Lin et al.: Sensitivity of models to bias and imprecision in life history parameters 
307 
Parameter magnifier (%) Parameter magnifier (%) 
Figure 1 
Relative changes determined from Monte Carlo simulation with data collected during 1998-2006 for female Japanese eel 
(Anguilla japonica) in the Kao-Ping River in southern Taiwan. The relative changes are measured as percentages for the 
means (shown in the left panels) and standard deviations (SD; shown in the right panels) of fishery-mortality-based refer- 
ence pointslFBRpg) in scenarios 2-6 of the yield-per-recruit model, where the mean and SD of the natural mortality (M and 
Em), the von Bertalanffy growth coefficient (K), the asymptotic length (L M ), and the multiplicative error in the growth curve 
(cgr) were under-specified from 5% to 95% and over-specified from 150% to 1000% (except forvalues of that went from 
50% to 200%). In each graph, the solid line indicates fishing mortality at which yield per recruit is at its maximum, and 
the dashed line indicates fishing mortality at which the increase of yield per recruit is only 10% of the increase of yield per 
recruit when fishing mortality is zero. Because of the large variation in RC values for the different scenarios, the scales of 
the y-axes differ greatly to allow changes to be seen. Under=underspecified parameters; Over=overspecified parameters; 
£M = the SD of M\ F cur =current fishing mortality and; e/p cur =SD of F cur . 
the RC changing from 70% to around 4000%. The pa- 
rameter K had high influence on the SD values of F^q% 
and ^ 50 %, with RC changing from 20% to 1500% (Fig. 
2, right panels). The parameters £m an d had con- 
siderable effects on the SD values of F%q% and F§q%, 
with RC ranging from 90% to 400% and from 40% to 
150%, respectively. The SDs of F^q% and ^ 50 % were not 
affected by £gr- 
Sensitivity of the composite risks of overfishing 
The composite risks of growth overfishing (F cur ex- 
ceeding F max and Fq 1 ) decreased with increasing 
means of M and K and did not depend on changes 
in L 0 o, £qr, and £m (Fig. 3). The risks of recruitment 
overfishing ( F cur exceeding F 30 % and F. 50 %) showed 
the greatest sensitivity to K followed by M. The risk 
of recruitment overfishing was affected also by chang- 
es in L«, but with less sensitivity. The parameters £m 
and £(jr did not contribute to noticeable changes in 
the risks of overfishing. The risks of both growth and 
recruitment overfishing showed strong sensitivity to 
both the mean and SD of F cur . As the mean of F cur 
increased from around 150% to 225%, the risks ap- 
proached 100%. The risks of overfishing decreased 
with increasing SD for F cur . 
