592 



Fishery Bulletin 89(4). 1991 



ft* 



/ N C\^ 





\'///\ Spawning ground 



Kya Australian fishing grounds 



Figure 1 



Geographical distribution of the 

 Australian and Japanese (inset) 

 southern bluefin tuna fisheries. Ar- 

 rows indicate movement of fish in 

 and away from Australian coastal 

 waters. Black squares on the inset 

 map indicate areas of higher Japa- 

 nese catch. 



During the 1980s, scientists became concerned that 

 the fishery was being overexploited, and recent stock 

 assessments have indicated the possibility that a decline 

 in recruitment to the surface fishery could have oc- 

 curred (Caton et al. 1990). These stock assessments 

 have been based almost exclusively on cohort analysis 

 (Gulland 1965) and other age-structured models, all of 

 which require an estimate of M. 



Despite its importance in stock assessment, the 

 natural mortality rate of southern bluefin tuna has 

 never been investigated in detail. Hayashi et al. (1969) 

 considered a value of 0.2/year to be appropriate, based 

 on a somewhat ad hoc comparison with growth param- 

 eter estimates. This value of M has been preferred in 

 most stock assessments, although a range of values has 

 also been used at various times (e.g., Hampton et al. 

 1984, Hampton and Majkowski 1986). More recently, 

 the application of the Hearn et al. (1987) method to 

 southern bluefin tagging data suggested that M could 

 be somewhat higher than the "traditional" value of 

 0.2/year, possibly in excess of 0.4/year. However, the 

 authors acknowledged that this was probably an over- 

 estimate because of likely non-compliance with various 

 assumptions. 



In this paper, selected southern bluefin tuna tagging 

 experiments are analysed for the purpose of deriving 

 estimates of M consistent with the tagging data. The 



Hearn et al. (1987) model (referred to as the HSH 

 model) and an extension of the Sibert (1984) model 

 (referred to as the SE model), which also provides 

 estimates of movement rates between geographically 

 separated fisheries and catchability coefficients, are ap- 

 plied. These models are tested using a simulation model 

 that reflects characteristics of the southern bluefin 

 population and fisheries. 



Tagging data 



Between 1959 and 1984, the CSIRO supervised the tag- 

 ging of more than 60,000 southern bluefin tuna, most- 

 ly aged 2-4 years, of which approximately 12,000 have 

 been recaptured and the tags returned. From these 

 data, four tagging experiments from which no further 

 returns are expected were defined (Table 1). Although 

 other groupings of the tagging data are possible, the 

 experiments were defined in this way so that they 

 represent reasonably homogeneous groups in terms of 

 release area, size of tagged fish, time period of release, 

 fishing method, and tagging personnel involved. All 

 fish were double-tagged using methods described by 

 Williams (1982). All returns from these experiments 

 were included in the analyses, with best estimates of 

 recapture times (normally the midpoint of the fishing 



