780 



Fishery Bulletin 91(4). 1993 



trial to trial, which correctly reflected the lack of inde- 

 pendence in the actual estimates. The kill values for 

 all other years were sampled independently. 



Results 



Estimates of abundance and kill 



The five years of the MOPS surveys resulted in 236 

 sightings of eastern spinner dolphins used in the abun- 

 dance estimate. The abundance estimate based solely 

 on these sightings was 568,100. Adding prorated num- 

 bers of unidentified spinner and unidentified dolphin 

 sightings resulted in a final estimate of 632,700, with 

 a CV of 0.167 (Table 1). The fisheries kill estimates 

 ranged from a high in 1961 of 138,000 to a low in 1983 

 of700(Table2). 



Contours of relative population size (N,./N h ) as a func- 

 tion of R„, and MNPL ranged from 0.35 to 0.55 (Fig. 2). 

 Relative population size increased with both R„, ( growth 

 rate) and MNPL (the amount of non-linearity in the 

 density-dependence response). The lowest relative 

 population size was 0.32, for the case of #,,,=0.00, (i.e., 

 no net growth in the population before fisheries kill 

 was included). The highest relative population size was 

 0.58 for the case of the highest growth rate and MNPL 

 (0.06 and 0.80, respectively). These low and high esti- 

 mates of relative population size correspond to esti- 

 mates of pre-exploitation abundance of 1,956,000 and 

 1,100,000, respectively. Relative population size in- 

 creased by approximately 0.03 for every increase of 

 0.01 in R m . The influence of MNPL was greater at 

 higher growth rates, as relative population size in- 

 creased by approximately 0.02 for every increase of 

 0.10 in MNPL at #,,=0.02, but increased by approxi- 



mately 0.05 for every increase of 0.10 in MNPL at 

 #,,,=0.06. There were no combinations of parameter 

 values such that relative population size was estimated 

 to be above MNPL. 



The upper 957c confidence limit for relative popula- 

 tion size as a function of R,„ and MNPL, based on the 

 sampling error of the abundance and kill estimates, 

 ranged from 0.45 to 0.91 (Fig. 3). The upper confidence 

 limit was always above MNPL when R,„ was greater 

 than 0.046 (Fig. 3, shaded region). The lower 95% con- 

 fidence limit for relative population size as a function 

 of R„, and MNPL, ranged from 0.22 to 0.36 (Fig. 4). 



All population trajectories declined until 1973 (Fig. 

 5), after which the estimated fisheries kill declined 

 substantially (Table 2). For the highest growth rate, 

 the population trajectory showed an increasing trend 

 from 1976 to 1988 (Fig. 5, line C), whereas for the 

 lowest growth rate the model resulted in a relatively 

 stable population level between 1976 and 1988 (Fig. 5, 

 line A). 



The confidence limits around relative population size 

 broadened with increasing R m . For example, for a 

 MNPL of 0.60, the confidence limits ranged from 0.23 

 to 0.44 for #,,,=0.00, whereas they ranged from 0.33 to 

 0.72 for #,,=0.06 (Fig. 6). As in Smith (1983), relative 

 population size was a linear function of R m . 



Discussion 



For all parameter values of R r „ and MNPL equal to 

 those in Smith ( 1983), estimates of relative population 

 size were higher in this analysis. For example, for 

 R„ =0.03 and MNPL=0.65, Smith (1983) reported a rela- 

 tive population size of 0.20 versus a result of 0.42 



