Priority marine feature surveys within the Small Isles MPA and surrounding waters

4. Discussion

4.1 Priority marine feature species and habitat components

The six-year Marine Scotland survey programme outlined in this report provides a baseline record of the SMI MPA at a spatial coverage and temporal resolution previously unavailable. The data analysed for this report strengthen the evidence base underpinning the Scottish MPA network and enable the further development and refinement of models, such as connectivity models, that can inform the establishment of future management measures to support an ecologically coherent network. Making these data available for such future applications is essential, thus the full dataset is archived and stored in line with the Scottish Government's Open Data Strategy and data relating to PMFs were provided to the Geodatabase of Marine features adjacent to Scotland (GeMS) curated by NatureScot. Access to the full dataset and associated information can be gained via the Marine Scotland Data Portal; DOI: 10.7489/1614-1.

The areas under survey in this study extend to waters in the SMI MPA that are deeper than previously recorded within the GeMS database, or visited in subsequent surveys (O'Dell et al., 2021), providing additional evidence on the distribution of PMF species or habitat components within the MPA. What is significant about the SMI MPA is that parts of the Sound of Canna have been not been exposed to high levels of activity from bottom-contacting towed gear historically, possibly due to challenging seabed topography and the navigational impediments created by a long-established dredge spoil disposal site (Howson et al., 2012). In this localised area the dredge spoil disposal site acts as a de facto refugium from the action of bottom-contacting towed gear (Shephard et al., 2012; Stirling et al., 2016). However, such refugia do not exist across the entirety of the MPA.

The findings detailed in this report confirm that the Sound of Canna has higher densities of some PMF species and habitat components, such as S. pallida, P. anguicomus and L. celtica, compared to other locations within the SMI MPA. For some species, the only occurrences within the SMI MPA were within the Sound of Canna. Modiolus modiolus was only recorded from two survey boxes within the Sound of Canna and Atrina fragilis was only recorded from three Sound of Canna boxes within the MPA and one box south of the MPA boundary.

Based on the survey results, F. quadrangularis appeared to show changes in box-level density over time. The lowest box-level densities for F. quadrangularis were consistently recorded after 2014, with densities of F. quadrangularis for all survey boxes within the SMI MPA having post-2014 values lower than 0.100 n.m^-2. The quantitative assessment of changes in F. quadrangularis over time for two survey boxes in the south of the Sound of Canna (S07 and S08) identified significant decreases in the density of F. quadrangularis between 2014 and all subsequent survey years (2015, 2016, 2017) (see Annex D of the Annex Materials document, Greathead et al. (2023)). These two boxes occurred within the boundaries of the MPA where fishing activity was apparent.

The two other species that appeared to show a decrease in density over time, A. fragilis and M. modiolus, were only encountered at a few sites and in generally low numbers. These apparent decreases in density could reflect inherent variation in the underlying data, a natural change in the suitability of environmental conditions for these species, or a response to a potential increase in human activities. Testing for these potential effects is not possible with the current dataset, particularly considering that the surveys were originally designed as the "before" stage of a BACI to assess the impact of management measures and not to provide a detailed view of all habitats and species within the SMI MPA.

The low power for detecting changes in A. fragilis and M. modiolus density reflects the challenges associated with repeatedly surveying these species over time. Both species can occur singly and in large aggregations, with A. fragilis reliant on soft sediment, whilst M. modiolus can colonise a range of substrates (Tyler-Walters and Wilding, 2017; Tyler-Walters, 2007). Patchily distributed individuals or aggregations are more likely to be missed by standard benthic survey methods relying on the drift of the vessel during the deployment of equipment, such as the surveys conducted on the MRV Alba na Mara. Consequently, it is possible that the apparent decreases in density for A. fragilis and M. modiolus reflect spatial gaps in survey coverage where patchily distributed individuals or aggregations were missed. From an MPA management perspective, it is worth noting that A. fragilis within S03 and S04 occurs within the localised de facto refugium provided by the dredge spoil disposal site, whereas S15 is located outside of this and has the potential to be subjected to higher fishing pressure.

For other species, such as A. sarsi, P. multiplicatus, S. pallida, P. anguicomus and L. celtica, there were no consistent patterns in density apparent across boxes or years. The quantitative assessment of changes in S. pallida over time for two survey boxes in the south of the Sound of Canna (S06 and S67) did not detect any significant change in S. pallida density between 2015, 2016 and 2017 (see Annex D). Some of the variation in density for A. sarsi, P. multiplicatus, S. pallida, P. anguicomus and L. celtica may, again, be an artefact of spatial gaps in survey coverage. Challenges in consistently surveying these species are linked to their biological traits, for example, a solitary lifestyle on the seafloor for the tube-dwelling anemones A. sarsi and P. multiplicatus; and a reliance on patchily distributed hard substrates such as rocks and boulders for S. pallida and P. anguicomus (Wilson, 2007; 2008). Previous work suggests that L. celtica has a reliance on shelly gravel (Rowley, 2007), although the SMI MPA surveys also encountered this species on other substrates. Hard substrates were often highly localised in the SMI MPA, interspersed by larger areas of habitat that would be regarded as unsuitable for many species requiring hard substrate. Ultimately, the lack of consistent density patterns for A. sarsi, P. multiplicatus, S. pallida, P. anguicomus and L. celtica points to high variability in the spatial distribution of these species on the seafloor, which complicates assessments of change in density over space and time and the use of sentinel hypothesis-based monitoring in their conservation.

The high variation in density and abundance of some target features over space and time in the Small Isles region suggests that the baseline data presented here may have limited power to inform monitoring efforts, on their own. This finding highlights the difficulties in using standard photographic and video survey methods to robustly assess the wide range of PMF species and habitat components within the SMI MPA. Subsequently, power analyses of wider monitoring efforts across the Scottish MPA network, including the study at the Small Isles, are needed to ensure that benthic monitoring can detect ecological change. The detailed F. quadrangularis and S. pallida case study (Annex D) goes some way towards demonstrating what is possible in terms of detecting significant change in survey boxes, where sufficient data are available.

Recent power analyses conducted for other marine environments indicate that very large areas with multiple replicates may need to be surveyed to detect changes in benthic species in response to anthropogenic impacts (Ardron et al., 2019). Consequently, where species are reliant on habitats with patchy distributions that are difficult to survey repeatedly over time, consideration is needed as to how the baseline data collected could be used to inform monitoring and assessment of future MPA management measures. For some species or habitats, the resolution of the data collected may not be sufficient to detect a change or deterioration in status over time or space, thus sentinel hypothesis-based monitoring may be less effective for some species.

4.2 The impact of fishing on Funiculina quadrangularis densities

The action of bottom-contacting towed gears is widely acknowledged as the main pressure causing activity on marine benthic ecosystems (Halpern et al., 2008; Hiddink et al., 2017; Rijnsdorp et al., 2018; Moffat et al., 2020), with the ability to cause large-scale changes in species distribution (e.g., Tillin et al., 2006; Hinz et al., 2009). However, careful design of baseline and monitoring surveys within MPAs is needed to determine if changes in density or distributions of seabed organisms and habitats result from the removal of anthropogenic pressures, such as fishing, instead of other coincidental factors.

The survey locations used in the study of the distribution of F. quadrangularis were selected based on suitable habitat (derived from SDMs initially but confirmed by observation) and the intensity of activity from bottom-contacting towed gear. However, despite the employment of a stratified design to limit the influence that environmental niche might play in their distribution, no effect of fishing on the density of F. quadrangularis was found. Anecdotally, this finding is supported by the existence of high densities of F. quadrangularis recorded in the Minch; nearly four times that recorded elsewhere despite this area being exposed to high levels of activity from bottom-contacting towed gear year-on-year (Annex A, Table A.11). The significant effect of environmental variables such as mud, depth, slope, and curvature on the density of F. quadrangularis found in our study accords with SDMs produced for this species around the West of Scotland (Greathead et al., 2015), although it should be noted that the stratified design employed in our study is likely to reduce their influence in the model.

Nevertheless, due to its erect profile and inability to retract into the substrate when disturbed, F. quadrangularis is thought to be particularly vulnerable to the action of the otter trawl used by the commercially important Nephrops norvegicus fishery (Greathead et al., 2005), which causes abrasion at the surface and deeper into the first few centimetres the seabed (Eigaard et al., 2016). In another Marine Scotland study looking at different seapen species, there was evidence of an association between trawling and a reduction in the abundance of seapens, although the authors acknowledged that the number of stations used was only just sufficient to detect this association (Harrald et al., 2018). Funiculina quadrangularis is a known bycatch species and has been recorded in the cod-ends of trawl nets used in the assessment of the Nephrops fishery (Milligan and Neil, 2010) and in other parts of trawl gear, such as the wings and top and bottom panels of the extension (Glendinning, 2012). Assessment of F. quadrangularis bycatch based on cod-end collection alone estimated a mean of 21 seapens being collected every hour of trawling (Milligan and Neil, 2010). A later survey including F. quadrangularis caught in other sections of the gear noted that bycatch of F. quadrangularis could be highly variable, with one trawl collecting 410 seapens from the cod end and 30 from other sections of gear, resulting in a combined mean gear value of 92 F. quadrangularis collected each hour of trawling time (Glendinning, 2012).

Variability in the number of seapens collected as bycatch by Glendinning (2012), despite using the same gear and methodology for survey, may suggest that F. quadrangularis can have a patchy distribution with localised areas of high density, which would complicate efforts to assess natural or fishing-induced changes in abundance of this species over space and time. This variability could account for the lack of a detectable change where we looked at F. quadrangularis density in the Sea of Hebrides (including the SMI MPA), Minch and Inner Sound. In that instance, bycatch could be expected to be dependent on the overlapping distribution and intensity of fishing with the abundance and distribution of F. quadrangularis. Considering that seapens were observed in our study to sometimes be lying at an oblique angle to the seabed when currents were high, it is possible that bycatch events may also be dependent on the direction of the prevailing current in relation to towing. Previous evidence does suggest that the sensitivity of other erect seapens to trawling impacts may vary depending on physical aspects such as their profile (Hill et al., 2000). It is also important to note that there is no information available on the age structure of the seapens observed on the seafloor in our study, thus it is not possible to determine if seapen mortality resulting from periodic fishing activity leads to seapens with a younger age structure occurring in areas that are more heavily fished.

In contrast to our findings relating to seapen densities, it should be noted that a recent study of F. quadrangularis distribution across the UK continental shelf found that the species is more likely to be recorded in areas where there had been no surface abrasion from fishing activity (Downie et al., 2021). The authors fitted SDMs for two separate eco-regions, and then used the model from one eco-region to predict to the distribution of F. quadrangularis in the other eco-region. Differences in fishing activity between the regions was proposed by the authors as a possible explanation for the poor model transferability that was found. Qualitative comparison of predictions and actual presences suggested F. quadrangularis may be restricted to sandier habitats in the Greater North Sea, compared to the muddier habitats the species prefers in the Celtic Seas Region (Downie et al., 2021). However, the hypothesis that fishing pressure caused the differences in habitat preference between the two regions was not formally tested by the authors. What is pertinent to our study is that Downie et al. (2021) only considered the effect of towed fishing activity on the distribution of F. quadrangularis at larger regional scales, compared to our smaller scale analysis conducted across multiple locations.

Vessel monitoring systems position data, such as those used by Downie et al. (2021) and in our study, typically have a polling rate of once every one or two hours (Shepperson et al., 2018). For our study, this limited spatial resolution of fishing activity equates to a 0.05 x 0.05 degree grid, approximately 15 km² at 60 °N latitude. The coarse resolution of these aggregated VMS data in relation to the spatial scale of seapen records makes it difficult to quantitatively analyse the impact of fishing activity on distributions, even at regional scales (Downie et al., 2021). For example, one of the dredge spoil disposal sites in the Sound of Canna, had an annual SAR of > 3 between 2014 – 2016 according to the aggregated VMS data, even though these areas cannot be fished because of depth and topography. A threshold of 0.43 SAR (equivalent to 0.5 hours of trawling within a km over a year) is typically used as a cut-off for significant adverse impacts (ICES, 2020); where fishing effort exceeds this, the biomass of seapens, sponges and certain corals on the seafloor is low. Conversely, areas containing suitable habitat that are subject to fishing effort below the 0.43 SAR threshold would be expected to have greater seapen biomass present that could subsequently be at risk from further fishing activities (NAFO, 2016). Considering that VMS data is likely to underrepresent true levels of fishing activity for vessels under 12 m, vessels that are known to be active in all three study sites (Kafas et al., 2014), this will further confound potential associations between towed fishing and seapen density.

4.3 The future surveys of priority marine features

The development of survey equipment in-house by Marine Scotland delivered marked improvements to the survey design. The ability to switch between lander and drop frame using an integrated frame meant that surveys could be altered quickly to match survey to sea state. Considering the vast amount of footage (65 hours) taken during the SMI MPA study, the future introduction of dedicated benthic image analysis and annotation software is expected to deliver further quality assurance and archival advantages for assessing changes in species density and distribution over space and time. Engaging with the wider scientific community in the development of standardised approaches for collecting, annotating, analysing and archiving imagery data through ongoing national (e.g. The Big Picture Project: JNCC, 2021) and international (e.g. The Challenger 150 Megafaunal Image-Based Technical Working Group: Challenger 150, 2021) initiatives will also ensure the datasets generated can be combined and shared to improve the evidence base for an ecologically coherent Scottish MPA network.

The development of Autonomous Underwater Vehicles (AUVs) that can record seabed imagery across a more tightly defined transect, and target survey locations more precisely than cameras towed from a vessel, may help to counteract the high levels of variation in box-level species density associated with habitat heterogeneity. Issues relating to water clarity and excessive towing speeds when using a drop frame may be improved with the development of AUV technology. Such technology could provide uninhibited access to the survey site, which is currently a very real impediment to traditional survey techniques in locations where static fishing gear is deployed. AUV platforms can also collect accurate geospatial information on the seafloor environment using attached sensor arrays, supporting more accurate predictions of where suitable habitats may occur for species of interest. However, AUVs can have limited ability to conduct imagery surveys on topographically complex habitats, such as rock walls, and the quality of imagery data collected will still be constrained by the resolution and lighting of the camera system mounted on the AUV platform (JNCC, 2018).

The baseline data collection design presented in this report is novel in that it uses several published (Greathead et al., 2015; Stirling et al., 2016) and unpublished SDMs to direct survey effort, in addition to traditional records of observation. Using these SDMs meant that survey effort was better directed across previously un-surveyed locations, focussing on those that presented habitat conditions suited to the features of interest. This technique enabled the identification of the first A. fragilis aggregation located outside the Sound of Canna in Scottish waters in recent times (Stirling et al., 2016), southeast of the island of Muck. The newly discovered A. fragilis aggregation, which lies to the south of the SMI MPA, falls within the Sea of Hebrides Nature Conservation MPA. This MPA was designated in December 2020, although it is not likely to provide additional protection to the A. fragilis aggregation, given that the protected features for this MPA are basking sharks, minke whales, fronts, and the marine geomorphology of the Scottish Shelf seabed (NatureScot, 2020a). However, there is a commitment to implement protection for PMFs (such as A. fragilis) outside of MPAs and proposals consulted on in 2018 included the location of the A. fragilis aggregation to the southeast of Muck within a draft PMF management zone (NatureScot, 2018a).