Abstract. - Two models were fit 

 to data from four experiments in 

 which tagged southern bluefin tuna 

 Thunnus maccoyii were released in 

 Australian coastal waters and recap- 

 tured in the Australian surface fish- 

 eries and the Japanese longline fish- 

 ery. The principal objective of the 

 analysis was to estimate the instan- 

 taneous rate of natural mortality (M). 

 Movement rates and catchability co- 

 efficients were also estimated using 

 the second model. The first model 

 (HSH) was fit to exact recapture 

 times assuming, inter alia, that the 

 tagged population was extinct at the 

 time of the last tag return. The sec- 

 ond model (SE) was fit to grouped 

 data classified by two release fish- 

 eries and three recapture fisheries, 

 explicitly incorporating movement 

 between the geographically separ- 

 ated release fisheries and permanent 

 emigration from the release fisheries 

 into the Japanese longline fishery. 

 Using the HSH model, estimates of 

 M ranged from 0.20 to 0.42/year for 

 the different experiments if full 

 reporting of recaptured tags was 

 assumed. The estimates decreased 

 slightly as assumed reporting rate 

 was decreased. The SE model yielded 

 estimates of M ranging from 0.20 to 

 0.23/year with M constrained to be 

 equal in each of the recapture fish- 

 eries. Unconstrained-M estimates 

 were obtained which suggested higher 

 levels of natural mortality in the re- 

 lease fisheries; however, these esti- 

 mates were considered unreliable be- 

 cause of their large standard errors 

 and high degree of confounding with 

 other parameters of the model. Sim- 

 ulation trials indicated that input 

 parameters used to generate simu- 

 lated tag-return data sets could be 

 accurately retrieved using the SE 

 model. However, the HSH model 

 produced positively biased estimates 

 of M because of the low level of ap- 

 parent fishing mortality in the Japa- 

 nese longline fishery. 



Estimation of Southern Bluefin 

 Tuna Thunnus maccoyii Natural 

 Mortality and Movement Rates 

 from Tagging Experiments 



John Hampton 



CSIRO Division of Fisheries. Marine Laboratories 



GPO Box 1538. Hobart, Tasmania 7001, Australia 



Present address: South Pacific Commission. B P. D5. Noumea Cedex. New Caledonia 



Manuscript accepted 15 April 1991. 

 Fishery Bulletin, U.S. 89:591-610 (1991). 



The rate at which members of a fish 

 population die from causes other than 

 fishing, the so-called instantaneous 

 rate of natural mortality (M), has a 

 large bearing on the population and 

 fishery dynamics of an exploited spe- 

 cies. Natural death is rarely directly 

 observable in fish populations, mak- 

 ing M notoriously difficult to esti- 

 mate, and much attention has been 

 given over the years to devising ex- 

 perimental and statistical techniques 

 to do so. The experimental technique 

 most frequently used is tagging and 

 recapture, the object of which, as 

 stated by Beverton and Holt (1957), 

 is ". . .to set up and examine the 

 properties of an 'experimental' popu- 

 lation of marked fish in which certain 

 parameters that would be difficult or 

 impossible to estimate in the 'natural' 

 population can be determined with 

 some accuracy". 



A variety of statistical techniques 

 has been developed to estimate M 

 and other parameters from tagging 

 experiments. One of the simplest is 

 that of Gulland (1955), who derived 

 maximum likelihood estimators for M 

 and the fishing mortality rate (F) 

 (both assumed constant over time) 

 for a completed tagging experiment 

 (i.e., one in which no tagged fish re- 

 main alive at the time of the last tag 

 recapture). Hearn et al. (1987) gener- 

 alized the Gulland (1955) method by 

 allowing F to vary with time. Other 

 models based on log-linear regression 

 (Sandland 1982) or maximum likeli- 

 hood estimation (Seber 1973, Wether- 



all 1982) normally require M and F 

 to be constant; however, some of the 

 more sophisticated methods allow F 

 to vary with time as a function of 

 fishing effort or catch (e.g., Lucas 

 1975, Kleiber et al. 1987). As an ex- 

 tension of these single-fishery models, 

 Sibert (1984) developed a two-fishery 

 model in which estimates of M, F, 

 and the rates of movement between 

 the fisheries could be obtained. 



The southern bluefin tuna Thunnus 

 maccoyii for many years has been the 

 subject of an important fishery in 

 southern Australian waters, where 

 juveniles form surface schools and 

 are captured principally by pole-and- 

 line and purse-seine gears. In addi- 

 tion, a large fleet of Japanese long- 

 liners has targeted the adults in the 

 higher, southern latitudes of the In- 

 dian, Southern, Pacific, and Atlantic 

 Oceans since 1952. 



Tagging experiments conducted by 

 the Australian Commonwealth Scien- 

 tific and Industrial Research Organ- 

 ization (CSIRO) have demonstrated 

 that these fisheries exploit a common 

 stock (Hampton 1989), and that there 

 is considerable movement of juvenile 

 fish between fishing grounds off the 

 south coast of western Australia 

 (WA), in the Great Australian Bight 

 off south Australia (SA), and off the 

 south coast of New South Wales 

 (NSW). Throughout the juvenile phase, 

 southern bluefin move away from the 

 Australian coast and subsequently 

 become recruited to the Japanese 

 longline fishery (Fig. 1). 



591 



