Gotelli and Colwell, Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness

Species richness is a common measurement in describing diversity and community structure. It is an important tool in conservation efforts, which frequently aim to maximize an area’s richness and thereby, increase its biodiversity. Despite its simplicity as a concept, species richness, or the number of species present, is often a very difficult measurement to accurately obtain and is even harder to properly apply to an ecological study. Gotelli and Colwell describe some more methods for obtaining species richness data as well as common pitfalls in their utilization.

Presented in Figure 1, are four alternate taxon sampling curves based on two dichotomies. The first being the unit of measurement: individual- or sample-based; and the second being the method of collection: accumulation or rarefication curves. Individual-based simply determines the number of species looking at randomly decided individuals in each plot while sample-based sets random samples and determines the number of species in each and pools the collected data. There exists also a “hybrid” method (not shown in Fig. 1), “m-species list”, in which the first m number of species encountered in a sampling area are listed and is repeated for multiple samples, plotting the number of new species against the sample number. Accumulation curves function similarly in that with each new sampling effort, the new species found are plotted until, ideally, no new species are being recorded and the curve reaches an asymptote. Alternatively, rarefcation curves repeatedly sample the collection of a certain number of samples/individuals, recording the number of species observed with consecutively increasing (or decreasing) sample numbers. When plotted, this too will ideally reach an asymptote, as a some number of increasing samples should ultimately contain all species present.

These curves, which have the potential to accurately describe the species richness of the area in question, need to be applied with caution. Figure 2 shows that a rarefication curve, when rescaled from samples to individuals, gives conflicting data. It is important to compare richness based on individuals rather than samples given the possibility of datasets to differ in their mean number of individuals per sample. There are pitfalls associated with the “m-species list” method as well when comparing species rich to species poor sites (Figure 3). This method requires a greater sampling effort in the species- poor site in order to reach the required number of m species. And should ideally level out sooner than the species rich site. However, in a case where sampling stops before an asymptote is reached, both sites will appear to have similar richness.

Category-subcategory ratios, such as species to individual ratios, are often used to make richness comparisons. Figure 4 shows two examples of pitfalls in using this value. This ratio is dependent on species density, and as seen in 4(a), samples with equal species richness may still show different ratios. Also, the ratio assumes that richness increases linearly with abundance so that it is possible to obtain equal ratios when the richness is in fact different (4(b)). Figure 5 and 6 explain the pitfalls in using a species/genus ratio. This ratio will be greater for a larger sample than for a smaller sample. Traditionally, a low ratio is attributed to strong intergenic competition, but more recent evidence suggests that this is not always the case. Figure 7 compares two rarefication curves (one individual and one sample based), which are used to determine species richness or density. In the individual based curve (a), both treatments must be “rarefied” to the same number of individuals (white dot) to compare the richness as the complete samples (black dots) give the species density, while in the sample based curve (b), the complete samples give the richness, so the treatments must be rarified to the same number of samples to compare the species density.

One final issue of importance is the use of asymptotic estimators in instances where exhaustive sampling is impractical. Extrapolating from rarefication curves does not provide a reliable estimate, as it is essentially a tool for interpolation. Nonparemetric estimators are a better option. One example is based on the number of rare species- a larger amount indicates a greater likelihood of species not represented by the dataset. There are some asymptotic estimators, which have proven accurate when used on small datasets for which the richness is known. In applying these estimators to larger datasets, which have not begun to level off, however, an asymptote is not always reached. In these cases the estimator serves as a lower bound estimate of richness.