Scooter Assisted Large Area Diver-based (SALAD) surveys to assess size and abundance of crown-of-thorns starfish in northern GBR (JCU)

Full description

This dataset records Scooter Assisted Large Area Diver-based (SALAD) surveys of Crown of Thorns Starfish (CoTS) in the northern Great Barrier Reef from 2019 - 2025. Using scooters allows large coral reef areas to be surveyed, while allowing divers to slow down and spend more time to locate cryptic CoTS where scars on coral are visible. This dataset records the density of CoTS over the survey tracks and size measurements of individual CoTS. This dataset comprises of two inter-related tables. The primary table (GBR_JCU_SALAD-CoTS-surveys_2019-2025_Tracks.csv) provides data at the level of individual SALAD tows (generally two parallel surveys conducted in each time and place), which shows the size and position of the survey area and reports on both the number of CoTS recorded as well as the number of distinct sets of feeding scars (where CoTS were not detected). This allows for estimates of recorded densities (number of CoTS sighted per hectare), inferred densities (number of CoTS as well as sets of feeding scare per hectare), and detectability (recorded density/ inferred density). The secondary table (GBR_JCU_SALAD-CoTS-surveys_2019-2025_CoTS.csv) provides specific information on individual CoTS that were recorded during SALAD surveys, including the time, depth, size and exposure of each starfish. The unique Track codes can be used to link data from the two different tables. The GPS coordinates provided are taken from top side GPS (ie from boat), when divers enter and exit the water, which is the most appropriate marks for replicating specific tracks. Transect length however, is determined using a towed GPS. The towed GPS is attached to the diver using a 15-25m tether, such that there is a discrepancy between position of the CoTS and the recorded mark/track that varies depending on depth and wind direction. Tracks should not therefore be used to establish the explicit position of the transect or CoTS, but do provide the best estimate of transect distance. The search paths for each of two buddy divers conducted simultaneous surveys are considered independent, as they are separated by depth/ habitat. However, the distance travelled is assumed to be the same for both divers. This dataset provides the full time series of surveys that were supported through multiple grants. The following is a record of the supporting funding programs and regions that were sampled each year: Year Funding program: Regions 2019 LIRS (2019 CoTS research grant): Lizard 2020 LIRS (2020-2021 CoTS research grant): Cairns, Lizard 2021 LIRS (2020-2021 CoTS research grant): Cairns, Lizard; CCIP (Early-investment project): Charlotte, Grenville 2022 CCIP (Project P-06): Townsville; CCIP (Project P-04): Cairns, Lizard, Charlotte, Grenville 2023 CCIP (Project D-02): Townsville; CCIP (Project P-04): Cairns, Lizard, Charlotte, Grenville 2024 CCIP (Project P-04): Cairns, Lizard, Charlotte, Grenville 2025 NESP (Project 5.4): Cairns, Lizard, Charlotte, Grenville Where: Townsville - Townsville region Cairns - Cairns region Lizard - Lizard Island region Charlotte - Princess Charlotte Bay region Grenville - Cape Grenville region Methods: This project used scooter-assisted large area diver-based (SALAD) surveys to assess local densities of CoTS, following Chandler et al (2023). SALAD surveys were undertaken using Yamaha (500Li) Seascooters. These underwater scooters greatly increased the area that could be surveyed, while allowing complete autonomy over the direction and speed of movement. Most notably, the speed of movement varied in accordance with visibility and habitat complexity to maximise detection of CoTS. During each scooter survey, divers would traverse a section of reef searching for feeding scars and CoTS within a 5-m wide belt, with each diver surveying 2.5 m either side of their median plane. Two buddy divers surveying simultaneously along the same reef tract remain > 5 m apart, providing independent estimates of CoTS densities and generally from different depth zones and habitats. For the most part, one diver would survey the shallow reef crest (1-3m depth depending on the tide) while the other diver would simultaneously survey along the reef slope (4-7m depth); adjusting as necessary to remain within sight of each other. The proximity of divers and survey paths varied according to visibility and habitat structure. Where practicable, divers were attached to a surface float that housed a small, waterproof GPS unit. Aside from recording GPS co-ordinates for the start and end of each surveyed track, GPS co-ordinates were recorded at 30 second intervals to provide a detailed record of the pathway taken by each diver. These data were used to determine the distance travelled by each diver and the overall search area, necessary to estimate the local densities of CoTS. Multiple distinct SALAD surveys were undertaken at each reef, whereby each survey represented the extensive searching of a prescribed area of reef by a single researcher and during a single dive (averaging 67.85 minutes duration 1.3SE). Most scooter surveys were conducted along continuous reef margins, whereby researchers gradually progressed along the reef contour throughout the course of the survey. However, some surveys were conducted on distinct patch reefs, where the divers carefully searched the entire circumference of the reef and terminated the survey when they returned to the starting point. During SALAD surveys, most CoTS were located only after first sighting apparent feeding scars. All apparent feeding scars were carefully inspected to rule out tissue loss due to other factors (e.g., coral disease or Drupella spp.). If confident that tissue loss was caused by CoTS, careful searching was undertaken within the immediate area (up to 10m from the most recent feeding scar) to locate the starfish. In instances where CoTS could not be found (and were presumably hidden within the reef matrix), each distinct cluster of scars was used to infer the local presence of an individual CoTS, assuming multiple feeding scars within the same general vicinity were considered to be caused by a single CoTS. Counts of distinct sets of feeding scars were summed together with the number of CoTS recorded and divided by search area to gain overall inferred CoTS densities, following Chandler et al. (2023). Detectability was therefore estimated based on the number of CoTS that were sighted (recorded density) divided by the inferred density for each SALAD survey. For every CoTS detected, divers recorded i) the size of the starfish (maximum diameter, cm), ii) the time of observation to cross-reference with time-based records of GPS co-ordinates from towed GPS and thereby record the approximate location of each CoTS, iii) depth, iv) the proportion of the starfish that was visible from directly above as a measure of exposure, which will inform likelihood of detecting the same starfish using alternative survey methods, and v) whether the CoTS was actively feeding, as well as taxonomic identity (mostly, genera) of all corals in the immediate vicinity that had feeding scars (conspicuous evidence of recent tissue loss over a relatively large and continuous portion of the colony). Limitations of the data: Estimates of CoTS density recorded during this study are not comparable with data obtained using other survey methods (e.g., manta tows), given that recorded densities of CoTS are based on intensive localised searching for CoTS. Even so, the detectability of CoTS using this survey method was highly constrained. Visual surveys based solely on sightings of individual CoTS are therefore likely to substantially under-estimate local densities. Inferred densities (which explicitly account for the starfish that are not detected) may provide a more accurate estimate of the local density of CoTS, though there are also limitations and potential biases to this approach. Most critically, it is challenging to discern CoTS feeding scars from other causes of coral mortality (e.g., feeding activities of Drupella spp. and other corallivorous invertebrates) without carefully inspecting conspicuous coral injuries. It is also very difficult to distinguish distinct sets of feeding scars (attributable to each individual starfish), especially if CoTS densities are very high, or coral mortality caused by CoTS is compounded by other disturbances (e.g., coral bleaching). We contend that detectability reported in this study is most likely underestimated for two reasons: i) we do not (and cannot) account for detectability of feeding scars, and ii) it was assumed that distinct clusters of feeding scars were caused by a single CoTS. There may also be additional CoTS concealed within the reef matrix without corresponding evidence of recent feeding activity. These biases are also expected to scale with CoTS density, wherein it will be increasingly challenging to distinguish between distinct sets of feeding scars attributable to individual starfish. Accounting for feeding scars will, therefore, have greatest utility for resolving low to moderate densities of CoTS, which are very important for detecting and managing the initiation of population irruptions. During preparation of this dataset for publication, we identified some data consistency issues across the two spreadsheets (GBR_JCU_SALAD-CoTS-surveys_2019-2025_Tracks.csv and GBR_JCU_SALAD-CoTS-surveys_2019-2025_CoTS.csv). These include mismatches between reported CoTS/scar counts in the two tables, Track IDs appearing in GBR_JCU_SALAD-CoTS-surveys_2019-2025_CoTS.csv but not in GBR_JCU_SALAD-CoTS-surveys_2019-2025_Tracks.csv, and discrepancies in location names and reef assignments. Additionally, some survey records (e.g., TET-003D) show geographic positions that are displaced from their reported reef locations. These issues are unlikely to affect the overall usage of this dataset, but users who require fully reconciled data are encouraged to contact the dataset authors directly. Format of the data: The primary data file is a single spreadsheet (GBR_JCU_SALAD-CoTS-surveys_2019-2025_Tracks.csv) that reports on the specific start and end location for each individual SALAD survey, noting that these vary greatly in time and length. The distance travelled is also not a straight line between the start and end co-ordinates (e.g., sometimes loop back to starting point), but is determined based on the specific track and explicitly measured using a towed GPS. The specific GPS track files have mostly been saved but are not currently part of this data set. The second spreadsheet (GBR_JCU_SALAD-CoTS-surveys_2019-2025_CoTS.csv) reports on the information collected for individual CoTS, including size and exposure (see data dictionary). Data dictionary: GBR_JCU_SALAD-CoTS-surveys_2019-2025_Tracks.csv: Year – Survey year. Date – Date of the survey in DD/MM/YYYY. Track – Identifier for the SALAD survey track. Each diver has a separate identifier. Region – Name of the region that the survey was conducted. Values of: Lizard, Cairns, Cape Grenville, Princess Charlotte Bay, Townsville. Reef – Name of the reef where the survey was conducted. Examples: Lizard, North Direction, MacGillvray, Eagle, Moore, Thretford, Elford, U/N 11-049. Site – Local name of the survey location. Examples: Big Vickies (outside), Watson's Bay, Washing Machine, North Direction NE, Crescent Reef - Mooring, Great Adventures - Trap Reef. Start_Time, End_Time – Start and end of the survey in HH:MM. This was not record for surveys in 2019. Time – Survey time in minutes CoTS (Number of CoTS) – Explicit number of Crown-of-thorns starfish (CoTS) recorded on each distinct SALAD survey Scars (Number of Scars) – Number of distinct sets of feeding scars considered to indicate the presence of a CoTS, that was not detected Total – The combined number of COTS and scars, considered to better reflect the absolute number of CoTS likely to be present within the area surveyed Detect (Detectability) – Proportion of Total number of CoTS (inferred density) that were actually detected. Detect = CoTS/Total. Start_Lat, Start_Lon – Starting location of the survey in decimal latitude and longitude. End_Lat, End_Lon – Ending location of the survey in decimal latitude and longitude. Note that the start and end can correspond to the same location if the survey was conducted around the complete perimeter of a patch reef. Distance – Distance in metres of the survey transect. Area – Estimated area in square metres of the survey track based on a 5 m swatch width. Area = Distance*5. RecDensity (Recorded density) – The estimated density of CoTS, based on the actual number of COTS detected within the actual areal extent of the SALAD survey (scaled to 1 hectare) RecDensity = (CoTS/Area)*10000 InfDensity (Inferred density) – The estimated density that takes account of both CoTS detected and distinct feeding scars (Total) within the actual areal extent of the SALAD survey (scaled to 1 hectare) InfDensity = (Total/Area)*10000 GBR_JCU_SALAD-CoTS-surveys_2019-2025_CoTS.csv: Year – Survey year. Track – Identifier for the track that the CoTS were observed on. Region – Region of the survey Reef – Name of the reef Temp – Temperature of the water as measured using the diver’s diver computer. Time – Time that the observation about the CoTS or Scar was recorded Depth – Depth of the observation in metres taken from the dive computer CoTS_Scar (CoTS/Scar) – Records whether we did detect CoTS, as distinct to sets of feeding scars where no CoTS detected; where CoTS were detected we provide information on Size, Exposure and Feeding Size – Maximum diameter of individual CoTS in mm measured in situ from tips of opposite arms Exposed – % of starfish that is visible from directly above Feeding – Indicates whether CoTS were actively feeding, as evident based on stomach everted through oral opening. Values of Y, N or NA. Collected – Undocumented. eAtlas Processing: The original data were provided as two Excel Spreadsheets each containing a single table. These were exported as CSV files for subsequent processing. The attribute names were adjusted, in collaboration with the dataset author, to be compatible with conversion to shapefiles. Positive latitude values in GBR_JCU_SALAD-CoTS-surveys_2019-2025_Tracks.csv were negated (e.g. for track ELF-003D from 16.92379 to -16.92379) and the longitude values for MOO-017D and MOO-017S were corrected from 156.25075 to 146.25075. References: Pratchett MS, Caballes CF, Burn D, Doll PC, Chandler JF, Doyle JR, Uthicke S (2022) Scooter-assisted large area diver-based (SALAD) visual surveys to test for renewed outbreaks of crown-of-thorns starfish (Acanthaster cf. solaris) in the northern Great Barrier Reef. A report to the Australian Government by the COTS Control Innovation Program (32 pp). Chandler JF, Burn D, Caballes CF, Doll PC, Kwong SLT, Lang BJ, Pacey KI, Pratchett MS (2023) Increasing densities of Pacific crown-of-thorns starfish (Acanthaster cf. solaris) at Lizard Island, northern Great Barrier Reef, resolved using a novel survey method. Scientific Reports 13: 19306. Pratchett MS, Doll PC, Chandler JF, Doyle JR, Uthicke S, Dubuc A, Caballes CF (2025) Early detection of renewed population irruptions of Pacific crown-of-thorns starfish (Acanthaster cf. solaris) on Australia’s Great Barrier Reef. A report to the Australian Government by the COTS Control Innovation Program (53 pp). Pratchett MS, Baird AH, Burn D, Caballes CF, Chandler JF, Garing M, Lang BJ, Levering LT, Pacey KI, Doll PC (In review) Detectability and exposure of Pacific crown-of-thorns starfish (Acanthaster cf. solaris) during emerging population irruptions in the northern Great Barrier Reef, Australia. Coral Reefs. Location of the data: This dataset is filed in the eAtlas enduring data repository at: data\custodian\2025-2027-NESP-MaC-5\5.4_COTS-adaptive-management\data\GBR_JCU_SALAD-CoTS-surveys_2019-2025

Lineage

Maintenance and Update Frequency: asNeeded

Notes

Credit
A portion of this dataset was funded by Australian Government Department of Climate Change, the Environment, Energy & Water (DCCEEW) through the NESP Marine and Coastal Hub. In addition to NESP (DCCEEW) funding, this project is matched by an equivalent amount of in-kind support and co-investment from project partners and collaborators.

Purpose
Effectively sampling low-density populations of crown-of-thorns starfish (CoTS) necessitates surveying large areas of reef habitat, which has traditionally been achieved using manta-tow methods. However, manta-tow surveys have limited capacity to detect CoTS, which are mostly concealed beneath corals or within the reef matrix. Conversely, intensive visual surveys greatly increase the detection of CoTS, but are inherently limited in their spatial extent. In this study, we used underwater scooters to substantially increase the spatial extent of visual surveys undertaken by SCUBA divers, but without constraining the capacity to stop and search for CoTS within complex reef habitats, and whenever feedings scars are observed (see Chandler et al. 2023). This novel sampling method provides unprecedented and highly resolved data on the densities of CoTS during the widely anticipated start of the fifth outbreaks of CoTS in the northern GBR since the 1950s.