Garrison: Interactions between marine mammals and pelagic longllne fistiing gear 



413 



of pilot whale interactions with longline 

 gear. A linear term for water temperature 

 was also highly significant and, on inspec- 

 tion of initial regression results, there 

 was a unimodal relationship between 

 water temperature and interaction rate. 

 Water temperature ( chi-square = 8.53, 

 df=l, P=0.0035j and its second-order term 

 (chi-square=7.87, df=l, P=0.0050) were 

 included in the model and were statisti- 

 cally significant. There was also an appar- 

 ent nonlinear relationship with mainline 

 length, with a step increase in interac- 

 tion rate for mainlines greater than 20 

 nautical miles in length. The continuous 

 mainline length variable was replaced 

 with a binary response (mainline >20 

 miles). This variable was highly signifi- 

 cant (chi-square=13.24, df=l, P=0.0003i 

 and provided greater explanatory power 

 and a better model fit than the continu- 

 ous variable. The final model was highly 

 significant (total chi-square=146.67, df=6, 

 P<0.0001l, and both examination of resid- 

 uals and goodness-of-fit tests indicated 

 strong model fit. 



Interactions with pilot whales were 

 much more likely to occur in the MAB 

 fishing area than in other regions. Both 

 mainline length and damage to sword- 

 fish catch on the set also significantly 

 increased the probability of observing a 

 pilot whale interaction (Fig. 3). There was 

 a strong decrease in the probability of a 

 pilot whale interaction with increasing 

 distance from the 200-m isobath (Fig. 4), 

 and this trend was apparent in the MAB 

 and other regions. No pilot whale interac- 

 tions were observed in sets greater than 

 30 km away from the 200-m isobath. The 

 correlation with water temperature was 

 generally weaker than that for other fac- 

 tors; however, there was a peak in inter- 

 action rates in warm waters between 70 

 and 80°F in the MAB and between 75 and 85°F in the 

 other geographic areas. This warm water peak was as- 

 sociated with late summer and fall in these regions. 



Logistic regression results: Risso's dolphins 



Risso's dolphin interaction rates were significantly 

 higher in the MAB (chi-square=4.56, df=l, P=0.0328) 

 and the NEC fishing areas (chi-square = 13.03, df=l, 

 P= 0.0003) than in areas farther south. The highest 

 interaction rates overall were observed in the NEC. As 

 with pilot whales, distance from the 200-m isobath (chi- 

 square = 4.58, df=l, P=0.0306) and damage to swordfish 

 catch (chi-square=4.68, df=l, P=0.0297) were important 

 explanatory factors. However, overall, the strength of 

 these effects was weaker than that for pilot whales. 



A 



m 



I 



 Pilot whales 



D Risso's dolptiin 



D Other marine mammals 



Q 



tnM] 



1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 



Year 



B 



Ini 



Lll l l^ 



Jan-Feb Mar-Apr May-Jun Jul-Aug Sep-Oct Nov-Dec 



Months 



Figure 2 



Annual (A) and seasonal (Bl patterns in observed marine mammal 

 interaction rates (number of positive sets/total number of observed 

 sets) for pilot whales (Globicephala spp.), Risso's dolphins {Grampus 

 griseus). and other marine mammal species. 



The mainline length effect, although improving the 

 explanatory power of the model, was not a statistically 

 significant factor (chi-square = 3.284, df=l, P=0.07). 

 The use of fish bait, either alone or in addition to squid 

 bait, significantly decreased the probability of observ- 

 ing a Risso's dolphin interaction (chi-square=4.33, df=l, 

 P=0.0375). The overall model was highly significant 

 (chi-square=71.68, df=6, P<0.0001) and provided a good 

 fit to the data. 



The bait effect was apparent only in the observed 

 data from the MAB region and the southern Atlantic 

 areas and was not observed in the NEC region, but the 

 positive correlation with damage to swordfish catch 

 was observed across all regions (Fig. 5). The negative 

 correlation with distance from the 200-m isobath for 

 Risso's dolphins was somewhat weaker than that for 



