Townsend Cromwell are applicable to these 

 species groups: 



Most birds of these groups will be observed 

 at these distances. Some farther from the ship 

 than the stated distance will be seen, but others 

 that are nearer will probably be missed. 



Overestimates of maximum distances of ob- 

 servation result in calculated population es- 

 timates that are conservative. 



Table 6 gives population estimates of each 

 species for the study area during the cruise on 

 which each species was most abundant. The 

 figure for birds per linear mile- was weighted 

 by the effective area of survey per linear m.ile 

 of observation of each species group, and the 

 resulting value for birds per square mile was 

 multiplied by the approximate area covered by 

 the study to yield a population estimate for each 

 species. 



Analysis of density by subareas . --Depending 

 upon its islands of origin, breeding schedule, 

 and pelagic habitat "preference," each species 

 should show distinctive patterns of relative 

 density in various parts of the study area in 

 different seasons. To test the data for such 

 patterns so that they can be demonstrated num- 

 erically as well as visually, the study area was 

 divided into eight subareas of about equal size. 

 A visual appraisal of distribution patterns in- 

 dicated that the land masses of the Hawaiian 



157°30'W 155° ISftQ- 



?6°30 N 



® 



® 



® 



(D 



157"30 



MrtO' 







® 



® 



® 



152°30 



-22° 



i4no 



Figure 28. — Location of eight 5° x 4° subdivi- 

 sions of the study area. Each subdivision 

 contains ca. 72,000 square miles (24.6 million 

 ha.) . 



group should be included in one subarea, that 

 an area of low density east of the Hawaiian 

 group should be in another, and that areas of 

 relatively high density to the south should be in 

 still others. Accordingly, quadrangles of 5° 

 latitude by 4° longitude (ca. 72,000 sq. miles or 

 24.6 million ha,) were selected as being prob- 

 ably the most meaningful in terms of homo- 

 geneity of distribution within each subarea (fig. 

 28). Densities in terms of birds per linear 

 mile of observations for the more numerous 

 regularly occurring species were calculated 

 for the four seasons, rather than for each 

 month, to minimize inequalities in sampling and 

 to simplify the presentation of the data. Spring 

 was taken to mean March-May; summer, June- 

 August; fall, September-November; and winter, 

 December -February. 



70 



