Methodology
To identify moult in puffins, we have developed a new method combining
multiple data streams from geolocator loggers to more accurately
identify year-round behaviours of seabirds. While this seems to work
reasonably well for dual-equipped individuals, it also highlights some
shortcomings of using standard single-deployment geolocators to identify
fine-scale behaviours. Halpin et al. (2021) show how species’ behaviour
can unpredictably influence location estimates using light-level
geolocators. Leg-tucking in alcids presents a similar problem for the
interpretation of behaviours from saltwater immersion loggers on the
same devices (Fayet et al. 2017, Linnebjerg et al. 2014). We provide a
method to partially correct for this behavioural classification issue
using concurrent light and immersion data. The limitations of this
partial correction are reflected in the low success rate of moult
inference in single-logger birds. Because puffins and some other alcids
spend very little time in flight even when not undergoing moult (see
results, Dunn et al. 2020), flightless moult is impossible to identify
without relatively accurate behavioural data, and the few individuals
for which moult periods were detected using a single logger are likely
biased towards times of the year when leg tucking behaviour is less
prevalent. Despite these constraints, our methods provide insights into
the behaviour and life-history traits of a threatened species, and
progresses our knowledge surrounding the timing and location of a highly
vulnerable period in their annual cycle.
Over the last 10 – 15 years, hundreds of alcids have been tagged with a
single geolocator throughout their biogeographic range (Fayet et al.
2017, Reiertsen et al. 2021), but our method does not have the power to
identify moult in a sufficient proportion of individuals to robustly
investigate population-wide patterns. More involved methods, for
instance using machine learning to identify flightless stopovers
(Guilford et al. 2009), usually require large amounts of pre-assigned
training data to confidently infer behaviour, but may even then be
liable to misclassification due to individual- or colony-level
differences in behaviour (Bennison et al. 2018). Finer resolution data,
such as from accelerometers, would allow us to identify flight with much
more confidence (e.g. Patterson et al. 2019). GPS loggers would record
far more accurate locations, potentially allowing us to identify imposed
residency due to flightless moult. To date, none of these alternative
devices are small or efficient enough for year-round deployment on
puffins. Geolocators that record temperature can also be used to help
correct for leg-tucking (Elliot & Gaston 2014, Dunn et al. 2020),
though like the light-based corrections used in this study,
temperature-based corrections do not fully capture all instances of
leg-tucking. A ventrally mounted immersion switch would provide a truer
representation of flight/non-flight behaviour, but despite being light
enough for long-term deployment, techniques to mount devices long term
on the body, as opposed to on a leg-ring, that do not affect the bird’s
performance, have not been developed (Lameris et al. 2018). For now,
dual-equipped geolocators are probably the most viable method to
investigate the flightless moult of puffins and other alcids. As
technology improves and devices become smaller, the combined weight of
two loggers will have less impact on an animal. Stable isotope analysis
of moult feathers may also be used to coarsely gauge the location of the
most recent primary moult (e.g. St. John Glew et al. 2018) and to
validate geolocator based findings. To complement this, a relatively
accurate geolocator-informed moult timing and location tells us where
and when flight feathers were formed, allowing us to analyse the trophic
position of food consumed during feather formation using stable isotope
analysis (St. John Glew et al. 2019). It may also provide information on
the prevalence of toxic chemicals in marine food webs at the time and
location of moult by looking at chemical composition of feathers (Fort
et al. 2016).