Atlantic leatherback migratory paths and temporary residence areas

Abstract

Background: Sea turtles are long-distance migrants with considerable behavioural plasticity in terms of migratory patterns, habitat use and foraging sites within and among populations. However, for the most widely migrating turtle, the leatherback turtle Dermochelys coriacea, studies combining data from individuals of different populations are uncommon. Such studies are however critical to better understand intra- and inter-population variability and take it into account in the implementation of conservation strategies of this critically endangered species. Here, we investigated the movements and diving behaviour of 16 Atlantic leatherback turtles from three different nesting sites and one foraging site during their post-breeding migration to assess the potential determinants of intra- and inter-population variability in migratory patterns.

Methodology/principal findings: Using satellite-derived behavioural and oceanographic data, we show that turtles used Temporary Residence Areas (TRAs) distributed all around the Atlantic Ocean: 9 in the neritic domain and 13 in the oceanic domain. These TRAs did not share a common oceanographic determinant but on the contrary were associated with mesoscale surface oceanographic features of different types (i.e., altimetric features and/or surface chlorophyll a concentration). Conversely, turtles exhibited relatively similar horizontal and vertical behaviours when in TRAs (i.e., slow swimming velocity/sinuous path/shallow dives) suggesting foraging activity in these productive regions. Migratory paths and TRAs distribution showed interesting similarities with the trajectories of passive satellite-tracked drifters, suggesting that the general dispersion pattern of adults from the nesting sites may reflect the extent of passive dispersion initially experienced by hatchlings.

Conclusions/significance: Intra- and inter-population behavioural variability may therefore be linked with initial hatchling drift scenarios and be highly influenced by environmental conditions. This high degree of behavioural plasticity in Atlantic leatherback turtles makes species-targeted conservation strategies challenging and stresses the need for a larger dataset (>100 individuals) for providing general recommendations in terms of conservation.

Figure 1. Movements of 16 leatherback turtles.

Reconstructed movements of 16 Argos-tracked leatherback turtles during their migration in the Atlantic Ocean from 2005 to 2008. Twelve SRDLs were deployed on gravid females nesting in Panama (n = 3, PAyear-ID), Suriname and French Guiana complex (n = 6, SUyear-ID and FGyear-ID, respectively), and Gabon (n = 3, GAyear-ID). Four others were deployed on leatherback turtles incidentally captured by Uruguayan fisheries (pelagic longlines and coastal bottom-set gillnets) in international waters of the Southwest Atlantic and in Kiyú, Uruguay, respectively (URyear-ID). For each turtle, transit and Temporary Residence Areas (TRAs) are identified by dotted and solid lines, respectively. Each TRA is identified by a number in black and white, for neritic and oceanic domains, respectively (see M&M for details).

Figure 2. Fidelity to Temporary Residence Areas.

Illustrative examples of fidelity to Temporary Residence Areas (TRAs) in leatherback turtles during their pluri-annual migration. After nesting in Gabon in March 2006, GA06-1 reached a first oceanic TRA (TRA13) by May 2006 (right insert, dark blue track) that she reached again by May 2007 (light blue track) after a counter-clockwise long loop in the middle South-equatorial Atlantic. After being released in the Rio de la Plata estuary in October 2006 (left insert, brown track), UR06-3 moved southward into oceanic water before coming back to her neritic TRA: the Rio de la Plata estuary (TRA21) by February 2007 (red track) that she reached again by January 2008 (orange track) after migrating north toward Brazilian waters close to the Victoria-Trinidad seamount chain. Each year, UR06-3 resided during 3 months in the Rio de la Plata estuary (TRA21).

Figure 3. Trajectories for satellite-tracked drifters.

Map of trajectories for satellite-tracked drifters released in the vicinity of leatherback turtle tagging sites. Filled circles show the location of the tagging sites. Dotted circles show the starting point of the drifter tracks. Drifters were selected to indicate possible drift scenarios from the tagging sites (Panama, Suriname, French Guiana, Gabon and International waters off the Uruguayan coasts) to the main Temporary Residence Areas of the leatherback turtles identified in this study.

Figure 4. Migration paths and oceanographic parameters.

a- Migration path in relation to weekly sea level anomaly (MSLA) of an Argos-tracked leatherback turtle (PA05-2) nesting in Panama in July 2005. The fine line represents the turtle's track from 10/10/2005 to 20/02/2006 (TRA11), while the bold line represents the week from the 30/12/2005 to the 06/01/2006 concurrent to MSLA map. b- Migration path in relation to weekly absolute dynamic topography (MADT) of an Argos-tracked leatherback turtle (FG05-2) nesting in French Guiana in July 2005. The fine line represents the turtle's track from 01/10/2005 to 24/02/2006 while the bold line represents the week from the 25/10/2005 to the 01/11/2005 (TRA 1) concurrent to MADT map. c- Migration path in relation to chlorophyll a surface concentration of an Argos-tracked leatherback turtle (GA06-1) nesting in Gabon in March 2006. The fine line represents the turtle's track from 04/03/2006 to 21/02/2007 while the bold line represents the period from the 01/06/2006 to the 30/06/2006 (TRA 13) concurrent to [Chla] map.

Figure 5. Variation in diving behaviour and velocities between areas.

Diving behaviour and velocities in transit areas (filled dots), oceanic TRAs (filled crossed squares) and neritic TRAs (filled crossed triangles) for three Argos-tracked leatherback turtles nesting in Suriname (SU05-1) and French Guiana (FG05-2 and FG05-5) during their migrations in 2005. Differences between track sections were statistically tested using Kruskal-Wallis test followed by a post-hoc Bonferroni test. Different letters indicate significant (p<0.05) differences among areas. Values are expressed as mean ± SD.