SAMPLING METHODS "WITHIN COLONY" 



When only a few colonies in an area require censusing, the researcher 

 needs to know which of several methods to employ for a given species (or 

 group) under certain habitat conditions. Usually the expense of aircraft will 

 demand some type of ground method (boat or on foot). 



Of all ground methods, a total nest count using a marking method to 

 prevent duplication is the most accurate. However, the accuracy of such 

 "total" nest counts may be misleading (Drury 1973) unless a second count is 

 made and a correction factor is applied, such as the Lincoln (mark-recapture) 

 Index (Overton 1971). The accuracy of a one-search "total" count was compared 

 to the "true-total" derived from a two-search count using the Lincoln Index 

 (Table 19). Eight sample plots ranging from 0.06 to 1.10 ha were selected in 

 various densities of vegetation in seven gull colonies. In each plot, all 

 nests found on the first search were marked. A second search v/as then made, 

 and a separate count of marked and unmarked nests was made. A qualitative 

 assessment of nest dispersion (uniform, patchy) was also recorded. 



The error range of 4-22% is large considering the relatively small sample 

 plots. Vegetation density rather than plot size seems to be the more impor- 

 tant factor in determining error rates. The "1 ight"-vegetated plots had 4, 7, 

 and 9% errors, while those in "moderate-dense" plots ranged from 7-22% (mean 

 15%). In summary, the accuracy of alleged "total" nest counts may be more 

 apparent than real unless a second count is made, admittedly a \/ery time-con- 

 suming process. However, the benefit one gains in achieving greater accuracy 

 may be offset by causing greater disturbance, especially in large, dense 

 colonies where several hours may be required to mark e^^ery nest. 



An alternative to a total nest count is a sample count, using the sample 

 to derive a total population estimate. This approach is especially valuable 

 where the habitat density or colony size renders a total count impractical. 

 In general, two types of sampling schem.es may be employed in the field: plot 

 (area) and plotless (distance) methods (Pielou 1969). Plot methods include 

 quadrat, strip, or belt transect sampling, where (ideally) random "plots" are 

 sampled and used to extrapolate the "true" population. Plotless methods com- 

 monly used are nearest-neighbor distance (Clark and Evans 1954, Lloyd 1967), 

 point-centered quarter, and line-intercept (Smith 1966). With these methods, 

 distances are measured and converted to aerial measures. A disadvantage of 

 all except the belt transect method is that the total colony areas to be 

 censused must be calculated. Surveying or m.easuring from aerial photographs 

 can be expensive and time-consuming. Both plot (belt transect) and plotless 

 (point-centered quarter) methods were tested in gull and heron-egret colonies 

 to test the efficacy of these sampling methods under different regimes of 

 habitat and nest dispersion pattern (Appendix A). Gull and heron colonies 

 were chosen because their size and inaccessibility (especially in the case of 

 herons) often preclude total count methods. In the largest colonies, a large 

 sample plot was used. 



Both sampling methods yielded highly variable population estimates, 

 depending on vegetation and nest dispersion patterns (Table 20). The point- 

 centered quarter method was reliable only when nests were uniformly distri- 

 buted and vegetation v/as light (except at Eagle Island). 



35 



