1). Evidently 0.9 kg monofilament attachment did 

 not offer sufficient resistance for the hook to pierce 

 the jaw and tag the fish. Furthermore, it did not take 

 an exceptionally large fish to detach a tag because the 

 returned tag with 2.7 kg monofilament was removed 

 from a 3.2 kg fish. 



Hook type and the hook type-tagging location in- 

 teraction approached the 0.05 significance level as 

 sources of variation in the ANOVA (Table 2) because 

 higher proportions of straight hooks were detached, 

 except for tagging longlines set at the east Hudson 

 site where slightly higher proportions of circle hooks 

 were lost (Table 3). These results suggest that fish 

 are more easily hooked by straight hooks. However, 

 all tags returned had circle hooks (Table 1); thus, 

 although straight hooks tagged more fish, they ap- 

 parently did not remain in the jaws as well. 



The returned tags (7) represent about 27c of the 

 maximum number theoretically deployed (384). 

 However, because an unknown number of tags were 

 observed to be lost due to fouling during setting and 

 retrieval, the true rate of return is >2%. This return 

 rate is comparable with that reported for marine tag- 

 ging studies on relatively deep-dwelling reef fishes 

 using conventional tags applied at the surface 

 (Grimes etal. 1982b). Our tagging technique appears 

 to be useful over relatively long periods. Tagged fish 

 were at liberty from 115 d (0.32 yr) to 577 d (1.6 yr) 

 (Table 4). Similarly, Phillip's (1968) only detachable 

 tag return was from a marked kelp bass at liberty 

 about 2 yr. All of our returns suggest that tilefish in 

 the vicinity of Hudson Canyon are relatively seden- 



test monofilament). Studies using this procedure 

 could be relatively inexpensive because the major ex- 

 pense in most marine fish-tagging studies — vessel 

 time — would be eliminated. However, problems with 

 the detachable tagging technique may make its use 

 questionable for determining population parameters 

 other than movement. As with other tagging pro- 

 cedures, mortality of tagged animals may be in- 

 creased, especially since tags are placed in the mouth 

 and could impair feeding. However, all recaptured 

 animals in our study were reported in good condition 

 with no obvious scars, wounds, or other signs of 

 stress. Gut hooking (swallowed hooks) may also 

 cause additional tagging mortality. In a longline 

 assessment study (Grimes et al. 1982a) about 47c of 

 all hooked fish seen (42) from a submersible were 

 gut-hooked. 



Unlike conventional tagging procedures, the 

 researcher using detachable tags does not know what 

 species (and their relative numbers) were marked, 

 other than the target species. This was not a problem 

 in our tilefish study because this fishery is virtually 

 monospecific; if detachable tags are administered via 

 a fishery, as in the case we described, tagging data 

 can be adjusted according to the relative abundance 

 of species in the catch. 



It may also be possible to use detachable tags to es- 

 timate other population parameters, such as total 

 mortality, if sufficient return data are available and 

 assuming that tags are not lost from fish over the ex- 

 perimental period. For example, mortality could be 

 estimated either from the ratio of numbers of tagged 



tary, which might be expected given that tilefish in- 

 habit (and presumably construct) extensive burrows 

 (Able et al. 1982). 



This tagging procedure may represent one of the 

 few workable procedures presently available for in- 

 vestigating deep-dwelling fish. Optimal tag design 

 could be determined by a preliminary study, as we 

 have demonstrated (e.g., the optimal detachable tag 

 for tilefish is constructed with a circle hook, serially 

 numbered and addressed, and attached with 1.8 kg 



TABLE 4. — Returns from tilefish tagged in the west 

 Hudson Canyon, 1979. n.a. = not available. 



665 



