|
|
 |
|
|
| |
Communication at the NABS Annual meeting, Keystone, 2000
(152) EFFECTS OF FIELD SAMPLING METHOD AND SUBSAMPLING ON PERFORMANCE OF RIVPACS-TYPE MODELS.
J.D. Ostermiller and C.P. Hawkins. Department of Fisheries and Wildlife, Utah State University, Logan. UT 84322-5210
We often infer the condition of stream biota from samples collected within a reach. However, both field sampling and lab processing can influence estimates of biotic structure. We examined how type of field sampling, laboratory subsampling, and taxonomic resolution influenced the performance of RIVPACS-type models. We also examined variation among replicate samples in O/E values, the model's measure of biological condition. Samples were collected in summer (1998) from 93 reference streams in western Oregon and Washington. Both fixed-area, riffle habitat and 10-minute, multi-habitat invertebrate samples were collected. All samples were first subsampled (>300 individuals) and then processed in their entirety. We then created 12 models from all combinations of taxonomic resolution (family and genus), sampling method, and sorting process. We also examined variability among O/E values from samples collected at 5 replicate reaches in each of 5 streams. Model error was lowest for families and when both types of field samples were combined. However, models based on subsamples performed better than those based on whole sorts. About half of the error in O/E estimates was related to sampling. These analyses imply that assessment errors can be minimized when multiple habitats are sampled and a fixed number of individuals are identified.
Presented at 4:30 PM on Tuesday, May 30, 2000 in Bioindicators and Rapid Bioassessments II
|
|
|
|
|
|
|
