The differences in adult-to-nest ratios led rne to consider the sources of 

 variation. Much of the variation between States ir.ay be due to observer bias 

 differences or to changes in habitat resulting in visibility differences. To 

 test observer error, aerial estimates of gull colonies were compared using 

 estimates made by two experienced wildlife biologists (Observers 1 and 2) 

 during a 1-day census in southern Maine on 31 May (Figure 11). Estimates of 

 Observer 2 were then compared with those of another, Observer 3, for all 

 Massachusetts colonies obtained on 1 and 6 June (Figure 12). Since Observer 2 

 censused both States, the results reveal both inter-observer differences and 

 State differences. The difference between estimates by Observers 2 and 3 are 

 slight compared to those between Observers 1 and 2. The two regressions for 

 Observer 2 are only slightly different (b = 0.35 vs. 0.30). 



These results might be interpreted as follows: (1) there is little 

 difference in adult/nest ratios between gull colonies in Maine and those in 

 Massachusetts, (2) there can be a major difference in estimates by different 

 observers. Without verification of the "true" number present (e.g. using 

 photographs), it is impossible to say which observer is more accurate. How- 

 ever, a minimum of one bird per nest should be present because of incubation 

 or brooding duties. The slopes and y-intercepts can be better understood if 

 the raw data are examined (Tables 24 & 25). At the large colonies. Observers 

 2 and 3 usually underestimiate the number of birds by a large factor. Observer 

 1, however, probably over estimates the larger colonies. 



Major habitat differences permitted a comparison of the adult-to-nest 

 count relationship among a number of gull colonies in New England. Eleven 

 colonies in Rhode Island, Connecticut, and Massachusetts, with light to mod- 

 erate cover of beach grass or on bare rock, were contrasted with six 

 Massachusetts colonies in dense herbaceous or shrub vegetation. Regression 

 analysis indicated that about 25% more gulls per nest are estimated when the 

 colony occupies a lightly-vegetated island (slope, b = 0.67) rather than a 

 densely-vegetated one (slope, b = 0.53). On many islands in Boston Harbor and 

 on the north shore of Massachusetts, a large percentage of gulls nest under 

 dense vegetation and are easily missed during an aerial inventory. This 

 contributes significantly to bird-to-nest ratios (Figures 11 & 12) being much, 

 lower than the 1 1.0 value expected if at least one adult is always in 

 attendance at a nest. 



Species Correction Factors (Aerial-Ground) 



Comparing adults estimated from the air with nests counted at certain 

 colonies provided the basis for many of the correction factors used in deter- 

 mining the number of breeding pairs for colonies at which nests were not 

 counted (Table 26). Laughing gull nest counts were only obtained at three 

 colonies so regression analysis was not used to obtain a correction factor. 

 The factor obtained for common terns was applied to other similar terns. 

 Aerial photographs of colonies of black skimmers, royal, and sandwich terns 

 showed that incubating and non-incubating birds (mates) could be distinguish- 

 ed. Portnoy (1977) found that both members of a black skimmer pair often 

 attend the nest during incubation so a bird-to-nest ratio of 0.5 was used. 

 Most often, only one adult attends the nest in royal-sandwich tern colonies so 

 a 1.0 factor was applied. 



44 



