Project 0809/1

Project Title: Increasing the accuracy of dugong population estimates from aerial surveys by quantifying dugong diving behaviour

Chief Investigators: Prof Helene Marsh, Prof Rhondda Jones and Dr Ivan Lawler

Since the mid 1980s, aerial surveys have been used to estimate the distribution and relative abundance of dugongs over the vast and remote coastal waters of tropical Australia. The probability of detecting the dugongs that are present in a survey area at the time of an aerial survey is composed of three components: 1) the probability that an area will be surveyed (sampling fraction); 2) the probability of an animal being near enough to the surface to be seen by an observer (availability bias), and 3) the probability of detecting an animal given that it is available (perception bias). Both availability and perception biases are heterogeneous both between and within surveys and methods have been developed to estimate the different components of dugong detection probability. Based on the analysis of Chilvers et al. (2004), the availability correction factors developed by Pollock et al. (2006) were based on dugong diving records across the diel cycle, despite aerial surveys being conducted only between 8am and 4pm. 

We analysed data from Timed Depth Recorders (TDRs) from dugongs tracked in Hervey Bay, Queensland in the context of fine scale telemetry, habitat mapping, and remotely sensed data to increase the capacity of aerial surveys to estimate dugong abundance by improving the correction factor for dugongs that are unavailable because of water turbidity.  We analysed dive recordings from Mk 7 and 9 TDRs, Australia equipped with satellite PTT/GPS tags. The location fixes from the GPS tags were used to determine the water depths experienced by the tracked dugongs in order to examine the effect of water depth on the dugongs’ use of the water column. We used logistic regression to predict the proportion of time a dugong would be available to observers in aircraft by estimating the effects of experienced water depth, time of day (0-8am/8-4pm/4-12am) and tidal direction (Incoming/Outgoing) using 1,403,248 TDR records for which all variables were available (~ 2 months of data for each of six animals). 

The dugongs were divided into two groups for the analysis depending on the water depth they occupied (0-6m; >7m). The logistic regression models indicated that all factors affected the availability of dugongs from survey aircraft. Once the effect of tide factor was removed, the effect of time-of-day was weak, although it was still significant for all cases. Water depth had the strongest effect; dugongs are significantly less available when in water 4-11m deep than in shallower or deeper water. Thus some of the differences between successive aerial surveys of the same area are likely to have resulted from dugongs being distributed differently across the depth gradient within a survey area between surveys. In shallow waters, availability is also influenced by tidal direction; dugong are understandably closer to the surface on outgoing tides. Prior to this analysis fluctuations in dugong abundance in a time series of surveys of the same area was attributed to dugongs migrating in and out of a survey area.

The results indicate that the accuracy of the aerial survey estimates of dugong abundance could be improved by stratifying the aerial survey blocks by water depth and in shallow water correcting for incoming versus outgoing tidal conditions. The difference is likely to be considerable as the area of potential dugong habitat between 4-11m deep is vast e.g .~15,500 km² for the Great Barrier Reef region and is likely to be particularly important for Torres Strait. The reanalysis of the aerial survey data based on this research may lead to a significant reassessment of dugong population estimates from aerial surveys.