Comparability in the detection of harbour porpoise
The C-POD and F-POD were co-deployed for a continuous period of 444 days
between April 2021 and July 2022. Harbour porpoise detections were
recorded on 94% of days on the C-POD (419 days), and 98% on the F-POD
(433 days).
Across all deployments, total C-POD detections were lower than those
recorded using the F-POD (Figure 1). The margin of difference was highly
dependent on the detection metric and the train quality classification
category examined, with smallest differences noted when combining all
(visually validated) quality categories. While there was a significant
correlation between the detections on the co-deployed PODs at all
temporal scales, PODs were in greatest agreement at the broadest scale
metric of detection positive days (filter HiMod r= 0.86,
p<0.05; HiModLo r=0.84, p<0.05, Appendix Table S2),
and least comparable at the scale of number of clicks per hour (Figure
1, Appendix Table S2). Although variable across time, the increased
capacity for click detection by the F-POD was evident, with the F-POD
detecting 10 times or more the number of clicks detected by the C-POD
(Figure 1, Appendix Table S1). Considering the “Hi” filter alone
showed limited comparability between the C-POD and the F-POD, with
significant but weak correlation (Appendix Table S2). Moreover, using
only the high-quality classification filter meant that on average 75%
of the F-POD DPH were not detected by the C-POD within the same quality
grouping. Using the HiMod or HiModLo groupings increased the
comparability of PODs, but the F-POD still consistently recorded more
harbour porpoise detections overall. There was a small proportion of
detection positive hours recorded on the C-POD that were not matched by
the F-POD. However, these visually validated C-POD detections often
matched unclassified NBHF clicks (i.e. not defined to be click trains by
any of the quality classes) on the F-POD. Furthermore, most unmatched
C-POD detections were weaker trains of Low-quality, occurring during
periods of increased ambient noise, and not classified by the more
conservative F-POD algorithms.
Over total 15-month period the PODs were deployed, a detection ratio of
1.38 was calculated using detection positive hours. Seasonal variability
in this detection ratio occurred with PODs least comparable in the
spring-summer when detection rates where lowest (ratios: Spring: 1.52,
Summer: 1.48, Autumn: 1.07, Winter: 1.37, Figure 2). Despite the
detection differences, both PODS similarly identified temporal patterns
of occurrence at hourly scales (DPM and DPH) using the HiMod or HiModLo
train quality groupings. There was a decrease in detections from April
to July and consistently more detections throughout the winter months
(Figure 2).
Harbour porpoise foraging behaviour was detected by both of the
co-deployed PODs throughout the deployment period. Buzz positive hours
(BPH), i.e. hours where foraging buzzes were identified, made up a low
proportion of the total recording hours, particularly for the C-POD
which detected less than a third of the BPH detected by the F-POD.
Foraging buzzes were found to account for approximately 8% of the total
clicks recorded by the C-POD, compared to 26% for the F-POD. The number
of buzz positive hours per day was found to vary seasonally, reflecting
the temporal patterns shown in overall harbour porpoise detections
(Figure 2), with low counts of BPH from May till July, and peaks shown
throughout winter as well as for August (Figure 2).