294 Monitoring Threatened Species and Ecological Communities
probability of an animal being recorded by an observer, given that the animal is
available (perception bias). Perception bias is estimated in mark–recapture
framework following Pollock et al. (2006).
The focus here is on availability bias. Two methodologies for estimating
dugong abundance are compared, illustrating the importance of accounting for
variable detection probability. The Pollock method (Pollock et al. 2006), which
has been used in most dugong stock assessments since 2000, accounts for the
effects of environmental conditions but assumes that the time dugongs spend at
or near the surface is homogeneous across water depths. The more recent
Hagihara method (Hagihara et al. 2014, 2016) accounts for the effects of
environmental conditions, but also accounts for changes in the diving patterns of
the target species with water depth.
In most circumstances, the detectability of individuals declines with increasing
distance from an observer, and this decline is accounted for in the estimation
algorithms inherent in line transect methodologies. However, Pollock et al. (2006)
found no such decline in the detection of dugongs across a 200 m wide strip. Thus
strip transects, which are logistically simpler than line transects, have been used in
aerial surveys for dugongs because they typically surface very cryptically and for
only a few seconds.
To minimise environmental sources of measurement ‘noise’, limits were set for
weather and sea surface conditions within which surveys could take place (e.g. no
surveys during rain or rough seas). In Torres Strait, most dugong sightings are of
one or two animals. The few groups of ≥10 dugongs were counted in circling mode
to obtain accurate counts.
Availability detection probability was estimated using information on: (1) the
depth of the detection zone in which dugongs are visible from above-water
observers; and (2) the proportion of time animals spend in that zone. The
availability detection probabilities estimated empirically by Pollock et al. (2006)
were used for the estimates of abundance and density calculated using the Pollock
method. The depth of the detection zone varies with environmental conditions
such as sea state, turbidity and glitter at the surface, categorised by a composite
environmental conditions index (Table 22.1) in the Hagihara method. Hagihara et
al. (2016) then estimated the proportion of time animals spend in each detection
Table 22.1. Environmental conditions index, a composite index of sea state, turbidity and glitter on the
surface.
The description of water quality has been adopted from Pollock et al. (2006).
Environmental Conditions Index Water quality Depth range Visibility of the seafloor
1 Clear Shallow Clearly visible
2 Variable Variable Partially visible
3 Clear Deep Not visible
4 Turbid Variable Not visible