Comparing cetacean abundance estimates derived from spatial models and design-based line transect methods

Spatial modelling is increasingly being used as an alternative to conventional design-based line transect sampling to estimate cetacean abundance. This new method combines line transect sampling with spatial analysis to predict animal abundance based on the relationship of animals observed to environmental factors. It presents several advantages including: (1) the ability to use data collected from 'platforms of opportunity', (2) the ability to estimate abundance for any defined subarea within the study area, and (3) the possibility for increased precision if covariates explain sufficient variability in the data. One study has been conducted to compare spatial modelling with conventional line transect methods, but the use of covariates in the detection function and the inclusion of school size have not previously been investigated. In the present study, the density of striped dolphins Stenella coeruleoalba was estimated in western Mediterranean waters using spatial distance sampling models applying generalised additive models (GAMS). This estimate was compared with density values previously estimated in the same area using conventional line transect methods. The densities estimated were very similar: 0.494 animals km(-2) (coefficient of variation, CV = 0.16) using spatial models and 0.489 animals km(2) (CV = 0.19) using conventional line transect methods. Densities were also similar when they were calculated in stratified areas defined during the original line transect study. The precision of the estimates from spatial modelling was higher than that of the estimates obtained from conventional line transect analysis, particularly in the subareas. The results confirm that spatial modelling is a good approach for estimating cetacean abundance.