23 



the age composition* are randomly assigned to years that have no age 

 composition data and how production ratios are calculated. The 

 escapement in 1975 is the parent escapement of age 3 fish returning in 

 1978, ag« 4 fish in 1979, and age 5 fish in 1980. Since no age 

 composition data is available for 1978, a year with age composition data 

 is randomly chosen (1985) and that age composition is assigned to 1978. 

 Thus 5.4% of the escapement in 1978 (640 fish) is calculated to be 

 comprised of age 3 fish, or 34.6 fish. This results in an estimate of 

 1,000 fish escapement in 1975 producing 34.6 age 3 fish in 1978, or a 

 parent escapement to age specific return escapement of 0.0346. In a 

 similar manner, the ratio of the 1975 parent escapement to 1979 age 4 

 return escapement is 0.3610 and the ratio of the 1975 parent escapement 

 to the 1980 age 5 return escapement is 0.1170. This process is 

 continued for parent escapements through 1989 for age 5 returns, 1990 

 for age 4 returns, and 1991 for age 3 returns. The random selection of 

 age composition for years with unltnown age composition is repeated for 

 each of the 1,000 simulations. 



Table 9 demonstrates the forecast procedure for one simulation. A 

 parent escapement in 1992 will produce age 3 fish in 1995. Randomly, 

 one of the age 3 parent to age specific return escapement ratios 

 (0.0679) is chosen, which when multiplied by the 1992 escapement, yields 

 an age 3 escapement of 37.3 fish. Li)(ewise, the parent escapement in 

 1991 of 318 fish, after randomly choosing an age 4 parent to age 

 specific return escapement ratio of 0.8510, will produce 270.6 for the 

 age 4 e8cap>ement, and the parent eBcap>ement in 1990 of 78 fish, after 

 randomly choosing an age 5 parent to age specific return escapement 

 ratio of 0.1420, will produce 11.1 in the age 5 escapement. This 

 results in a total escapement of 319 fish in 1995. This process is 

 continued through the year 2089 to produce a simulated escapement of 899 

 fish. If the escapement had decreased to the specified level of 

 extinction (either 1 or 30 fish) the simulation is terminated and the 

 population is considered extinct. 



We also applied the exponential diffusion model (Dennis et al. 1991) to 

 the same four time series to estimate extinction likelihoods because 

 that is the method Waplea et al. (1991) used when Sna)ce River fall 

 chinoo)c were first listed. 



Waples et al. (1991) used running averages with the exponential 

 diffusion model; we used the actual escapements. Running the 

 exponential diffusion model provides a comparison to the results of the 

 bootstrap method and creates a linJc to earlier evaluations of the 

 extinction risk that Snake River fall chinook salmon face. 



The outlook for the Snake River fall Chinook salmon population using 

 escapement data truncated in 1990 is very pessimistic using either the 

 bootstrap method (Table 10 and Figures 7-11) or the exponential 

 diffusion model approach (Table 11). The results from the bootstrap 

 analysis estimates that the probability of extinction is 97.0% or 98.3% 

 (depending if the 1975-1990 or 1980-1990 data set is used, respectively) 

 if extinction is defined as 1 fish. If extinction is defined as 30 fish 

 (see Crammer and Neeley; 1993 for an explanation of the rational of this 

 value) then all simulations using the data through 1990 resulted in the 

 abundance falling below the extinction criteria. Using the exponential 

 diffusion model with data only through 1990, the probability of the 

 population abundance falling below extinction levels of 1 or 30 always 

 exceeded a probability of 99%. 



However, if the 1991 through 1994 escapement data is included in the 

 analysis, the outlook for the Snake River fall chinook salmon population 

 is substantially more optimistic. In the bootstrap analysis, none of 

 the simulations resulted in population abundances approaching 1 fish at 

 the end of 96 years. In fact, the probability of reducing the 

 population to 30 fish or less was 0.3%, 5.0%, and 4.4% for the 1980-1994 

 data series, the 1975-1994 data series, and the weighted analysis using 

 the 1975-1994 data series; respectively. None of these probabilities 

 meet the criteria for an endangered species as specified by Thompson 

 (1991). In fact, there is less than an 8% chance of the population 

 being fewer than 300 fish in 2089, less than a 22% chance of the 

 population being fewer than 1,000 fish in 2089, and over one-half of the 



