FISHERY BULLETIN: VOL. 81, NO. 2 



sus results was not addressed in Reilly et al. (1980). 

 We report here a quantitative appraisal of this effect, 

 and account for it in our abundance estimation. 



METHODS 



Field Methods: Shore Census 



The exact seasonal duration of the annual census 

 changed only slightly from year to year, but it usually 

 began on or before 10 December and ended on or af- 

 ter 6 February (59 d). The watch was conducted be- 

 tween 0700 and 1 700 h, 7 d a week, by two observers 

 who alternated 5-h shifts. 



The observers watched to the north for southward 

 swimming whales to come into view. At first sighting 

 of a whale or group of whales the time was recorded 

 and an initial estimate was made of the number of 

 whales in the group. The whales were kept under ob- 

 servation until they were directly offshore from the 

 station, usually about 0.5 h later. At that time a final 

 estimate of the number present was recorded, along 

 with the time and an estimate of the distance of the 

 animals offshore. Distance estimates were classified 

 in seven intervals: 0-0.25; 0.25-0.50; 0.50-0.75; 0.75- 

 1.0;1.0-1.5; 1.5-2.0; 2.0+ mi. Beaufort Sea state, 

 wind direction, and notes on visibility conditions 

 were recorded continuously throughout the day. 

 Binoculars (7 X 50) were used regularly. Beginning in 

 1978-79, visibility conditions were assigned one of 

 six ordinal categories (Table 1) for each pod ob- 

 served. For data prior to 1978-79, visibility conditions 

 were classified to within these categories during the 

 analysis, based upon information recorded sys- 

 tematically during the censuses. 



TABLE 1. — Gray whale census — Granite Canyon visibility codes. 



Code 



Condition 



Description 



04 



05 

 06 



Fair 



Poor 



None 



Clear day. or high clouds. No glare. Horizon visi- 

 ble. Effective sighting distance = 3+ mi 



Clear or some cloud cover. Some glare, surface 

 ripple. Effective sighting distance = 2-3 mi. 



Some fog, haze, low clouds. Some interference 

 from chop, surf, or glare. Effective sighting dis- 

 tance = 1-2 mi. 



Fog, full overcast, light rain, haze with glare. Fre- 

 quent whitecaps. Effective sighting distance = 

 0.5-1 mi. 



Moderate rain or fog, large surf, bad glare, etc. 

 Effective sighting distance = 0.25-0.5 mi. 



Combination of conditions make it very difficult 

 or impossible to see even the closest (within 0.5 

 mi.) whales. Heavy rain, dense fog, near dark- 

 ness, etc. 



Analysis Methods: Shore Census 



Occasionally during the censusing, only one of the 

 standard two sighting times per group (when first 



seen and when directly offshore) was recorded. Fre- 

 quently when an observer came on duty at 0700 h 

 there were whales directly offshore and no "north 

 time" was recorded. In addition, at the end of the day 

 at 1 700 h, whales which had not yet passed directly in 

 front of the station were often sighted to the north, 

 and no "south time" was recorded. To correct 

 for missing time records, a mean difference between 

 the two times was calculated for each observer in- 

 dividually. Missing time records were then generated 

 from this average, and the single time record avail- 

 able. The time when the animals were directly 

 offshore was then used to categorize data for time of 

 day analyses. Only sightings with this time falling 

 between 0700 and 1700 h were used for abun- 

 dance estimation. 



The results of the 1978-79 and 1979-80 half-day 

 observation periods were investigated by analysis of 

 variance (ANOVA) for differences between observ- 

 ers and between morning vs. afternoon periods on 

 rate of recording animals, as was previously done 

 (Reilly et al. 1980) for the 1967-68 through 1977-78 

 data. We also examined the two most recent censuses 

 for possible changes in hourly rates of recorded 

 counts, as done previously for the 1967-68 through 

 1977-78 data. Again, we looked for significant de- 

 pressions in the counts both at the ends of the 5-h ob- 

 server periods (as an indication of observer fatigue) 

 and at the beginning and end of the day (as an indica- 

 tion of daylight-mediated change in migration 

 rate) . 



For any migratory species which can be censused 

 feasibly from a fixed point, the distribution of daily 

 counts, transformed to proportions for each migra- 

 tion, can be viewed profitably as a time-density dis- 

 tribution and modeled by various probability density 

 functions (Mundy 1979). We previously assumed a 

 normal distribution (Reilly et al. 1980) for all years 

 pooled. Problems with this approach were that mean 

 days between years were not equal and that a slight 

 but consistent skewness occurred causing lack of fit. 

 Consequently we have replaced the normal distribu- 

 tion with the more flexible gamma distribution (Pear- 

 son's Type III; Bury 1975) and modeled each year 

 separately. The time-density model for each migra- 

 tion was then employed in three ways: 



1) To estimate the number of whales having passed 

 the station before the first and after the last day 

 of the census (the "tails"). 



2) As a standard for comparison with observed dai- 

 ly results, in a determination of if, and to what 

 degree, conditions associated with the six visi- 

 bility categories affect census results. 



270 



