Multiple species telemetry provides insight into habitat preferences of large river fish
We evaluated movement and habitat use of 105 individual fish representing six species (freshwater drum Aplodinotus grunniens, common carp Cyprinus carpio, smallmouth buffalo Ictiobus bubalus, channel catfish Ictalurus punctatus, hybrid striped bass Morone saxatilis X M. chrysops and flathead catfish Pylodictus olivaris) over approximately 36 months within one pool of a regulated large river. These species were selected based previously established numerical dominance. Our hypotheses were that a) movement rate would vary in response to species, season, and channel morphometrics; and b) habitat within the channel would be used differentially by fish. A total of 1096 discrete observations were collected along with detailed habitat information. Aggregate fish data were subsetted to include only fish with > 8 observations. Observations for 79 individual fish were randomly selected for the analysis cohort, reducing the dataset to 598 observations as the source for the training data. We used displacement of fish as a metric of movement to examine patterns and relationships. Specific fish-habitat interactions considered for analysis were proximity to tributaries, proximity to shore, depth, and substrate character based on average particle size. We used an information-theoretic approach paired with weights-of-evidence predictive modelling to develop a surface indicative of aggregate species habitat preference.
We were able to demonstrate a seasonal component to fish movement, with freshwater drum and common carp having higher seasonal rates of movement than other species. We were also able to demonstrate differential habitat use. Contrast for the weights of evidence approach was highest for depth, although fish were located across all depths. Fish were associated with the shoreline in higher proportion than they were associated with the thalweg, while the majority of observations were made in the thalweg region. Sediment structure (particle size) and rate of movement had no interaction, although more fish were observed in areas of smaller-sized particles. The association of fish with tributaries had no seasonal component. Periods of high river discharge appeared to limit fish movement, even when discharge between sampling events was taken into account. The results of the probabilistic model indicate that there are several discrete regions with a high (80-100%) probability of fish occurrence, while the rest of the habitat averaged between 40-60% probability of occurrence. The results of this study indicate that using multiple species of fish as a basis for habitat modelling yields may allow managers to predict the amount of suitable habitat for fish in regulated large rivers.