Lindley and Mohr; Impact of population manipulations of Morone saxatilis on Oncorhynchus tshawytscha 327 



log(^,) of only 2.6x10"^ per generation less than 

 that at a stock size near zero. At the recovery 

 target population size of 20,000 adult winter-run 

 Chinook salmon, in contrast, the median estimate 

 of /3 corresponds to a population growth rate re- 

 duction of 0.13 per generation. 



The fit of the model was assessed by comparing 

 the observed spawning escapement data series to 

 the posterior predictive distribution of W (Gelman 

 et al., 1995), which was estimated by drawing 

 10,000 samples from p(0\-) and a normal(0,cT^) 

 and applying Equations 1-3. Figure 5 shows the 

 observed data and boxplots of the posterior pre- 

 dictive distributions for the data points. Observed 

 escapement in 1980 and 1991 was below the fifth 

 percentile of the distribution for predicted escape- 

 ment for those years. Winter-run chinook salmon 

 returning in 1980 and 1991 were born during 

 the drought years of 1976-77 and 1987-88. The 

 1976-77 drought was particularly severe; there 

 were very low river flows and water temperatures 

 exceeded 21°C during the winter-run chinook 

 salmon egg incubation period, well above the 

 50% mortality temperature of 16°C reported for 

 chinook salmon (Alderice and Velsen, 1978). The 

 association between these outliers and droughts 

 suggests that the model does not accurately handle 

 an important source of risk. The estimate of a was 

 influenced by the 1980 and 1991 escapements, but because 

 critically dry years appear to reduce survival more than wet 

 years increase it (as suggested by the lack of large positive 

 deviations in Fig. 5), estimates of absolute extinction risk 

 may be optimistic. Our focus, however, is on the relative risk 

 of extinction under different management scenarios. 



Extinction risk estimation and stocking plan analysis 



Figure 6 shows the cumulative distributions of quasi-extinc- 

 tion and recovery probabilities under the three striped bass 

 stocking levels predicted by the density-dependent model. 

 Winter-run chinook salmon have a 28% chance of becoming 

 quasi-extinct and a 11% chance of recovering to more than 

 20,000 adults in 50 years, if no striped bass stocking were 

 to occur (Table 2). If a striped bass stocking program were 

 to stabilize the striped bass population at 700,000 adults, 

 the probability of quasi-extinction in 50 years would rise 

 from 28% to 30% (5=1.9%, 0.9 CI=|1.2%, 2.6%]), and the 

 probability of recovery would decline from 11% to 10%. 

 A future adult bass population of 3.0x10*^ would raise the 

 chance of winter-run chinook salmon quasi-extinction to 

 55% (5= 27.7%, 0.9 CI=125.4%, 30.1% 1 1 and lower the recov- 

 ery probability to 3.8%. If, on the other hand, striped bass 

 predation could be eliminated completely, the probability 

 of quasi-extinction would decline to 23'7( (5=-4.5%, 0.9 

 CI=[-5.6%, -3.4%]) and the probability of recovery within 

 50 years would rise to 14%. 



The probability of quasi-extinction according to the 

 density-independent model is quite similar to that of the 

 density-dependent model, but the predicted probability of 

 recovery is substantially higher with density independence 



1970 



1975 



1980 



1985 

 Year 



1990 



1995 



2000 



Figure 5 



Posterior predictive distributions (gray boxes and whiskers) and 

 observed winter chinook salmon spawning escapement (circles). 

 Gray boxes cover the middle 0.50 percentile interval, and whiskers 

 represent the middle 0.90 percentile interval. 



(Table 2). Extinction probability is somewhat more sensi- 

 tive to striped bass predation in the density-independent 

 model. This greater sensitivity to striped bass abundance 

 results from the higher estimate for the bass predation rate 

 parameter and the lack of compensation in the density- 

 independent model. 



The density-dependent model indicates that without 

 further population growth rate increases, winter-run 

 chinook salmon are unlikely to reach the recovery bench- 

 mark: recovery will not occur within 20 years, and there 

 is less than an \Y'< chance of reaching the 20,000 adult 



