Discussion:
To the best of our knowledge, this study is the first to evaluate the performance of co-deployed C-POD and F-POD devices in a field setting for monitoring harbour porpoise. Previous studies have evaluated the performance of the C-POD with other types of full bandwidth recorders (incl. Soundtrap, DMON), with the C-POD typically performing well with an overall high degree of accuracy (Roberts and Read, 2015; Sarnocinska et al., 2016; Jacobson et al., 2017). Our results suggest that at appropriate temporal scales, C-PODs provide comparable results to the newer F-PODs and with certain caveats, F-PODs would be suitable replacements for C-PODs in existing monitoring programmes as C-PODs reach the end of their serviceable lifetime.
Data from click detectors such as C-PODs, its predecessor T-PODs, and now the F-PODs have been used to monitor small cetaceans, as well as their responses to anthropogenic activities in numerous settings including pile driving, seismic surveys, and fisheries deterrent devices (e.g. Philpott et al., 2007; Thompson et al. , 2013; Omeyeret al. , 2020; Todd et al., 2020). Our results show that the F-POD consistently detects more echolocation clicks and foraging buzzes than the C-POD across the temporal scales of minutes, hours and days, as well as all train quality classification groupings. Lower detection rates by the C-POD is to be expected due to advances in F-POD electronics and software to capture more information on individual echolocation clicks and enhance train detection (Chelonia Ltd, 2022). This poses a potential issue for researchers engaged in long-term monitoring, with questions about how comparable different POD types may be, potentially affecting time-series as C-PODs are eventually replaced by F-PODs. This study shows that both C-PODs and F-PODs detected similar patterns of occurrence and echolocation activity. As indicated by Garrod et al. (2018), detection metrics at a minimum of an hourly scale are representative of relative occurrence, enabling temporal trends to be determined. Detections at the broader scale of detection positive days were found to match best between both PODs with little discrepancy between them. Therefore, for direct comparison between C-POD and F-POD data, detection positive days, and using combined classifications of Hi Mod and Lo, is the only detection metric recommended. However, such a metric would be insufficient for identifying fine-scale temporal patterns of occurrence of behaviour in response to factors such as diurnal changes in prey availability or tidal state, both known to influence harbour porpoise occurrence and feeding behaviour (Schaffeld et al., 2016; Zein et al., 2019)
Our results highlighted that F-PODs appear to be much more capable at identifying harbour porpoise click trains with confidence than the C-POD (i.e. classified as high quality by the KERNO classifier). However, when considering the combined train quality classification groupings (i.e. HiMod and HiModLo) the two PODs are substantially more comparable. Researchers considering using a time series consisting of data from C-PODs and F-PODs for analysis of temporal trends should consider using the combined classifications, provided extensive visual validation is followed, particularly for low quality trains to eliminate possible false positive detections. The enhanced train detection specified by the manufacturer has also been demonstrated in our results with F-POD continually detecting more harbour porpoise detections than the C-POD by a factor of 1.38 across the deployment period. Comparability between the PODs was however found to be variable between seasons, with the highest detection ratio in spring and summer making detection rates on the PODs less comparable. Detection ratios such as explored here could be investigated further within monitoring programmes looking to transition to the use of F-PODs. Understanding how the devices compare in various deployment sites can help for the interpretation of long-term data beyond the lifespan of the C-POD and avoid misinterpretation of the data (for example interpreting a false increase in occurrence due to differing device sensitivities).
Investigating spatial and temporal patterns in species occurrence is often the crux of ecological monitoring (e.g., Jones et al., 2014; Williamson et al., 2017; Zein et al., 2019). Generalised additive models were used to investigate whether detections from both types of POD have the capacity to identify the same temporal drivers of porpoise occurrence and foraging activity. The occurrence models for both PODs highlighted the same temporal patterns, and same environmental predictors with very similar effect sizes, suggesting that analyses using data from both C-PODs and F-PODs will not be affected greatly by POD type. However, it would be prudent to include POD type as a fixed factor in any such analysis. Conversely, the models investigating feeding buzzes did not provide analogous results. No temporal patterns were found using the C-POD data, possibly due to the much lower detection rates of feeding buzzes using C-PODs, with the F-POD detecting three times as many foraging buzzes in comparison, and 10 times as many echolocation clicks overall. The higher detection rates of feeding buzzes by the F-POD enabled detection of temporal patterns including an increase in foraging buzzes from autumn to winter, and an increase in foraging activity during the day. The specific nature of the relationships and their ecological context is outside of the scope of the current study, but the contrasting ability of the PODs to detect feeding buzzes is particularly relevant in the context of integrating F-PODS into long-term datasets. The increased click detection capacity of the F-POD now enables fine-scale analysis of foraging or social behaviours (demonstrated by high click rates (Clausen et al.,2010)), that has perhaps been missed or underestimated using C-PODs. Additionally, F-PODs were found to be less effected by environmental noise levels within the 20-160KHz noise band, as indicated byNall (Clausen et al., 2018). It is plausible that decreased detections on the C-POD during periods of increasing environmental noise is a consequence of detector performance, which has been overcome during the development of the F-POD in conjunction with the increased click detection ability.
Long-term datasets and consistency of monitoring methods throughout the duration of monitoring programmes are important to enable long-term trends to be identified, particularly in areas of high conservation importance. Changing controllable factors such as monitoring equipment can skew our understanding of these long-term trends and in turn make it more difficult to interpret a change in habitat use, or behaviour of a species, which can be detrimental in the event of a disturbance or imminent threat. Static acoustic monitoring using PODs has been an integral part of cetacean monitoring programmes exploring habitat use and behaviour. Our results show than the C-POD and the F-POD are consistently comparable at the broad scale of identifying porpoise presence, and produce similar results when modelling environmental correlates of occurrence. However, the C-POD failed to detect the more nuanced patterns detected by the F-POD, particularly when investigating foraging behaviour versus occurrence. On account of its greater sensitivity and increased detection rates for harbour porpoise the F-POD certainly can be a useful tool to integrate into acoustic monitoring programmes. While the introduction of F-PODs into long-term time series is unlikely to change our understanding of the environmental drivers of occurrence, it is advisable that detections from one POD type should not be directly compared with detections from another as this could give erroneous results of increased occurrence due to differing detection rates rather than a true increase in individuals. Furthermore, any studies transitioning between PODs or combining C-POD and F-POD data should consider including POD type as a factor when conducting time-series. While the current study only investigated comparability in POD performance for detecting acoustic activity of harbour porpoise, it is likely that analogous differences would be seen for other cetacean species recorded by PODs. This study has given insights that the F-POD will be invaluable for future monitoring of harbour porpoise and other cetacean species, however care and consideration must be taken to make C-POD data adaptable for the integration into future studies.