Statistical models which unify line transect and mark recapture models for
closed heterogeneous populations are presented. The models extend conventio
nal line transect models by including a mark recapture component which allo
ws relaxation of the conventional assumption that detection on the tracking
is certain. They generalize existing mark recapture models by including a
statistical model for the distribution of explanatory variables which affec
t detectability or catchability, and they allow this distribution to be est
imated simultaneously with detection probabilities. In a line transect cont
ext, independent observers represent the capture occasions; The issue of un
modelled heterogeneity is addressed and composite survey methods are propos
ed which allow observable heterogeneity to be modelled while at the same ti
me minimising unobserved heterogeneity by design. "Asymmetric" designs, in
which the areas searched by the independent observers is separated and one
observer sets up trials from which the other platform's detection function
is estimated, are proposed. Such designs have substantial advantages over d
esigns which treat the two platforms symmetrically. Recaptures/resightings
are assumed to occur without error. In practice this is seldom a realistic
assumption and the methods presented here could be improved by incorporatin
g uncertainty associated with recapture/resighting identification. Neverthe
less, the methods provide a flexible framework for estimating animal abunda
nce combining line transect and mark recapture methods.