During the fall of 1999, a fish diversion system utilising high-frequency sound was installed at the Annapolis Tidal Generating Station in Nova Scotia to test its the effectiveness of reducing the passage of fish, primarily anadromous Alosa, through the turbine tube. The fish diversion system tested was a band-limited random noise signal projected into the turbine forebay by 4 transducers mounted across intake. The signal pulse was presented at a 33% duty cycle (0.5 seconds on, followed by 1 second off) with the majority of energy focused between 122 and 128kHz. The 160 dB threshold was reached at a distance of 10 to 12 m from the intake face.

The system was partially effective for Alosa, but not for the other 11 species for which data were analysed. With the system turned on, the rate of passage of American shad through the turbine tube was 35% less than when the system was turned off. The rate of blueback herring passage apparently only decreased (26%) if the largest catches are excluded from the analysis, suggesting that effectiveness is decreased when large schools of fish are present. The rate of passage through the fishway located nearest the turbine increased by factors of 3.6 times for American shad and 4.1 times for alewife when the diversion system was turned on. The diversion system apparently did not affect the rate of fish passage through the fishway located by the sluice gates. The effectiveness of the system could potentially be improved by angling the barrier towards the fishway located near the turbine.   

On an opportunistic basis we explored methods of estimating turbine mortality at the TGS. As an alternative to direct measurement of handling mortality, a significant source of error in turbine mortality studies, we used a logistic model to relate the probability of death to the duration of the net deployment. The intercept in this model is an estimate of turbine mortality. The model appeared to provide believable estimates for robust species, but requires that an offset be fitted for fish that are more susceptible to damage from handling. Preliminary results suggest that turbine mortality is between 0.0 and 6.3 % for alewife, sea lamprey, Atlantic silversides, pipefish, winter flounder and windowpane. Some species, such as Atlantic silverside, show a preference for passing seaward through the new fishway. 

Development of behavioural guidance systems can be costly and are without a guarantee of success. Ultimately, the effectiveness of a fish diversion system should be measured using the population-level response. For a species with a high compensatory capacity and low mortality associated with passage at the causeway (fishway usage coupled with turbine mortality), the effect of diversion at the population-level may be negligible. If a multi-species diversion system is to be developed, target species should be chosen based on a population-level risk assessment, to avoid the developmental cost for species for which the benefit may be minor.