Designing mark-recapture studies to reduce effects of distance weighting on movement distance distributions of stream fishes

Mark-recapture studies generate biased, or distance-weighted, movement data because short distances are sampled more frequently than long distances. Using models and field data, we determined how study design affects distance weighting and the movement distributions of stream fishes. We first modeled distance weighting as a function of recapture section length in an unbranching stream. The addition of an unsampled tributary to one of these models substantially increased distance weighting by decreasing the percentage of upstream distances that were sampled. Similarly, the presence of multiple tributaries in the field study resulted in severe bias. However, increasing recapture section length strongly affected distance weighting in both the model and the field study, producing a zone where the number of fish moving could be estimated with little bias. Subsampled data from the field study indicated that longer median (three of three species) and maximum distances (two of three species) can be detected by increasing the length of the recapture section. The effect was extreme for bluehead chub Nocomis leptocephalus, a highly mobile species, which exhibited a longer median distance (133 m versus 60 m), a longer maximum distance (1,144 m versus 708 m), and a distance distribution that differed in shape when the full (4,123-m recapture section) and subsampled (1,978-m recapture section) data sets were compared. Correction factors that adjust the observed number of movements to undersampled distances upwards and those to oversampled distances downwards could not mitigate the distance weighting imposed by the shorter recapture section. Future studies should identify the spatial scale over which movements can be accurately measured before data are collected. Increasing recapture section length a priori is far superior to using post hoc correction factors to reduce the influence of distance weighting on observed distributions. Implementing these strategies will be especially important in stream networks where fish can follow multiple pathways out of the recapture section.