SEARES-01489; No of Pages 9 Journal of Sea Research xxx (2016) xxx–xxx Contents lists available at ScienceDirect Journal of Sea Research journal homepage: www.elsevier.com/locate/seares GPS-tracking and colony observations reveal variation in offshore habitat use and foraging ecology of breeding Sandwich Terns R.C. Fijn a,⁎, J. de Jong a, W. Courtens b, H. Verstraete b, E.W.M. Stienen b, M.J.M. Poot a,1 a b Bureau Waardenburg bv, Consultants for Environment & Ecology, P.O. Box 365, 4100 AJ Culemborg, The Netherlands Research Institute for Nature and Forest (INBO), Kliniekstraat 25, 1070 Brussels, Belgium a r t i c l e i n f o Article history: Received 7 March 2016 Received in revised form 11 November 2016 Accepted 27 November 2016 Available online xxxx Keywords: Thalasseus sandvicencis Foraging strategy GPS-logger Resting Behaviour Enclosure Multi-method approach a b s t r a c t Breeding success of seabirds critically depends on their foraging success offshore. However, studies combining atsea tracking and visual provisioning observations are scarce, especially for smaller species of seabirds. This study is the first in which breeding Sandwich Terns were tracked with GPS-loggers to collect detailed data on foraging habitat use in four breeding seasons. The maximum home range of individual Sandwich Terns comprised approximately 1900 km2 and the average foraging range was 27 km. Trip durations were on average 135 min with average trip lengths of 67 km. Actual foraging behaviour comprised 35% of the time budget of a foraging trip. Substantial year-to-year variation was found in habitat use and trip variables, yet with the exception of 2012, home range size remained similar between years. Food availability, chick age and environmental conditions are proposed as the main driving factors between inter- and intra-annual variations in trip variables. Our multi-method approach also provided geo-referenced information on prey presence and we conclude that future combining of colony observations and GPS-loggers deployments can potentially provide a near complete insight into the feeding ecology of breeding Sandwich Terns, including the behaviour of birds at sea. © 2016 Elsevier B.V. All rights reserved. 1. Introduction A wide array of techniques to study the offshore distribution and activities of seabirds is available: from direct observations from ships or airplanes to the use of electronic tracking devices. The latter can provide detailed insights into flight paths of individual birds, which allows the spatial and temporal distributions of seabirds at sea to be identified (e.g. Weimerskirch et al., 2002; Kotzerka et al., 2010) and often offers information on foraging behaviour of individual birds (e.g. Wilson et al., 2002; Cooke et al., 2004; Shamoun-Baranes et al., 2011). From the variety of tracking methods available, GPS-loggers provide the highest resolution and accuracy. Due to the relatively large size of the batteries, these devices were until recently fairly heavy, and have consequently only been applied on larger seabird species (e.g. Weimerskirch et al., 2002; Gremillet et al., 2004; Gyimesi et al., 2011; Kotzerka et al., 2010). With the ongoing miniaturization of GPS-loggers, also smaller seabird species can now be equipped with appropriately sized tags. Similar to tracking techniques, there is also a large variety of methods available to determine the diet (Barrett et al., 2007) and ⁎ Corresponding author. E-mail address: r.c.fijn@buwa.nl (R.C. Fijn). 1 Present address: Statistics Netherlands, P.O. Box 24500, 2490 HA The Hague, The Netherlands. breeding parameters of seabirds. For species that carry their prey in plain sight (e.g. terns, alcids), chick-feeding diets can be studied by observing returning adults in the colony. One of these is the Sandwich Tern Thalasseus sandvicencis, a single-prey loader that breeds colonially on the coasts of Europe. Sandwich Terns are generally known for their specialisation in prey choice (Stienen et al., 2000, Courtens et al., this volume), and provisioning rates, foraging duration and diet have been extensively studied in chick-rearing terns (e.g. Garthe and Kubetzki, 1998; Stienen et al., 2000; Dies and Dies, 2005; Stienen et al., 2015). Nevertheless, where adults collect the food for their chicks or how Sandwich Terns exploit their offshore foraging habitat remains largely unknown. Some observational and VHF radio tracking studies exist (e.g. Stienen, 2006; Baptist and Leopold, 2010; Perrow et al., 2011; Poot et al., 2014) and these have revealed that Sandwich Terns use shallow waters close to the coast as well as areas tens of kilometres offshore. Yet most of these studies, the exception being Perrow et al., 2011, were unable to track complete foraging trips of terns at sea. In this study, a new generation of miniature GPS-loggers was used on Sandwich Terns for the first time in order to expand on the conventional colony research and to focus on offshore habitat use, time budgets and behaviour during foraging trips. By doing so, we were able to collect novel quantitative data on home range and foraging trip characteristics and also to map the offshore origin of prey items of known species and size for the first time. http://dx.doi.org/10.1016/j.seares.2016.11.005 1385-1101/© 2016 Elsevier B.V. All rights reserved. Please cite this article as: Fijn, R.C., et al., GPS-tracking and colony observations reveal variation in offshore habitat use and foraging ecology of breeding Sandwich Terns, J. Sea Res. (2016), http://dx.doi.org/10.1016/j.seares.2016.11.005 2 R.C. Fijn et al. / Journal of Sea Research xxx (2016) xxx–xxx 2. Material and methods 2.1. Colony-based research Fieldwork was done in 2012–2015 in two locations in the southwestern part of the Netherlands within the Natura 2000-SPA Haringvliet. In 2012, 2013 and 2015 the Sandwich Tern colony was located at the Scheelhoek Nature Reserve (N51°49′ E04°04′), and in 2014 the colony was located at the nearby (5 km to the east) Slijkplaat (N51°48′ E04°09′). Each year the Natura 2000-SPA Haringvliet holds a colony of approximately 1500–3500 pairs of Sandwich Terns on either Scheelhoek or Slijkplaat. A representative part of the colony was fenced off with an enclosure, which encompasses both nests in the core of the colony, where the fittest pairs breed as well as nests towards the periphery of the colony (c.f. Stienen et al., 2000). By doing so the chicks of the nests in this part remained in the same place and could easily be studied. A small hide was erected next to the enclosure, which allowed direct observations of prey deliveries to the chicks. Two times a week, two observers conducted a total of 8 h of observational protocols. During these protocols, provisioning rates, trip durations, prey species and prey length (relative to bill-length) were recorded. We determined foraging trip duration for 931 trips longer than 5 min for individually recognizable adults (ringed, colour dyed or specific head moult characteristics) by determining the time the adult left and subsequently returned to the colony with recognizable prey (N = 905). 2.2. GPS-tracking A total of 34 adult Sandwich Terns were captured in May and June of 2012–2015 and fitted with GPS-loggers. Birds were either captured on the nest during the last week of incubation or early chick-rearing stages (n = 12) with walk-in traps, or in the later stages of chick-rearing with spring traps (n = 22). Captures were spread in time throughout the breeding season to sample birds in different stages of the breeding cycle since it was not feasible to cover the entire chick-rearing period due to the limited battery capacity of the loggers. Age of chicks from adults deployed with GPS-loggers was unknown and estimated taken as the mean hatching date of eggs in the enclosure. Captured birds were ringed with both a uniquely numbered metal ring and a fieldreadable plastic colour-ring. Distinct body parts of most of these birds were also treated with colour-dye (picric acid or silver nitrate) to allow easy recognition in the field. All birds were equipped with GPS-loggers (Ecotone ALLE-55 GPSUHF, ~ 4 g, L:35 × W:15 × H:10 mm), that can take up to ~ 400 GPSfixes on one battery load, depending on environmental conditions and sampling interval. All loggers stored GPS positions with 5 min intervals to the device memory and included date, time, latitude, longitude and GPS-speed. Data were automatically transferred to base stations placed in the colony from a distance up to ~ 100 m. All loggers were programmed to start recording 48 h after capture and had different duty cycles throughout the day. Some recorded data on many days with a duty cycle of only 6 h per day, while other devices were programmed with a longer cycle (12–16 h) to allow data collection over the entire day. Seven of these loggers were attached to feathers on the back with TESA tape (No. 4651; Beiersdorf AG, Hamburg, Germany) following the methodology described by Wilson et al. (1997). Sandwich Terns turned out to be aggressive towards the tape deployments and were capable of removing the loggers by plucking and biting the taped feathers and resulting in substantial tag loss (4 out of the initial 7 tags). The other 27 loggers were therefore attached with a specially designed backpack loop harness following the design described by Kenward (1985). We used a thick and supple fishing elastic (Preston Innovations Slip Elastic, diameter 1.4–2.2 mm). This made the harness strong and flexible but also ensured that the harnesses were shed after 2 to 3 months due to degradation under influence of the sun (Fijn et al., in preparation). The weight of these loggers including rings and attachment material (5.8 g) is well within the range of the generally accepted limit of 3% of the body mass (Phillips et al., 2003; Vandenabeele et al., 2011a) of the Sandwich Terns in our study (average weight of 241 ± 13.4 g; range min 210–max 270 g; ~2.4%). Handling time (capture to release) was approximately 15 min for tape deployments to 10 min for harness deployments. A total of 24 out of 34 loggers successfully transferred positional data to the base station placed in the colony. Five birds had shed the logger prematurely, probably due to pulling out the feathers to which the tape was fixed in 2012 (N = 3), or due to the use of an experimental weak link in the harness design in 2015 (N = 2). The fate of the other five missing loggers is unknown. Of the 24 loggers that provided data, one recorded a behavioural response due to failed breeding (Fijn et al., 2014) and two others malfunctioned and only transferred a part of a single foraging trip. These three loggers were left out of the analyses. A total of 154 foraging trips were logged by the remaining 21 loggers (Table 1). Based on flight direction, flight speed and habitat characteristics, individual GPS positions were classified into six categories: 1. resting near colony, 2. guarding in colony, 3. resting outside the colony, 4. commuting to the foraging area, 5. foraging, and 6. commuting to the colony. Resting was defined as a combination of speed b0.5 m/s, and more than two subsequent fixes in close proximity on solid ground. Commuting was defined as a straight lined flight path away or towards the colony and flight speeds N 0.5 m/s. Foraging was defined as clustered positions in habitat away from land and flight speeds N 0.5 m/s. Distances between individual fixes were calculated in ArcGIS (Esri, version 10.2). Trips were either recorded completely (from start to end in the colony), nearly complete (only small parts of the trip missing when flying away or towards the colony), or incomplete (complete commutes or parts of foraging missing) and were marked accordingly in the database. Incomplete trips were the consequence of trips that exceeded the programmed duty cycle of the logger or suffered from premature battery depletion. Habitat use and home range in this paper was defined based on complete, nearly complete and incomplete trips (21 loggers, 154 trips). To determine maximum foraging distance or foraging range (straight line from the most distant position to the colony) and trip length (sum of all individual distances between positions for one foraging trip), we used 101 complete and near-complete trips from 20 loggers (Table 1), extrapolating missing parts of the nearly complete track with a straight line to the colony. As a consequence, trip length is expected to be slightly underestimated. The logger that was left out of the analysis recorded only two incomplete trips. For the calculations of trip duration (time difference between start and end of trip), we used only the trips that were recorded completely from start to end in the colony (65 complete trips of 16 loggers, Table 1), as trip duration can only be calculated when both the start and end of a trip was known. 2.3. Statistical analyses We used Kernel Density Estimation (Worton, 1989) to determine habitat use based on individual location fixes. A kernel density map of all locations characterised as foraging was generated in ArcGIS with a search radius of 3000 m. Minimum Convex Polygons were created around all points (Kernohan et al., 2001) to estimate home range of individual birds, separate years, and all birds combined. To overcome pseudoreplication issues, as individuals were tracked for several successive trips, data were analysed using mixed-model ANOVA (using the function lme in the R package nlme [Pinheiro et al., 2015]) with trip duration, range, etc. as dependent variables, methodology and year as fixed factor, and individual as a random factor. Nonparametric statistical tests (Spearman r) were used to check for correlations between trip variables. Non-parametric tests were also performed to test for differences in trip duration, trip length and foraging range between years (Kruskal-Wallis test) with post-hoc pairwise multiple Please cite this article as: Fijn, R.C., et al., GPS-tracking and colony observations reveal variation in offshore habitat use and foraging ecology of breeding Sandwich Terns, J. Sea Res. (2016), http://dx.doi.org/10.1016/j.seares.2016.11.005 R.C. Fijn et al. / Journal of Sea Research xxx (2016) xxx–xxx 3 Table 1 Sample sizes and trip characteristics of foraging trips of breeding Sandwich Terns Thalasseus sandvicencis with GPS-loggers (21 loggers) and marked individuals (931 trips) observed during food protocols in the colony in four consecutive years. Presented are averages ± standard deviation (minimum–maximum). GPS-loggers Colony obs. Average trip duration (min) Year Bird ID # of trips Incomplete Near complete Complete Home range (km2) Average foraging range (km) Average trip length (km) Average trip duration (min) 2012 B-N32 B-N34 B-N72 B-N73 B-N74 B-N75 W-NH6 W-NK3 W-NKT B-N01 L-N02 L-N06 L-N07 L-N12 L-N17 W-N1C Y-N02 Y-N05 Y-N07 Y-N15 Y-N16 5 7 2 6 4 6 6 15 5 2 7 3 8 10 8 4 14 12 6 12 12 154 30 26 42 56 7 ±4 – 3 1 1 3 5 1 1 4 2 2 1 3 3 5 1 4 3 3 2 5 53 13 6 17 17 3 ±1 2 – 1 1 1 0 5 4 – – – – 1 4 1 3 5 – – 4 4 36 5 9 9 13 2 ±2 3 4 – 4 – 1 – 10 1 – 5 2 4 3 2 – 5 9 3 6 3 65 12 11 16 26 4 ±3 38 193 242 78 306 277 1057 1122 424 571 447 116 348 1433 365 809 969 1980 987 842 695 3714 893 2085 2170 2526 633 ± 497 12 14 28 11 31 18 40 35 38 – 22 32 23 28 25 42 22 37 33 24 29 33 ± 31 ± 65 24 ± 87 38 96 ± 78 ± 102 – 55 ± 81 ± 60 ± 67 ± 54 ± 95 ± 50 ± 97 ± 78 ± 59 ± 88 ± 89 ± 55 (51–152) 68 ± 34 (31–109) – 31 ± 10 (20–41) – 82 – 130 ± 61 (36–256) 298 – 99 ± 49 (32–165) 170 ± 45 (138–202) 116 ± 107 (33–272) 77 ± 79 (32–168) 65 ± 39 (37–92) – 157 ± 105 (38–249) 203 ± 105 (42–389) 229 ± 97 (151–338) 105 ± 86 (31–251) 241 ± 153 (140–417) 27 ± 13 (4–61) 67 ± 35 (9–155) 134 ± 95 (20–417) 931 377 206 171 177 105 ± 71 (5–398) 5 ±2 9 ±6 6 ±3 11 ± 3 3 2 2 3 1 5 2 4 3 6 3 5 189 ± 109 868 ± 386 584 ± 431 1095 ± 509 15 36 28 28 35 84 67 72 62 ± 39 (20–152) 145 ± 77 (36–298) 108 ± 68 (32–272) 179 ± 109 (31–417) 109 ± 80 (5–398) 103 ± 57 (10–270) 119 ± 74 (9–378) 90 ± 61 (9–265) 2013 2014 2015 Total Total 2012 Total 2013 Total 2014 Total 2015 Average (±sd) Average 2012 Average 2013 Average 2014 Average 2015 ±2 ±2 ±1 ±1 ±1 ±1 ±2 ±1 ±1 ±6 ±1 ±2 ± 1 (11−12) ± 9 (4–21) ± 7 (4–23) ± 11 (22–50) ± 10 (12–51) ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± comparison tests (pairwise comparisons using Tukey and Kramer (Nemenyi) test with Tukey-Dist approximation for independent samples) using the R-Package PMCMR (Pohlert, 2014). Logistic regression was used to study relationships between proportional data and trip duration. All statistical analyses were carried out in R (R Core Team, 2015) and most graphs were made with ggplot2 (Wickham, 2009). 3. Results 3.1. Habitat use, home range and trip length All trips were oriented in a northwest direction, but diverged into different directions (between north and southwest) as soon as the birds left the Haringvliet estuary. The combined foraging area of all tracked birds in all years comprised N 3700 km2 (Minimum Convex Polygon), with a maximum individual home range of 1980 km2. Home ranges were approximately 900 km2 in 2012 (based on 30 trips) and ranged between 2000 and 2500 km2 in 2013–2015 (based on 26, 42 and 56 trips respectively). Sandwich Terns foraged in different types of habitat from shallow waters close to the coast to deeper waters up to N 60 km from the colony (Table 1, Fig. 1). Exposed sand banks, beaches and mud flats were used as resting habitat but also the beaches of the breeding island as well as the protective rock dam around the colony were used. The average foraging range over all years is 27 ± 13 km (4–61) and considerable variation in the location of core foraging areas was found (Fig. 1). In 2012 the Kernel Density Estimates of foraging locations shows high usage of coastal areas, which is confirmed by lower foraging ranges and shorter trip lengths compared to the other three years when the birds foraged considerably further offshore. In 2013, almost all 7 (13−30) 3 (30−33) 11 (9–35) 15 (8–45) 12 (11–36) 3 (38–45) 13 (10–42) 14 (12–61) 10 (21–39) 14 (11–58) 4 (23–35) 8 (4–31) 10 (12–51) 12 (8–45) 13 (10–61) ± ± ± ± 7 (24–42) 21 (9–52) 18 (9–55) 32 (48–137) 29 (25–141) 19 (27–80) 27 (62–100) 40 (19–121) 42 (17–134) 24 (28–76) 8 (90–105) 32 (21–95) 36 (24–155) 20 (55–94) 36 (24–135) 30 (47–127) 20.9 (9–87) 29.8 (25–141) 31.9 (17–134) 37.1 (21–155) foraging locations were further away from the colony, again coinciding with the largest average range and longest trip lengths. The largest geographical range in foraging locations was found in 2015, taking into account that this was also the year with the largest number of recorded trips. Sandwich Terns travelled on average 67 ± 35.2 km (9–155) and 75% of all trips was between 20 and 80 km. Trip length and foraging range were significantly correlated (Spearmans r = 0.95, N = 65, P b 0,001) and as a consequence variation in trip length between years was found (Table 1, Fig. 2). 3.2. Foraging trip duration Average trip duration of birds with GPS-loggers was 134 ± 94 min (range: 20–417). Almost 75% of the trips took between 30 and 180 min (Fig. 3). Both trip length (linear regression r = 0.88, N = 65, P b 0,001) and foraging range from the colony (r = 0.76, N = 65, P b 0,001) were highly correlated with trip duration. The longest average trip durations for birds with GPS-loggers were found in 2015. Trip duration of Sandwich Terns recorded in the colonies was on average 105 ± 71 min (range: 5–398, N = 931) with also approximately 75% of the trips between 30 and 180 min. The longest average trip durations for birds in the colonies were found in 2014. Due to the differences in sample size, timing of sampling and the location of the colony, no interannual comparisons were performed for each method separately. In 2013 and 2015 significantly longer trips were found for birds with GPS-loggers compared to trip durations recorded in the colony, whereas in 2012 significantly shorter trips were found for birds with loggers (Mixed ANOVA F1,120 = 6.94, P b 0.01). Please cite this article as: Fijn, R.C., et al., GPS-tracking and colony observations reveal variation in offshore habitat use and foraging ecology of breeding Sandwich Terns, J. Sea Res. (2016), http://dx.doi.org/10.1016/j.seares.2016.11.005 4 R.C. Fijn et al. / Journal of Sea Research xxx (2016) xxx–xxx Fig. 1. Kernel Density Estimates calculated for foraging Sandwich Terns Thalasseus sandvicencis (21 birds, 154 trips) using location data collected in four consecutive breeding seasons with GPS-loggers on birds breeding in two different colonies (Scheelhoek: 2012, 2013 and 2015 [dot], Slijkplaat: 2014 [cross]). Bathymetry: © Copyright EMODnet Bathymetry 2016. 3.3. Time budgets during foraging trips 3.4. Geo-referenced prey items Of the complete trips (n = 65), birds spent an average of 32 ± 14% (range: 2–74) of the time in transit from the colony to the foraging areas, 36 ± 17% (range: 7–80) at foraging locations seemingly actively searching for prey for either self-provisioning or for the chick, and 28 ± 13% (range: 5–66) in transit back to the colony. The remaining 4 ± 11% (range: 0–59) was spent resting on sand banks and beaches (Fig. 4). Resting occurred mainly during longer trips (Fig. 4) and the proportion of a trip that was spent on resting increased significantly with increasing trip durations (Logistic regression t = 2.957, P b 0.01). No significant relationship was found between the foraging proportion of a trip and trip duration (t = 1.547, P = 0.13). Over all years combined, the chick diet of Sandwich Terns as recorded from the hide consisted of 76.8% Clupeidae, 20.1% Ammodytidae and 3.1% other species (N = 905). Out of these prey items, 12 fish were brought to the chicks by tagged birds (10 in 2012, 2 in 2013). Of these, ten were clupeids Clupeidae sp. with body lengths of 8 to 12 cm and two were sandeels Ammodytidae sp. of 16 cm. An approximate catch location could be determined from the GPS data for six of these fish (four clupeids, two sandeels), based on the last fix classified as ‘foraging’ within the trip. Three out of the four clupeids were small (8, 9 and 9 cm) and were caught relatively nearby (~10 km) on short trips (29 ± 16.1 min), whereas a larger clupeid (12 cm) was caught much further Please cite this article as: Fijn, R.C., et al., GPS-tracking and colony observations reveal variation in offshore habitat use and foraging ecology of breeding Sandwich Terns, J. Sea Res. (2016), http://dx.doi.org/10.1016/j.seares.2016.11.005 R.C. Fijn et al. / Journal of Sea Research xxx (2016) xxx–xxx 40 40 2012 30 30 20 20 10 10 0 40 2013 0 40 2014 30 30 20 20 10 10 5 2015 0 0 0-20 20-40 40-60 60-80 80-100 100-120 >120 km 0-20 20-40 40-60 60-80 80-100 100-120 >120 km 25 Frequency of occurrence 20 15 10 5 0 0-20 20-40 40-60 60-80 80-100 100-120 >120 km Trip length Fig. 2. Frequency distribution of trip length (km) of Sandwich Terns Thalasseus sandvicencis with GPS-loggers. away (N25 km away from colony) on a longer (incompletely recorded) trip (N61 min) (Fig. 5). Also, both (large) sandeels were caught further offshore (N 35 km from colony) during long trips (N150 min). The observations on prey delivery from the hide showed that average trip durations were significantly longer with increasing fish lengths for both Clupeidae (Logarithmic regression R2 = 0.89, N = 695, P b 0.001) and Ammodytidae (R2 = 0.69, N = 182, P b 0.01) (INBO unpublished data). 4. Discussion This study is the first to successfully map foraging tracks and at-sea distribution of 24 Sandwich Terns with GPS-loggers and reveals new insights in foraging behaviour and offshore feeding sites used by adults during chick rearing. We confirmed premature logger-loss for five birds. The fate of five other loggers that didn't transfer data to our base stations is unknown. Several explanations are possible; the five birds carrying these loggers might have lost their device away from the colony or the loggers encountered a technical failure. Also, these birds may have been mistakenly captured as breeders but were in fact non-breeding residents, or deserted their nest/chick, or otherwise failed to breed successfully. 4.1. Habitat use and home range Our study showed that Sandwich Terns used core foraging areas up to 40 km from the coast (60 km from the colony). Foraging ranges in our study were higher than previous estimates based on visual observations, VHF tracking, or boat-based tracking (e.g. Fasola and Bogliani, 1990 [Mediterranean, Italy], Stienen, 2006 [Dutch Wadden Sea], Perrow et al., 2011 [Norfolk, UK]), which were generally within 25 km from the colony. More locally, Baptist and Meininger (1984) estimated that foraging ranges of Sandwich Terns in our study area (in the currently deserted colony of Hompelvoet) were on average 25 km (maximum 35 km). This might indicate that Sandwich Terns in the Delta area have to travel further than conspecifics in other colonies to find appropriate food items, although it can also not be ruled out that some birds equipped with GPS-loggers performed longer foraging trips due to brood loss. In contrast to the generally reported shorter distances, some studies from the UK and Germany reported or estimated even larger maximum foraging ranges than in our study (up to 72 km, references in Langston, 2010, Eglinton and Perrow, 2014). The most often used compilation of data on seabird foraging ranges suggests an average range for Sandwich Terns of 11.5 ± 4.7 km, a mean maximum of 49 ± 7.1 km, and a maximum range of 54 km (Thaxter et al., 2012), and foraging ranges in our study were at the upper end of these estimates. This discrepancy can partly be explained by our colony location being approximately 6 km inshore (and 11 km in 2014), but is most likely due to the difference in local food availability between the Netherlands and the UK. Surprisingly, no apparent differences in core foraging habitat nor in home range size and trip lengths or durations were found between 2014, when the colony was located 5 km further inshore, and 2013 or 2015. Assuming that food availability is broadly similar Please cite this article as: Fijn, R.C., et al., GPS-tracking and colony observations reveal variation in offshore habitat use and foraging ecology of breeding Sandwich Terns, J. Sea Res. (2016), http://dx.doi.org/10.1016/j.seares.2016.11.005 R.C. Fijn et al. / Journal of Sea Research xxx (2016) xxx–xxx 2013 10 0 0 50 50 10 80 -2 20 -1 -1 0 12 0 40 m in -2 21 0 >2 40 80 -2 -1 15 0 18 0 20 -1 -1 90 12 0 -6 0 60 -9 0 0 90 0 30 5 birds, 26 trips 28 birds, 380 trips -3 0 10 10 10 50 20 -9 0 20 -3 0 30 30 -6 0 25 birds, 172 trips 40 60 40 0 2015 5 birds, 16 trips m in 10 40 20 >2 40 20 -2 30 21 0 30 2014 2 birds, 11 trips 28 birds, 206 trips 40 18 0 40 50 50 4 birds, 12 trips 28 birds, 380 trips -1 2012 15 0 50 30 6 30 Frequency (% of all trips) GPS-logger trips (16 birds, 65 trips) Colony observations (107 birds, 936 trips) 25 20 15 10 5 0 0 - 30 30 - 60 60 - 90 90 - 120 120 - 150 150 - 180 180 - 210 210 - 240 >240 min Trip duration Fig. 3. Frequency distribution of trip duration (min) of Sandwich Terns Thalasseus sandvicencis with GPS-loggers (black bars) and observed in the colony (grey bars). between those three years, this indicates that such an additional distance does not seem to have major effects on foraging location choice. No longer trip durations of birds with GPS-loggers were found in Commute towards colony Resting 2014, which is most likely due to the large variation in trip durations in the GPS-data combined with the relatively short additional flight time (average flight speeds during commutes were approximately Foraging/searching Commute out of colony 100% Percentage of trip time budget 80% 60% 40% 20% 0% 0 - 60 60 - 120 120 - 180 180 - 240 Trip duration (min) > 240 min Average Fig. 4. Time budgets for complete trips of Sandwich Terns classified in 5 trip durations and an average time budget for all trips. Please cite this article as: Fijn, R.C., et al., GPS-tracking and colony observations reveal variation in offshore habitat use and foraging ecology of breeding Sandwich Terns, J. Sea Res. (2016), http://dx.doi.org/10.1016/j.seares.2016.11.005 R.C. Fijn et al. / Journal of Sea Research xxx (2016) xxx–xxx Fig. 5. Six catch locations of identified prey items for chisk of Sandwich Terns with GPSloggers in 2012 (clupeids) and 2013 (sandeels). Within the fish sizes are to scale. Bathymetry: © Copyright EMODnet Bathymetry 2016. 35 km/h corresponding to 17 min for 10 km additional flying for an average trip duration of 134 ± 95 min). Nevertheless, longer trip durations were found during colony observations, with durations in 2014 being the longest of all years (Table 1). Stienen et al. (2015) suggested that such an increased investment in foraging time is compensated by increasing non-attendance of the chick, resulting in stable provisioning rates and thus no effects on survival and breeding success, which is reflected in the breeding success of 2014, which was the highest of all four breeding seasons (INBO unpublished data). In all years, breeding success in a reference part of the colony was relatively good when compared to other nearby breeding locations and previous breeding seasons (INBO unpublished data). 4.2. Year-to-year variation in foraging range Substantial year-to-year variation in foraging range was obvious in our data (Kruskal-Wallis χ23 = 24.95, P b 0.001) however, due to various reasons these interannual comparisons in habitat use are difficult to interpret. Firstly, prey availability and local feeding conditions around colonies are known to be a driving force behind variation in foraging ranges, as has been previously shown in two neighbouring colonies of Caspian Terns Hydroprogne caspia (Patterson, 2012), a fairly similar species of tern in terms of size and foraging strategy. However, since food availability varies between years in our study area (Tulp et al., 2014, Tien et al., this volume) and within years due to tide (Couperus et al., 2016) and weather (Thorpe, 1978, own observations), and given our relatively small sample size, interannual comparisons are difficult to ascertain. Secondly, our GPS-sampling varied within seasons with samples taken during different chick ages. Colony-based research however, showed that trip durations (and most likely the correlated trip lengths and maximum range) increase with increasing chick age due to the fact that older chicks need (and are able to handle) larger prey items (e.g. Stienen et al., 2000, Stienen, 2006 and references herein). In line with these previous findings also our data from the hide resulted in a significant positive relationship between average prey length and trip duration (INBO unpublished data). Results from fish monitoring in our study area suggest that larger individuals of the most important prey items of Sandwich Terns (Ammodytidae sp. and Clupeidae sp.) reside further from the coast compared to smaller fish (Bureau Waardenburg unpublished data, Tulp et al., 2014), and furthermore many Sandwich Terns with large prey items were sighted well offshore during aerial and boat-based surveys (Bureau Waardenburg observations). This all suggests that Sandwich Terns likely need to increase their foraging 7 range during the course of the season in order to catch larger prey items. This was also confirmed, although based on a limited sample size, in our study (Fig. 5.), and such a gradual shift in prey choice and thus trip duration over the season makes interannual comparisons challenging. Moreover, prey availability is also likely to be influenced by abiotic factors, such as tide, as was shown in a nearby and similar habitat (Couperus et al., 2016). Also weather conditions can influence the availability of prey, as was seen in our study in 2013 for example, when cold spring weather seemed to have delayed the arrival of available (to the Sandwich Terns) clupeids in our study area, which directly influenced diet choice and foraging behaviour (INBO unpublished data). Intra-annual variation in foraging ranges between and within individual birds would be a key subject for further study. At present, examining the spatial and temporal differences between foraging locations of individual Sandwich Terns, or the role of factors such as tide, weather, timing, etc. on trip duration, suggests some relationship, but only by further combining colony observations and GPS-deployments might these gaps be addressed. 4.3. Variation in trip durations between methods Trip duration of breeding Sandwich Terns has been previously extensively studied, but only through direct observations of marked individuals in the colonies (Stienen et al., 2000) or with boat-based tracking (Perrow et al., 2011). Boat-based tracking yields data similar to GPS-tracking, but our trip durations were longer than those found by Perrow et al., 2011. This is partly due to the generally more inland location of our study colony (between 7 and 12 km) compared to the colonies studied by Perrow et al. (2011) that are directly on the coast, but also due to a bias towards tracking mainly shorter trips with boats, compared to tracking all trips with GPS-loggers. Trip durations recorded with GPS-loggers were generally slightly longer than those found during observations in the colonies; a reason for which could be the presence of the logger. Previous research suggests several effects of the deployment of loggers to various bird species, including reduced foraging efficiency resulting in longer trips (e.g. Vandenabeele et al., 2011b). Average trip durations varied between years for both methods but these results are difficult to interpret statistically, as variation is extremely large and the mechanisms behind this variation are diverse. Similar to the mechanisms driving foraging range (§4.2), specific dietary needs of the chicks or the adults, and changes in prey availability due to tide and weather can also influence trip durations. However, weather can also strongly impact trip duration via inhibited foraging efficiency during adverse weather conditions (e.g. Stienen et al., 2000 and references herein, Baptist and Leopold, 2010). Were any effects of the physical presence of the loggers present, these might be accentuated during adverse weather conditions, such as higher wind speeds. Factors such as e.g. water temperature, precipitation and wind speed are also known to vary greatly between years and we were therefore unable to examine these in further detail with regards to interannual variation. Variables such as trip length, foraging range and time budgets cannot be studied with colony-based observations. On the other hand, colony-based observations do allow data on prey delivery to the chicks and information on prey choice to be collected. A combination of colonybased observations and GPS-loggers is needed in order for foraging data and provisioning data to be linked (see Section 4.4). Our study shows that variation between individuals and within seasons is large and that this clouds interannual variation in studies with small sample sizes. Increasing sample sizes in logger research, or concentrating on single breeding stages (e.g. chicks) potentially reduces this variation and leads to better interpretation of the results. 4.4. Time budgets during foraging trips In contrast to numerous studies on seabird species that regurgitate food for their chicks (e.g. Ropert-Coudert et al., 2004; Chivers et al., Please cite this article as: Fijn, R.C., et al., GPS-tracking and colony observations reveal variation in offshore habitat use and foraging ecology of breeding Sandwich Terns, J. Sea Res. (2016), http://dx.doi.org/10.1016/j.seares.2016.11.005 8 R.C. Fijn et al. / Journal of Sea Research xxx (2016) xxx–xxx 2012), activity budgets of single-prey loaders are merely limited to auk species (e.g. Uttley et al., 1994; Tremblay et al., 2003), and time budgets of breeding terns have only been studied once before (Patterson, 2012). Our results suggest that approximately 35% of a foraging trip is spent on active diving and searching for food for self-provisioning and for chick feeding, which is less than the 40–70% found for Caspian Tern (Patterson, 2012). Since Caspian Terns forage closer to the colony, commuting time is much shorter compared to that of Sandwich Terns (10– 25% versus 60% for Sandwich Terns) and leaving a larger part of the trip for foraging and resting (20–35% vs. 4% respectively). Similarly the proportion foraging during a trip is expected to be larger for Sandwich Terns breeding nearer to the foraging site (e.g. UK colonies) than those more inland (our study). Flexibility in time budgets of foraging activities has been found for Sandwich Terns (Stienen et al., 2015). The absence of a significant positive relationship in our GPS data between foraging proportion and trip duration (t = 1.547, P = 0.13), in combination with high correlations between trip duration, trip length and foraging range, suggest that an increase in trip duration is not caused by increasing foraging duration at nearby locations but results from foraging locations that are further from the colony. This is supported by the fact that Sandwich Terns not only commuted longer with increasing trip durations, but also foraged for longer on these trips. As trip duration and foraging range are positively correlated, this means that it takes longer to search for suitable (probably larger) prey-items further from the colony (and thus further offshore, see also Section 4.5). Far-ranging trips thus ask for investment in both commuting and foraging time, which suggest that larger prey items, that might live further offshore, are more difficult to catch than other (randomly selected) prey items. As a consequence, adult Sandwich Terns have to deal with both increasing energetic and time expenditures with increasing chick age, which in turn reflected in increasing non-attendance of the chicks (c.f. Stienen et al., 2015). 4.5. Geo-referenced prey items Although sample sizes were small, our results suggest that Sandwich Terns can be used as a ‘sampling’ tool to map offshore prey distribution. These results were merely opportunistically collected as most of our GPS-deployments were on birds outside the enclosures. Also small sample sizes were the results of a combination of small battery life of the loggers and limited opportunity to do continuous feeding protocols. A continuously recording and remotely operated camera would improve the output of such a set-up considerably without disturbance to the colony. 4.6. Conclusion Sandwich Terns in the southwestern part of the Netherlands proved to be highly specialized foragers with relatively large foraging ranges compared to conspecifics from other colonies, yet also substantial variability in foraging locations was found. This might indicate the availability and exploitation of wide-ranging prey-items, but our data suggest a more complex foraging strategy exploiting nearby foraging locations to catch smaller prey items for self-provisioning or to feed young chicks and farther offshore locations to catch suitable (larger) prey items for their older chicks. Drivers behind this strategy remained unknown probably due to our relatively small sample sizes, but there are indications that the (inland) location of the colony, as well as food availability, chick age, or even weather and tide conditions play a role. Also the question whether this foraging strategy is consistent and repeatable within and between years was beyond the scope of the current study but would be a prerequisite to unravel the foraging strategy of our Sandwich Terns. Acknowledgements This study was part of the monitoring programme into the effects of the compensation measures designed for the construction of the seaward expansion of the Rotterdam Harbour (‘Tweede Maasvlakte’). This programme (PMR-NCV) was initiated by the Dutch Ministry of Infrastructure and the Environment and commissioned by Deltares (G. van der Kolff, T. Prins, A. Boon and J. Reijnders) and Rijkswaterstaat WVL (M. van Eerden and K. Borst). Bureau Waardenburg, INBO Research Institute for Nature and Forest, and Delta Project Management carried out the research on terns within this project in a consortium with a lead of IMARES Wageningen-UR (I. Tulp and H. Heessen). Tracking of Sandwich Terns was performed under project licence for animal procedures AVD401002015102 of the Central Authority for Scientific Procedures on Animals (CCD). The authors would like to thank M. Van de Walle, N. Vanermen and P. Wolf for help in the field. Fieldwork was carried in nature reserves of Natuurmonumenten (NM) and Staatsbosbeheer (SBB) and W. van Steenis, H. Meerman, J. de Roon, M. Broere (NM) and R. in ‘t Veld, N. de Bruin, A. Wesdorp (SBB) are thanked for their cooperation and hospitality. We thank T. Boudewijn, M. Collier (both Bureau Waardenburg) and two anonymous reviewers for their input to improve the manuscript. References Baptist, H.J.M., Meininger, P.L., 1984. Ornithologische verkenning van de Voordelta van Zuidwest-Nederland, 1975–1983. Rijkswaterstaat Deltadienst Hoofdafdeling Milieu en Inrichting (In Dutch). Baptist, M.J., Leopold, M.F., 2010. Prey capture success of sandwich terns Sterna sandvicensis varies non-linearly with water transparency. Ibis 152, 815–825. Barrett, R.T., Camphuysen, C.J., Anker-Nilssen, T., Chardine, J.W., Furness, R.W., Garthe, S., Huppop, O., Leopold, M.F., Montevecchi, W.A., Veit, R.R., 2007. Diet studies of seabirds: a review and recommendations. ICES J. Mar. Sci. 64, 1675–1691. Chivers, L.S., Lundy, M.G., Colhoun, K., Newton, S.F., Houghton, J.D.R., Reid, N., 2012. Foraging trip time-activity budgets and reproductive success in the black-legged kittiwake. Mar. Ecol. Prog. Ser. 456, 269–277. Cooke, S.J., Hinch, S.G., Wikelski, M., Andrews, R.D., Kuchel, L.J., Wolcott, T.G., Butler, P.J., 2004. Biotelemetry: a mechanistic approach to ecology. Trends Ecol. Evol. 19, 334–343. Couperus, B., Gastauer, S., Fässler, S.M.M., Tulp, I., van der Veer, H.W., Poos, J., Poos, J., 2016. Abundance and tidal behaviour of pelagic fish in the gateway to the Wadden Sea. J. Sea Res. 109, 42–51. Courtens, W., Verstraete, H., Vanermen, N., van de Walle, M., Stienen, E.W.M., Faecal samples reveal the diet of breeding adult Sandwich Terns in Belgium and the southern part of the Netherlands. J. Sea Res. this volume. Dies, J.I., Dies, B., 2005. Kleptoparasitism and host response in a sandwich tern colony in eastern Spain. Waterbirds 28, 167–171. Eglinton, S.M., Perrow, M.R., 2014. Literature Review of Tern (Sterna & Sternula spp.) Foraging Ecology. ECON Ecological Consultancy Ltd. (Contract ref. C13-0204-0686, Report to JNCC). Fasola, M., Bogliani, G., 1990. Foraging ranges of an assemblage of Mediterranean seabirds. Colon. Waterbirds 72–74. Fijn, R.C., Wolf, P.A., Courtens, W., Verstraete, H., Stienen, E.W.M., Iliszko, L., Poot, M.J.M., 2014. Post-breeding prospecting flights of adult sandwich terns Thalasseus sandvicensis. Bird Study 61, 566–571. Fijn, R.C., Collier, M.P., Gyimesi, A., Clark, N., Poot, M.J.M., Experiences using a new type of short-term harness to track Sandwich Terns. Thalasseus sandvicensis with GPSloggers, in preparation. Garthe, S., Kubetzki, U., 1998. Diet of sandwich terns Sterna sandvicensis on Juist (Germany). Sula 12, 13–19. Gremillet, D., Dell'Omo, G., Ryan, P.G., Peters, G., Ropert-Coudert, Y., Weeks, S.J., 2004. Offshore diplomacy, or how seabirds mitigate intra-specific competition: a case study based on GPS tracking of Cape gannets from neighbouring colonies. Mar. Ecol. Prog. Ser. 268, 265–279. Gyimesi, A., Boudewijn, T.J., Poot, M.J.M., Buijs, R.J., 2011. Habitat use, feeding ecology and reproductive success of Lesser black-baked gulls breeding in Lake Volkerak. Bureau Waardenburg Report 10-234, Culemborg, the Netherlands. Kernohan, B.J., Gitzen, R.A., Millspaugh, J.J., 2001. Analysis of animal space use and movements. In: Millspaugh, J.J., Marzluff, J.M. (Eds.), Radiotracking and Animal Populations. Academic, San Diego, California, USA, pp. 125–166. Kenward, R.E., 1985. Raptor radio-tracking and telemetry. In: Newton, I., Chancellor, R.D. (Eds.), Conservation Studies on Raptors. I.C.B.P. 5, pp. 409–420 Technical Publication. Kotzerka, J., Garthe, S., Hatch, S.A., 2010. GPS tracking devices reveal foraging strategies of black-legged kittiwakes. J. Ornithol. 151, 459–467. Langston, R.H.W., 2010. Offshore wind farms and birds: Round 3 zones, extension to Round 1 and Round 2 sites and Scottish Territorial waters. RSPB Research Report No 39, Sandy, UK. Patterson, A., 2012. Breeding and Foraging Ecology of Caspian Terns Nesting on Artificial Islands in the Upper Klamath Basin, California. MSc. thesis. Oregon State University. Please cite this article as: Fijn, R.C., et al., GPS-tracking and colony observations reveal variation in offshore habitat use and foraging ecology of breeding Sandwich Terns, J. Sea Res. (2016), http://dx.doi.org/10.1016/j.seares.2016.11.005 R.C. Fijn et al. / Journal of Sea Research xxx (2016) xxx–xxx Perrow, M.R., Skeate, E.R., Gilroy, J.J., 2011. Visual tracking from a rigid-hulled inflatable boat to determine foraging movements of breeding terns. J. Field Ornithol. 82, 68–79. Phillips, R.A., Xavier, J.C., Croxall, J.P., 2003. Effects of satellite transmitters on albatrosses and petrels. Auk 120, 1082–1090. Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., EISPACK, Heisterkamp, S., van Willigen, B., R Core Team, 2015. nlme: linear and Nonlinear mixed effect models. R Package Version 3, pp. 1–124. Pohlert, T., 2014. The Pariwise Multiple Comparison of Mean Ranks Package (PMCMR) (R package). Poot, M.J.M., et al., 2014. Perceel Vogels – Grote stern en visdief. In: Prins, T.C., van der Kolff, G.H. (Eds.), PMR Monitoring natuurcompensatie Voordelta: eindrapport 1e fase 2009–2013 deel B. Deltares rapport 1200672-000-ZKS-0043. http:// kennisonline.deltares.nl/product/30737 (In Dutch). Core Team, R., 2015. R: A Language and Environment for Statistical Computing. R foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org/. Ropert-Coudert, Y., Gremillet, D., Kato, A., Ryan, P.G., Naito, Y., Le Maho, Y., 2004. A finescale time budget of cape gannets provides insights into the foraging strategies of coastal seabirds. Anim. Behav. 67, 985–992. Shamoun-Baranes, J., Bouten, W., Camphuysen, C.J., Baaij, E., 2011. Riding the tide: Intriguing observations of gulls resting at sea during breeding. Ibis 153, 411–415. Stienen, E.W.M., 2006. Living with gulls: trading off food and predation in the Sandwich Tern Sterna sandvicensis. Alterra Scientific Contributions 15. Rijksuniversiteit Groningen, Groningen, The Netherlands PhD dissertation. Stienen, E.W.M., van Beers, P.W.M., Brenninkmeijer, A., Habraken, J.M.P.M., Raaijmakers, M.H.J.E., van Tienen, P.G.M., 2000. Reflections of a specialist: patterns in food provisioning and foraging conditions in Sandwich Terns Sterna sandvicensis. Ardea 88, 33–88. Stienen, E.W.M., Brenninkmeijer, A., Courtens, W., 2015. Intra-specific plasticity in parental investment in a long-lived single-prey loader. J. Ornithol. 156, 699–710. Thaxter, C.B., Lascelles, B., Sugar, K., Cook, A.S.C.P., Roos, S., Bolton, M., Langston, R.H.W., Burton, N.H.K., 2012. Seabird foraging ranges as a preliminary tool for identifying candidate marine protected areas. Biol. Conserv. 156, 53–61. 9 Thorpe, J.E., 1978. Rhythmic Activity in Fishes. Academic Press, London. Tien, N.S.H., Craeymeersch, J., van Damme, C., Couperus, A.S., Adema, J., Tulp, I., Burrow distribution of three sandeel species relates to sediment composition, water velocity and beam trawl fishing, in Dutch coastal waters. J. Sea Res. this volume. Tremblay, Y., Cherel, Y., Oremus, M., Tveraa, T., Chastel, O., 2003. Unconventional ventral attachment of time-depth recorders as new method for investigating time budget and diving behaviour of seabirds. J. Exp. Biol. 206, 1929–1940. Tulp, I.Y.M., et al., 2014. Perceel Vis. In: Prins, T.C., van der Kolff, G.H. (Eds.), PMR Monitoring natuurcompensatie Voordelta: eindrapport 1e fase 2009–2013 deel B. Deltares rapport 1200672-000-ZKS-0043 . http://kennisonline.deltares.nl/product/30737 (In Dutch). Uttley, J.D., Walton, P., Monaghan, P., Austin, G., 1994. The effects of food abundance on breeding performance and adult time budgets of guillemots Uria aalge. Ibis 136, 205–213. Vandenabeele, S.P., Shepard, E.L., Grogan, A., Wilson, R.P., 2011a. When thee per cent may not be three per cent; device-equipped seabirds experience variable flight constraints. Mar. Biol. 159, 1–14. Vandenabeele, S.P., Wilson, R.P., Grogan, A., 2011b. Tags on seabirds: how seriously are instrument-induced behaviours considered? Anim. Welf. 20, 559–571. Weimerskirch, H., Bonadonna, F., Bailleul, F., Mabille, G., Dell'Omo, G., Lipp, H.-P., 2002. GPS tracking of foraging albatrosses. Science 295, 1259. Wickham, H., 2009. ggplot2: Elegant Graphics for Data Analysis. Springer, New York. Wilson, R.P., Pütz, K., Peters, G., Culik, B., Scolaro, J.A., Charrassin, J.-B., Ropert-Coudert, Y., 1997. Long-term attachment of transmitting and recording devices to penguins and other seabirds. Wildl. Soc. Bull. 25, 101–106. Wilson, R.P., Gremillet, D., Syder, J., Kierspel, M.A.M., Garthe, S., Weimerskirch, H., SchaferNeth, C., Scolaro, J.A., Bost, C.-A., Plotz, J., Nel, D., 2002. Remote-sensing systems and seabirds: Their use, abuse and potential for measuring marine environmental variables. Mar. Ecol. Prog. Ser. 228, 241–261. Worton, B.J., 1989. Kernel methods for estimating the utilization distribution in homerange studies. Ecology 70, 164–168. Please cite this article as: Fijn, R.C., et al., GPS-tracking and colony observations reveal variation in offshore habitat use and foraging ecology of breeding Sandwich Terns, J. Sea Res. (2016), http://dx.doi.org/10.1016/j.seares.2016.11.005