Linking wildlife tracking data with environmental features to improve understanding of dugong diving ecology and population size estimates

Hagihara, Rie (2015) Linking wildlife tracking data with environmental features to improve understanding of dugong diving ecology and population size estimates. PhD thesis, James Cook University.

PDF (Thesis)
Download (2MB) | Preview
View at Publisher Website:


The dugong (Dugong dugon) is a coastal marine mammal of conservation concern with a sub-tropical and tropical range extending from East Africa to the Solomon Islands and Vanuatu. Australia is the dugong's stronghold and the site of most modern research. Compared to the research on the dugong's horizontal space use and movement patterns, little is known of the dugong's diving behaviour. Application of behavioural information to large-scale monitoring studies of population abundance has been minimal. My research combined data collected from a variety of technologies and platforms (satellite/GPS wildlife tracking, remote and benthic sensing, aerial survey) to study dugong diving behaviour and improve aerial survey estimates of dugong abundance.

The objective of this thesis was to maximise the usage of wildlife tracking data to improve fine-scale knowledge of the dugong diving ecology and to apply this information to improve the methodology to estimate dugong abundance. I achieved these objectives by: 1) developing an empirical procedure to maximise the correct identification of dives recorded by time-depth recorders (TDRs); 2) advancing understanding of fine-scale dugong diving behaviour by linking dive records with fine-scale spatial movement data and habitat descriptions; and 3) improving aerial survey estimates of dugong population size by accounting for their heterogeneous diving and surfacing behaviours. Details of my results follow.

Aim 1: Develop an empirical procedure to identify dives in shallow-diving aquatic wildlife such as the dugong

Dives from coastal aquatic animals can be difficult to interpret because the shallow nature of their dives relative to the resolution of TDRs often precludes the reliable identification of the different phases of a dive (e.g., descent, bottom, and ascent). I developed an empirical procedure to determine the thresholds for: 1) the zero-offset correction (ZOC) for surface calibration; and 2) the maximum dive depth (dive threshold (DT)). This empirical approach increased the reliability of dive identification and was essential to subsequent interpretations of dugong diving behaviour (Aim 2).

Aim 2: Advance insights into fine-scale dugong diving ecology

I used statistical models based on dive parameters identified using the methodology developed for Aim 1. Dugongs are benthic feeders that primarily consume seagrass. Thus dugongs reaching in the vicinity of the seafloor where seagrass is present are more likely to be feeding on seagrass than the seafloor where no seagrass is present. However, behavioural inference from dives per se was not attempted because several behaviours can occur over seagrass meadows.

Using 8 dive metrics (descent rate, bottom time, vertical displacement, maximum depth, ascent time, ascent rate, asymmetry and ascent rate divided by descent rate), I performed a series of logistic regression models to predict dives that achieved the two criteria: a) mid-water dives that did not provide the dugong with access to the seafloor and dives that did; and b) dives that enabled the dugong to access the seafloor in areas with seagrass and without seagrass. These criteria were determined from a bathymetric model, tidal records, and a seagrass model from shallow banks of Moreton Bay, The logistic regression models showed that compared to dives that had a high likelihood of accessing the seafloor (seafloor dives), dives that had a high likelihood of not reaching the seafloor (mid-water dives) were characterised by shorter bottom times, a larger degree of vertical displacement (presumably the result of active tail movements) during the bottom phase, and slower ascent rates. The profiles of these mid-water dives included U-, V- and other shapes (Fig.1).

The dugongs that had a high likelihood of accessing the seafloor in locations supporting seagrass transited quickly between the surface and the seafloor and maximised the time spent on the substratum, presumably maximising nutrient return. The profiles of such dives were mostly classified as square-shaped and less frequently U-shaped. Dugongs undertaking seafloor dives in locations without seagrass also spent a long time on the bottom but were sluggish in all phases of the dive, including the transits between the surface and the bottom. These dives generally had U-shaped profiles (with some square profiles). The dive shapes in the three groups overlapped supporting my assumption that inferences about dive function on a broad classification of dive shapes given the data I examined is not possible.

Aim 3: Estimate dugong population size that is more robust by accounting for their heterogeneous diving and surfacing behaviours

The current aerial survey methodology used to estimate dugong population sizes at extensive spatial scales accounts for availability bias (animals that are present in survey transects but not visible) due to water turbidity and sea state but assumes constant dugongs' diving and surfacing patterns. To improve availability bias estimates (availability detection probabilities), particularly to account for heterogeneous availability bias, I first estimated availability detection probability by combining data from dugongs fitted with TDRs, GPS satellite tracking units, and fine-scale bathymetric models (Chapter 5). I found availability detection probabilities varied with water depth. All dugongs in clear shallow water (e.g., <1 m) are presumed to be available for detection and the availability bias in these shallow waters was not estimated experimentally. The probability of a dugong being available was next highest in water up to 5 m deep (0.60 to 0.87), followed by water ≥25 m deep (0.58 to 0.85), and lowest in water 5 to 25 m deep (0.34 to 0.69). These depth-specific availability corrections should be more accurate and increasing the likelihood of detecting actual change in a population size.

Using correction factors that incorporated the dugong depth-specific availability detection probabilities, I improved estimates of dugong population abundance over three survey regions (Chapter 6). In Moreton Bay, the abundance estimates based on depth-independent (constant) and depth-specific availability detection probabilities were similar because a high proportion of dugongs were sighted in clear shallow water where all animals were potentially available for detection. In Hervey Bay, the abundance estimate based on the depth-specific availability detection probabilities was lower than the estimate using the constant availability detection probabilities, because more than 50% of dugongs were sighted in clear deep water where the estimated depth-specific availability detection probabilities were higher than the depth-independent estimates. In Torres Strait, the difference in the estimated abundance between the two methodologies was large (>3500 dugongs; 28%). Many dugongs were sighted in waters 5-25 m deep in this region and the depth-specific availability estimates were smaller than the estimates independent of water depth, leading to the larger abundance estimate.


The results of my research have not only significantly improved understanding of the diving behaviour of dugongs and led to improved estimates of dugong abundance in heterogeneous environments but have also demonstrated methodological advances that should have wider application to shallow-diving aquatic wildlife whose studies are often hampered by coarse resolution of TDRs and affinity of the animals to shallow waters.

Item ID: 41267
Item Type: Thesis (PhD)
Keywords: abundance estimation; aerial survey; availability bias; behavior; behaviour; benthic; coastal animal; dive analysis software; dive analysis; dive threshold; diving; dugong ecology; dugong population; dugong; generalized linear mixed model (GLMM); geographical distribution; Hervey Bay; marine mammal; Moreton Bay; population estimates; population; Shoalwater Bay; Sirenia; time-depth recorder (TDR); zero-offset correction (ZOC)
Related URLs:
Additional Information:

Publications arising from this thesis are available from the Related URLs field. The publications are:

Chapter 3: Hagihara, Rie, Jones, Rhondda, Sheppard, James, Hodgson, Amanda, and Marsh, Helene (2011) Minimizing errors in the analysis of dive recordings from shallow-diving animals. Journal of Experimental Marine Biology and Ecology, 399 (2). pp. 173-181.

Chapter 5: Hagihara, Rie, Jones, Rhondda E., Grech, Alana, Lanyon, Janet M., Sheppard, James K., and Marsh, Helene (2014) Improving population estimates by quantifying diving and surfacing patterns: a dugong example. Marine Mammal Science, 30 (1). pp. 348-366.

Date Deposited: 01 Dec 2015 06:22
FoR Codes: 05 ENVIRONMENTAL SCIENCES > 0502 Environmental Science and Management > 050202 Conservation and Biodiversity @ 50%
06 BIOLOGICAL SCIENCES > 0608 Zoology > 060801 Animal Behaviour @ 50%
SEO Codes: 97 EXPANDING KNOWLEDGE > 970106 Expanding Knowledge in the Biological Sciences @ 50%
97 EXPANDING KNOWLEDGE > 970105 Expanding Knowledge in the Environmental Sciences @ 50%
Downloads: Total: 379
Last 12 Months: 10
More Statistics

Actions (Repository Staff Only)

Item Control Page Item Control Page