A quantitative assessment of continuous versus structured methods for the detection of marine mammals and seabirds via opportunistic shipboard surveys

Marine monitoring efforts are increasingly supported by opportunistic shipboard surveys. However, opportunistic survey methods often require adaptation to suit the vessel and the operations being conducted onboard. Whilst best-practice techniques for surveying marine wildlife on vessels of opportunity are yet to be established, testing and development of alternative methods can provide means for capturing ecological information in otherwise under-surveyed areas. Explicitly, survey methods can be improved while baseline ecological data for new regions are gathered simultaneously. Herein, we tested different survey approaches on a vessel of opportunity in a remote offshore area where little is known about the community composition of top-order marine vertebrate predators: western and south-western Tasmania, Australia. We found that continuous surveys provide greater species counts than structured “snapshot” surveys over the course of a voyage, but that structured surveys can be more practical when managing factors such as observer fatigue. Moreover, we provide a baseline dataset on the marine vertebrate community encountered in western and south-western Tasmania. This information will be critically important for industry and conservation management objectives, and is key to our understanding of the offshore ecosystem around Tasmania.

these fisheries and seals 41 , there remains little to no information on the community composition of marine toporder vertebrate predators (defined here as cetaceans, pinnipeds, and seabirds) in the region.
Hereby, this study seeks to address three key knowledge gaps: (1) Are there differences in species detectability between continuous and structured (snapshot) shipboard surveys of marine vertebrates; (2) How do continuous and structured surveys compare in terms of time/cost-effectiveness and relevance for monitoring and management purposes; and (3) Which species compose the marine vertebrate community in offshore waters from western/south-western Tasmania?Beyond these questions, this paper also aims to introduce a valuable, open-access data source that can be used for many species-specific and climate related questions.

Results
After traversing 451.87 nautical miles, and performing 252.33 h of survey effort, we identified 45 species of marine vertebrate predator (seabirds, and marine mammals), and 11,645 individuals to species level in the survey area (survey area shown in Fig. 2-see "Methods" section.Species summaries shown in Table 1.Numbers reported exclude incidental, fish, and plastic records).
When assessing the number of species recorded by each survey protocol to species level identification, structured surveys detected 38 species, and continuous surveys detected 40 species (Table 1).When expanding this assessment to include higher orders such as genera (i.e., those that could not be identified to a species level), structured surveys detected 45 order/species, and continuous surveys detected 53 order/species (Table 1).These differences were significant (LRT, p < 0.01; Wald's F Test, F = 318.71,p < 0.01, Table 2).
Of the species detected, 33 were detected by both surveys (29 birds, 4 mammals), 5 were detected by structured surveys only (3 birds, 2 mammals), and 7 were detected by continuous surveys only (4 birds, 3 mammals) (Table 1).Twenty-six of the detected species are classified as least concern, and 19 are threatened with extinction or worse (as per the IUCN Red List of Threatened Species, https:// www.iucnr edlist.org/ last accessed: 17/06/2024).Whilst most species of seabird were detected via both survey protocols, marine mammals were largely detected via the continuous survey protocol (Table 1).Incidental observations recorded 22 birds and 4 marine mammals (including Long-finned Pilot Whales Globicephala melas, and an unidentified Minke Whale species).Marine mammals were mostly seen on and near the continental shelf between 0 and 2350 m (Fig. 1).

Comparing survey methods
Here, we compare differences between two standard shipboard marine mammal and seabird survey protocols on vessels of opportunity to contrast their relative effectiveness.Whilst models accounted for the large disparity in survey effort, and despite structured surveys generally providing a greater number of detections per effort (Table 1), continuous survey protocols offer a higher likelihood of detecting a higher diversity of species than a structured survey protocol (LRT, p < 0.01, Wald's F Test, F = 318.71,p < 0.01, Table 2).Such marginal differences observed in the total number of species detected by each protocol in this study (n = 2, Table 1) may lead readers to conclude that structured surveys are more efficient (especially when considering the amount of effort afforded to each protocol-~ 60 h for structured and ~ 190 h for continuous).However, it is important to note that when considering the duration of the voyage, the observed disparity in species richness between the two detection methods would expand over time.Intuitively, given infinite time and space, both protocols would eventually encompass all species present in a given area.Realistically however, it is more probable that continuous observations would capture every detection recorded by structured protocols, while the reverse may not hold true.Despite this, there are several additional considerations that should be made prior to the commencement of any survey: (1) Target species As evidenced by our dataset, day-to-day variation in species occurrence, and abundance is present.Seabird abundance is known to fluctuate with weather and sea conditions, and whilst it's implausible to predict the abundance of any given species on any given day, fluctuations in abundance can be compensated for over time provided there is repeated survey effort (where the magnitude of abundance becomes more valuable than precise abundance estimates 42 ).Notwithstanding abundance, the detectability and occurrence of different species is something observers can account for and address in their study design.Pointedly, occurrence is related to seasonality and environmental drivers 43 .Surveys can therefore be temporally and spatially targeted to maximise the likelihood of encountering target species.However, when dealing with species that are rare, sparse, or inhabit subsurface environments (such as Cetacea or Pinnipedia) 32 , species detectability becomes the major consideration.
For example, in our study the intra-species variation between survey protocols was minimal for seabirds, but considerable for mammals.This is likely an artefact of the survey protocols.That is, both groups (seabirds and marine mammals) are wide-ranging predators-but the capacity for flight can enhance the detection likelihood of seabirds 44 , and therefore with increasing effort, reduces the chance of an observer 'overlooking' a species' presence within a region.Conversely, the detection of marine mammals (especially individuals, i.e., those that are not part of a pod/herd) require an observer to be looking in the mammal's direction at the exact moment that it surfaces 45 .This is easier with surface rafting species such as pinnipeds, or porpoising species such as Delphinidae, but is difficult with species such as Balaenoptera who only come to the surface for short periods of time 32,46 .Consequently, continuous surveys inherently increase the chances of encountering marine mammals, and our results reflect this.However, our incidental records capture an even greater diversity of marine mammals, suggesting that marine mammal diversity is easily under sampled-and perhaps snapshot surveys (i.e., time-restricted) are inadequate for the detection of mammals' in lieu of continuous watch.As observers' ability to detect species is affected (regardless of viewing conditions) by external factors such as training, experience, motivation, and fatigue 47 , we believe it is important and beneficial to design survey methods for projects on vessels of opportunity in consultation with the designated observers.It is also important to consider the accessibility of viewing platforms to observers (relevant to the vessel-e.g., are there restricted access areas that would impede sightings?), and the aims of a project piggybacking on vessels of opportunity: As the data being collected on these vessels are not the primary focus of the voyage, how can observers collect data in a way that increases output and simultaneously addresses the proposed research question?For instance, it has been found that tourist vessels provide a suitable platform for baleen whale observations, but that abundance can only be estimated provided that 12 or more vessels collect data 26 .Finally, it is good practice to consider variations in survey method, or alternate survey techniques.For example, when monitoring for cetaceans, there are a number of methods used (mark-recapture, line-transect, acoustic monitoring, etc.)-each with a distinct set of advantages and disadvantages 10 .Not all these methods are applicable on vessels of opportunity, and some are not appropriate for the monitoring of different clades of marine vertebrate, but many of these approaches can be developed, adapted, and applied to Southern Ocean studies.
For this study, it was the firm consensus of all observers that maintaining watch for all marine mammals and seabirds was less fatiguing than maintaining watch for cetaceans only (based on their subjective feelings of tiredness-rather than quantitative measurements).This was because data entry remained relatively constant when observing for all marine vertebrates, which meant that the observers felt more engaged with the task, and were actively stimulated for the watch period.Conversely, when observing for cetaceans only, there were long periods between sightings, and boredom led to fatigue.Though it was not explicitly tested within this study, future studies could pursue a similar approach by assessing/measuring observer fatigue and sighting efficiency as the primary research question.
(3) Time Though structured surveys detected fewer species when compared to continuous surveys, structured surveys presented several benefits that should also be considered on vessels of opportunity.Firstly, structured surveys freed up observer time and allowed them to assist with other operations onboard the scientific vessel.Secondly, they allowed observers to make decisions on when to make best use of their time and perform surveys.For example, there were days when the vessel returned to the east coast to shelter from storms, and surveys were not required.This is a decision that would likely be overlooked, and not possible if continuous surveys were planned for the entire voyage.Finally, structured surveys left observers far less fatigued, and because they have a set area being surveyed, can be used to generate density estimates comparable to other surveys 10,48 .
Unfortunately, there were insufficient observers onboard to implement a paired survey approach, which involves conducting each survey protocol simultaneously.To effectively employ this approach, it would be beneficial to have a dedicated project with a primary focus on testing methods.We strongly encourage future studies to consider this if it is feasible.We further encourage future studies to measure distance to all sightings (regardless of protocol) to enable density estimates via distance sampling.In our study, distances to cetaceans were measured as per EBPC requirements, but we see no barrier as to why this couldn't also be applied to seabirds.Understanding the environmental drivers of abundance and richness was also beyond the scope of this study, but could be assessed by future work.As an initial assessment, it appears that richness is affected by detectability (indicated  www.nature.com/scientificreports/by the high F-values of protocol and cloud cover in Table 2), and features associated with biological productivity (such as bathymetry and fluorescence 49 , Table 2).Fortunately, the habitat coverage (i.e., bathymetry)was reasonably balanced between the two survey types (Fig. 2).By utilizing weighting based on effort (Table 1) and employing our statistical approach, we were still able to make some initial and meaningful comparisons-namely, with regards to our research questions, we can answer the first two such that (1) there are differences in species detectability between survey protocols, and (2) efficacy is related to purpose, and purpose needs to be defined prior to an opportunistic voyage to derive the best outcomes for research.

Describing the seabird and marine-mammal community off West Tasmania
Beyond findings relevant to survey protocols, our study provides an important and detailed baseline capturing marine vertebrate fauna occurrence off the west and south-west coast of Tasmania.In line with our third research question, we recorded 45 species to a species level with the survey protocols described within this paper (Table 1).However, if we include incidental records (not included in this study, but part of the open access dataset available via request at https:// data.csiro.au/, or by contacting the authors), an even greater number of detections (n = 47, and 70 if higher orders are included) is realised.Notably, some incidental records clearly describe species not detected by the continuous or structured survey protocols but not identified to a species level resolution (e.g., Minke Whale sp.Balaenoptera).In itself, this provides some reason for more survey effort in the study area as there are more species to be detected within the region.Another, arguably more compelling reason for additional survey effort, is to account for seasonality and the subsequent changes in community composition due to seasonal migrants 50 .That is, due to the operational nature of the primary project to which these opportunistic surveys were attached, dates were selected during the season known to have the lowest chance of cetacean activity for seismic reflection purposes.Ideally, to comprehensively understand the marine community ecology of western Tasmania, surveys will have to occur during months of high biological activity, when the acoustic environment is far less disturbed.
Despite surveys occurring in-between the peak seasons for Austral winter and Austral summer seabird migrants in south-east Australia (i.e., it is expected that fewer species were present compared to winter and summer 50 ), of the 11,645 individuals we identified to a species level, the majority were either albatrosses or petrels (Table 1).Of these, 12 species (9 albatrosses and 3 petrels) are listed as threatened, endangered, or critically endangered (as per the IUCN Red List of Threatened Species).Additionally, there was a high number of detections for the globally vulnerable (as per the IUCN Red List of Threatened Species) Sperm Whale Physeter macrocephalus.Given that seabirds are the most threatened group of birds in the world, with more than one third of species at risk of extinction, and many more showing population decline 51,52 , our observations suggest the west coast of Tasmania may be an important foraging or migratory pathway for many seabird species.Hereby, it warrants further research attention to quantify the importance of the region as an ecological hotspot.
In fact, in the absence of formal pre-existing information regarding the seabird and marine mammal community off the west coast of Tasmania, we looked to citizen science initiatives such as eBird 53 and Birdata 54 to assess the novelty of our output.Unfortunately, when using Birdata, it was unclear which surveys were conducted by whom, and knowing that Birdata imports from eBird (for example, see: https:// ebird.org/ region/ AU/ post/ birdl ife-surve ys/ last accessed: 19/06/2024), it was difficult to disentangle personal incidental records from our voyage to previous surveys.However, using eBird, we found that this dataset detects 18 species of seabird previously unreported in the study region-emphasising the importance of this dataset and its contribution to conservation management.

Perspectives
With considerable interest in offshore wind-power infrastructure around the southern coasts of Australia, gathering information on marine vertebrate communities is paramount to making informed management decisions 36 .In Europe, where offshore wind-farms have been in operation for more than 20 years, aerial surveys are often conducted as a means to assess marine vertebrate associations with planned or established turbines 55,56 .Whilst aerial surveys are somewhat better at capturing high density counts of marine vertebrates at-sea (particularly seabirds), vessel-based observations provide far greater species resolution 8 -a disparity that is increasing over time due to improvements in photographic technologies, and the ability to assess digital images during, or after, a sighting 34 .Such resolution is vital to Australian marine ecosystem studies given the diversity and migration of many endangered vertebrate species persisting in Australian and Southern Ocean waters 57 .
Furthermore, high level species resolution is beneficial to long-term monitoring programs concerned with the impacts of climate change and species movement resulting from niche expansion or contraction 58 .Long-term monitoring programs provide insight towards macroecological trends not always detected at the breeding colony or site-specific level 43 .When coupled with tracking studies-particularly those that focus on sentinel species (i.e., predators that act as bioindicators for climate or ecosystems 2 )-behavioural insights that are individual, species, and sex-specific can be realised [59][60][61][62] .
Finally, high resolution species data assist with the validation (or eventual invalidation if enough surveys are performed with zero observations) of mechanisms such as Important Bird and Biodiversity Areas (IBAs) 63 .IBAs are part of a data driven Birdlife International programme focussed on conserving avian species, globally 64 .In Australian marine settings, a similar mechanism known as Biologically Important Areas (BIAs) are used to identify important areas for seabirds, but is also extended in this context to include marine mammals and reptiles, as well as some sharks and fishes.Biologically Important Areas are a major planning tool used by the Commonwealth of Australia to manage permitting of Commonwealth waters and in Australian Marine Parks based on species occurrence 65 , and they are currently under review (as announced the on Australian government website: www.nature.com/scientificreports/https:// www.dcceew.gov.au/ envir onment/ marine/ marine-speci es/ bias, last accessed: 14/07/2023).However, for many species, these BIAs are underpinned with very little quantitative data (often just expert opinion) and hence surveys such as this are particularly important for validation and improvement of species distribution maps.
Our study provides validifying information pertinent to 14 of the 16 seabird species identified within Australia's south-east marine region BIA.The two species not captured by this study are coastal, and whilst present in our incidental records, are generally not present in deep offshore waters where much of the survey effort occurred.With climate driven poleward shifts observed in the distribution of many Australian marine species, it remains unknown how or if some ecosystems will adapt 66 .Accordingly, we believe the dataset presented by this study is particularly valuable given the capacity of top-order predator data to detect bottom-up effects 3 .

Conclusions
Our data provide an important baseline in what we hope to establish as an important long-term monitoring site, and help to inform management decisions such as those surrounding BIA assessments.Given the species richness and abundance of marine vertebrates, we believe targeted tracking studies, and surveys across different seasons will increase our understanding of the region, and we encourage anyone working within the area to build upon our existing dataset (open access dataset available via request at https:// data.csiro.au/, or by contacting the authors).We consider this knowledge generation time-critical as marine top-order predators are susceptible to climate effects such as marine heat waves 67 , and Tasmania is projected to experience extreme changes in seasurface temperature due to anthropogenic climate change 68 .Finally, our study highlights the need for careful consideration of methods prior to boarding a vessel of opportunity, and the need to consult voyage managers (crew, staff, scientists, tour-operators, or otherwise) to understand what survey protocols are possible within the bounds of their operations.

Continuous survey protocol
As per Australian federal legislation (EPBC Act Policy Statement 2.1-Interaction between offshore seismic exploration and whales), one MMO was required on watch for marine mammals of interest the entire time that seismic operations were underway.As permit conditions require the designated MMO to collect only marine mammal data, it was decided between the team of four MMOs onboard that one to two MMOs would maintain MMO watch, whilst two MMOs would perform continuous surveys for marine vertebrate predators (mammals and seabirds).The description of methods herein relates to the marine vertebrate predator observers.Both observers were on watch for the entirety of the survey period, except for a staggered 30-min lunch break.This means that the observer effort for continuous surveys is double the survey hours minus 1 h when accounting for lunch (effort = (2 × time)-1).Continuous surveys began at first light, and continued until sunset, or until the seismic line was completed (pre-sunset).Typically, these surveys lasted 8 or 9 h.Notably, seismic operations were occasionally delayed or terminated due to cetacean encounters.When performing continuous surveys, observers were not constrained in where or how far they could look.However, they remained in constant communication to notify one-another of observed animals in an attempt to avoid double counts of individuals.

Structured survey protocol
Structured surveys occurred when seismic operations were not underway, and followed a similar approach to other studies with similar species assemblages 69 .For these surveys, observers did not have a say on where the ship was headed-so ship activity ranged from stationary (when coring operations were underway), to steaming (when in transit, or when dredging operations were underway).During these surveys, each observer watched a 90-degree section over the bow of the ship.One observer would watch the starboard side, and the other would watch the port side.Animals were only recorded if they were seen within a 1 km radius of the observer (essentially, this method used a 2-km strip transect with 1-km either side of the ship, and distances were calculated with the use of reticle binoculars).This limit was chosen as large swells could obstruct the horizon for observers/ obstruct animals behind waves, and the observers wanted to be able to maintain visuals on individual animals to avoid double counts (notably, this hindrance exists within the continuous protocol, but as there are no visual constraints on distance for continuous surveys, observers could maintain watch on an individual beyond high swell until it reappeared).Explicitly, observers would notify one another when an animal was passing by their 90-degree section and moving into the other observers' section.For example, if observer 1 spotted a Shy Albatross on the starboard side traversing across the bow towards the port side of the ship, they would notify observer 2 and let them know to avoid double count of that individual.
Where possible, these surveys occurred within the same daylight period, three times a day: 0800-0900, 1200-1300, and 1600-1700 AEST.Notably, daylight savings ended during this voyage.Survey hours were subsequently adjusted to keep daylight hours biologically consistent between surveys, and effort was calculated as double the recorded time (2 observers, so effort = 2 × time).

Accounting for shipboard followers
For each of the survey methods, both observers tried to account for shipboard followers (i.e., seabirds that return to, or follow a vessel during survey periods) as much as possible.This was done by maintaining visuals on circling birds, but also by noting differences in individual plumages or moult patterns.If an individual with no distinguishing features left the 1 km survey radius (during structured surveys) and later returned, it was likely re-counted.
Given the survey protocol for shipboard followers remained constant, the proportion of followers per species remains constant between surveys and comparisons in abundance are proportionate.That said, based on our observations of the birds whose plumage was unique, very few individuals return after leaving the one km radius.

Incidental observations
Species observed outside of survey protocols were occasionally reported as incidental observations.These observations tended to be collected for species that are known to be less common, data-deficient, difficult to observe, or simply charismatic.They were not included in this analysis but are mentioned in the discussion.

Analyses and data handling
All data analyses and handling occurred in Google Sheets (Google LLC.(2018)), Microsoft Excel (Microsoft Corporation.(2018).Microsoft Excel.Retrieved from https:// office.micro soft.com/ excel), GlobalMapper (version 24.1, see: https:// www.bluem arble geo.com/ global-mapper/), and the R statistical environment (version 4.3.0, 70).Explicitly, data were collected in a shared Google Sheet, and later downloaded to Microsoft Excel where they were screened for typographical errors.Once cleaned, the data were saved in a .csvfile and imported to GlobalMapper

Figure 1 .
Figure 1.Marine mammals observed from the RV Investigator during the IN2023_V02 voyage.Panel A depicts all sightings-with each species described by the colours in the legend below.This colour key extends into panel B and C which depict pinniped and cetacean sightings, respectively.For panels B and C, a white boundary depicts the Tasmanian coastline.A GEBCO_2022 Bathymetric product is displayed under data.This figure was created with GlobalMapper (version 24.1, see: https:// www.bluem arble geo.com/ global-mapper/). https://doi.org/10.1038/s41598-024-68512-6 All data were collected during the CSIRO Marine National Facilities voyage IN2023_V02 onboard the CSIRO Research Vessel RV Investigator.This voyage traversed 4193 nautical miles in waters near to the west and southwest coast of Tasmania, Australia (Fig.2).All work occurred within the bounding box 41° 34′ 56.4″ S, 139° 17′ 06.3″ E, to 44° 37′ 59.6″ S, 147° 23′ 23.0″ E, between the 24th of March, and the 30th of April, 2023.Marine vertebrate surveys were opportunistic in the sense that geophysical surveys were the primary focus of the voyage.Vertebrate observations were made from the observation deck (28.15 m ASL) or bridge (25.5 m ASL).At both stations, observers could see to the horizon in all directions provided that environmental conditions permitted visibility.When seismic operations were underway, two 210 cubic inch airguns were towed at a depth of 6 m.

Figure 2 .
Figure 2. Elevation map of study area (west and south-west Tasmania) with ship tracks (4193 nautical miles) shown in cream, continuous survey effort shown in black, structured survey effort shown in orange, and marine national parks shown in red.A separate legend for bathymetry (depth, GEBCO_2022) is shown to the right of the figure.This figure was created with GlobalMapper (version 24.1, see: https:// www.bluem arble geo.com/ global-mapper/).

Table 1 .
Summary of all observations (that were identifiable to species level) made on board the RV Investigator during the IN2023_V02 voyage for continuous and structured survey protocols.For all columns, protocol type is reflected in the position and colour of text, separated by a pipe/vertical bar "|": Left = Structured (purple), and Right = Continuous (orange).Higher values in the "Number ÷ effort" column are shown with bold text (structured effort = 60 h, continuous effort = 192.33h).Species names reflect IOC World Bird List v14.1.

Table 2 .
Summary table of Wald's F test (type II) on the full model.Whilst environmental variables are reported as part of the output, they were only included for the likelihood ratio test (see "Methods" section).Explicitly, understanding the environmental drivers of species trends is beyond the scope of this study, but is something that can be achieved with the dataset we are presenting.
Marine vertebrate surveys were conducted alongside Marine Mammal Observer (MMO) duties associated with seismic operations.Observers were equipped with binoculars (10 × 42 magnification), reticle binoculars (7 × 50 mag.),DSLR (100-400 mm) or digital (80 × zoom) cameras.Underway global positioning system (GPS), and environmental data were acquired in real time from the onboard ship-sensors.There were two survey protocols: 1. Continuous (which were performed when seismic operations required MMOs on watch for the entire seismic survey period); and 2. Structured snapshots.Each protocol was conducted underway, and considerations pertaining to them, is described in detail below: