FISHERY BULLETIN: VOL. 86, NO. 3 



surveys the declination angle to marine mammals 

 was measured when the animals were perpendicular 

 to the aircraft using hand-held inclinometers. 

 Declination angles were not measured during the 

 1984 survey. 



Distance to the coast was monitored using 

 declination angles (19° for 0.61 km and 6° for 1.85 

 km). On the 1984 survey, this distance was 0.61 

 km for the entire survey. During the 1985 survey, 

 we surveyed at both 0.61 and 1.85 km from the 

 coast. In 1984, the coast was taken to be the outer 

 limit of the surf zone. In 1985, the coast was taken 

 to be the outer limit of the surf zone or, if kelp 

 beds were present, the outer margin of those 

 beds. 



Shore Observation Methods 



Observations of harbor porpoise diving behavior 

 were made from rocky headlands in northern Ore- 

 gon (Tillamook Head, Neahkahnie Mountain, Cape 

 Meares, and Cape Lookout) immediately before the 

 second aerial survey (7-11 September 1985). Ob- 

 servers were equipped with 7 x 50 binoculars with 

 compasses and ocular reticles and a single 20 x 120 

 binocular. Ventilation data were collected whenever 

 possible and included the number of animals at the 

 surface and the length of time spent at the surface. 

 Observations were recited aloud by the observer and 

 were written down by a second person or were 

 recorded onto magnetic tape. The ventilation cycle 

 typically consisted of a period with several surfacing 

 rolls and breaths (which we call a surfacing series) 

 followed by a much longer period of submergence 

 (which we call a dive). This dive cycle corresponds 

 to ventilation pattern B as described by Watson and 

 Gaskin (1983) for harbor porpoise in the Bay of 

 Fundy and the pattern described by Taylor and 

 Dawson (1984) for porpoise in Glacier Bay. 



Helicopter Observation Methods 



Behavioral observations were also made by three 

 observers in a 4-passenger, jet-turbine helicopter. 

 Upon locating a group of harbor porpoise, a fluores- 

 cein dye marker was dropped and the helicopter 

 hovered or circled slowly above the group at an 

 altitude of approximately 300 m. The number of 

 animals, the time they were visible at the surface, 

 and the dive times were recorded, along with infor- 

 mation on cloud cover, sea state, and water turbid- 

 ity. Each behavioral session was given a subjective 

 rating based on how well the observers could follow 

 the group and obtain accurate dive times. Only ses- 



sions with good or excellent ratings were included 

 in analyses. 



Probability of 

 Missing Submerged Animals 



Given that a porpoise would be within the visual 

 range of an observer, the probability that it will be 

 at the surface during the passage of the aircraft is 

 related to the average time it spends at the surface, 

 s, the average time spent below the surface, d, and 

 the window of time during which it is within the 

 visual range of an observer, t. This probability was 

 calculated as 



Pr (being visible) 



s -I- t 

 s + d 



(2) 



The probability of missing a submerged animal is 

 equal to the complement of this value. 



Density Estimation 



Density of harbor porpoise was estimated as the 

 number of animals seen divided by the area searched 

 (Equation (1)). This raw density estimate was ad- 

 justed by dividing by the probability that an animal 

 would be visible from the air at any given instant 

 (Equation (2)). The area searched was estimated as 

 the swath widths times the lengths of the transects. 

 Transect lengths were calculated as the sum of the 

 great circle distances between successive position 

 fixes. Densities were calculated for each of the eight 

 statistical regions used by Barlow (1988) (Fig. 2). 



The statistical difference in harbor porpoise 

 density between different sighting conditions or 

 different areas was tested using the raw density 

 estimates. Density estimates for short transects 

 were frequently zero, thus violating the parametric 

 assumptions of normally distributed, homoscedastic 

 error. Nonparametric tests were therefore chosen 

 for density comparisons. In discussing statistical 

 tests, a transect segment refers to the length of 

 transect line between two successive position fixes 

 and are typically <20 km. The measured variables 

 relating to sighting conditions are constant within 

 a segment, and because each sighting is accom- 

 panied by a new position fix, a segment will contain 

 at most, one sighting. 



Whenever applicable, the Wilcoxon paired-sample 

 test (Wilcoxon 1945) was used to test one factor 

 while controlling for as many other factors as possi- 

 ble. Ten paired measures of density were created 



436 



