Richness and Evenness

One of the simplest ways to compare two different assemblages of animal bone is to look at them in terms of richness and evenness. Richness is the number of categories, either taxonomic identification or skeletal element, used to describe your assemblage. Evenness is the relative quantity of each of these categories in the assemblage. Since MNI is usually reported for any faunal assemblage, we can look at richness and evenness using its taxonomic categories.

Qualitatively, richness and evenness can be estimated by placing two tables of NISP or MNI values adjacent to each other. Use MNI whenever possible, since NISP comparisons may just be comparing how fragmented the assemblages are. Also, as mentioned previously, it is wise to focus on order-, family-, or genus-level identifications, not species-level identifications, when comparing data produced by different analysts. Simply reduce species-level identifications to genus-level identifications, and genus-level identifications to order-level identifications if necessary. If mutually exclusive categories were used<>, it should be easy to calculate totals for each level of identification. Then count the number of categories used to describe each animal bone assemblage. For example, when comparing the faunal assemblages of several different archaeological sites in New York, I reduced all identifications to taxonomic order and then compiled the data in a table (table). By counting the total number of taxonomic orders used to describe each assemblage, the Engelbert Site shows the highest richness and Frontenac Island and Sackett show the lowest richness.

NISP Table

Table. Comparison of several faunal assemblages from New York State by NISP. Some sites have been analyzed be multiple analysts. In those cases, the analysts name appears after the site name. In this example the Engelbert site was described using the most taxonomic orders and is therefore assemblage with the most richness.

Richness does not take into account how many individuals are present within each of these taxonomic groups; this is the role of evenness. Evenness allows for a general understanding of the patterns of animal interaction that created the assemblage. A very even assemblage, one in which the same number of individuals represents each genus, shows little selection of animals or little preference for one animal over another. A very uneven assemblage, one where a single genus is represented by many more individuals than all others, shows a high degree of selection or specialization. Again, MNI is preferred over NISP for this calculation. For example, an assemblage that is made up of five cows (Bos sp.), five pigs (Sus sp.), and five sheep (Ovis sp.) is a very even assemblage that shows general exploitation of domestic species. An assemblage that is made up of five cows (Bos sp.), one pig (Sus sp.), and twelve sheep (Ovis sp.) is an uneven assemblage that shows a focus on sheep utilization over other domestic species. Using the data from New York provided in table 6, most of these assemblages are very uneven due to the high numbers of Artiodactyla (in these cases, deer and elk) when compared to all other taxonomic orders. The Engelbert assemblage is more even than the others because of the high Anura NISP, which comes close to the Artiodactyl value. This suggests that both Artiodactyla and Anura were selected for.

The examples given here have used richness and evenness in a qualitative sense, a relative measure based on appearance. This approach can be quite useful for quickly exploring the data to see what aspects of an assemblage may be more or less important to the research questions at hand. If the qualitative approach suggests trends that prove to be important, the quantitative approach, meaning the use of exact numbers in formulae, should be employed in order to correct for differences in sample size between assemblages being compared. Richness and evenness as statistics can be calculated using a variety of formulas. For more information on these statistics, see Leonard and Jones (1989).