LO: MORTALITY RATES OF NORTHERN ANCHOVY 



Table 2.— Estimates from multi-equation model (MEM), single-equation model 

 (SEM), and maximum likelihood estimator (MLE) for anchovy egg and larval mor- 

 tality {a and /?), and their standard error (SE) based upon 1980-83 field data where 

 K is number of age groups and m is number of tows used in each model. For both 

 MEM and SEM, nonlinear regression (NR), linear regression (LR) and weighted 

 nonlinear regression (WNR) estimates are given. 



plications where it is preferable to estimate IMRs 

 for egg and larvae simultaneously (e.g., simulation 

 studies of mortality at all life stages). The SEM is 

 preferable for modeling the mortality curves 

 through all life stages because it eliminates the 

 multiple estimates that occur at the endpoint of each 

 life stage (Fig. 2). In addition, the SEM allows 



estimation of the IMRs for all life stages even when 

 data for some life stages are inadequate for indepen- 

 dent estimation of a life-stage-specific IMR. In com- 

 paring NR and LR, the estimates of IMRs from 

 these two procedures were similar, despite the dif- 

 ferent assumptions about the error term. One com- 

 plication of using LR is that the abundance for any 



403 



