weight was lost eight through postdepositional solution or erosion of the shells or through losses of small shell fragments during archaeological recovery. The method allows the entire shell assemblage to be used in the calculations, rather than just a few whole shells, increasing the precision of the estimate or reducing the shell sample size necessary for a reasonable estimate. 

PROSPECTS

Future studies may be able to test the reliability, precision, and applicability of various methods for estimating mean shell size in midden samples, as well as the interpretive value of the results they provide. Methods may be tested both through the excavation and analysis of new midden samples and through studies of existing curated collections, if the methods used in the collections’ recovery were sufficiently intensive and unbiased.