Migration routes and behaviour of Atlantic salmon and sea trout around operational wind turbine arrays in the Moray Firth, Scotland

4. Discussion

Salmon smolt migration to the marine environment

Successful emigration of salmon post-smolts into the Moray Firth was generally rapid (median ~5 days after release) and showed little inter-annual variation. Given that loss rates during the riverine stage can be high, between 0.3 and 7.0% per km migration distance in river systems (Jepsen et al., 1998; Thorstad et al., 2012), the annual average freshwater loss rate of 1% km^-1 in the current study was minimal, and slightly lower than that reported for the nearby river Conon (2.2% km^{- 1}) (Newton et al., 2021). However, after leaving freshwater, losses of salmon post-smolts in Y2 increased dramatically. The detection records of 10 individuals ended in Wick Bay, with two clearly preyed on by mammals. The remaining individuals generally moved quickly through this area, entering the Moray Firth during darkness. Previous studies highlighted the role of tidal currents in estuarine transit (Lacroix et al., 2004; Lothian et al., 2018), but the timing of movements through Wick Bay was apparently unrelated to tide, perhaps due to the comparatively small size of this 1.5 km long inlet, beyond which fish access open coastal waters.

Migration path/ bearing

On average, 44% of salmon post-smolts did not head directly north after leaving Wick Bay to take the shortest straight-line route to feeding grounds, but instead swam east, south-east, and to a lesser extent, directly south. While post-smolts that travelled through this first part of the Moray Firth on bearings of 88° through to 204° were intercepted by the marine receiver array with high probability, they may have passed through without detection due to the ≥500 m spacing of receivers. Therefore, 44% represents the minimum proportion of smolts taking these easterly to southerly trajectories. It has been suggested that such individuals at some point make a course adjustment and turn north on a trajectory to the Norwegian Sea/Greenland (Main et al., 2022; Ounsley et al., 2020). Directly crossing the North Sea and following the Norwegian Coastal Current north is potentially more energetically efficient than the alternate proposed route whereby smolts follow the coast north then head north-west to enter the Slope Current (Newton et al., 2021). The current findings support the growing evidence that the former is the preferred migration route for a significant proportion of salmon smolts leaving the east coast of Scotland. While smolts migrating along this route may gain an energetic advantage, it brings a greater likelihood of fish passing through one or multiple OWDs.

Post-smolt association with the offshore wind developments

The two operational OWDs investigated within the current study lie to the south-east of the River Wick, with the closest turbines located 19 km south-east from the estuary mouth. The minimum encounter rate of emigrating smolts with these OWDs was highly consistent between study years, with 18 and 20% of smolts that entered the Moray Firth migrated through the turbine zones.

Known residency in the OWDs was of short duration, with post-smolts spending a median of 36 minutes (maximum 5 h) among the turbines. This is what would be expected with transitory movement through the area, and the majority of reconstructed swim paths indeed showed relatively unidirectional movements, predominantly in easterly, southerly and south-easterly directions, resulting in 64% of last detections within Moray East. However, some individuals were in the vicinity of, and potentially in, the OWDs for extended periods: up to 22 h in Y1 and 51 h in Y2. Most acoustic telemetry studies on Atlantic salmon post-smolts in the marine environment have employed receivers configured in lines across potential migration routes (e.g. Lilly et al., 2024; Rodger et al., 2025). While effective for quantifying migration success and the proportion of individuals using each route, such arrays do not capture more complex movements. Use of the PrePARED grid array in the current study therefore provided a valuable opportunity to detect multi-directional movements and recurrent behaviour among migrating post-smolts, and these were apparent in the reconstructed swim paths of three individual fish. It was not possible to determine whether such movements were in response to factors associated with the OWDs in particular or simply reflected natural migration behaviour. Either way, these convoluted swim paths inevitably increased the exposure time of the post-smolts to the OWDs and resulted in long trajectories within the turbine zones; over 45 km in one case.

One area of concern about OWD impact pathways on diadromous fish is the potential for enhanced predation rate due to the aggregation of predators around scour protection, fixed pile and foundation tower structures (Bicknell, Gierhart, Newton, et al., 2025). Deducing possible predation events, and indeed any mortality, from acoustic telemetry data is an often overlooked yet important consideration due to the risk of introducing bias into data interpretation, particularly when inferring behaviours (Klinard & Matley, 2020). The use of depth and temperature sensors in Y2 of the current study provided valuable insight in this regard, with no evidence of predation events among the component of salmon post-smolts with sensors that were tracked in the marine array. There are, however, limitations to these data. Predation may have occurred undetected, for example, by avian predators targeting the shallow swimming post-smolts during which the tag would likely be ingested and subsequently undetectable as the bird flies away, or by other ectothermic fish with swim speeds and depths not atypical of the smolts themselves. Where depth and temperature data were not available, the deduction of possible predation events relied solely on the examination of reconstructed swim paths, looking for erratic behaviours and unusually fast swimming speed, which is inherently less reliable, particularly when using an array in which receiver spacings greatly exceed detection distance. Therefore, while the current study provided no indication of heightened predation on post-smolts within the OWDs, this topic warrants further investigation, ideally employing the emerging technologies such as prey-to-predator switching tags specifically designed to address questions about predation (Jacoby & Piper, 2025; Lennox et al., 2023).

Sea trout association with the offshore wind developments

Translocation of tagged adult sea trout into the centre of the two turbine zones in Beatrice and into the “Control Area” provided an opportunity to study the movements and behaviour of a statistically robust number of individuals within the marine environment. Although on average fish appeared to disperse outside of the receiver array within a few days and did not return, a substantial proportion of individuals in both Y1 (35%) and Y2 (42%) persisted in the area for over a week and re-entered the detection zones multiple times; one individual revisited this zone over 4 months after its first detection. While these often complex detection records provided valuable behavioural insights and demonstrated the value of this novel approach, it is important to acknowledge the inherent compromises associated with the translocation approach i.e. additional fish handling, transportation and acclimation between capture and release locations. Capture in a nearby marine aggregation site (Wester sea pools and bay), where fish were in saline water and had been using marine habitats (evidenced by the presence of sea lice on most individuals), followed by the slow acclimation between capture site water and release site water during the boat transportation were seen as biologically reasonable and pragmatic approaches to obtaining large amounts of behavioural data from a greater number of individuals than could have reasonably been expected if relying on in-site capture of study animals around the open water marine study site (OWDs).

Sea trout have been shown to exhibit high variation in their use of marine waters (Eldøy et al., 2015; Nevoux et al., 2019; Thorstad et al., 2016). Although they predominantly remain in coastal areas rather than travelling to distant feeding grounds offshore, long-distance migrations in the order of 100s of kilometres are not uncommon (Bekkevold, et al 2020; Birnie-Gauvin et al., 2019). Recent studies employing Data Storage Tags (DSTs) also suggest that sea trout make much greater use of deep water while in the marine environment than previously thought; a behaviour that the authors postulated was likely exploiting food resources (Artero et al., 2025; Kristensen et al., 2019). Depth data from the current study strongly support this, with over half of recorded depths at 40 m or greater. The deep diving behaviour of sea trout kelts has also been linked to using deeper water as a thermal refuge during warmer periods (Eldøy et al., 2017; Kristensen et al., 2018). Sea trout dwelling predominantly at the surface and making regular dives to around 20 m is commonly described in the literature and considered to be the result of fish foraging activity and the avoidance of the surface layers during the brightest daylight hours to reduce predation risk (Artero et al., 2025; Kristensen et al., 2018; Rikardsen et al., 2007). This behaviour was observed in the current study, but more prevalent was a tendency to dive to 40-60 m depth, often close to the sea floor, and remain there for extended periods, sometimes several days. Importantly, depth use characteristics in the OWDs were not demonstrably different from those in the “Control Area”, suggesting that fish were unlikely to be adopting this deep diving behaviour in response to the turbine structures and the potential increase of predation events associated with wind farm infrastructure.

Reduced marine survival is implicated as one cause of the decline in sea trout populations observed worldwide (Butler & Walker, 2007; ICES, 2013). As for salmon post-smolts, there is concern that sea trout could be the subject of heightened predation within the OWDs due to potential predator ‘hotspots’ among the structures. Conversely, sea trout are themselves predators of smaller fish and crustaceans (Knutsen et al., 2001; Rikardsen et al., 2006), which can be abundant in the artificial reefs created by OWDs (Svendsen et al., 2022 and references therein). The avoidance of predators in the first instance, or the attraction to rich foraging in the second, may be expected to have influenced the residency patterns and activities of the tagged sea trout. The results showed no difference in residency duration between Beatrice OWD and the Control Area, nor any difference in exploratory activity. Of the 62 adult sea trout with sufficient detection records to analyse, there was one presumed predation event, which was due to an ectothermic predator; no predation by mammals was detected. In the absence of consistent depth and temperature recording, such as that provided by DSTs, there is an inherent risk of predation bias, and it cannot be ruled out that some detections denote the movements of predators rather than sea trout. Ingested tags are usually expelled after a few days (Kristensen et al., 2019), but this can be substantially longer, up to 34 days for shark species for example (Brunnschweiler, 2009). Thorough examination of depth and temperature data in the context of reconstructed spatial movements rendered the risk of predation bias low for fish with relatively numerous detections; for fish with relatively few detections or detections with large temporal spread, the risk was increased. Accepting these limitations, data were of sufficient quantity and resolution to conclude that while predation of sea trout was observed, it was not widespread among study animals.