Remotely sensing motion: the use of multiple biologging technologies to detect fine-scale, at-sea behaviour of breeding seabirds in a variable Southern Ocean environment

Schoombie, Stefan

Remotely sensing motion: the use of multiple biologging technologies to detect fine-scale, at-sea behaviour of breeding seabirds in a variable Southern Ocean environment

Doctoral Thesis

2021

Abstract

The at-sea behaviour of seabirds, such as albatrosses and petrels (order Procellariiformes), is difficult to study because they spend most of their time on the ocean and have extremely large ranges. In the early 2000s, behavioural studies of seabirds were dominated by diving patterns of diving birds or spatial studies from satellite telemetry. Recent advances in biologging technologies have opened up new avenues for studying the at-sea behaviour of farranging seabirds in their natural environment. Bio-logging devices are now small enough to be attached to flying seabirds where multiple sensors record data at infrasecond sampling rates. These data can be used to infer, inter alia, body posture, activity (e.g. flapping, takeoff, landing, etc.), magnetic heading and spatial distribution at a resolution that was not previously possible. Bio-logging devices are battery powered and a tradeoff exists between the length of deployments and sampling frequencies, however not a lot of study has been done on what the effect of coarse sampling rates are on data quality. Together with the masses of data that are generated by bio-logging devices, analytical tools have also become available to extract useful metrics from the data. This thesis utilized some of the latest bio-logging technology to study the at-sea behaviour of several procellariiforms, breeding on Marion, Gough and Nightingale Islands, from finescale data. After describing the loggers used and the methods of deployment in Chapter 2, I assess the effect that sampling rates have on metrics derived from GPS loggers in Chapter 3. This was done by sub-sampling GPS tracks recorded at 1-s sampling intervals, showing the effect that different sampling intervals have on metrics, including the total distance travelled and behavioural states derived from path length and turning angles. I show that for larger sampling intervals, the total distance travelled will be underestimated at varying degrees depending on flight sinuosity. By varying sampling rates when estimating behavioural states, I show that moderate (10–30 min) sampling intervals may produce better results. I explore the limitations of low-cost GPS loggers for fine-scale analyses and conclude that specialized loggers are most likely required when sampling at intervals < 1 s. In Chapter 4 I use specialized loggers in the form of tri-axial magnetometer, and video loggers and describe two novel methods to extract roll angles of albatrosses during dynamic soaring flight. Animal body angles are normally extracted by using tri-axial accelerometer data, but their dynamic soaring flight mode inhibits the use of these methods. I show how magnetometer data are independent of dynamic movement and can be used to estimate roll angles of flying seabirds. This method is validated from bird-borne video footage where the horizon is used as a proxy for the bird's roll angle and I describe a method to automatically extract such angles using computer vision techniques. These new methods are then applied to data collected from Wandering Albatrosses Diomedea exulans in Chapter 5, showing how the birds vary their roll angle in response to changing winds. Additionally, flapping flight was identified from patterns in the vertical axis (heave) of a tri-axial accelerometer and I show how Wandering Albatrosses may be flapping more than expected. By coupling flapping and roll angles I show that flapping, on occasion, occur at the upper turn of the dynamic soaring cycle, a period previous thought devoid of flaps. These results also suggest possible sexual differences, where males seem to flap more often than females and limit their take-offs to favourable wind conditions. Lastly, in Chapter 6 I use the same methods as in the previous two chapters to compare the fine-scale flight of six Procellariiformes species breeding on Marion, Gough and Nightingale Islands. I show how these species have varied flight patterns where they respond differently to wind patterns, most likely driving their distribution and eventual foraging areas. As expected, smaller species seem to be more manoeuvrable allowing them to rapidly roll at extreme angles in strong winds while tolerating light winds by increasing the amount of time spent flapping. Breeding location also played a role as birds from the Tristan da Cunha archipelago flapped more often and flew in lighter winds than Marion Island birds. In summary, Chapter 7 discusses how, using a multisensor approach, bio-logging technology can be effectively used to study the fine-scale behaviour of flying seabirds. Each of the loggers have their own limitations and it is important to take these into account when analyzing the data. I describe two new methods for extracting roll angles from dynamic soaring seabirds and show how individuals from several species vary roll angle and flapping flight in response to changing winds. This thesis highlights the varying behavioural strategies that flying seabirds use in the Southern Ocean, showing that individual species and populations may respond differently to changing environmental conditions.

Keywords

Biological Sciences

Reference:

Collections

PhD / Doctoral

Full item page