Parrish and Kazama Ghost fishing in the Hawaiian lobster fishery 



723 



Oc^ 5.42, P 0.02). This was not true for the field trials 

 (x^ 2.06, P 0.15), and mortalities in the laboratory and 

 field were significantly different (x" 4.74, P 0.03). 

 Consequently, the two test situations were evaluated 

 separately, with all animals that died excluded from the 

 trap-occupation analysis. 



Contingency tables were used to test for differences 

 in the exit distributions for various groups of the data 

 (Fienberg 1987). Within each species, the distributions 

 of exits were first compared between replicate trap 

 strings within the same habitat type and were found 

 not to be significantly different (spiny lobster— high 

 relief, x" 3.22, P 0.50; even substrate, x" 10.00, P 

 0.19; slipper lobster-high relief, x" 3.33, P 0.50; even 

 substrate x" 9.52, P 0.22). Therefore, the distributions 

 of exits for the two replicates were pooled within each 

 habitat type. Within each species, exits were then com- 

 pared for the effect of the two habitat types and were 

 found not significantly different (spiny lobster— x" 

 10.81, P0.21; slipper lobster-x" 4.53, P 0.72). The 

 distributions of two habitat types were then pooled 

 within each species, and exits of the two species were 

 not significantly different (x^ 16.93, P 0.08). Conse- 

 quently, the distributions of the two species in all field 

 trials were combined (A'^ 126 lobsters after 2 early mor- 

 talities) for further comparisons. 



Within each species, exits observed in the tank were 

 compared with the field data pooled by replicate and 

 habitat type and were not significantly different (spiny 

 lobster-x^ 14.42, P0.21; slipper lobster-x^ 13.63, 

 P 0.09). Exits of spiny and slipper lobsters in the tank 

 were not significantly different (x" 11.55, P 0.32), and 

 the data were subsequently pooled (A^ 59 lobsters after 

 5 early mortalities). A comparison of exits of all lobsters 

 in the tank (pooled) and all lobsters in the field (pooled) 

 showed a significant difference (x" 23.889, P 0.03). 



Half of the 126 spiny and slipper lobsters stocked in 

 the field and 33% of the lobsters stocked in the labor- 

 atory exited within 48 hours after being placed in the 

 traps. Ninety percent or more of the exits in both tank 

 and field trials occurred within the first 16 days. All 

 field animals had left by day 30, and all laboratory 

 animals by day 56. The overall exit pattern of the 

 lobsters suggested an exponential model. The data 

 were fitted to the log-transformed exponential function 



In, (Nt/N,) = bt, 



where Nt is the number of lobsters remaining after 

 time t from Nq lobsters initially stocked. The param- 

 eter b was estimated with a log-linear regression pro- 

 cedure for the field traps (b= -0.16/day; r^ 0.992, 

 SE 0.284, P<0.001) and for the laboratory traps (b = 

 -0.094/day, r^ 0.961, SE 0.632, P<0.001; Fig. 2). 



Stocked spiny and slipper lobsters exited and re- 

 entered field traps in at least 13 instances; 6 lobsters 

 returned to the same trap. One spiny lobster was 

 observed exiting three traps within 6 days. 



Discussion 



Trap stability 



The lack of structural damage and appreciable move- 

 ment of the plastic traps in the field contrasts with the 

 popular opinion of experienced fishermen that lost 

 traps break up and roll off the banks into waters 

 beyond the depth range of lobsters. Fishermen routine- 

 ly report movement of trap strings as a result of power- 

 ful swells moving across the commercial banks. Despite 

 the frequently observed movement of the groundline 

 by swells at the Oahu study site, it is likely that the 

 study site does not fully duplicate the power of the long- 

 term swells common in the NWHI. Lost traps may not 

 shift on the bottom as much as actively fished traps. 

 Buoys and interconnecting polypropylene line provide 

 additional lift and resistance to water motion; there- 

 fore, fully rigged strings of traps are more likely to 

 move than isolated traps severed from the groundline. 

 A 1990 systematic diving survey of 33 sites around 2 

 of the prominent NWHI commercial lobster fishing 

 banks revealed only 2 mangled derelict traps (F. Par- 

 rish, unpubl. data). The failure to locate large amounts 

 of lost gear may be partly explained by this survey be- 

 ing incidental to other work. Total gear losses in 1989 

 averaged about 1 trap/nm- over the total estimated 

 area of the lobster fishing grounds (Landgraf et al. 

 1989). Lost gear could be heavily concentrated in a few 

 of the more intensively fished areas that the survey 

 may have missed. 



A trap manufacturer (Fathoms Plus Marine Imple- 

 mentation) has made available a corrodible pin which 

 is intended to allow the halves of plastic traps to even- 

 tually fall apart once the pin deteriorates. The fact that 

 our pinless trap remained relatively intact for 6 months 

 in the field suggests that the synthetic plastic clips on 

 the trap roof will continue to hold the trap together, 

 especially for fisheries conducted in calmer seas. 



Mortality and movement of lobsters 



Seven deaths among the 192 spiny and slipper lobsters 

 within the 56-day study represent low mortality when 

 compared with the natural mortality estimates from 

 the fishery population modeling by Haight and Polovina 

 (1992). Extrapolation of the experiment's percentage 

 of mortality from 2 months to 1 year (22%) is close to 

 half the fishery's annual estimated natural mortality 

 (40%). The fact that only animals that began the trials 



