In Rivers Inlet, on the British Colombia coast, the sockeye salmon population relies on naturally spawning salmon to maintain the population. To maintain large salmon populations, it is thus necessary for a certain number (somewhere between 250,000 and 1.5 x 10⁶) of fish to escape fishing each year. Because of the economic importance of the fishery, it is important for the number of salmon returning each year to be as large as possible. For this reason, and because very low salmon populations were recorded several times between the late 1950s and mid-1970s, the Department of Fisheries and Oceans (DFO) decided to experiment with management, halting fishing in the inlet in the hopes of rebuilding salmon populations and learning how many spawning salmon were necessary to yield the maximum population size. Unfortunately, the inherently large variability in salmon populations and uncertain methods for counting salmon meant that five years into the project, they still were not able to determine whether more spawning salmon were resulting in higher populations. Thus, a workshop was held by the DFO in which their own scientists worked with fishing industry representatives to design a new plan for managing the salmon harvest. Participants used mathematical models, either hand calculations or simple computer simulations, to examine the effects of different harvest strategies on salmon populations. These models structured the communication about different potential options, and allowed them to reach agreement on a new strategy in which fishing was allowed to resume. This strategy also pursued the original AM experiment by occasionally allowing large numbers of spawners to escape fishing, but over a much longer time frame (60-80 years). The result of collaboration in this case was not ideal from a scientific, and perhaps a long-term economic, standpoint. The authors suggested that the original lack of collaboration and inadequate assurances of meaningful results were largely responsible for the lack of support for the experimental policy. The authors recommended that collaboration be included throughout the various stages of the AM process.