An approach to minimise bias in estimation of the lengthfrequency distribution of bowhead whales (Balaena mysticetus) from aerial photogrammetric data

Résumé

Past attempts to estimate the length structure of the Bering-Chukchi-Beaufort (B-C-B) stock of bowhead whales have yielded quite different results from one study to the next because of size segregation on the summering areas or because of size segregation during the spring migration combined with inconsistent sampling during the migration period. A new approach is presented to document the length-frequency distribution of the B-C-B stock using length measurements from 3,107 whale images collected during the spring migrations of 1985, 1986, 1989-1992 and 1994. This method provides estimates of the proportion of calves (length <6m), subadults (length 6-<13m) and adults (length !13m). The data from all years are combined by weekly period and a bootstrap sampling procedure is used to construct the lengthfrequency by week. The distributions for each week are then combined to obtain the overall distribution, with each week’s contribution being in proportion to the fraction of the migration estimated from ice-based census studies to pass during that week. Corrections for differential detectability of mother/calf pairs and for calves born after they pass Point Barrow, Alaska, are allowed for in the analysis. This new approach eliminates some of the biases that affected past attempts to estimate the length-frequency distribution for the B-C-B population of bowhead whales. It is robust to inclusion or exclusion of data for any given year and the time interval chosen to define repeat images. The new approach estimates a slightly higher proportion of subadults and lower proportion of adults in the population than most previous studies. The proportion of calves is also lower, but that is suspected to result from our inability to accurately estimate the proportion of the migration late in the season when many of the mother-calf pairs pass Point Barrow. These late season migrants have not been accounted for during past photogrammetry studies or the ice-based census. Although the results do not differ substantially from those of most previous studies, sensitivity analyses indicate that several biases existed in the previous methods, but largely cancelled each other out.