Multivariate analysis of biologging data reveals the environmental determinants of diving behaviour in a marine reptile

Abstract

Diving behaviour of 'surfacers' such as sea snakes, cetaceans and turtles is complex and multi-dimensional, thus may be better captured by multi-sensor biologging data. However, analysing these large multi-faceted datasets remains challenging, though a high priority. We used high-resolution multi-sensor biologging data to provide the first detailed description of the environmental influences on flatback turtle (Natator depressus) diving behaviour, during its foraging life-history stage. We developed an analytical method to investigate seasonal, diel and tidal effects on diving behaviour for 24 adult flatback turtles tagged with biologgers. We extracted 16 dive variables associated with three-dimensional and kinematic characteristics for 4128 dives. K-means and hierarchical cluster analyses failed to identify distinct dive types. Instead, principal component analysis objectively condensed the dive variables, removing collinearity and highlighting the main features of diving behaviour. Generalized additive mixed models of the main principal components identified significant seasonal, diel and tidal effects on flatback turtle diving behaviour. Flatback turtles altered their diving behaviour in response to extreme tidal and water temperature ranges, displaying thermoregulation and predator avoidance strategies while likely optimizing foraging in this challenging environment. This study demonstrates an alternative statistical technique for objectively interpreting diving behaviour from multivariate collinear data derived from biologgers.

Keywords: biologging; dive type; diving behaviour; multi-sensor data; sea turtle.

Figure 1.

Surface movement paths from filtered GPS locations (CRS: WGS 84) for flatback turtles (n = 24) at Yawuru Nagulagun Roebuck Bay, Western Australia, visualized using ArcMap (v. 10.6). For each individual, GPS locations were filtered based on number of satellites (greater than 6.0) and altitude (0–150.0 m) to reduce potentially erroneous locations. Track colours indicate sampling season: Austral summer, warm; winter, cool.

Figure 2.

Representative timeseries of flatback turtle diving during (a) winter (Turtle ID 9) and (b) summer (Turtle ID 31). Inset panel shows concurrent depth, overall dynamic body acceleration (ODBA; proxy for locomotory activity), pitch (body angle) and heading (travel path direction with respect to North) calculated from the multi-sensor data, from which dive variables were derived (table 1 for full explanation of variables).

Figure 3.

Cluster analysis output for (a) k-means and (b) hierarchical clustering. Elbow and silhouette plots (k = 1 : 10) and cluster plots (k = 2 : 7) were assessed for each method, to determine optimal number of clusters for n_dives = 4128 from standardized dive variables (refer table 1), derived from multi-sensor data collected from adult flatback turtles (n = 24).

Figure 4.

Representative dives for multivariate dive features for flatback turtles. Reconstructed three-dimensional dives representative of high (left column) and low (right column) value scores for each of the retained principal components, PC1–PC5. Black arrow marks dive path start and travel direction, path colour scale for each PC as per interpreted eigenvector loading values, where PC1 = depth, PC2 = activity and body angle (descent), PC3 = activity and body angle (ascent), PC4 = activity and duration (bottom) and PC5 = tortuosity (ascent) (see electronic supplementary material, figure S7). Cartesian coordinates (distance eastwards and northwards from a starting point of 0) were estimated from dead reckoned accelerometer and magnetometer data (20–50 Hz) using the dead reckoning wizard within Framework4 [83]; Vectorial Dynamic Body Acceleration (VeDBA) was selected as a proxy for locomotory speed and path anchored to begin at the turtles' capture location. Dive paths were visualized in three-dimensional space by plotting the combined Cartesian coordinates with the corresponding depth data. Note, all representative dives were selected from Turtle ID 9 and axes scales vary between dives to aid visualization.

Figure 5.

Generalized additive mixed models of the diving behaviour of flatback turtles (n = 24) at Yawuru Nagulagun Roebuck Bay, Western Australia. Solid line represents the most parsimonious model with individual specified as a random effect (±s.e. represented by shaded bands and dashed lines; summer, darker shading; winter, lighter shading; table 4). There were clear seasonal, diel and tidal patterns in diving behaviour. Note, to aid visualization the Y-axes scales differ between models and for parametric effects. Model colour corresponds to interpretation of PCs by their eigenvector loadings (see electronic supplementary material, figure S7 and figure 4).