Migration patterns and putative spawning habitats of Atlantic halibut (Hippoglossus hippoglossus) in the Gulf of St. Lawrence revealed by geolocation of pop-up satellite archival tags

Migration patterns and putative spawning habitats of Atlantic halibut (Hippoglossus hippoglossus)... Abstract Characterizing migratory behaviours contributes to the sustainable management of marine fishes by resolving stock structure and identifying the timing and locations of events within fish life cycles. The migratory behaviour of Atlantic halibut (Hippoglossus hippoglossus) in the Gulf of St. Lawrence (GSL), Canada was characterized over an annual cycle using pop-up satellite archival tags (n = 15). Daily probability density functions of individual halibut positions were estimated using a geolocation model specifically developed to track demersal fish species in the GSL. Reconstructed migration routes (n = 8) revealed that Atlantic halibut displayed seasonal migrations, moving from deeper offshore waters in the winter to shallower nearshore waters in the summer. Variability in migratory behaviours was observed among individuals tagged at the same location and time. One individual resided year round in the vicinity of the tagging site, three individuals displayed homing behaviour, and four individuals did not return to the tagging site. The identification of presumed spawning rises for two individuals suggested that spawning of Atlantic halibut occurred in the GSL. Although based on a limited number of individuals, these results suggest that Atlantic halibut in the GSL forms a philopatric population, supporting the current separate management of this stock from the adjacent Scotian Shelf and southern Grand Banks stock. Introduction Knowledge of individual fish movements benefits the sustainable management of fishery resources. Characterizing fish migratory behaviours helps in determining population structure (Cadrin et al., 2013), developing advanced stock assessment models (Cadigan, 2016), and designing marine protected areas (Grüss et al., 2011). Recent advances in fish tracking technologies have greatly facilitated the study of fish movement (reviewed by Hussey et al., 2015), and have contributed to the management of fishery resources; for instance, by improving the design of fishery closure to preserve spawning aggregations (Le Bris et al., 2013a), or by facilitating the implementation of dynamic spatial zoning to reduce bycatch (Hobday et al., 2010). Atlantic halibut (Hippoglossus hippoglossus) is one of the most valuable fish species per unit weight in Atlantic Canada fisheries (http://www.dfo-mpo.gc.ca/stats/commercial/sea-maritimes-eng.htm) but little is known about its migration patterns, habitat preferences, and stock structure, especially in the Gulf of St. Lawrence (GSL). This knowledge gap is partly explained by the low catchability of adult Atlantic halibut in the multispecies bottom-trawl surveys carried out by Fisheries and Oceans Canada (Bourdages et al., 2016). In Atlantic Canada, the species is managed as two separate stocks: the GSL stock, Northwest Atlantic Fisheries Organisation (NAFO) divisions 4RST; and the stock spanning the Scotian Shelf and southern Grand Banks (SSGB stock), NAFO divisions 3NOPs4VWX5Zc. These stock boundaries were defined based on several conventional tagging studies (McCracken, 1958; Stobo et al., 1988; den Heyer et al., 2012) which suggested extended movements across the Gulf of Maine, the Scotian Shelf, and the southern Grand Banks regions, and limited movements between these regions and the GSL. Conventional tagging studies often provide limited insight about fish movements and distributions (Bolle et al. 2005) because of the lack of data between tagging and recapture locations, and because they can reflect the spatial distribution of fishing effort to a greater extent than the spatial distribution of the resources. Complementary information is thus needed to fully characterize fish movements and population structure. Population structure arises from spatial or temporal segregations of spawning groups. Precise knowledge of spawning times and locations and connectivity among spawning groups is necessary for stock identification (Zemeckis et al., 2017). Atlantic halibut in the SSGB stock have been shown to perform seasonal migrations from summer feeding areas on the continual shelf to fall-winter potential spawning areas in deep water along the continental slope (Neilson et al., 1993; Armsworthy et al., 2014). Collection of male and female gonads in ripening and spent-recovering stages in May and June in the GSL led to speculation that halibut spawning occurs from February to April in the GSL (Kohler, 1967). Unfortunately, the lack of data during the winter period has precluded precise determination of halibut spawning times and locations in the GSL. Such seasonal data have been difficult to collect due to the formation of an ice cover and to the absence of a winter halibut fishery in the GSL. Advances in electronic tagging technologies including pop-up satellite archival tags (PSATs) offer new opportunities to study the migratory behaviour of Atlantic halibut throughout the year. Patterns in the vertical movements of fish recorded by electronic tags can indicate specific behaviours such as foraging or spawning. For instance, sudden change in the diving behaviour of Atlantic bluefin tuna (Thunnus thynnus) has been inferred to correspond to transitions from feeding to spawning (Block et al., 2001). Meanwhile, prolonged shifts in depth and temperature distributions of Atlantic cod (Gadus morhua) likely indicate that they are migrating from feeding to spawning areas (Grabowski et al., 2011). In the closely related Pacific halibut (Hippoglossus stenolepis), regularly spaced abrupt ascents and descents in the water column as observed in PSATs’ depth time series have been interpreted as spawning rises based on comparison with SCUBA observation (Seitz et al., 2005). Abrupt ascents suggestive of spawning events were also observed in Atlantic halibut equipped with PSATs (Armsworthy et al., 2014). Using PSATs could thus help identify Atlantic halibut feeding and spawning times and areas in the GSL. An inherent limitation in using PSATs on ground fish is the difficulty in estimating daily positions of tagged individuals. PSATs do not provide direct observations of individual positions between tagging and recapture. In addition, groundfish may be located at depths where it is difficult to obtain reliable light signals to use in traditional light-based geolocation methods (e.g. Sibert et al., 2003). One solution is to infer movements in and out of a specific area by visually comparing oceanographic characteristics (e.g. depth and temperature) observed inside and outside the area with environmental data recorded by tags (Nielsen and Seitz, 2017). Another solution is to statistically compare data recorded by the tags with spatial grids of oceanographic variables (e.g. tidal amplitude and phase, depth, temperature, salinity) to estimate fish daily positions (Andersen et al., 2007; Pedersen et al., 2008). The latter has been used successfully to track Atlantic cod equipped with data-storage tags in the GSL (Le Bris et al., 2013b), but has yet to be used to track Atlantic halibut. In this study, we used PSATs to characterize the annual migratory behaviours of Atlantic halibut in the GSL. Movement patterns were reconstructed using a geolocation model, specifically developed to track demersal fish species in the GSL. We hypothesized that, similar to the migratory behaviour of Atlantic halibut in other regions, individuals from the GSL would migrate from offshore deeper overwintering habitats to summer inshore shallower feeding habitats. We also hypothesized that halibut would reside year-round and spawn within the GSL. Material and methods Tagging operations We deployed a total of 20 PSATs (X-Tags, Microwave Telemetry Inc., Columbia, MD) on large (>100 cm fork length) Atlantic halibut. Tagging operations were carried out on-board commercial fishing vessels on 1–2 October 2013 at a single location off Port au Choix, Newfoundland (Figure 1, Table 1). Halibut were caught using longlines equipped with #16 circle hooks and baited with cut Atlantic herring, Clupea harengus. Soak times ∼10 h at depths between 180 and 220 m. Individuals larger than 100-cm fork length that appeared healthy were equipped with PSATs. Tags were attached using sterilized titanium darts linked to the tags by a 180-kg test monofilament tethers covered with black polyolefin shrink wrap tubing. The dart was inserted under the pterygiophores on the dorsal, eyed-side of the fish following methods developed on the closely related Pacific halibut, H. stenolepis (Seitz et al., 2005). Table 1. Tagging and recapture information for the 20 tags deployed on Atlantic halibut at a single location (50°36′N–57°34′W) on 1 or 2 October 2013. Tag ID  Fork length (cm)  Programmed deployment (months)  Pop-up   Type of data  Data (%)  Days of data  Distance (km)  Total distance (km)  Daily distance (km)  Date  Latitude (°N)  Longitude (°W)  914c  116  3  04 Jan 2014  NA  NA              915  130  3  12 Dec 2013  50°43′  57°25′  Transmitted  1  72  16      916  129  3  02 Jan 2014  50°23'  57°46′  Transmitted  1  94  28      917  113  6  03 Apr 2014  48°38′  61°05′  Transmitted  1  184  335      918  122  6  02 Apr 2014  48°71  60°11′  Transmitted  13  183  263      919  113  6  07 Apr 2014  50°21′  58°29′  Transmitted  5  188  44      920  126  12  01 Oct 2014  50°47′  57°23′  Archived    365  28  1463  4.02  921  121  12  02 Oct 2014  47°39′  57°19′  Transmitted  35  187  327      922+  108  12  NA  NA  NA              923  121  12  01 Oct 2014  49°49′  62°08′  Transmitted  41  206  336  1195  3.32  924  125  12  01 Oct 2014  50°39′  57°47′  Archived    365  16  284  0.77  925  122  12  02 Oct 2014  50°01′  65°51′  Transmitted  40  234  592  1709  4.69  926  139  12  02 Oct 2014  50°45′  57°28′  Archived    365  19  814  2.24  927  125  12  06 Oct 2014  51°40′  55°54′  Archived    370  166  1065  2.89  928+  146  12  NA  NA  NA              929+  165  12  NA  NA  NA              930b  115  12  NA  NA  NA              931  108  12  01 Oct 2014  50°42′  57°46′  Archived    365  18  1541  4.23  932a  140  12  24 Jul 2014  51°08′  56°59′  Archived    297  73  643  2.19  933  145  12  20 Dec 2013  50°42′  57°37′  Transmitted  7  80  13      Tag ID  Fork length (cm)  Programmed deployment (months)  Pop-up   Type of data  Data (%)  Days of data  Distance (km)  Total distance (km)  Daily distance (km)  Date  Latitude (°N)  Longitude (°W)  914c  116  3  04 Jan 2014  NA  NA              915  130  3  12 Dec 2013  50°43′  57°25′  Transmitted  1  72  16      916  129  3  02 Jan 2014  50°23'  57°46′  Transmitted  1  94  28      917  113  6  03 Apr 2014  48°38′  61°05′  Transmitted  1  184  335      918  122  6  02 Apr 2014  48°71  60°11′  Transmitted  13  183  263      919  113  6  07 Apr 2014  50°21′  58°29′  Transmitted  5  188  44      920  126  12  01 Oct 2014  50°47′  57°23′  Archived    365  28  1463  4.02  921  121  12  02 Oct 2014  47°39′  57°19′  Transmitted  35  187  327      922+  108  12  NA  NA  NA              923  121  12  01 Oct 2014  49°49′  62°08′  Transmitted  41  206  336  1195  3.32  924  125  12  01 Oct 2014  50°39′  57°47′  Archived    365  16  284  0.77  925  122  12  02 Oct 2014  50°01′  65°51′  Transmitted  40  234  592  1709  4.69  926  139  12  02 Oct 2014  50°45′  57°28′  Archived    365  19  814  2.24  927  125  12  06 Oct 2014  51°40′  55°54′  Archived    370  166  1065  2.89  928+  146  12  NA  NA  NA              929+  165  12  NA  NA  NA              930b  115  12  NA  NA  NA              931  108  12  01 Oct 2014  50°42′  57°46′  Archived    365  18  1541  4.23  932a  140  12  24 Jul 2014  51°08′  56°59′  Archived    297  73  643  2.19  933  145  12  20 Dec 2013  50°42′  57°37′  Transmitted  7  80  13      For all tags, distance is the linear distance between tagging and the tag pop-up locations. For tags that were geolocated, total distance is the distance travelled by the individual over the tracking period, estimated using the modes of probability density functions of daily fish locations. % of data indicates the percentage of transmitted data from non-physically recovered tags that were received via the ARGOS satellite system. a Indicates a tag that was recaptured by a harvester before the programmed pop-up date. b Indicates when a tag did not report. c Indicates a tag that reported but did not transmit data. Figure 1. View largeDownload slide Map of study area. Star shows the tagging location of Atlantic halibut, H. hippoglossus. White diamonds, triangles, and circles show pop-up locations of PSATs programmed to release 3, 6, and 12 months after tagging. Grey contours show 200, 400, and 500 m isobaths. Black lines delineate Northwest Atlantic Fisheries Organization (NAFO) divisions. Figure 1. View largeDownload slide Map of study area. Star shows the tagging location of Atlantic halibut, H. hippoglossus. White diamonds, triangles, and circles show pop-up locations of PSATs programmed to release 3, 6, and 12 months after tagging. Grey contours show 200, 400, and 500 m isobaths. Black lines delineate Northwest Atlantic Fisheries Organization (NAFO) divisions. PSATs recorded depth (±0.34–5.38 m resolution from 0 to 1296 m), temperature (±0.16–0.23°C resolution), and light intensity levels (±4× 10−5 Lux resolution at 555 nm) at 2-min intervals throughout the deployment period. On pre-programmed dates, PSATs released from the animal, floated to the surface and transmitted data to the Argos satellite system. When tags popped-off and surfaced, the first transmissions provided accurate GPS position which were considered fish’s final locations. We programmed three pop-up dates to meet two objectives. First, in order to obtain fish position at the onset (January) and near the end (April) of the presumed spawning season (DFO, 2015), three PSATs were programmed to pop-up 3 months after tagging, and three PSATs were programmed to pop-up 6 months after tagging (Table 1). Second, in order to obtain information on fish migratory behaviour over the course of an entire year, 14 PSATs were programmed to pop-up after 12 months. In October 2014, 12 months after tag deployment, we conducted a tag recovery mission using a CLS ARGOS RXG-134 goniometer (CLS America Inc., Lanham, MD, USA). The goniometer antenna mounted on a commercial fishing vessel received Argos pings broadcasted from floating PSATs, indicating signal direction and relative strength (Fisher et al., in review). Signal direction and strength enabled us to approach floating PSATs until we located them visually and could physically recover them. Physically recovered tags provided the full 2-min resolution archived data. When a tag could not be physically recovered, we used compressed subsets of recorded data transmitted to the Argos satellite system. The resolution of transmitted data varied among from 15, 30, and 60 min, depending upon total time at liberty. Because of the difference in data resolution between the “archived” and “transmitted” datasets, we separated depth and temperature records in two groups based on data resolution. Within each data group (archived and transmitted), data were pooled and the monthly and seasonal means and standard deviations of depth and temperature were calculated. Geolocation Preliminary analyses revealed that light levels recorded by the PSATs could not be used for halibut geolocation because individuals were generally distributed too deep to obtain reliable light information. Similarly, the tidal location method (Metcalfe and Arnold, 1997) could not be used due to the low tidal amplitude in the GSL (Le Bris, 2014). Instead, we used a geolocation model based on daily maximum depth and associated bottom temperature, taking advantage of the Gulf’s strong spatial gradients in bathymetry (Figure 1) and bottom temperature, and of the temporal persistence in bottom temperature gradients (Figure 2). The model assumed that fish visited the seafloor at least once a day and that the daily maximum depth recorded by the tag corresponded to the seafloor, a reasonable assumption for demersal flatfish like Atlantic halibut. Figure 2. View largeDownload slide Quarterly bottom temperature in the GSL in 2014. Based on available data, the winter (January–March) grid was developed using March data, the spring (April–June) grid using June data, the summer (July–September) grid using July–August data and the fall (October–December) grid using November data. Figure 2. View largeDownload slide Quarterly bottom temperature in the GSL in 2014. Based on available data, the winter (January–March) grid was developed using March data, the spring (April–June) grid using June data, the summer (July–September) grid using July–August data and the fall (October–December) grid using November data. Bottom temperature data Seasonal bottom temperature grids were computed using conductivity, temperature and depth (CTD) data obtained by Fisheries and Oceans Canada (Galbraith et al., 2015). Temperature profiles obtained from CTD casts were averaged into depth intervals of 1 m and spatially interpolated for each depth interval onto a 2-km resolution grid using the Barnes algorithm. Minimum and maximum temperatures measured in each of nine oceanographic subareas of the Gulf (Galbraith et al., 2015) were used to bind horizontal interpolation of temperature. The bottom temperature was then obtained for each grid-point by selecting the interpolated temperature at the depth level corresponding to the bathymetry from the Canadian Hydrographic Service (the method is fully described in Tamdrari et al., 2012). Finally, the standard deviation (SD) of the bathymetry value associated with each grid cell was estimated based on the bathymetry of the eight adjacent grid cells:   σz=1n∑i=1nzi-z-, (1) where zi is the bathymetry of the adjacent cell i with n = 8, and z- is the mean bathymetry of the eight adjacent cells. The same approach was used to estimate the SD of bottom temperature for each grid cell. High bathymetry and bottom temperature SD were found in areas of marked topographic changes such as along the slope of the channels and in coastal areas. Based on available data, we created bottom temperature grids for the months of November 2013, March 2014, June 2014, August 2014, and November 2014. March bottom temperatures were used for the winter (January–March), June bottom temperatures for the spring (April–June), August bottom temperatures for the summer (July–September), and November bottom temperatures for the fall (October–December, Figure 2). Geolocation model The geolocation model used to reconstruct migration routes of Atlantic halibut was based on the hidden Markov model (HMM) initially developed to track Atlantic cod in the North Sea (Pedersen et al., 2008; Thygesen et al., 2009). HMMs consist of two coupled stochastic models: the process model and the observation model. For each day, the process model simulated the fish movement using the diffusion equation:   ∂φx,t∂t=D∂2φx,t∂x2 + ∂2φx,t∂y2, (2) where φx,t was the probability density of the fish position (i.e. the probability that the fish was located at position x = [x,y] at time t = {1,…,T}), and D represented the diffusion rate. This partial differential equation was discretized in space onto the 2-km resolution grid (i.e. 331 × 476 regular grid cells) and time (number of recording days T) and solved using finite differences. The observation model assigned a likelihood value to each grid cell based on the match between the maximum depth and associated temperature recorded by the tag (z, tp) at day t with bathymetry and bottom temperature values from the grid (uz, utp):   Lz,tp|x= ∫z-Δzz+ΔzNz;uzx,σzxdz . ∫tp-Δtptp+ΔtpNtp;utpx,σtpxdtp (3) where Δtp and Δz were temperature and depth errors associated with tag measurements, σtp and σz were the SD of bottom temperature and bathymetry (Equation 1), and Nu,σ was a normal distribution of mean u and standard deviation σ. Based on information from the tag manufacturer, errors associated with tag measurements Δz and Δtp were set at 5.38 m and 0.23°C. For the last time step, a likelihood value of 1 was assigned to the grid cell where pop-up occurred or where the fish was caught, and 0 to other grid cells. Refined estimations of daily fish position were obtained by multiplying the probability density from the movement model with the likelihood value from the observation model and dividing by the normalization constant γt (Thygesen et al., 2009). Cells containing land were automatically assigned a zero probability. The diffusivity parameter was estimated by minimizing the negative log-likelihood function using Matlab v8.5.0 (The Mathworks, Natick, MA) fminbnd function:   LD= -∏t=1Tlog γt (4) Finally, a backward sweep of the model was performed (smoothing filter) to further refine location estimates (Thygesen et al., 2009). Incomplete data transmission from floating PSATs to the orbiting satellite can cause gaps in depth and temperature time series for an entire day or for several consecutive days. The geolocation model was thus adjusted to accommodate for gaps in transmitted time series. If temperature or depth was not available for a given day, the observation model used only the normal distribution corresponding to the available variable (i.e. depth or temperature, Equation 3). If both depth and temperature data were missing on a given day, the probability density function on that day was estimated using the movement model only (Equation 2) and was not refined with the observation model. Identification of putative spawning and feeding areas In order to identify potential feeding and spawning behaviours, we visually inspected depth and temperature data recorded by each PSAT. Specifically, we identified abrupt ascents and descents indicative of potential spawning rises (Seitz et al., 2005), and for marked, prolonged shifts in bathymetric and temperature distributions indicative of migration from and to feeding or spawning areas (Grabowski et al., 2011). Locations of specific behaviours were then estimated by normalizing daily probability density functions as follows:   φx,Γ= 1m1Γ∑j=1m∑tΓφix,t, (5) where Γ was the number of days t during which specific depth patterns indicative of spawning or feeding activity were observed and m was the number of individuals j displaying these depth patterns. Results Tracking success Of the 20 PSATs placed on Atlantic halibut, 14 popped off at the pre-programmed date, and 1 PSAT was captured in the fishery before its pre-programmed pop-up date (Table 1). Four PSATs did not report, and one reported but provided no data. Of the 14 PSATs that popped off at the pre-programmed time, 5 were physically recovered, which provided access to the 2-min temporal data resolution for 365 days (archived data). The one PSAT (#932) that was retrieved through the fishery also provided 2-min data for 296 days. The six PSATs programmed to pop-up 3 (n = 3) and 6 (n = 3) months after tagging transmitted data to satellites; however, low percentages (1–13%) of these data were received. Data transmission was probably impeded by the unusually thick and extensive ice cover that occurred in the GSL during the winter 2013–2014 (Galbraith et al., 2015). Finally, three PSATs recording data for 12 months could not be physically recaptured, and 35, 40, and 41% of the transmitted data were received (Table 1). These reception rates of transmitted data could be explained by numerous factors such as the position of the Argos satellites or the orientation of the tag antenna during transmission. Using the geolocation model, we were able to reconstruct the tracks of the six PSATs with archived data, and the tracks of two of the three tags with 12-month transmitted data (#923 and #925). We could not reconstruct the track of tag #921. The diffusivity parameter had an unusually high value leading to unrealistic daily movements. As expected, for the two tags with only transmitted datasets (#923 and #925), reconstructed tracks showed high uncertainties around daily geolocation estimates. Depth and bottom temperature error plots suggested that the model performed well (Figure 3 and Supplementary Figure S1). Indeed, for most of the tags, estimated depths and bottom temperatures matched closely with daily maximum depths and associated temperatures recorded by PSATs. For tags #927 and #931, however, estimated bottom temperatures differed from recorded temperatures mostly in July, when recorded temperatures were highly variable (Figure 3 and Supplementary Figure S1). These differences may be explained by the difficulty in capturing the high spatial and temporal variability in temperatures in nearshore waters during the summer when constructing bottom temperature grids based on CTD casts from scientific surveys. Such error might have introduced bias in estimates of summer distributions but is unlikely to have affected estimated distributions in other seasons. Figure 3. View large Download slide Daily maximum depth (left panels) and associated temperature (right panels) recorded by the PSATs (black lines) and estimated by the geolocation model (grey lines) for four Atlantic halibut. Full archived data were available for tags #924, #920, and #927. Transmitted data only were available for tag #925. Similar plots for the other four geolocated Atlantic halibut are provided in Supplementary Figure S1. Refer to the online version of the manuscript for a colored version of the figure. Figure 3. View large Download slide Daily maximum depth (left panels) and associated temperature (right panels) recorded by the PSATs (black lines) and estimated by the geolocation model (grey lines) for four Atlantic halibut. Full archived data were available for tags #924, #920, and #927. Transmitted data only were available for tag #925. Similar plots for the other four geolocated Atlantic halibut are provided in Supplementary Figure S1. Refer to the online version of the manuscript for a colored version of the figure. Seasonal distribution Reconstruction of daily locations of tagged Atlantic halibut showed seasonal variability in geographic distributions (Figure 4), depth preferences (Figure 5a), and temperature associations (Figure 5b). Atlantic halibut remained associated with the relatively deep (mean = 243 ± 32 m, max = 435 m, min = 128 m) and warm (mean = 5.8 ± 0.3°C, min = 1.1°C, max = 7.1°C) waters of the Esquiman Channel during winter. In the spring, halibut were closely associated with the 200 m isobath along the Gulf’s Channels (mean depth 197 ± 57 m, max = 333 m, min = 8 m; mean temperature 4.8 ± 1.4°C, min = −0.9°C, max = 9.3°C). In the summer, halibut were mostly distributed in inshore, shallow waters (mean = 113 ± 69 m, max = 269 m, min = 5 m) along the west coast of Newfoundland, the Quebec North Shore, and Anticosti Island, where water temperature was highly variable (Figure 5, mean = 5.2 ± 2.9°C, min =−1.3°C, max = 17.5°C). In the fall, migratory halibut progressively returned to relatively deep (208 ± 31 m, max = 309 m, min = 102 m) and warm (mean = 5.4 ± 0.6°C, min = 2.9°C, max = 11.7°C) waters prior to overwintering. Transmitted data show depth and temperature distributions remarkably similar to tags with archived data, except during the summer when tags with transmitted data recorded deeper depths than tags with archived data (Figure 5). Figure 4. View largeDownload slide Seasonal probability density functions of the position of the geolocated Atlantic halibut (n = 8); each panel depicts all eight fish during the specified season. The colour bar indicates the proportion of the density functions encompassed by coloured contours. Figure 4. View largeDownload slide Seasonal probability density functions of the position of the geolocated Atlantic halibut (n = 8); each panel depicts all eight fish during the specified season. The colour bar indicates the proportion of the density functions encompassed by coloured contours. Figure 5. View largeDownload slide Monthly mean (a) depths and (b) temperatures from the archived (black dots) and transmitted data (white dots). Error bars show one standard deviation of the mean. Figure 5. View largeDownload slide Monthly mean (a) depths and (b) temperatures from the archived (black dots) and transmitted data (white dots). Error bars show one standard deviation of the mean. Individual variability in migratory behaviour Migratory behaviour varied among tagged individuals (Figure 6 and Supplementary Figure S2). Based on the mode of daily probability density functions, one halibut (#924) travelled an estimated distance of less than 1 km day−1 and stayed within a 30-km radius of the tagging location throughout the year (Figure 6a). Evidence of movements from recorded depth time series ruled out the possibility that this fish was dead (Figure 3). Three (#920, #926, and #931) individuals travelled on average a total distance of 1273 km over the year from their tagging location to distant winter and summer areas but returned to 22 km on average of their initial release site (Table 1, Figure 6b and Supplementary Figure S2a and b). Halibut #927 and #932 displayed similar annual migration patterns characterized by an endpoint in the Strait of Belle Isle at 73 and 166 km north of the tagging location (Figure 6c and Supplementary Figure S2c). Finally, two individuals (#923 and #925) did not return to the Northeast GSL but rather migrated through the Anticosti Channel to complete their annual cycle in the northwest GSL (NAFO Division 4S) at 336 and 592 km from the original tagging location (Figure 6d and Supplementary Figure S1d). The full migration track of halibut #921 could not be reconstructed, but its tag popped off in NAFO Division 3Ps outside of the GSL (Figure 1). Figure 6. View largeDownload slide Probability density functions of four Atlantic halibut demonstrating the inter-individual variability in migratory behaviour. The colour bar indicates the proportion of the density functions encompassed by coloured contours. The black circle represents the initial tagging location in October 2013 and the black diamond shows pop-up location in October 2014. (a) Halibut #924: resident. (b) Halibut #920: homing. (c) Halibut #927: summer area in Strait of Belle Isle. (d) Halibut #925: summer area west of Anticosti Island. Figure 6. View largeDownload slide Probability density functions of four Atlantic halibut demonstrating the inter-individual variability in migratory behaviour. The colour bar indicates the proportion of the density functions encompassed by coloured contours. The black circle represents the initial tagging location in October 2013 and the black diamond shows pop-up location in October 2014. (a) Halibut #924: resident. (b) Halibut #920: homing. (c) Halibut #927: summer area in Strait of Belle Isle. (d) Halibut #925: summer area west of Anticosti Island. Putative spawning and feeding behaviours Visual inspection of depth time series recorded by PSATs revealed three particularly interesting patterns, the first two likely representing spawning behaviour and the third one feeding behaviour. Firstly, numerous consecutives abrupt ascents and descents in the water column were observed from 1st January to 28th February in tag #931 (Figure 7a). This pattern occurred when the halibut was distributed at the deepest depth (∼300 m) observed in the tag’s depth time series and was located at the junction of the Laurentian and Esquiman Channels (Figure 8a and b). Secondly, six abrupt ascents and descents ranging from 25 to 80 m in magnitude and from 4 to 20 min in duration were observed from 9th February to 25th February in tag #932 (Figure 7b). Three days consistently separated these rises. Rises were observed when the individual was distributed between 250 and 300 m deep in the Esquiman Channel (Figure 8a and b). Thirdly, four individuals showed shifts in their distributions in June from ∼200 to <100 m deep waters (Figure 8a). When distributed in shallow waters, small daily variations in recorded depths were accompanied by large fluctuations in recorded temperatures (Figure 8a). Individuals were geolocated in nearshore waters along the west coast of Newfoundland, the Strait of Belle Isle, and the Quebec North Shore when displaying this pattern (Figure 7c). Figure 7. View largeDownload slide Presumed spawning rises for halibut #931 (a) and #932 (b). Figure 7. View largeDownload slide Presumed spawning rises for halibut #931 (a) and #932 (b). Figure 8. View largeDownload slide (a) Depth (black lines) and temperature (grey lines) time-series recorded by PSATs #932, #931, #920, and #927. Circles indicate periods when putative spawning behaviours were observed: February–February 25 for tag #932 and January 1–February 28 for tag #931. Rectangles indicate periods when putative feeding behaviours were observed: 6 July–16 August for tag #920, 26 June–6 October for tag #927, 22 June–26 August 26 for tag #931, and 24 June–1 October for tag #932. (b) Probability density functions of fish positions during putative spawning behaviours. (c) Probability density functions of fish positions during putative feeding behaviours. The colour bar indicates the proportion of the density function encompassed by coloured contours. Light grey lines show 200 and 500 m isobaths. Tag IDs are indicated on probability density function maps. Figure 8. View largeDownload slide (a) Depth (black lines) and temperature (grey lines) time-series recorded by PSATs #932, #931, #920, and #927. Circles indicate periods when putative spawning behaviours were observed: February–February 25 for tag #932 and January 1–February 28 for tag #931. Rectangles indicate periods when putative feeding behaviours were observed: 6 July–16 August for tag #920, 26 June–6 October for tag #927, 22 June–26 August 26 for tag #931, and 24 June–1 October for tag #932. (b) Probability density functions of fish positions during putative spawning behaviours. (c) Probability density functions of fish positions during putative feeding behaviours. The colour bar indicates the proportion of the density function encompassed by coloured contours. Light grey lines show 200 and 500 m isobaths. Tag IDs are indicated on probability density function maps. Discussion Using a geolocation model that compared depth and temperatures recorded by PSATs with seasonal bottom temperature and bathymetry grids, we provided a detailed reconstruction of annual movement patterns of Atlantic halibut in the GSL. Reconstructed tracks revealed seasonal migrations within the GSL with individual variability in migratory behaviour, and identified potential halibut spawning grounds within the GSL. Daily probability density functions of fish positions showed that, generally, GSL Atlantic halibut spent the winter in the relatively deep and warm waters of the Esquiman and Laurentian Channels, migrated in the spring towards shallower nearshore waters where they spent the summer, and migrated back in the fall to their winter deep offshore habitat. These distribution patterns were consistent with seasonal distribution of past commercial catches in the GSL, indicating higher abundance of halibut in relatively deep waters in spring and fall, and in shallower waters in the summer (McCracken, 1958; Archambault and Grégoire, 1996). Distribution patterns were also consistent with catches from scientific surveys, indicating that halibut occupy the Gulf’s deep channels in winter and are located at depths around 200 m in the spring (DFO, 2015). Seasonal migrations from winter deeper water to summer shallower water appears to be a general characteristic of this species as it has been observed in the SSGB stock (Stobo et al., 1988; Armsworthy et al., 2014), the Gulf of Maine (Sigourney et al., 2006), and Norwegian waters (Godo and Haug, 1988). The winter migration to deeper waters may correspond to a spawning migration. In this study, we observed rapid ascents in the depth profiles recorded by two PSATs during the months of January and February. By comparing the spawning behaviour in flatfish species observed from SCUBA diving with depth patterns from high-resolution data collected with PSATs, Seitz et al. (2005) characterized these events as spawning rises in the closely related Pacific halibut. Specifically, tagged Pacific halibut displayed seven rapid ascents and descents, ∼65 h apart, then ascended to a shallower depth after the final spawning rise (Seitz et al., 2005). In our study, halibut #932 displayed six abrupt ascents and descents at 3-day intervals in February, and ascended to a shallower depth after the final rise. Atlantic halibut is a batch spawner and laboratory studies have shown that females require 2–4 days to hydrate successive batches of eggs (Methven et al. 1992, Finn et al. 2002). Consistent with observed spawning behaviour in Pacific halibut and with timing of hydration of egg batches in Atlantic halibut, we suggest that the rises observed in tag #932 corresponded to release of egg batches by a female Atlantic halibut. In contrast, tag #931 displayed numerous, more frequent and irregular rises in January and February than tag #932. Release of milt has been observed in January and February in captive male Atlantic halibut collected in Newfoundland waters (Methven et al., 1992), and frequent irregular rises have been observed during spawning events in other groundfish species such as Atlantic cod (Brawn, 1961) and haddock, Melanogrammus aeglefinus (Hawkins et al., 1967). We suggest that the frequent irregular rises displayed over a 2 months period by individuals #931 corresponded to milt release events from a male Atlantic halibut. We did not sex individuals in this study but further work combining PSATs and individuals of known sex will provide better understanding of sex-specific spawning behaviours in Atlantic halibut. Spawning times and locations of Atlantic halibut have historically been difficult to identify. In the northeast Atlantic spawning is thought to take place at depths of 700–1000 m on the continental slope from December to March with the peak of spawning activity in February (Haug, 1990). In Atlantic Canada, studies based on bottom trawl surveys and commercial catches have indicated that spawning occurs in late winter and early spring on Nova Scotia banks, from late winter to late spring on Newfoundland’s southern Grand Banks, and from February to April in the GSL (McCracken, 1958; Kohler, 1967). More recent histology and visual assessments of halibut gonad maturity from bottom trawl research surveys and commercial catches further indicated that peak spawning on the Scotian Shelf and southern Grand Banks might occur in November and December in deep slope waters (Neilson et al., 1993). A recent study using PSATs observed putative spawning rises between October and January at depths of 800–1000 m, presumably on the slope of the southern Grand Bank (Armsworthy et al., 2014). The presumed spawning rises in our study occurred in January and February at depth of 250–300 m in the Esquiman Channel and at the intersections of the Esquiman and Laurentian Channels. If the observed abrupt rises in PSATs depth time series actually reflect spawning rises, our results suggest that Gulf halibut are spatially and temporally segregated from Atlantic halibut from the SSGB during spawning. Only 2 of the 12 individuals that produced data during the spawning season showed a presumed spawning behaviour. Three factors may explain the low proportion of spawning individuals in our study. First, for tags with only transmitted data (n = 6), the lower temporal resolution and the presence of gaps in the depth time series may have masked abrupt ascents and descents in depth time series (Fisher et al., in review). Second, some fish tagged in this study may have been immature. Length at 50% maturity in GSL Atlantic halibut is 92 cm FL males and 130 cm FL for females (DFO, 2015). Length of tagged individuals varied from 108 to 165 FL; thus, it is possible that some females tagged in this study were immature. Third, in numerous iteroparous fish species, some individuals skip spawning during one or several years (Rideout et al., 2005). It has been suggested, based on the lack of evidence of spawning migrations in PSATs data, that skipped spawning might occur in both Pacific and Atlantic halibut (Loher and Seitz, 2008; Seitz et al., 2016). Skipped spawning might have contributed to the low proportion of tagged individuals displaying presumed spawning rises; however, there is currently no histological evidence of skipped spawning in Atlantic halibut. Beside abrupt ascents and descents in PSATs depth profiles, we also observed a marked shift in depth distributions in four halibut in the summer. With a remarkable consistency in timing, the four halibut crossed the cold intermediate layer, a mid-depth water layer with temperature below 0°C (Galbraith et al., 2015), during the last week of June and the first week of July to occupy 100 m shallow inshore waters. This pattern is very similar to a shift in distribution observed in Atlantic cod that was interpreted as a transition from spawning to feeding areas (Grabowski et al., 2011). While in inshore waters, halibut from this study experienced large fluctuations in water temperature over a limited depth range. Such fluctuations in temperature may reflect movements across the thermocline, as the maximum depth of the thermocline in the summer 2014 in the northeast Gulf was ∼75 m (Galbraith et al., 2015). The summer halibut locations estimated in this study are known to be important feeding grounds for Atlantic herring (C. harengus) and Atlantic cod (Dufour and Ouellet, 2007), two of the main fish prey species of halibut in the summer (Denis Chabot, Fisheries and Oceans Canada, Mont-Joli, unpublished diet composition data from scientific bottom trawl surveys). While further diet composition studies are needed to verify this hypothesis, we suggest that halibut migrate to inshore shallow waters in the summer to feed. Reconstructed migration tracks revealed substantial variability in migratory behaviours among individuals, including two different site fidelity behaviours. Three individuals migrated back to within 25 km of their tagging location after travelling as much as 1500 km over the course of the year, indicating a homing behaviour. In contrast, one individual adopted a resident behaviour and stayed within 30 km of its tagging location year round. Several studies using conventional tagging have indicated site-fidelity in Atlantic halibut (McCracken, 1958; Stobo et al. 1988); however, because conventional tagging data are limited to tagging and recapture positions, whether such site fidelity corresponded to either a homing or a residency behaviour could not be determined. More recently, the near continuous depth time series recorded by PSATs on Atlantic halibut in a Norwegian Fjord indicated that individuals did not cross the fjord’s sill, revealing year-round residency (Seitz et al., 2014). Other studies using PSATs suggested potential year-round residency in nearshore waters in the Gulf of Maine (Seitz et al., 2016) and along the continental slope in Atlantic Canada (Armsworthy et al, 2014); however, seasonal migrations could not be ruled out. Using a combination of acoustic tags, PSATs and a collection of oceanographic measurements, a recent study showed that both local residency and homing could occur within a group of fjord-dwelling Pacific halibut tagged in a single experiment (Nielsen and Seitz, 2017). Daily probability distributions of fish positions estimated in this study indicates that the two site fidelity behaviours can occur within a group of Atlantic halibut tagged at the same location and time. Structure in marine populations can be maintained by a spatial and/or temporal segregation of spawning components. We observed potential spawning in January–February in the GSL, which contrasts with presumed spawning from October to January along the continental slope south of the Grand Banks (Armsworthy et al. 2014). These differences in spawning time and location suggest the presence of a philopatric population of Atlantic halibut in the GSL and support the current separate management of Atlantic halibut stocks in eastern Canada. A previous genetic analysis on Atlantic halibut could not detect population structure in the Northwest Atlantic (Reid et al. 2005). The lack of genetic differentiation can be explained by the sampling outside of the spawning season, or by some mixing between populations preventing genetic difference but not necessarily preventing a level of population structure relevant to fisheries management. Another mechanism that can maintain population structure is the retention or return of spawning products to their natal area. In this study, all but one fish displayed high-site fidelity to the Gulf, consistent with the low mixing rates inside and outside the Gulf observed in several conventional tagging studies (McCracken, 1958; Stobo et al., 1988; den Heyer et al. 2012). This high-site fidelity supports the notion of a Gulf halibut population. Mixing could; however, occur through the dispersal of eggs and larvae. A recent study speculated that eggs spawned on the continental slope south of the Grand Banks could drift into the GSL (Armsworthy et al., 2014). They noted that Atlantic halibut eggs are neutrally buoyant in water masses where temperature and salinity ranged between 4.5 and 7.0°C and 33.8 and 35.0 (Haug et al., 1984), indicating that halibut eggs spawned south of Grand Banks should be found in the relatively deep waters of the Laurentian Channel. The deep waters of the Laurentian Channel travel from the continental slope towards the GSL at velocity of ∼1 cm s−1 (Gilbert, 2004). Hence, it would take close to a year for passive drifting eggs and larvae to travel from the continental slope to the entrance of the Gulf. This is significantly longer than the 6–7-month period required for halibut larvae to complete metamorphosis and the juveniles to settle in nursery areas (Trumble et al., 1993). Future studies combining genetics, ichthyoplankton collections, and numerical modelling of egg and larval transport would help evaluate potential connectivity between spawning grounds and the presence of population structure in Atlantic halibut in the northwest Atlantic. Conclusion and perspectives The high-commercial value of the Atlantic halibut fishery, the healthy status of the two Canadian stocks (DFO, 2015; Trzcinski and Bowen, 2016), and the contrasting perception of stock health on both sides of the Hague line (Shackell et al., 2016), has led to increased research effort to better understand migration, distribution, and stock structure of this commercially important species (e.g. Armsworthy et al., 2014,Seitz et al., 2016). Contributing to this effort, our study provided a detailed characterization of year-long movements of Atlantic halibut within the GSL and of potential spawning in the Gulf’s deep channels in January and February. However, given the low sampling size and the unique tagging location and time, this study should be considered a first step in understanding variability in halibut migratory behaviour and spawning locations in the GSL, and evaluating the connectivity between the GSL and SSGB stocks. Supplementary data Supplementary material is available at the ICESJMS online version of the manuscript. Acknowledgements We thank J. Spingle, E. Carruthers, D. Kamada, K. Krumsick, and the Fish, Food and Allied Workers for their important contributions to field work logistics. We also express our gratitude to Port au Choix halibut harvesters R. Dobbin, F. Dobbin, L. Gaslard, R. Gaslard, and P. Gaslard, who contributed to halibut and tagging and tag recovery operations. Funding This study was funded by the Newfoundland and Labrador Department of Fisheries and Aquaculture and by the Research and Development Corporation of Newfoundland and Labrador Ignite Grants to J.A.D.F. and D.R. References Andersen K. H., Nielsen A., Thygesen U. H., Hinrichsen H.-H., Neuenfeldt S. 2007. Using particle filter to geolcoate Atlantic cod (Gadus morhua) in the Baltic Sea, with special emphasis on determining uncertainty. Canadian Journal of Fisheries and Aquatic Sciences , 64: 618– 627. Google Scholar CrossRef Search ADS   Archambault D., Grégoire F. 1996. Revue des données historiques de pêche du flétan atlantique du golfe du Saint-Laurent (Divisions de l'OPANO 4RST) (1893-1995).  Department of Fisheries and Oceans, Canadian Stock Assessment Secretariat Research Document 96/56. Armsworthy S. L., Trzcinski M. K., Campana S. E. 2014. Movements, environmental associations, and presumed spawning locations of Atlantic halibut (Hippoglossus hippoglossus) in the northwest Atlantic determined using archival satellite pop-up tags. Marine Biology , 161: 645– 656. Google Scholar CrossRef Search ADS   Block B. A., Dewar H., Blackwell S. B., Williams T. D., Prince E. D., Farwell C. J., Boustany A., et al.   2001. Migratory movements, depth preferences, and thermal biology of Atlantic Bluefin Tuna. Nature , 293: 1310– 1314. Bolle L. J., Hunter E., Rijnsdorp A. D., Pastoors M. A., Metcalfe J. D., Reynolds J. D. 2005. Do tagging experiments tell the truth? Using electronic tags to evaluate conventional tagging data. ICES Journal of Marine Science , 62: 236– 246. Google Scholar CrossRef Search ADS   Bourdages H., Brassard C., Desgagnés M., Galbraith P., Gauthier J., Légaré B., Nozères C., et al.   2016. Preliminary results from the groundfish and shrimp multidisciplinary survey in August 2015 in the Estuary and northern Gulf of St. Lawrence. Department of Fisheries and Oceans Canada Science Advisory Secretariat Research Document 2016/004. Brawn, V. M. 1961. Reproductive behaviour of the cod (Gadus callarias L.). Behaviour, 18: 177–197. Cadigan N. G. 2016. A state-space stock assessment model for northern cod, including under-reported catches and variable natural mortality rates. Canadian Journal of Fisheries and Aquatic Sciences , 73: 296– 308. Google Scholar CrossRef Search ADS   Cadrin S. X., Kerr L. A., Mariani S. eds. 2013. Stock Identification Methods: Applications in Fishery Science . Academic Press, Cambridge. den Heyer C. E., Armsworthy A., Wilson S., Wilson G., Bajona L., Bond S., Trzcinski M. K. 2012. Atlantic halibut all-sizes tagging program summary report for 2006 to 2011. Canadian Technical Report Fisheries and Aquatic Sciences , 2992: vii+34p. DFO. 2015. Stock assessment of Atlantic halibut of the Gulf of St. Lawrence (NAFO Divisions 4RST) for 2013 and 2014. Department of Fisheries and Oceans Canada Science Advisory Secretariat Science Advisory Report 2015/023. Dufour R., Ouellet P. 2007. Estuary and Gulf of St. Lawrence marine ecosystem overview and assessment report. Canadian Technical Report of Fisheries and Aquatic Sciences , 2744E: vii+112p. Finn R. N., Østby G. C., Norberg B., Fyhn H. J., 2002. In vivo oocyte hydration in Atlantic halibut (Hippoglossus hippoglossus); proteolytic liberation of free amino acids, and ion transport, are driving forces for osmotic water influx. Journal of Experimental Biology , 205: 211– 224. Google Scholar PubMed  Fisher J. A. D., Robert D., Le Bris A., Loher T. In review. Pop-up satellite archival tag (PSAT) temporal data resolution affects interpretations of spawning behaviour and vertical habitat use. Animal Biotelemetry  Galbraith P. S., Chassé J., Nicot P., Caverhill C., Gilbert D., Pettigrew B., Lefaivre D., et al.   2015. Physical oceanographic conditions in the Gulf of St. Lawrence in 2014. DFO Canadian Science Advisory Secretariat Research Document , 2015: v + 82 p. Gilbert D. 2004. Propagation of temperature signals from the northwest Atlantic continental shelf edge into the Laurentian Channel. ICES CM , 2004: N:7. Godo O. R., Haug T. 1988. Tagging and recapture of Atlantic halibut (Hippoglossus hippoglossus L.) on the continental shelves of Eastern Canada, and off Western and Eastern Greenland. Journal of Northwest Atlantic Fisheries Science , 8: 25– 31. Google Scholar CrossRef Search ADS   Grabowski T. B., Thorsteinsson V., McAdam B. J., Marteinsdóttir G. 2011. Evidence of segregated spawning in a single marine fish stock: sympatric divergence of ecotypes in Icelandic cod? PLoS ONE , 6: e17528. Google Scholar CrossRef Search ADS PubMed  Grüss A., Kaplan D. M., Guénette S., Roberts C. M., Botsford L. W. 2011. Consequences of adult and juvenile movement for marine protected areas. Biological Conservation , 144: 692– 702. Google Scholar CrossRef Search ADS   Haug T. 1990. Biology of the Atlantic halibut, Hippoglossus hippoglossus (L. 1758). Advance in Marine Biology , 26: 1– 70. Google Scholar CrossRef Search ADS   Haug T., Kjørsvik, Solemdal P. 1984. Vertical distribution of Atlantic halibut (Hippoglossus hippoglossus) eggs. Canadian Journal of Fisheries and Aquatic Sciences , 41: 798– 804. Google Scholar CrossRef Search ADS   Hawkins, A. D., Chapman, K. J., and Symonds, D. J. 1967. Spawning of haddock in captivity. Nature, 215: 923–925. Hobday A. J., Hartog J. R., Timmiss T., Fielding J. 2010. Dynamic spatial zoning to manage southern bluefin tuna (Thunnus maccoyii) capture in a multi-species longline fishery. Fisheries Oceanography , 19: 243– 253. Google Scholar CrossRef Search ADS   Hussey N. E., Kessel S. T., Aarestrup K., Cooke S. J., Cowley P. D., Fisk A. T., Harcourt R. G., et al.   2015. Aquatic animal telemetry: a panoramic window into the underwater world. Science , 348: 1221– 1231. Google Scholar CrossRef Search ADS   Kohler A. C. 1967. Size at maturity, spawning season, and food of Atlantic halibut. Journal of Fisheries Research Board of Canada , 24: 53– 66. Google Scholar CrossRef Search ADS   Le Bris A. 2014. Mechanisms and consequences of variation in the migratory behaviour of Atlantic cod (Gadus morhua L.) in the northern Gulf of St. Lawrence. PhD dissertation, Memorial University of Newfoundland, St. John’s, NL. Le Bris A., Fréchet A., Wroblewski J. S. 2013a. Supplementing electronic tagging with conventional tagging to redesign fishery closed areas. Fisheries Research , 148: 106– 116. Google Scholar CrossRef Search ADS   Le Bris A., Fréchet A., Galbraith P. S., Wroblewski J. S. 2013b. Evidence for alternative migratory behaviours in the northern Gulf of St Lawrence population of Atlantic cod (Gadus morhua L.). ICES Journal of Marine Science , 70: 793– 804. Google Scholar CrossRef Search ADS   Loher T., Seitz A. C. 2008. Characterization of active spawning season and depth for eastern Pacific halibut (Hippoglossus stenolepis), and evidence of probable skipped spawning. Journal of Northwest Atlantic Fisheries Science , 41: 23– 36. Google Scholar CrossRef Search ADS   McCracken F. D. 1958. On the biology and fisheries of the Canadian Atlantic Halibut, Hippoglossus hippoglossus L. Journal of Fisheries Research Board of Canada , 15: 1269– 1311. Google Scholar CrossRef Search ADS   Metcalfe J. D., Arnold G. P. 1997. Tracking fish with electronic tags. Nature , 387: 665– 666. Google Scholar CrossRef Search ADS   Methven D. A., Crim L. W., Norberg B., Brown J. A., Goff G. P., Huse I. 1992. Seasonal reproduction and plasma levels of sex steroids and vitellogenin in Atlantic halibut (Hippoglossus hippoglossus). Canadian Journal of Fisheries and Aquatic Sciences , 49: 754– 759. Google Scholar CrossRef Search ADS   Nielsen J. K., Seitz A. C. 2017. Small-scale interannual site fidelity suggests the potential utility of protected areas for management of Pacific halibut. ICES Journal of Marine Science , doi:10.1093/icesjms/fsx040. Neilson J. D., Kearney J. F., Perley P., Sampson H. 1993. Reproductive biology of Atlantic halibut (Hippoglossus hippoglossus) in Canadian waters. Canadian Journal of Fisheries and Aquatic Sciences , 50: 551– 563. Google Scholar CrossRef Search ADS   Pedersen M. W., Righton D., Thygesen U. H., Andersen K. H., Madsen H. 2008. Geolocation of North Sea cod (Gadus morhua) using hidden Markov models and behavioural switching. Canadian Journal of Fisheries and Aquatic Sciences , 65: 2367– 2377. Google Scholar CrossRef Search ADS   Rideout R. M., Rose G. A., Burton M. P. M. 2005. Skipped spawning in female iteroparous fishes. Fish and Fisheries , 6: 50– 72. Google Scholar CrossRef Search ADS   Seitz A. C., Evans M. D., Courtney M. B., Kanwit J. K. 2016. Continental shelf residency by adult Atlantic halibut electronic tagged in the Gulf of Maine. Journal of Northwest Atlantic Fisheries Science , 48: 33– 40. Google Scholar CrossRef Search ADS   Seitz A. C., Michalsen K., Nielsen J. L., Evans M. D. 2014. Evidence of fjord spawning by southern Norwegian Atlantic halibut (Hippoglossus hippoglossus). ICES Journal of Marine Science , 71: 1142– 1147. Google Scholar CrossRef Search ADS   Seitz A. C., Norcross B. L., Wilson D., Nielsen J. L. 2005. Identifying spawning behavior in Pacific halibut, Hippoglossus stenolepis, using electronic tags. Environmental Biology of Fishes , 73: 445– 451. Google Scholar CrossRef Search ADS   Shackell N. L., Frank K. T., Nye J. A., den Heyer C. E. 2016. A transboundary dilemma: dichotomous designations of Atlantic halibut in the Northwest Atlantic. ICES Journal of Marine Science , 73: 1798– 1805. Google Scholar CrossRef Search ADS   Sibert J. R., Musyl M. K., Brill R. W. 2003. Horizontal movements of bigeye tuna (Thunnus obesus) near Hawaii determined by Kalman filter analysis of archival tagging data. Fisheries Oceanography , 12: 141– 151. Google Scholar CrossRef Search ADS   Sigourney D. B., Ross M. R., Brodziak J., Burnett J. 2006. Length at age, sexual maturity and distribution of Atlantic halibut, Hippoglossus hippoglossus L., off the Northeast USA. Journal of Northwest Atlantic Fisheries Science , 36: 81– 90. Google Scholar CrossRef Search ADS   Stobo W. T., Neilson J. D., Simpson P. G. 1988. Movements of Atlantic halibut (Hippoglossus hippoglossus) in the Canadian North Atlantic. Canadian Journal of Fisheries and Aquatic Sciences , 45: 484– 491. Google Scholar CrossRef Search ADS   Reid D. P., Pongsomboon S., Jackson T., McGowan C., Murphy C., Martin-Robichaud D., Reith M. 2005. Microsatellite analysis indicates an absence of population structure among Hippoglossus hippoglossus in the north-west Atlantic. Journal of Fish Biology , 67: 570– 576. Google Scholar CrossRef Search ADS   Tamdrari H., Castonguay M., Brêthes J.-C., Galbraith P. S., Duplisea D. E. 2012. The dispersal pattern and behaviour of Atlantic cod (Gadus morhua) in the northern Gulf of St. Lawrence: results from tagging experiments. Canadian Journal of Fisheries and Aquatic Sciences , 69: 112– 121. Google Scholar CrossRef Search ADS   Thygesen U., Pedersen M., Madsen H. 2009. Geolocating fish using hidden Markov models and data storage tags. In Tagging and Tracking of Marine Animals with Electronic Devices, pp. 277–293 . Ed. by Nielsen J., Arrizabalaga H., Fragoso N., Hobday A., Lutcavage M., Sibert J.. Springer, Dordrecht. Trumble R. J., Neilson J. D., Bowering W. R., McCaughran D. A. 1993. Atlantic halibut (Hippoglossus hippoglossus) and Pacific halibut (H. stenolepis) and their North American fisheries. Canadian Bulletin of Fisheries and Aquatic Sciences , 227: 84. Trzcinski, M. K., and Bowen, W. D. 2016. The recovery of the Atlantic halibut: a large, long-lived, and exploited marine predator. ICES Journal of Marine Science, doi:10.1093/icesjms/fsv266. Zemeckis D. R., Liu C., Cowles G. W., Dean M. J., Hoffman W. S., Martins D., Cadrin S. X. 2017. Seasonal movements and connectivity of an Atlantic cod (Gadus morhua) spawning component in the western Gulf of Maine. ICES Journal of Marine Science , doi:10.1093/icesjms/fsw190. © International Council for the Exploration of the Sea 2017. All rights reserved. For Permissions, please email: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png ICES Journal of Marine Science Oxford University Press

Migration patterns and putative spawning habitats of Atlantic halibut (Hippoglossus hippoglossus) in the Gulf of St. Lawrence revealed by geolocation of pop-up satellite archival tags

Loading next page...
 
/lp/ou_press/migration-patterns-and-putative-spawning-habitats-of-atlantic-halibut-F0VmlgwDZl
Publisher
Oxford University Press
Copyright
© International Council for the Exploration of the Sea 2017. All rights reserved. For Permissions, please email: journals.permissions@oup.com
ISSN
1054-3139
eISSN
1095-9289
D.O.I.
10.1093/icesjms/fsx098
Publisher site
See Article on Publisher Site

Abstract

Abstract Characterizing migratory behaviours contributes to the sustainable management of marine fishes by resolving stock structure and identifying the timing and locations of events within fish life cycles. The migratory behaviour of Atlantic halibut (Hippoglossus hippoglossus) in the Gulf of St. Lawrence (GSL), Canada was characterized over an annual cycle using pop-up satellite archival tags (n = 15). Daily probability density functions of individual halibut positions were estimated using a geolocation model specifically developed to track demersal fish species in the GSL. Reconstructed migration routes (n = 8) revealed that Atlantic halibut displayed seasonal migrations, moving from deeper offshore waters in the winter to shallower nearshore waters in the summer. Variability in migratory behaviours was observed among individuals tagged at the same location and time. One individual resided year round in the vicinity of the tagging site, three individuals displayed homing behaviour, and four individuals did not return to the tagging site. The identification of presumed spawning rises for two individuals suggested that spawning of Atlantic halibut occurred in the GSL. Although based on a limited number of individuals, these results suggest that Atlantic halibut in the GSL forms a philopatric population, supporting the current separate management of this stock from the adjacent Scotian Shelf and southern Grand Banks stock. Introduction Knowledge of individual fish movements benefits the sustainable management of fishery resources. Characterizing fish migratory behaviours helps in determining population structure (Cadrin et al., 2013), developing advanced stock assessment models (Cadigan, 2016), and designing marine protected areas (Grüss et al., 2011). Recent advances in fish tracking technologies have greatly facilitated the study of fish movement (reviewed by Hussey et al., 2015), and have contributed to the management of fishery resources; for instance, by improving the design of fishery closure to preserve spawning aggregations (Le Bris et al., 2013a), or by facilitating the implementation of dynamic spatial zoning to reduce bycatch (Hobday et al., 2010). Atlantic halibut (Hippoglossus hippoglossus) is one of the most valuable fish species per unit weight in Atlantic Canada fisheries (http://www.dfo-mpo.gc.ca/stats/commercial/sea-maritimes-eng.htm) but little is known about its migration patterns, habitat preferences, and stock structure, especially in the Gulf of St. Lawrence (GSL). This knowledge gap is partly explained by the low catchability of adult Atlantic halibut in the multispecies bottom-trawl surveys carried out by Fisheries and Oceans Canada (Bourdages et al., 2016). In Atlantic Canada, the species is managed as two separate stocks: the GSL stock, Northwest Atlantic Fisheries Organisation (NAFO) divisions 4RST; and the stock spanning the Scotian Shelf and southern Grand Banks (SSGB stock), NAFO divisions 3NOPs4VWX5Zc. These stock boundaries were defined based on several conventional tagging studies (McCracken, 1958; Stobo et al., 1988; den Heyer et al., 2012) which suggested extended movements across the Gulf of Maine, the Scotian Shelf, and the southern Grand Banks regions, and limited movements between these regions and the GSL. Conventional tagging studies often provide limited insight about fish movements and distributions (Bolle et al. 2005) because of the lack of data between tagging and recapture locations, and because they can reflect the spatial distribution of fishing effort to a greater extent than the spatial distribution of the resources. Complementary information is thus needed to fully characterize fish movements and population structure. Population structure arises from spatial or temporal segregations of spawning groups. Precise knowledge of spawning times and locations and connectivity among spawning groups is necessary for stock identification (Zemeckis et al., 2017). Atlantic halibut in the SSGB stock have been shown to perform seasonal migrations from summer feeding areas on the continual shelf to fall-winter potential spawning areas in deep water along the continental slope (Neilson et al., 1993; Armsworthy et al., 2014). Collection of male and female gonads in ripening and spent-recovering stages in May and June in the GSL led to speculation that halibut spawning occurs from February to April in the GSL (Kohler, 1967). Unfortunately, the lack of data during the winter period has precluded precise determination of halibut spawning times and locations in the GSL. Such seasonal data have been difficult to collect due to the formation of an ice cover and to the absence of a winter halibut fishery in the GSL. Advances in electronic tagging technologies including pop-up satellite archival tags (PSATs) offer new opportunities to study the migratory behaviour of Atlantic halibut throughout the year. Patterns in the vertical movements of fish recorded by electronic tags can indicate specific behaviours such as foraging or spawning. For instance, sudden change in the diving behaviour of Atlantic bluefin tuna (Thunnus thynnus) has been inferred to correspond to transitions from feeding to spawning (Block et al., 2001). Meanwhile, prolonged shifts in depth and temperature distributions of Atlantic cod (Gadus morhua) likely indicate that they are migrating from feeding to spawning areas (Grabowski et al., 2011). In the closely related Pacific halibut (Hippoglossus stenolepis), regularly spaced abrupt ascents and descents in the water column as observed in PSATs’ depth time series have been interpreted as spawning rises based on comparison with SCUBA observation (Seitz et al., 2005). Abrupt ascents suggestive of spawning events were also observed in Atlantic halibut equipped with PSATs (Armsworthy et al., 2014). Using PSATs could thus help identify Atlantic halibut feeding and spawning times and areas in the GSL. An inherent limitation in using PSATs on ground fish is the difficulty in estimating daily positions of tagged individuals. PSATs do not provide direct observations of individual positions between tagging and recapture. In addition, groundfish may be located at depths where it is difficult to obtain reliable light signals to use in traditional light-based geolocation methods (e.g. Sibert et al., 2003). One solution is to infer movements in and out of a specific area by visually comparing oceanographic characteristics (e.g. depth and temperature) observed inside and outside the area with environmental data recorded by tags (Nielsen and Seitz, 2017). Another solution is to statistically compare data recorded by the tags with spatial grids of oceanographic variables (e.g. tidal amplitude and phase, depth, temperature, salinity) to estimate fish daily positions (Andersen et al., 2007; Pedersen et al., 2008). The latter has been used successfully to track Atlantic cod equipped with data-storage tags in the GSL (Le Bris et al., 2013b), but has yet to be used to track Atlantic halibut. In this study, we used PSATs to characterize the annual migratory behaviours of Atlantic halibut in the GSL. Movement patterns were reconstructed using a geolocation model, specifically developed to track demersal fish species in the GSL. We hypothesized that, similar to the migratory behaviour of Atlantic halibut in other regions, individuals from the GSL would migrate from offshore deeper overwintering habitats to summer inshore shallower feeding habitats. We also hypothesized that halibut would reside year-round and spawn within the GSL. Material and methods Tagging operations We deployed a total of 20 PSATs (X-Tags, Microwave Telemetry Inc., Columbia, MD) on large (>100 cm fork length) Atlantic halibut. Tagging operations were carried out on-board commercial fishing vessels on 1–2 October 2013 at a single location off Port au Choix, Newfoundland (Figure 1, Table 1). Halibut were caught using longlines equipped with #16 circle hooks and baited with cut Atlantic herring, Clupea harengus. Soak times ∼10 h at depths between 180 and 220 m. Individuals larger than 100-cm fork length that appeared healthy were equipped with PSATs. Tags were attached using sterilized titanium darts linked to the tags by a 180-kg test monofilament tethers covered with black polyolefin shrink wrap tubing. The dart was inserted under the pterygiophores on the dorsal, eyed-side of the fish following methods developed on the closely related Pacific halibut, H. stenolepis (Seitz et al., 2005). Table 1. Tagging and recapture information for the 20 tags deployed on Atlantic halibut at a single location (50°36′N–57°34′W) on 1 or 2 October 2013. Tag ID  Fork length (cm)  Programmed deployment (months)  Pop-up   Type of data  Data (%)  Days of data  Distance (km)  Total distance (km)  Daily distance (km)  Date  Latitude (°N)  Longitude (°W)  914c  116  3  04 Jan 2014  NA  NA              915  130  3  12 Dec 2013  50°43′  57°25′  Transmitted  1  72  16      916  129  3  02 Jan 2014  50°23'  57°46′  Transmitted  1  94  28      917  113  6  03 Apr 2014  48°38′  61°05′  Transmitted  1  184  335      918  122  6  02 Apr 2014  48°71  60°11′  Transmitted  13  183  263      919  113  6  07 Apr 2014  50°21′  58°29′  Transmitted  5  188  44      920  126  12  01 Oct 2014  50°47′  57°23′  Archived    365  28  1463  4.02  921  121  12  02 Oct 2014  47°39′  57°19′  Transmitted  35  187  327      922+  108  12  NA  NA  NA              923  121  12  01 Oct 2014  49°49′  62°08′  Transmitted  41  206  336  1195  3.32  924  125  12  01 Oct 2014  50°39′  57°47′  Archived    365  16  284  0.77  925  122  12  02 Oct 2014  50°01′  65°51′  Transmitted  40  234  592  1709  4.69  926  139  12  02 Oct 2014  50°45′  57°28′  Archived    365  19  814  2.24  927  125  12  06 Oct 2014  51°40′  55°54′  Archived    370  166  1065  2.89  928+  146  12  NA  NA  NA              929+  165  12  NA  NA  NA              930b  115  12  NA  NA  NA              931  108  12  01 Oct 2014  50°42′  57°46′  Archived    365  18  1541  4.23  932a  140  12  24 Jul 2014  51°08′  56°59′  Archived    297  73  643  2.19  933  145  12  20 Dec 2013  50°42′  57°37′  Transmitted  7  80  13      Tag ID  Fork length (cm)  Programmed deployment (months)  Pop-up   Type of data  Data (%)  Days of data  Distance (km)  Total distance (km)  Daily distance (km)  Date  Latitude (°N)  Longitude (°W)  914c  116  3  04 Jan 2014  NA  NA              915  130  3  12 Dec 2013  50°43′  57°25′  Transmitted  1  72  16      916  129  3  02 Jan 2014  50°23'  57°46′  Transmitted  1  94  28      917  113  6  03 Apr 2014  48°38′  61°05′  Transmitted  1  184  335      918  122  6  02 Apr 2014  48°71  60°11′  Transmitted  13  183  263      919  113  6  07 Apr 2014  50°21′  58°29′  Transmitted  5  188  44      920  126  12  01 Oct 2014  50°47′  57°23′  Archived    365  28  1463  4.02  921  121  12  02 Oct 2014  47°39′  57°19′  Transmitted  35  187  327      922+  108  12  NA  NA  NA              923  121  12  01 Oct 2014  49°49′  62°08′  Transmitted  41  206  336  1195  3.32  924  125  12  01 Oct 2014  50°39′  57°47′  Archived    365  16  284  0.77  925  122  12  02 Oct 2014  50°01′  65°51′  Transmitted  40  234  592  1709  4.69  926  139  12  02 Oct 2014  50°45′  57°28′  Archived    365  19  814  2.24  927  125  12  06 Oct 2014  51°40′  55°54′  Archived    370  166  1065  2.89  928+  146  12  NA  NA  NA              929+  165  12  NA  NA  NA              930b  115  12  NA  NA  NA              931  108  12  01 Oct 2014  50°42′  57°46′  Archived    365  18  1541  4.23  932a  140  12  24 Jul 2014  51°08′  56°59′  Archived    297  73  643  2.19  933  145  12  20 Dec 2013  50°42′  57°37′  Transmitted  7  80  13      For all tags, distance is the linear distance between tagging and the tag pop-up locations. For tags that were geolocated, total distance is the distance travelled by the individual over the tracking period, estimated using the modes of probability density functions of daily fish locations. % of data indicates the percentage of transmitted data from non-physically recovered tags that were received via the ARGOS satellite system. a Indicates a tag that was recaptured by a harvester before the programmed pop-up date. b Indicates when a tag did not report. c Indicates a tag that reported but did not transmit data. Figure 1. View largeDownload slide Map of study area. Star shows the tagging location of Atlantic halibut, H. hippoglossus. White diamonds, triangles, and circles show pop-up locations of PSATs programmed to release 3, 6, and 12 months after tagging. Grey contours show 200, 400, and 500 m isobaths. Black lines delineate Northwest Atlantic Fisheries Organization (NAFO) divisions. Figure 1. View largeDownload slide Map of study area. Star shows the tagging location of Atlantic halibut, H. hippoglossus. White diamonds, triangles, and circles show pop-up locations of PSATs programmed to release 3, 6, and 12 months after tagging. Grey contours show 200, 400, and 500 m isobaths. Black lines delineate Northwest Atlantic Fisheries Organization (NAFO) divisions. PSATs recorded depth (±0.34–5.38 m resolution from 0 to 1296 m), temperature (±0.16–0.23°C resolution), and light intensity levels (±4× 10−5 Lux resolution at 555 nm) at 2-min intervals throughout the deployment period. On pre-programmed dates, PSATs released from the animal, floated to the surface and transmitted data to the Argos satellite system. When tags popped-off and surfaced, the first transmissions provided accurate GPS position which were considered fish’s final locations. We programmed three pop-up dates to meet two objectives. First, in order to obtain fish position at the onset (January) and near the end (April) of the presumed spawning season (DFO, 2015), three PSATs were programmed to pop-up 3 months after tagging, and three PSATs were programmed to pop-up 6 months after tagging (Table 1). Second, in order to obtain information on fish migratory behaviour over the course of an entire year, 14 PSATs were programmed to pop-up after 12 months. In October 2014, 12 months after tag deployment, we conducted a tag recovery mission using a CLS ARGOS RXG-134 goniometer (CLS America Inc., Lanham, MD, USA). The goniometer antenna mounted on a commercial fishing vessel received Argos pings broadcasted from floating PSATs, indicating signal direction and relative strength (Fisher et al., in review). Signal direction and strength enabled us to approach floating PSATs until we located them visually and could physically recover them. Physically recovered tags provided the full 2-min resolution archived data. When a tag could not be physically recovered, we used compressed subsets of recorded data transmitted to the Argos satellite system. The resolution of transmitted data varied among from 15, 30, and 60 min, depending upon total time at liberty. Because of the difference in data resolution between the “archived” and “transmitted” datasets, we separated depth and temperature records in two groups based on data resolution. Within each data group (archived and transmitted), data were pooled and the monthly and seasonal means and standard deviations of depth and temperature were calculated. Geolocation Preliminary analyses revealed that light levels recorded by the PSATs could not be used for halibut geolocation because individuals were generally distributed too deep to obtain reliable light information. Similarly, the tidal location method (Metcalfe and Arnold, 1997) could not be used due to the low tidal amplitude in the GSL (Le Bris, 2014). Instead, we used a geolocation model based on daily maximum depth and associated bottom temperature, taking advantage of the Gulf’s strong spatial gradients in bathymetry (Figure 1) and bottom temperature, and of the temporal persistence in bottom temperature gradients (Figure 2). The model assumed that fish visited the seafloor at least once a day and that the daily maximum depth recorded by the tag corresponded to the seafloor, a reasonable assumption for demersal flatfish like Atlantic halibut. Figure 2. View largeDownload slide Quarterly bottom temperature in the GSL in 2014. Based on available data, the winter (January–March) grid was developed using March data, the spring (April–June) grid using June data, the summer (July–September) grid using July–August data and the fall (October–December) grid using November data. Figure 2. View largeDownload slide Quarterly bottom temperature in the GSL in 2014. Based on available data, the winter (January–March) grid was developed using March data, the spring (April–June) grid using June data, the summer (July–September) grid using July–August data and the fall (October–December) grid using November data. Bottom temperature data Seasonal bottom temperature grids were computed using conductivity, temperature and depth (CTD) data obtained by Fisheries and Oceans Canada (Galbraith et al., 2015). Temperature profiles obtained from CTD casts were averaged into depth intervals of 1 m and spatially interpolated for each depth interval onto a 2-km resolution grid using the Barnes algorithm. Minimum and maximum temperatures measured in each of nine oceanographic subareas of the Gulf (Galbraith et al., 2015) were used to bind horizontal interpolation of temperature. The bottom temperature was then obtained for each grid-point by selecting the interpolated temperature at the depth level corresponding to the bathymetry from the Canadian Hydrographic Service (the method is fully described in Tamdrari et al., 2012). Finally, the standard deviation (SD) of the bathymetry value associated with each grid cell was estimated based on the bathymetry of the eight adjacent grid cells:   σz=1n∑i=1nzi-z-, (1) where zi is the bathymetry of the adjacent cell i with n = 8, and z- is the mean bathymetry of the eight adjacent cells. The same approach was used to estimate the SD of bottom temperature for each grid cell. High bathymetry and bottom temperature SD were found in areas of marked topographic changes such as along the slope of the channels and in coastal areas. Based on available data, we created bottom temperature grids for the months of November 2013, March 2014, June 2014, August 2014, and November 2014. March bottom temperatures were used for the winter (January–March), June bottom temperatures for the spring (April–June), August bottom temperatures for the summer (July–September), and November bottom temperatures for the fall (October–December, Figure 2). Geolocation model The geolocation model used to reconstruct migration routes of Atlantic halibut was based on the hidden Markov model (HMM) initially developed to track Atlantic cod in the North Sea (Pedersen et al., 2008; Thygesen et al., 2009). HMMs consist of two coupled stochastic models: the process model and the observation model. For each day, the process model simulated the fish movement using the diffusion equation:   ∂φx,t∂t=D∂2φx,t∂x2 + ∂2φx,t∂y2, (2) where φx,t was the probability density of the fish position (i.e. the probability that the fish was located at position x = [x,y] at time t = {1,…,T}), and D represented the diffusion rate. This partial differential equation was discretized in space onto the 2-km resolution grid (i.e. 331 × 476 regular grid cells) and time (number of recording days T) and solved using finite differences. The observation model assigned a likelihood value to each grid cell based on the match between the maximum depth and associated temperature recorded by the tag (z, tp) at day t with bathymetry and bottom temperature values from the grid (uz, utp):   Lz,tp|x= ∫z-Δzz+ΔzNz;uzx,σzxdz . ∫tp-Δtptp+ΔtpNtp;utpx,σtpxdtp (3) where Δtp and Δz were temperature and depth errors associated with tag measurements, σtp and σz were the SD of bottom temperature and bathymetry (Equation 1), and Nu,σ was a normal distribution of mean u and standard deviation σ. Based on information from the tag manufacturer, errors associated with tag measurements Δz and Δtp were set at 5.38 m and 0.23°C. For the last time step, a likelihood value of 1 was assigned to the grid cell where pop-up occurred or where the fish was caught, and 0 to other grid cells. Refined estimations of daily fish position were obtained by multiplying the probability density from the movement model with the likelihood value from the observation model and dividing by the normalization constant γt (Thygesen et al., 2009). Cells containing land were automatically assigned a zero probability. The diffusivity parameter was estimated by minimizing the negative log-likelihood function using Matlab v8.5.0 (The Mathworks, Natick, MA) fminbnd function:   LD= -∏t=1Tlog γt (4) Finally, a backward sweep of the model was performed (smoothing filter) to further refine location estimates (Thygesen et al., 2009). Incomplete data transmission from floating PSATs to the orbiting satellite can cause gaps in depth and temperature time series for an entire day or for several consecutive days. The geolocation model was thus adjusted to accommodate for gaps in transmitted time series. If temperature or depth was not available for a given day, the observation model used only the normal distribution corresponding to the available variable (i.e. depth or temperature, Equation 3). If both depth and temperature data were missing on a given day, the probability density function on that day was estimated using the movement model only (Equation 2) and was not refined with the observation model. Identification of putative spawning and feeding areas In order to identify potential feeding and spawning behaviours, we visually inspected depth and temperature data recorded by each PSAT. Specifically, we identified abrupt ascents and descents indicative of potential spawning rises (Seitz et al., 2005), and for marked, prolonged shifts in bathymetric and temperature distributions indicative of migration from and to feeding or spawning areas (Grabowski et al., 2011). Locations of specific behaviours were then estimated by normalizing daily probability density functions as follows:   φx,Γ= 1m1Γ∑j=1m∑tΓφix,t, (5) where Γ was the number of days t during which specific depth patterns indicative of spawning or feeding activity were observed and m was the number of individuals j displaying these depth patterns. Results Tracking success Of the 20 PSATs placed on Atlantic halibut, 14 popped off at the pre-programmed date, and 1 PSAT was captured in the fishery before its pre-programmed pop-up date (Table 1). Four PSATs did not report, and one reported but provided no data. Of the 14 PSATs that popped off at the pre-programmed time, 5 were physically recovered, which provided access to the 2-min temporal data resolution for 365 days (archived data). The one PSAT (#932) that was retrieved through the fishery also provided 2-min data for 296 days. The six PSATs programmed to pop-up 3 (n = 3) and 6 (n = 3) months after tagging transmitted data to satellites; however, low percentages (1–13%) of these data were received. Data transmission was probably impeded by the unusually thick and extensive ice cover that occurred in the GSL during the winter 2013–2014 (Galbraith et al., 2015). Finally, three PSATs recording data for 12 months could not be physically recaptured, and 35, 40, and 41% of the transmitted data were received (Table 1). These reception rates of transmitted data could be explained by numerous factors such as the position of the Argos satellites or the orientation of the tag antenna during transmission. Using the geolocation model, we were able to reconstruct the tracks of the six PSATs with archived data, and the tracks of two of the three tags with 12-month transmitted data (#923 and #925). We could not reconstruct the track of tag #921. The diffusivity parameter had an unusually high value leading to unrealistic daily movements. As expected, for the two tags with only transmitted datasets (#923 and #925), reconstructed tracks showed high uncertainties around daily geolocation estimates. Depth and bottom temperature error plots suggested that the model performed well (Figure 3 and Supplementary Figure S1). Indeed, for most of the tags, estimated depths and bottom temperatures matched closely with daily maximum depths and associated temperatures recorded by PSATs. For tags #927 and #931, however, estimated bottom temperatures differed from recorded temperatures mostly in July, when recorded temperatures were highly variable (Figure 3 and Supplementary Figure S1). These differences may be explained by the difficulty in capturing the high spatial and temporal variability in temperatures in nearshore waters during the summer when constructing bottom temperature grids based on CTD casts from scientific surveys. Such error might have introduced bias in estimates of summer distributions but is unlikely to have affected estimated distributions in other seasons. Figure 3. View large Download slide Daily maximum depth (left panels) and associated temperature (right panels) recorded by the PSATs (black lines) and estimated by the geolocation model (grey lines) for four Atlantic halibut. Full archived data were available for tags #924, #920, and #927. Transmitted data only were available for tag #925. Similar plots for the other four geolocated Atlantic halibut are provided in Supplementary Figure S1. Refer to the online version of the manuscript for a colored version of the figure. Figure 3. View large Download slide Daily maximum depth (left panels) and associated temperature (right panels) recorded by the PSATs (black lines) and estimated by the geolocation model (grey lines) for four Atlantic halibut. Full archived data were available for tags #924, #920, and #927. Transmitted data only were available for tag #925. Similar plots for the other four geolocated Atlantic halibut are provided in Supplementary Figure S1. Refer to the online version of the manuscript for a colored version of the figure. Seasonal distribution Reconstruction of daily locations of tagged Atlantic halibut showed seasonal variability in geographic distributions (Figure 4), depth preferences (Figure 5a), and temperature associations (Figure 5b). Atlantic halibut remained associated with the relatively deep (mean = 243 ± 32 m, max = 435 m, min = 128 m) and warm (mean = 5.8 ± 0.3°C, min = 1.1°C, max = 7.1°C) waters of the Esquiman Channel during winter. In the spring, halibut were closely associated with the 200 m isobath along the Gulf’s Channels (mean depth 197 ± 57 m, max = 333 m, min = 8 m; mean temperature 4.8 ± 1.4°C, min = −0.9°C, max = 9.3°C). In the summer, halibut were mostly distributed in inshore, shallow waters (mean = 113 ± 69 m, max = 269 m, min = 5 m) along the west coast of Newfoundland, the Quebec North Shore, and Anticosti Island, where water temperature was highly variable (Figure 5, mean = 5.2 ± 2.9°C, min =−1.3°C, max = 17.5°C). In the fall, migratory halibut progressively returned to relatively deep (208 ± 31 m, max = 309 m, min = 102 m) and warm (mean = 5.4 ± 0.6°C, min = 2.9°C, max = 11.7°C) waters prior to overwintering. Transmitted data show depth and temperature distributions remarkably similar to tags with archived data, except during the summer when tags with transmitted data recorded deeper depths than tags with archived data (Figure 5). Figure 4. View largeDownload slide Seasonal probability density functions of the position of the geolocated Atlantic halibut (n = 8); each panel depicts all eight fish during the specified season. The colour bar indicates the proportion of the density functions encompassed by coloured contours. Figure 4. View largeDownload slide Seasonal probability density functions of the position of the geolocated Atlantic halibut (n = 8); each panel depicts all eight fish during the specified season. The colour bar indicates the proportion of the density functions encompassed by coloured contours. Figure 5. View largeDownload slide Monthly mean (a) depths and (b) temperatures from the archived (black dots) and transmitted data (white dots). Error bars show one standard deviation of the mean. Figure 5. View largeDownload slide Monthly mean (a) depths and (b) temperatures from the archived (black dots) and transmitted data (white dots). Error bars show one standard deviation of the mean. Individual variability in migratory behaviour Migratory behaviour varied among tagged individuals (Figure 6 and Supplementary Figure S2). Based on the mode of daily probability density functions, one halibut (#924) travelled an estimated distance of less than 1 km day−1 and stayed within a 30-km radius of the tagging location throughout the year (Figure 6a). Evidence of movements from recorded depth time series ruled out the possibility that this fish was dead (Figure 3). Three (#920, #926, and #931) individuals travelled on average a total distance of 1273 km over the year from their tagging location to distant winter and summer areas but returned to 22 km on average of their initial release site (Table 1, Figure 6b and Supplementary Figure S2a and b). Halibut #927 and #932 displayed similar annual migration patterns characterized by an endpoint in the Strait of Belle Isle at 73 and 166 km north of the tagging location (Figure 6c and Supplementary Figure S2c). Finally, two individuals (#923 and #925) did not return to the Northeast GSL but rather migrated through the Anticosti Channel to complete their annual cycle in the northwest GSL (NAFO Division 4S) at 336 and 592 km from the original tagging location (Figure 6d and Supplementary Figure S1d). The full migration track of halibut #921 could not be reconstructed, but its tag popped off in NAFO Division 3Ps outside of the GSL (Figure 1). Figure 6. View largeDownload slide Probability density functions of four Atlantic halibut demonstrating the inter-individual variability in migratory behaviour. The colour bar indicates the proportion of the density functions encompassed by coloured contours. The black circle represents the initial tagging location in October 2013 and the black diamond shows pop-up location in October 2014. (a) Halibut #924: resident. (b) Halibut #920: homing. (c) Halibut #927: summer area in Strait of Belle Isle. (d) Halibut #925: summer area west of Anticosti Island. Figure 6. View largeDownload slide Probability density functions of four Atlantic halibut demonstrating the inter-individual variability in migratory behaviour. The colour bar indicates the proportion of the density functions encompassed by coloured contours. The black circle represents the initial tagging location in October 2013 and the black diamond shows pop-up location in October 2014. (a) Halibut #924: resident. (b) Halibut #920: homing. (c) Halibut #927: summer area in Strait of Belle Isle. (d) Halibut #925: summer area west of Anticosti Island. Putative spawning and feeding behaviours Visual inspection of depth time series recorded by PSATs revealed three particularly interesting patterns, the first two likely representing spawning behaviour and the third one feeding behaviour. Firstly, numerous consecutives abrupt ascents and descents in the water column were observed from 1st January to 28th February in tag #931 (Figure 7a). This pattern occurred when the halibut was distributed at the deepest depth (∼300 m) observed in the tag’s depth time series and was located at the junction of the Laurentian and Esquiman Channels (Figure 8a and b). Secondly, six abrupt ascents and descents ranging from 25 to 80 m in magnitude and from 4 to 20 min in duration were observed from 9th February to 25th February in tag #932 (Figure 7b). Three days consistently separated these rises. Rises were observed when the individual was distributed between 250 and 300 m deep in the Esquiman Channel (Figure 8a and b). Thirdly, four individuals showed shifts in their distributions in June from ∼200 to <100 m deep waters (Figure 8a). When distributed in shallow waters, small daily variations in recorded depths were accompanied by large fluctuations in recorded temperatures (Figure 8a). Individuals were geolocated in nearshore waters along the west coast of Newfoundland, the Strait of Belle Isle, and the Quebec North Shore when displaying this pattern (Figure 7c). Figure 7. View largeDownload slide Presumed spawning rises for halibut #931 (a) and #932 (b). Figure 7. View largeDownload slide Presumed spawning rises for halibut #931 (a) and #932 (b). Figure 8. View largeDownload slide (a) Depth (black lines) and temperature (grey lines) time-series recorded by PSATs #932, #931, #920, and #927. Circles indicate periods when putative spawning behaviours were observed: February–February 25 for tag #932 and January 1–February 28 for tag #931. Rectangles indicate periods when putative feeding behaviours were observed: 6 July–16 August for tag #920, 26 June–6 October for tag #927, 22 June–26 August 26 for tag #931, and 24 June–1 October for tag #932. (b) Probability density functions of fish positions during putative spawning behaviours. (c) Probability density functions of fish positions during putative feeding behaviours. The colour bar indicates the proportion of the density function encompassed by coloured contours. Light grey lines show 200 and 500 m isobaths. Tag IDs are indicated on probability density function maps. Figure 8. View largeDownload slide (a) Depth (black lines) and temperature (grey lines) time-series recorded by PSATs #932, #931, #920, and #927. Circles indicate periods when putative spawning behaviours were observed: February–February 25 for tag #932 and January 1–February 28 for tag #931. Rectangles indicate periods when putative feeding behaviours were observed: 6 July–16 August for tag #920, 26 June–6 October for tag #927, 22 June–26 August 26 for tag #931, and 24 June–1 October for tag #932. (b) Probability density functions of fish positions during putative spawning behaviours. (c) Probability density functions of fish positions during putative feeding behaviours. The colour bar indicates the proportion of the density function encompassed by coloured contours. Light grey lines show 200 and 500 m isobaths. Tag IDs are indicated on probability density function maps. Discussion Using a geolocation model that compared depth and temperatures recorded by PSATs with seasonal bottom temperature and bathymetry grids, we provided a detailed reconstruction of annual movement patterns of Atlantic halibut in the GSL. Reconstructed tracks revealed seasonal migrations within the GSL with individual variability in migratory behaviour, and identified potential halibut spawning grounds within the GSL. Daily probability density functions of fish positions showed that, generally, GSL Atlantic halibut spent the winter in the relatively deep and warm waters of the Esquiman and Laurentian Channels, migrated in the spring towards shallower nearshore waters where they spent the summer, and migrated back in the fall to their winter deep offshore habitat. These distribution patterns were consistent with seasonal distribution of past commercial catches in the GSL, indicating higher abundance of halibut in relatively deep waters in spring and fall, and in shallower waters in the summer (McCracken, 1958; Archambault and Grégoire, 1996). Distribution patterns were also consistent with catches from scientific surveys, indicating that halibut occupy the Gulf’s deep channels in winter and are located at depths around 200 m in the spring (DFO, 2015). Seasonal migrations from winter deeper water to summer shallower water appears to be a general characteristic of this species as it has been observed in the SSGB stock (Stobo et al., 1988; Armsworthy et al., 2014), the Gulf of Maine (Sigourney et al., 2006), and Norwegian waters (Godo and Haug, 1988). The winter migration to deeper waters may correspond to a spawning migration. In this study, we observed rapid ascents in the depth profiles recorded by two PSATs during the months of January and February. By comparing the spawning behaviour in flatfish species observed from SCUBA diving with depth patterns from high-resolution data collected with PSATs, Seitz et al. (2005) characterized these events as spawning rises in the closely related Pacific halibut. Specifically, tagged Pacific halibut displayed seven rapid ascents and descents, ∼65 h apart, then ascended to a shallower depth after the final spawning rise (Seitz et al., 2005). In our study, halibut #932 displayed six abrupt ascents and descents at 3-day intervals in February, and ascended to a shallower depth after the final rise. Atlantic halibut is a batch spawner and laboratory studies have shown that females require 2–4 days to hydrate successive batches of eggs (Methven et al. 1992, Finn et al. 2002). Consistent with observed spawning behaviour in Pacific halibut and with timing of hydration of egg batches in Atlantic halibut, we suggest that the rises observed in tag #932 corresponded to release of egg batches by a female Atlantic halibut. In contrast, tag #931 displayed numerous, more frequent and irregular rises in January and February than tag #932. Release of milt has been observed in January and February in captive male Atlantic halibut collected in Newfoundland waters (Methven et al., 1992), and frequent irregular rises have been observed during spawning events in other groundfish species such as Atlantic cod (Brawn, 1961) and haddock, Melanogrammus aeglefinus (Hawkins et al., 1967). We suggest that the frequent irregular rises displayed over a 2 months period by individuals #931 corresponded to milt release events from a male Atlantic halibut. We did not sex individuals in this study but further work combining PSATs and individuals of known sex will provide better understanding of sex-specific spawning behaviours in Atlantic halibut. Spawning times and locations of Atlantic halibut have historically been difficult to identify. In the northeast Atlantic spawning is thought to take place at depths of 700–1000 m on the continental slope from December to March with the peak of spawning activity in February (Haug, 1990). In Atlantic Canada, studies based on bottom trawl surveys and commercial catches have indicated that spawning occurs in late winter and early spring on Nova Scotia banks, from late winter to late spring on Newfoundland’s southern Grand Banks, and from February to April in the GSL (McCracken, 1958; Kohler, 1967). More recent histology and visual assessments of halibut gonad maturity from bottom trawl research surveys and commercial catches further indicated that peak spawning on the Scotian Shelf and southern Grand Banks might occur in November and December in deep slope waters (Neilson et al., 1993). A recent study using PSATs observed putative spawning rises between October and January at depths of 800–1000 m, presumably on the slope of the southern Grand Bank (Armsworthy et al., 2014). The presumed spawning rises in our study occurred in January and February at depth of 250–300 m in the Esquiman Channel and at the intersections of the Esquiman and Laurentian Channels. If the observed abrupt rises in PSATs depth time series actually reflect spawning rises, our results suggest that Gulf halibut are spatially and temporally segregated from Atlantic halibut from the SSGB during spawning. Only 2 of the 12 individuals that produced data during the spawning season showed a presumed spawning behaviour. Three factors may explain the low proportion of spawning individuals in our study. First, for tags with only transmitted data (n = 6), the lower temporal resolution and the presence of gaps in the depth time series may have masked abrupt ascents and descents in depth time series (Fisher et al., in review). Second, some fish tagged in this study may have been immature. Length at 50% maturity in GSL Atlantic halibut is 92 cm FL males and 130 cm FL for females (DFO, 2015). Length of tagged individuals varied from 108 to 165 FL; thus, it is possible that some females tagged in this study were immature. Third, in numerous iteroparous fish species, some individuals skip spawning during one or several years (Rideout et al., 2005). It has been suggested, based on the lack of evidence of spawning migrations in PSATs data, that skipped spawning might occur in both Pacific and Atlantic halibut (Loher and Seitz, 2008; Seitz et al., 2016). Skipped spawning might have contributed to the low proportion of tagged individuals displaying presumed spawning rises; however, there is currently no histological evidence of skipped spawning in Atlantic halibut. Beside abrupt ascents and descents in PSATs depth profiles, we also observed a marked shift in depth distributions in four halibut in the summer. With a remarkable consistency in timing, the four halibut crossed the cold intermediate layer, a mid-depth water layer with temperature below 0°C (Galbraith et al., 2015), during the last week of June and the first week of July to occupy 100 m shallow inshore waters. This pattern is very similar to a shift in distribution observed in Atlantic cod that was interpreted as a transition from spawning to feeding areas (Grabowski et al., 2011). While in inshore waters, halibut from this study experienced large fluctuations in water temperature over a limited depth range. Such fluctuations in temperature may reflect movements across the thermocline, as the maximum depth of the thermocline in the summer 2014 in the northeast Gulf was ∼75 m (Galbraith et al., 2015). The summer halibut locations estimated in this study are known to be important feeding grounds for Atlantic herring (C. harengus) and Atlantic cod (Dufour and Ouellet, 2007), two of the main fish prey species of halibut in the summer (Denis Chabot, Fisheries and Oceans Canada, Mont-Joli, unpublished diet composition data from scientific bottom trawl surveys). While further diet composition studies are needed to verify this hypothesis, we suggest that halibut migrate to inshore shallow waters in the summer to feed. Reconstructed migration tracks revealed substantial variability in migratory behaviours among individuals, including two different site fidelity behaviours. Three individuals migrated back to within 25 km of their tagging location after travelling as much as 1500 km over the course of the year, indicating a homing behaviour. In contrast, one individual adopted a resident behaviour and stayed within 30 km of its tagging location year round. Several studies using conventional tagging have indicated site-fidelity in Atlantic halibut (McCracken, 1958; Stobo et al. 1988); however, because conventional tagging data are limited to tagging and recapture positions, whether such site fidelity corresponded to either a homing or a residency behaviour could not be determined. More recently, the near continuous depth time series recorded by PSATs on Atlantic halibut in a Norwegian Fjord indicated that individuals did not cross the fjord’s sill, revealing year-round residency (Seitz et al., 2014). Other studies using PSATs suggested potential year-round residency in nearshore waters in the Gulf of Maine (Seitz et al., 2016) and along the continental slope in Atlantic Canada (Armsworthy et al, 2014); however, seasonal migrations could not be ruled out. Using a combination of acoustic tags, PSATs and a collection of oceanographic measurements, a recent study showed that both local residency and homing could occur within a group of fjord-dwelling Pacific halibut tagged in a single experiment (Nielsen and Seitz, 2017). Daily probability distributions of fish positions estimated in this study indicates that the two site fidelity behaviours can occur within a group of Atlantic halibut tagged at the same location and time. Structure in marine populations can be maintained by a spatial and/or temporal segregation of spawning components. We observed potential spawning in January–February in the GSL, which contrasts with presumed spawning from October to January along the continental slope south of the Grand Banks (Armsworthy et al. 2014). These differences in spawning time and location suggest the presence of a philopatric population of Atlantic halibut in the GSL and support the current separate management of Atlantic halibut stocks in eastern Canada. A previous genetic analysis on Atlantic halibut could not detect population structure in the Northwest Atlantic (Reid et al. 2005). The lack of genetic differentiation can be explained by the sampling outside of the spawning season, or by some mixing between populations preventing genetic difference but not necessarily preventing a level of population structure relevant to fisheries management. Another mechanism that can maintain population structure is the retention or return of spawning products to their natal area. In this study, all but one fish displayed high-site fidelity to the Gulf, consistent with the low mixing rates inside and outside the Gulf observed in several conventional tagging studies (McCracken, 1958; Stobo et al., 1988; den Heyer et al. 2012). This high-site fidelity supports the notion of a Gulf halibut population. Mixing could; however, occur through the dispersal of eggs and larvae. A recent study speculated that eggs spawned on the continental slope south of the Grand Banks could drift into the GSL (Armsworthy et al., 2014). They noted that Atlantic halibut eggs are neutrally buoyant in water masses where temperature and salinity ranged between 4.5 and 7.0°C and 33.8 and 35.0 (Haug et al., 1984), indicating that halibut eggs spawned south of Grand Banks should be found in the relatively deep waters of the Laurentian Channel. The deep waters of the Laurentian Channel travel from the continental slope towards the GSL at velocity of ∼1 cm s−1 (Gilbert, 2004). Hence, it would take close to a year for passive drifting eggs and larvae to travel from the continental slope to the entrance of the Gulf. This is significantly longer than the 6–7-month period required for halibut larvae to complete metamorphosis and the juveniles to settle in nursery areas (Trumble et al., 1993). Future studies combining genetics, ichthyoplankton collections, and numerical modelling of egg and larval transport would help evaluate potential connectivity between spawning grounds and the presence of population structure in Atlantic halibut in the northwest Atlantic. Conclusion and perspectives The high-commercial value of the Atlantic halibut fishery, the healthy status of the two Canadian stocks (DFO, 2015; Trzcinski and Bowen, 2016), and the contrasting perception of stock health on both sides of the Hague line (Shackell et al., 2016), has led to increased research effort to better understand migration, distribution, and stock structure of this commercially important species (e.g. Armsworthy et al., 2014,Seitz et al., 2016). Contributing to this effort, our study provided a detailed characterization of year-long movements of Atlantic halibut within the GSL and of potential spawning in the Gulf’s deep channels in January and February. However, given the low sampling size and the unique tagging location and time, this study should be considered a first step in understanding variability in halibut migratory behaviour and spawning locations in the GSL, and evaluating the connectivity between the GSL and SSGB stocks. Supplementary data Supplementary material is available at the ICESJMS online version of the manuscript. Acknowledgements We thank J. Spingle, E. Carruthers, D. Kamada, K. Krumsick, and the Fish, Food and Allied Workers for their important contributions to field work logistics. We also express our gratitude to Port au Choix halibut harvesters R. Dobbin, F. Dobbin, L. Gaslard, R. Gaslard, and P. Gaslard, who contributed to halibut and tagging and tag recovery operations. Funding This study was funded by the Newfoundland and Labrador Department of Fisheries and Aquaculture and by the Research and Development Corporation of Newfoundland and Labrador Ignite Grants to J.A.D.F. and D.R. References Andersen K. H., Nielsen A., Thygesen U. H., Hinrichsen H.-H., Neuenfeldt S. 2007. Using particle filter to geolcoate Atlantic cod (Gadus morhua) in the Baltic Sea, with special emphasis on determining uncertainty. Canadian Journal of Fisheries and Aquatic Sciences , 64: 618– 627. Google Scholar CrossRef Search ADS   Archambault D., Grégoire F. 1996. Revue des données historiques de pêche du flétan atlantique du golfe du Saint-Laurent (Divisions de l'OPANO 4RST) (1893-1995).  Department of Fisheries and Oceans, Canadian Stock Assessment Secretariat Research Document 96/56. Armsworthy S. L., Trzcinski M. K., Campana S. E. 2014. Movements, environmental associations, and presumed spawning locations of Atlantic halibut (Hippoglossus hippoglossus) in the northwest Atlantic determined using archival satellite pop-up tags. Marine Biology , 161: 645– 656. Google Scholar CrossRef Search ADS   Block B. A., Dewar H., Blackwell S. B., Williams T. D., Prince E. D., Farwell C. J., Boustany A., et al.   2001. Migratory movements, depth preferences, and thermal biology of Atlantic Bluefin Tuna. Nature , 293: 1310– 1314. Bolle L. J., Hunter E., Rijnsdorp A. D., Pastoors M. A., Metcalfe J. D., Reynolds J. D. 2005. Do tagging experiments tell the truth? Using electronic tags to evaluate conventional tagging data. ICES Journal of Marine Science , 62: 236– 246. Google Scholar CrossRef Search ADS   Bourdages H., Brassard C., Desgagnés M., Galbraith P., Gauthier J., Légaré B., Nozères C., et al.   2016. Preliminary results from the groundfish and shrimp multidisciplinary survey in August 2015 in the Estuary and northern Gulf of St. Lawrence. Department of Fisheries and Oceans Canada Science Advisory Secretariat Research Document 2016/004. Brawn, V. M. 1961. Reproductive behaviour of the cod (Gadus callarias L.). Behaviour, 18: 177–197. Cadigan N. G. 2016. A state-space stock assessment model for northern cod, including under-reported catches and variable natural mortality rates. Canadian Journal of Fisheries and Aquatic Sciences , 73: 296– 308. Google Scholar CrossRef Search ADS   Cadrin S. X., Kerr L. A., Mariani S. eds. 2013. Stock Identification Methods: Applications in Fishery Science . Academic Press, Cambridge. den Heyer C. E., Armsworthy A., Wilson S., Wilson G., Bajona L., Bond S., Trzcinski M. K. 2012. Atlantic halibut all-sizes tagging program summary report for 2006 to 2011. Canadian Technical Report Fisheries and Aquatic Sciences , 2992: vii+34p. DFO. 2015. Stock assessment of Atlantic halibut of the Gulf of St. Lawrence (NAFO Divisions 4RST) for 2013 and 2014. Department of Fisheries and Oceans Canada Science Advisory Secretariat Science Advisory Report 2015/023. Dufour R., Ouellet P. 2007. Estuary and Gulf of St. Lawrence marine ecosystem overview and assessment report. Canadian Technical Report of Fisheries and Aquatic Sciences , 2744E: vii+112p. Finn R. N., Østby G. C., Norberg B., Fyhn H. J., 2002. In vivo oocyte hydration in Atlantic halibut (Hippoglossus hippoglossus); proteolytic liberation of free amino acids, and ion transport, are driving forces for osmotic water influx. Journal of Experimental Biology , 205: 211– 224. Google Scholar PubMed  Fisher J. A. D., Robert D., Le Bris A., Loher T. In review. Pop-up satellite archival tag (PSAT) temporal data resolution affects interpretations of spawning behaviour and vertical habitat use. Animal Biotelemetry  Galbraith P. S., Chassé J., Nicot P., Caverhill C., Gilbert D., Pettigrew B., Lefaivre D., et al.   2015. Physical oceanographic conditions in the Gulf of St. Lawrence in 2014. DFO Canadian Science Advisory Secretariat Research Document , 2015: v + 82 p. Gilbert D. 2004. Propagation of temperature signals from the northwest Atlantic continental shelf edge into the Laurentian Channel. ICES CM , 2004: N:7. Godo O. R., Haug T. 1988. Tagging and recapture of Atlantic halibut (Hippoglossus hippoglossus L.) on the continental shelves of Eastern Canada, and off Western and Eastern Greenland. Journal of Northwest Atlantic Fisheries Science , 8: 25– 31. Google Scholar CrossRef Search ADS   Grabowski T. B., Thorsteinsson V., McAdam B. J., Marteinsdóttir G. 2011. Evidence of segregated spawning in a single marine fish stock: sympatric divergence of ecotypes in Icelandic cod? PLoS ONE , 6: e17528. Google Scholar CrossRef Search ADS PubMed  Grüss A., Kaplan D. M., Guénette S., Roberts C. M., Botsford L. W. 2011. Consequences of adult and juvenile movement for marine protected areas. Biological Conservation , 144: 692– 702. Google Scholar CrossRef Search ADS   Haug T. 1990. Biology of the Atlantic halibut, Hippoglossus hippoglossus (L. 1758). Advance in Marine Biology , 26: 1– 70. Google Scholar CrossRef Search ADS   Haug T., Kjørsvik, Solemdal P. 1984. Vertical distribution of Atlantic halibut (Hippoglossus hippoglossus) eggs. Canadian Journal of Fisheries and Aquatic Sciences , 41: 798– 804. Google Scholar CrossRef Search ADS   Hawkins, A. D., Chapman, K. J., and Symonds, D. J. 1967. Spawning of haddock in captivity. Nature, 215: 923–925. Hobday A. J., Hartog J. R., Timmiss T., Fielding J. 2010. Dynamic spatial zoning to manage southern bluefin tuna (Thunnus maccoyii) capture in a multi-species longline fishery. Fisheries Oceanography , 19: 243– 253. Google Scholar CrossRef Search ADS   Hussey N. E., Kessel S. T., Aarestrup K., Cooke S. J., Cowley P. D., Fisk A. T., Harcourt R. G., et al.   2015. Aquatic animal telemetry: a panoramic window into the underwater world. Science , 348: 1221– 1231. Google Scholar CrossRef Search ADS   Kohler A. C. 1967. Size at maturity, spawning season, and food of Atlantic halibut. Journal of Fisheries Research Board of Canada , 24: 53– 66. Google Scholar CrossRef Search ADS   Le Bris A. 2014. Mechanisms and consequences of variation in the migratory behaviour of Atlantic cod (Gadus morhua L.) in the northern Gulf of St. Lawrence. PhD dissertation, Memorial University of Newfoundland, St. John’s, NL. Le Bris A., Fréchet A., Wroblewski J. S. 2013a. Supplementing electronic tagging with conventional tagging to redesign fishery closed areas. Fisheries Research , 148: 106– 116. Google Scholar CrossRef Search ADS   Le Bris A., Fréchet A., Galbraith P. S., Wroblewski J. S. 2013b. Evidence for alternative migratory behaviours in the northern Gulf of St Lawrence population of Atlantic cod (Gadus morhua L.). ICES Journal of Marine Science , 70: 793– 804. Google Scholar CrossRef Search ADS   Loher T., Seitz A. C. 2008. Characterization of active spawning season and depth for eastern Pacific halibut (Hippoglossus stenolepis), and evidence of probable skipped spawning. Journal of Northwest Atlantic Fisheries Science , 41: 23– 36. Google Scholar CrossRef Search ADS   McCracken F. D. 1958. On the biology and fisheries of the Canadian Atlantic Halibut, Hippoglossus hippoglossus L. Journal of Fisheries Research Board of Canada , 15: 1269– 1311. Google Scholar CrossRef Search ADS   Metcalfe J. D., Arnold G. P. 1997. Tracking fish with electronic tags. Nature , 387: 665– 666. Google Scholar CrossRef Search ADS   Methven D. A., Crim L. W., Norberg B., Brown J. A., Goff G. P., Huse I. 1992. Seasonal reproduction and plasma levels of sex steroids and vitellogenin in Atlantic halibut (Hippoglossus hippoglossus). Canadian Journal of Fisheries and Aquatic Sciences , 49: 754– 759. Google Scholar CrossRef Search ADS   Nielsen J. K., Seitz A. C. 2017. Small-scale interannual site fidelity suggests the potential utility of protected areas for management of Pacific halibut. ICES Journal of Marine Science , doi:10.1093/icesjms/fsx040. Neilson J. D., Kearney J. F., Perley P., Sampson H. 1993. Reproductive biology of Atlantic halibut (Hippoglossus hippoglossus) in Canadian waters. Canadian Journal of Fisheries and Aquatic Sciences , 50: 551– 563. Google Scholar CrossRef Search ADS   Pedersen M. W., Righton D., Thygesen U. H., Andersen K. H., Madsen H. 2008. Geolocation of North Sea cod (Gadus morhua) using hidden Markov models and behavioural switching. Canadian Journal of Fisheries and Aquatic Sciences , 65: 2367– 2377. Google Scholar CrossRef Search ADS   Rideout R. M., Rose G. A., Burton M. P. M. 2005. Skipped spawning in female iteroparous fishes. Fish and Fisheries , 6: 50– 72. Google Scholar CrossRef Search ADS   Seitz A. C., Evans M. D., Courtney M. B., Kanwit J. K. 2016. Continental shelf residency by adult Atlantic halibut electronic tagged in the Gulf of Maine. Journal of Northwest Atlantic Fisheries Science , 48: 33– 40. Google Scholar CrossRef Search ADS   Seitz A. C., Michalsen K., Nielsen J. L., Evans M. D. 2014. Evidence of fjord spawning by southern Norwegian Atlantic halibut (Hippoglossus hippoglossus). ICES Journal of Marine Science , 71: 1142– 1147. Google Scholar CrossRef Search ADS   Seitz A. C., Norcross B. L., Wilson D., Nielsen J. L. 2005. Identifying spawning behavior in Pacific halibut, Hippoglossus stenolepis, using electronic tags. Environmental Biology of Fishes , 73: 445– 451. Google Scholar CrossRef Search ADS   Shackell N. L., Frank K. T., Nye J. A., den Heyer C. E. 2016. A transboundary dilemma: dichotomous designations of Atlantic halibut in the Northwest Atlantic. ICES Journal of Marine Science , 73: 1798– 1805. Google Scholar CrossRef Search ADS   Sibert J. R., Musyl M. K., Brill R. W. 2003. Horizontal movements of bigeye tuna (Thunnus obesus) near Hawaii determined by Kalman filter analysis of archival tagging data. Fisheries Oceanography , 12: 141– 151. Google Scholar CrossRef Search ADS   Sigourney D. B., Ross M. R., Brodziak J., Burnett J. 2006. Length at age, sexual maturity and distribution of Atlantic halibut, Hippoglossus hippoglossus L., off the Northeast USA. Journal of Northwest Atlantic Fisheries Science , 36: 81– 90. Google Scholar CrossRef Search ADS   Stobo W. T., Neilson J. D., Simpson P. G. 1988. Movements of Atlantic halibut (Hippoglossus hippoglossus) in the Canadian North Atlantic. Canadian Journal of Fisheries and Aquatic Sciences , 45: 484– 491. Google Scholar CrossRef Search ADS   Reid D. P., Pongsomboon S., Jackson T., McGowan C., Murphy C., Martin-Robichaud D., Reith M. 2005. Microsatellite analysis indicates an absence of population structure among Hippoglossus hippoglossus in the north-west Atlantic. Journal of Fish Biology , 67: 570– 576. Google Scholar CrossRef Search ADS   Tamdrari H., Castonguay M., Brêthes J.-C., Galbraith P. S., Duplisea D. E. 2012. The dispersal pattern and behaviour of Atlantic cod (Gadus morhua) in the northern Gulf of St. Lawrence: results from tagging experiments. Canadian Journal of Fisheries and Aquatic Sciences , 69: 112– 121. Google Scholar CrossRef Search ADS   Thygesen U., Pedersen M., Madsen H. 2009. Geolocating fish using hidden Markov models and data storage tags. In Tagging and Tracking of Marine Animals with Electronic Devices, pp. 277–293 . Ed. by Nielsen J., Arrizabalaga H., Fragoso N., Hobday A., Lutcavage M., Sibert J.. Springer, Dordrecht. Trumble R. J., Neilson J. D., Bowering W. R., McCaughran D. A. 1993. Atlantic halibut (Hippoglossus hippoglossus) and Pacific halibut (H. stenolepis) and their North American fisheries. Canadian Bulletin of Fisheries and Aquatic Sciences , 227: 84. Trzcinski, M. K., and Bowen, W. D. 2016. The recovery of the Atlantic halibut: a large, long-lived, and exploited marine predator. ICES Journal of Marine Science, doi:10.1093/icesjms/fsv266. Zemeckis D. R., Liu C., Cowles G. W., Dean M. J., Hoffman W. S., Martins D., Cadrin S. X. 2017. Seasonal movements and connectivity of an Atlantic cod (Gadus morhua) spawning component in the western Gulf of Maine. ICES Journal of Marine Science , doi:10.1093/icesjms/fsw190. © International Council for the Exploration of the Sea 2017. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Journal

ICES Journal of Marine ScienceOxford University Press

Published: Jan 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off