Satellite tracking reveals sex-specific migration distance in green turtles (Chelonia mydas)

Abstract

Satellite tracking is a key tool for studying sea turtles in the wild. Most tracking has been performed on adult females however, leaving knowledge gaps regarding other population segments, such as adult males. By satellite tracking 12 male green turtles (Chelonia mydas) at a breeding site in West Africa, we describe their movements from the breeding to the foraging grounds and compare migrations with those of 13 females tracked in the same season. During the mating period, some males remained near the focal nesting site, while others performed exploratory movements, apparently to visit other nearby rookeries. Males migrated on average shorter distances to foraging grounds (377 km, range 50–1081, n = 9) compared to females (1038 km, range 957–1850, n = 11]). Importantly, male foraging areas overlapped with previously described areas for females, suggesting sex-specific migration distances are not derived from differences in habitat selection. Strong support for differential migration by sex in sea turtles has hitherto been found in just one other species, but indications are that it may be a general feature in this group. These findings have important implications for our understanding of the interplay between reproductive roles and movement ecology of these emblematic animals.

1. Introduction

Sea turtles spend most of their lives at sea, making observation of their movements and behaviour a challenge. In recent decades, satellite tracking has become a key tool for unravelling enigmatic aspects of their ecology [1]. Dozens of satellite-tracking studies on sea turtles are published each year on a wide variety of topics, such as habitat selection, navigation and interactions with human activities [2–4]. As female sea turtles are readily captured while nesting on beaches, they are particularly amenable to the deployment of tracking devices. However, focusing tracking efforts on adult females have left the movement ecology of immature and male turtles poorly understood [3]. To date, satellite tracking of adult males has been conducted on all seven sea turtle species, however, often with sample sizes of as few as one or two individuals [5]. Further tracking of adult males can therefore help address long-standing questions, such as whether sea turtles display sex-specific migratory behaviour or use different foraging habitats.

Results from tracking studies on loggerhead (Caretta caretta), hawksbill (Eretmochelys imbricata) and Kemp's ridley turtles (Lepidochelys kempii) suggest that male sea turtles tend to remain closer to breeding grounds than females [6–8]. Additionally, reconstructive paternity analyses and satellite tracking indicate that, in some populations, male sea turtles also return to breed at shorter intervals than do females [9,10]. It has been hypothesized that if males breed more frequently than females, then foraging closer to the breeding grounds may confer fitness benefits by both maximizing mating opportunities and optimizing travel costs [11]. In green turtles (Chelonia mydas), evidence from both population censusing and tracking studies suggest that the sexes use similar foraging habitats [12,13]. However, as few male green turtles have successfully been tracked from breeding sites to foraging grounds, it remains uncertain whether males of this species choose foraging grounds closer to breeding sites [12,14,15].

Here we report on the satellite tracking of male green turtles in West Africa. Our aim was to describe movement behaviours of this under-studied population segment, from the breeding season through migration and into the post-breeding foraging period. To do so, we first quantify the ranging distances and space use areas of males during breeding, when they are presumably attempting to mate with receptive females. Next, we test whether males and females breeding in the same year migrate different distances to the foraging grounds. Finally, we assess whether sex differences in migratory distance can be explained by differential habitat use, by comparing the locations of foraging grounds used by each sex.

2. Methods

(a) . Device deployment

In July and November of 2021, 25 green turtles (12 males and 13 females) were fitted with satellite transmitters to record their seasonal movements. Turtles were tracked from Poilão Island (10°52'08″N 15°43'30″W) in the Bijagós Archipelago of Guinea-Bissau, the largest green turtle rookery in the Eastern Atlantic Ocean [16]. Turtles were fitted with Lotek F6G-376B FastGPS devices, which derive both GPS and ARGOS locations, transmitting the data via the ARGOS satellite system. The peak laying season at Poilão is August–September [17], and as mating generally occurs prior to peak nesting activity [18], males were captured in July. From a boat, mating pairs were spotted near Poilão, a fishing net was cast to catch the male which was equipped with a transmitter either onboard or on the nearby beach. Transmitters were deployed on females in November of the same year, either during nesting or when found stranded in intertidal rock pools after nesting (see [19] for details).

(b) . Tracking data processing

Low-accuracy positions were prevalent in the tracking data (e.g. 52% ARGOS B positions), so we used a series of steps to maximize information content while reducing spatial error effects. First, we removed locations with greater than 5 km h⁻¹ travel speed and less than 20° turning angle using the R package argosfilter [20]. Next, we applied the stricter McConnell speed filter set at 5 km h⁻¹ with the trip R package to remove further outliers [21]. Then, we visually examined plots of net displacement from the tagging site (i.e. Poilão Island) for each turtle to discern the temporal extent of each seasonal period: breeding (mating/inter-nesting), migration and post-nesting foraging (electronic supplementary material, figure S1). We identified migration initiation as a steep and persistent increase in displacement from Poilão without a return movement. We demarcated the beginning of the foraging period as a flattening-off of displacement following migration. We then fitted a continuous-time random walk model for each individual and seasonal period, using the R package crawl, to incorporate spatial error and temporal sampling heterogeneity into estimating locational probability [22]. We used the horizontal spatial error estimates published in the foiegras R package for each location class [23]. We then predicted the travel path at an interpolated sampling interval of 2 h, the mean interval in the raw data.

(c) . Breeding period space use

To describe male space use patterns during breeding we (i) calculated the maximum distance each turtle travelled away from Poilão, (ii) identified loop trips and range-shifts and (iii) estimated home range sizes. We ran kernel density estimation on the model-predicted tracks for each turtle to estimate individual core areas (i.e. 50% utilization distribution (UD)) and home ranges (95% UD) during breeding using the R package track2KBA [24]. For home range estimation, we used a smoothing parameter of 1 km, using the reference bandwidth method. Before fitting crawl models to the breeding period data, we applied an additional filter to remove outlier locations, identified using the core deviation tool in the ctmm R package [25].

(d) . Migration distance

To compare migration distance by sex, we calculated two metrics for each individual completing a migratory journey: (i) the great circle distance between the breeding site (Poilão Island) and the foraging area and (ii) the cumulative distance travelled during migration. As the distribution of migration distances was highly non-normal, we tested for differences using the Mann–Whitney U Test. To visualize the migration routes, we mapped the crawl-predicted path of each turtle. Before fitting crawl models to the migration period data, we used an additional filter to remove positions located more than 500 m inland.

As females in 2021 were tagged in November, towards the end of peak laying period, we tested whether migration departure date affected the probability of a female turtle migrating to a distant breeding location (i.e. Mauritania). To do so, we analysed 26 migrations performed in two earlier years (2018 and 2020), reported in a previous study [19]. We then fitted a logistic regression with migration destination as a binary variable (short/long) and departure date and deployment year as explanatory variables.

(e) . Post-breeding foraging habitat

To assess whether the sexes use different foraging habitats, potentially explaining differences in migration distance, we overlapped the tracks of males during foraging on areas identified in a previous study from the tracks of 35 foraging females [19]. We calculated the number of males which used known female foraging areas and the percentage of points falling within those areas, and for males not using female foraging areas, we calculated the distance to the closest known area.

3. Results

The duration of data transmission varied across individuals, spanning a median of 109 days (min = 17 days, max = 158 days, n = 25, table 1). The devices of four individuals (three males and one female) stopped transmitting prior to migration and one failed during migration (one female), leaving data for nine males and 11 females from breeding through migration and into foraging. During breeding, males ranged a median maximum distance of 13.9 km (min = 5.7 km, max = 99.5 km, n = 12) from Poilão Island. Two males shifted away from Poilão before migration, and two others performed short-duration looping trips (figure 1). Additionally, individual 213039 left Poilão shortly after device attachment, making long looping movements in the Bijagós, before migrating (figure 1). For the 11 males showing periods of stationarity during breeding (i.e. excluding ID 213039), 50% core-use areas and 95% home ranges averaged 9 km² (min = 6 km², max = 28 km²) and 92 km² (min = 32 km², max = 210 km²) in size, respectively. Tracked males migrated away from the breeding grounds after a mean of 40 days (min = 8.2 days, max = 62 days, n = 9).

Table 1.

Summary of sample of green turtles tracked from the rookery location on Poilão Island, Guinea-Bissau.

					migration
ID	sex	curved carapace length (cm)	device deployment date	tracking duration (d)	start date	destination	great circle distance (km)	total distance (km)
213020	male	89	04 Jul 2021	109	27 Aug 2021	Guinea-Bissau	86	131
213021	male	92.8	04 Jul 2021	40	—	—	—	—
213037	male	92.5	07 Jul 2021	17	—	—	—	—
213038	male	100	07 Jul 2021	107	10 Sept 2021	Mauritania	1081	1361
213039	male	97	08 Jul 2021	62	18 Jul 2021	Senegal	384	452
213040	male	95	06 Jul 2021	112	28 Aug 2021	Senegal	379	487
213041	male	93	05 Jul 2021	128	14 Aug 2021	Guinea-Bissau	50	71
213042	male	93.2	06 Jul 2021	76	12 Aug 2021	Senegal	343	448
213043	male	88	05 Jul 2021	60	12 Aug 2021	The Gambia	289	388
213044	male	83	06 Jul 2021	28	—	—	—	—
213045	male	106.3	08 Jul 2021	89	25 Aug 2021	Mauritania	1079	1239
213046	male	92.6	09 Jul 2021	68	03 Aug 2021	Senegal	377	464
						median	377	452
224389	female	103	14 Nov 2021	121	29 Dec 2021	Mauritania	1104	a
224390	female	102	14 Nov 2021	141	14 Nov 2021	Mauritania	1080	1736
224391	female	106.4	15 Nov 2021	158	27 Nov 2021	Mauritania	1028	1242
224392	female	100	15 Nov 2021	67	—	—	—	—
224393	female	98.4	15 Nov 2021	158	15 Nov 2021	Mauritania	1056	1335
224394	female	89.3	15 Nov 021	158	15 Nov 2021	Mauritania	1043	1454
224395	female	105	17 Nov 2021	156	17 Feb 2022	Ghana	1850	2379
224396	female	98.5	18 Nov 2021	155	27 Nov 2021	Mauritania	1023	1386
224397	female	99.6	19 Nov 2021	154	30 Dec 2021	Mauritania	1035	1301
224398	female	97.5	20 Nov 2021	45	22 Dec 2021	b	—	—
224399	female	98	21 Nov 2021	152	03 Dec 2021	Mauritania	1038	1248
224400	female	102.6	21 Nov 2021	151	01 Jan 2022	Mauritania	992	1294
224401	female	100.3	23 Nov 2021	107	10 Jan 2022	Mauritania	957	a
						median	1038	1335

^aKnown foraging location, but exact route unknown due to data gaps.

^bDevice failed during migration, in Northern Senegal.

Figure 1.

Examples of reconstructed travel paths of three male green turtles during breeding. Turtles were captured near Poilão Island (bottom-right corner). Inset figures show the displacement distance travelled from the capture site over time. (a) An individual that remained stationary around Poilão before migrating. (b) An individual that shifted its range close to another known breeding site on Meio Island. (c) An individual that left Poilão shortly after device attachment and performed a longer movement to visit several known green turtle rookeries, before migrating northwards.

Of the nine males tracked throughout migration, two remained in the Bijagós Archipelago, five travelled to coastal waters of The Gambia and Senegal, and two went to the Banc d'Arguin in northern Mauritania (figure 2). For females, 10 of 11 individuals migrated to northern Mauritania, while the remaining female travelled to a hitherto unknown foraging area for this population near Accra, Ghana (figure 2). The median distance to the foraging grounds used by males was significantly shorter than females, both in terms of the great circle distance (W = 82, p = 0.01, n = 20) and total distance travelled (W = 76, p = <0.001, n = 20, table 1). There was no significant effect of advancing season on the probability of a female turtle migrating to Mauritania, rather than moving a shorter distance to The Gambia/Senegal or remaining within the larger Bijagós Archipelago (z = 1.02, estimate = 0.18, s.e. = 0.01, p = 0.3, electronic supplementary material, figure S2).

Figure 2.

The comparison of post-breeding migrations of nine male (a) and 11 female (b) green turtles breeding in 2021. Coloured lines show the re-constructed travel paths of individual turtles between the breeding grounds at Poilão Island (yellow diamond) and foraging grounds (yellow circles). The histogram (c) illustrates the great circle distance between the breeding site and the foraging ground of each individual. The paths of two females that foraged in Mauritania are not displayed (ID 224389 and 224401), due to large temporal gaps. One female travelled southwards to Ghana, a hitherto unknown foraging location for this population, whose travel route is shown in (d).

During foraging, tracking devices on males transmitted a median of 32 days (min = 3 days, max = 73 days, n = 9), revealing that seven of nine males overlapped with foraging areas used by females in 2018–2020 (electronic supplementary material, figure S3). The remaining two males settled within 2.8 km and 6.1 km of known female foraging areas.

4. Discussion

By tracking a robust sample of male green turtles from a breeding site to the foraging grounds, we reveal novel behaviours of this under-studied population segment. During the breeding period, we found that males can both remain near a single rookery or perform exploratory movements, presumably to visit other breeding sites [10]. We show for the first time in green turtles that males tend to migrate shorter distances to foraging areas compared to females. Importantly, the foraging sites chosen by males were also known foraging areas for females, indicating that differences in migratory behaviour are not likely caused by differential foraging habitat selection. We also identified a novel foraging ground for females of this population in Ghana, located further away than all previously known sites.

Differing migration distances between the sexes are a common feature in birds [26], and evidence is mounting to suggest the same in sea turtles [15]. Such sex-specific behaviours may reflect reproductive role specialization, which in turn may be influenced by cost–benefit trade-offs related to the frequency of breeding, migration distance and foraging habitat quality [8]. For example, if male sea turtles breed more frequently than females, they may choose to forage closer to breeding sites to reduce the time and energy costs of migration [6]. Migrating shorter distances may also allow males to maximize mating opportunities, either by intercepting females en route to breeding sites or by scheduling their arrival according to the timing of female receptiveness [11,27]. Evidence from at-sea surveys of breeding loggerheads suggests that males attempting to mate outside the receptive period are normally unsuccessful, implicating a role for female receptiveness in driving male behaviour during migration and breeding [28]. However, spatio-temporal patterns of female sexual receptiveness in the wild remain poorly understood [29].

Sex-specific migration distances could have implications for population structuring at foraging sites. In Nicaragua, catch records from sea turtle fisheries identified male-biased sex ratios at foraging grounds located closer to the main source rookery at Tortuguero, Costa Rica [30,31]. Understanding whether sex ratios are biased at foraging grounds is relevant to understanding wider population dynamics, as mortality rates can differ across sites due to varying exposure to threats [32]. Females previously tracked from Poilão indicate that the foraging destinations used can vary among annual breeding cohorts, with four of four turtles migrating to Mauritania in 2001, seven of 15 in 2018, none of eight in 2019 and six of 12 in 2020 [19,33]. However, without contemporaneous data on male movements, it remains unclear whether sex-specific migration distances remain consistent despite annual variation in foraging destinations. A study tracking loggerheads of both sexes across several years found that males consistently foraged in areas located closer to breeding sites [6], suggesting such differences can persist across years in sea turtles. As the distances green turtles migrate vary widely across populations, it would be relevant to compare movements between the sexes at other sites, for example to understand potential relationships between sex, migratory distance, and foraging habitat (e.g. oceanic versus coastal, [14,34]). Continuing to expand efforts to study breeding males as well as juveniles is crucial, both to gaining a more complete picture of sea turtle ecology and to informing their effective conservation.

Ethics

The animal study was reviewed and approved by Orgão Responsável pelo Bem-estar Animal do ISPA (ORBEA – ISPA). Research permits were granted by the Instituto da Biodiversidade e das Áreas Protegidas (IBAP), of Guinea-Bissau. Fieldwork involving sea turtles followed recommended guidelines (NMFS-SFC, 2008), to reduce possible disturbance to the turtles, and were conducted by trained personnel.

Data accessibility

Data derivatives and analysis code are available in a Zenodo repository (Beal et al. [35]); tracking data are stored in a different Zenodo repository (Beal et al. [36]), which will become available after a one-year embargo period, ending 31 August 2023. The tracking datasets analysed herein can alternatively be requested from the authors on Movebank (www.movebank.org) under the study names ‘Chelonia mydas_bijagos_females_2021’ (study ID: 2173907778), and ’Chelonia mydas_bijagos_males_2021’ (study ID: 2173900562).

Supplementary material is available online [37].

Authors' contributions

M.B.: conceptualization, data curation, formal analysis, methodology, software, visualization and writing—original draft; P.C.: conceptualization, funding acquisition, methodology, project administration, resources, supervision and writing—review and editing; A.R.: project administration, resources and writing—review and editing; A.J.P.: resources and writing—review and editing; C.B.: resources and writing—review and editing; J.M.: formal analysis, methodology, software and writing—review and editing; C.S.: project administration, resources and writing—review and editing; E.S.: project administration, resources and writing—review and editing; R.P.: conceptualization, data curation, funding acquisition, methodology, project administration, supervision and writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This study was funded by the MAVA Foundation through the projects ‘Consolidation of sea turtle conservation at the Bijagós Archipelago, Guinea-Bissau’ and ‘Tortue d'Arguin’; the Regional Partnership for Coastal and Marine Conservation (PRCM), through the project ‘Survies des Tortues Marines’; the ‘La Caixa’ Foundation (ID 100010434) through a fellowship awarded to A.R.P. (LCF/BQ/PR20/11770003); the Fundação para a Ciência e a Tecnologia, Portugal, through a grant (UIDB/04292/2020 and UIDP/04292/2020) awarded to MARE and the project LA/P/0069/2020 granted to the Associate Laboratory ARNET.