Behavioural studies of diving birds have reported that the ratio of di
ve duration to the duration of the subsequent period on the surface di
splays a characteristic relation to dive duration. For short dives, th
e dive to surface ratio increases with dive duration, whereafter the r
elation peaks, and for longer dives decreases with increasing dive dur
ation. Such a relationship is not a general prediction of existing mar
ginal value models which have been used to predict optimal diving beha
viour. This may be because the smooth curve used to describe the oxyge
n gain rate of individuals after surfacing is not a good reflection of
the respiratory physiology of birds. Here we argue that on physiologi
cal grounds, the oxygen gain curve for avian divers will not be smooth
, but will have two distinct regions (representing oxygen recovery in
the respiratory tract, and in haemoglobin and myoglobin, respectively)
. Modifying two of the classical diving models by incorporating such a
kinked curve causes them to predict the humped relationship between d
ive to surface ratio and dive duration under many circumstances. We al
so present data on the duration of dives and surface periods from thre
e species of diving seabirds: the shag, Phalacrocorax aristotelis, the
black guillemot, Cepphus grylle and the common guillemot, Uria aalge.
All three species showed a humped relationship for dive to surface ra
tio as a function of dive duration. In line with the predictions of ou
r model, when oxygen stores on surfacing were greatly depleted, the di
ve to surface ratio peaked at short dive durations. (C) 1998 The Assoc
iation for the Study of Animal Behaviour.