Methodological bias in home range and mobility estimates when locating radio-tagged trout, Salmo trutta, at different time intervals

Radio tracking has been extensively used to enhance our knowledge on the mo vement and home range of fish in general and salmonids in particular. Howev er, the use of various temporal sampling protocols is likely to overlook fi sh movements, and produce experimental artefacts, the amplitude of which is unknown and may compromise comparison of fish behaviours revealed by diffe rent protocols. Starting from a day-by-day tracking study of brown treat in Belgian waters (Aisne stream, nine fish, minimum 39 daily locations per fi sh), we re-examined their home range and mobility, through a subsampling pr ocess, as if the fish had been located at longer time intervals (2-14 days) . The estimates from subsamples were compared to the original data set in o rder to quantify the corresponding reduction of accuracy from observed data , and how far this could be predicted on the basis of locating frequency. T he results clearly indicate that all intervals longer than one day generall y generate substantial biases (reduction of accuracy from 0 to 82% for home range and from 5 to 92% for mobility) but these can be partly corrected th rough the use of predictive models. This analysis demonstrates that any com parison between studies relying on different locating frequencies can gener ate some ambiguity when interpreting biological phenomena or geographical d ifferences. (C) 2000 Ifremer/CNRS/INRA/IRD/Cemagref/Editions scientifrques et medicales Elsevier SAS.