| Japanese rivers have experienced various alteration of morphology, mainly for the purpose of flood control. Such morphological alteration seems to have significant influences on fish behavior, but the influence is not clearly understood quantitatively. The purpose of this paper is to try to assess the influences of river morphology changes on fish behavior quantitatively by using telemetry method and fish behavior mmodel. The study site is located in midstream of the Chikuma River, Japan. Chikuma River is a part of the Shinano River which flows down northward in center of Japan. To develop a fish behavioral model driven by the velocity distribution, we analyzed relationships between simulated current distribution and measured fish behavior by telemetry. In the research field, the location of four individuals of Hemibarbus labeo(test fish) were continuously tracked about every 3 minutes from 24 February to 25 August, 2005, using the Advanced Telemetry system (ATS) which we have developed. Measured fish behaviors and simulated velocity distribution are correlated to develop a fish behavioral model such that simulated flow can predict fish behavior on GIS platform. Here in after, we describe fishes on computer as virtual fishes. Past river morphology change was presumed by using topographic maps of and around the Chikuma River in 1896, 1964 and 2000. Flows during floods at above three periods were simulated using the presumed river morphologies. The behavioral pattern of virtual fish was compared for three periods. Tracked fish behavior with ATS indicated that the test fish prefer pools where average velocity is 0.6 m/s, and average water depth is about 0.5 m/s in ordinary water level. ATS also showed that the test fish avoided to be flushed downstream by utilizing low velocity area which occurred on floodway during floods. Results of simulated flood flows for the three periods indicated that velocity distribution in 1896 was conspicuously different. Comparing to the morphology in 1964 and 2000, the river in 1896 meandered and had many types of landscape (e.g., pool and riffle). The river in 1896 flooded over river terrace and frequently formed large floodplains. Virtual fish behavior simulated by the model indicated that virtual fishes could utilize much more pools in 1896 comparing to the river in 2000 during ordinary water level and they could easily find low velocity area during flood, which minimize the probability of being flushed. Our assessment showed that the construction of continuous dike along the river banks limited areas that are suitable for native fishes. |
