Adopting a model-based approach for density estimation represents a departure from traditional line transect methodology, for which unbiased estimation relies on a survey design with randomized trackline locations. Although it is desirable to have a set of tracklines with sufficient spatial spread to provide representative coverage of the survey area, spatial modelling does not require that the tracklines are designed according to a formal survey sampling scheme. If the spatial coverage of the nonrandom data is adequate, spatial models can yield estimates of abundance from sightings data collected from 'platforms of opportunity' like ferries or survey vessels which allow collection of large quantities of nonrandom data at a cost substantially lower than that of a properly designed survey. Possibility of bias in abundance estimation due to model misspecification and need for sophisticated analysis methods are the chief disadvantages.
展开▼