32 



Fishery Bulletin 97(1), 1999 



(0 

 XI 



o 



0.1 0.12 014 0.16 0.18 02 0.22 "0.24 



0.06 008 0.1 012 0.14 0.16 18 0.2 22 0.24 0.26 



q X 10^[1/trapx month)] 



2.000 2,600 3.000 3,500 4.000 4 500 5.000 



Stock biomass in year 1969-70 (t) Maximum sustainable yield (t) 



Figure 5 



The distributions of the parameters estimated with the bootstrapped least median of squared errors (LMSE) for the period of 

 1969-70 to 1993-94. 



stock biomass and CPUE is the same for the major- 

 ity of years. 



Compared with traditional least-squares estima- 

 tion methods, the LMSE method tends to fit the 

 model to the majority of the data, and its estimates 

 of parameters are not affected by atypical data ob- 

 served in a few years (Chen et al., 1994). Thus, if 

 atypical observations arise, parameter estimation 

 with the LMSE will not be affected greatly. Because 

 the LMSE is not sensitive to atypical data, these data 

 tend to be far from the LMSE-estimated CPUEs 

 (Figs. 2 and 3) and thus are readily detected. How- 

 ever, it is important to determine why an observed 

 CPUE is far from the predicted CPUE. Such a prac- 

 tice requires extensive background information on 

 the fishery. Apart from exceptionally large observa- 

 tion errors, atypical observations may arise from 

 unusual environmental conditions and substantial 

 changes in fishing methods and locations. The na- 

 ture of atypical observations resulting from unusual 



environmental conditions differs from those result- 

 ing from unusual observation errors. These atypical 

 data should be considered separately from those ob- 

 served under normal conditions in the stock assess- 

 ment process because such unusual conditions last 

 only for a short period of time in the development of 

 the fishery (Chen et al., 1994). For example, the ob- 

 served CPUE in 1971-72 was much higher than the 

 LMSE-estimated CPUE (Fig. 4). This might result 

 from the fishery being expanded in that year to in- 

 clude aggregations of lobsters on the previously 

 unexploited slopes of the continental shelf. This 

 would result in a high q for that year. Clearly, such a 

 high level of CPUE cannot be sustained for long and 

 should not be interpreted as an indicator of large 

 stock biomass for that year CPUEs observed from 

 years 1917 to 1924 were much lower than the LMSE 

 predicted ( Fig. 2 ). This finding may be related to large 

 observation errors in CPUEs. For this period, the 

 number of fishing boats (other than trawlers or Dan- 



