Modelling the effects of dams on freshwater fish distributions in Hokkaido, Japan

1. We examined the effects of habitat fragmentation caused by dams on freshwater fish species using a database of 7848 fish presence/absence surveys, conducted between 1953–2003 in Hokkaido, Japan. 2. A series of generalised linear models showed that for 11 of 41 taxa examined, the probability of occurrence had been influenced either negatively (eight taxa) or positively (three taxa) by the presence of a dam downstream from their habitat. 3. Maps of modelled predictions revealed that dams had had widespread negative impacts on certain taxa, while for other taxa the impact was limited to specific basins. Two of the three taxa whose probability of occurrence was increased in areas above dams have long been transplanted into reservoirs in Japan. 4. For four of the eight taxa whose probability of occurrence was reduced above dams and all three taxa whose probability of occurrence increased above dams, the temporal length of habitat isolation (i.e. the number of years between dam construction and sampling) was also a significant predictor of the probability of occurrence. This pattern indicates that these taxa experienced a gradual rather than an instantaneous population impact as a result of dam construction. 5. The eight taxa whose probability of occurrence was reduced as a consequence of dams all exhibit migratory life cycles. Although migratory taxa are probably more susceptible to the negative effect of dams, we could not detect significant relationships between migration life histories and the effect of fragmentation by dams. 6. These analyses enable stream and fisheries managers to quantify the impacts of habitat fragmentation because of dams for individual species. The spatially explicit nature of our analyses also enables identification of the areas of the impact at broad geographical scales. Using our results, managers can take effective conservation and restoration measures to predict, mitigate or remove the impact of dams. For example, our results can be used to prioritise dams for removal or to predict losses of biodiversity and ecosystem services in advance of dam construction.