Effect of unmodelled heterogeneity upon abundance estimation using line and point transects

It is widely recognised that unmodelled heterogeneity in detection probability can cause significant negative bias in mark recapture abundance estimates. By contrast, distance sampling methods are supposed to be much less affected, so long as detection at zero distance is certain, due to an estimator property called pooling roubstness. We explore mathematically the property of pooling robustness, and assess by simulation the level of heterogeneity needed for the property to break down. We found that the bias in line transect estimates of abundance were small.