SAMPLING SPECIES RICHNESS
• Simulate a population of 1000 individuals composed of various species. • Calculate species richness by sampling. • Determine how community composition affects species richness estimates. • Develop a bootstrap analysis of how sample size affects species richness estimates.
Imagine you are a conservation biologist conducting surveysof insect species in previously unstudied areas. Your mission is to estimate the number of species occurring in different habitat types across a large region. The number of species that occurs in a particular area is called its species richness, and it is just one of many measures of biodiversity. A practice known as a rapid biodiversity assessment is currently being used by many conservationorganizations to survey the biodiversity of plants and animals before pristine habitats are altered and developed (see, for example, http://www.conservation.org/RAP/Default.htm). Assume there are 10 locations that must be sampled in a short period of time. How many samples should you take at each site to estimate the number of insect species in a location before moving onto the next location? Time andfunding are short and you will not be able to do a complete survey of the insect biota. A basic problem is that it is nearly impossible to count every single species in a community. If funding and time were unlimited, you might conduct a complete census and enumerate all of the species in the community. However, this is not often the case; instead you must settle for sampling the community andestimating its species richness based on this sample of individuals. Estimating species richness by sampling presents some major challenges. First, you are likely to miss some species. And second, although the more you sample in a particular area the more likely you are to find new, previously unsampled species, there is a point of diminishing returns that must be considered in your sampling efforts.For example, consider a community that consists of 1000 insect species, and you sample insects by sweeping the vegetation with a net. In your first sweep, you capture 25 species. In your second sweep, you capture 30 species, but 20 of
these were already captured in the first sweep. Thus, with 2 samples your total species richness is 35 (25 new species recorded with thefirst sweep, and 10 new species recorded with the second sweep). With each sweep (sample), the chances of adding a new, previously unsampled species decreases. At some point it becomes cost-effective to move
Species Richness as a Function of Sample Size
80 70 60 50 40 30 20 10 0
Cumulative number of species sampled
Number of samples
to the nextlocation and start sampling anew. In the example shown in Figure 1, taking 15 samples will yield more or less the same species richness estimate as taking 18 or 20 samples. What factors will determine the shape of a sampling curve such as Figure 1? One factor is the distribution of the individuals within the community. If the community consists of 100 species, but 90% of the total individuals are fromspecies 1, most of our samples will consist of species 1, and we may have to take many samples to encounter one of the rarer species. In contrast, if the numbers of individuals in the community are more or less evenly distributed across 100 species, so that no single species dominates the community, you may not have to sample as much because all species are equally abundant. Another generalproblem with sampling is that you will never really know how well your species richness estimate measured the true species richness in a community. After all, this is what you are trying to estimate with your sampling. With advances in computing, however, it is now possible to ask the question, “If we take a different, random sample from a community with a known number of species, how does the species...