Factors affecting perpendicular sighting distances on shipboard line-transect surveys for cetaceans

Abstract

Factors that affect cetacean perpendicular sighting distances are investigated using a Generalised Additive Modelling (GAM) framework to analyse 8,203 sightings of 34 cetacean species seen on 200,000km of shipboard line-transect surveys in the eastern Pacific in 1986-96. Perpendicular sighting distance is modelled as a non-linear function of the following predictor variables: species; an a priori species grouping; the logarithm of group size; Beaufort sea state; presence of rain or fog; sighting cue; sun glare; geographic stratum; observer; ship; year; cruise; and, in 1991-96, visibility and swell height. Based on Akaike Information Criteria (AIC), the best model for 1986-96 included all variables except rain/fog code. For the 1991-96 data, swell height anomaly was also important and replaced ship and year in the best-fit model. For independent subsets of the data, GAM coefficients were highly correlated, indicating that many of the same factors were acting in different areas and at different times. Species and sighting methods (25x binoculars vs unaided eye) had the largest effects on perpendicular sighting distances. The a priori species groups captured much, but not all of the among-species differences. Two other species-related factors (group size and sighting cue) were also important in all models. Factors related to search conditions (Beaufort sea state and swell height anomaly) and to the searchers themselves (individual observer) were also important. We anticipate that this information on the relative magnitudes of factors affecting perpendicular sighting distance can be used to improve both design and analysis of line-transect data.