Oral Presentation Australian Society of Fish Biology and Oceania Chondrichthyan Society Conference 2016

State-space modelling methods for telemetry data: an application to surface-association behaviours in tropical tuna (#10)

Joe Scutt Phillips 1 , Graham M Pilling 2 , Bruno Leroy 2 , Karen Evans 3 , Thomas Usu 4 , Kurt M Schaefer 5 , Chi H Lam 6 , Simon Nicol 7
  1. Climate Change Research Centre, University of New South Wales, Sydney, NSW, Australia
  2. Oceanic Fisheries Programme, Secretariat of the Pacific Community, Nouméa, New Caledonia
  3. Ocean and Atmosphere Flagship, Commonwealth Scientific and Industrial Research Organisation, Hobart, Tasmania, Australia
  4. National Fisheries Authority, Port Moresby, Papua New Guinea
  5. Inter-American Tropical Tuna Commission, La Jolla, CA, USA
  6. Large Pelagics Research Center, University of Massachusetts, Amherst, MA, USA
  7. Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra, NSW, Australia

Objectively classifying the behavioural time-series obtained from animal telemetry experiments requires factors such as temporal autocorrelation, varying time-scales, high-resolution data and underlying motivation to be considered. Hidden Markov models (HMMs), a form of state-space model, provide a useful classification tool for such data. HMMs assume that the observed patterns in movement data are generated by an unseen Markov process, which switches between an observation model of mixed distributions, each with different parameters.

Here, we describe the design and numerical parameter estimation of these models for time-series of telemetry data, and present an example application examining the vertical behaviour of two species of tropical tuna. Tuna are known to reside near the surface around drifting and anchored fish aggregating devices (FADs) over periods of time that range from days to months. This residence makes them more vulnerable to fishers by increasing their horizontal encounter rate with vessels, and by causing a switch in vertical movement in favour of extended association with shallow waters. Estimating HMMs on 80 time-series from electronic tags implanted in bigeye (Thunnus obesus) and yellowfin (T. albacares) tuna at liberty in the Western and Central Pacific Ocean, we defined and objectively identified periods of surface-association that are typical in residence around floating object.

The proportion of time exhibiting surface-association was generally lower than has been previously observed, although intra- and inter-individual variability was high. Surface-association was clear and prolonged for many fish following release, but prior to recapture, surface-association events were non-existent or shorter than two days for 85% of bigeye and 74% of yellowfin. These results suggest the vertical component of floating-object association may be less important than short-term horizontal concentration when increasing probability of capture. Controlling the number of sets made on FADs may be a more effective management measure than the current approach of periodic closures.