Direct Evidence of Swimming Demonstrates Active Dispersal in the Sea Turtle Lost Years

Transcription

1 Report Direct Evidence of Swimming Demonstrates Active Dispersal in the Sea Turtle Lost Years Highlights d We developed the first direct test of the passive migration hypothesis in sea turtles Authors Nathan F. Putman, Katherine L. Mansfield d d d Drifter tracks diverged substantially from those of small turtles Analyses unequivocally demonstrate that swimming influences sea turtle distributions Location-dependent and species-specific turtle behavior were observed Correspondence nathan.putman@gmail.com In Brief The role of ocean currents in the dispersal of marine animals is well known, but the contribution of swimming behavior is poorly understood. Putman and Mansfield show that oriented swimming by young sea turtles in the wild substantially influences their movement compared to what would be expected based on ocean circulation processes alone. Putman & Mansfield, 2015, Current Biology 25, May 4, 2015 ª2015 Elsevier Ltd All rights reserved

2 Current Biology Report Direct Evidence of Swimming Demonstrates Active Dispersal in the Sea Turtle Lost Years Nathan F. Putman 1, * and Katherine L. Mansfield 1,2 1 Southeast Fisheries Science Center, National Marine Fisheries Service, 75 Virginia Beach Drive, Miami, FL 33149, USA 2 Department of Biology, University of Central Florida, Orlando, FL 32816, USA *Correspondence: nathan.putman@gmail.com SUMMARY Although oceanic dispersal in larval and juvenile marine animals is widely studied, the relative contributions of swimming behavior and ocean currents to movements and distribution are poorly understood [1 4]. The sea turtle lost years [5] (often referred to as the surface-pelagic [6] or oceanic [7] stage) are a classic example. Upon hatching, young turtles migrate offshore and are rarely observed until they return to coastal waters as larger juveniles [5]. Sightings of small turtles downcurrent of nesting beaches and in association with drifting organisms (e.g., Sargassum algae) led to this stage being described as a passive migration during which turtles movements are dictated by ocean currents [5 10]. However, laboratory and modeling studies suggest that dispersal trajectories might also be shaped by oriented swimming [11 15]. Here, we use an experimental approach designed to directly test the passive-migration hypothesis by deploying pairs of surface drifters alongside small green (Chelonia mydas) and Kemp s ridley (Lepidochelys kempii) wild-caught turtles, tracking their movements via satellite telemetry. We conclusively demonstrate that these turtles do not behave as passive drifters. In nearly all cases, drifter trajectories were uncharacteristic of turtle trajectories. Speciesspecific and location-dependent oriented swimming behavior, inferred by subtracting track velocity from modeled ocean velocity, contributed substantially to individual movement and distribution. These findings highlight the importance of in situ observations for depicting the dispersal of weakly swimming animals. Such observations, paired with information on the mechanisms of orientation, will likely allow for more accurate predictions of the ecological and evolutionary processes shaped by animal movement. RESULTS AND DISCUSSION We combined in situ and modeling approaches to test the passive-migration hypothesis in wild-caught, surface-pelagic turtles [6]. Green (Chelonia mydas; n = 24) and Kemp s ridley (Lepidochelys kempii; n = 20) turtles were tracked by satellite telemetry [16, 17] in the northern and eastern Gulf of Mexico along with two types of simultaneously deployed surface drifters (Figure 1). We directly assessed whether these small turtles ( cm straight carapace length [SCL]) drifted passively by comparing separation distances between turtles and drifters to separation distances between the drifter pairs. If the movement of turtles were primarily the result of ocean circulation processes, we would expect that separation distances would be similar among groups. However, separation distances between turtles and drifters were significantly greater than separation distances between pairs of drifters (Wilcoxon signed-rank test, p < 0.034, for days 2 14) (Figures 2A and 2B), indicating that turtle movement is not a passive process. We then used high-resolution ocean circulation model output [18] and virtual particle-tracking software [19] to extract behavior from turtle tracks. Apparent swimming velocity was derived by subtracting modeled ocean velocity from track velocity [20 22]at 2-day steps along the tracks. This common approach attributes any difference between the modeled velocity and track velocity to swimming velocity [22]. Though the model characterized the oceanic conditions reasonably well (Figure 2C), by tracking drifters and turtles simultaneously we could consider how the model s inability to fully resolve ocean velocity influences our interpretations of swimming behavior (Figure 3). For instance, the direction of residual velocity in drifters was typically westward in latitudes north of 27.5 N(Figure 3H). Owing to this systematic bias, the true swimming speed of turtles swimming westward would likely be less than calculated (or attributable to passive drift), whereas the true swimming speed of turtles swimming eastward would actually be greater than calculated. The mean headings of both green and Kemp s ridley turtles were eastward in this region, suggesting that contributions of turtle swimming behavior to net movement may be underestimated across this area (Figures 3B and 3E). In latitudes south of 27.5 N, the direction of residual velocity in drifters was fairly random (Figure 3I). In this case, directional swimming from turtles would not likely be an artifact introduced by the model, but rather a true orientation preference. In these more southern latitudes, green turtles continued to orient eastward, whereas Kemp s ridley turtles shifted to northward swimming (Figures 3C and 3F). These calculated swimming directions suggest active orientation by turtles, rather than bias introduced by our analytical methods. If the movement of turtles were primarily the result of ocean circulation processes, we would expect swimming speeds to be similar for drifters and turtles. Calculated swim speeds did not differ between the two drifter types (Mann-Whitney U test, Current Biology 25, , May 4, 2015 ª2015 Elsevier Ltd All rights reserved 1221

3 Figure 1. Tracks of Turtles, Drifters, and Virtual Particles Released from the Same Locations Each panel depicts the paths traveled by drifters (blue), virtual particles tracked within hindcast ocean circulation model output (purple), green turtles (green), and Kemp s ridley turtles (red) released on the same date and location (white circles). Gray shading indicates bathymetry, and the thin gray line delineates the continental shelf (200 m isobath). Further details can be found in Table S1, and plots for all release events can be found in Figure S1. (A) A Kemp s ridley turtle tagged on August 13, 2011 maintained position within a restricted area along the West Florida Shelf, whereas drifters and particles traveled north and off the shelf. (B) A green turtle tagged on October 20, 2012 traveled south-southwest, whereas drifters and particles moved east-southeast. (C) Green turtles tagged on May 19, 2014 traveled predominantly eastward (n = 3). One of the turtles traveled north and west, similar to drifters. Virtual particles drifted north and beached along the Louisiana coast. (D) Turtles tagged on May 21, 2014 generally moved to the east, with two Kemp s ridley turtles trending southward before turning north. Drifters traveled north and beached on the Louisiana coast, as did many virtual particles. Some virtual particles traveled east and south but continued out of the Gulf of Mexico without turning back northward. U 211,162 = 17976, p = 0.388) and thus were pooled for comparison to turtles. Swim speeds of green turtles (median = m/s, n = 221 steps) and Kemp s ridley turtles (median = m/s, n = 263 steps) were faster than those of drifters (median = m/s, n = 375 steps) (green versus drifter: Mann-Whitney U test, U 374,221 = 49933, p = ; Kemp s ridley versus drifter: Mann-Whitney U test, U 374,263 = 5376, p = 0.048), further indicating that turtles were actively swimming. Moreover, swim speeds of both species were at the lower range of speeds reported for similar size and stage loggerhead turtles (Caretta caretta) (central tendencies range from 0.15 m/s to 0.3 m/s [23 25]), which implies that other turtle species could possess similar control over their movements. This method of simultaneously tracking organisms, drifters, and virtual particles [26] provides direct, environmental context for dispersal trajectories and produces new insights on turtle behavior in the open sea. Our findings support preliminary reports that young green turtles are more active than Kemp s ridley turtles [6]. Swim speeds of green turtles were faster than Kemp s ridley turtles (Mann-Whitney U test, U 263,221 = 32739, p = 0.016). Likewise, the swimming orientation of green turtles (median of individual s Rayleigh r value = 0.46) tended to be more directional than Kemp s ridley turtles (median of individual s Rayleigh r value = 0.283) (Mann-Whitney U test, U 20,24 = 297, p = 0.04). The track headings of individual green turtles were better predicted by calculated swimming orientation (median circular-circular correlation r = 0.403) than by the direction of modeled ocean currents (median circular-circular correlation r = 0.112) (Wilcoxon signedrank test, W = 80, p = 0.045, n = 24 turtles). For Kemp s ridley turtles, track headings were predicted equally well by calculated orientation (median circular-circular correlation r = 0.222) and the direction of modeled ocean currents (median circular-circular correlation r = 0.208) (Wilcoxon signed rank test, W = 71, p = 0.528, n = 20 turtles). Considering that both species were caught in the same areas, large-scale differences in distribution might be mediated primarily by species-specific behavior (Figure 3), rather than by ocean conditions encountered. The consistently oriented swimming in green turtles likely facilitates the long-distance movements known in this species [27, 28], whereas Kemp s ridley turtles behavior appears to promote retention in their primary range within the Gulf of Mexico [29 31]. Contrary to expectations [23], we detected no relationship between turtle size (SCL) and swimming speeds (median or maximum) (Spearman R < 0.271, p > , n = 24 for green, n = 20 for Kemp s ridley). We suspect that while swimming ability increases as turtles grow [23], different environmental and endogenous factors influence swimming activity and thus the swim speeds inferred along the tracks of turtles. Surprisingly, one of the environmental factors most likely to influence swimming speed and activity, water temperature [32] (which ranged from 21 Cto31 C along the tracks of turtles), had limited predictive value (green Spearman R = 0.006, p = 0.917, n = 235 steps; Kemp s ridley Spearman R = 0.074, p = 0.190, n = 312 steps). It is possible that conditions associated with microhabitat (e.g., the presence of floating algae or the density of prey items) [6, 8] might obscure a relationship between swimming speed and temperature that would otherwise be observed, or that physiological performance can be maintained over a relatively wide range of temperatures if changes are gradual and acclimation can occur [23, 24, 33]. It was recently reported that young, lab-reared loggerhead turtles orient their movement into current flows (i.e., are rheotactic) upon release into the ocean [25]. For the turtles that we tracked, circular-circular correlations between swimming direction and current direction were extremely weak but tended to be positive (Kemp s ridley: r = 0.071, p > 0.05, n = 312 steps; green: r = 0.033, p > 0.05, 1222 Current Biology 25, , May 4, 2015 ª2015 Elsevier Ltd All rights reserved

4 n = 235 steps) and thus are inconsistent with a hypothesis of rheotaxis as a ubiquitous explanation for orientation in the open sea [25]. The lack of relationship between swimming speed and endogenous (body size) and environmental (water temperature, current direction) variables that have traditionally been assumed to be important in the locomotion process of marine animals [5 9, 25, 26, 32, 33] implies that some aspects of sea turtle behavior and movement should be reconsidered. Our findings may be useful in designing laboratory and field experiments to clarify the correlates of turtle swimming velocity as well as the sensory cues used by turtles to navigate in the wild [11, 12, 34, 35]. Given (1) the major divergence between the trajectories of turtles and the trajectories of drifters and particles (Figures 1 and 2), (2) faster swim speeds of turtles compared to drifters, and (3) our finding that swimming orientation of turtles differed dramatically from what would be expected if they drifted with ocean currents (Figure 3), it is clear that surface-pelagic stage turtles actively control their movements. Indeed, it appears that even at the smallest sizes, swimming has important influences on subsequent distribution and fitness of turtles [14, 36]. How, then, should one interpret the reported agreement between models assuming passive drift and observed distributions (e.g., [27, 37, 38])? In part, the congruence in previous studies is likely due to sparse in situ data available for turtles, resulting in wide confidence intervals for observations that will often overlap with model predictions [39]. Additionally, the temporal and spatial scales over which dispersal is examined in models are broader than in our tracking data. For instance, plotting the relative density of turtle, drifter, and virtual particle location data suggests some broadscale agreement in distributions (Figure 4). Specifically, virtual particle distribution overlaps with most of the dispersal pathways observed in green and Kemp s ridley turtles (Figure 4). However, our results suggest that differences in relative density between passive particles and turtles are attributable to species-specific swimming behavior. This is consistent with an earlier modeling study suggesting that currents largely dictate which pathways are available to juvenile sea turtles but that magnetic navigation behavior influences the proportion of turtles following particular dispersal pathways [13]. Thus, numerical models simulating passive drift in turtles are not without merit for the production of a null hypothesis of oceanic distribution [27], but we now clearly demonstrate that swimming behavior is an important component for explaining how oceanic-stage turtles achieve observed distributions. In particular, our results show that directional swimming Figure 2. Separation Distances for Turtles, Drifters, and Virtual Particles (A) The separation distance (y axis) over a 14-day period (x axis) between green turtles and drifters (dashed green line = median; solid green line = mean; thin lines = 95% confidence interval). For context, the separation distances between pairs of drifters are also plotted (blue lines, following the same conventions as described above, are displayed in each panel). Values in parentheses show the number of comparisons for calculations (number of turtleto-drifter comparisons on the left; number of drifter-to-drifter comparisons on the right; these values decrease over time due to tag loss). (B) Conventions as in (A), but with separation distances between Kemp s ridley turtles and drifters (red lines). Statistical differences were found between the separation distances of both turtle species and drifters (green: Wilcoxon signed-rank test, p < , for each day; Kemp s ridley: Wilcoxon signedrank test, p < 0.034, for each day), indicating that turtle movement is not a passive process. (C) Conventions as in (A), but with separation distances between virtual particles and drifters (purple). Separation distances between virtual particles and drifters were initially greater than for drifter pairs (days 2 and 4, Wilcoxon signed-rank test, p < 0.01) but were otherwise statistically indistinguishable (Wilcoxon signed-rank test, 0.07 < p < 0.6). Current Biology 25, , May 4, 2015 ª2015 Elsevier Ltd All rights reserved 1223

5 Figure 3. Location-Dependent Orientation Behavior Swimming orientation was inferred by subtracting modeled ocean velocity from track velocity between successive points along 24 green turtle tracks (A C), 20 Kemp s ridley turtle tracks (D F), and 26 surface drifter tracks (G I) (see Figure S2 for an example). (A, D, and G) Mean direction of residual velocity (swimming orientation) within 2 latitude 3 2 longitude bins (arrows), plotted over tracks. Arrows size is scaled relative to the circular standard deviation of the mean (larger arrows = smaller standard deviation); arrow coloration is scaled to the number of track steps within each bin (bins with less than two track steps are not plotted). (B, C, E, F, H, and I) Circular histograms in which length of bars is proportional to the number of track segments where swimming was oriented within a given 15 range. The outer triangle indicates the mean heading; geographic north is denoted N. (B) Swimming orientation along tracks of green turtles north of latitude 27.5 N (mean heading = 119, Rayleigh r = 0.318, n = 208 steps). (C) Swimming orientation along tracks of green turtles south of latitude 27.5 N (mean heading = 89, Rayleigh r = 0.202, n = 27 steps). (E) Swimming orientation along tracks of Kemp s ridley turtles north of latitude 27.5 N (mean heading = 102, Rayleigh r = 0.063, n = 238 steps). (F) Swimming orientation along tracks of Kemp s ridley turtles south of latitude 27.5 N (mean heading = 344, Rayleigh r = 0.313, n = 74 steps). (H) Swimming orientation along tracks of surface drifters north of latitude 27.5 N (mean heading = 281, Rayleigh r = 0.190, n = 377 steps). (I) Swimming orientation along tracks of surface drifters south of latitude 27.5 N (mean heading = 148, Rayleigh r = 0.027, n = 137 steps). (even if relatively weak) (Figure 3) strongly shapes the movement and distribution of organisms at sea. Thus, data on the mechanisms of orientation [3, 13, 40] may be as important as information on swimming speed and pelagic-larval duration [1, 41] for predicting the ecological and evolutionary processes shaped by animal movement. EXPERIMENTAL PROCEDURES Capture and Tagging of Sea Turtles All research was conducted in compliance with the protected species laws of the United States and under NOAA Fisheries and University of Central Florida IACUC approvals (Atlantic , 13-37W) and National Marine Fisheries permits (1551 and 16733). We deployed 44 solar-powered satellite tags (Microwave Telemetry, Inc.; 9.5 g) on green and Kemp s ridley juvenile sea turtles captured offshore in the northern and eastern Gulf of Mexico between July 2011 and June Turtles were between 14.1 and 29.9 cm straight carapace length (Table S1), corresponding roughly to an age of 1 to 2 years [42]. Turtles were captured from a vessel platform; observers on the vessel searched for turtles along oceanographic fronts where turtles aggregate [5 8] (typically associated with Sargassum algae in blue water located at least 100 km offshore). Sampling occurred offshore of Marco Island, Sarasota, Panama City, and Pensacola (Florida); Orange Beach (Alabama); and Venice (Louisiana). All turtles were measured, weighed, and flipper tagged using standard protocols [43]. For Kemp s ridley turtles (n = 20), the attachment technique included pre-treating the turtle s shell with manicure acrylic and 1224 Current Biology 25, , May 4, 2015 ª2015 Elsevier Ltd All rights reserved

6 Figure 4. Distribution of Turtles, Drifters, and Virtual Particles Relative density from track data of green turtles (A), drifters released with green turtles (B), virtual particles tracked from green turtle release sites (C), Kemp s ridley turtles (D), drifters released with Kemp s ridley turtles (E), and virtual particles tracked from Kemp s ridley turtle release sites (F). Coloration is log 10 scaled for comparisons of relative density with different numbers of tracks and locations between panels. Although tracks of individual turtles are poorly represented by drifters and virtual particles (Figure 1), there is some agreement in broad-scale distributions. Specifically, virtual particles overlap with most of the dispersal pathways observed in turtles. The differences in relative density are likely attributable to swimming behavior by turtles. attaching two strips of 3 5 mm neoprene to either side of the turtle s vertebral ridge using veterinary or toupee/hair extension glue [16]. Clear aquarium silicone was used to affix the tag to the neoprene [16]. For green turtles (n = 24), the attachment technique included a thin, malleable marine adhesive. Both attachment methods are flexible and allow for some animal growth without detaching. Individuals were released on the same day as captured, and in the same general area. We deployed two types of drifters at the same locations and times as we tagged the turtles: (1) surface Eddie drifters with drogues extending to 1 m depth, and (2) very-near-surface Kathleen drifters (ballasted 5- gallon buckets, 36.7 cm in depth) ( MainPage/lob/driftdesign.html). These drifters were used because surfacepelagic turtles in the Gulf of Mexico spend more than 90% of their time in the uppermost 1 m of the water column [6]. There were a total of 15 releases, 13 of which involved deploying a pair of drifters and 1 to 10 turtles, and 2 of which involved individual Kemp s ridley turtles that were opportunistically captured without drifters onboard (direct comparison between these turtles and drifters was not possible, but track data were otherwise treated the same). A rectangle ( , 75 km 2 ) centered at the latitude and longitude of each deployment location served as a release site for 1,000 virtual particles tracked within the surface layer of Gulf of Mexico Hybrid Coordinate Ocean Model (HYCOM) output (hourly snapshots at 0.04 spatial resolution, extracted from The duration of particle advection was determined by the duration of longest turtle track from a particular release site. Particles were advected at 30 min intervals through the HYCOM output using the Runge-Kutta fourth-order method applied in ICHTHYOP v.2 particle-tracking software [19]. Location data from turtles were imported into seaturtle.org s Satellite Tracking and Analysis Tool [44] for filtering. Location data from the satellite tags were derived from Argos location data and were archived and filtered using standard methods [18]. Positional data were further extracted from tracks of turtles at 48 hr intervals ( steps ) using only the best-quality location data (classified as 0, 1, 2, or 3, for which location errors are typically less than 5 km [45]). We obtained 235 steps from 24 green turtles and 312 steps from 20 Kemp s ridley turtles. Data from 13 Eddie and 13 Kathleen drifters were pooled for analysis, from which 542 steps were obtained. This sub-sampling of data allowed for standardization of track data used in subsequent analysis. Analyses We computed separation distances (rhumb lines) between pairs of drifters to determine what divergence would be expected due to ocean circulation processes. We then computed the separation distances between turtles and each of the drifters at 2-day intervals over a 14-day period [22]. The separation distance between a turtle and each drifter was subtracted from the separation distance between the two corresponding drifters. The Wilcoxon signed-rank test compared the separation distance of drifter pairs with turtles and drifter separation distances each day (2, 4, 6,. 14). We also tested the accuracy of modeled ocean velocity fields used in later analyses by computing separation distances between each drifter and the closest virtual particle [46]. To simplify the visual depiction of the results, we computed the median and mean (±95% confidence interval) separation distance for each day (Figure 2). A rectangle (75 km 2 ) centered at the latitude and longitude of each 48 hr location along the tracks of turtles and drifters served as the release site for 200 virtual particles within the surface layer of Gulf of Mexico HYCOM output. This area was chosen to account for any error in location data, as in [46]. For five locations outside of this domain, global HYCOM output (daily snapshots at 0.08 spatial resolution) was used. The duration of particle advection was determined by the duration between successive points along the track; particles were advected at 15 min intervals through the HYCOM output using ICHTHYOP v.2 software as described above [20]. The particle closest to the next point along the track was used to calculate the apparent ocean current velocity, derived from the straight-line distance between the starting location of the particle and its ending location [46]. The particle vector was subtracted from the track vector (also derived from the straight-line distance between the successive locations) to compute the apparent swimming velocity. We hypothesized that if divergence along the tracks of turtles were primarily the result of model error, Mann-Whitney U tests would find no difference between the swim speeds of turtles and drifters. For these analyses, we assumed statistical independence among the 2-day steps along tracks. For 22 green turtles, 17 Kemp s ridley turtles, and 19 drifters, the 2-day period s swimming speed was not correlated with subsequent speeds (R 2 < 0.55, p > 0.05, for each). For the tracks of 2 green turtles, 3 Kemp s ridley turtles, and 7 surface drifters showing some autocorrelation (R 2 < 0.47, p < 0.04, for each), track steps were sequentially halved (every second step was removed) to eliminate remaining autocorrelation (R 2 < 0.38, p > 0.05, for each). In these analyses, steps were reduced to 221 for green turtles, 263 for Kemp s ridley Current Biology 25, , May 4, 2015 ª2015 Elsevier Ltd All rights reserved 1225

7 turtles, and 375 for drifters. The full track data were used for analyses not directly testing whether turtle swim speeds differed from drifters. SUPPLEMENTAL INFORMATION Supplemental Information includes two figures, one table, and Supplemental Experimental Procedures and can be found with this article online at dx.doi.org/ /j.cub AUTHOR CONTRIBUTIONS N.F.P. performed modeling, conducted quantitative analyses, and contributed to study design. K.L.M. conducted the fieldwork turtle capture, tagging, drifter deployment, and tag data collection paying homage with regular, personal offerings to Neptune/Poseidon. N.F.P. and K.L.M. wrote the manuscript. ACKNOWLEDGMENTS We thank M. Bresette, J. Gorham, K. Aderhold, the Strike Zone Too crew, S. Hirama, B. Witherington, R. Hardy, E. Seney, J. Guertin, C. Mott, R. Welsh, D. Bagley, C. Long, R. Chabot, D. Witherington, and C. Crady for field assistance. J. Manning, D. Eggleston, and R. He provided suggestions for use of drifters. E. Putman, U. Lateiner, and F. Maderspacher provided editorial suggestions on the manuscript. Considerable thanks are also due to the makers of meclizine and scopolamine. Funding and/or support were provided by NOAA Fisheries, Southeast Fisheries Science Center, the National Research Council Research Associateship Program, the NOAA Oil Spill Supplemental Spend Plan, North Carolina State University s Initiative for Biological Complexity, and Inwater Research Group. Received: November 8, 2014 Revised: January 17, 2015 Accepted: March 10, 2015 Published: April 9, 2015 REFERENCES 1. Largier, J. (2003). Considerations in estimating larval dispersal distances from oceanographic data. Ecol. Appl. 13, S71 S Hill, A.E. (1991). A mechanism for horizontal zooplankton transport by vertical migration in tidal currents. Mar. Biol. 111, Staaterman, E., Paris, C.B., and Helgers, J. (2012). Orientation behavior in fish larvae: a missing piece to Hjort s critical period hypothesis. J. Theor. Biol. 304, Young, C.M., He, R., Emlet, R.B., Li, Y., Qian, H., Arellano, S.M., Van Gaest, A., Bennett, K.C., Wolf, M., Smart, T.I., and Rice, M.E. (2012). Dispersal of deep-sea larvae from the intra-american seas: simulations of trajectories using ocean models. Integr. Comp. Biol. 52, Carr, A.F. (1987). New perspectives on the pelagic stage of sea turtle development. Conserv. Biol. 1, Witherington, B., Hirama, S., and Hardy, R. (2012). Young sea turtles of the pelagic Sargassum-dominated drift community: habitat use, population density, and threats. Mar. Ecol. Prog. Ser. 463, Bolten, A.B. (2003). Active swimmers passive drifters: the oceanic juvenile stage of loggerheads in the Atlantic system. In Loggerhead Sea Turtles, A.B. Bolten and B.E. Witherington, eds. (Smithsonian Institution Press), pp Witherington, B.E. (2002). Ecology of neonate loggerhead turtles inhabiting lines of downwelling near a Gulf Stream front. Mar. Biol. 140, Luschi, P., Hays, G.C., and Papi, F. (2003). A review of long-distance movements by marine turtles, and the possible role of ocean currents. Oikos 103, Monzon-Arguello, C., Lopez-Jurado, L.F., Rico, C., Marco, A., Lopez, P., Hays, G.C., and Lee, P.L.M. (2010). Evidence from genetic and Lagrangian drifter data for transatlantic transport of small juvenile green turtles. J. Biogeogr. 37, Lohmann, K.J., Cain, S.D., Dodge, S.A., and Lohmann, C.M.F. (2001). Regional magnetic fields as navigational markers for sea turtles. Science 294, Putman, N.F., Endres, C.S., Lohmann, C.M.F., and Lohmann, K.J. (2011). Longitude perception and bicoordinate magnetic maps in sea turtles. Curr. Biol. 21, Putman, N.F., Verley, P., Shay, T.J., and Lohmann, K.J. (2012). Simulating transoceanic migrations of young loggerhead sea turtles: merging magnetic navigation behavior with an ocean circulation model. J. Exp. Biol. 215, Putman, N.F., Scott, R., Verley, P., Marsh, R., and Hays, G.C. (2012). Natal site and offshore swimming influence fitness and long-distance ocean transport in young sea turtles. Mar. Biol. 159, Scott, R., Marsh, R., and Hays, G.C. (2012). A little movement orientated to the geomagnetic field makes a big difference in strong flows. Mar. Biol. 159, Mansfield, K.L., Wyneken, J., Rittschoff, D., Walsh, M., Lim, C.W., and Richards, P. (2012). Satellite tag attachment methods for tracking neonate sea turtles. Mar. Ecol. Prog. Ser. 457, Mansfield, K.L., Wyneken, J., Porter, W.P., and Luo, J. (2014). First satellite tracks of neonate sea turtles redefine the lost years oceanic niche. Proc. Biol. Sci. 281, Chassignet, E.P., Hurlburt, H.E., Smedstad, O.M., Halliwell, G.R., Hogan, P.J., Wallcraft, A.J., Baraille, R., and Bleck, R. (2007). The HYCOM (Hybrid Coordinate Ocean Model) data assimilative system. J. Mar. Syst. 65, Lett, C., Verley, P., Mullon, C., Parada, C., Brochier, T., Penven, P., and Blanke, B. (2008). Lagrangian tool for modelling ichthyoplankton dynamics. Environ. Model. Softw. 23, Gaspar, P., Georges, J.Y., Fossette, S., Lenoble, A., Ferraroli, S., and Le Maho, Y. (2006). Marine animal behaviour: neglecting ocean currents can lead us up the wrong track. Proc. Biol. Sci. 273, Luschi, P., Benhamou, S., Girard, C., Ciccione, S., Roos, D., Sudre, J., and Benvenuti, S. (2007). Marine turtles use geomagnetic cues during opensea homing. Curr. Biol. 17, Fossette, S., Putman, N.F., Lohmann, K.J., Marsh, R., and Hays, G.C. (2012). A biologists guide to assessing ocean currents: a review. Mar. Ecol. Prog. Ser. 457, Revelles, M., Carreras, C., Cardona, L., Marco, A., Bentivegna, F., Castillo, J.J., Martino, G., Mons, J.L., Smith, M.B., Rico, C., et al. (2007). Evidence for an asymmetrical size exchange of loggerhead sea turtles between the Mediterranean and the Atlantic thought the Straits of Gibraltar. J. Exp. Mar. Biol. Ecol. 349, Abecassis, M., Senina, I., Lehodey, P., Gaspar, P., Parker, D., Balazs, G., and Polovina, J. (2013). A model of loggerhead sea turtle (Caretta caretta) habitat and movement in the oceanic North Pacific. PLoS ONE 8, e Kobayashi, D.R., Farman, R., Polovina, J.J., Parker, D.M., Rice, M., and Balazs, G.H. (2014). Going with the flow or not: evidence of positive rheotaxis in oceanic juvenile loggerhead turtles (Caretta caretta) in the South Pacific Ocean Using Satellite Tags and Ocean Circulation Data. PLoS ONE 9, e Fossette, S., Gleiss, A.C., Chalumeau, J., Bastian, T., Armstrong, C.D., Vandenabeele, S., Karpytchev, M., and Hays, G.C. (2015). Currentoriented swimming by jellyfish and its role in bloom maintenance. Curr. Biol. 25, Putman, N.F., and Naro-Maciel, E. (2013). Finding the lost years in green turtles: insights from ocean circulation models and genetic analysis. Proc. Biol. Sci. 280, Lopez-Castro, M.C., Bjorndal, K.A., Kamenov, G.D., Zenil-Ferguson, R., and Bolten, A.B. (2013). Sea turtle population structure and connections between oceanic and neritic foraging areas in the Atlantic revealed through trace elements. Mar. Ecol. Prog. Ser. 490, Current Biology 25, , May 4, 2015 ª2015 Elsevier Ltd All rights reserved

8 29. Shaver, D.J., Hart, K.M., Fujisaki, I., Rubio, C., Sartain, A.R., Peña, J., Burchfield, P.M., Gamez, D.G., and Ortiz, J. (2013). Foraging area fidelity for Kemp s ridleys in the Gulf of Mexico. Ecol. Evol. 3, Collard, S.B., and Ogren, L.H. (1990). Dispersal scenarios for pelagic posthatchling sea turtles. Bull. Mar. Sci. 47, Carr, A. (1980). Some problems of sea turtle ecology. Am. Zool. 280, Cavallo, C., Dempster, T., Kearney, M.R., Kelly, E., Booth, D., Hadden, K.M., and Jessop, T.S. (2015). Predicting climate warming effects on green turtle hatchling viability and dispersal performance. Funct. Ecol. Published online January 2, Spotila, J.R., and Standora, E.A. (1985). Environmental constraints on the thermal energetics of sea turtles. Copeia 1985, Lohmann, K.J., Putman, N.F., and Lohmann, C.M.F. (2012). The magnetic map of hatchling loggerhead sea turtles. Curr. Opin. Neurobiol. 22, Mott, C.R., and Salmon, M. (2011). Sun compass orientation by juvenile green sea turtles (Chelonia mydas). Chelonian Conserv. Biol. 10, Scott, R., Biastoch, A., Roder, C., Stiebens, V.A., and Eizaguirre, C. (2014). Nano-tags for neonates and ocean-mediated swimming behaviours linked to rapid dispersal of hatchling sea turtles. Proc. Biol. Sci. 281, Blumenthal, J.M., Abreu-Grobois, F.A., Austin, T.J., Broderick, A.C., Bruford, M.W., Coyne, M.S., Ebanks-Petrie, G., Formia, A., Meylan, P.A., Meylan, A.B., and Godley, B.J. (2009). Turtle groups or turtle soup: dispersal patterns of hawksbill turtles in the Caribbean. Mol. Ecol. 18, Putman, N.F., Mansfield, K.L., He, R., Shaver, D.J., and Verley, P. (2013). Predicting the distribution of oceanic-stage Kemp s ridley sea turtles. Biol. Lett. 9, Mansfield, K.L., and Putman, N.F. (2013). Oceanic habits and habitats Caretta caretta. In The Biology of Sea Turtles, Volume III, J. Wyneken, K.J. Lohmann, and J.A. Musick, eds. (CRC Press), pp Putman, N.F., Jenkins, E.S., Michielsens, C.G.J., and Noakes, D.L.G. (2014). Geomagnetic imprinting predicts spatio-temporal variation in homing migration of pink and sockeye salmon. J. R. Soc. Interface 11, Cowen, R.K., and Sponaugle, S. (2009). Larval dispersal and marine population connectivity. Annu. Rev. Mar. Sci. 1, Turtle Expert Working Group (2000). Assessment Update for the Kemp s Ridley and Loggerhead Sea Turtle Populations in the Western North Atlantic. NOAA Technical Memorandum NMFS-SEFSC-444 (United States Department of Commerce). species/tewg2000.pdf. 43. NationalMarineFisheriesServiceSoutheast Fisheries Science Center (2008). Sea Turtle Research Techniques Manual. NOAA Technical Memorandum NMFS-SEFSC-579 (United States Department of Commerce). sefsc.noaa.gov/turtles/tm_579_sefsc_strtm.pdf. 44. Coyne, M.S., and Godley, B.J. (2005). Satellite Tracking and Analysis Tool (STAT): an integrated system for archiving, analyzing and mapping animal tracking data. Mar. Ecol. Prog. Ser. 301, Costa, D.P., Robinson, P.W., Arnould, J.P.Y., Harrison, A.-L., Simmons, S.E., Hassrick, J.L., Hoskins, A.J., Kirkman, S.P., Oosthuizen, H., Villegas-Amtmann, S., and Crocker, D.E. (2010). Accuracy of ARGOS locations of Pinnipeds at-sea estimated using Fastloc GPS. PLoS ONE 5, e Putman, N.F., and He, R. (2013). Tracking the long-distance dispersal of marine organisms: sensitivity to ocean model resolution. J. R. Soc. Interface 10, Current Biology 25, , May 4, 2015 ª2015 Elsevier Ltd All rights reserved 1227

9 Current Biology Supplemental Information Direct Evidence of Swimming Demonstrates Active Dispersal in the Sea Turtle Lost Years Nathan F. Putman and Katherine L. Mansfield

10

11

12

13

14

15 Figure S1, Related to Figure 1. Tracks of turtles, drifters, and virtual particles released from the same locations. Each panel depicts the paths traveled by drifters (blue), virtual particles tracked within a hind-cast ocean circulation model (purple), green turtles (green), and Kemp s ridley turtles (red) released on the same date and location (white circles). Gray shading indicates bathymetry and the thin gray line delineates the Continental Shelf. The release number (e.g., Release 1) listed to the right of each panel corresponds to the Release Number designated in Table S1. The number of turtles of each species and drifters tracked during a given release are also listed to the right of each panel. For each release, 1000 virtual particles were tracked from the site of deployment. Tracks of drifters and virtual particles are truncated to the duration of the longest tracked turtle of a given release. (A) Release 1 August 13, (B) Release 2 August 13, (C) Release 3 September 16, (D) Release 4 July 17, (E) Release 5 July 18, (F) Release 6 July 29, (G) Release 7 October 20, (H) Release 8 October 22, (I) Release 9 May 22, (J) Release 10 May 23, (K) Release 11 May 19, (L) Release 12 May 20, (M) Release 13 May 21, (N) Release 14 June 3, (O) Release 15 June 4, 2014.

16

17 Release Number Turtle ID Species SCL (cm) Release Date Release Latitude Release Longitude Turtle Track Duration (days) Bucket Drifter ID Bucket Track Duration (days) Eddie Drifter ID Eddie Track Duration (days) Lk /13/ Lk /13/ a Lk /16/ None NA None NA Cm /17/ Cm /17/ a Lk /18/ a Lk /18/ a Lk /29/ None NA None NA Cm /20/ Cm /22/ Cm /22/ Cm /22/ Cm /22/ Cm /22/ a Cm /22/ Lk /22/ Lk /22/ Lk /22/ a Lk /22/ a Lk /22/ Cm /23/ Cm /23/ a Cm /23/ a Cm /23/ Lk /23/ Lk /23/ Lk /23/

18 Release Number Turtle ID Species SCL (cm) Release Date Release Latitude Release Longitude Turtle Track Duration (days) Bucket Drifter ID Bucket Track Duration (days) Eddie Drifter ID Eddie Track Duration (days) Cm /19/ Cm /19/ Cm /19/ Cm /19/ Cm /20/ Cm /20/ Lk /20/ Cm /21/ Cm /21/ Lk /21/ Lk /21/ Lk /21/ Cm /3/ Cm /3/ Cm /4/ Lk /4/ Lk /4/ Turtles released at the same date and location share the same release number. = Kemp s ridley ( ), = Green ( ). SCL = straight carapace length.