Marine Biology

, Volume 161, Issue 6, pp 1455–1466

Protected species use of a coastal marine migratory corridor connecting marine protected areas

  • Kellie L. Pendoley
  • Gail Schofield
  • Paul A. Whittock
  • Daniel Ierodiaconou
  • Graeme C. Hays
Original Paper

DOI: 10.1007/s00227-014-2433-7

Cite this article as:
Pendoley, K.L., Schofield, G., Whittock, P.A. et al. Mar Biol (2014) 161: 1455. doi:10.1007/s00227-014-2433-7


The establishment of protected corridors linking the breeding and foraging grounds of many migratory species remains deficient, particularly in the world’s oceans. For example, Australia has recently established a network of Commonwealth Marine Reserves, supplementing existing State reserves, to protect a wide range of resident and migratory marine species; however, the routes used by mobile species to access these sites are often unknown. The flatback marine turtle (Natator depressus) is endemic to the continental shelf of Australia, yet information is not available about how this species uses the marine area. We used a geospatial approach to delineate a coastal corridor from 73 adult female flatback postnesting migratory tracks from four rookeries along the north-west coast of Australia. A core corridor of 1,150 km length and 30,800 km2 area was defined, of which 52 % fell within 11 reserves, leaving 48 % (of equivalent size to several Commonwealth Reserves) of the corridor outside of the reserve network. Despite limited data being available for other marine wildlife in this region, humpback whale migratory tracks overlapped with 96 % of the core corridor, while the tracks of three other species overlapped by 5–10 % (blue whales, olive ridley turtles, whale sharks). The overlap in the distribution ranges of at least 20 other marine vertebrates (dugong, cetaceans, marine turtles, sea snakes, crocodiles, sharks) with the corridor also imply potential use. In conclusion, this study provides valuable information towards proposing new locations requiring protection, as well as identifying high-priority network linkages between existing marine protected areas.


Marine protected areas (MPAs) are now being widely introduced around the world and are often designated, at least partly, to help protect migratory animals where they seasonally aggregate to breed or forage (e.g. cetaceans, Hooker et al. 1999; sharks, Kinney and Simpfendorfer 2009; sea turtles, Schofield et al. 2013a). However, migratory species are also at risk during migration along corridors connecting breeding and foraging habitats (Shillinger et al. 2008; Womble and Gende 2013). Yet, while studies are beginning to identify key corridors used by marine wildlife (Mumby 2006; Block et al. 2011; Olavo et al. 2011), the protection of such areas remains primarily hypothetical (but see Lipcius et al. 2003; Fernandes et al. 2005; King and Beazley 2005; Guzman et al. 2008) or experimental (Holland 2012). The benefits of connecting protected habitats, including isolated areas, involve potentially reducing the risk of extinction by increasing species and population persistence, improving population sizes, enhancing species diversity and/or raising genetic exchange (Newmark 1987; Parks and Harcourt 2002; Hilty et al. 2006).

Failure to implement the protection of wildlife migratory corridors on land, sea or air generally arises because of a combination of wildlife, logistical (capacity to enforce and associated economic costs), stakeholder and/or political issues (Boersma and Parrish 1999; Hyrenbach et al. 2000; Shillinger et al. 2008; Womble and Gende 2013). For instance, the consistency of animal migratory routes may be subjected to variation at both individual and population-level scales, complicating the delineation of key protection zones (for overview, see Akesson and Hedenstrom 2007; Agardy et al. 2011). This issue is exacerbated in avian or marine species that traverse open expanses of ocean (e.g. Shaffer et al. 2006; Schofield et al. 2013b), because wind and ocean currents cause drift, with course correction being difficult in the absence of visual cues, such as landmasses (e.g. Berger 2004; Alerstam et al. 2006; Broderick et al. 2007; Hays et al. 2010; Hawkes et al. 2011). It is difficult to manage (i.e. monitor and regulate) potentially detrimental human activities across vast areas (e.g. Hyrenbach et al. 2000; Hooker et al. 2011). Migratory routes often traverse stakeholder properties (e.g. Innes et al. 1998; Cherney and Clark, 2009), airways/waterways heavily used by commercial shipping, natural energy stations (i.e. wind and wave) and areas used by the armed forces (e.g. Mullen et al. 2013; Firestone et al. 2008) or important fisheries resources (e.g. Zappes et al. 2013). Finally, migratory animals rarely remain within one country; hence, the establishment of international management cooperation and agreed protocols is critical (Shillinger et al. 2008). Recently, more studies are tracking increasingly large numbers of terrestrial, avian and marine wildlife to accurately infer population-level movement patterns (e.g. Borger et al. 2006; Pinaud 2007; Schofield et al. 2013b); however, most proposed areas for protection continue to be based on the single-species approach, rather than at the ecosystem level (Hooker et al. 1999; King and Beazley 2005). At the governmental level, the minimal investment (or fewest hurdles to overcome) for the maximal output is logically sought (Shogren et al. 1999); hence, the delineation of single corridors supporting multiple species might be more likely to be considered over multiple corridors supporting single species (Baumgartner 2004).

In 2012, Australia announced the establishment of a network of Commonwealth Marine Reserves, in addition to existing State reserves, covering 36 % of the nation’s marine area (Australian Government; Supplementary Fig. 1). This network represents an ‘ecosystems’ approach to coastal marine management, by protecting a mosaic of interconnected ecosystem types/habitats and associated biota (McNeill 1994; Fernandes et al. 2005; Russ et al. 2008). Yet, many of these reserves are discontinuous, lacking connecting corridors, despite being designed to protect a number of highly mobile species (Marsh et al. 1993), such as dugong (Marsh et al. 1999), marine turtles (e.g. Limpus et al. 1992; Wallace et al. 2010), whales and dolphins (e.g. Double et al. 2010, 2012a, b; Bejder et al. 2012), sharks (Wilson et al. 2006; Heithaus et al. 2007; Speed et al. 2010) and commercially important fish stocks (Fernandes et al. 2005). This absence of corridors is partially explained by many species migrating to oceanic (pelagic) habitats (e.g. blue whales, Branch et al. 2007; whale sharks, Sequeira et al. 2013; sea turtles, Wallace et al. 2010), rather than along the coast of Australia, making viable corridors difficult to establish. Furthermore, recent research has suggested that species occupying higher latitudes invest in more extensive migrations compared with those occupying tropical regions (i.e. lower latitudes) (Laurel and Bradbury 2006).

Contradictory to these two statements, the flatback marine turtle is endemic to the Australian continental shelf (Pritchard 1997), exhibiting both extensive longitudinal (112–152°E) and latitudinal (4–27°S) migratory movement between breeding and foraging grounds along the west, north and east coasts of Australia (Marsh et al. 1993; Wallace et al. 2010), reaching as far as Papua New Guinea (Limpus et al. 1983; Prince 1998). This species is considered vulnerable in Western Australia (Wallace et al. 2010), due to predation by wildlife (dingoes and introduced red fox), fisheries bycatch and consumption (e.g. of eggs) by indigenous peoples; however, the Red List of the International Union for Conservation of Nature (IUCN) categorises this species as data deficient, and hence difficult to assess. As the flatback sea turtle remains in coastal habitats throughout its life history, it could be used as a focal species to model a coastal migratory corridor (King and Beazley 2005) connecting Australia’s MPAs and may incidentally encompass movement of other migratory species.

Here, we analysed 73 adult female flatback tracking datasets from four rookeries located along the north-west coast of Australia between 2005 and 2012 to (1) delineate the migratory corridor used by these individuals, (2) determine the extent of connectivity and overlap of this corridor with existing State and Commonwealth Marine Reserves and (3) establish the potential benefits of such a corridor to other species through the evaluation of tracking/distribution data for other wildlife in the published literature. Limited information about flatback sea turtles has previously been available until this study; hence, here, we evaluate the extent to which the migratory route of an endemic species receives protection by the existing and planned network of marine reserves along the north-western continental margin of Western Australia.

Materials and methods

Study area and target species

An estimated 20 000 female flatbacks nest along the west, north and east coasts of Australia, spanning latitudes of 4–27°S (Fig. 1a; Wallace et al. 2010). Nesting occurs during the austral summer months (November–January), with up to four clutches being laid at approximately 13.8 ± 0.6 day intervals in a single season (Pendoley et al. 2014), and females returning to breed every 1–3 years (Hamann et al. 2003; Limpus 2009). North-west Australia hosts four major flatback breeding regions that fall within existing and/or planned MPAs (Fig. 1a; Wallace et al. 2010). However, several flatback nesting rookeries fall outside of protection zones. Four such rookeries were studied within the framework of various environmental impact assessments of industrial activity (for more detail, see Pendoley et al. 2014). These rookeries included Barrow Island (20.81°S latitude, 115.45°E longitude), Mundabullangana (20.41°S, 118.07°E), Port Hedland (20.31°S, 118.58°S) and Thevenard (21.46°S, 115.02°E) (Fig. 1b). A recent study has modelled that Barrow and Mundabullangana support 1 512 and 1 461 nesting females annually, with insufficient data being available to calculate this information for the other two nesting sites (Pendoley et al. 2014).
Fig. 1

a Major (red circles) and minor (yellow circles) flatback nesting sites in Australia (based on Wallace et al. 2010), b the flatback nesting sites from which the female flatback turtles departed in the current study and c an adult female flatback turtle carrying a transmitter

Turtle instrumentation

Between 2005 and 2012, 100 adult female flatback turtles were equipped with satellite transmitters at the four specified rookeries on the north-west coast of Australia (Barrow Island, n = 59 turtles; Mundabullangana, n = 3; Port Hedland, n = 30; Thevenard Island, n = 8), of which 73 were successfully tracked to foraging areas (see Supplementary Table 1 for turtle and transmitter information; Barrow Island, n = 44; Mundabullangana, n = 2; Port Hedland, n = 21; Thevenard Island, n = 6). Four different models of satellite transmitter were attached to females: KiwiSat101 and Fastloc GPS-Argos transmitters from Sirtrack Ltd., MK-10 from Wildlife Computers and Satellite-Relayed Data Loggers from St Andrews Sea Mammal Research Unit (for transmission details, see Pendoley et al. 2014). All transmitters were attached to the carapace of females on the beaches immediately following nesting. All units were attached using a harness method adapted from Sperling and Guinea (2004; Fig. 1c). In brief, each unit was mounted on a polycarbonate plate lined with grooved neoprene padding to allow water flow beneath the plate. The unit was then positioned on the flatback turtle using a harness threaded through six slots present on the polycarbonate plate. Each unit was positioned on the central anterior portion of the flatback turtle carapace, covering approximately the first and second vertebral scutes. The harness had six straps made from nylon seatbelt webbing, which were secured using velcro. Zinc staples held the straps in place and served as a deliberate ‘weak link’ that gradually corroded. This method was proved viable by the return of females between 1 and 3 years after unit attachment in 2005; however, some individuals had evidence of carapace wear. Hence, as with device attachments to sea turtles in general, it is important to aim to quantify the impacts (e.g. drag) of the attachments on flatback turtles (e.g. Witt et al. 2011; Jones et al. 2013). Units provided either Global Positioning System (GPS) quality locations (n = 65) and/or Argos quality locations (n = 8) relayed either via the Argos satellite system or relayed via the mobile phone network. External flipper tags and PIT identifiers were used to distinguish individual turtles and confirm that no turtles were tracked for more than one season. Transmitters were not attached to turtles with signs of recent injury, emaciation or flipper trauma/loss (leeches were not observed on any turtles).

Data processing

To reconstruct migration tracks, we used the highest Argos quality location classes (1, 2 and 3; Hays et al. 2001) and we used GPS quality locations based on six or more satellites (Schofield et al. 2013a, b). In addition, we removed locations that necessitated an unrealistically high speed of travel (>5 km h−1; Luschi et al. 1998) or turning angles >25° (as acute turning angles are usually indicative of erroneous locations; Hawkes et al. 2011). Because of differences in data volume per turtle, the datasets were further filtered to allow comparative analysis among all turtle datasets. This adjustment is important to prevent data point bias to a specific site as a result of certain individuals. Hence, the median location within each day for transmitted and retrieved archival data was selected for each turtle (Swihart and Slade 1985; Makowski et al. 2006; Tremblay et al. 2006; Schofield et al. 2013a) to conduct objective analyses of spatial and temporal area use. In addition, we only retained the transiting portions of the tracks to remove any bias to nontransiting sites, such as the breeding sites (including internesting movement) and foraging sites. Such transient sites (where turtles remained for a few days to weeks during migration, possibly foraging) and final foraging sites were identified (and removed) by individuals slowing down and remaining in fixed areas for extended periods of time (minimum of 5 days; see Hays et al. 2010), using a combination of displacement distance and changes in speed of travel (Blumenthal et al. 2006; Schofield et al. 2010). Depth values were extracted from the Australian bathymetry and topography grid (Whiteway 2009) to determine mean seabed depth traversed by each turtle track.

Migratory corridor delineation

To delineate the corridor used by turtles, the filtered locations were plotted using the World Geodetic System (WGS84; 1984) in ArcGIS (version 10.1, ESRI®) software. Data were projected using the GDA 1994 Geoscience Australia Lambert projection for spatial analysis. To create kernel density estimates (KDE), first, we conducted least square cross validation (LSCV) to determine smoothing parameters (bandwidth) (Rodgers et al. 2007). We selected a 20 km cell size for the KDE analysis based on (1) the large geographical extent of the data analyses and track density, (2) the trade-off between computational speed and resolution and (3) because it provided a comparable resolution for the cumulative track length estimate described later in this section. Turtle locations were combined, and the KDE grid was derived using the kernel density tool available in the Spatial Analyst ArcGIS Extension (ESRI®). Grid values were extracted to point turtle locations. Extracted records were queried to determine the volume of the KDE distribution at 25, 50, 75 and 95 % and subsequently used to threshold the KDE raster to determine utilisation area estimates for the polygons used for the display.

To determine corridor use, the track locations were converted to polylines sequenced by logged time stamp to produce a cumulative track for each turtle. A regular 20-km grid mesh was generated for the study area using the Repeating Shapes ArcGIS extension tool (Jenness 2012). To determine the cumulative track length of all turtles that passed through each 20 km vector grid, a spatial join was used. Cumulative track length was used rather than the count of the number of individuals to better reflect corridor path use. This approach ensures that turtles in a given grid were not equally weighted, such as a turtle traversing just 100 m versus a turtle transiting the entire 20 km cell. The results were classified into four categories based on class breaks using intervals of 1 standard deviation from the mean; specifically >1.5 = very high use (termed core); 1.5–0.5 = high use; 0.5 to −0.5 = intermediate use; less than −0.5 = low use. The four raster categories generated here were used for the quantification of all subsequent analyses (Supplementary Fig. 2).

Migratory corridor overlap with MPAs

GIS layers for the State Marine Reserves intersecting with the turtle track data were obtained from the Department of Environment and Conservation (n = 5, total area 10 047 km2). The GIS layers for the Commonwealth Marine Reserves (n = 18, total area 487 477 km2) were obtained from the Department of Sustainability, Environment, Water, Population and Communities ( Cells with cumulative track values in the grid mesh that intersected with individual State and Commonwealth Reserves and within zoning schemes were identified. Cumulative track length and the proportion of marine reserve containing track cells were quantified.

Migratory corridor benefit to other marine wildlife

All peer-reviewed literature in the Web of Knowledge and Google Scholar, along with publically available reports on the Internet, was searched for marine mammal, marine reptile and shark species in Australia to obtain information about species distributions and migratory tracking datasets. When distribution ranges were not available from the published literature, they were obtained from the Australian Government Species Profile and Threats Database (SPRAT, For the marine species for which tracking data were available, the tracks were reconstructed and overlaid on the corridor delineated from the flatback turtle tracks.

The migratory tracks of other species (n = 6) that were located for this region were reconstructed using Google Earth (Supplementary Fig. 3). Then, using ArcGIS (version 10.1, ESRI®), we generated presence (value 1) and absence cells (value 0) for each species, based on reconstructed migratory track positions. We then added all species’ extents together to determine cumulative space use. For example, a cell value of seven would indicate tracks for all species (i.e. the six other species plus the flatback) intersecting that specific cell. We then calculated the percentage of migratory species tracks that intersected with turtle tracks and the proportion of tracking cells for each species falling within the marine reserve network.


Migratory movement patterns

Of the 73 adult female flatback turtles tracked to their foraging grounds, 11 remained within 100 km of the frequented rookeries (i.e. residents, mean 65 ± 25 km, range 10–95 km), four migrated south-west (all from Port Hedland) and 58 migrated north-east (Fig. 2a; Supplementary Table 1). The four turtles that travelled south-west from Port Hedland migrated an average 400 km (mean 405 ± 29 km, range 370–440 km). Of the 58 turtles that migrated north-east, the majority (45 %, n = 26) travelled 500–1,000 km, 22 % (n = 13) travelled 100–500 km, 24 % (n = 14) travelled 1,000–1,500 km and 9 % (n = 5) travelled >1,500 km. All turtles combined had a latitudinal range of 10–22°S and longitudinal range of 114–141°E. In general, the females from all four nesting sites were of a similar size range, with an average curved carapace length of 90 cm (SD ± 2.5; range 85–99 cm). Postnesting migration away from the rookeries spanned a period from 22 November to 27 January, for all years combined.
Fig. 2

a The postnesting migratory tracks of 73 adult female flatbacks tracked from the four nesting sites (yellow circles from west to east: Thevenard, n = 5; Barrow Island, n = 44; Mundabullangana, n = 2; Port Hedland, n = 22) between 2005 and 2012. b The 25 % (red), 50 % (pink), 75 % (orange) and 95 % (yellow) kernel density estimates (KDEs) of the migratory tracks using ArcGIS (version 10.1, ESRI®) software. The bandwidth was calculated using LSCV = 35 km. Because the tracks were linear (i.e. unidirectional), and all turtles did not depart from the same rookery or arrive at the same foraging area, the 25–50 % KDEs were biased to the rookeries (where all tracks began), and hence were not representative of the corridor; therefore, the 75 % KDE was more representative of the broader movement patterns of these four rookeries. c Area estimates based on cumulative track length using a 20-km grid mesh generated by the Repeating Shapes ArcGIS extension. Cumulative track length was used rather than the count of the number of individuals to better reflect corridor path use. Class breaks using intervals of 1 standard deviation from the mean were used to divide the data into four SD categories: very high use (termed core) >1.5 (red), high use 0.5–1.5 (pink), intermediate use = −0.5–0.5 (orange) and low use less than −0.5 (yellow)

Migratory corridor delineation

The 50 % KDE was biased to the marine area directly fronting the rookeries (Fig. 2b)because this is where the transmitters were attached , hence, where the maximum density of tracks occurred; therefore, this measure could not be used to delineate a linear migratory corridor. In comparison, the 75 % KDE better presented the corridor, extending a length of 1 200 km and covering an area of 89 519 km2. The maximum extent of the corridor (95 % KDE) extended 2 760 km in length and covered an area of 235 488 km2.

Cumulative track length in each 20 km vector grid produced a more refined corridor (Fig. 2c). For instance, the core-use area of the corridor (Fig. 2c) was 1 150 km in length, which was similar to the 75 % KDE, but covered an area of 30 800 km2 (i.e. 66 % smaller than the 75 % KDE). The maximum extent of the corridor accounting for all grid cells with turtle tracks present was 3 600 km in length and covered an area of 301 600 km2.

The maximum seabed depth traversed by each track had a mean of 127 m (±20; max. range 50–127 m), with the core corridor area ranging between 50 and 500 m seabed depth, while the maximum depth of the overall corridor extended to 1 000 m. All turtles remained on the continental margin, with a maximum track distance from shore of 125 km (±35; max. range 36–125 km).

Migratory corridor overlap with MPAs

The adult female flatback corridor connected from six (two State and four Commonwealth; core-use) to 11 (three State and eight Commonwealth; all cumulative track length categories) marine reserves along the west to north coast of Australia (Fig. 3a,b; Supplementary Table 2). Compared with the total extent of the corridor, we calculated that 52 % of the core area is encompassed by the existing/proposed marine reserved, and hence afforded protection. The corridor cells fell into 100 % of five of the reserves and 35–85 % of a further four reserves (Fig. 3b; Supplementary Table 2). In comparison, the core corridor area covered 16–62 % of any given reserve (Fig. 3a,b; Supplementary Table 2). Therefore, an additional 14 800 km2 would be required to establish a core corridor that includes the red cells of the corridor that fall outside the reserves in Fig. 3a and connects the reserves. This addition is similar in size to the Eighty-Mile Beach or Arafura Commonwealth Reserves (Supplementary Table 2).
Fig. 3

a The locations of the State and Commonwealth Marine Reserves (black outlined areas) in relation to the cumulative track length estimate for each 20 km grid in Fig. 2c delineating the adult female flatback corridor. Numbering corresponds to the State and/or Commonwealth Reserves: 1 Shark Bay, 2 Carnarvon Canyon, 3 Gascoyne, 4 Ningaloo, 5 Montebello, 6 Dampier, 7 Eighty-Mile Beach, 8 Argo-Rowley Terrace, 9 Mermaid Reef, 10 Roebuck, 11 Kimberley, 12 Ashmore Reef, 13 Cartier Island, 14 Oceanic Shoals, 15 Joseph Bonaparte Gulf, 16 Arafura, 17 Arnhem, 18 Wessel, 19 Limmen, 20 Gulf of Carpentaria and 21 West Cape York. See Supplementary Fig. 1 for more detailed maps, b Percentage of State and Commonwealth Reserves (ordered from west to east) that encompassed the flatback corridor based on the cumulative track length estimate for each 20 km grid (black bars = >1.5 SD, very high use (core); grey bars = all cumulative track cells). See Supplementary Table 2 and Supplementary Fig. 1 for the full terms of abbreviations and MPA information; numbers correspond to the reserves presented in part a. c Cumulative overlap of the migratory tracks of the other species’ in relation to the flatback corridor. Presence (1) and absence (0) cells were generated for each species (n = 7, including the female flatbacks of the current study) based on the reconstructed track positions (n = 79; Supplementary Fig. 3) of published papers, and then, all species extents were added (1 species = yellow grids; 2 species = orange grids; 3 species = pink grids; and 4 species = red grids). The bold black perimeter indicates the 95 % KDE of the flatback corridor

Migratory corridor benefit to other marine wildlife

The general species distribution ranges indicated that at least 20 species of marine mammals (dugong, cetaceans), marine reptiles (sea turtles, saltwater crocodile, sea snakes) and sharks overlapped with the corridor (Supplementary Table 3); however, this information does not confirm species use of this corridor.

Limited published tracking information was available for marine wildlife traversing this region. We located, reconstructed and overlaid the tracking datasets of 79 individuals; 40 humpback whales Megaptera novaeangliae; two blue whales Balaenoptera musculus; eight pygmy blue whales Balaenoptera musculus brevicauda; 12 olive ridley turtles Lepidochelys olivacea; four hawksbill turtles Eretmochelys imbricata; and 13 whale sharks Rhincodon typus (Fig. 3c; Table 1; Supplementary Fig. 3; Supplementary Table 3). Four of these species overlapped with the corridor (excluding the pygmy blue whales and hawksbill sea turtles). The greatest overlap in species area use (four species; Fig. 3c) was recorded in the western section of the corridor (between the Gascoyne/Ningaloo and Montebello marine reserves; Supplementary Fig. 1). Overall, 48 % of the corridor was used by two species, 6 % by three species and 2 % by four species.
Table 1

Tracking studies of other marine species in north-west Australia, showing the overlap with the flatback corridor delineated in the current study


Common name

Latin name

Tracking information

Overlap (%)





Humpback whale

Megaptera novaeangliae

40 individuals



Gales et al. (2009), Double et al. (2010, 2012a,b)


Blue whale

Balaenoptera musculus

2 individuals



Branch et al. (2007), Centre for Whale Research


Pygmy blue whale

Balaenoptera musculus brevicaudas

8 individuals



Gales et al. (2009), Double et al. (2012a)


Olive ridley sea turtle

Lepidochelys olivacea

12 individuals



Whiting et al. (2007), McMahon et al. (2007), Hamel et al. (2008)


Hawksbill sea turtle

Eretmochelys imbricata

4 individuals



Whiting et al. (2006), Hoenner et al. 2012


Whale shark

Rhincodon typus

13 individuals



Wilson et al. (2006), Sleeman et al. (2010), Sequeira et al. (2013)

The percentage overlap was calculated from the utilisation area estimates based on cumulative track length using a 20-km grid mesh generated by the Repeating Shapes ArcGIS extension. Core ≥1.5 SD category; Total = all SD categories combined

The humpback whale tracks (n = 40 tracks) overlapped with 96 % of the core flatback corridor and with 46 % of the total flatback corridor. The whale shark and blue whale tracks (n = 13, n = 2) also overlapped with 9 and 6 % of the core corridor, respectively (for both species, the overlap with the total corridor was 5 %). The olive ridley sea turtle overlapped with 9 % of the eastern end of the corridor (n = 12). However, these observations must be interpreted with caution, because population-level datasets were only available for humpbacks.


Based on the tracking data of 73 adult female flatbacks, we (1) demonstrated that individuals from four flatback marine turtle rookeries show fairly consistent patterns of migration along a coastal route separating breeding and foraging areas, (2) delineated a coastal corridor using geospatial analysis tools, (3) established the utility of the corridor to connect existing and proposed marine reserves in the region and (4) confirmed that other highly mobile marine species use this corridor. Through quantifying the extent to which this corridor is protected within an existing MPA network, we provide valuable information towards proposing new locations for protection, as well as identifying high-priority network linkages between existing MPAs.

Despite the adult female flatbacks in this study originating from four rookeries separated by up to 350 km, all tracked individuals moved along the same narrow tract of neritic waters in the north-west to northern regions of Australia. Humpback whales also exhibited highly similar linear movement patterns along this stretch of coastline, which represented the final leg in their >7,000 km migration to breeding grounds off Kimberley after departing foraging grounds in Arctic waters (Gales et al. 2009; Double et al. 2010, 2012a, b). Based on the published literature (Borger et al. 2006; Pinaud 2007; Schofield et al. 2013b), sufficient numbers of individuals from both species were tracked to make reliable population-level inferences (King and Beazley 2005). Similar levels of route fidelity along coastal tracts have been demonstrated for other marine wildlife across the world (Broderick et al. 2007; Bailey et al. 2009; Hays et al. 2010; Block et al. 2011; Hawkes et al. 2011; Schofield et al. 2013b). This phenomenon is attributed to the presence of fixed geophysical references along the coast that enable the animals to maintain a fixed course heading (Berger 2004; Alerstam et al. 2006). In contrast, oceanic movement patterns tend to be highly dispersed, as exemplified by the movement patterns of blue whales, pygmy blue whales and whale sharks tracked in our study region after departing the coast of Australia (Wilson et al. 2006; Gales et al. 2009; Sleeman et al. 2010; Double et al. 2012a; Sequeira et al. 2013), and in the wider Pacific (e.g. Block et al. 2011). The core route followed by the female flatbacks was around 1 500 km in length (with one individual travelling 2 650 km along the coast), covering broad longitudinal and latitudinal ranges of similar extents to other marine turtle species occupying more temperate latitudes (Broderick et al. 2007; Hawkes et al. 2011; Hays and Scott 2013; Schofield et al. 2013b). Thus, contradicting previous assumptions that species inhabiting lower latitudes (tropical) invest in less extensive long distance migrations (Laurel and Bradbury 2006). However, not all turtles migrated long distances, with some remaining resident around the breeding grounds, and an overall uniform distribution in migration distances. This observation indicates that suitable foraging habitat was, in fact, available along the entire length of the coast; hence, maybe resources are easily exhausted or transient (Bestley et al. 2010), with it ultimately being more energetically beneficial for individuals to move to lower latitudes further from the breeding grounds where richer and more reliable alternative sites are available (Hays and Scott 2013).

Habitat-use maps from tracking data may be biased towards the tagging site, i.e. the tagging site emerges as a high-use area as an artefact; hence, alternative approaches are often designed to obtain less biased habitat-use estimates (e.g. Maxwell et al. 2011), such as state-space models or fractal analysis. In addition, Maxwell et al. (2013) used the time-weighting method developed in Block et al. (2011) to weight later locations more than ones near the tagging site. In the current study, we found the greatest track density was, quite logically, at the rookeries where the transmitters were attached, even after removing all breeding site tracking locations prior to the onset of migration. Hence, we synthesised cumulative track lengths in 20 km grids, which were subsequently categorised into four categories based on standard deviation intervals. The core category produced a corridor of similar length to the 75 % KDE, but was much more refined (with a 66 % smaller area compared with the 75 % KDE), making it much more viable in terms of potential conservation application. Defining methods to identify corridors is critical for conservation. Throughout the literature, the most commonly cited strategy for long-term biodiversity conservation is increasing connectivity between protected areas (Heller and Zavaleta 2009); however, relevant boundaries must be identified (Agardy 1994; Hooker et al. 1999; Agardy et al. 2011) using analytical techniques that are reliable and repeatable for multiple species across multiple taxa (Chetkiewicz et al. 2006; Redfern et al. 2006; Robinson et al. 2011).

Tracking data may be used to delineate the relative importance of habitats used by multiple species (Maxwell et al. 2011, 2013). In the current study, we used presence/absence data from tracking datasets to identify the overlap in corridor use by various species. However, because of small samples sizes from each species, this technique risks delineating ‘hotspot’ areas that only overlap at the marginal edges of habitat (similar to the overlap of a 95 % utilisation distribution contour for multiple species; see Williams et al. 2013). Alternatively, Maxwell et al. (2013) created utilisation distributions for each species that was tracked, and then summed together to show where multispecies hotspots occur. Therefore, various methods are being developed to delineate core-use of multiple populations, species and/or taxa that have different start and end points (or transmission locations) at the seascape scale (Chetkiewicz et al. 2006; Block et al. 2011; Maxwell et al. 2011, 2013; Robinson et al. 2011).

In Australia, many endangered marine species are protected from targeted capture throughout much of their range. For example, aside from a low level of indigenous aboriginal harvest, flatback turtles are listed as vulnerable (Commonwealth Environmental Protection and Biodiversity Conservation Act 1999 and the West Australian Wildlife Conservation Act 1950); however, specific protection measures have not been implemented outside of MPAs. Hence, threats of mortality in unprotected corridors may still be high, for example from fishery bycatch (e.g. Lewison et al. 2004), boat strike (Hazel and Gyuris 2004) and industrial activity (Whiting et al. 2007). The efficacy of marine reserves at protecting species remains a hot topic both for sea turtles and other taxa (e.g. Chape et al. 2005; Bagchi et al. 2013; Cantu-Salazar et al. 2013). Hence, it would be interesting to try to assess bycatch rates inside and outside the marine reserves, as well as incidental boat strikes and other potential sources of mortality. In north-west Australia, there is huge growth in shipping, as part of mining activities, in addition to extensive petroleum and natural gas industrial activity in this region (Roberts et al. 2003; Bejder et al. 2012). For instance, the core area of the turtle migratory corridor falls directly over the Barrow and Thevenard area where the bulk of the North West Shelf oil and gas activity is focused, while Port Hedland is the largest shipping port in the country in terms of tonnage, followed by Dampier (which was visited by over 3 400 shipping vessels in 2006–2007). North-west Australia is also a major tourist destination, with associated boating and recreational fishing activities. Hence, despite the remoteness of these territories, there are still threats of incidental mortality of turtles outside of protected areas; therefore, the establishment of a protected coastal corridor would serve to safeguard the passage of these animals through high-risk areas (e.g. Hooker et al. 1999; Hyrenbach et al. 2000; Parks and Harcourt 2002; Hilty et al. 2006). We found that the existing structure of the marine reserves (State and Commonwealth) along the west to north coast of Australia encompassed 52 % of both the core and total corridor area. Hence, this reserve network provides intermittent protection to migrating marine turtles at certain legs of their journeys. However, even within these reserves, different zones exist with different levels of protection (Supplementary Table 2), although it has been proposed that the use of gillnets, trawls and longline fishing activity be prohibited in all zoning categories, from multiple use to sanctuaries (Australian Government Yet, existing and/or proposed regulations might require further adjustment and stronger enforcement to provide any form of protection benefit for flatbacks and other marine vertebrates transiting through these marine reserves.

Furthermore, to provide continuous protection to migrating flatbacks (among other species), these reserves would ideally need to be connected by a corridor. The core area of the corridor falling outside of the marine reserves covers an area of 14 800 km2; while this area is large, it is actually of equivalent size to the Eighty-Mile Beach or Arafura Commonwealth Reserves in this region. In addition, through our analysis of tracks of other marine wildlife, we found that humpback whale tracks overlapped with 96 % of the core flatback corridor, confirming the potential importance of this linear zone for other wildlife. However, while the two species used the same corridor, the timing of migration to and from breeding areas varies considerably. For instance, the northward and southward humpback migrations along the corridor occur between July and September (Gales et al. 2009; Double et al. 2010, 2012a, b), whereas flatback use extends from September to April (current study; Pendoley et al. 2014). Consequently, year-round protection measures would be required, taking the ecological needs of all species that use this corridor into account to be effective (Chape et al. 2005; Cantu-Salazar et al. 2013). For instance, it is also likely that a number of other species (at least 20, including mammals, reptiles and fishes) that have coastal distributions also use this corridor to varying degrees (Australian Government 2008); hence, information is required about their habitat-use and movement patterns to establish optimal protection measures. In addition, the humpback whale tracks could be integrated with the flatback tracks to extend the corridor westwards towards the Gascoyne and Ningaloo marine reserves, while the olive ridley tracks could extend the corridor eastwards towards the Great Barrier Reef National Park, linking at least five further State and Commonwealth Reserves in the west to north region of Australia. However, it is likely that the implementation of such a corridor would be highly complicated, particularly with the expanse of marine area already covered by existing reserves in this region of Australia. Alternative forms of protection (other than MPAs) could also be used, such as species- and threat-specific measures, like the closure of seasonal fisheries to reduce bycatch or reduced shipping speeds to minimise impacts with whales. The approach we have used here of assessing the overlap of animals and threats may be important to help drive conservation management and to prevent human activities from being shifted out of MPAs and into other important areas used by threatened species. This corridor also provides an example of how multispecies corridors could be designed and established in other coastal regions on the world using endemic species.

Most nations are working towards establishing comprehensive and representative marine protected areas, as part of national and international policies. The marine reserve network of Australia is pioneering on a worldwide scale (Fitzsimons 2011); however, focus remains on protecting areas where animals aggregate (i.e. breeding and foraging areas, Australian Government 2008), rather than shared wildlife migratory routes connecting these sites. Here, we used an endemic species to Australia, to delineate a coastal corridor that connects multiple MPAs and is used by other marine wildlife of conservation importance. We anticipate that, as the tracking datasets of other marine animals in this region of Australia are published, the importance of this corridor (as a whole or as part of a much longer corridor) connecting west and east Australia along the northern coastline will be realised. In conclusion, other, similarly overlooked, endemic species of other coastal regions around the world might also fit this paradigm and could be used in the establishment of multispecies corridors.


We thank Chevron Australia (D. Moro and R. Lagdon) and BHP Billiton (S. Mavrick) for the funding and logistical support for this project. Thanks to staff and volunteers at Pendoley Environmental for field support; notably, P. Tod, R. Murliss, N. Sillem, K. Ball, L. Claessen, T. Sunderland and N. Fitzsimmons. We thank P. Tod of Crackpots Ltd for supply of harnesses and attachment advice. Satellite attachment was conducted under the Department of Environment and Conservation Licence numbers: SF005670, SF006705, SF006706, SF007088, SF007143, SF007144, SF007641 and SF007643. GIS laboratory facilities at Deakin University, Warrnambool, Victoria were used for spatial analyses. We also thank the anonymous reviewers for their constructive suggestions to improve the manuscript.

Supplementary material

227_2014_2433_MOESM1_ESM.docx (1.4 mb)
Supplementary material 1 (DOCX 1448 kb)

Copyright information

© Springer-Verlag Berlin Heidelberg 2014

Authors and Affiliations

  • Kellie L. Pendoley
    • 1
  • Gail Schofield
    • 2
  • Paul A. Whittock
    • 1
  • Daniel Ierodiaconou
    • 2
  • Graeme C. Hays
    • 2
    • 3
  1. 1.Pendoley Environmental Pty LtdBooragoonAustralia
  2. 2.Centre for Integrative Ecology, School of Life and Environmental SciencesDeakin UniversityWarrnamboolAustralia
  3. 3.Department of BiosciencesSwansea UniversitySwanseaUK