Sampling variability in benthic studies may result from sampling device operation, physical features of the environment, laboratory sorting procedures, and biological features of study populations. Selected factors and procedures that influence variability, samplers affected, and proposed remedies are presented. Consequences of not considering autecological components in sampling designs are illustrated by analysis of larval counts of Cheumatopsyche pettiti (Banks), a multiple cohort caddisfly with an aggregated population. The range of mean numbers of C. pettiti was great with low sample numbers. Aggregation is more reliably measured at low sample numbers with the Index of Dispersion and the Mean Crowding Index than with the dispersion parameter k, the calculation of which from the maximum-likelihood equation, is integrally related to sample size. Nonrandom patterns of C. pettiti observed from samples collected in an Indiana, USA, stream riffle, may result from a failure to consider hyporheic distributions, spatial influences (e.g. sampling both favored and nonfavored microhabitats), instar-specific differences, and behavioral features. Variability in secondary production estimates of an aggregated population of Ceraclea ancylus (Vorhies) from a Kentucky, USA, stream indicated similar relationships to sample size.The size of the mean, the degree of aggregation, and the desired precision of the mean estimate will influence the number of samples required to estimate densities of benthic populations. Sample size requirements calculated from data reported in previous studies were high to achieve accepted levels of precision. Habitat stratification may reduce the numbers of samples required. Dicosmoecus gilvipes (Hagen) exhibited nonaggregated patterns and required fewer samples to estimate density in uniform substrate areas of a California, USA, river pool than did aggregated populations in both mixed substrate areas and the entire pool. Ceraclea ancylus required fewer samples for density estimates in stratified (by habitat or substrate type) than unstratified habitats, the fewest samples being necessary when the individual stone was the sampling unit. Judicious choice of study populations may permit larger numbers of samples to be collected and processed with reduced cost, as an alternative to stratification. Larvae of C. ancylus and D. gilvipes could be separated in the field; density underestimation due to a hyporheic population component was eliminated because of surface dwelling behavior or by choice of study sites; and compounded spatial distributions due to co-occurring instar-specific patterns were absent because the populations have a single cohort.Larger numbers of samples may be necessary than are generally taken in benthic studies. Further research is needed to assess variability in secondary production estimates and community diversity analyses. Improved methods for substrate surface area estimation and increased use of experimental approaches and sequential sampling techniques should be considered in future benthic sampling designs. Key words: sampling, benthos, aquatic, macroinvertebrate, Trichoptera, insect, experimental design, autecology, life history, variability
Read full abstract