ICES Journal of Marine Science: Journal du Conseil

Peschko,, Verena;Müller,, Sabine;Schwemmer,, Philipp;Mercker,, Moritz;Lienau,, Peter;Rosenberger,, Tanja;Sundermeyer,, Janne;Garthe,, Stefan

doi: 10.1093/icesjms/fsaa045pmid: N/A

Abstract Grey seals have become an important part of the ecosystem in the southern North Sea over the last 50 years. However, little is known about their spatial utilization of the German North Sea, especially in relation to the dispersal and behaviour of grey seal pups after weaning. We investigated these little-known aspects by recording the movements of 11 grey seal pups born at the largest German colony for 1–9 months after leaving the colony between 2015 and 2017. The individuals moved widely throughout the southern North Sea, including some individuals that moved far along the Danish coast or to Dutch and UK waters. A point process modelling approach revealed that pups increased their distance to Helgoland during the first 70 d at sea. The frequency of inferred foraging behaviour increased until week 7 and decreased afterwards, whereas fast travelling behaviour increased throughout the whole study period. These findings reflect the transition from naive to more experienced pups, with gradual increases in foraging effort, range and efficiency to account for their increasing energy demands for survival and growth. This study provides the first characterization of the dispersal, behaviour, and spatial utilization of recently weaned grey seal pups in the southern North Sea, which profoundly extents our knowledge of an increasingly important top predator in that area. Introduction The abundance of grey seals (Halichoerus grypus) along European mainland coasts in the North Sea has increased since 1967, after being almost absent or very low for several hundred years (Scheibel and Weidel, 1988; Härkönen et al., 2007; Brasseur et al., 2015). In 2019, 6538 hauled out grey seals were counted during the moulting season in the entire Wadden Sea area and 1370 in the German North Sea (Cremer et al., 2019). The largest colony in German waters is located on Helgoland (54°11′N, 7°55′E), which is a small island 48 km from the German coastline in the south-eastern North Sea (Figure 1). Pup numbers counted on Helgoland increased by 38% between 2014/2015 and 2015/2016, by 12% between 2016/2017 and 2017/2018, and by 20% between 2017/2018 and 2018/2019 (Brasseur et al., 2016, 2018a; Cremer et al., 2019). During the coordinated surveys, 387 pups were counted on Helgoland and 1684 pups were recorded in the entire Wadden Sea area in the breeding season 2018/2019 (Cremer et al., 2019). The pup counts represent an index of the pup production and not the total number of pups born, as pups are not born exactly at the same time (Brasseur et al., 2018b). This number is only available for colonies accessible from land, i.e. Helgoland, where a total of 425 births were recorded during land-based counts in 2018/2019 (Cremer et al., 2019). Grey seal pups are weaned within 15–18 d (Kovacs, 1987; Haller et al., 1996) and then left by their mother. Following a post-weaning fast during which they lose up to 25% of their body mass (Noren et al., 2008), the pups start to forage for themselves (Noren et al., 2008; Bennett et al., 2010). After leaving their colony, pups have an average of 36 d in which to develop successful foraging tactics before their protein reserves will become depleted and they face the risk of starvation (Bennett et al., 2007). This is therefore a crucial phase of their life cycle, when they need to develop successful foraging strategies to account for the high energy demands for survival and growth. An individual’s first-year survival, recruitment into the breeding population and reproductive success are strongly affected by their body condition during their early development (Beauplet et al., 2005; Bowen et al., 2015). Knowledge of their behaviour and habitat use, and changes in these in relation to the pup’s age, is therefore essential to understanding the important processes influencing the population and its adaptability in times of major environmental changes and increasing anthropogenic pressures. However, little is currently known about the spatial utilization of the German North Sea by seals, especially in terms of grey seal pup dispersal and behaviour after weaning. Figure 1. Open in new tabDownload slide Movement of individual grey seal pups in the North Sea after weaning in the years 2015, 2016 and 2017. Helgoland (= tagging location) is indicated by a black star. Figure 1. Open in new tabDownload slide Movement of individual grey seal pups in the North Sea after weaning in the years 2015, 2016 and 2017. Helgoland (= tagging location) is indicated by a black star. Weaned grey seal pups tagged at colonies along the UK east coast increased their foraging trip length and duration, as well as their dive duration and depth, during their first weeks at sea (Bennett et al., 2010; Carter et al., 2017). After an explorative phase, pups reduced their trip duration, commuting between haul-out areas and putative foraging locations (Carter et al., 2017). About 1.3% of pups born in colonies located along the east coast of the United Kingdom emigrate to the Netherlands, accounting for ∼35% of the annual growth of the Netherlands breeding population (Brasseur et al., 2015). This further underlines the mobility of grey seal pups, which may settle far from their colony of birth. Given the increasing importance of grey seals as top predators in the southern North Sea, information on the habitat use and behaviour of pups born in colonies located in the southern North Sea is crucial to understanding their potential influence on the population dynamics and to interpret future changes in these dynamics. Recording the movements of grey seal pups born on Helgoland and analysing the data using a point process model (PPM; Renner et al., 2015) will thus help to shed light on the dispersal of grey seal pups and their spatial utilization and behaviour, as well as changes in these aspects in line with increasing age of the pups. This study aimed to provide some basic information on the changes in spatial utilization and behaviour of grey seal pups during their first weeks of life. These results will eventually support decision-making with respect to grey seal conservation. Methods Eleven grey seal pups, aged ∼6–8 weeks, were equipped with satellite tags on the island of Helgoland between 2015 and 2017. Individuals were caught using a hand-held net, weighed, and measured. No sedation was used for device fixation. The fur was cleaned, the tag was attached to the fur on the upper back of the individual using epoxy glue (Devcon; Shannon, County Clare, Ireland), and the individuals were released once the glue had hardened. Tags were attached to a rectangular neoprene base with seven €2-sized small feet at its edges (Supplementary 1, Figure S1). Only these feet were glued to the fur to break up the glue patch. The tags would thus fall off during the next moult, if not before. This method was successfully tested and monitored for ∼1 year on a harbour seal (Phoca vitulina) in a seal rehabilitation centre. Seal catching and tagging were conducted in accordance with the German Protection of Animals Act and with the permission of the Ministry of Energy, Agriculture, the Environment, Nature and Digitalization (file number V 242-7224.121-37). Two tag types were used (Supplementary 2, Table S1 and Figure S2), both manufactured by Wildlife Computers (Redmond, King County, WA, USA): SPLASH (in 2015; 215 g) and SPOT tags (in 2016 and 2017; 119 g). Both were programmed to generate an Argos position when the animal surfaced, with a maximum of 400 uplinks per day (SPLASH), or 38/h with a maximum of 800 uplinks per day (SPOT), and to pause when the animal hauled out on land. No difference in tag performance was observed. Trip identification Argos positions were filtered using the Douglas Argos-Filter option in Movebank [settings: ‘distance angle rate’ filter, keep location class = 3, maxredun = 7, minrate = 10, ratecoeff = 25 (movebank.org; Douglas et al., 2012)]. Foraging trips for each individual were defined as a set of sequential locations at sea between two haul-out events. A location was identified as a haul-out location if it was within ≤1 km distance from the coast and/or located in areas of the Wadden Sea that may become dry at low tide. All other locations (>1 km distance from the coast or in tidal channels) were assumed to be part of foraging trips. A trip was defined to start at the first recorded position >1 km from the coast and to end at the last position >1 km from the coast. However, when interpreting these foraging trips, it has to be considered that they start relatively close to shore and thus could include positions that were related to waiting for haul outs to become available. Trips with less than three positions were excluded from the following analysis. Trip statistics, i.e. trip duration (h), total distance (km), and trip maximum and mean distances to Helgoland and to shore (km) were calculated for each individual foraging trip using the R package “trip” (Sumner et al., 2009; ,Sumner, 2011,, 2016, version 1.5.0). Data interpolation and behavioural classification State-space modelling (Jonsen et al., 2005; Patterson et al., 2008, and see Supplementary 3) was applied to regularize the dataset for the modelling approach and to increase the precision of our dataset, as spatial biases in the original data are considered during the path estimation. Before applying the state-space model, a possible dependency of the number of positions of the tagging duration was examined with different appropriate regression models and no significant effect was found. As the original data set comprised 30% of trips with >1 position/h and 46% with 0.5–1 positions/h (Supplementary 6, Table S3), the data set was regularized to 1-h intervals, with 720 min being the largest data gap that was allowed to be interpolated (for details see Supplementary 6). To test if the temporal resolution of the interpolation of the Argos positions did affect the behavioural analysis and the main results of the study, we conducted several robustness analyses (see Supplementary 6). We identified behavioural states using expectation–maximization binary clustering (EMbC; Garriga et al., 2016), a robust, non-supervised, multivariate clustering algorithm that minimizes prior assumptions and favours the semantic interpretation of the final clustering by splitting the input features in low and high values of speed and turning angle (Garriga et al., 2016). The classification does not depend on temporal or contextual correlations of behavioural states, hence the EMbC approach can be considered as a fine-scale behavioural segmentation method (Garriga et al., 2016). EMbC thus offers a new approach for the classification of behavioural states and has already been applied successfully in several studies (Mendez et al., 2017; Jones et al., 2018). The algorithm assigns each location to one of the four clusters (Supplementary 4, Table S2 and Figures S3 and S4): high velocity/low turn (HL), high velocity/high turn (HH), low velocity/low turn (LL), and low velocity/high turn (LH). HL was identified as fast directional movement and was interpreted as “fast travelling” behaviour, the two states with high turning angles (LH, HH) were merged and interpreted as “foraging” and LL was identified as slow directional movement and was interpreted as “slow travelling/resting” behaviour. EMbC analysis was conducted using the R package EMbC v2.0.1 (Garriga et al., 2019), and a smoother function was applied to take account of temporal associations of behavioural states. More details and a discussion of the method are provided in Supplementary 4. Statistical modelling Preparation of covariables ArcGIS (version 10.3; Environmental Systems Research Institute (ESRI), 2016) was used to calculate the means of the spatial environmental covariates for a grid with a spatial resolution of 5 × 5 km. Environmental variables included: (i) dist_coast = minimal distance to the mainland and large islands (except Helgoland); (ii) dist_Helgoland = minimal distance to the island of Helgoland; (iii) mean_depth = mean water depth; and (iv) slope = inclination of the seabed. Temporal variables drawn directly from the tracking dataset included: (v) year; (vi) week_tag = week after tagging date; (vii) tripid = the number of each foraging trip; and (viii) sealid = the individual seal. Furthermore, (ix) behaviour = identified by the EMbC method was used as a covariate in the modelling process. Point process models The statistical analysis of telemetry data investigating resource selection is often challenging, and various modelling strategies have previously been developed and discussed (Hooten et al., 2017). Popular approaches include (integrated) step selection functions (Thurfjell et al., 2014; Avgar et al., 2016) or point process approaches (Johnson et al., 2013; Renner et al., 2015). In addition to the true tracking locations, both these approaches use a number of contrasting points (“dummy points”, “pseudo-absences” or “available steps”) making the comparison of selected vs. available resources possible. Methods using contrasting points tend to produce better results than techniques using presence points alone (Brotons et al., 2004; Elith et al., 2006; Barbet-Massin et al., 2012). In the following analysis, we used and extended the spatio-temporal PPMs presented by Renner et al. (2015), because this approach naturally and automatically resolves many of the questions and pitfalls arising from the alternative approaches (Warton and Shepherd, 2010; Warton and Aarts, 2013; Renner et al., 2015; Hooten et al., 2017). For example, the role and number of dummy points were not chosen ad hoc but were deduced purely mathematically by the efficient estimation of an integral as a part of the PPM likelihood (Warton and Shepherd, 2010; Warton and Aarts, 2013). In addition, PPMs represent a generalization of many other frequently used methods (Johnson et al., 2008; Warton and Shepherd, 2010; Aarts et al., 2012). Finally, the PPM likelihood can be approximated by a mathematical method using standard generalized linear mixed modelling-regression software (Johnson et al., 2013; Renner et al., 2015), ensuring a flexible and individual implementation. More details of the modifications of the PPMs compared with the spatio-temporal PPMs as presented by Johnson et al. (2013) are shown in Supplementary 3. Model selection To ensure that the modelling approach was based on a robust dataset, and covering a period of profound changes in the pups behaviour (Carter et al., 2017), we only used data for the first 70 d of the tagging period. Continuous data were available for seven individuals during this period, while four individuals stopped transmitting data after 24–50 d (Supplementary 2, Figure S2). We analysed habitat use by grey seal pups and changes in this use throughout the tracking period using a generalized additive mixed model PPM (GAMM-PPM) on a dataset consisting of 9751 raw data points. Seal HG9 was removed because its dataset contained too few data points. When applying the GAMM-PPM to the tracking raw data, 78 008 dummy points were created. The optimal model regarding the set of fixed-effect predictors was selected by comparing 52 different models, based on the Akaike information criterion (AIC; Akaike, 1973). Models including (possibly nonlinear) interaction terms of the variables dist_coast, mean_depth, and slope in interaction with the week after tagging date were not selected through AIC analysis. Inspection of the results of the best models showed that all models had similar patterns of data, indicating that our main results were robust across different models. The best model included a second-order interaction term. Notably, the inclusion of interactions in a model means that lower order terms cannot be interpreted (Field et al., 2012); however, they are included to improve the model. The model enabled us to analyse if the pups showed increases or decreases in different behavioural categories in relation to (i) week after tagging (i.e. age of the pups) and (ii) distance to Helgoland, and if the distance to Helgoland generally changed with the week after tagging [see (1); interpretable interaction terms shown in bold]: Z ∼ ß+telogds,angle,logdt,k=c5,5,5+stripid,bs=re+ssealid,bs=re+syear,bs=re+behaviour+sweektag+sdistHelgoland+tiweektag,by=behaviour+tiweektag,distHelgoland+tidistHelgoland,by=behaviour.(1) Here, ß is the intercept, te() depicts a tensor product regression spline including the main effect, and ti() depicts a partial tensor product regression spline not interfering with (probably smooth) main effects. The Poisson distribution was chosen as probability distribution. Tensor–smooth interactions were favoured during AIC selection compared with models with cubic spline interactions applied to rescaled variables. Tripid, sealid, and year were included as random effects, indicated by the term bs = re. The term by = behaviour indicates that the splines were estimated independently for each behavioural category. Notably, a classical Poisson-GAMM was not fitted, but the response variable Z interplayed with appropriate regression weights such that a PPM likelihood was approximated (for more details, see Johnson et al., 2013). Standard errors and p-values were estimated conservatively. Model validation, numerical realization, and software PPM model-validation plots for the final GAMM-PPM were generated based on PPM-Pearson residuals (Baddeley and Turner, 2005; Baddeley et al., 2005). All statistical analyses were performed using the free statistical software R (R Core Team, 2019). Fitting of the state-space models was straightforward using the R package bsam (Jonsen et al., 2005; ,Jonsen, 2016). Spatial statistics were performed using spatstat (Baddeley and Turner, 2005), dummy-point meshes and trapezoid rule-based quadrature weights were created using mvQuad (Weiser, 2016), GAMM and generalized additive model (GAM) fits were performed using mgcv (Wood, 2006), and visualization was performed based on itsadug (van Rij et al., 2017). All the codes were programmed such that the main parts of the code could be run using parallel computing, using the parallel package and the bam() function from the mgcv package. Results Data for five female and six male grey seal pups were recorded from mid-January/beginning of February for periods of 24–267 d between 2015 and 2017 (Supplementary 2, Table S1). Individual grey seal pups from Helgoland moved widely throughout the southern North Sea, as visualized by their individual tracks (Figure 1). The German part of the North Sea was used intensely in all years. Two individuals moved along the Danish coast up to the Skagerrak/Kattegat area and one moved to Dutch and UK waters in 2015 and 2017, with individual HG12 settling at the UK coast after 9 weeks (Figure 1). In contrast, individuals stayed closer to the colony in 2016 compared with the other 2 years (Figure 1). Some individuals mainly stayed at the Helgoland colony (HG4 in 2016), but others frequently changed their haul-out locations (HG1, 2, 3, 8, and 11 in 2015 and 2017) or settled in areas along the coast, at least during the tracking period (HG6 and 7 in 2016). During their first trips, grey seal pups predominantly stayed in waters around the island (Supplementary 5, Figure S5). However, from the second week on, several individuals appeared to follow similar patterns dispersing far along the mainland coasts of Germany, Denmark and the Netherlands, and even to offshore and coastal areas of the UK. A total of 293 foraging trips were recorded (Table 1), with between 1 (HG9 and 11) and 70 trips per individual. The mean trip duration varied from 2.25 (HG4, SD ±2.74 d) to 30.22 d (HG11) and covered a mean distance of 89–1634 km. The high values were mainly recorded from three individuals (HG3, 9, and 11) for which we only recorded one to four foraging trips. The mean trip duration for all individuals was 8.84 d (SD ±6.26 d), and the mean total distance covered per trip was 386.3 km (Table 1). The mean maximum distance to the shore ranged from 22 (HG6) to 110 km (HG9). Bathymetric depth and distance from the coast were excluded from our model through the AIC (see below), but the data showed that the pups stayed in areas with a water depth of 0–40 m during the first weeks (Supplementary 5, Figure S6a). However, from week 4 onwards, some of the pups also explored areas with depths of >40 m (Supplementary 5, Figure S6b) and the maximum bathymetric depth of the explored areas in the following weeks was 511 m. While exploring their habitat, pups stayed in areas within 200 km of the coast. The proportions of 60-min intervals (Table 2) classified for each of the three behavioural categories varied among individuals: between 29 and 54% of locations were classified as slow travelling/resting (mean of all individuals: 42%), between 42 and 67% as foraging (mean: 52%), and between 4 and 10% as fast travelling (mean: 6%). As our robustness analysis revealed a dependency of absolute proportions of slow travelling/resting and fast travelling behaviour of the interpolation step lengths, the proportions of these behaviours have to be interpreted with caution (see Discussion and Supplementary 6 for more details). Table 1. Numbers and characteristics of trips per individual seal Year . Individual . Number of trips . Duration mean (d) . SD duration mean (d) . Total distance (km) . Max. distance to Helgoland (km) . Mean distance to Helgoland (km) . Max. distance to shore (km) . Mean distance to shore (km) . 2015 HG1 9 4.39 5.38 386.34 157.84 123.55 50.57 36.73 HG2 32 3.04 7.64 174.98 76.12 57.54 29.27 16.92 HG3 4 10.60 6.53 604.32 223.64 171.27 73.48 48.03 2016 HG4 70 2.25 2.74 89.22 22.43 15.14 46.54 39.96 HG5 32 5.40 6.74 193.39 77.43 48.37 62.02 37.84 HG6 52 2.76 5.10 152.22 114.07 101.95 22.15 14.08 HG7 33 4.48 4.97 205.63 93.64 72.62 49.64 29.01 2017 HG8 10 6.65 8.17 317.94 184.65 157.83 32.12 17.23 HG9 1 23.74 - 274.63 176.97 61.38 110.44 40.59 HG11 1 30.22 - 1 634.84 331.59 230.89 93.26 39.25 HG12 49 3.65 8.13 215.85 444.27 411.68 38.99 25.73 All seals – 293 8.84 6.26 386.30 172.97 132.02 55.32 31.40 Year . Individual . Number of trips . Duration mean (d) . SD duration mean (d) . Total distance (km) . Max. distance to Helgoland (km) . Mean distance to Helgoland (km) . Max. distance to shore (km) . Mean distance to shore (km) . 2015 HG1 9 4.39 5.38 386.34 157.84 123.55 50.57 36.73 HG2 32 3.04 7.64 174.98 76.12 57.54 29.27 16.92 HG3 4 10.60 6.53 604.32 223.64 171.27 73.48 48.03 2016 HG4 70 2.25 2.74 89.22 22.43 15.14 46.54 39.96 HG5 32 5.40 6.74 193.39 77.43 48.37 62.02 37.84 HG6 52 2.76 5.10 152.22 114.07 101.95 22.15 14.08 HG7 33 4.48 4.97 205.63 93.64 72.62 49.64 29.01 2017 HG8 10 6.65 8.17 317.94 184.65 157.83 32.12 17.23 HG9 1 23.74 - 274.63 176.97 61.38 110.44 40.59 HG11 1 30.22 - 1 634.84 331.59 230.89 93.26 39.25 HG12 49 3.65 8.13 215.85 444.27 411.68 38.99 25.73 All seals – 293 8.84 6.26 386.30 172.97 132.02 55.32 31.40 Mean values of trip duration (d), total distance travelled per trip (km), maximum and mean distances to Helgoland (km), and maximum and mean distances to the coast (km). Open in new tab Table 1. Numbers and characteristics of trips per individual seal Year . Individual . Number of trips . Duration mean (d) . SD duration mean (d) . Total distance (km) . Max. distance to Helgoland (km) . Mean distance to Helgoland (km) . Max. distance to shore (km) . Mean distance to shore (km) . 2015 HG1 9 4.39 5.38 386.34 157.84 123.55 50.57 36.73 HG2 32 3.04 7.64 174.98 76.12 57.54 29.27 16.92 HG3 4 10.60 6.53 604.32 223.64 171.27 73.48 48.03 2016 HG4 70 2.25 2.74 89.22 22.43 15.14 46.54 39.96 HG5 32 5.40 6.74 193.39 77.43 48.37 62.02 37.84 HG6 52 2.76 5.10 152.22 114.07 101.95 22.15 14.08 HG7 33 4.48 4.97 205.63 93.64 72.62 49.64 29.01 2017 HG8 10 6.65 8.17 317.94 184.65 157.83 32.12 17.23 HG9 1 23.74 - 274.63 176.97 61.38 110.44 40.59 HG11 1 30.22 - 1 634.84 331.59 230.89 93.26 39.25 HG12 49 3.65 8.13 215.85 444.27 411.68 38.99 25.73 All seals – 293 8.84 6.26 386.30 172.97 132.02 55.32 31.40 Year . Individual . Number of trips . Duration mean (d) . SD duration mean (d) . Total distance (km) . Max. distance to Helgoland (km) . Mean distance to Helgoland (km) . Max. distance to shore (km) . Mean distance to shore (km) . 2015 HG1 9 4.39 5.38 386.34 157.84 123.55 50.57 36.73 HG2 32 3.04 7.64 174.98 76.12 57.54 29.27 16.92 HG3 4 10.60 6.53 604.32 223.64 171.27 73.48 48.03 2016 HG4 70 2.25 2.74 89.22 22.43 15.14 46.54 39.96 HG5 32 5.40 6.74 193.39 77.43 48.37 62.02 37.84 HG6 52 2.76 5.10 152.22 114.07 101.95 22.15 14.08 HG7 33 4.48 4.97 205.63 93.64 72.62 49.64 29.01 2017 HG8 10 6.65 8.17 317.94 184.65 157.83 32.12 17.23 HG9 1 23.74 - 274.63 176.97 61.38 110.44 40.59 HG11 1 30.22 - 1 634.84 331.59 230.89 93.26 39.25 HG12 49 3.65 8.13 215.85 444.27 411.68 38.99 25.73 All seals – 293 8.84 6.26 386.30 172.97 132.02 55.32 31.40 Mean values of trip duration (d), total distance travelled per trip (km), maximum and mean distances to Helgoland (km), and maximum and mean distances to the coast (km). Open in new tab Table 2. Proportion of 60-min intervals classified for each behaviour Individual . % slow travel/rest . % foraging . % fast travel . HG1 29 67 4 HG2 39 55 6 HG3 49 45 6 HG4 38 57 5 HG5 42 48 10 HG6 35 59 6 HG7 46 49 5 HG8 46 49 5 HG11 54 42 4 HG12 40 50 10 Mean all (±SD) 42 (±7) 52 (±7) 6 (±2) Individual . % slow travel/rest . % foraging . % fast travel . HG1 29 67 4 HG2 39 55 6 HG3 49 45 6 HG4 38 57 5 HG5 42 48 10 HG6 35 59 6 HG7 46 49 5 HG8 46 49 5 HG11 54 42 4 HG12 40 50 10 Mean all (±SD) 42 (±7) 52 (±7) 6 (±2) Open in new tab Table 2. Proportion of 60-min intervals classified for each behaviour Individual . % slow travel/rest . % foraging . % fast travel . HG1 29 67 4 HG2 39 55 6 HG3 49 45 6 HG4 38 57 5 HG5 42 48 10 HG6 35 59 6 HG7 46 49 5 HG8 46 49 5 HG11 54 42 4 HG12 40 50 10 Mean all (±SD) 42 (±7) 52 (±7) 6 (±2) Individual . % slow travel/rest . % foraging . % fast travel . HG1 29 67 4 HG2 39 55 6 HG3 49 45 6 HG4 38 57 5 HG5 42 48 10 HG6 35 59 6 HG7 46 49 5 HG8 46 49 5 HG11 54 42 4 HG12 40 50 10 Mean all (±SD) 42 (±7) 52 (±7) 6 (±2) Open in new tab Model results The model revealed that the distance of the pups to Helgoland increased throughout the weeks after tagging [Figure 2 and Table 3, interaction term “ti(week_tag, dist_Helgoland)”, p < 2e−16]. The model revealed a significant change in the frequency of foraging behaviour in relation to week after tagging [Table 3; interaction term “ti(week_tag):foraging”, p < 0.001]. The frequency of foraging increased approximately until week 7 and decreased thereafter (Figure 3a). The interaction between the week after tagging and fast travelling behaviour revealed an increasing frequency for this behaviour with time [Table 3; interaction term “ti(week_tag):fast travelling”, p = 0.001] (Figure 3a). The model also revealed that slow travelling/resting [Table 3, interaction term “ti(dist_helgoland):slow travel/rest”, p = 0.013] and foraging behaviour [Table 3, interaction term “ti(dist_helgoland):foraging”, p = 0.010] increased significantly with increasing distance to Helgoland (Figure 3b). Figure 2. Open in new tabDownload slide Change in distance to Helgoland with week after tagging. Plot shows the partial (= isolated) effect of the interaction ti(week_tag, dist_Helgoland). The colour scale indicates relative differences in the smooth (blue = low preference, orange = high preference). Figure 2. Open in new tabDownload slide Change in distance to Helgoland with week after tagging. Plot shows the partial (= isolated) effect of the interaction ti(week_tag, dist_Helgoland). The colour scale indicates relative differences in the smooth (blue = low preference, orange = high preference). Figure 3. Open in new tabDownload slide Decrease/increase in behaviour with (a) week after tagging and (b) distance to Helgoland in kilometre. The plot shows the partial (= isolated) effect of each interaction separately. Figure 3. Open in new tabDownload slide Decrease/increase in behaviour with (a) week after tagging and (b) distance to Helgoland in kilometre. The plot shows the partial (= isolated) effect of each interaction separately. Table 3. Results for the best model Parametric coefficient . Estimate . Standard error . z-Value . Pr(>|z|) . Smooth terms . edfa . Ref.df . Chi square . p-Valueb . Intercept −6.647 0.093 −71.857 <2e−16*** Foraging −0.099 0.062 −1.615 0.106 Fast travel −0.160 0.123 −1.299 0.194 te(log_ds, angle, log_dt) 54.043 61.404 23 490.035 <2e−16*** s(trip_id) 66.525 81.000 2 817.805 <2e−16*** s(seal_id) 2.852 9.000 237.610 0.088 s(year) 0.486 2.000 29.768 0.237 s(week_tag) 3.416 4.185 49.013 0.000*** s(dist_Helgoland) 2.454 2.921 106.316 <2e−16*** ti(week_tag):slow travel/rest 1.807 2.042 2.641 0.241 ti(week_tag):foraging 2.689 2.902 63.065 0.000*** ti(week_tag):fast travel 3.453 3.768 16.807 0.001*** ti(week_tag, dist_Helgoland) 6.631 8.087 308.486 <2e−16*** ti(dist_Helgoland):slow travel/rest 2.161 2.405 8.239 0.013* ti(dist_Helgoland):foraging 2.044 2.275 8.550 0.010** ti(dist_Helgoland):fast travel 1.528 1.834 2.233 0.209 Parametric coefficient . Estimate . Standard error . z-Value . Pr(>|z|) . Smooth terms . edfa . Ref.df . Chi square . p-Valueb . Intercept −6.647 0.093 −71.857 <2e−16*** Foraging −0.099 0.062 −1.615 0.106 Fast travel −0.160 0.123 −1.299 0.194 te(log_ds, angle, log_dt) 54.043 61.404 23 490.035 <2e−16*** s(trip_id) 66.525 81.000 2 817.805 <2e−16*** s(seal_id) 2.852 9.000 237.610 0.088 s(year) 0.486 2.000 29.768 0.237 s(week_tag) 3.416 4.185 49.013 0.000*** s(dist_Helgoland) 2.454 2.921 106.316 <2e−16*** ti(week_tag):slow travel/rest 1.807 2.042 2.641 0.241 ti(week_tag):foraging 2.689 2.902 63.065 0.000*** ti(week_tag):fast travel 3.453 3.768 16.807 0.001*** ti(week_tag, dist_Helgoland) 6.631 8.087 308.486 <2e−16*** ti(dist_Helgoland):slow travel/rest 2.161 2.405 8.239 0.013* ti(dist_Helgoland):foraging 2.044 2.275 8.550 0.010** ti(dist_Helgoland):fast travel 1.528 1.834 2.233 0.209 Parametric coefficients and smooth terms are shown. Interaction terms that can be interpreted are indicated in bold. aedf = estimated degrees of freedom. bsignificance codes: 0 ‘***’, 0.001 ‘**’, 0.01 ‘*’, 0.05 ‘ . ’, 0.1 ‘ ’ Open in new tab Table 3. Results for the best model Parametric coefficient . Estimate . Standard error . z-Value . Pr(>|z|) . Smooth terms . edfa . Ref.df . Chi square . p-Valueb . Intercept −6.647 0.093 −71.857 <2e−16*** Foraging −0.099 0.062 −1.615 0.106 Fast travel −0.160 0.123 −1.299 0.194 te(log_ds, angle, log_dt) 54.043 61.404 23 490.035 <2e−16*** s(trip_id) 66.525 81.000 2 817.805 <2e−16*** s(seal_id) 2.852 9.000 237.610 0.088 s(year) 0.486 2.000 29.768 0.237 s(week_tag) 3.416 4.185 49.013 0.000*** s(dist_Helgoland) 2.454 2.921 106.316 <2e−16*** ti(week_tag):slow travel/rest 1.807 2.042 2.641 0.241 ti(week_tag):foraging 2.689 2.902 63.065 0.000*** ti(week_tag):fast travel 3.453 3.768 16.807 0.001*** ti(week_tag, dist_Helgoland) 6.631 8.087 308.486 <2e−16*** ti(dist_Helgoland):slow travel/rest 2.161 2.405 8.239 0.013* ti(dist_Helgoland):foraging 2.044 2.275 8.550 0.010** ti(dist_Helgoland):fast travel 1.528 1.834 2.233 0.209 Parametric coefficient . Estimate . Standard error . z-Value . Pr(>|z|) . Smooth terms . edfa . Ref.df . Chi square . p-Valueb . Intercept −6.647 0.093 −71.857 <2e−16*** Foraging −0.099 0.062 −1.615 0.106 Fast travel −0.160 0.123 −1.299 0.194 te(log_ds, angle, log_dt) 54.043 61.404 23 490.035 <2e−16*** s(trip_id) 66.525 81.000 2 817.805 <2e−16*** s(seal_id) 2.852 9.000 237.610 0.088 s(year) 0.486 2.000 29.768 0.237 s(week_tag) 3.416 4.185 49.013 0.000*** s(dist_Helgoland) 2.454 2.921 106.316 <2e−16*** ti(week_tag):slow travel/rest 1.807 2.042 2.641 0.241 ti(week_tag):foraging 2.689 2.902 63.065 0.000*** ti(week_tag):fast travel 3.453 3.768 16.807 0.001*** ti(week_tag, dist_Helgoland) 6.631 8.087 308.486 <2e−16*** ti(dist_Helgoland):slow travel/rest 2.161 2.405 8.239 0.013* ti(dist_Helgoland):foraging 2.044 2.275 8.550 0.010** ti(dist_Helgoland):fast travel 1.528 1.834 2.233 0.209 Parametric coefficients and smooth terms are shown. Interaction terms that can be interpreted are indicated in bold. aedf = estimated degrees of freedom. bsignificance codes: 0 ‘***’, 0.001 ‘**’, 0.01 ‘*’, 0.05 ‘ . ’, 0.1 ‘ ’ Open in new tab Discussion The results of this study provide the first detailed characterization of the dispersal and behaviour of recently weaned grey seal pups and their spatial utilization of the southern North Sea. Our modelling approach revealed how the behaviour of seal pups adapted during their first weeks at sea. This study revealed that grey seal pups from Helgoland dispersed long distances in their first weeks after weaning and moved widely throughout the southern North Sea. During their trips some of the pups covered distances from land, which are comparable to the movements of adult and subadult grey seals tagged in the Netherlands (Brasseur et al., 2017a), which intensely used offshore areas up to the Dogger Bank in the north, as far as Helgoland in the east and very regularly Dutch, UK, and Belgian waters. The German part of the North Sea was visited occasionally and, if so, predominantly between Helgoland and the East Frisian coast (Brasseur et al., 2017a). Several pups tagged in the present study, however, predominantly used areas which were not used by the individuals tagged in the Netherlands. The mean trip duration and distance for most individuals were comparable to that for pups in the north-western North Sea (Carter et al., 2017). The trip distances and durations demonstrated by young of the year (YOY) in the north-West Atlantic (Breed et al., 2011) and by recently weaned grey seals in the north-western North Sea (Carter et al., 2017) were longer than those recorded for adults of the respective study areas. This has been discussed in the context of foraging inexperience, intraspecific and interspecific competition, and exploration of their marine environment (Breed et al., 2011; Carter et al., 2017). For eight pups tagged on Helgoland, the amount of the time spent foraging was slightly higher than the time spent in slow travelling/resting, while fast travelling only occurred for a small amount of time. However, it has to be considered that especially the proportion of slow travelling/resting behaviour might be artificially high due to the interpolation scheme applied (see Supplementary 6 for a more detailed discussion). It is possible that the individuals travelled slowly or rested during these periods, but it is also possible that they changed direction and moved slowly while foraging. Thus, this behavioural class might include foraging behaviour that was not detected by the analysis as interpolation of data gaps of several hours to regular 1-h intervals might not depict the reality completely. In addition, short-term foraging events and shorter-term fast travelling events potentially are not represented appropriately when using a 1-h interpolation interval and potentially were more frequent than detected in our analysis. As a study by Russell et al. (2015) suggests, resting behaviour with very low speeds and high turning angles often occurs for long periods of time (multiple hours) on the surface and is presumably associated with food digestion (Russell et al., 2015). In our study, however, such intervals were assigned to foraging (LH but not HH). As 25% of the grey seals studied by Russell et al. (2015) spent over 5% of their time resting in offshore areas, presumably a comparable part of the intervals in our study was resting/food digestion at the surface rather than active foraging behaviour. Thus, the actual proportion of time spent foraging presumably was lower than detected in our analysis. However, due to the presumed association between resting and food digestion, this most probably does not strongly influence the results of the present study. However, for future studies, especially if particularly investigating periods of active foraging, it would thus be valuable to examine if interpreting part of the EMbC class LH (low velocity/high turning angles) as resting/food digestion rather than foraging is applicable. Although caution is needed when comparing different areas with presumably different conditions, our findings resembled the findings for grey seal YOY in Canadian waters, where the individuals also spent more time foraging than travelling (59 and 32%, respectively, Breed et al., 2011). Compared with adults and sub-adults of that region, the foraging ratio of the YOY was much lower, which was set in the context of presumably less-efficient foraging trips. However, the YOY described by Breed et al. (2011) were already 5 months old when tagged, while the individuals tagged in the current study were much younger and thus less experienced. Grey seal pups from Helgoland showed very different individual movement patterns, with some remaining more local than others. Most of the pups departed from the area around Helgoland during the first weeks of the tagging period and several pups dispersed far along the mainland coasts and towards the United Kingdom. This indicates intense exploration of other haul-out and foraging areas, possibly leading to later settlement in these areas (Brasseur et al., 2015). Dispersal and behaviour Based on the modelling approach, we were able to analyse different aspects of the grey seal pups’ habitat use and behaviour and to reveal changes occurring throughout the tagging period as the pups grew older. The increasing frequency of foraging behaviour until week 7 indicated that the pups gradually developed and applied this behaviour. The decrease in foraging behaviour after week 7 possibly indicated that individuals became more experienced and efficient in foraging (Bennett et al., 2010) and, thus, reduced their foraging effort. Our model also revealed that the pups increased their fast travelling behaviour with increasing age, again underlining their growing explorative behaviour with increasing age, but probably also indicating their ability to swim further and faster as they grow. This reflects the general transition from naive to more experienced pups, which gradually increase their foraging effort and range to account for their increasing energy demands for survival and growth (Bennett et al., 2007) but which also gradually increase their foraging efficiency (Bennett et al., 2010) and thus decrease their foraging effort after being several weeks at sea. Graphical analyses, generalized linear model (GLM) analyses and applying a multivariate analysis of variance (MANOVA, see Supplementary 6) showed that the behavioural changes with age of the pups were not affected by the interpolation scheme applied. Thus, interpolation does not affect our findings and the detected changes are robustly represented in the dataset. While exploring their habitat, the pups stayed within 200 km of the coast and successively used areas with deeper water. The diving ability of grey seal pups (i.e. dive depth and duration) has been shown to increase during the first months at sea (Noren et al., 2005; Bennett et al., 2010; Carter et al., 2017), and it can be expected that they might explore unknown and deeper areas as they grow older. Carter et al. (2017) showed that recently weaned grey seal pups born in colonies along the north-eastern coast of the UK stayed in areas of 55–80 m water depth during the first 40 d at sea, and that during this time, their dive depth increased from ∼15 m at the start to 50 m at day 40. The increase in distance to Helgoland could additionally be related to a change in habitat preference over time. In winter, prey density is higher in marine areas than in coastal waters, given that some fish species migrate towards deeper (warmer) waters (Woodhead, 1964; de Veen, 1978). This is also assumed to be potential reason for the increased foraging activity of harbour seals (Aarts et al., 2016) and grey seal YOY (Breed et al., 2011) in offshore waters during winter, as also shown for adult grey seals in the north-west Atlantic (Harvey et al., 2012). Furthermore, the preferences of juvenile and adult grey seals for shallow waters decreased during winter in the north-west Atlantic (Harvey et al., 2008). It is therefore likely that the pups encountered high prey densities in areas further offshore, which they were able to explore as soon as their diving ability increased. Our findings underline that the pups used areas more distant from their colony of birth as they grew older. They explored their habitat, probably searched for profitable foraging areas and dispersed throughout the southern North Sea. They would be expected to become more experienced and successful foragers after the first weeks (Bennett et al., 2010; Carter et al., 2017) and to thus increasingly exploit the foraging areas that they discovered in the meantime. Depending on their learning curve in the first weeks, but by 36 d after weaning at the latest, grey seal pups need to capture prey to replenish their protein reserves (Bennett et al., 2007) to survive and grow. This was reflected by the increasing foraging effort during the first few weeks, followed by a decrease as their foraging tactics presumably became more efficient. Competition Interspecific and intraspecific competition could additionally influence the pups’ preference for different habitats during different stages of their development. Adult grey seals in the north-west Atlantic used more coastal areas than juveniles (Harvey et al., 2008; Breed et al., 2011), while YOY were displaced from foraging habitats near the colony (Breed et al., 2013). Weaned grey seal pups from the north-western North Sea, which includes many grey and harbour seal haul outs (Jones et al., 2015), showed wide dispersal (Carter et al., 2017). Strong intraspecific and interspecific competition can affect pup behaviour in such areas (Breed et al., 2013). Recent studies have shown that many of the grey seal populations of the United Kingdom, especially in the north-western and northern areas appear to have reached carrying capacity through density-dependent pup survival (Russell et al., 2019; Thomas et al., 2019). However, in the central and especially the southern North Sea areas of the United Kingdom, pup production still strongly increases (Russell et al., 2019). As the exponential rate of increase in the southern North Sea region must in part be driven by females born elsewhere, this possibly indicates that pups from northern regions travel south in search of favourable foraging conditions and stay to pup (Russell et al., 2019). Both the German and the whole Wadden Sea grey seal populations are small compared with the UK population (Lonergan et al., 2011; Brasseur et al., 2018a) and pup as well as moult count numbers still increase (Brasseur et al., 2018b; Cremer et al., 2019), suggesting that intraspecific competition is likely to be lower than in UK waters. Our current findings showed that pups born in an area with presumably less intraspecific competition also dispersed far from their breeding grounds and repeatedly explored areas with lower seal densities (i.e. the Danish North Sea coast, Brasseur et al., 2017b, 2018a; Cremer et al., 2019). Nevertheless, the German harbour seal population is relatively large (16 145 individuals, 26 873 individuals in the entire Wadden Sea area; Galatius et al., 2017) and the entire Wadden Sea population recently potentially approaches carrying capacity (Brasseur et al., 2018b). Thus, the behaviour of grey seal pups in the south-eastern North Sea may also be influenced by interspecific competition. Conclusions This study investigated the behaviour of grey seals during the potentially most vulnerable phase of their life cycle, when naive individuals need to develop successful foraging strategies to cover their high energy demands. Knowledge of habitat preferences, behaviour, and their changes in relation to age is crucial to our understanding of the important processes driving population dynamics and to predict the populations’ ability to cope with environmental changes and anthropogenic stressors. Our results showed that grey seal pups use large areas of the southern North Sea, and they explore different North Sea regions and colonies closely, leading to possible recruitment into breeding populations other than their colony of birth. The increase in foraging frequency with time suggests that the pups progressively use their environment for foraging during the first weeks at sea, and its decrease reflects the transition from naive to more experienced foragers. They thus encounter various environmental conditions and potentially face a multitude of anthropogenic activities, even during their first weeks and months at sea. Our findings underline the complexity of early development in grey seal pups and highlight the need to consider their habitat requirements during the marine spatial planning process and to include them when investigating anthropogenic impacts. This study contributes to our knowledge of the spatial utilization of the German North Sea by an increasingly important top predator and thus provides important information for a broad audience, including researchers and those involved in management and decision-making in politics and industry. Acknowledgements We especially thank the team of the Seal Centre Friedrichskoog, Michael Janßen, Katharina Tilly, Rebecca Ballstaedt, and the team of “Verein Jordsand e.V.”, the Alfred-Wegener-Institute Helgoland, the Institute of Avian Research Helgoland, and the team of the Helgoland air field for supporting our work in the field. We also would like to thank for the very helpful comments of the anonymous reviewers of a former version of our manuscript. Funding Parts of the analysis were conducted in the framework of the MONTRACK project funded by the Federal Agency for Nature Conservation (project no.: Z 1.2-532 02/AWZ/2017/2). References Aarts G. , Cremer J., Kirkwood R., van der Wal J. T., Matthiopoulos J., Brasseur S. 2016 . Spatial distribution and habitat preference of harbour seals (Phoca vitulina) in the Dutch North Sea. Wageningen University & Research Centre, Wageningen Marine Research . Wageningen Marine Research Report Number C118/16. 43 pp. Aarts G. , Fieberg J., Matthiopoulos J. 2012 . Comparative interpretation of count, presence-absence and point methods for species distribution models . Methods in Ecology and Evolution , 3 : 177 – 187 . Google Scholar Crossref Search ADS WorldCat Akaike H. 1973 . Information theory and an extension of the maximum likelihood principle. In Proceeding of the Second International Symposium on Information Theory , pp. 267 – 281 . Ed. by Petrov B. N., Caski F.. Akademiai Kiado , Budapest . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Avgar T. , Potts J. R., Lewis M. A., Boyce M. S. 2016 . Integrated step selection analysis: bridging the gap between resource selection and animal movement . Methods in Ecology and Evolution , 7 : 619 – 630 . Google Scholar Crossref Search ADS WorldCat Baddeley A. , Turner R. 2005 . Spatstat: an R package for analyzing spatial point patterns . Journal of Statistical Software , 12 : 1 – 42 . Google Scholar Crossref Search ADS WorldCat Baddeley A. , Turner R., Møller J., Hazelton M. 2005 . Residual analysis for spatial point processes (with discussion) . Journal of the Royal Statistical Society: Series B (Statistical Methodology) , 67 : 617 – 666 . Google Scholar Crossref Search ADS WorldCat Barbet-Massin B. , Jiguet F., Albert C. H., Thuiller W. 2012 . Selecting pseudo-absences for species distribution models: how, where and how many? Methods in Ecology and Evolution , 3 : 327 – 338 . Google Scholar Crossref Search ADS WorldCat Beauplet G. , Barbraud C., Chambellant M., Guinet C. 2005 . Interannual variation in the postweaning and juvenile survival of subantarctic fur seals: influence of pup sex, growth rate and oceanographic conditions . Journal of Animal Ecology , 74 : 1160 – 1172 . Google Scholar Crossref Search ADS WorldCat Bennett K. A. , Speakman J. R., Moss S. E. W., Pomeroy P., Fedak M. A. 2007 . Effects of mass and body composition on fasting fuel utilisation in grey seal pups (Halichoerus grypus Fabricius): an experimental study using supplementary feeding . The Journal of Experimental Biology , 210 : 3043 – 3053 . Google Scholar Crossref Search ADS PubMed WorldCat Bennett K. A. , McConnell B. J., Moss S. E. W., Speakman J. R., Pomeroy P. P., Fedak M. A. 2010 . Effects of age and body mass on development of diving capabilities of gray seal pups: costs and benefits of the postweaning fast . Physiological and Biochemical Zoology , 83 : 911 – 923 . Google Scholar Crossref Search ADS PubMed WorldCat Bowen W. D. , den Heyer C. E., McMillan J. I., Iverson S. J. 2015 . Offspring size at weaning affects survival to recruitment and reproductive performance of primiparous gray seals . Ecology and Evolution , 5 : 1412 – 1424 . Google Scholar Crossref Search ADS PubMed WorldCat Brasseur S. , Cremer J., Czeck R., Galatius A., Jeß A., Körber P., Pund R., et al. 2018a . TSEG Grey Seal Surveys in the Wadden Sea and Helgoland in 2017-2018. Common Wadden Sea Secretariat, Wilhelmshaven, Germany. https://www.waddensea-worldheritage.org/sites/default/files/18-07-02_Greysealreport2018.pdf (last accessed 18 March 2020). Brasseur S. M. J. M. , Aarts G., Reijnders P. J. H., Kirkwood R. 2017a . Grey seal tracking reveals different behaviours of resident and transient population components. In Seals in Motion—How Movements Drive Population Development of Harbour Seals and Grey Seals in the North Sea . Ed. by Brasseur S. M. J. M.. PhD thesis, Wageningen University , Wageningen, the Netherlands . 176 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Brasseur S. M. J. M. , Czeck R., Galatius A., Jensen L. F., Jeß A., Körber P., Pund R. 2016 . TSEG Grey Seal Surveys in the Wadden Sea and Helgoland in 2015-2016. First Year of Almost Complete Monitoring by Aerial Surveys. https://www.waddensea-worldheritage.org/sites/default/files/2016_grey%20seal%20report_2015-2016.pdf (last accessed 18 March 2020). Brasseur S. M. J. M. , Czeck R., Galatius A., Jensen L. F., Jeß A., Körber P., Pund R., et al. 2017b . TSEG Grey Seal Surveys in the Wadden Sea and Helgoland in 2016-2017. General Growth but Local Drop in Numbers. Common Wadden Sea Secretariat, Wilhelmshaven, Germany. https://www.waddensea-worldheritage.org/sites/default/files/2017_greysealreport%202016-2017.pdf (last accessed 18 March 2020). Brasseur S. M. J. M. , Reijnders P. J. H., Cremer J., Meesters E., Kirkwood R., Jensen L. F., Jeβ A., et al. 2018b . Echoes from the past: regional variations in recovery within a harbour seal population . PLoS One , 13 : e0189674 . Google Scholar Crossref Search ADS WorldCat Brasseur S. M. J. M. , Van Polanen Petel T. D., Gerrodette T., Meesters E. H. W. G., Reijnders P. J. H., Aarts G. 2015 . Rapid recovery of Dutch gray seal colonies fueled by immigration . Marine Mammal Science , 31 : 405 – 426 . Google Scholar Crossref Search ADS WorldCat Breed G. A. , Bowen W. D., Leonard M. L. 2011 . Development of foraging strategies with age in a long-lived marine predator . Marine Ecology Progress Series , 431 : 267 – 279 . Google Scholar Crossref Search ADS WorldCat Breed G. A. , Bowen W. D., Leonard M. L. 2013 . Behavioral signature of intraspecific competition and density dependence in colony-breeding marine predators . Ecology and Evolution , 3 : 3838 – 3854 . Google Scholar Crossref Search ADS PubMed WorldCat Brotons L. , Thuiller W., Araujo M. B., Hirzel A. H. 2004 . Presence-absence versus presence-only modelling methods for predicting bird habitat suitability . Ecography , 27 : 437 – 448 . Google Scholar Crossref Search ADS WorldCat Carter M. I. D. , Russell D. J. F., Embling C. B., Blight C. J., Thompson D., Hosegood P. J., Bennett K. A. 2017 . Intrinsic and extrinsic factors drive ontogeny of early-life at-sea behaviour in a marine top predator . Scientific Reports , 7 : 15505 . Google Scholar Crossref Search ADS PubMed WorldCat Cremer J. , Brasseur S., Czeck R., Galatius A., Jeß A., Körber P., Pund R., Siebert U., Teilmann J., Bie Thøstesen C., Busch J. A. 2019 . EG-Seals Grey Seal Surveys in the Wadden Sea and Helgoland in 2018-2019 . Common Wadden Sea Secretariat , Wilhelmshaven, Germany . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC de Veen J. 1978 . On selective tidal transport in the migration of North Sea Plaice (Pleuronectes platessa) and other flatfish species . Netherlands Journal of Sea Research , 12 : 115 – 147 . Google Scholar Crossref Search ADS WorldCat Douglas D. C. , Weinzierl R., Davidson S., Kays R., Wikelski M., Bohrer G. 2012 . Moderating Argos location errors in animal tracking data . Methods in Ecology and Evolution , 3 : 999 – 1007 . Google Scholar Crossref Search ADS WorldCat Elith J. , Graham C. H., Anderson R. P., Dudik M., Ferrier S., Guisan A., Hijmans R. J., et al. 2006 . Novel methods improve prediction of species distributions from occurrence data . Ecography , 29 : 129 – 151 . Google Scholar Crossref Search ADS WorldCat Environmental Systems Research Institute (ESRI). 2016 . ArcGIS v. 10.3 ESRI . Redlands, CA . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Field A. , Miles J., Field Z. 2012 . Discovering Statistics Using R . Sage Publications Ltd , London . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Galatius A. , Brasseur S., Czeck R., Jeß A., Körber P., Pund R., Siebert U., et al. 2017 . Aerial Surveys of Harbour Seals in the Wadden Sea in 2017. Population Counts Still in Stagnation, but More Pups than Ever. https://www.waddensea-worldheritage.org/sites/default/files/2017_harboursealreport.pdf (last accessed 18 March 2020). Garriga J. , Palmer J. R. B., Oltra A., Bartumeus F. 2016 . Expectation-maximization binary clustering for behavioural annotation . PLoS One , 11 : e0151984 . Google Scholar Crossref Search ADS PubMed WorldCat Garriga J. , Palmer R. B., Oltra A., Bartumeus F. 2019 . EMbC: Expectation-Maximization Binary Clustering. R package version 2.0.3. https://CRAN.R-project.org/package=EMbC (last accessed 18 March 2020). Haller M. A. , Kovacs K. M., Hammill M. O. 1996 . Maternal investment by fast-ice breeding grey seals (Halichoerus grypus) . Canadian Journal of Zoology , 74 : 1531 – 1541 . Google Scholar Crossref Search ADS WorldCat Härkönen T. , Brasseur S., Teilmann J., Vincent C., Dietz R., Abt K., Reijnders P. 2007 . Status of grey seals along mainland Europe from the Southwestern Baltic to France . NAMMCO Scientific Publications , 6 : 57 – 68 . Google Scholar Crossref Search ADS WorldCat Harvey V. , Côté S. D., Hammill M. O. 2008 . The ecology of 3-D space use in a sexually dimorphic mammal . Ecography , 31 : 371 – 380 . Google Scholar Crossref Search ADS WorldCat Harvey V. , Hammill M. O., Swain D. P., Breed G. A., Lydersen C., Kovacs K. M. 2012 . Winter foraging by a top predator, the grey seal Halichoerus grypus, in relation to the distribution of prey . Marine Ecology Progress Series , 462 : 273 – 286 . Google Scholar Crossref Search ADS WorldCat Hooten M. B. , Johnson D. S., McClintock B. T., Morales J. M. 2017 . Animal Movement—Statistical Models for Telemetry Data . CRC Press , Boca Raton, Florida, USA. Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Johnson C. , Thomas D. L., Ver Hoef J. M., Christ A. 2008 . A general framework for the analysis of animal resource selection from telemetry data . Biometrics , 64 : 968 – 976 . Google Scholar Crossref Search ADS PubMed WorldCat Johnson D. S. , Hooten M. B., Kuhn C. E. 2013 . Estimating animal resource selection from telemetry data using point process models . Journal of Animal Ecology , 82 : 1155 – 1164 . Google Scholar Crossref Search ADS PubMed WorldCat Jones E. L. , McConnell B. J., Smout S., Hammond P. S., Duck C. D., Morris C. D., Thompson D., et al. 2015 . Patterns of space use in sympatric marine colonial predators reveal scales of spatial partitioning . Marine Ecology Progress Series , 534 : 235 – 249 . Google Scholar Crossref Search ADS WorldCat Jones T. B. , Patrick S. C., Arnould J. P. Y., Rodriguez-Malagon M. A., Wells M. R., Green J. A. 2018 . Evidence of sociality in the timing and location of foraging in a colonial seabird . Biology Letters , 14 : 20180214 . Google Scholar Crossref Search ADS PubMed WorldCat Jonsen I. 2016 . Joint estimation over multiple individuals improves behavioural state inference from animal movement data . Scientific Reports , 6 : 20625 . Google Scholar Crossref Search ADS PubMed WorldCat Jonsen I. D. , Flemming J. M., Myers R. A. 2005 . Robust state-space modelling of animal movement data . Ecology , 86 : 2874 – 2880 . Google Scholar Crossref Search ADS WorldCat Kovacs K. M. 1987 . Maternal behaviour and early behavioural ontogeny of grey seals (Halichoerus grypus) on the Isle of May, UK . Journal of Zoology , 213 : 697 – 715 . Google Scholar Crossref Search ADS WorldCat Lonergan M. , Duck C. D., Thompson D., Moss S., McConnell B. 2011 . British grey seal (Halichoerus grypus) abundance in 2008: an assessment based on aerial counts and satellite telemetry . ICES Journal of Marine Science , 68 : 2201 – 2209 . Google Scholar Crossref Search ADS WorldCat Mendez L. , Borsa P., Cruz S., de Grissac S., Hennicke J., Lallemand J., Prudor A., et al. 2017 . Geographical variation in the foraging behaviour of the pantropical red-footed booby . Marine Ecology Progress Series , 568 : 217 – 230 . Google Scholar Crossref Search ADS WorldCat Noren S. R. , Iverson S. J., Boness D. J. 2005 . Development of the blood and muscle oxygen stores in gray seals (Halichoerus grypus): implications for juvenile diving capacity and the necessity of a terrestrial postweaning fast . Physiological and Biochemical Zoology , 78 : 482 – 490 . Google Scholar Crossref Search ADS PubMed WorldCat Noren S. R. , Boness D. J., Iverson S. J., McMillan J., Bowen W. D. 2008 . Body condition at weaning affects the duration of the postweaning fast in gray seal pups (Halichoerus grypus) . Physiological and Biochemical Zoology , 81 : 269 – 277 . Google Scholar Crossref Search ADS PubMed WorldCat Patterson T. A. , Thomas L., Wilcox C., Ovaskainen O., Matthiopoulos J. 2008 . State-space models of individual animal movement . Trend in Ecology and Evolution , 23 : 87 – 94 . Google Scholar Crossref Search ADS WorldCat R Core Team. 2019 . R: A language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/ (last accessed 6 December 2019). Renner I. W. , Elith J., Baddeley A., Fithian W., Hastie T., Phillips S. J., Popovic G., et al. 2015 . Point process models for presence-only analysis . Methods on Ecology and Evolution , 6 : 366 – 379 . Google Scholar Crossref Search ADS WorldCat Russell D. J. F. , McClintock B. T., Matthiopoulos J., Thompson P. M., Thompson D., Hammond P. S., Jones E. L., et al. 2015 . Intrinsic and extrinsic drivers of activity budgets in sympatric grey and harbour seals . Oikos , 124 : 1462 – 1472 . Google Scholar Crossref Search ADS WorldCat Russell D. J. F. , Morris C. D., Duck C. D., Thompson D., Hiby L. 2019 . Monitoring long‐term changes in UK grey seal pup production . Aquatic Conservation: Marine and Freshwater Ecosystems , 29 : 24 – 39 . Google Scholar Crossref Search ADS WorldCat Scheibel W. , Weidel H. 1988 . Zum Vorkommen der Kegelrobben (Halichoerus grypus, Fabricius, 1791, Phocidae, Pinnipedia) Schleswig-Holstein . Zoologischer Anzeiger , 220 : 65 – 70 . OpenURL Placeholder Text WorldCat Sumner M. D. , Wotherspoon S. J., Hindell M. A. 2009 . Bayesian estimation of animal movement from archival and satellite tags . PLoS ONE , 4 : 10. OpenURL Placeholder Text WorldCat Sumner M. D. 2011 . _The Tag Location Problem_.Ph.D. thesis. University of Tasmania . https://eprints.utas.edu.au/12273/3/sumner.pdf> (last accessed 18 March 2020). Sumner M. D. 2016 . trip: Tools for the Analysis of Animal Track Data. R package version 1.5.0. https://CRAN.R-project.org/package=trip (last accessed 18 March 2020). Thomas L. , Russell D. J. F., Duck C. D., Morris C. D., lonergan M., Empacher F., Thompson D., et al. 2019 . Modelling the population size and dynamics of the British grey seal . Aquatic Conservation: Marine and Freshwater Ecosystems , 29 : 6 – 23 . Google Scholar Crossref Search ADS WorldCat Thurfjell H. , Ciuti S., Boyce M. S. 2014 . Applications of step-selection functions in ecology and conservation . Movement Ecology , 2 : 4 . Google Scholar Crossref Search ADS PubMed WorldCat van Rij J. , Wieling M., Baayen R., van Rijn H. 2017 . itsadug: Interpreting Time Series and Autocorrelated Data Using GAMMs. R package, version 2.3. https://cran.r-project.org/web/packages/itsadug/index.html (last accessed 18 March 2020). Warton D. I. , Shepherd L. C. 2010 . Poisson point process models solve the “pseudo-absence problem” for presence-only data . The Annals of Applied Statistics , 4 : 1383 – 1402 . Google Scholar Crossref Search ADS WorldCat Warton D. , Aarts G. 2013 . Advancing our thinking in presence-only and used-available analysis . Journal of Animal Ecology , 82 : 1125 – 1134 . Google Scholar Crossref Search ADS PubMed WorldCat Weiser C. 2016 . mvquad: Methods for Multivariate Quadrature (R package version 1.0-6). http://CRAN.R-project.org/package=mvQuad (last accessed 18 March 2020). Wood S. N. 2006 . Generalized Additive Models: An Introduction with R . Chapman and Hall , London . 416 pp. Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Woodhead P. M. J. 1964 . Changes in the behaviour of the sole, Solea vulgaris, during cold winters, and the relation between the winter catch and sea temperatures . Helgoländer Wissenschaftliche Meeresuntersuchungen , 10 : 328 – 342 . Google Scholar Crossref Search ADS WorldCat © International Council for the Exploration of the Sea 2020. All rights reserved. For permissions, please email: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)