Cooperative Fish and Wildlife Research Units Program: Arkansas
Education, Research and Technical Assistance for Managing Our Natural Resources

Little is known regarding fish movement in small to moderate sized, warm-water streams. In the late 1990’s a dam was proposed for upper Bear Creek, a tributary to the Buffalo National River. Barriers to fish dispersal can lead to negative impacts on stream fish assemblages. The objectives of this study were to determine 1) the seasonal fluctuations in fish species composition, relative abundance, and size structure and its relationship to environmental variables in Bear Creek headwaters and other Ozark headwater streams, 2) movement of fishes between Bear Creek and the Buffalo National River and its relationship to environmental variables, and 3) larval fish drift densities and rates in the upstream, middle, and downstream reaches of Bear Creek and its relationship to stream discharge. We sampled replicate habitats (pool, run, and riffle) in each of three Boston Mountain ecoregion streams and three Ozark Highland ecoregion streams seasonally (April, June, August and October) in 2002 and 2003. We used backpack electrofishing and three-pass removal techniques to determine fish population abundances, capture probabilities, and density estimates. We used the Zippin removal estimator via maximum-likelihood to estimate the number of each species and these population estimates were divided by habitat unit area (for RDA) and habitat unit volume (for ANOVA) to derive fish densities. We used redundancy analysis (RDA) to examine relationships between fish species relative densities and explanatory variables, especially season by site interactions. To determine factors driving the patterns found in the RDA’s, we used univariate ANOVA to examine the effect of season and stream on total fish density, total fish numbers, and habitat volume. Length-frequency histograms of the common species were compared among seasons using Kolmogorov-Smirnov tests to examine seasonal shifts in size classes and to determine whether larger size classes were moving into headwater streams to spawn. Seasonal species extinction, colonization, turnover rates, and detection probabilities in Bear Creek were calculated using program COMDYN. To determine fish movement between Bear Creek and the Buffalo River, two-way weirs were installed between the Buffalo River and the downstream reach on Bear Creek in 2002 and 2003. Multiple regression models were used to examine the relationship between the number of fish caught each day and abiotic predictor variables date, discharge, water temperature, and daily flow difference. We determined movement of larval fish (5-25mm total length) by sampling nocturnal stream drift with replicate nets set at upper, middle and lower reaches of Bear Creek in spring 2002 and 2003. Drift densities were compared among sites and times using 95% confidence intervals. Redundancy analysis showed a significant relationship between fish species densities and site by season combinations as explanatory variables in both 2002 and 2003. In both years fish species densities were well separated by ecoregion. In both years there were greater densities of central stonerollers, creek chubs, whitetail shiners and slender madtoms in Boston Mountain streams and greater densities of banded sculpins, hornyhead chubs and lampreys in Ozark Highland streams. Significant seasonal changes in fish species densities also occurred, especially in 2002 and in Boston Mountain streams. Seasonal shifts in fish densities were driven largely by reductions in habitat volume from April to October in Boston Mountain streams and by changes in fish numbers in Ozark Highland streams. Species detection probabilities were high so it was unlikely that species were missed during sampling. Although fish species richness increased from April to October in Boston Mountain streams in 2002, overall fish species richness and species composition in headwater streams was temporally stable with low numbers of species and little change in species composition over time. In 2002, capture probabilities for total fish numbers were significantly greater in October than in August in Boston Mountain streams, whereas capture probabilities did not differ significantly among seasons in Ozark Highland streams. In 2003, capture probabilities for total fish numbers were significantly greater in August and October than in June in Boston Mountain streams and significantly greater in August and October than in April in Ozark Highland streams. Simulations showed that population estimates should not be biased within the range of capture probabilities (0.30-0.60) and population sizes (50-300) that we observed. Length frequency distributions changed among seasons in all headwater streams, but there was little evidence that larger individuals moved in to spawn in the spring. The main seasonal change in length frequency histograms was due to the growth of individuals in various age classes. In 2002, the total number of fish moving upstream or downstream through weirs between Buffalo River and Bear Creek did not exceed 100 fish day-1, whereas in 2003 the total number of fish moving upstream or downstream through weirs between Buffalo River and Bear Creek exceeded 100 fish day-1 on several occasions. In 2003, large numbers of Notropis pilsbryi and Campostoma anomalum moved upstream from Buffalo River into Bear Creek during April and June. Upstream movement of Notropis pilsbryi was significantly negatively related to date and upstream movement of Campostoma anomalum was significantly negatively related to date, discharge and temperature. Larval fish drift densities were extremely high at the lower Bear Creek site and decreased at the middle and upper reaches of Bear Creek. The number of larval fish drifting into Buffalo River from Bear Creek at the lower Bear Creek site ranged from 30,000 individuals per night in April 2003 to 1,260,000 individuals per night in May 2002. Larval fish drift densities did not increase substantially during storm flows in May 2002, but increased significantly during nighttime despite reduced discharge. Ecoregion and hydrology appear to play a major role in structuring the fish assemblage of upper Bear Creek and other Boston Mountain and Ozark Highland streams. In the case of Boston Mountain streams, highly variable discharge with high flows in spring and extensive stream drying in summer led to strong seasonal changes in fish species densities with significantly increased densities in summer and autumn. In Ozark Highland streams, hydrology is important because in this karst ecoregion numerous springs lead to more seasonally stable flows and less extreme seasonal shifts in fish species densities. Major unidirectional movements from Buffalo River to Bear Creek of several common fish species during breeding season and extremely high larval fish densities suggest that Bear Creek acts as spawning and nursery habitat for multiple species of fish. Larval drift densities suggest that despite extensive drying, permanent portions of the middle and upper reaches of Bear Creek also act as spawning and nursery areas for some fish species.