Melinda G Conners1, Théo Michelot2, Eleanor I Heywood3, Rachael A Orben4, Richard A Phillips5, Alexei L Vyssotski6, Scott A Shaffer7, Lesley H Thorne3. 1. School of Marine and Atmospheric Sciences, Stony Brook University, Stony Brook, NY, 11794, USA. melinda.conners@stonybrook.edu. 2. Centre for Research into Ecological and Environmental Modelling, University of St Andrews, St Andrews, KY169LZ, UK. 3. School of Marine and Atmospheric Sciences, Stony Brook University, Stony Brook, NY, 11794, USA. 4. Department of Fisheries and Wildlife, Oregon State University, Hatfield Marine Science Center, 2030 SE Marine Science Dr., Newport, OR, 97365, USA. 5. British Antarctic Survey, Natural Environment Research Council, High Cross, Madingley Road, Cambridge, CB3 0ET, UK. 6. Institute of Neuroinformatics, University of Zurich and Swiss Federal Institute of Technology (ETH), 8057, Zurich, Switzerland. 7. Department of Biological Sciences, San Jose State University, San Jose, CA, 95192-0100, USA.
Abstract
BACKGROUND: Inertial measurement units (IMUs) with high-resolution sensors such as accelerometers are now used extensively to study fine-scale behavior in a wide range of marine and terrestrial animals. Robust and practical methods are required for the computationally-demanding analysis of the resulting large datasets, particularly for automating classification routines that construct behavioral time series and time-activity budgets. Magnetometers are used increasingly to study behavior, but it is not clear how these sensors contribute to the accuracy of behavioral classification methods. Development of effective classification methodology is key to understanding energetic and life-history implications of foraging and other behaviors. METHODS: We deployed accelerometers and magnetometers on four species of free-ranging albatrosses and evaluated the ability of unsupervised hidden Markov models (HMMs) to identify three major modalities in their behavior: 'flapping flight', 'soaring flight', and 'on-water'. The relative contribution of each sensor to classification accuracy was measured by comparing HMM-inferred states with expert classifications identified from stereotypic patterns observed in sensor data. RESULTS: HMMs provided a flexible and easily interpretable means of classifying behavior from sensor data. Model accuracy was high overall (92%), but varied across behavioral states (87.6, 93.1 and 91.7% for 'flapping flight', 'soaring flight' and 'on-water', respectively). Models built on accelerometer data alone were as accurate as those that also included magnetometer data; however, the latter were useful for investigating slow and periodic behaviors such as dynamic soaring at a fine scale. CONCLUSIONS: The use of IMUs in behavioral studies produces large data sets, necessitating the development of computationally-efficient methods to automate behavioral classification in order to synthesize and interpret underlying patterns. HMMs provide an accessible and robust framework for analyzing complex IMU datasets and comparing behavioral variation among taxa across habitats, time and space.
BACKGROUND: Inertial measurement units (IMUs) with high-resolution sensors such as accelerometers are now used extensively to study fine-scale behavior in a wide range of marine and terrestrial animals. Robust and practical methods are required for the computationally-demanding analysis of the resulting large datasets, particularly for automating classification routines that construct behavioral time series and time-activity budgets. Magnetometers are used increasingly to study behavior, but it is not clear how these sensors contribute to the accuracy of behavioral classification methods. Development of effective classification methodology is key to understanding energetic and life-history implications of foraging and other behaviors. METHODS: We deployed accelerometers and magnetometers on four species of free-ranging albatrosses and evaluated the ability of unsupervised hidden Markov models (HMMs) to identify three major modalities in their behavior: 'flapping flight', 'soaring flight', and 'on-water'. The relative contribution of each sensor to classification accuracy was measured by comparing HMM-inferred states with expert classifications identified from stereotypic patterns observed in sensor data. RESULTS: HMMs provided a flexible and easily interpretable means of classifying behavior from sensor data. Model accuracy was high overall (92%), but varied across behavioral states (87.6, 93.1 and 91.7% for 'flapping flight', 'soaring flight' and 'on-water', respectively). Models built on accelerometer data alone were as accurate as those that also included magnetometer data; however, the latter were useful for investigating slow and periodic behaviors such as dynamic soaring at a fine scale. CONCLUSIONS: The use of IMUs in behavioral studies produces large data sets, necessitating the development of computationally-efficient methods to automate behavioral classification in order to synthesize and interpret underlying patterns. HMMs provide an accessible and robust framework for analyzing complex IMU datasets and comparing behavioral variation among taxa across habitats, time and space.
Authors: Christopher C Wilmers; Barry Nickel; Caleb M Bryce; Justine A Smith; Rachel E Wheat; Veronica Yovovich Journal: Ecology Date: 2015-07 Impact factor: 5.499
Authors: Rory P Wilson; Luca Börger; Mark D Holton; D Michael Scantlebury; Agustina Gómez-Laich; Flavio Quintana; Frank Rosell; Patricia M Graf; Hannah Williams; Richard Gunner; Lloyd Hopkins; Nikki Marks; Nathan R Geraldi; Carlos M Duarte; Rebecca Scott; Michael S Strano; Hermina Robotka; Christophe Eizaguirre; Andreas Fahlman; Emily L C Shepard Journal: J Anim Ecol Date: 2019-06-27 Impact factor: 5.091
Authors: Graeme C Hays; Helen Bailey; Steven J Bograd; W Don Bowen; Claudio Campagna; Ruth H Carmichael; Paolo Casale; Andre Chiaradia; Daniel P Costa; Eduardo Cuevas; P J Nico de Bruyn; Maria P Dias; Carlos M Duarte; Daniel C Dunn; Peter H Dutton; Nicole Esteban; Ari Friedlaender; Kimberly T Goetz; Brendan J Godley; Patrick N Halpin; Mark Hamann; Neil Hammerschlag; Robert Harcourt; Autumn-Lynn Harrison; Elliott L Hazen; Michelle R Heupel; Erich Hoyt; Nicolas E Humphries; Connie Y Kot; James S E Lea; Helene Marsh; Sara M Maxwell; Clive R McMahon; Giuseppe Notarbartolo di Sciara; Daniel M Palacios; Richard A Phillips; David Righton; Gail Schofield; Jeffrey A Seminoff; Colin A Simpfendorfer; David W Sims; Akinori Takahashi; Michael J Tetley; Michele Thums; Philip N Trathan; Stella Villegas-Amtmann; Randall S Wells; Scott D Whiting; Natalie E Wildermann; Ana M M Sequeira Journal: Trends Ecol Evol Date: 2019-03-14 Impact factor: 17.712
Authors: Rory P Wilson; Craig R White; Flavio Quintana; Lewis G Halsey; Nikolai Liebsch; Graham R Martin; Patrick J Butler Journal: J Anim Ecol Date: 2006-09 Impact factor: 5.091
Authors: Ran Nathan; Orr Spiegel; Scott Fortmann-Roe; Roi Harel; Martin Wikelski; Wayne M Getz Journal: J Exp Biol Date: 2012-03-15 Impact factor: 3.312
Authors: K Yoda; Y Naito; K Sato; A Takahashi; J Nishikawa; Y Ropert-Coudert; M Kurita; Y Le Maho Journal: J Exp Biol Date: 2001-02 Impact factor: 3.312
Authors: Matthew D Taylor; Luke McPhan; Dylan E van der Meulen; Charles A Gray; Nicholas L Payne Journal: PLoS One Date: 2013-11-18 Impact factor: 3.240