Marine Environmental Research 101 (2014) 69e80
Contents lists available at ScienceDirect
Marine Environmental Research journal homepage: www.elsevier.com/locate/marenvrev
Great skua (Stercorarius skua) movements at sea in relation to marine renewable energy developments H.M. Wade a, *, E.A. Masden a, A.C. Jackson a, b, C.B. Thaxter c, N.H.K. Burton c, W. Bouten d, R.W. Furness e, f a
Environmental Research Institute, North Highland College, University of the Highlands and Islands, Thurso, UK Cornwall College Newquay, Wildflower Lane, Trenance Gardens, Newquay, Cornwall, UK British Trust for Ornithology, The Nunnery, Thetford, Norfolk, UK d Computational Geo-Ecology, Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Amsterdam, The Netherlands e College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, UK f MacArthur Green, 95 South Woodside Road, Glasgow, UK b c
a r t i c l e i n f o
a b s t r a c t
Article history: Received 12 June 2014 Received in revised form 14 September 2014 Accepted 15 September 2014 Available online 17 September 2014
Marine renewable energy developments (MREDs) are an increasing feature of the marine environment. Owing to the relatively small number of existing developments and the early stage of their associated environmental monitoring programmes, the effects of MREDs on seabirds are not fully known. Our ability to fully predict potential effects is limited by a lack of knowledge regarding movements of seabirds at sea. We used GPS tracking to improve our understanding of the movements at sea of a protected seabird species breeding in Scotland, the great skua (Stercorarius skua), to better predict how this species may be affected by MREDs. We found that the overlap of great skuas with leased and proposed MREDs was low; particularly with offshore wind sites, which are predicted to present a greater risk to great skuas than wave or tidal-stream developments. Failed breeders overlapped with larger areas of MREDs than breeding birds but the overall overlap with core areas used remained low. Overlap with wave energy development sites was greater than for offshore wind and tidal-stream sites. Comparison of 2011 data with historical data indicates that distances travelled by great skuas have likely increased over recent decades. This suggests that basing marine spatial planning decisions on short-term tracking data could be less informative than longer-term data. © 2014 Elsevier Ltd. All rights reserved.
Keywords: Environmental impact Remote sensing Seabirds GPS Marine spatial planning Wind power Wave power Bio-logging
1. Introduction Marine renewable energy developments (MREDs) are an increasing feature of the marine environment. The rapid development of these industries is being driven by the need to reduce carbon emissions and to increase energy security. In Scotland, The Crown Estate and the Scottish Government are working in partnership to realise the Scottish Government's target of generating 100% of electricity from renewable sources by 2020 (Scottish Government, 2011). An advanced marine plan has been developed for Scotland with sites leased to offshore wind, wave and
* Corresponding author. Environmental Research Institute, North Highland College, University of the Highlands and Islands, Ormlie Road, Thurso, KW14 7EE, UK. E-mail address: [email protected] (H.M. Wade). http://dx.doi.org/10.1016/j.marenvres.2014.09.003 0141-1136/© 2014 Elsevier Ltd. All rights reserved.
tidal-stream energy developers, with additional sites proposed for future development (Scottish Government, 2013a). Owing to the small number and limited extent of existing MREDs, the effects of offshore wind, wave and tidal-stream installations on seabirds are not fully known (Fox et al., 2006; McCluskie et al., 2013; Witt et al., 2012). This is despite several leased and proposed MRED sites in north-east Scotland being located in close proximity to special protection areas (SPAs) designated specifically to safeguard breeding populations of internationally important seabird species (JNCC, 2014a) (Fig. 1). SPA legislation (EC Birds Directive 2009/147/EC) requires that any development does not damage the integrity of protected seabird populations (Scottish Natural Heritage, 2014). Some of the potential negative effects of MREDs on seabirds include collision with devices (Everaert and Stienen, 2007; Langton et al., 2011); displacement from areas where devices are located (Larsen and Guillemette, 2007; McDonald et al., 2012); increased flight costs if devices act
70
H.M. Wade et al. / Marine Environmental Research 101 (2014) 69e80
Fig. 1. Locations of breeding colonies (yellow circles; defined by number of Apparently Occupied Territories (AOT)) and special protection areas (blue polygons) designated for breeding seabird populations in north-east Scotland, and the proximity of leased and proposed marine renewable energy development sites (solid polygons ¼ leased sites; crosshatched ¼ proposed sites; purple ¼ offshore wind sites; green ¼ wave sites; red ¼ tidal-stream sites). Study colonies indicated with black arrows. See Supplementary Table 1 for colony sizes and year of AOT count. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
as barriers to movement (Desholm and Kahlert, 2005; Masden et al., 2009, 2010); alteration of foraging habitats if devices alter currents and sediment movements (Gill, 2005; Langhamer et al., 2010; McCluskie et al., 2013); and disturbance from increased vessel traffic (Bellefleur et al., 2009; Garthe and Hüppop, 2004; Ronconi and Clair, 2002; Schwemmer et al., 2010). Potential positive effects include the increased provision of loafing and roost structures at sea (Deurs et al., 2012; Grecian et al., 2010); and increased foraging opportunities if devices act as fish aggregation devices (FADs) (Langhamer and Wilhelmsson, 2009; Wilhelmsson et al., 2006), artificial reefs (Langhamer et al., 2009; Reubens et al., 2013a, 2013b) or de facto marine protected areas (dfMPAs) (Grecian et al., 2010; Inger et al., 2009). Although knowledge of seabird ecology and behaviour allows moderately robust predictions of which species of seabirds are most likely to be at risk from impacts (Furness et al., 2013, 2012; Garthe and Hüppop, 2004), a lack of knowledge regarding seabird movements at sea
limits our ability to identify and quantify potential effects of MREDs on populations (McCluskie et al., 2013). The great skua (Stercorarius skua) is a protected species breeding in Scotland from April to late August/early September, with numerous breeding colonies located in close proximity to leased and proposed MREDs. Currently we know very little about the movements of this species at sea, which makes it difficult to predict how they may be affected by MREDs. Great skuas have a restricted global distribution with over 60% of the global population (ca.16,000 pairs) breeding in Scotland at the most southern part of their range and migrating to winter off Iberia and northwest Africa (Magnusdottir et al., 2012; Mitchell et al., 2004). This internationally important seabird population is experiencing low levels of breeding success (Orkney Bird Report Committee, 2013; Shetland Bird Club, 2013). In Orkney, the breeding population of great skuas has undergone a 23% reduction over the last decade (Meek et al., 2011) and numbers at the largest colonies in Shetland have
H.M. Wade et al. / Marine Environmental Research 101 (2014) 69e80
also decreased (JNCC, 2014b). These declines are likely to have been driven by a reduced sandeel (Ammodytes marinus) abundance (Caldow and Furness, 2000), decreases in fisheries discarding and possibly the impacts of climate change (Oswald et al., 2008, 2011), which have resulted in lower prey availability for great skuas (Hamer et al., 1991; Votier et al., 2004; Meek et al., 2011). This evidence suggests that great skuas are already under pressure from changing climate and human activities such as fishing, potentially in both their breeding and wintering foraging areas. As such, this species is at risk from additional impacts of MREDs that could compound existing declines in productivity and population sizes. Existing knowledge suggests that great skuas are most likely to be at risk from collision with offshore wind turbines, owing to their tendency to spend the majority of their time at sea in flight (except at night when they sit on the water or return to the colony), and because they fly at altitudes overlapping with turbine rotors (Furness et al., 2013; Garthe and Hüppop, 2004). The species is less likely to be negatively affected by wave and tidal-stream MREDs as they are surface feeders and so unlikely to collide with devices on or below the water surface (Furness et al., 2012). Birds struggling to find sufficient prey or birds from larger breeding colonies travel further to forage than those from smaller colonies or when food is abundant (Bertrand et al., 2012; Lewis et al., 2001; Wakefield et al., 2013). Therefore, great skuas that travel over greater distances and range more widely increase their probability of encountering and potentially being affected by a MRED. Conversely, if MREDs are located in close proximity to breeding colonies, breeding birds that are constrained to return to their nest and limited in the area over which they can forage (Shaffer et al., 2003), may repeatedly encounter and be affected by MREDs (Masden et al., 2010). Given this evidence, different subsections of great skua populations may be affected in different ways by MREDs. To investigate how great skuas may be affected by MREDs we used tracking technology to improve our understanding of how this species uses the marine environment. We investigated the movements of breeding birds and failed breeders from two of the largest and best studied colonies in the UK (Fig. 1 and Supplementary Table 1) to assess differences according to colony location and breeding status of the bird (actively breeding or failed breeder). We also aimed to investigate the movements of birds according to their nest status (incubating, chick-rearing, following successful fledging of a chick and following a failed breeding attempt). The overlap of great skuas with leased and proposed MREDs was quantified for birds during different breeding stages and from two different colonies. In addition, we compared tracking data with historical data to assess whether great skua movements have changed over time, with the intention of exploring the implications of predicting the effects of MREDs on species based on short-term tracking data. 2. Materials and methods 2.1. Data collection Solar-powered GPS data loggers were fitted to 20 breeding great skuas in the period from 03 June 2011 to 14 June 2011. The loggers were attached for the duration of the breeding season. Ten loggers were fitted to birds from Hoy, Orkney, UK (58 520 N, 3 240 W) and ten loggers fitted to birds from Foula, Shetland, UK (60 80 N, 2 50 W). The study sites represent the largest breeding colonies for great skuas in Orkney and Shetland and are located in close proximity to leased and proposed MRED sites (Fig. 1 and Supplementary Table 1). Birds were caught on the nest using a remote-controlled noose trap and logger attachment took a mean of 25 min (range ¼ 19e32 min). UvA-BiTs GPS data loggers (Bouten et al., 2013) weighing ca. 25 g (500 m beyond a pre-established colony perimeter and lasting >20 min were defined as ‘trips’. Trips covering distances >2000 km (n ¼ 2), which were an order of magnitude longer than the next longest trips, were excluded from analyses as they biased utilisation distribution (UD) calculations. When comparing movements of birds from Hoy and Foula, trips were categorised as occurring either during breeding or following a failed breeding attempt. When comparing movements of birds from Hoy only, trips were defined in more detail as occurring during incubation, chickrearing, following successful fledging of a chick or following a failed breeding attempt. For each trip, we calculated the trip duration (time elapsed between departure and return to the colony), the foraging range (the maximum distance reached from the colony), and the total distance travelled (summation of distances between GPS points along the route). The number of trips per day were defined as the number of times a bird left the colony on a foraging trip. When birds did not leave the colony or were away from the colony on a foraging trip lasting longer than a day, a value of zero trips undertaken that day was recorded. Trip metrics were calculated using R package ‘trip’ ver.1.1-17 (Sumner, 2013) and Geospatial Modelling Environment ver.0.7.2.0 (Beyer, 2012). 2.3. Utilisation distributions To estimate the core area used by birds in each category (outlined in Section 2.2) we used kernel density estimates (KDE) to calculate the 50% UD (Harris et al., 1990). Owing to variable intervals between GPS fixes and missing GPS locations, data were linearly interpolated to establish regular 10 min time intervals between GPS points. Interpolation of GPS locations eliminated overestimation and bias in habitat/area use as a result of clustering of short-interval GPS fixes in one area. KDEs were calculated using all location fixes within each category (outlined in Section 2.2) using a fixed smoothing parameter of 10,000 m and a grid size of 2,500 m. The smoothing parameter and grid size were identified as most appropriately representing the original data through visual assessment of a series of UDs calculated with a range of bandwidths. For birds in each category, we calculated the percentage overlap of the 50% UD with leased and proposed MRED sites (Scottish Government, 2013b). Analyses were performed using R packages ‘adehabitatLT’ ver.3.14 (Calenge, 2006), ‘adehabitatHR’ ver.4.10 (Calenge, 2006), and ArcGIS (ArcMap ver.10. ESRI, USA). 2.4. Variation in movements Multiple trips were recorded for each bird. To account for nonindependence of data as a result of repeated sampling, we used generalized linear mixed models (GLMM) with bird identity included as a random effect. In analyses comparing the movements
72
H.M. Wade et al. / Marine Environmental Research 101 (2014) 69e80
Fig. 2. Kernel density estimates of great skua movements from Foula (AeB) and Hoy (CeD) during breeding (A, C) and following breeding failure (B, D) until 16 July 2011.
of breeding and failed breeders from Hoy and Foula, models were constructed using a Gaussian distribution in R package ‘lme4’ ver.1.1-5 (Bates et al., 2012). In analyses comparing the movements of birds from Hoy during the breeding season (during incubation, chick-rearing, following successful fledging of chicks or following a failed breeding attempt), models were constructed using a Gaussian distribution in R package ‘nlme’ ver.3.1-109 (Pinheiro et al., 2013). This different model structure was required to compare the movements of birds from the Hoy colony alone as there was a greater variance in trips incorporating freshwater bathing or visits to probable club sites (sites where non-breeders congregate; Furness, 1987) than in trips that went directly out to sea, which was not evident when jointly analysing the Hoy and Foula movements. To account for this heterogeneity in variance that was attributed to trip type, a fixed variance structure was
applied. In all models, trip length (km), duration (min) and maximal distance from the colony (km) were response variables and were log-transformed prior to statistical analysis to conform to assumptions of normality. Significance was assessed using likelihood ratio tests (LRTs) following backward step-wise model selection (Crawley, 2007). All analyses were implemented in R ver.3.0.1 (R Core Team, 2013). 2.5. Historical data Productivity on Foula, and data on nest attendance and trip characteristics during chick rearing, were extracted from published and unpublished literature from 1974 to 2011 (Catry, 1997; Furness, 1977; JNCC, 2014b; SOTEAG, 2012). Trip duration (time away from
H.M. Wade et al. / Marine Environmental Research 101 (2014) 69e80
nest), the number of trips undertaken per day and nest attendance were recorded during nest watches. Where available, productivity data for Hoy were also extracted (JNCC, 2014b).
3. Results Of the 20 loggers deployed, data from 17 were analysed. The remaining three loggers (all from Hoy) did not charge their batteries sufficiently for GPS fixes to be taken regularly or the tagged bird left the colony very soon after logger attachment. The duration of data collection ranged from 23 to 106 days. A total of 2039 trips were recorded. Despite tagged birds not returning to their colony to breed the following year, monitoring of the control group of 20 breeding individuals (ten from each colony) found no adverse effects of the logger and harness attachment on territory attendance or breeding productivity during the 2011 breeding season (Thaxter et al., in review-a).
3.1. Comparison of movements of breeding birds and failed breeders from Hoy and Foula A total of 651 foraging trips were recorded for 17 birds of known breeding status from Hoy and Foula between 03 June 2011 and 15 July 2011.
3.1.1. Trip characteristics Failed breeders from each colony spent more time at sea, travelled over greater distances and undertook fewer trips per day than birds engaged in a breeding attempt. Birds from Foula tended to spend more time at sea and travelled further than birds from Hoy (Supplementary Table 2; Figs. 2 and 3A).
73
3.1.2. Variation in movements between colonies and according to breeding status The duration of trips differed according to whether a bird was breeding or had failed in its breeding attempt (GLMM: Х 2 ¼ 33.35, d.f. ¼ 1, p < 0.001); with breeding birds spending less time away from the colony than failed breeders (Fig. 3B). The length of trips and the maximum distance reached from the colony during trips also differed according to breeding status (GLMM: Х 2 ¼ 12.04, d.f. ¼ 1, p < 0.001; GLMM: Х 2 ¼ 9.12, d.f. ¼ 1, p < 0.01 respectively) (Fig. 3C and D). The duration of trips also differed according to colony (GLMM: Х 2 ¼ 12.35, d.f. ¼ 1, p < 0.001); with birds from Foula undertaking trips that lasted longer than trips by birds from Hoy (Fig. 3B). The distance covered during trips and the maximum distance reached from the colony during trips differed according to colony (GLMM: Х 2 ¼ 18.15, d.f. ¼ 1, p < 0.001; GLMM: Х 2 ¼ 18.37, d.f. ¼ 1, p < 0.001 respectively) (Fig. 3C and D). 3.1.3. Overlap with marine renewables There was no overlap of core areas with leased or proposed offshore wind sites for birds from Foula. A very low overlap with leased sites for offshore wind development was observed for failed breeders from Hoy (
Contents lists available at ScienceDirect
Marine Environmental Research journal homepage: www.elsevier.com/locate/marenvrev
Great skua (Stercorarius skua) movements at sea in relation to marine renewable energy developments H.M. Wade a, *, E.A. Masden a, A.C. Jackson a, b, C.B. Thaxter c, N.H.K. Burton c, W. Bouten d, R.W. Furness e, f a
Environmental Research Institute, North Highland College, University of the Highlands and Islands, Thurso, UK Cornwall College Newquay, Wildflower Lane, Trenance Gardens, Newquay, Cornwall, UK British Trust for Ornithology, The Nunnery, Thetford, Norfolk, UK d Computational Geo-Ecology, Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Amsterdam, The Netherlands e College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, UK f MacArthur Green, 95 South Woodside Road, Glasgow, UK b c
a r t i c l e i n f o
a b s t r a c t
Article history: Received 12 June 2014 Received in revised form 14 September 2014 Accepted 15 September 2014 Available online 17 September 2014
Marine renewable energy developments (MREDs) are an increasing feature of the marine environment. Owing to the relatively small number of existing developments and the early stage of their associated environmental monitoring programmes, the effects of MREDs on seabirds are not fully known. Our ability to fully predict potential effects is limited by a lack of knowledge regarding movements of seabirds at sea. We used GPS tracking to improve our understanding of the movements at sea of a protected seabird species breeding in Scotland, the great skua (Stercorarius skua), to better predict how this species may be affected by MREDs. We found that the overlap of great skuas with leased and proposed MREDs was low; particularly with offshore wind sites, which are predicted to present a greater risk to great skuas than wave or tidal-stream developments. Failed breeders overlapped with larger areas of MREDs than breeding birds but the overall overlap with core areas used remained low. Overlap with wave energy development sites was greater than for offshore wind and tidal-stream sites. Comparison of 2011 data with historical data indicates that distances travelled by great skuas have likely increased over recent decades. This suggests that basing marine spatial planning decisions on short-term tracking data could be less informative than longer-term data. © 2014 Elsevier Ltd. All rights reserved.
Keywords: Environmental impact Remote sensing Seabirds GPS Marine spatial planning Wind power Wave power Bio-logging
1. Introduction Marine renewable energy developments (MREDs) are an increasing feature of the marine environment. The rapid development of these industries is being driven by the need to reduce carbon emissions and to increase energy security. In Scotland, The Crown Estate and the Scottish Government are working in partnership to realise the Scottish Government's target of generating 100% of electricity from renewable sources by 2020 (Scottish Government, 2011). An advanced marine plan has been developed for Scotland with sites leased to offshore wind, wave and
* Corresponding author. Environmental Research Institute, North Highland College, University of the Highlands and Islands, Ormlie Road, Thurso, KW14 7EE, UK. E-mail address: [email protected] (H.M. Wade). http://dx.doi.org/10.1016/j.marenvres.2014.09.003 0141-1136/© 2014 Elsevier Ltd. All rights reserved.
tidal-stream energy developers, with additional sites proposed for future development (Scottish Government, 2013a). Owing to the small number and limited extent of existing MREDs, the effects of offshore wind, wave and tidal-stream installations on seabirds are not fully known (Fox et al., 2006; McCluskie et al., 2013; Witt et al., 2012). This is despite several leased and proposed MRED sites in north-east Scotland being located in close proximity to special protection areas (SPAs) designated specifically to safeguard breeding populations of internationally important seabird species (JNCC, 2014a) (Fig. 1). SPA legislation (EC Birds Directive 2009/147/EC) requires that any development does not damage the integrity of protected seabird populations (Scottish Natural Heritage, 2014). Some of the potential negative effects of MREDs on seabirds include collision with devices (Everaert and Stienen, 2007; Langton et al., 2011); displacement from areas where devices are located (Larsen and Guillemette, 2007; McDonald et al., 2012); increased flight costs if devices act
70
H.M. Wade et al. / Marine Environmental Research 101 (2014) 69e80
Fig. 1. Locations of breeding colonies (yellow circles; defined by number of Apparently Occupied Territories (AOT)) and special protection areas (blue polygons) designated for breeding seabird populations in north-east Scotland, and the proximity of leased and proposed marine renewable energy development sites (solid polygons ¼ leased sites; crosshatched ¼ proposed sites; purple ¼ offshore wind sites; green ¼ wave sites; red ¼ tidal-stream sites). Study colonies indicated with black arrows. See Supplementary Table 1 for colony sizes and year of AOT count. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
as barriers to movement (Desholm and Kahlert, 2005; Masden et al., 2009, 2010); alteration of foraging habitats if devices alter currents and sediment movements (Gill, 2005; Langhamer et al., 2010; McCluskie et al., 2013); and disturbance from increased vessel traffic (Bellefleur et al., 2009; Garthe and Hüppop, 2004; Ronconi and Clair, 2002; Schwemmer et al., 2010). Potential positive effects include the increased provision of loafing and roost structures at sea (Deurs et al., 2012; Grecian et al., 2010); and increased foraging opportunities if devices act as fish aggregation devices (FADs) (Langhamer and Wilhelmsson, 2009; Wilhelmsson et al., 2006), artificial reefs (Langhamer et al., 2009; Reubens et al., 2013a, 2013b) or de facto marine protected areas (dfMPAs) (Grecian et al., 2010; Inger et al., 2009). Although knowledge of seabird ecology and behaviour allows moderately robust predictions of which species of seabirds are most likely to be at risk from impacts (Furness et al., 2013, 2012; Garthe and Hüppop, 2004), a lack of knowledge regarding seabird movements at sea
limits our ability to identify and quantify potential effects of MREDs on populations (McCluskie et al., 2013). The great skua (Stercorarius skua) is a protected species breeding in Scotland from April to late August/early September, with numerous breeding colonies located in close proximity to leased and proposed MREDs. Currently we know very little about the movements of this species at sea, which makes it difficult to predict how they may be affected by MREDs. Great skuas have a restricted global distribution with over 60% of the global population (ca.16,000 pairs) breeding in Scotland at the most southern part of their range and migrating to winter off Iberia and northwest Africa (Magnusdottir et al., 2012; Mitchell et al., 2004). This internationally important seabird population is experiencing low levels of breeding success (Orkney Bird Report Committee, 2013; Shetland Bird Club, 2013). In Orkney, the breeding population of great skuas has undergone a 23% reduction over the last decade (Meek et al., 2011) and numbers at the largest colonies in Shetland have
H.M. Wade et al. / Marine Environmental Research 101 (2014) 69e80
also decreased (JNCC, 2014b). These declines are likely to have been driven by a reduced sandeel (Ammodytes marinus) abundance (Caldow and Furness, 2000), decreases in fisheries discarding and possibly the impacts of climate change (Oswald et al., 2008, 2011), which have resulted in lower prey availability for great skuas (Hamer et al., 1991; Votier et al., 2004; Meek et al., 2011). This evidence suggests that great skuas are already under pressure from changing climate and human activities such as fishing, potentially in both their breeding and wintering foraging areas. As such, this species is at risk from additional impacts of MREDs that could compound existing declines in productivity and population sizes. Existing knowledge suggests that great skuas are most likely to be at risk from collision with offshore wind turbines, owing to their tendency to spend the majority of their time at sea in flight (except at night when they sit on the water or return to the colony), and because they fly at altitudes overlapping with turbine rotors (Furness et al., 2013; Garthe and Hüppop, 2004). The species is less likely to be negatively affected by wave and tidal-stream MREDs as they are surface feeders and so unlikely to collide with devices on or below the water surface (Furness et al., 2012). Birds struggling to find sufficient prey or birds from larger breeding colonies travel further to forage than those from smaller colonies or when food is abundant (Bertrand et al., 2012; Lewis et al., 2001; Wakefield et al., 2013). Therefore, great skuas that travel over greater distances and range more widely increase their probability of encountering and potentially being affected by a MRED. Conversely, if MREDs are located in close proximity to breeding colonies, breeding birds that are constrained to return to their nest and limited in the area over which they can forage (Shaffer et al., 2003), may repeatedly encounter and be affected by MREDs (Masden et al., 2010). Given this evidence, different subsections of great skua populations may be affected in different ways by MREDs. To investigate how great skuas may be affected by MREDs we used tracking technology to improve our understanding of how this species uses the marine environment. We investigated the movements of breeding birds and failed breeders from two of the largest and best studied colonies in the UK (Fig. 1 and Supplementary Table 1) to assess differences according to colony location and breeding status of the bird (actively breeding or failed breeder). We also aimed to investigate the movements of birds according to their nest status (incubating, chick-rearing, following successful fledging of a chick and following a failed breeding attempt). The overlap of great skuas with leased and proposed MREDs was quantified for birds during different breeding stages and from two different colonies. In addition, we compared tracking data with historical data to assess whether great skua movements have changed over time, with the intention of exploring the implications of predicting the effects of MREDs on species based on short-term tracking data. 2. Materials and methods 2.1. Data collection Solar-powered GPS data loggers were fitted to 20 breeding great skuas in the period from 03 June 2011 to 14 June 2011. The loggers were attached for the duration of the breeding season. Ten loggers were fitted to birds from Hoy, Orkney, UK (58 520 N, 3 240 W) and ten loggers fitted to birds from Foula, Shetland, UK (60 80 N, 2 50 W). The study sites represent the largest breeding colonies for great skuas in Orkney and Shetland and are located in close proximity to leased and proposed MRED sites (Fig. 1 and Supplementary Table 1). Birds were caught on the nest using a remote-controlled noose trap and logger attachment took a mean of 25 min (range ¼ 19e32 min). UvA-BiTs GPS data loggers (Bouten et al., 2013) weighing ca. 25 g (500 m beyond a pre-established colony perimeter and lasting >20 min were defined as ‘trips’. Trips covering distances >2000 km (n ¼ 2), which were an order of magnitude longer than the next longest trips, were excluded from analyses as they biased utilisation distribution (UD) calculations. When comparing movements of birds from Hoy and Foula, trips were categorised as occurring either during breeding or following a failed breeding attempt. When comparing movements of birds from Hoy only, trips were defined in more detail as occurring during incubation, chickrearing, following successful fledging of a chick or following a failed breeding attempt. For each trip, we calculated the trip duration (time elapsed between departure and return to the colony), the foraging range (the maximum distance reached from the colony), and the total distance travelled (summation of distances between GPS points along the route). The number of trips per day were defined as the number of times a bird left the colony on a foraging trip. When birds did not leave the colony or were away from the colony on a foraging trip lasting longer than a day, a value of zero trips undertaken that day was recorded. Trip metrics were calculated using R package ‘trip’ ver.1.1-17 (Sumner, 2013) and Geospatial Modelling Environment ver.0.7.2.0 (Beyer, 2012). 2.3. Utilisation distributions To estimate the core area used by birds in each category (outlined in Section 2.2) we used kernel density estimates (KDE) to calculate the 50% UD (Harris et al., 1990). Owing to variable intervals between GPS fixes and missing GPS locations, data were linearly interpolated to establish regular 10 min time intervals between GPS points. Interpolation of GPS locations eliminated overestimation and bias in habitat/area use as a result of clustering of short-interval GPS fixes in one area. KDEs were calculated using all location fixes within each category (outlined in Section 2.2) using a fixed smoothing parameter of 10,000 m and a grid size of 2,500 m. The smoothing parameter and grid size were identified as most appropriately representing the original data through visual assessment of a series of UDs calculated with a range of bandwidths. For birds in each category, we calculated the percentage overlap of the 50% UD with leased and proposed MRED sites (Scottish Government, 2013b). Analyses were performed using R packages ‘adehabitatLT’ ver.3.14 (Calenge, 2006), ‘adehabitatHR’ ver.4.10 (Calenge, 2006), and ArcGIS (ArcMap ver.10. ESRI, USA). 2.4. Variation in movements Multiple trips were recorded for each bird. To account for nonindependence of data as a result of repeated sampling, we used generalized linear mixed models (GLMM) with bird identity included as a random effect. In analyses comparing the movements
72
H.M. Wade et al. / Marine Environmental Research 101 (2014) 69e80
Fig. 2. Kernel density estimates of great skua movements from Foula (AeB) and Hoy (CeD) during breeding (A, C) and following breeding failure (B, D) until 16 July 2011.
of breeding and failed breeders from Hoy and Foula, models were constructed using a Gaussian distribution in R package ‘lme4’ ver.1.1-5 (Bates et al., 2012). In analyses comparing the movements of birds from Hoy during the breeding season (during incubation, chick-rearing, following successful fledging of chicks or following a failed breeding attempt), models were constructed using a Gaussian distribution in R package ‘nlme’ ver.3.1-109 (Pinheiro et al., 2013). This different model structure was required to compare the movements of birds from the Hoy colony alone as there was a greater variance in trips incorporating freshwater bathing or visits to probable club sites (sites where non-breeders congregate; Furness, 1987) than in trips that went directly out to sea, which was not evident when jointly analysing the Hoy and Foula movements. To account for this heterogeneity in variance that was attributed to trip type, a fixed variance structure was
applied. In all models, trip length (km), duration (min) and maximal distance from the colony (km) were response variables and were log-transformed prior to statistical analysis to conform to assumptions of normality. Significance was assessed using likelihood ratio tests (LRTs) following backward step-wise model selection (Crawley, 2007). All analyses were implemented in R ver.3.0.1 (R Core Team, 2013). 2.5. Historical data Productivity on Foula, and data on nest attendance and trip characteristics during chick rearing, were extracted from published and unpublished literature from 1974 to 2011 (Catry, 1997; Furness, 1977; JNCC, 2014b; SOTEAG, 2012). Trip duration (time away from
H.M. Wade et al. / Marine Environmental Research 101 (2014) 69e80
nest), the number of trips undertaken per day and nest attendance were recorded during nest watches. Where available, productivity data for Hoy were also extracted (JNCC, 2014b).
3. Results Of the 20 loggers deployed, data from 17 were analysed. The remaining three loggers (all from Hoy) did not charge their batteries sufficiently for GPS fixes to be taken regularly or the tagged bird left the colony very soon after logger attachment. The duration of data collection ranged from 23 to 106 days. A total of 2039 trips were recorded. Despite tagged birds not returning to their colony to breed the following year, monitoring of the control group of 20 breeding individuals (ten from each colony) found no adverse effects of the logger and harness attachment on territory attendance or breeding productivity during the 2011 breeding season (Thaxter et al., in review-a).
3.1. Comparison of movements of breeding birds and failed breeders from Hoy and Foula A total of 651 foraging trips were recorded for 17 birds of known breeding status from Hoy and Foula between 03 June 2011 and 15 July 2011.
3.1.1. Trip characteristics Failed breeders from each colony spent more time at sea, travelled over greater distances and undertook fewer trips per day than birds engaged in a breeding attempt. Birds from Foula tended to spend more time at sea and travelled further than birds from Hoy (Supplementary Table 2; Figs. 2 and 3A).
73
3.1.2. Variation in movements between colonies and according to breeding status The duration of trips differed according to whether a bird was breeding or had failed in its breeding attempt (GLMM: Х 2 ¼ 33.35, d.f. ¼ 1, p < 0.001); with breeding birds spending less time away from the colony than failed breeders (Fig. 3B). The length of trips and the maximum distance reached from the colony during trips also differed according to breeding status (GLMM: Х 2 ¼ 12.04, d.f. ¼ 1, p < 0.001; GLMM: Х 2 ¼ 9.12, d.f. ¼ 1, p < 0.01 respectively) (Fig. 3C and D). The duration of trips also differed according to colony (GLMM: Х 2 ¼ 12.35, d.f. ¼ 1, p < 0.001); with birds from Foula undertaking trips that lasted longer than trips by birds from Hoy (Fig. 3B). The distance covered during trips and the maximum distance reached from the colony during trips differed according to colony (GLMM: Х 2 ¼ 18.15, d.f. ¼ 1, p < 0.001; GLMM: Х 2 ¼ 18.37, d.f. ¼ 1, p < 0.001 respectively) (Fig. 3C and D). 3.1.3. Overlap with marine renewables There was no overlap of core areas with leased or proposed offshore wind sites for birds from Foula. A very low overlap with leased sites for offshore wind development was observed for failed breeders from Hoy (
