large percentage of small males (including many smaller than 15 cm) were 

 ripe during the spawning season and thus were easily sexed. But when 

 spent, these fish may have been sexed as unknown because their scales 

 were so small and weakly ctenoid that they couldn't be felt. This could 

 have accounted, in part, for the large increase of recaptured unknowns 

 and decline in the number of males. 

 Other assumptions or restrictions 



6. Estimates are reliable only when 9% or more of the population is 

 sampled (Bishop and Sheppard 1973; Southwood 1978). 



7. For accurate results the coefficient of variation (SE of N/N) 

 should be less than 5% (Rof f 1973) . 



8. Probability of survival (phi) greater than 1.00 indicates a major 

 error (Southwood 1978). 



Except for 197 5 (6.6%), 11.6 to 23.4% of each year's estimated 

 population was sampled (Tables 2-8) . The coefficients of variation, 

 however, ranged from 8.0% in 1976 to 36.6% in 1977. Roff (1973), however, 

 did note that the 5% level of precision was almost never obtained in 

 tagging studies, so this restriction is probably of little concern. 



Begon (1979) pointed out that estimates of survival probability 

 (phi) of greater than 1 as well as the calculation of negative numbers 

 joining (B, usually referred to as births in most Jolly models) were 

 biologically impossible values. He stated that since phi and B are 

 estimates, they are subject to error and that imperfect data produces 

 impossible estimates. He recommended that the best estimate of any phi 

 greater than 1 would be 1.00 and that for negative B would be 0. Negative 

 gains should not be interpreted as losses and vice versa. 



23 



