Who lives in the open sea? Distribution and densities of surfacing marine megafauna in three subregions of the South Pacific (New Caledonia, Wallis and Futuna, and French Polynesia)

Study area and survey designs

The geographical extent of the sampling design was chosen to cover a latitudinal and longitudinal gradient and a diversity of marine habitats, but also with regard to the availability of suitable airfields and aircraft maintenance facilities (Fig. 1). Surveys were conducted during the southern tropical wet season (from mid-October to early May), in order to maximise good weather conditions for aerial detection (especially wind speed). The total survey area covers 2.5 million km² throughout the tropical western and central South Pacific Ocean (Fig. 1). It overlaps with three biogeographic provinces in the Longhurst system (Longhurst 2010), which aims to represent the broad-scale processes that drive ocean productivity. First, the Archipelagic Deep Basin has a very large latitudinal extent from 22°N to 35°S, a complex topography including deep basins and shelves, and includes New Caledonia and the Coral Sea at its eastern boundary. The second province is the Western Pacific Warm Pool, a ‘typical’ tropical system, with an oligotrophic mixed layer and known to be at the heart of the El Niño Southern Oscillation (ENSO) mechanism (Longhurst 2010) and including Wallis and Futuna on its south-eastern border. Finally, the South Pacific Subtropical Gyre, characterised by nutrient depleted waters and often described as the most uniform and stable region of the open ocean (Longhurst 2010) which includes French Polynesia.

Fig. 1.

Study area and track of effort for REMMOA aerial surveys in the Tropical Pacific Ocean. (a) New Caledonia in 2014, (b) Wallis and Futuna EEZ in 2015, and (c) French Polynesia in 2010, with bold line corresponding to effort covered two times. Top left: the Longhurst Biogeographical Provinces (from http://www.marineregions.org) with (a) in the Archipelagic deep basins, (b) in the Western Pacific warm pool, and (c) in the South Pacific gyre. a–c maps are at the same scale.

The New Caledonia subregion (25–17°S, 160–171°E) is part of the south-west Pacific, which is considered as the centre of major water pathways from the subtropics to the equator and to southern latitudes (Ganachaud et al. 2007). New Caledonia’s EEZ is included in the Natural Park of the Coral Sea, the fifth largest MPA in the world (1.3 million km²). The study area includes oceanic waters located outside of the barrier reef of the main island up to 250 km to the west, including the shallow banks of Landsdowne and Fairway, and the Loyalty and Walpole Islands to the east, Bélep and d’Entrecasteaux Reefs to the north, and the Isle of Pines in to the south (Fig. 1). This subregion encompasses a variety of habitats including slope, shallow banks, seamounts, and abyssal plains. It was surveyed from October to December 2014. The Chesterfield-Bampton and Bellona submarine plateaus were not surveyed due to their distance from the nearest airfield (Table 1).

The second subregion comprises most of the EEZ of Wallis and Futuna (WF, 15–10°S, 4–9°W). It includes the volcanic islands of Wallis and Futuna and the uninhabited Alofi, as well as some seamounts to the north of Rothuma, Cook’s Dome and Waterwhich Sandbank, and deep offshore waters to the south (Fig. 1). This subregion was surveyed between late November 2014 and mid-January 2015.

Finally, the third subregion which makes up the second largest maritime domain in the South Pacific, is French Polynesia (26–6°S, 154–133°W). Five main archipelagos, all oriented northwest–southeast, create major obstacles to ocean currents in this area. Due to the size of the EEZ, the subregion has been divided into four distinct survey blocks (an isolated group of strata, Fig. 1). The central Tuamotu Archipelago is the largest and includes numerous islands and atolls. Adjacent to it and at its southeastern border is the more dispersed Gambier Archipelago, both of which are included in the Tuamotu-Gambier survey block (TUA-GAM). To the west and south are two major island groups separated by wide deep passages: the Society and Austral archipelagos (Martinez et al. 2009). The French Polynesia subregion exhibits a strong latitudinal gradient in primary production, from the depleted south, to the northern Marquesas Archipelago (Table 1). The western part of the latter benefits from contrasting strong primary production compared to the surrounding oceanic waters, due to a remarkable island mass effect (Martinez and Maamaatuaiahutapu 2004). In addition, the complex geomorphology of the Marquesas archipelago creates a diversity of marine habitats (Galzin et al. 2016). This subregion was surveyed from January to early April 2011.

Survey design in each subregion was pre-determined by airfield locations on islands and aircraft endurance, then the accessible spatial extent was divided into smaller blocks, which were in turn stratified into bathymetric strata: slope (<2000 m) and oceanic (>2000 m). Survey design was optimised with a zigzag track layout, performed with Distance 6.2 software (Thomas et al. 2010) using Equal space zigzag for the New Caledonia and Wallis and Futuna subregions. In French Polynesia tracks were designed manually using a geospatial software (ArcMap from Esri), the survey plan was carried out and some transects were flown a second time on the same axes (56% of the effort in this subregion; Fig. 1) to minimise logistical constraints for the pilots of the three aircraft, with a time lag between crossings of 1 day to 3 months (average 16 days). For the three subregions, the effort/area ratio was not homogeneous between strata, in order to collect enough sightings for specific habitats for distribution modelling. Details are available in the field reports (Van Canneyt et al. 2011, 2015).

Aerial survey protocol

Surveys were conducted using a standard line-transect methodology (Buckland et al. 2001), as previously used for marine mammals (Hammond et al. 2013), but adapted for a multi-target protocol (Laran et al. 2017a; Lambert et al. 2019). To achieve the planned effort in a relatively short period of time, we simultaneously used three twin-engine Britten Norman Islander (BN-2 high-winged aircrafts with fixed landing gear). Transects were flown at 90 knots (167 km.h⁻¹) with a target altitude of 600 feet (183 m). The survey team consisted of 12 to 19 different observers per subregion (with the same team in New Caledonia and Wallis and Futuna), including five observers consistent across the three subregions. Observers attended a 1-week training course before participating in the surveys, which included a few training flights. Each flight typically consisted of two observers scanning the sea surface on either side of the aircraft with the naked eye in order to survey close to the track line through both bubble windows. Both were connected by intercom to the data recorder and the pilot. To minimise bias due to observer fatigue, crew members rotated approximately every hour, with an off-duty observer on board. Data were collected using the SAMMOA software (SAMMOA 2014), which provides an audio recording that can be used for post-validation. Observers recorded all environmental conditions impacting the detection of animals: sea state (Beaufort scale), turbidity (from 0 for clear water to 2 for turbid water, where animals are only visible very close to the surface), cloud cover (0–8 octas), glare angle (0–360°) and severity (from 0 with no glare to 3 a strong glare that affects the ability to detect sightings), and a subjective ranking of sighting conditions (excellent, good, medium, poor). These covariates were recorded for each transect and whenever any of them changed. Observers scanned the sea surface and sub-surface as well as the air column searching for all targeted taxa. Maximum collection distance from the transect was 200 m for seabirds, jellyfish, plankton (as Trichodesmium mats) and marine debris using a strip transect methodology, assuming that all available individuals were seen within the 200 m on either side of the transect, while for boat sightings two 500 m strip were considered. Both distances were marked on the landing gear. In addition, surfacing flying fish species, from Exocoetidae family, were counted using two manual counters and aggregated by section of transect. For marine mammals, sea turtles, and elasmobranchs, perpendicular distance was measured with clinometers when the animal was abeam. Number of individuals in each sighting, and presence of juvenile were also recorded. All taxa were identified to the lowest taxonomic level possible. Several taxa that could not be distinguished from the air, such as small sized delphinids, hard-shelled turtles or ‘grey’ terns, were grouped together for analyses (see Supplementary Table S1–S4 for the list of taxa). In this study, only sightings of marine mammals, seabirds, sea turtles and large fish were considered.

The aircraft stayed on the track line except when circle-back manoeuvres were engaged in order to check identification, group size or to collect digital images for some specific marine mammal sightings. The effort was stopped for a few minutes and resumed when the aircraft rejoined the track at the stopping point. Flights were scheduled only in good weather condition (Beaufort scale <4) with the Zygrib interface (www.zygrib.org) for visualisation of NOAA-GFS standard weather forecast grib files.

Data analysis

A total of 167 422 km of transects were flown across the three subregions. To visualise the spatial distribution of identified taxa, the number of individuals encountered per kilometre was mapped on a 60 km × 60 km grid (to optimise effort homogeneity among cells).

For density estimates, only effort and sightings collected under good conditions (sea state <4 Beaufort and subjective condition equal to, or better than, ‘Medium’ on both sides of the aircraft) were retained thereby limiting variation in perception bias.

The strip transect methodology for seabirds assumes perfect detection within 2 m × 200 m bands (each side of the aircraft). For other taxa, sightings with larger perpendicular distances were truncated (≈5% of sightings) and excluded (Buckland et al. 2001), and detection functions were fitted. The effective strip widths (ESW) were estimated for all species or groups of species using Distance sampling methodology (Thomas et al. 2010), with the R package PelaCDS (ver. 0.1.0.9002) from Pelaverse package environment¹. Data from all three subregions were pooled, as these surveys share the same protocol, training procedures and partially the same observers across subregions.

Hazard rate and half normal models were fitted to perpendicular distances, and for each taxa the model that minimised the Akaike Information Criterion (AIC) was selected as the detection function (see Supplementary Fig. S1). Multiple Covariate Distance Sampling analysis (MCDS, Marques and Buckland 2004) was tested using sea state, turbidity, cloud cover and glare severity as covariates. Mean group size and encounter rate were estimated for each bathymetric strata for seven marine mammal taxa, eleven seabird groups, three elasmobranch groups and the hard-shelled sea turtles. And then densities were re-estimated (weighted according to strata area), as the coefficient of variation, following Buckland et al. (2001).

The detection probability on the track line is typically composed of the perception and the availability bias. No specific protocol was implemented to estimate the perception bias and it was kept as constant as possible (standardised protocol, same aircraft types, partly same observers). Regarding the availability bias, information on diving behaviour and the proportion of time spent visible at surface of the different species was collected from the literature and used as the correction (Thomson et al. 2012). Correction factors applied to density estimates were assumed to be equal to 1 for seabird taxa and summarised on Table 2 for other groups. The same proportions were used as in the previous study in the south-west Indian Ocean (see Laran et al. (2017a), Table S1); supplemented by an availability correction factor of 0.834 for dugongs estimated in New Caledonia, using a global value for all water depths (Hagihara et al. 2016, 2018) and 0.25 as the mean percentage of time spent in the top 5 m for more than 60 individuals of whale sharks (Rhincodon typus) (Andrzejaczek et al. 2022).

To examine regional patterns in the marine megafaunal community, we compared relative densities of 22 different taxa, corrected (for availability bias only) in six survey areas considered as sample sites: New Caledonia, Wallis and Futuna, and French Polynesia divided among Austral, Society, Tuamotu-Gambier (TUA-GAM) and Marquesas. For each, the community assemblage was investigated with Principal Component Analysis (PCA) using the multivariate analysis R package FactoMineR of R 2.7 (Lê et al. 2008).

Marine mammals

A total of 19 species of marine mammals were identified at the species level (or genus level for Kogia spp) from 602 sightings (5169 individuals) across the three subregions (see Table S1 for details by species and subregion). Results were pooled into taxonomic groups, with the exception of the dugong, because of the general difficulty of identifying species from the air. Selected taxa were: Balaenopteridae, Ziphiidae, Kogiidae, Physeteridae, small Delphininae (S. longirostris, Stenella attenuata, Delphinus spp) large Delphininae (Tursiops truncatus, Lagenodelphis hosei, Steno bredanensis, T. aduncus), small Globicephalinae (Peponocephala electrea, Feresa attenuata), large Globicephalinae (Globicephala macrorhynchus, Pseudorca crassidens, Orcinus orca or the well distinctive Grampus griseus). The minimum diversity was encountered in Austral, with only five distinct taxonomic groups, increasing north-ward and west-ward to 18 groups in New Caledonia, including four species of baleen whales and the dugong. Delphininae account for a quarter of the sightings, except in Wallis and Futuna (41%). Deep divers, such as Physeteridae, Kogiidae and Ziphiidae accounted for one quarter of sightings in Wallis and Futuna to one-third in the other two subregions (Fig. 2). Occurrences of Globicephalinae were about the same in all subregions. Balaenopteridae were seen from October to January in New Caledonia and Wallis and Futuna, but not at all in French Polynesia, which was surveyed later, between January and early May.

Fig. 2.

Taxonomic composition of sightings of marine mammals (on transect of effort) collected over the three subregions.

Marine mammals, in terms of individuals per kilometre of effort, exhibited a patchy distribution in French Polynesia, favouring nearshore waters, with the exception of the western part of Marquesas, where cells with the maximum index of individuals per kilometre were found even offshore (Fig. 3). In New Caledonia and Wallis and Futuna, their distribution is more homogeneous. Among taxonomic groups, deep divers were the most evenly distributed with juveniles (about 1/3 to 1/2 the size of adults) encountered in all subregions (Fig. S3). Higher densities of Ziphiidae and Kogiidae were estimated in the Marquesas survey block, with 0.08 and 0.06 individuals/km², respectively (corrected density, Table 3). Large and small Delphininae preferred waters around islands, except in Wallis and Futuna and west of the Marquesas (French Polynesia) where they were also encountered offshore. Juveniles were encountered in all three subregions (Fig. S3). Wallis and Futuna had the highest corrected density of small Delphininae with 0.05 individuals/km² (CV = 31%), twice that of New Caledonia, while the density of large Delphininae peaked in New Caledonia (0.03 individuals/km², CV = 38%). The encounter rate of Globicephalinae peaked west of New Caledonia, in the Coral Sea (west of Grande Terre Island, where several sightings included juveniles), and west of the Marquesas, but the higher densities were estimated in Wallis and Futuna (0.07 individuals/km², CV = 55%) and New Caledonia (0.06 individuals/km², CV = 55%), with values at least three times higher than in central and northern French Polynesia (Table 3).

Fig. 3.

Spatial distribution of individuals visually detected per kilometre for marine mammals in the three Pacific’s subregions: New Caledonia, Wallis and Futuna and French Polynesia during REMMOA surveys. Each pooled on a 60 km × 60 km grid cell and given in individuals per kilometre. Cells accounting for a minimum effort of 50 km.

Cumulative density (corrected for availability) of marine mammals was minimal in the oligotrophic waters of Austral (Fig. 4). It increased towards the north of French Polynesia, culminating in the Marquesas, where it was similar to that of New Caledonia and, to a lesser extent, Wallis and Futuna. New Caledonia had the more diverse composition with similar estimates for Delphininae, Globicephalinae, and deep divers (Ziphiidae, Physeteridae and Kogiidae). Wallis and Futuna was dominated by small Delphininae and Globicephalinae (mainly Risso’s dolphin, melon-headed whales and pygmy killer whales). And among the deep divers, Longman’s beaked whales (Indopacetus pacificus) were particularly abundant, while in the Marquesas Ziphiidae and Kogiidae were dominant.

Fig. 4.

Cumulative densities (×10² individuals/km²) across the six distinctive survey blocks of the tropical Pacific Ocean, corrected for availability bias considering averaged proportion of time spend at surface for (a) cetaceans, (c), turtles, sharks and rays; while (b) seabird densities were not corrected. (NC, New Caledonia; WF, Wallis-Futuna; and for French Polynesia AUS, Austral, SOC, Society; TUA-GAM, Tuamotu-Gambier and MAR, Marquesas).

Seabirds

Of the 25 842 seabird sightings (73 310 individuals), 4094 could be assigned to single species and 16 011 were identified at the genus level (eight genera). The remaining sightings (5737) could only be identified at higher taxonomic levels (Table S2). Species with only one sighting record were considered to be rare and were excluded from tables and analyses. For consistency, sightings were pooled into 11 different taxonomic groups: ‘Brown’ terns, ‘Grey’ terns, White terns, Noddies, Hydrobatidae, ‘Grey’ petrels, ‘Brown’ petrels, Shearwaters, Sulidae, Phaethontidae, Fregatidae and silver gull (Larus novaehollandiae) for New Caledonia only (see Tables 4 and S3 for detail on species).

Taxonomic diversity appeared to be roughly similar across subregions, with most categories found in all three subregions (Table 4). Encounter rate of seabirds (in individuals/km) showed contrasting patterns, with high values in the north of the TUA-GAM survey block (French Polynesia) and the south-west coast of New Caledonia (Fig. 5). Similarly, the cumulative density of seabirds peaked in TUA-GAM, followed by Marquesas, New Caledonia and Society survey blocks (Fig. 4). New Caledonia hosts the highest densities of shearwaters and ‘Grey’ petrels, with densities decreasing along a west-east longitudinal gradient, decreasing by one or two orders of magnitude in Wallis and Futuna and French Polynesia (Table 4; Fig. S4). This presence seems to be rather coastal and coincides with the breeding of petrels (Tahiti petrel, Pseudobulweria rostrata; black-winged petrel, Pterodroma nigripennis and Gould’s petrel, P. leucoptera) or shearwaters (wedgetailed, Ardenna pacifica; short-tailed shearwaters, Ardenna tenuirostris) breeding or migrating during the beginning of the monitoring in December. Conversely, noddies, white terns, Sulidae and Fregatidae were most abundant in French Polynesia compared to western subregions. Phaethontidae stood out as the only group with homogeneous densities in all three subregions.

Fig. 5.

Spatial distribution of individuals visually detected per kilometre for seabirds in the three Pacific’s subregions: New Caledonia, Wallis and Futuna and French Polynesia during REMMOA surveys. Each pooled on a 60 km × 60 km grid cell and given in individuals/km. Cells accounting for a minimum effort of 50 km.

Elasmobranchs

A total of 702 sightings (905 individuals) of elasmobranchs were collected during line transect efforts (Table S4). The highest occurrence (individuals/km) was observed along the west coast of New Caledonia and west of Marquesas (Fig. 6), while only a few encounters were observed in the south of French Polynesia. Both Marquesas and New Caledonia have similar densities of rays (Table 3), while shark density peaked in New Caledonia, then in TUA-GAM (Fig. 4).

Fig. 6.

Spatial distribution of individuals visually detected per kilometre for elasmobranchs (sharks including whale shark and rays), in the three Pacific’s subregions: New Caledonia, Wallis and Futuna and French Polynesia during REMMOA surveys. Each pooled on 60 km × 60 km grid cell and given in individuals per kilometre. Cells accounting for a minimum effort of 50 km. Sightings of whale sharks were plotted (pink diamond).

Only a few conspicuous taxa could be identified at the species level from the air, notably the whale shark, only encountered as single individuals. Although encountered in all three subregions, whale sharks were particularly abundant along the west coast of Grande Terre (New Caledonia) and in the waters of Wallis and Futuna (surveyed from October to January; Fig. 6), where the density was highest (Table 3).

Sea turtles

A total of the 260 sea turtle sightings (equivalent to 287 individuals) were collected during the transect effort. Only three were identified as leatherback turtles (Dermochelys coriacea) (Table S4), one in each of the three subregions, indicating the scarcity of this relatively distinctive species, at least during the survey period.

Hard-shelled sea turtles were encountered in all subregions, mostly in coastal waters, where the index of individuals per km peaked (Fig. 7). Global density of hard-shelled sea turtles was highest in New Caledonia, followed by Wallis and Futuna and TUAM-GAM, although densities were almost four times lower (Table 3, Fig. 4).

Fig. 7.

Spatial distribution of individuals visually detected per kilometre for hard-shelled sea turtles (with sightings of leatherback turtles) in the three Pacific’s subregions: New Caledonia, Wallis and Futuna and French Polynesia during REMMOA surveys. Each pooled on a 60 km × 60 km grid cell and given in individuals per kilometre. Cells accounting for a minimum effort of 50 km. Sightings of leatherback turtles were plotted (brown diamond).

Megafauna assemblages

When considering cumulative corrected densities for all taxa (Fig. 4), consistent general patterns emerged, with the western-most subregion (New Caledonia) peaking for marine mammals, turtles, and elasmobranchs, the Marquesas for marine mammals and the TUA-GAM for seabirds only.

From the PCA conducted over the six survey blocks representing the three subregions and including densities of 22 different vertebrate taxa (seven marine mammal groups, hard-shelled sea turtles, three groups of elasmobranchs and 11 seabird groups), biogeographical provinces were well isolated and the latitudinal gradient was highlighted. PCA axis 1 explained 40% of the variance and discriminated New Caledonia with the dugong, large Delphininae, silver gull, ‘grey’ petrels, shearwaters, ‘grey’ terns and hard-shelled turtles, apart from other subregions (Fig. 8), as visible on the species correlations for the first dimensions of the PCA (Fig. S5). PCA axis 2, discriminated the northern of French Polynesia (Marquesas) from the southern (Austral), accounting for 26% of the variance with higher contribution of Ziphiidae, Kogiidae, Fregatidae, Hydrobatidae and rays associated with Marquesas. The next two following axes explained 17% and 13% of the variance respectively, with association of Wallis and Futuna with small Delphininae, Globicephalinae, whale shark, ‘brown’ tern and Phaethontidae while TUA-GAM were associated with Sulidae, noddies and sharks, contrary to the Austral archipelago associated with Physeteridae.

Fig. 8.

Result of the first two axes (a) and the third and fourth axes (b) of the Principal Component Analysis obtained with the corrected densities of 22 groups of megafauna over six distinct survey blocks of the three Pacific subregions. Survey blocks discriminate by axes are underlined and associated species (contribution to the PCA dimension greater than 9) are represented by silhouette (from https://www.phylopic.org/).

Marine mammals

Interestingly, the corrected abundance estimate (for availability bias only) for dugongs in our study is 660 individuals (CV: 32%), which is consistent with previous estimates from dedicated aerial surveys ranging from 649 to 898 (CV: 30% and 25%, respectively) in November 2011 and 2012 (Cleguer et al. 2017). Given that the sampling design of the REMMOA survey was not dedicated to lagoon habitats and that we applied a coarse availability bias correction, it is likely that using a more refined bias correction, with sighting location and specificity of each protocol could reduce the difference between these results.

Prior to our REMMOA survey, only four marine mammal species were known to occur in the waters around Wallis and Futuna (Doremus et al. 2010; Miller 2023). The present study allowed 13 new species or genera to be added to this list (Table S1). The distribution of this remarkable diversity denotes a widespread use of the EEZ by marine mammals.

Of the 15 odontocete species already reported in the French Polynesia subregion (Reeves et al. 1999; Gannier 2000, 2002, 2004, 2009; Gannier and West 2005; Gannier and Petiau 2006; Oremus et al. 2012), 14 were recorded during the REMMOA aerial surveys in 2011. We recorded some species in archipelagos where they had not been observed before, especially around the Austral and Gambier Archipalago, which were poorly documented before (cf. Table S1). Our standardised survey effort allows for the first time to clearly visualise the strong latitudinal gradient for Delphininae and Globicephalinae, confirming the previously mentioned high diversity of the Marquesas (Gannier 2009). The waters of the Society Archipelago show a high number of individuals per kilometre for marine mammals, and is confirmed by coastal transects carried out around some islands during the REMMOA survey (but not included in the present analyses as overlapping spatially (see Laran et al. 2012) are consistent with previous results (Gannier 2000). Our study emphasises the importance of the Society Archipelago for cetacean conservation, particularly for spinner dolphins or rough-toothed dolphins which show site fidelity between groups and little movement between archipelagos (Oremus et al. 2007, 2012). When considering densities, corrected for availability bias, Kogiidae and Ziphiidae are almost as abundant as Delphininae and Globicephalinae combined in the Marquesas. In addition, juveniles were observed in half of the 13 sperm whale encounters, suggesting the Marquesas Archipelago is a nursing ground.

Seabirds

All three subregions studied in this study include important seabird colonies. However, comparing published figures for breeding populations with the abundances estimated at sea is not straightforward. Indeed, many open-ocean seabird populations include abundant non-breeding segments (immatures and sabbatical adults) that are not bound to colonies and can therefore wander over large stretches of the ocean. As a consequence, seabirds observed at sea in any of the three subregions considered here can be composed of a combination of local breeders foraging at sea, a fraction of local non-breeders foraging in the area surrounding their colony of origin, and a number of non-breeders from other regions of the Pacific and Indian Oceans, keeping in mind that the latter may outnumber the former. Therefore, an exact match between abundance at sea and the size of local colonies cannot be expected. In contrast, past land-based seabird monitoring has been heterogeneous in the three subregions studied, both spatially (uneven geographic coverage) and temporally (intermittent survey efforts). Available data are generally scarce, leaving considerable uncertainty about the distribution, abundance and phenology of breeding seabird species, and limiting the comparability of census data. Therefore, this information on seabirds at sea is less accurate than direct counts at colonies, but is still important because it provides information on major foraging areas and may make it possible to identify areas of increased presence that may correspond to unknown colonies or resting sites.

There are two main limitations in this study. First, the phenology of the species encountered and the short timeframe of the survey. Tropical seabird communities generally host species with both seasonal and aseasonal breeding activity (e.g. Baudat-Franceschi 2011, 2012), therefore observations collected over short periods of time (here 2–4 months) may not reflect the full diversity and/or abundance of breeding species in a given region. Second, in relation to their specific foraging behaviour, with some having both parents travelling short distances multiple times per day (e.g. brown noddy, Anous stolidus; roseate tern, Sterna dougallii) (Maxwell et al. 2016; Pratte et al. 2021), while others have parents that alternately travel away from the colony over several days (e.g. wedge-tailed shearwater, Tahiti petrel; Ravache et al. 2020; Weimerskirch et al. 2020).

The REMMOA surveys covered approximately one-third of the New Caledonian EEZ. Due to practical and logistical constraints, the survey tracks did not include the Chesterfield-Bampton submarine plateau and its archipelago, which supports a large number of breeding seabirds (Spaggiari et al. 2007; Borsa et al. 2022). Observations showed a wide use of the surveyed areas by seabirds, with Procellariidae and Sulidae species mostly found in pelagic waters, according to their known foraging distances (Ravache et al. 2020; Weimerskirch et al. 2020). With the exception of ‘brown’ terns, encounters of terns and noddies were higher near the mainland, probably reflecting their shorter foraging distances. Shearwaters were as expected, the most abundant species group, as the wedge-tailed shearwater, is by far the most abundant species among New Caledonia’s breeding seabirds (0.5–1 million breeding pairs; (Baudat-Franceschi 2006, 2011, 2012; Barré et al. 2011). Short-tailed shearwaters, may have also been observed passing through New Caledonian waters during their seasonal migration, although they do not breed locally (Baudat-Franceschi 2012). ‘Grey’ petrels (mainly Gould’s petrel, and black-winged petrel) were found in slightly higher numbers than expected, with an estimated abundance of 23 000–42 000 individuals compared to an approximate total colony size of 1100–10 100 breeding pairs (Spaggiari et al. 2007). Among the Laridae, the ‘brown’ terns (mainly sooty terns (Onychoprion fuscatus), and a few bridled terns (Onychoprion anaethetus) were the most abundant species group (95% IC: 43 000–160 000 individuals), but their estimates are lower than the estimated population size (~300 000 individuals for the sooty tern, Barré et al. 2011). This is probably due to the absence of Chesterfield Reef from our study area, where 90 000–100 000 breeding pairs of sooty terns are concentrated (Baudat-Franceschi 2011). Sulidae, Fregatidae and noddies show similar discrepancies, while the numbers of ‘grey’ terns (roseate tern, black-naped tern, Sterna sumatrana; fairy tern Sternula nereis; and greater-crested tern, Thalasseus bergii) are close to land-based estimates, as all species breed mostly near the mainland in the areas included in the survey (Baudat-Franceschi 2006, 2012; Berr et al. 2020, 2021, 2022). White terns (here Gygis alba) and Phaethontidae (Phaethon lepturus and Phaethon rubricauda) are the only species for which there are important differences between at-sea observations and land-based counts that cannot be explained by the extent of the surveyed area. Both had much higher numbers in aerial surveys (5000–13 600 and 7500–14 700 individuals, respectively) than in land observations (~1500 and ~2000 suspected breeding pairs respectively; Barré et al. 2011). This discrepancy may reflect the cryptic on-land behaviour of the two species groups (seasonal breeding by isolated pairs in either cliffs or tall trees) and/or foraging by individuals from other locations (Fiji, Vanuatu) in New Caledonian waters.

Published data on seabirds from the Wallis and Futuna archipelago are scarce, as the only reliable monitoring campaigns were conducted in 1985–1986, 2014 and 2018–2019 (Thibault et al. 2014, 2015; Berr and Mathivet 2018; Berr 2019). For species groups except noddies, the abundance estimates derived from the REMMOA surveys are much higher than the land counts. The estimates for noddies (95% IC: 13 500–80 000 individuals) are consistent with surveys conducted in 2018 which reported a colony of 15 000–30 000 breeding pairs on islets north of the Wallis lagoon (Berr and Mathivet 2018). For other species, the estimated at-sea densities suggest that a large number of non-breeding seabirds use the waters of Wallis and Futuna for foraging or pass through them during migration. The frequent observations of shearwaters at sea (95% IC:8900–24 100 individuals) may explain the temporary presence of mixed species nesting on Fiji Islands (Watling 2001), as procellariid species are thought to have been nearly extirpated from all the islands of Wallis and Futuna (Thibault et al. 2014; Berr and Mathivet 2018; Berr 2019).

The geographical extent of French Polynesia (2.5 M km² EEZ) and the number of islands (>100 main islands and >1000 islets in total) contribute to the scarcity of survey data on most seabird breeding sites (Thibault and Bretagnolle 1999). Therefore, density estimates derived from the REMMOA surveys provide some new information on seabird distribution and abundance. White terns and noddies are the dominant species group among Polynesian seabirds, as they are found at all of our survey blocks, with particularly high densities in the northern Tuamotu archipelago. Sulidae (mainly red-footed, Sula sula and brown boobies, Sula leucogaster) are also abundant in the Tuamotu archipelago, in agreement with previous land-based surveys (Thibault and Bretagnolle 1999). The high abundance of seabirds in the northern Tuamotu and southern Marquesas can be linked to the distribution of their prey: within French Polynesian waters, epipelagic micronekton biomass is highest in these two areas (Bertrand et al. 2002). In contrast, the southern Austral Islands have very few birds compared to the rest of the EEZ, similar to other megafaunal species.

Elasmobranchs

Approximately 50 shark species are known to occur in the waters of New Caledonia, where they inhabit reef-lagoon and open-ocean ecosystems throughout their life cycle (fishbase.org, Boussarie et al. 2018). Previous studies using baited cameras and environmental DNA analyses revealed higher diversity and density of sharks in remote areas, particularly in northern New Caledonia, Chesterfield and d’Entrecasteaux (Bakker et al. 2017; Juhel et al. 2018). The REMMOA surveys highlight a distribution of elasmobranchs throughout the Grande Terre with a clear preference for the west coast. The most common ray species encountered during the flights was the spotted eagle ray (Aetobatus narinari) while among manta rays, the reef species (Mobula alfredi) is dominant in New Caledonia (Lassauce 2021). This corresponds well with the location of sightings inside the lagoon during the survey in New Caledonia, unlike the other two subregions (see https://pelabox.univ-lr.fr/pelagis/PelaObs/ for species distribution).

Knowledge of elasmobranchs in the Wallis and Futuna EEZ is also very limited, but 10 species of sharks and five species of rays inhabit the waters around the islands (Williams et al. 2006). These include several pelagic species such as whale shark, longfin shark (Isurus paucus) and manta ray. Several sightings of devil rays (Mobula spp) were also collected during the survey. In contrast to New Caledonia, no hammerhead sharks (Sphyrna spp) were observed, but 20 whale sharks were detected during transects. Higher encounter rates (of individuals) were estimated around Wallis Island (including the lagoon). Pelagic and migratory species such as whale sharks and Mobula spp were ubiquitous in the oceanic and shallow waters north of the surveyed area. Noting that aerial surveys detect only animals at or near the sea-surface, when all species were combined, nearly 4400 sharks (95% IC: 1400–14 300) and about 900 rays (95% IC: 320–3000) were estimated for the surveyed area (corrected for availability bias only), all species combined. For the endangered whale shark, the encounter rate recorded in the Wallis and Futuna subregion is the highest of all surveyed areas covered by the REMMOA project. This area may represent a favourable zone for juveniles, as most sightings were of small individuals (5 m or less). However, none of our three subregions were highlighted in the global map of whale shark hotspots, revealed by spatial distribution of tracked animals (Womersley et al. 2022).

Since 2006, the entire French Polynesian EEZ has been declared a shark sanctuary for all species (Decree No. 396 of the Environmental Code 28 April 2006 and the mako shark (Isurus spp) since 2012). The sanctuary prohibits fishing and trade in products of all fish belonging to the taxon Elasmobranchii, with the exception of rays (Ward-Paige 2017). Approximately 30 species of elasmobranchs have been reported through citizen science in the Marquesas (Séguigne et al. 2023). The Tuamotu-Gambier survey block has a density of sharks equivalent to New Caledonia, mainly encountered around the Tuamotu Archipelago. Regarding rays, this study highlighted a high density of manta rays in the Marquesas Archipelago, where the oceanic manta ray, and the reef manta ray, co-occur (Carpentier et al. 2019). Despite the protected status, there are concerns about bycatch in fisheries and tourism, including the development of feeding sites (Carpentier et al. 2019).

Sea turtles

In New Caledonia waters, leatherback sea turtle presence had only been mentioned in the Loyalty Archipelago by a telemetry study (Benson et al. 2011), and to our knowledge, the REMMOA survey reported the first sighting within the Grande Terre lagoon. The aerial surveys occurred during the nesting season of the critically endangered North-east Indian Ocean and East Pacific Ocean subpopulations (Petro et al. 2007; Robinson et al. 2014), which may result in underestimation of leatherback turtle presence. The REMMOA survey was conducted at the beginning of the nesting season for green (Chelonia mydas) and loggerhead (Caretta caretta) turtles, which occurs from late October to late March (Read et al. 2013; Fournière et al. 2015; Barbier et al. 2023; Fretey et al. 2023). Both species migrate between foraging and nesting areas as part of their life cycle (Read et al. 2014). There are no estimates for these species in the area to compare with our corrected abundance of 14 600 individuals (95% IC: 10 600–21 200) excluding the Chesterfield and Bellona Reefs, recently identified as major breeding rookeries (Fretey et al. 2023). REMMOA therefore provides a much needed first assessment at sea of the sea turtle community in New Caledonia.

Only the green turtle has been regularly observed in the Wallis and Futuna area, with no mention of nesting sites on the main islands (SPREP 2007) but a rising awareness of potential sites on insular, small and scattered nesting grounds (Berr 2019; B. Bouchard, pers. comm.). Our study complements the knowledge on the distribution of hard-shelled sea turtles in the area. They were mainly encountered found in the Wallis lagoon, and less frequently around Futuna and Alofi (neither of which has a lagoon) or offshore. The presence of the leatherback sea turtle was also attested with two encounters in the north of the study area.

Five species of sea turtles are in French Polynesia waters, green, hawksbill (Eretmochelys imbricata), leatherback, loggerhead and olive ridley turtle (Lepidochelys olivacea). Scientific knowledge on the distribution of sea turtles is limited, the size of the territory and the dispersion of the islands being a major obstacle. The two regularly reported species are: (1) the green turtle, which nests from November to April, with Tetiaroa (included in the Society surveyed block) being the largest known nesting site across the Windward Islands in the Society Archipelago (Petit et al. 2013; Laloë et al. 2020); and (2) and the hawksbill turtle. The flights took place at the end of the green turtle nesting season. Two other important nesting rookeries are monitored and classified as territorial reserves: Scilly and Bellinghausen Atolls (out of the study area). In addition to bycatch threats, destruction of nesting sites, pollution, and predation of hatchlings by dogs, rats or other introduced predators (Gronwald et al. 2019), the consumption of turtle meat is a traditional practice still on going in this subregion (Brikke 2009; Rudrud 2010), while prohibited by the environmental regulations of French Polynesia (arrêté 466 CM of 22 March 2018).

Conservation of marine megafauna

Marine megafaunal species are diverse and charismatic. They provide a source of protein for human populations and/or support regional economies and cultures and are an integral part of biodiversity. These species are valuable as ecological indicators and may play pivotal roles in assessments of marine ecosystem health (Estes et al. 2016). They influence their ecosystems in various way as consumers, prey, detritus sources, and nutrient vectors, and their impacts extend to all species through direct and indirect interactions over large time and space scales from coastal to oceanic food webs, providing top-down effects on marine food webs that can control community stability (Estes et al. 2016). Most megafaunal species are long-lived and slow to reproduce making them particularly vulnerable to all sorts of environmental change, including climate change, whose effects on vertebrates have consequences for marine ecosystem structure and services (Sydeman et al. 2015). Large-scale climate phenomena such as the ENSO, the major year-to-year climate signal on earth involves both the ocean and atmosphere. This variability in oceanographic mechanisms affects ecological and biological processes, and in turn foraging patterns, breeding performance and thus population dynamics of marine top predators (Nowicki et al. 2019). Oceanic and atmospheric conditions in the tropical Pacific fluctuate somewhat irregularly between warm El Niño phases and cold phases called La Niña in which surface waters cool across the tropical Pacific. Among the three subregions, French Polynesia was sampled by the REMMOA survey during a weak to moderate La Niña event, while Wallis and Futuna and New Caledonia were sampled during weak El Niño conditions (2007–2020 ONI values from https://psl.noaa.gov/data/correlation/oni.data; cf. Fig. S6). This difference in general oceano-climatic conditions could have an effect on the distribution of megafauna and on our observations. Long-term datasets are crucial to assess extreme climatic events and potential impacts of climate change, but few are available due to logistical difficulties and financial costs (e.g. Simmonds and Isaac 2007; Bost et al. 2015). In our project, the three subregions were surveyed only once, with no opportunity to test for the effects of extreme climatic events or global change. However, in a context of rapid change in marine ecosystems, our results provide a large-scale baseline, but the need for more regular sampling in the future is critical for monitoring change in the ocean.

There is a trend to establish MPAs based on their marine megafauna, and particularly mammals or seabirds, and to use top predators as ecological indicators or focal species (Hooker and Gerber 2004; Tetley et al. 2022). Knowledge of the distribution and abundance of megafauna is important to better understand and manage the impacts of anthropogenic activities. The ability to monitor progress towards such goals has been constrained by the lack of robust data on MPAs. A multi-species monitoring approach provides an opportunity to collect data on multiple taxa simultaneously, which can improve cost-effectiveness and help to better understand and predict how taxa are likely to respond to or be affected by disturbances. In addition this approach can identify areas where more dedicated surveys would be important in term of the unknown presence of a species, a hot spot or human interactions.

MPAs are increasing in number and size around the world as a tool for pelagic conservation (Game et al. 2009). However, most MPAs have been identified as too small to encompass the complete home ranges of most highly mobile species, as shown by tracking data for five megafaunal taxa (Conners et al. 2022). Our multi-taxa study may also inform future designations. Indeed 30 by 30 worldwide initiative was agreed at the COP15 meeting of the Convention on Biological Diversity (https://www.cbd.int/meetings/COP-15), to designate 30% of Earth’s land and ocean areas as protected by 2030 and there is now a global call to increase the coverage of MPAs. France is a major actor in marine conservation due to its numerous overseas territories scattered across the Oceans, and especially in the three subregions surveyed in our study, which account for 64% of the French EEZ (French Polynesia, 47%; New Caledonia, 14%; Wallis and Futuna, 3%). Nonetheless, according to regulation-based classification system (Horta e Costa et al. 2016), France currently uses a case specific, loose approach to define strong protection, with no criteria for the representativeness of protection across the French Ocean basins (Claudet et al. 2021). Due to their status as overseas local authorities, French Polynesia, New Caledonia and Wallis and Futuna have full authority to manage and protect the environment in their EEZ (Schaffar 2011). A standardised recurring protocol could help to assess some species of marine megafauna (marine mammals, seabirds, sea turtles and large fish) as in the case, for example, in the European the Marine Strategy Framework Directive.

More recently, the Important Marine Mammal Areas (IMMAs) initiative, launched by the Marine Mammal Protected Areas Task Force of the International Union for the Conservation of Nature (Tetley et al. 2022), identifies discrete portions of habitat that are important to one or more marine mammal species, and have the potential to be delineated and managed for conservation. REMMOA surveys have contributed to the designation of several IMMAs, which were identified through a biocentric expert process independent of political and socio-economic pressure or concerns (Notarbartolo di Sciara and Hoyt 2020). Four IMMAs have been identified in New Caledonia, and three in French Polynesia, while the one for Wallis and Futuna is still under revision (www.marinemammalhabitat.org/imma-eatlas/). Similar initiatives are underway for sharks and rays (Important Sharks and Rays Areas, ISRA, https://sharkrayareas.org/) and sea turtles (Important Marine Turtles Areas, IMTA). The aerial multi-taxa protocol used here represents a powerful tool for measuring marine megafauna abundance and potential long-term changes by repeating this protocol on a regular basis.