Movement patterns of temperate wrasses (Labridae) within a small Marine Protected Area.

The movement patterns of three commercially important wrasse (Labridae) species inside a small marine protected area (~ 0.15 km2 ) on the west coast of Norway were analysed over a period of 21 months. The mean distance between capture and recapture locations varied between 10-187 meters and was species and season specific. The extent of movement was not related to body size or sex. These results imply that a network of small strategically located marine protected areas can be used as management tools to protect wrasses from size- and sex-selective fishing mortality. This article is protected by copyright. All rights reserved.


Main text:
Species in the wrasse family (Labridae) are important mesopredators on rocky reefs in the Northeastern Atlantic (Costello, 1991;Norderhaug et al., 2005). They prey on a wide range of invertebrates and constitute an important part of the diet for larger predatory fish and seabirds (Bourlat et al., 2021;Dehnhard et al., 2021;Östman et al., 2016). Several of these wrasses are facultative cleaners in the wild (Breen, 1996;Hilldén, 1983;Potts, 1973), a behaviour that is exploited to combat salmon louse Lepeophtheirus salmonis (Krøyer 1837) infestation in salmonid aquaculture in Norway and elsewhere (Skiftesvik et al., 2013;Henly et al., 2021).
The Norwegian wrasse fishery has expanded substantially during the last decade, which has resulted in reduced abundance and body size in exploited populations of goldsinny Ctenolabrus rupestris (L.) and corkwing wrasse Symphodus melops (L.) (Halvorsen et al., 2017a(Halvorsen et al., , 2017b. These two species make up ~90 % of reported catch (Halvorsen et al., 2020).
The rock cook Centrolabrus exoletus (L.) is also frequently caught, but is typically discarded as bycatch. The larger Ballan wrasse (Labrus berggylta Ascanius 1767) is much less abundant, but is highly valued as a cleaner fish (Skiftesvik et al., 2014).
The fishery for wrasse is selective by species, size and sex, which increases the risk of overfishing (Halvorsen et al., 2020(Halvorsen et al., , 2016Kindsvater et al., 2020). Depletion of wrasses can have wider ecological consequences through top-down effects on smaller grazers and molluscs, or via bottom up effects on larger piscivores, since wrasses constitute an important part of the diet of seabirds and gadoid fishes (Dehnhard et al., 2021;Kraufvelin et al., 2020;Östman et al., 2016). However, most of the coastal regions affected by this fishery lack historical population data (Halvorsen et al., 2017a;Skiftesvik et al., 2015), making impact assessments difficult. No-take marine protected areas (MPAs) in regions with high fishing pressure represent valuable unfished reference areas and can play an important role in providing data to support an ecosystem based management strategy (Bourlat et al., 2021; This article is protected by copyright. All rights reserved. Costello, 2014;Halvorsen et al., 2017a). For example, length frequency distributions in and outside no-take MPAs can be used to estimate natural mortality and fishing mortality; essential parameters in stock assessment models (Wilson et al., 2014). At the same time, MPAs provide the opportunity to monitor indirect effects on important prey or predators of these wrasses in absence of fishing pressure (Bourlat et al., 2021).
We characterized the movement patterns of S. melops, C. rupestris and C. exoletus -the three most abundant wrasses in Norway and tested for interspecific and intraspecific differences.
Movement patterns may differ between these species due to, for example, differences in life history traits, diet and vertical distribution (Halvorsen et al., 2020;Sayer et al., 1995Sayer et al., , 1996. Further, males of S. melops build nests out of filamentous algae and provide paternal care, which may influence site-fidelity and sex-specific movements during the reproductive period of May-July (Halvorsen et al., 2017b;Potts, 1985). Information about fish movement, including ranges and site fidelity, is key to guide the design of MPAs that can protect the natural phenotypic variation of these species (Halvorsen et al., 2017b). We conducted a markrecapture survey on a small island within a designated wrasse MPA in western Norway ( Fig.   1) during the period 2017-2019. To ensure spatially similar sampling effort, we divided the areas into 16 smaller zones and alternated between sampling odd and even numbered zones each day (8 zones per day). The location and extent of zones were determined based on natural landmarks (e.g. a large rock, a small bay or other distinct change in the coastline) and distribution of suitable habitat (~hard substrate, visually assessed; see supporting information for the zones shown on map).
There were five sampling periods during which wrasse were collected using fyke nets; 2017: August/September (1); 2018: May (2) -July (3) -September (4); 2019: May (5). The fyke nets were set perpendicular to the shore, left overnight, and hauled the next day. Passing fish encountering the leader of the fyke net are guided into the cod-end; thus, capture depends on This article is protected by copyright. All rights reserved. fish movement. Each zone was visited three times during the first sampling period. Two different types of fyke nets were used in equal proportions; a short type specialized for catching cod (5 m single leader, 55 cm diameter entrance ring and leader mesh size of 30 mm) and a longer, more fine meshed type specialized for wrasses (leader: 7.8 m leader, entrance ring 70 cm diameter, 11 mm mesh size). The fine-meshed fyke nets captured a wider size range of the target species and, therefore, it was decided to continue with this type only in later sampling periods, during which each zone was sampled twice. Data from an additional sampling period carried out with pots in September 2018 (two weeks after the previous fyke net sampling) was also included in the analysis. These were standard wrasse pots baited with 40-80 g frozen prawns, Pandalus borealis (Krøyer, 1838) (pots were two chambered, 70 × 40 × 28 cm, 11 mm mesh size, 60 × 90 mm elliptical entrances, 12 mm wide escape openings).
Although the use of baited gear may influence the movement of reef fish (Bacheler et al., 2018), we assumed that this had little influence on our results given a balanced distribution of gear in the sampling area. Inclusion of pot-data improved statistical power, and the main conclusions (e.g. significance of model coefficients) were not affected if pot-data was excluded. The sampling effort was distributed as described for fyke nets and the traps were soaked overnight.
A handheld GPS (GARMIN GPSMAP 78s) was used to record the position of the gear. Upon capture, all fish were identified to the species level, and a handheld reader (BioMark HPR) was used to check for the presence of a passive integrated transponder tag (PIT, 2,12x12 mm, RFID solutions). Wrasses were visually sexed and measured for total length to the nearest mm, and those with a total length of ≥ 10 cm were PIT-tagged, following the procedure described for S. melops in Halvorsen et al (2017b). No fish were tagged during the pot sampling. All fish were carefully released at the site of capture. The method of capture, tagging and handling of live fish complies with the Norwegian food safety authorities' animal This article is protected by copyright. All rights reserved.
welfare laws, guidelines and policies as approved by the Norwegian Animal Research Authority (Project IDs: 8715 and 15307).
We PIT-tagged 3721 wrasse and recaptured 265 (195 S. melops, 57 C. rupestris, 13 C. exoletus; supporting information, Table 1). Capture-recaptures from the same sampling period were excluded. All data analysis was conducted using R v. 4.03 (R-Core team, 2020). The Rpackage gdistance (van Etten, 2017) was used to calculate the shortest waterway distance between all capture-recapture positions (Kersula & Seitz, 2019). We first converted our map polygon to a raster with 1818 x 1381 cells (1 x 1 meters), with each assigned a value corresponding to land (0) or sea (1). Movement between cells was constrained by a transition matrix with up to 16 possible directional movements to neighbour cells, while restricting movement to cells in the sea. The distance moved by an individual was calculated as the leastcost path distance between the cell of capture and the cell of recapture. Since sampling was restricted to a relatively small area, we estimated the expected mean movement under scenarios of completely free and random movement between all of the capture locations. To this end, the full list of coordinates of gear placements was duplicated ten times and then randomly paired. The mean distance between these simulated random movements was 289 meters (max 655 m). We assumed that there was minimal migration in or out of the study system, which is surrounded by waters of at least 20 meters depth (Figure 1). The shortest distance to the nearby island was ~85 meters (from the northernmost land point), but this stretch of water has sandy bottom, which is unfavourable habitat for wrasse such that they are unlikely to occupy or traverse it (Costello, 1991;Sayer et al., 1993;Gjøsaeter, 2002). A generalized linear model (glm; gamma-distribution, log-link) was used to test for interspecific differences in distance moved (m) between capture and recapture locations. The model included two factors and their interaction; Species (C. rupestris, C. exoletus, S. melops) and Interval (within year, including those captured and recaptured in May -July -September This article is protected by copyright. All rights reserved. 2018; and between years, including those captured and recaptured in different sampling years). This distinction is important because wrasses are territorial during spring and summer but hibernate or migrate to deeper water during winter (Costello, 1991;Hilldén, 1981;Sayer, 1999). Intraspecific differences in movement patterns were assessed using glm, including sex and body size as covariates. Model selection was carried out from among eight models, defined a priori (including the null model), by comparing their Akaike Information Criterion values, corrected for small sample sizes (AICc) (Burnham & Anderson, 2004). The highest ranked models did not include sex or size for any of the species (supporting information, Table 2). However, we acknowledge that the small sample sizes for C. rupestris and C. exoletus provides limited statistical power for detecting fine-scale differences. For both of these species, models including sex ranked second in both species (ΔAICc < 2), so we cannot exclude the possibility of subtle sexual differences in movement that could be have been detected with a larger data set. The underlying statistical assumptions in the glm models  Table 1). Thus, the small study area (~ 0.15 km 2 ) appears to provide sufficient space for natural movements for these two species. C. exoletus, had the highest mean distance moved (189 m, between years), but the 95 % confidence intervals were wide and overlapped the mean estimates for S. melops and for the simulated random movements (289 m). Thus, we acknowledge that a larger study area and This article is protected by copyright. All rights reserved.
sample size is necessary for assessing the movement of C. exoletus over a winter (between years). Nevertheless, like C. rupestris, the C. exoletus moved significantly shorter distances in the summer (25 m; within year).
It is possible that the longer between-year movements seen in C. exoletus and C. rupestris is related to territory relocation following hibernation in deeper waters (Hilldén, 1981). The depth range of C. rupestris (0-40 m) and C. exoletus (0-50 m) are wider than for S. melops (0-18 m; Halvorsen et al., 2020); and the two former species also show a seasonal shift in distribution towards deeper waters from June to October, which is not seen for the S. melops ( Figure 4 in Halvorsen et al., 2020). This is consistent with diving observations at 5, 10 and 18 m, where C. exoletus and C. rupestris was only observed in deep locations (10 and 18 m) during winter (February-March), while both had the highest abundance at 5 meters depth from May-June (Skog et al., 1994). S. melops was rarely observed at 10 m and never at 18 m in that survey. These studies suggest that C. rupestris and C. exoletus have a higher scope for vertical movements, while the S. melops are more confined to horizontal movements when searching for food or other resources and might remain in the same home range for winter hibernation.
The very high residency of C. rupestris is consistent with previous studies -individuals remain in the same territories/refuges across several months and even consecutive years (Hilldén, 1981;Sayer, 1999). Acoustic telemetry would be useful to obtain more detailed insights into home-range and habitat use of these species, for instance in revealing the location and duration of winter hibernation. Relatively small acoustic tags are now available (e.g. V5, VEMCO, Nova Scotia, Canada), but they would probably be too large for the majority of C. rupestris and C. exoletus caught in this study. Moreover, these small tags have a relatively short battery life (~months) which would limit the applicability of this method in comparing movement between different seasons in the same study.
This article is protected by copyright. All rights reserved.
Body size did not affect movement for any of these species (this study), nor in the larger L.
bergylta (Villegas-Ríos et al., 2013). This implies that these temperate wrasses maintain the size and location of the home range as they grow and age. Thus, the low mobility and affinity to shallow water is probably a strong determinant of population structure for these wrasses.
The Norwegian coastline is rugged, with deep fjords and numerous rocky reefs and islands (Simensen et al., 2021), which implies a fine-scaled structure of many isolated and demographically closed populations because of natural barriers. This is probably a major reason for the considerable differences in growth, age and abundance that have been observed at spatial scales of less than 5 km for these wrasses, and this has been related to variation in fishing pressure (Halvorsen et al., 2017a;Olsen et al., 2019). Hence, rebuilding a natural species composition and size distribution in isolated populations may be slow if there is no exchange of juveniles or adults with nearby areas.
MPAs are well suited to protect coastal species from overfishing (Fernández-Chacón et al., 2015;Sørdalen et al., 2020). For S. melops, a very small MPA (600 m coastline, comparable to the size of our study area) can provide protection for S. melops with moderate rates of spillover (fishing mortality MPA: 6-9 % vs Fished: 36-40 %; Halvorsen et al., 2017b). Since movement does not depend on size or sex (this study), a network of such small MPAs may be an effective approach to ensure that intensively fished regions support productive wrasse populations with natural sex-ratios and can also potentially benefit fisheries through spillover. This is something that should be assessed in future studies, along with dispersal and connectivity patterns during the egg and larval stages. Such information can be used to inform the optimal spatial design of a MPA network that also ensures sufficient exchange of recruits between fished areas and neighbouring MPAs (Baetscher et al., 2019;Harrison et al., 2020).
On the other hand, adult spill-over rates should be minimal if MPAs are to be used as unfished reference areas for scientific monitoring. This can be achieved if the MPA This article is protected by copyright. All rights reserved.
boundaries align with natural barriers to movement for wrasses (depth and/or larger stretches of sandy bottom), as is the case for the MPA in this study, or if MPAs are considerably larger, as in earlier studies where MPAs have been used to assess the impact of wrasse fisheries (Halvorsen et al., 2017a; four MPAs, mean size 2.8 km 2 ) and (Bourlat et al., 2021; single MPA, 2.6 km 2 ).
These observations on wrasse movement are relevant for another potential challenge in this fishery; the survival of smaller wrasses discarded as bycatch (Halvorsen et al., 2017b;Skiftesvik et al., 2014). In Norway, fishers are obligated to return any bycatch immediately at sea close to shore, although not necessarily at the site of capture (Halvorsen et al., 2020). The extent of this practice has not yet been assessed, but we highlight the potential of local negative consequences, such as higher predation risk for individuals that attempt to home or, in case of settlement in a new location, of increased competition for territories or food. This may be a more serious problem for C. rupestris, for which smaller individuals can make up almost half of the catch (Halvorsen et al., 2020). Further, the very high residency of C. rupestris implies that it would be practically difficult to release those caught as bycatch within their home range. Homing ability should, therefore, be further investigated, along with consequences for fitness and social interactions if translocated wrasses establish in a new area.