Preliminary ecological results
Using the SealNet facial recognition software package and a small,
initial dataset sampled across two years (2019 and 2020) in Casco Bay,
we identified four individuals in the datasets from both years,
indicating a small degree of local site fidelity across years during the
months of June and July. All four seals were found on the haul-out site,
Brandt Ledges, in 2019. In 2020, the four seals were photographed again
within 1-3 nautical miles of Brandt Ledges: one was photographed on
Branning Ledges, two were photographed at the Mitchell Field site, and
one was photographed on the Whaleboat Island site. This result supports
previous results suggesting site fidelity among harbor seals off the
coast of NE Scotland . It is also interesting to note that two of the
individuals found in the dataset from both years, Clove and Petal, were
found together initially on Brandt Ledges on one day in 2019, and then
found together again in 2020 at the Mitchell Field site. These results
suggest that SealNet software may be useful in future long-term studies
of social relationships in harbor seals. Previous studies have examined
competitive relationships among harbor seals , however further research
is needed to examine other questions related to social behavior,
including social fidelity, persistence of family groups, and other
social dynamics.
Our preliminary ecological results suggest some site-fidelity of harbor
seals in Middle Bay as well as site-fidelity to neighboring haul-out
sites within the bay. A more extensive ecological study is underway to
determine the degree of site fidelity and spatial connectivity of
haul-out sites in this region. In addition, our results provide an
estimated population size of approximately four thousand seals utilizing
Middle Bay, although more extensive photographic data will help refine
this population estimate. Current estimates of harbor seal abundance are
outdated, suggesting a population of 38,014 individuals in the whole of
Maine in 2001 ( followed by an aerial survey done in 2012 determining a
population of 75,834 individuals (). Accurate local and regional
population estimates are imperative to understanding dynamics of seal
abundance in relationship to anthropomorphic and climate changes to
coastal marine environments as well as the impact of an increasing great
white shark population.
The use of facial recognition software to identify individuals in wild
populations is a relatively new area of research being primarily
utilized in studies of land mammals such as lemurs and brown bears
(Crouse et al., 2017). Our research extends the use of such methods to
marine mammal species. Facial biometrics are not the only measure that
can be used for automated identification of seals. For example, a
recent, groundbreaking study utilized pelage markings found on the seals
coat to identify grey seal individuals near Wales (Langley et al.,
2020). Given that coat patterns may change across seasons during molting
or over time in harbor seals, facial biometrics may offer and additional
and/or more reliable method of identification. Thus, developing facial
recognition techniques for harbor seals allows for a rapid, non-invasive
means for detailed study of an economically and ecologically important
species. Importantly, researchers can customize the software and the GUI
to suit their own needs at each step of data collection—training the
face detector for additional species, modifying the alignment procedure,
or preprocessing images for face recognition.