We compiled satellite telemetry data from marine megafauna satellite tagged in the North-West Marine Region of the Australian Exclusive Economic Zone (EEZ) (henceforth referred to as ‘north-west Australia’) (Figure 1, Supplementary Table S2). The North Marine Region was also included in our analyses as many individuals tagged in north-west Australia moved there (Figure 1). These two regions encompass an area totalling 2,205,740 km² and covering 38% of the Australian EEZ.

We compiled a list of threats to these taxa in north-west Australia based on relevant literature (Gordon et al., 2003, Commonwealth of Australia, 2017; Hart et al., 2018; Commonwealth of Australia, 2020; Reynolds et al., 2022; Womersley et al., 2022). Threat categories were selected as those associated with pressures managed under the Environment Protection and Biodiversity Conservation Act 1999 (Cth) (EPBC Act) in Australia and that are directly associated with anthropogenic activities or processes (Table 1; see Supplementary Table S1 for definitions of commonly used terms). Seven threat categories were ultimately defined: oil spill, underwater noise, vessel strike, bycatch, entanglement, artificial light, and coastal habitat degradation. We were restricted to pressures that matched species distributions within the Australian EEZ in space and time, and consequently, for some threats, there were no available spatial data at the resolution required for our analysis. Most notably these included marine debris and ghost nets for the threat of entanglement. Although incomplete, we considered the use of the most current data available for pressures as the most appropriate measure of human activity in the region to understand current and future exposure to threats (Table 1).

Once the threat categories were defined, we searched for existing spatial layers and data that would be representative of those threats (Table 1) and processing of these is described below.

All pressures that were considered representative of each of the seven threats as per Table 1 were then overlaid, and normalised grid cell values were summed and normalised again between 0 and 1 (referred to as cumulative pressure) to allow for inter-taxa comparisons. The number of co-occurring pressures (referred to as number of pressures) in each grid cell were also counted and mapped. Hence, the maximum cumulative pressure value for a grid cell for each taxon was 1 and the maximum value for number of pressures varied between 6 and 9 for each taxon as not all taxa were affected by all pressures (Table 1).

We also mapped the distribution of each threat category identified in Table 1 by overlaying all spatial layers that were related to each category and then summed the pressure intensity values and normalised them between 0 and 1 to allow for comparison among threat categories. We also counted and mapped the number of co-occurring threat categories in a grid cell within our study region (0 to 7).

Spatial overlap between marine megafauna taxa and pressures was used to estimate two indices of exposure to threats for each taxon in north-west Australia. We adapted the method developed by Halpern et al. (2008) and Maxwell et al. (2013) using the two formulas based on cumulative pressure and occupancy of taxa, and the number of pressures and percentage of animals in a grid cell. As this is an exposure analysis, the risk, magnitude, and frequency of potential impacts, as well as the sensitivity of species, were not considered in the calculations of exposure.

For exposure intensity:

Where the exposure intensity (EI) for each taxon is the overlap between taxa occupancy and cumulative pressure in a grid cell. S is the occupancy in a grid cell of taxon i; And CP is the normalised cumulative pressure as the sum of intensity values for all pressures interacting with taxon i. The maximum theoretical value of EI is 1 if the maximum taxa occupancy (1) overlapped with a maximum cumulative pressure (1).

And for exposure in number:

Where the exposure in number (EN) for each taxon was the overlap between percentage of individuals and number of pressures in a grid cell. N was the percentage of tagged individuals of taxon i in a grid cell (0 – 100), and NP was the number of pressures that interacted with taxon i in each grid cell. The maximum number of N is 100% but the value of NP varies among taxa. Thus, the maximum theoretical value of EN would be 600 (inter-nesting turtles), 700 (migratory turtles, whales, and whale sharks) or 900 (foraging turtles) if all individuals from each taxon overlapped with all pressures affecting the taxon in a grid cell (maximum of six, seven and nine pressures per taxon, respectively, Table 1). We standardised EN values for each taxon to between 0 and 1 to match the scale of EI values for our cumulative threat analyses, but we also report raw EN values for comparison with the maximum theoretical value for each taxon.

We mapped the exposure (intensity and number) per taxa group to identify areas of high exposure. We then ranked all grid cell values of EI and EN from largest to smallest, and split the cells encompassing the top 50% of the cumulative distribution (i.e. high exposure) from the bottom 50% (after Soanes et al., 2013). We classified the area encompassing the top 50% of cells (highest exposure) as core exposure areas (EI and EN values equivalent to the top 50% quartiles).

We calculated the proportion of the spatial distribution of each taxon within core exposure areas and the proportion of the taxa distributions overlapping with 0 to n co-occurring threats categories (maximum of 7; Table 1). We also calculated the overlap between core exposure areas for each taxon and Australian Marine Parks (Commonwealth regulated), marine reserves (State regulated), Indigenous Protected Areas (co-regulated by Indigenous groups), and Biologically Important Areas (BIAs). In Australia, BIAs are defined as areas where biologically important behaviour occurs and are identified in National Recovery Plans or Marine Bioregional Plans under the EPBC Act (Commonwealth of Commonwealth of Australia, 2012; Commonwealth of Australia, 2015b; Commonwealth of Australia, 2017). These BIAs were extracted from the Australian National Conservation Values Atlas (NCV Atlas; Commonwealth of Australia, 2015a) (http://www.environment.gov.au/webgis-framework/apps/ncva/ncva.jsf). We also assessed overlap with Critical Habitat areas designated by the Recovery Plan for Marine Turtles in Australia (20 km buffer for green and hawksbill turtles and 60 km for flatback turtles, around key rookeries; Commonwealth of Australia, 2017) for inter-nesting turtles.

All analyses used software R version 4.1 (R Core Team, 2019), ArcGIS 10.8 and ArcGIS ArcPro v 2.8 (ESRI, 2020).

As our compiled satellite tracking data spanned over multiple years (2000 – 2020), and a reasonable sample size was achieved (median of 49, ranging from 19 to 251 tracks per species; Supplementary Table S2), the calculated distributions were assumed to be representative of the typical spatial distribution of a taxon. All taxa were distributed throughout the study area off north-west Australia and extended into northern Australia, from shallow coastal waters (below the 50 m bathymetric contour) to offshore oceanic areas (beyond the 200 m bathymetric contour) (Figures 2A, B, 3A, B). During inter-nesting, high use areas (occupancy ≥0.5) occurred adjacent to the nesting beaches (Figure 2A) with a large percentage of inter-nesting turtles (>20%) dispersed over larger areas (Supplementary Figure S1A, Supplementary Figure S2F). Distributions of marine turtles during migration and foraging largely occurred over continental shelf waters (<200 m depth) (Figure 2B; Supplementary Figures S3A, B). Only limited migratory movements occurred in oceanic areas (water depths > 200 m) or outside the Australian EEZ (Supplementary Figure S3A, B), with a median water depth of 53 m during migration, ranging from 1 to 6765 m. Areas of highest occupancy (>0.5) during migration mostly occurred in coastal waters off mainland Australia in the Pilbara region and near the nesting sites where turtles were satellite tagged (Supplementary Figures S2A, B and Supplementary Figures S3A, B). Areas of highest occupancy (>0.5) during foraging occurred as discrete grid cells dispersed in coastal waters (Supplementary Figure S1B). The distributions based on the number of turtles in a grid cell mirrored the foraging occupancy distributions, however, high use areas (>15% migrating turtles overlapping in a grid cell, and >4% foraging turtles) were larger for the former (occupancy ≥ 0.5) (Figure 2B and Supplementary Figures S2C, D).

There was a clear distinction between use of coastal and oceanic habitats by whales (Figure 3A; Supplementary Figure S4A) that was driven by species and also season, with northerly migrating humpbacks mostly using coastal habitats whereas southerly migrating humpback whales (Bestley et al., 2019) and pygmy blue whales used habitats further offshore with pygmy blue whales movements predominantly occurring at and off the shelf edge. Whales had relatively low occupancy as migration (fast movement) was generally more prevalent than residency behaviour such as foraging, resting and reproduction (slower movement) (Supplementary Figure S4A). Discrete grid cells of higher occupancy occurred near Port Hedland (humpback whales) (Supplementary Figure S5A) and in the Banda Sea (outside Australian EEZ; pygmy blue whales) (Figure 3A). However, a large number of whales overlapped in the Ningaloo Reef region (Figure 3A; Supplementary Figure S5B). Areas with high percentages of whales in a grid cell also occurred along the continental shelf and shelf edge of north-west Australia (Figure 3A; Supplementary Figure S5B).

High occupancy and percentage of whale sharks occurred off Ningaloo Reef where most sharks were tagged (Figure 3B; Supplementary Figure S5A, B), with the largest percentage of sharks (>40%) found in both offshore and inshore waters from Ningaloo Reef to Barrow Island (Supplementary Figure S4B, Supplementary Figure S5D).

All three taxa (turtles, whales, and sharks) overlapped in the continental shelf waters off the Pilbara coast, where most tags (≈66%) were deployed (Supplementary Table S2), but also the southern Kimberley region (Figure 4A). All taxa also overlapped in restricted areas within offshore waters between coastal Pilbara and Scott Reef (Figure 4A). At least two taxa occurred in the southern extent of the study area between Ningaloo Reef and Shark Bay, northern Kimberley and offshore areas in the Pilbara and Kimberley regions (Figure 4A). Areas with only one taxon represented individual migratory paths from tagged individuals or foraging areas used by individual turtles (Figure 4A).

The distribution of cumulative pressures and number of pressures varied with each taxon. For taxa that displayed offshore movements such as whales, whale sharks and migrating turtles, higher cumulative pressures and a larger number of pressures occurred beyond the shelf break in Western Australia (Figures 3C, D; Supplementary Figures S3C, D, S4C, D). For coastal taxa, including nesting and foraging turtles, a higher number of pressures and higher cumulative pressure intensity occurred very close to shore (Figures 2C, D; Supplementary Figures S1C, D). The maximum number of co-occurring pressures was 9 for foraging turtles, 7 for migratory turtles, whales and whale sharks, and 6 for inter-nesting turtles (Figures 2, 3; Supplementary Figures S1, S3).

We also mapped the distribution of each of the seven threat categories listed in Table 1 (Figure 5) and all threats combined (Figure 4B). Coastal modification was limited to within the 50 m bathymetric contour (Figure 5A). Vessel strike, underwater noise and oil spill threats were distributed throughout the Australian EEZ (Figures 5B–D), with highest threat intensity for a vessel strike occurring along intensive shipping routes around the Pilbara region near large ports (Figure 5B). The underwater noise and oil spill threat distribution was greatly influenced by the number of spatial layers for these threats related to the oil and gas industry (Table 1), with higher intensity near the shelf break (200 m bathymetric contour) and offshore waters between the Ningaloo Reef and Rowley Shoals, as well as near Cartier Reef and Joseph Bonaparte Gulf (Figures 5C, D). The distributions for entanglement and bycatch threats largely reflected the spatial footprint of the commercial fishing and aquaculture industries (Figures 5E, F), whereas the threat from artificial light was highest near ports, major towns and at some offshore oil and gas production facilities, e.g., offshore of the northern Kimberley and Pilbara regions (Figure 5G). All seven threats overlapped in shelf waters of the Pilbara coast between Barrow Island and Rowley Shoals and isolated coastal areas in the Northern Territory and Queensland (Figure 4B). Four to six threats categories overlapped in shallow coastal areas and around Rowley Shoals and Scott Reef, whereas most of the offshore waters (> 200 m) had three co-occurring threats (Figure 4B).

Most species included in our analysis have wide-ranging distributions that extend beyond the Australian EEZ (Wilson et al., 2006; Sleeman et al., 2010; Double et al., 2014; Reynolds et al., 2017; Möller et al., 2020; Ferreira et al., 2021; Thums et al., 2022, Commonwealth of Australia, 2017). However, with the exception of pygmy blue whales, most taxa had highest occupancy, and high percentage of individuals, in coastal waters of north-west Australia, with this area having overlap of all three taxa (Figure 4A).

Overall, number of pressures and cumulative pressures for each taxon were highest on the continental shelf, mid-shelf and slope waters of the Pilbara region (114°E to 117°E), with hotspots around large coastal towns such as Dampier, Port Hedland, Broome and Darwin, where there are large ports and regional towns supporting the resources industry. These high threat values in coastal areas are consistent with the concentration of highest threat and impact levels within coastal zones seen in Australia (Ostwald et al., 2021) and elsewhere globally (Halpern et al., 2009; Maxwell et al., 2013; Trew et al., 2019), as threats resulting from human activities both on land and in the marine environment co-occur in coastal regions (Halpern et al., 2008). In our study area, some offshore areas (>200 m depth) also showed high cumulative pressure and number of pressures for whales and whale sharks. This was due to multiple threats associated with the presence of oil and gas production facilities and commercial fisheries overlapping in this area (Shipping, i.e., artificial light, strike, oil spill and underwater noise).

Areas of high exposure to threats were largely concentrated in coastal waters throughout the Pilbara region and offshore waters across north-west Australia, reflecting the patterns reported for taxa distribution and cumulative threats. When compared to the maximum theoretical exposure scores (1 for EI and 600 – 900 for EN), our exposure values for individual taxon were relatively low to moderate (EI between 0.3 – 0.5 and EN between 32 – 354). This is likely a result of animal distribution showing high occupancy (~1) in a comparatively small area of the total study region thus driving these values (see Figure 2 and Supplementary Figures S1–S4). It also reflects that for most of the extent of taxa distributions we have limited data, particularly for migratory behaviours, resulting in low overlap of individual tracks.

However, even a relatively low to moderate exposure score still implies the presence of multiple threats (Table 2 and Figures 4–6), and the cumulative impact of this exposure could affect animal behaviour and/or population survivorship. For example, the risk of vessel strike is a widespread threat for whale sharks throughout their distribution (Reynolds et al., 2022; Womersley et al., 2022), with strikes occurring in both oceanic and coastal waters. Even within protected areas such as the Ningaloo Reef Marine Park, 39% of sharks identified between 2008 and 2013 exhibited some form of scarring with an increase in number of fresh lacerations year-to-year suggesting increasing boat strikes over time (Lester et al., 2020). Globally, areas of high collision risk (i.e., where core areas of whale shark occurrence and shipping associated threats exist) were positively correlated with reported mortalities of whale sharks due to collision, with vessel strike mortality being probable for at least one tracked shark (Womersley et al., 2022). Thus, even this reported level of exposure may result in negative effects on populations. Risk of vessel strike is also important for whales, with Australia representing 15% of reported vessel strikes involving large whales worldwide (Peel et al., 2018). Additionally, records of entanglement of baleen whales in fishing gear starting in 2010 (Groom and Coughran, 2012) indicate this threat occurs throughout Western Australia, with a maximum record of over 30 whales entanglements in 2013, leading to implementation of gear modification that reduced this statistic to ~9 whale entanglements/year between 2014 – 2017 (How et al., 2021).

Not surprisingly, core exposure areas for coastal species such as foraging and nesting turtles, and humpback whales, occurred in shallow areas of the continental shelf near large cities, ports and areas targeted by coastal fisheries in the Pilbara region. Even large sections of core exposure areas for taxa that display offshore migratory movements were found on the shelf, particularly near Ningaloo Reef, an important seasonal aggregation site for juvenile whale sharks (Wilson et al., 2001; Meekan et al., 2006; Sleeman et al., 2010; Norman et al., 2016) and the deeper waters of the Ningaloo area are an important foraging area along the migratory route of pygmy blue whales in the Eastern Indian Ocean (Double et al., 2014; Möller et al., 2020; Thums et al., 2022). Humpback whales with calves use the Exmouth Gulf area to the east of Ningaloo Reef for resting (Jenner et al., 2001) and neonate calves occur at Ningaloo Reef, with this region possibly representing the southern-most extent of the humpback whale calving grounds along the Western Australian coast (Salgado Kent et al., 2012; Irvine et al., 2017). The importance of the Ningaloo region, a World Heritage Area, for marine megafauna is also evident from the distribution of inter-nesting, migrating and foraging turtles. This area is also used by a wide range of other threatened megafauna not included in this study, such as dugongs (Dugong dugon), reef and oceanic manta rays (Mobula alfredi and birostris), killer whales (Orcinus orca), bottlenose dolphins (Tursiops spp.), Australian humpback dolphins (Sousa sahulensis), Australian snubfin dolphins (Orcaella heinsohni) and several species of cetaceans and sharks (Preen et al., 1997; Sleeman et al., 2007; Allen et al., 2012; Brown et al., 2014; Ferreira et al., 2015; Hanf et al., 2016; Speed et al., 2016; Raudino et al., 2018), but telemetry data for these species are largely absent or else minimal, and/or restricted in spatial coverage.

Additionally, core exposure areas calculated using EN were relatively larger than those of EI, indicating that a large percentage of the tagged animals overlapped with many pressures (at least three or four out of seven) and were consequently exposed to multiple threats (Figure 6). This is because the percentage of individuals is a more conservative metric to define important areas when limited sample size is available (Ferreira et al., 2021; Thums et al., 2022), particularly for the extensive spatial scale used here. As a result, core exposure areas identified using EN are more useful to highlight geographical areas where caution is needed for future management. Our exposure intensity index, however, can easily be used in to analyses of cumulative impact and risk at appropriate spatial scales and for specific species, where other information such as species vulnerability are also accounted for in recognition that impact and vulnerability indices vary with each species (Maxwell et al., 2013) and also life stage. For example, inter-nesting turtles are vulnerable to different threats and at different intensities compared to migratory or foraging turtles (Poloczanska et al., 2009; Fuentes et al., 2011; Hart et al., 2018; Fossette et al., 2021b).

Although core exposure areas represented less than 14% of most taxa distributions, some of these areas extended over ~5° of longitude, spanning hundreds of kilometres and encompassing remote areas on and off the shelf. However, comparatively, in the Gulf of Mexico, 38% of the area used by foraging turtles overlapped with high levels of threats (Hart et al., 2018) and in the Equatorial Guinea, over 25% of suitable habitat for marine mammals had high impact scores (Trew et al., 2019). The relatively low percentage of area exposed to high levels of threats found here in comparison to other highly industrialised regions such as the Gulf of Mexico indicates that management and restrictions of some human activities (e.g., time area closures, boat propellor guards, fishing gear modifications, and restriction of harmful activities within protected areas) have the potential to be considered preemptively. However, we note that many of the activities we assess are of economic and social value, and so a trade-off exists between species protection and human activities. In addition, research prioritisation (e.g., via vulnerability/risk assessments) within this area would likely be feasible at this level of exposure and provide essential information needed for recovery plans and species and spatial management.

The reported exposure of marine turtles, whales and whale sharks to threats occurred despite Australia having a large network of marine protected areas throughout these marine regions. The overlap between core exposure areas and existing spatial protection (Australian marine parks, state-managed marine reserves, and BIAs) indicates that these taxa are still exposed to high levels of threats even when inside the boundaries of designated protection areas. Although some human activities, such as commercial fishing and other natural resource exploration pressures, are excluded from most MPAs, these areas are still exposed to other threats. This includes habitat modification from coastal development, underwater noise from nearby oil and gas production facilities, shipping lanes and recreational boating, as well as artificial light from nearby ports or urban centres. We acknowledge that some impacts of some of these activities may be mitigated (e.g., Oil Pollution Emergency Plan [OPEP] for oil and gas regulations, Ship Oil Pollution Emergency Plan [SOPEP] for large commercial vessels and Bycatch and Discarding Workplans for commercial fisheries), and so a useful next step would be to undertake a cumulative risk assessment of priority areas and species. Although such a process would not be possible across the entire extent of our study area because considerable gaps exist in the understanding of taxa-specific impacts and mitigations of human pressure at spatiotemporal scales relevant to the activity and species distribution (Cordes et al., 2016; Elliott et al., 2019), and thus species-specific risk assessments may be more feasible. Doing so would highlight existing mitigation and areas within species distribution that require further mitigation of the major threats, or to support the design of new areas for protection. This is particularly relevant for threats such as boat strike, entanglement, and coastal modification to migratory humpback whales and marine turtles, especially in marine reserves and World Heritage Areas such as Ningaloo Reef. Additionally, we showed that inter-nesting core exposure areas are almost completely encompassed within turtle inter-nesting Critical Habitat. Considering the importance of these habitats for species survival (Commonwealth of Australia, 2017), fine-scale assessments of cumulative impact and risk could inform on the effectiveness of current mitigation efforts and inform on the need for further mitigation.

Our analysis suggests that the industrial development of north-west Australia is largely responsible for the high threat exposure within the region through intense shipping, oil and gas production facilities, onshore infrastructure, ports and population centres that service their work force. Industrial developments are assessed and regulated through legal frameworks by the States and Territories and/or the Australian Government. An impact assessment process is required for every new development proposal that has the potential to harm a listed threatened species (or habitat) and BIAs available in the National Conservation Values Atlas are often used by regulators to assess the overlap of proposed developments with the distributions of threatened and migratory species. Our detailed analysis will therefore benefit industry by providing distribution maps across behaviours and taxa groups (e.g., Figures 2, 3). This is particularly useful as many existing BIAs have been designed based on limited data combined with expert input rather than quantitative analysis (Thums et al., 2022). Managers will equally benefit from these outputs and the ability to assess and identify areas where a new industrial development might add pressure to an area that is already highly exposed to threats. This is important as impact assessments are usually conducted on an individual project level and, further, existing legislated mitigation and management measures do not take into consideration the potential cumulative impact of multiple co-occurring pressures and projects. Our analysis also highlights areas within megafauna distributions that have lower exposure to threats that could be targeted for the design of new protected areas or enhanced protection to ensure relatively pristine habitats are conserved. For example, low exposure areas within the Kimberley region, as well as the waters off Eighty Mile Beach, were important for turtles and whales, whereas the southern Ningaloo Reef was important for whale sharks and whales (see Figures 2, 3).

Distributions of marine species are dynamic, particularly in the case of highly mobile megafauna, and the definition of spatial distributions can be highly dependent on the sample size of tracked animals (Shimada et al., 2020; Ferreira et al., 2021). It is important to recognise that there are some inherent biases that exist in our tracking datasets, as almost all of our satellite transmitters were deployed within the North-West Marine Region, and some of our tracking datasets were sex and age biased (only adult female turtles) and others seasonally biased (only pygmy blue whales during outward migration) (Figure 1; Supplementary Table S2). Despite attempting to reduce bias towards tagging locations in the calculation of time spent in area and proportion of individuals in grid cells, it is expected that a larger number of tags will be active (and therefore, a larger amount of data is being transmitted) closer to the tagging location (Block et al., 2011), which results in some underestimation of distribution beyond the tagging site. In addition, animal telemetry data provides presence‐only observations with no information about absence or abundance of species. Thus, a lack of movement data in an area does not necessarily mean that it is not used by the species or taxa. This is particularly true when sample sizes of tracking data are low relative to the size of the population, as was the case for whales and whale sharks in our study. Hence, the exposures calculated here are potentially under-estimated. Even for marine turtles, large numbers of tagged animals (n = ~100) may still not be sufficient to describe the full extent of the foraging distribution (Ferreira et al., 2021), however, they are likely to be representative of hotspot areas. For this reason, we did not down-sampled the available tracking data for flatback turtles (n = 251 individuals) relative to the other turtle species considered in this study (n = 96 and 42 individuals for green and hawksbill turtles, respectively). Although the larger flatback turtle dataset may have created a flatback turtle bias in the location of the highest value of turtle exposure, the overall distribution of turtle exposure was not impacted by this bias. Alongside these caveats of sample size and bias, it is important to consider that many of the threats considered here are not static and will likely change in both spatial and temporal extent, as will other threats not accounted for here, such as climate change and marine debris.

Our study focused on threats that are able to be managed at local and regional scales. Although climate change is a major threat for marine megafauna (Halpern et al., 2008; Maxwell et al., 2013; Lewison et al., 2014; Bowler et al., 2020), pressures associated with this threat – including carbon emissions, global industrialisation, large scale agriculture/farming and deforestation – have a global distribution and are not restricted to the marine environment. Associated mitigation actions are subsequently required to be implemented at a much larger scale than the regional context of our study. Similarly, the threat of plastic ingestion for marine megafauna as a result of the concentration of marine debris has been increasingly gaining attention as a widespread, key pressure on species and populations (see review Kühn and van Franeker, 2020). However, spatial data on distributions of marine debris and plastics within the Australian EEZ remains limited (but see Reisser et al., 2013; Olivelli et al., 2020) and detailed mapping at more relevant spatial scales is urgently needed to assess the exposure of marine megafauna to this threat. Once this work is completed in the future it can be incorporated into the spatial threat exposure analysis we present here.