An overview of sampling issues in species diversity and abundance surveys

Homer S. Swingle was a pioneer in pond management and professor of fisheries at Auburn University. Results from his scientific research on the principles of fish interactions in lakes and ponds, and methods to improve fish production are used worldwide, and he is credited with improving lives of countless people by increasing the amount of food available to those in developing countries and providing recreational opportunities to a multitude of anglers. His advice was sought by US presidents, prime ministers, the United Nations and countries across the globe (Byrd 1973). His biological work was groundbreaking, yet there is little evidence that Homer Swingle used sampling procedures when conducting his work. His major works contain no reports of sampling error with his estimates (Swingle 1950, 1952) and his contemporaries have pointed out that he actively avoided subsampling in favor of census. His methods for enumerating a population were simple. At the conclusion of his pond experiments, he would census every fish involved, either by draining or treating the pond with piscicide (fish poison). Being a student on Swingle’s piscicide crew meant counting every dead fish as it floated to the water’s surface in the hot Alabama sun over a 3-day period. Those involved in the third-day pickup were the unluckiest, and often the least senior members of the crew—a smelly task indeed! While a complete census of all experimental or survey subjects provides the most complete information, few of us have the luxury of counting every subject in a population to estimate population characteristics. It simply costs too much, takes too much time, or is infeasible in some other way. Therefore we have to sample a portion of the population to estimate diversity or population parameters such as abundance. For example, a marine biologist estimating the number of clams in the Gulf of California cannot hope to count every one—he will expand an estimate calculated from a sample of plots. A plant biologist measuring the biomass of grasses in a field will not weigh every grass blade in the field—she will estimate the biomass through a sample of plots containing grass. A wildlife biologist will not count all mice in a national park to determine relative density of different species. Density will be determined by a subset of captures from different areas. To ensure the most useful information, a biologist must be familiar with basic sampling issues. Here we discuss some issues that all biologists should know when sampling diversity, abundance, or other parameters. Consideration of these issues allows the biologist to sample in a manner that, while not a perfect reflection of the population, will provide the best representation of the true population as possible.