|
Résumé :
|
The key difficulty in assessing animal numbers from the air is that not all animals are seen by the observers. Methods for estimating detection probabilities, or accounting for imperfect detection, are reviewed including double surveys, use of sightability models, mark-resight, and mark-recapture. The assumptions needed for each method are considered as well as issues concerning survey design. For closed-population mark-recapture modelling particular attention is given to multiple observer studies. An emphasis is that an assumption of complete independence in double-observer studies is rarely justifiable and that independent observers will generally only satisfy an assumption of conditional independence and not complete independence.
|