Brock and Ward: Octopus bycatch and lobster mortality In the South Australian rock lobster fishery 439 



J. edwardsii (O. maorum), (Gardener 5 ), and a localized 

 study in the New Zealand fisheries for J. edwardsii 

 (O. maorum) found the proportion of the lobster catch 

 killed by octopus to be as high as 10% (Ritchie 6 ). The 

 estimates of lobster mortality from these other studies 

 should be treated with caution however because the cur- 

 rent study is the only one that documents within-trap 

 lobster mortality from a fishery-wide data set. 



The general linear modeling approach that we used to 

 determine the factors associated with M L has some limi- 

 tations. For example, the logbook data for the SARLF, 

 like the monitoring data for most other fisheries, are 

 not completely independent, and interdependence among 

 observations can bias estimates of parameters. Simi- 

 larly, some of the factors in the model, notably CPUE L 

 and CPUE are partially correlated. In addition, the 

 large number of observations and degrees of freedom 

 tend to make most factors significant. We considered 

 all of these issues when interpreting the results of the 

 analyses and used the mean square (MS) values to rank 

 the importance of factors. 



In both zones, inter- and intra-annual fluctuations 

 in M L largely reflect the effects of CPUE and CPUE L . 

 The broad trends in annual CPUE have largely cor- 

 responded to those for M, with peaks in both generally 

 synchronous in both fishing zones. In the SZ, the gen- 

 eral increase in M L since 1983 appears to result from 

 the increase in CPUE, which has more than doubled 

 over this period. This assessment is supported by catch- 

 rate data from individual MFAs. The two MFAs in the 

 SZ that have had the greatest increases in CPUE L over 

 the last 5 years (56 and 58) have also had the highest 



5 Gardener, C. 2002. Personal commun. Tasmanian Aqua- 

 culture and Fisheries Institute, Private Bag 49, Hobart, 

 Tasmania 7001. 



6 Ritchie. L. D. 1972. Octopus predation on trap-caught rock 

 lobster — Hokianga area, N.Z. September-October 1972. New 

 Zealand Marine Department. Fisheries Technical Report 81, 

 40 p. (Available from Ministry of Fisheries, 101-103 The 

 Terrace, Wellington, New Zealand, 1020.] 



corresponding increase in M L . Increases in CPUE L are 

 likely to elevate M L by both increasing the probability 

 of octopus encountering traps containing lobsters and 

 the number of lobsters in traps entered by octopus. 



However, M L is also positively correlated with soak- 

 time, especially in the SZ. This finding is consistent 

 with patterns observed in the New Zealand fishery for 

 J. edwardsii 6 and reflects the increased opportunities 

 for octopus predation when pots containing lobsters 

 remain in the water for longer periods. In the SZ, fish- 

 ermen return to port each day and choose to fish or not 

 to fish each day according to factors such as weather 

 and price; therefore, although a 24-h soak period is still 

 most common, soak times can range from one to five 

 days. In the NZ, fishermen remain at sea for extended 

 periods and consequently soak times longer than 24 

 hours are rare. 



There was considerable variation in the fishery data, 

 especially in the southern zone. It is likely that much 

 of this variation is related to the large geographical 

 extent of the fishery as opposed to fishing practises. 

 Across the fishery lobster growth rates and subsequent 

 catch rates vary greatly (McGarvey et al., 1999a). For 

 example, since 1991, the CPUE L in MFAs 56 and 58 

 have been twice those of MFAs 51 and 55 in the SZ. 

 Although the variation in CPUE between the zones 

 has been similar, the higher variability in CPUE L in 

 the SZ is reflected in the variation in M L . 



Data spanning 17 years and covering about 50.000 

 km 2 represent one of the few long-term and large-scale 

 data sets on the distribution and abundance of an octo- 

 pus species (Hernandez-Garcia et al., 1998: Quetglas et 

 al., 1998). The paucity of octopus studies on these scales 

 reflects the logistical constraints of fishery-independent 

 surveys of octopus populations and the poor and incon- 

 sistent methods generally used to record fishery catch 

 and effort data (Boyle and Boletsky, 1996). The few data 

 that are available on the distribution patterns of octo- 

 pus have been obtained mainly from small commercial 

 fisheries and CPUE ( , has been included as a measure 

 of relative abundance (Defeo and Castilla, 1998; Her- 

 nandez-Garcia et al., 1998). This approach has proven 

 useful, but several potential biases must be considered 

 when CPUE data are being interpreted: these include 



1) changes in fishing methods and efficiency over time; 



2) the distribution pattern (e.g., random or aggregated) 

 of the species under consideration; and 3) spatiotempo- 

 ral fluctuations in catchability (Richards and Schnute, 

 1986; Rose and Kulka, 1999). There are several reasons 

 why the data from the SARLF may provide a useful 

 measure of the relative abundance of octopus over these 

 spatial and temporal scales. Most importantly, the ba- 

 sic unit of effort in the fishery, the trap, has remained 

 unchanged since 1983. Furthermore, although O. mao- 

 rum is retained as bycatch and kills J. edwardsii in 

 traps, it is neither targeted nor avoided by fishermen, 

 and fishing effort is thus relatively independent of its 

 distribution patterns because the economic effects of 

 both the sale of octopus bycatch and the costs of lobster 

 predation are relatively small compared to the primary 



