A Remote Science Experience: Struggles and Successes


Using accelerometer data to estimate prey capture attempts of pinniped species around the world


HOW IT ALL BEGAN...

SARS-CoV-2 spread around the world, causing a global lockdown

A REMOTE PROFESSIONAL PRACTICE

Australian Sea Lion

Neophoca cinereal

Northern Fur Seal

Callorhinus ursinus

South American Fur Seal

Arctocephalus australis

Antarctic Fur Seal

Arctocephalus gazella


Due to the global lockdown, the possibility of traveling to Australia to work at Deakin University disappeared, and with it, the possibility of pinniped-tagging field work. Disappointed but not defeated, I decided to make the most out of this situation and dedicate my time into data analysis and learning how to interpret accelerometer data. What better time to further learn how to use the R programming language and other data analysis software than during quarantine?

Tri-axial accelerometer (Cefas Techonology Limited)

During the 8 weeks of the professional practice, it was my job to interpret data from tri-axial accelerometers for 26 individuals of the four pinniped species above, and estimate the number of prey capture attempts for each of their dives. It was also my responsibility to extract the dive profiles from Time-Depth Recorders (TDR) in R Studio using the DiveMove package.

All tagged sea lion/ fur seals for this study were adult females with suckling pups. This was done in order to guarantee their return to the beach where the initial tagging occurred for safe removal of their tags after a full foraging trip. Each trip could last between a few hours to a couple of days-long, species dependant.

Each of the individuals was tagged with A) a tri-axial accelerometer B) an animal-borne video camera (crittercam), C) a TDR and a GPS datalogger.

THE WORK PROCESS

Initially, I began with online tutorials, trial-and-error experimenting in the programs that I had to learn how to use for the data analysis, and the reading of many scientific publications on the topic. As I read about similar studies performed on pinniped in the past and about the methodology of sea lion/fur seal tagging, I began day dreaming of doing it myself.














DayDreamLR.mov

DATA ANALYSIS

Igor Pro Analysis

  • The dive data and the accelerometer data were previously aligned.

  • aY and aX axis represent the sway motion of the pinniped's head (x or y depend on accelerometer placement).

  1. First, I applied a high pass filter to the sway axis to isolate head movements from full body movements.

  2. Next, I selected an appropriate threshold to determine which sway peaks reflected a true prey capture attempt (PCAs).

  3. Finally, I combined PCAs with a calculated time threshold which separated head shakes from one prey capture and another (btw. 4-6 seconds).

  4. This allowed me to create a graphical representation and export a file with an accurate PCA count for each of the sea lion/fur seal individuals.


R Studio Analysis

  1. With DiveMove R package, I created a TDR plot and data file with the sea lion/fur seal dive profiles.

  2. I zero-offset corrected each file.

  3. Finally, I set a filter to only count dives below 2m depth and with a minimum dive time of 12 seconds as foraging dives and exported this data for further analysis.

All this data will contribute to a global study on foraging behaviour of pinniped species from around that globe which is being led my project supervisor, Dr John Arnould, at Deakin University.

FINDINGS AND LESSONS LEARNED

NFS-PreyHandling.mov

Prey handling varies between benthic and pelagic foragers.

  • Those that feed in the water column (pelagic foragers) tend to catch smaller prey and feed throughout the dive.

  • In contrast, those that feed on the sea floor (benthic foragers) search for larger prey and usually carry it to the surface for feeding.

  • Some pinniped species, such as the Northern Fur Seals, use both types of foraging and thus made the analysis more challenging.

Watch the video to check this out!

Note the head shakes with every catch the Northern Fur Seal accomplishes or attempts. These result in the peaks in the sway axis of the accelerometer data which allowed me to interpret her foraging behaviour underwater.

PP Results.docx

This document summarises the chosen sway and time thresholds for each of the species from the graphs created with Igor Pro data and most importantly it summarises the average prey capture attempts per dive for each of the species. Scroll to see results in detail.

RESULTS:

  • AUSL attempted around 1 PCAs per dive

  • NFS attempted around 3 PCAs per dive

  • SAFS attempted around 5 PCAs per dive

  • AFS attempted around 3 PCAs per dive


Trivia: From this data and the info above, can you guess which species are benthic and/or pelagic foragers?

For trivia answers click here: "Credits & Acknowledgments" page.

CONCLUSION

Accomplishing a scientific study from home proved to be more challenging than having done so in an office environment or in the field. The infinite R errors, the glitches with data downloads and slow internet, plus the easily swayed attention span were a few of the challenges I encountered during this professional practice.

Nonetheless, I enjoyed the experience and was thrilled to learn a skill I had been wanting to for a while: how to interpret accelerometer data.

These incredible technological advances are furthering scientists ability to understand how animals behave in places where we can't easily observe them, such as underwater. Being able to read waves from head movements and interpret them as certain aspects of behaviour is incredible.