Supervised accelerometry analysis can identify prey capture by penguins at sea

Abstract

Determining where, when and how much animals eat is fundamental to understanding their ecology. We developed a technique to identify a prey capture signature for little penguins from accelerometry, in order to quantify food intake remotely. We categorised behaviour of captive penguins from HD video and matched this to time-series data from back-mounted accelerometers. We then trained a support vector machine (SVM) to classify the penguins' behaviour at 0.3 s intervals as either 'prey handling' or 'swimming'. We applied this model to accelerometer data collected from foraging wild penguins to identify prey capture events. We compared prey capture and non-prey capture dives to test the model predictions against foraging theory. The SVM had an accuracy of 84.95±0.26% (mean ± s.e.) and a false positive rate of 9.82±0.24% when tested on unseen captive data. For wild data, we defined three independent, consecutive prey handling observations as representing true prey capture, with a false positive rate of 0.09%. Dives with prey captures had longer duration and bottom times, were deeper, had faster ascent rates, and had more 'wiggles' and 'dashes' (proxies for prey encounter used in other studies). The mean (±s.e.) number of prey captures per foraging trip was 446.6±66.28. By recording the behaviour of captive animals on HD video and using a supervised machine learning approach, we show that accelerometry signatures can classify the behaviour of wild animals at unprecedentedly fine scales.

Convergent Foraging Tactics of Marine Predators with Different Feeding Strategies across Heterogeneous Ocean Environments

“…(Sims et al, 1997;Sims and Quayle, 1998). Moreover, a positive correlation between prey intake and duration of the bottom phase has also been found for harbor seals (Phoca vitulina; Baechler et al, 2002) and for penguins (Carroll et al, 2014). Although other functions, such as resting and traveling, have been proposed for U-shaped dives in marine turtles and seals (Hochscheid et al, 1999;Baechler et al, 2002;Seminoff et al, 2006), such dive functions seem to be less likely to apply to large predatory fish.…”

“…Linking movement patterns to habitat use remains, however, a challenging task. Detailed records of prey abundance and distribution and accurate indices of feeding are difficult to obtain for the majority of species and although visual assessment of prey capture is possible for some species (Seminoff et al, 2006;Elliott et al, 2008), in most cases, indirect parameters have been used as a proxy (e.g., gastric or visceral temperature changes, mouth/beak opening or head/jaw movement, accelerometer signatures; Sepulveda et al, 2004;Gleiss et al, 2011aGleiss et al, , 2013Nakamura et al, 2011Nakamura et al, , 2015Carroll et al, 2014;Nakamura and Sato, 2014). For efficient foraging by predators, patterns of habitat use are assumed to reflect the distribution, density and quality of prey resources (Stephens and Krebs, 1986;Austin et al, 2006;Carroll et al, 2017).…”

“…The time an optimally foraging animal spends in a specific area is assumed to be related to its prey richness (Stephens and Krebs, 1986;Mori et al, 2005), hence U dives are thought to correspond to prey patch exploitation because this type is characterized by a bottom phase of prolonged duration at a relatively constant depth, likely reflecting foraging on aggregated prey patches (Austin et al, 2006). Indeed, some studies in pinnipeds and penguins have shown a positive correlation between food intake and high duration of the bottom phase in U-shaped dives (Carroll et al, 2014;Viviant et al, 2014). Whale sharks have also been found to dive at steeper angles during U-shaped dives, thus maximizing time foraging in a horizontally restricted area (Gleiss et al, 2011b).…”

Using accelerometers to remotely and automatically characterize behavior in small animals

“…Many simultaneous recordings of paired behavioral observations and accelerometer readings must be collected to determine the correct mapping. Machine-learning techniques have successfully been used to complete this task (Carroll et al, 2014;Bidder et al, 2014;Escalante et al, 2013;Gao et al, 2013;Grünewälder et al, 2012;Martiskainen et al, 2009;McClune et al, 2014;Nathan et al, 2012;Sakamoto et al, 2009). Previously, machine-learning algorithms have not modeled sequential correlations between behaviors and have not allowed for flexible lengths of behavioral segments, two constraints that may limit system accuracy.…”

Measuring abnormal rotational movements in free-swimming fish with accelerometers: implications for quantifying tag- and parasite-load

“…Such data can be used to indirectly quantify variation in behaviour, energetics and physiology, and to infer how animals interact with each other and their environment (Cooke et al, 2004) for habitat modelling and conservation management (Bograd et al, 2010;Whitney et al, 2010). For example, micro-storage accelerometer tags allow for remote measurements of fine-scale movements and behaviour among free-swimming fish in time and space in controlled mesocosm environments (Gleiss et al, 2010;Broell et al, 2013;Noda et al, 2014;Wright et al, 2014;Broell and Taggart, 2015), as well as in the wild (Kawabe et al, 2003a,b;Tsuda et al, 2006;Whitney et al, 2010;Carroll et al, 2014).…”