mm but the commercial nets were limited to 

 meshes ranging 110 to 130 mm. In 1979 we also 

 obtained some data from commercial catcher boats 

 of the mother ship Nojima Maru , but these were 

 used only in determining geographical differences 

 in catch rates. Nets were set at 16:30 h and hauled 

 at 02:30 h on Hoyo Maru No. 67, similarly to com- 

 mercial boats; on other cruises deployment and 

 retrieval were at 18:00 h and 03:00 h. Nets con- 

 sisted of panels, called tans, 50 m by 6 m deep that 

 were sewn together in series, with floats attached 

 to the topline and lead weights attached to the 

 bottom line of each panel. The stretched mesh thus 

 hung vertically like a curtain from the sea surface 

 down to a depth of 6 m. A commercial net was 

 usually about 15 km long, while those on our ves- 

 sels were about 6.5 km (130 tans). A commercial 

 fleet of 45 catcher boats (plus a mother ship) de- 

 ployed its nets in parallel lines on a grid, each net 

 about 5 nmi from neighboring ones. For several 

 sets in the central Bering Sea, Hoyo Maru No. 67 

 fished within the configuration of a commercial 

 fleet. A complete review of the Japanese drift gill 

 net fishery is contained in Fredin et al. (footnote 

 2). 



On all cruises except that of the Nojima Maru 

 catcher boats, we noted entangled birds as the nets 

 were being retrieved. Carcasses from the first 

 cruise of Hoyo Maru No. 67 were saved by Jones, 

 frozen, shipped to the United States, and, along 

 with those from Nojima Maru, were later iden- 

 tified by Ainley; those from other cruises were 

 identified immediately by Ainley and DeGange. 

 On most sets the mesh size and the depth at which 

 birds were entangled were recorded. 



Using data from the Hoyo Maru No. 67 and the 

 Nojima Maru catcher boats (see Figure 1), we com- 

 pared catch rates to distance south of the Aleutian 

 Islands. Data from Oshoro Maru and Hokusei 

 Maru were not used because they did not use com- 

 mercial nets. However, a comparison of catch rates 

 to the distance north of the Aleutians used data 

 exclusively from Oshoro Maru because we had no 

 data from commercial nets close enough to the 

 islands. The shape of the regression curve would 

 not be affected but the asymptotes would be dis- 

 placed downward due to the low catch rates of the 

 research nets used. 



On two cruises in 1979, Ainley and DeGange 

 censused seabirds that occurred within 300 m of 

 one forequarter of the ship whenever we steamed 

 at 9-12 kn during daylight; birds circling or follow- 

 ing the ship were ignored. Distance traveled dur- 



ing a given time period (a transect) multiplied by 

 the 300 m transect width provided an estimate of 

 ocean area surveyed. Area divided into the 

 number of birds counted resulted in an index of 

 bird density per square kilometer. For 16 sets on 

 the July cruise of Hoyo Maru No. 67 and 5 sets on 

 the 1979 cruise of Oshoro Maru, densities of net- 

 able seabird species (i.e., those caught during the 

 cruises) were correlated to the number caught per 

 tan (110-130 mm mesh only) at each set. Density 

 indices at netting areas were averages of the 

 counts immediately before arrival at and another 

 after departure from the netting site. Only five of 

 Oshoro Marus sets were used because other 

 studies took her far away from netting areas at 

 very low speed before transits at cruising speeds 

 began. We equilibrated our estimates of bird den- 

 sity based on simultaneous but independent 

 counts made aboard Oshoro Maru, 25-27 July 

 1979. 



In the following discussions oceanographic re- 

 gions are approximate because they are based on 

 geographic locality as shown in Favorite et al. 

 (1976), and not on sophisticated oceanographic 

 measurements taken at the time of bird observa- 

 tions. Hourly sea-surface temperatures, however, 

 did aid somewhat in confirming the boundaries 

 between certain domains. 



Results and Discussion 



Species Observed 



We recorded 31 species of seabirds during our 

 at-sea observations but only 15 of these were 

 caught in the nets (Table 1). For certain species, 

 their pelagic distribution was reflected in the gill 

 net catch by statistically significant correlations 

 (P<0.05) between density estimates and number 

 of birds caught. For example, in the Bering Sea 

 and in Western Subarctic waters south of the 

 Aleutians, the short-tailed shearwater greatly 

 outnumbered the sooty shearwater, both in den- 

 sity and in the gill nets, but farther south to just 

 north of the Subarctic Front their numerical im- 

 portance in the catch and in the censuses was 

 reversed. Several other species also showed differ- 

 ences in their distributions, for example, ancient 

 murrelet and crested auklet. 



A bird's foraging methods influenced its suscep- 

 tibility to being caught (Table 2). Of the 13 species 

 we observed that feed beneath the sea surface by 

 diving or pursuit plunging (see Ainley 1977 for 



802 



