Comparison of closing and passing mode from a line-transect survey of delphinids in the eastern Tropical Pacific Ocean

Résumé

Line-transect ship surveys are the primary method used to estimate abundance of pelagic cetaceans. However, survey methods are often modified from traditional methods because observers must approach cetacean groups to identify species and estimate group size. Returning to the trackline after approaching a school dramatically reduces the amount of effective survey time, so ships often resume survey effort at the sighting but parallel to the original trackline (closing mode). Survey effort is no longer independent of group locations, and it is unclear how such methodological modifications affect overall abundance estimates. This research presents the results of a study designed to determine the effects of closing mode methods on abundance estimation for cetacean species in the eastern tropical Pacific. Species identification and group size estimation in closing mode are compared with results using survey techniques where the ship does not approach or slow down to investigate a sighting (passing mode). Both empirical data and simulations were used to compare group encounter rates in the two modes and to better understand the mechanisms that might lead to an encounter rate bias in closing mode. As seen in similar studies, observers are able to identify to the species level less frequently in passing mode (81% vs 57% of sightings), and point estimates of delphinid group size were 58% lower in passing mode than closing mode at distances between 1.0 and 5.5km from the trackline. In addition, uncertainty in group size both within and between observers was higher in passing mode. Closing mode delphinid group encounter rates were generally 20–25% lower than passing mode delphinid group encounter rates. Simulations showed the empirically lower encounter rates in closing mode are due to a loss in detection probability caused by the stop-start nature of the survey method. The closing mode encounter rate bias is greater when groups are in fewer and/or tighter clusters and when overall group density is higher. Methodological adjustments and analytical solutions to improve group size estimation and species identification in passing mode and reduce closing mode encounter rate bias are analytically complex and would also result in the loss of important additional life history data. Nevertheless, such avenues should be explored further.