The region of interest is located in the SW Indian Ocean, from the Northern Seychelles Islands, to the Mascarene Islands, and the Mozambique Channel extending over 5 million km² from latitude 2–24°S and longitude 40–60°E (Figure 1). It encompasses extensive parts of the Exclusive Economic Zone (EEZ) of the five countries of the Indian Ocean Commission (Comoros, Madagascar, Mauritius, La Réunion, and Seychelles). Our study area covered three distinct contrasting ecoregions of the Longhurst classification: the East African coastal province, the Indian South Subtropical Gyre, and the Indian Monsoon Gyre regions (Longhurst, 1998, Figure 1). Effort was split into six survey blocks totaling 1.4 million km² that were selected on ecosystemic and logistical grounds (Figure 1; Table 1) with: (1) three survey blocks located along the west coast of Madagascar in the Eastern African ecoregion: the Northern Mozambique Channel block (NMC, 275,000 km²) around the Comoros Archipelago and the Glorieuses Islands (France) which was covered from mid-December to the beginning of January; the Central Mozambique Channel (CMC, 123,000 km²), which includes Juan de Nova Island (France) and the Southern Mozambique Channel (SMC, 153,000 km²) with Europa Island and Bassas da India atoll (France), both of which were covered from end of January to the beginning of February; (2) one survey block including in the Seychelles the Granitic Islands (central position on the Seychelles Bank) and north of Amirantes Islands, representing the productive Indian Monsoon Gyre (SE, 294,000 km²), all covered from end of March to beginning of April; and (3) the last two survey blocks covered the oligotrophic Indian South Subtropical Gyre from the end of February to the end of March: these spanned the Tromelin Island (France) to the East Coast of Madagascar (TM, 153,000 km²) and around the Mascarene Islands (MAS, 407,000 km²) including La Réunion (France) and the Island of Mauritius including its outer Island, S^t Brandon. Each block was sub-divided into bathymetric strata: shelf (<200 m), slope (from 200 to 2,000 m) and oceanic (>2,000 m) strata (Figure 1). This subdivision resulted in a total of 22 strata.

The survey was conducted during the austral summer (December 2009–April 2010), in order to guarantee good conditions for aerial sighting detection. In consequence some species such as large migratory whales undertake annual migration to high latitude summer feeding areas and have been almost not encountered during the survey. But for some other highly migratory species the chosen survey period represents a key explanatory parameter in interpreting their distribution and relative abundance in each survey block.

A total survey line effort of c. 89,000 km was conducted among survey blocks and strata in such a way that habitats expected to have lower density of top predators (Chla-depleted oceanic waters) would receive proportionately more effort than habitats with high expected cetacean densities. In addition search effort was optimized with a zigzag track layout. To minimize logistical constraints for planes and increase effort, several transects were flown twice or more but only effort collected for two replicates of the same transect were retained for the analysis. No ethics approval was needed as the research did not involve animal subjects.

Surveys were carried out following a standard line-transect methodology (Buckland et al., 2001) with aircraft speed (167 km h⁻¹/90 knots) and altitude (182 m/600 feet) similar to previous large-scale aerial surveys dedicated to marine mammals in European waters (Hammond et al., 2013) or to megafauna (e.g. Laran et al., in press). This approach represents a good compromise between safety constraints (related to low-level flights) and the choice to extend the survey to non-mammal taxa including seabirds, sea turtles, and large elasmobranchs. Nonetheless, seabird sightings were collected following a strip-transect methodology in order to minimize disrupting the attention of observers. Transects were flown using high-wing aircraft (BN2) equipped with bubble windows. The survey crew consisted of two trained observers searching with the naked eye and a navigator collecting data on a laptop computer equipped with “VOR” software developed for the aerial part of the SCANS-II survey (Hammond et al., 2013). The aircraft's position was recorded every 2 s using an on-board GPS device. Beaufort Sea state, glare severity, turbidity, cloud cover, and subjective sighting conditions (an overall subjective assessment of the detection conditions: good, moderate, or poor as for small Delphinids) were recorded at the beginning of each transect and whenever any of these parameters changed. A fourth, off duty crew member was also present to enable the rotation of crew members every hour to limit any loss of vigilance due to observer fatigue. Perpendicular distances obtained from clinometers were collected by observers for marine mammals, sea turtles, and elasmobranchs. For seabird data collected in strip transect mode, all encounters located within 200 m of the aircraft's track line (marked on the landing gear) were assumed to be detected. Species identification was made to the lowest taxonomic level whenever possible, but groupings were inevitable for several taxa that could not be told apart from the air. For marine mammals nine groups were considered: small Delphininae, large Delphininae, small Globicephalinae, Risso's dolphin, large Globicephalinae, beaked whale, sperm whale, Kogia spp., and dugongs (cf. Supplementary Table 2 for the list of species). For seabirds: seven groups comprised brown terns, gray terns, noddies, petrels and shearwaters, tropicbirds, boobies, frigatebirds (cf. Supplementary Table 3 for the list of species). Finally a “hard-shelled group,” mainly green turtles in the area (Bourjea, 2015) and leatherback turtles, were considered in addition to manta rays, unidentified rays, whale sharks, hammerhead sharks, and unidentified sharks (cf. Supplementary Table 4 for the list of species).

Individuals encountered were mapped across the six survey blocks for the main taxa over a grid of 60 × 60 km to optimize homogenous effort among cells, in order to visualize their distribution as determined by the aerial survey in the region. From the detection data on each species in each cell, cumulative taxonomic richness was modeled with occupancy models (MacKenzie et al., 2002) over the entire region using averages of depth, slope, and distance to 200 m isobaths within each cell. Occupancy modeling was undertaken in order to better describe any hotspots for megafauna diversity in the area.

For density estimation, 83,726 km of line transect effort were retained for analysis. This effort was mainly conducted in slope (36%) and oceanic strata (52%) and under good sea state (95% of the effort in Beaufort conditions ≤3, 74% with Beaufort conditions of 0–2) thereby limiting variation in perception bias. The strip transect methodology for seabirds implied perfect detection within 2 × 200 m bands (i.e., one on either side of the aircraft). For other taxa, sightings with larger perpendicular distances were truncated (≈5% of sightings) and were excluded (Buckland et al., 2001). Detection curves and effective strip half-widths (esw) were estimated for eight marine mammal taxa, five elasmobranch taxa, the hard-shelled group, and leatherback sea turtles using Distance sampling software (Thomas et al., 2010). When the effect of meteorological conditions (sea state, glare severity, or subjective sighting condition) or group size were statically significant (at 5% level), Multiple Covariate Distance Sampling was used (Marques and Buckland, 2004). The best models were selected using the Akaike Information Criterion (AIC). Mean or regressed pod size against g(x) (when significant), was estimated for each survey block and each strata within a block, if significant (Z-Test).

Detection probability on the track line, g(0), is typically composed of the perception bias (the proportion of animals available for detection at the surface on the transect line but missed by the observers) and the availability bias (the proportion of animals present on the transect line but not available for detection at the visible subsurface or surface). We only corrected for the latter using crude estimates from the literature (Penguiness book web site, Ropert-Coudert and Kato, 2012): the average diurnal proportion of time spent at the surface for the different relevant species or groups of species were used as a correction factor for availability bias (Table 2 and Supplementary Table 1 for details).

To examine regional patterns in the megafauna community we compared, as sampling stations, 22 spatial units corresponding to bathymetric strata with a sufficient amount of effort (i.e., >900 km) among the survey blocks (Figure 1; Table 1). For each stratum the community assemblage was investigated with Principal Component Analysis (PCA) using the ADE4 package for R, after centering and standardizing the estimated densities of 18 groups of megafauna: eight cetacean groups, five seabird groups, three groups of elasmobranchs, and two of turtles. Hammerhead sharks were included within the sharks group and manta rays with rays.

Nineteen marine mammal taxa were identified during the entire survey (Supplementary Table 2). The highest diversity was encountered in the North Mozambique Channel (NMC) and Seychelles (SE) blocks, with 14 taxa. Predicted occupancy over the entire region suggested hotspots of taxonomic richness in the Mozambique Channel and Seychelles with more than five taxa in most of the cells, compared to only 2–3 taxa on average in the Mascarene (MAS) and Tromelin (TM; Figure 2A) regions surveyed.

Higher numbers of individuals per unit effort were also obtained in the Mozambique Channel and Seychelles (Figure 3A). There were substantial variations among survey blocks: Delphininae and Globicephalinae were more prevalent in the Mozambique Channel and the Seychelles than in the Mascarene. In contrast deep divers (i.e., sperm whales, Kogia spp. or beaked whales) did not show much difference between survey blocks (Supplementary Figure 1). The dugong (Dugong dugon) was recorded exclusively along the northwestern coast of Madagascar and one sighting in the Comoros (NMC).

Across the eight detection functions derived for cetacean species, subjective sighting condition, and glare severity were the only two statistically significant covariates (Supplementary Figure 3). Higher densities of marine mammals (uncorrected for availability bias) were found in the Seychelles and Mozambique Channel blocks (Table 3). The density of Delphininae peaked at 0.10 individual km⁻² in Mozambique Channel and Seychelles, with small Delphininae avoiding the SMC and large Delphininae the NMC. Small Globicephalinae reached densities >0.10 individual km² in the NMC and CMC blocks, but densities of large Globicephalinae peaked in the Indian Monsoon Gyre survey block (SE; 0.07 individual km⁻²). Beaked whale densities showed less contrasting densities between survey blocks with a maximum occurrence in the Mozambique Channel. Densities of sperm whales peaked in the TM and SE survey blocks. Finally the density of Kogia was highest in the SE survey block (0.002 individual km⁻²).

A total of 12 seabird species or groups of species were identified (Supplementary Table 3). Apart from the Hydrobatidae all groups were recorded in every survey block. Seabird taxonomic richness patterns contrasted with those of marine mammals (Figure 2B). Although, seabirds were abundant across the whole region occupancy modeling predicted the Seychelles as a hotspot, while the middle latitudes of Madagascar (CMC and TM) were a “colder” spot. With respect to sightings of seabirds, the highest number of individuals per unit of effort was encountered in the central Mozambique Channel (CMC; Figure 3B). Extensive variations were found between survey blocks for most taxa (Table 4, Supplementary Figure 2). “Brown” terns (Onychoprion spp.) were detected in all surveyed areas but peaked in CMC with a density of 4.3 individual km⁻². “Gray” terns (Sterna spp., Thalasseus bergii, Gygis alba) were preferentially encountered in shelf waters off Madagascar while including G. alba in the Seychelles resulted in a more oceanic distribution pattern. “Gray” tern densities over the Mozambique Channel and the Seychelles survey blocks (0.5–0.9 individual km⁻²) were 10 times higher than east of Madagascar (Table 4). Noddies had their strongholds in Seychelles, while Procellariidae were the most abundant seabird grouping around La Réunion Island. Tropicbirds, boobies, and frigatebirds showed densities ≤0.05 individual km⁻² at their highest levels.

Only a few conspicuous taxa could be identified at the species or genus level from the air (e.g., whale sharks Rhyncodon typus, manta rays Manta birostris spp., or leatherback turtle Dermochelys coriacea; Supplementary Table 4). Five different detection functions were estimated for elasmobranchs and two for marine reptiles (Supplementary Figure 3).

Elasmobranchs were more frequently encountered in the Mozambique Channel and the Seychelles (Figure 3C). Whale sharks were only reported in the Central and North Mozambique Channel and east of Madagascar (TM), with densities (uncorrected for availability bias) of 0.05–0.1 × 10⁻² individual km⁻² (Table 5). Hammerhead and “other sharks” were encountered in all survey blocks, with pooled estimated densities varying from almost nil in the Mascarene to 1.2 × 10⁻² individual km⁻² in the Central Mozambique Channel. Manta and other rays were found in all survey blocks with maximum densities in the Mozambique Channel and the Seychelles (Table 5).

Leatherback turtles were almost only found in the Central Mozambique Channel where their density (uncorrected for availability bias) reached 0.18 × 10⁻² individual km⁻², whereas hard-shelled turtles were encountered in all survey blocks, mostly in coastal waters (Figure 3D) and with maximum densities estimated in the North and Central Mozambique Channels (NMC and CMC, Table 5).

In order to account for extensive inter-taxon differences in time spent visible at or close to the surface, corrections for availability bias were made for cetaceans, sharks, rays, and turtles. When considering cumulative corrected densities for all taxa, consistent general patterns emerged. The Central Mozambique Channel block emerged as the primary regional density hotspot for pelagic megafauna, ranking first for all general megafauna categories: cetaceans, seabirds, elasmobranchs and sea turtles (Figure 4). This region was followed in order of importance by the other blocks of the Mozambique Channel as well as the Seychelles. In contrast, survey blocks to the east of Madagascar always ranked low for any broad categories of megafauna. More nuanced results emerged when considering more specific sub-categories of megafauna, as seen for procellariids in the Mauritius and La Réunion survey blocks, or for sharks and hammerhead sharks in the East of Madagascar-Tromelin (Figure 4).

From the PCA conducted across the 22 stations (Figure 5) with densities of 18 distinct vertebrate taxa (corrected for availability, except for seabirds), the first axis (PC1; 23% of total variance) discriminated between two geographic groups: the eutrophic east of Madagascar (MAS and TM) associated with procellariids and tropicbird densities, and the oligotrophic Mozambique Channel and Seychelles associated with sharks and “brown” tern densities. PC2, which accounted for 21% of the variance. This was partly explained by an offshore-inshore gradient and it was negatively correlated with densities of Kogia spp., large Globicephalinae, Risso's dolphins, and small Delphininae but positively correlated with densities of elasmobranchs and sea turtles. The Mascarene Archipelago was discriminated from other strata and characterized by higher Procellariid and tropicbird densities. A high density of Kogia singled out the slope habitat around the Seychelles. Megafauna assemblages (with respect to their densities) were thus clearly structured across the SW Indian Ocean.

In visual surveys of pelagic megafauna, potential survey biases are of at least three types: availability bias, perception bias, and responsive behavior. For the latter we made the assumption that, given altitude and speed of the aircraft, species could not react prior to being in the observers' field of vision. Regarding availability and perception biases, a double platform protocol could estimate both parameters for a selection of species, but this approach was not feasible during our survey. Except for seabirds, availability bias was tentatively corrected on the basis of dive pattern, using the average proportion of time spent close to the surface and considering that the sighting time-window (≈4 s) was instantaneous relative to the duration of most surface/dive cycles. Nevertheless, to improve the relative density estimates of several species, using availability bias from the scientific literature can be improved upon, for example by using dive pattern data (a function of season, bathymetry, time of the day, etc.).

Regarding perception bias, variations in detectability introduced by the sighting condition were dealt with through daily selection of transect lines to achieve excellent sighting conditions most of the time. Secondly the effect of weather-related covariates on detection functions, when statistically significant, was accounted for in six out of 14 taxa. Nevertheless, due to spatio-temporal heterogeneity in both dive patterns and bottom color, particularly in shallow waters, additional availability, and perception biases probably affect detection rate, particularly for sea turtles, dugongs (Hagihara et al., 2014), or coastal Delphininae. It is currently unclear whether these biases would on average be negative or positive.

For seabird data collected via strip transect methods, in the absence of an available and accepted methodology we did not attempt any correction and therefore our results should be considered as conservative estimates. Nonetheless, orders of magnitude of seabird densities across the three survey blocks of the Mozambique Channel (NMC, CMC, and SMC, 2.9 individual km⁻²) were similar to previous summer ship-survey estimates in the same area (Jaquemet et al., 2014).

The distribution observed for cetaceans generally concur with previous findings. Of the 25 species of odontocetes already encountered in the region (Marsh et al., 2003; Best, 2007; Kiszka et al., 2009a), we observed 18 during our survey. Cetacean studies in this area have mostly relied on vessel-based surveys in coastal areas and only a few have been conducted in offshore areas (Ballance and Pitman, 1998; Dulau-Drouot et al., 2008; Kiszka et al., 2008) and none of these previous surveys allow a comprehensive comparison with the present study. The distribution of the vulnerable dugong along the northwest coast of Madagascar was highlighted during this study, and a relative abundance of 100 dugongs was estimated without any correction factors incorporated (Laran et al., 2012). We also demonstrated the first occurrence of Kogia spp. in the central and south Mozambique Channel and of the common bottlenose dolphin around Tromelin Island.

Occupancy modeling predicted marine mammal diversity to peak south of the Mozambique Channel and, more moderately, around the Seychelles. Regarding marine mammal relative abundance, except for migratory baleen whales the present work provides a summer snapshot of taxonomic richness in the whole region. Salient results included the preponderance of Delphininae in the Central and South Mozambique Channel, especially the larger Delphininae species such as the bottlenose dolphin. The northern part of the Mozambique Channel was dominated by small Globicephalinae, mainly the melon-headed whale. In the East of Madagascar to Tromelin, the cetacean community composition was more even, with low densities for all taxa with the exception of sperm whales for which large summer aggregations have previously been reported north and east of Madagascar (Kasuya and Wada, 1991). Deep divers were found in all survey blocks with maximum densities in the Mozambique Channel for beaked whales and around the Seychelles for Kogia spp.

The SW Indian Ocean represents a region of major importance for seabirds with 31 breeding species accounting for an estimated 7.4 million pairs without considering non-reproductive segments of the populations or non-breeding species in the area. These are concentrated in the Seychelles, the Mozambique Channel and the Mascarene archipelago (Le Corre et al., 2012). The on-land distribution of the main breeding colonies can clearly explain most of our results. Most at-sea studies of seabirds in the SW Indian Ocean have relied on telemetry, and five major oceanic foraging hotspots are identified from which three include the major breeding colonies listed above (Le Corre et al., 2012). The comparison of seabird relative abundance at sea with colony counts is not straight forward nor easy, notably because breeding phenology varies between sites (Jaquemet et al., 2007). However, ranking our survey blocks according densities at sea is consistent with the ranking of the major breeding sites on the basis of colony counts; with Juan de Nova (Central Mozambique Channel) first, followed by Europa (South Mozambique Channel) and the Seychelles. However, our survey block in the north of the Mozambique Channel did not encompass the other major breeding site of Aldabra. A good correspondence between our Figure 2B was obtained with an earlier map of species richness obtained from the distribution of seven tracked species (Le Corre et al., 2012), except for northwest of Madagascar—Comoros and Seychelles, where we predicted a higher seabird taxonomic richness.

Conducting the survey in the summer had a significant impact on observed seabird distribution and density estimates, because of species—and sometimes site-specific breeding phenology and migration patterns. For example, around Europa sooty tern (Onychoprion fuscatus) winter density was estimated to 4.7 individual km⁻² from a vessel (Jaquemet et al., 2005), while we obtained 1.5 individual km⁻² during our summer aerial survey, but with a greater proportion of Sterna spp. (“gray” terns) than in the ship-based survey (Jaquemet et al., 2005). This could result either from a differential perception bias between dark- and light-backed species or from seasonal changes in their distribution. Nevertheless, in the Central Mozambique Channel our maximum density of “brown” terns (mainly sooty terns) accounted for 88% of the total seabird density in this survey block, in relative agreement with species composition of the main breeding site where sooty terns represent 99% of breeding seabirds (Le Corre and Jaquemet, 2005). Our estimates of “brown” tern relative abundance obtained in all survey blocks, except in the Seychelles where sooty terns breed from July to October then disperse, were strongly correlated (r²: 0.96) with the corresponding numbers of breeding pairs of sooty terns (Feare et al., 2007).

The main limitation of aerial surveys for seabirds lies in species identification: groupings based on visual similarity may lead to ecologically heterogeneous categories, or the main species encountered might also vary between survey blocks but this not be captured by broader group-based categorization.

The present work provides the first large scale and synoptic survey for sea turtles and elasmobranchs in the SW Indian Ocean. Previous knowledge mostly came from surveys of sea turtle breeding beaches, coastal observations, or telemetry (Lauret-Stepler et al., 2007; Dalleau et al., 2012; Bourjea, 2015). For elasmobranchs, except for whale shark (Rowat et al., 2009), the main sources of data in the SW Indian Ocean have come from fisheries monitoring, local photo-identification programmes, and a few telemetry studies (Kiszka et al., 2009b). Most of these species are notoriously difficult to monitor from a vessel due to their generally inconspicuous surface behavior. Therefore, quantitative comparisons between regions or species are not readily feasible. However, higher densities of sharks and rays previously found in the Mozambique Channel (Couturier et al., 2012), or whale shark occurrence around the north of Madagadcar (Rowat, 2007) are consistent with some of our results. Similarly, the distribution of hard-shelled sea turtles observed during our survey is consistent with breeding sites in the area (Dalleau et al., 2012; Bourjea, 2015) or satellite tracking showing transit across the open ocean with clear preferences for foraging sites in shallow, euphotic coastal waters (Hays et al., 2014). By comparison, the sizeable densities of the critically endangered leatherback turtle encountered in the shelf strata of West coast of Madagascar confirm its coastal occurrence as observed along South African coast and Mozambique (Robinson et al., 2016). If individuals nesting in South Africa forage in the study area (Robinson et al., 2016) their relationships with known nesting sites need to be explored.

The pelagic megafauna dealt with in this work included a variety of charismatic species that share a number of important conservation issues. Among the species encountered 13 are listed as critically endangered, endangered, vulnerable, or nearly threatened on the IUCN Red List, and nine are listed as data deficient (Supplementary Tables 2–4). Among marine mammals the populations of three species (Indo-Pacific bottlenose dolphin, Indo-Pacific humpback dolphin, and dugong) are either decreasing or fragmented in the SW Indian Ocean. And several species of seabirds are mostly restricted to the region, in particular two endemic petrels of La Réunion Island: the Barau's petrel (Pterodroma baraui) and the Mascarene petrels (Pseudobulweria aterrima).

Various anthropogenic threats have been identified for marine megafauna in in the SW Indian Ocean, such as incidental or direct catches in fisheries for marine mammals, elasmobranchs, or sea turtles (Kiszka et al., 2008; Cerchio et al., 2009; Amandé et al., 2010; Bourjea et al., 2014), the direct utilization of turtle and seabird eggs (Cheke, 2001), and underwater noise arising from oil, gas or other geophysical exploration (Southall et al., 2013). However, fisheries bycatch has been identified as the primary driver of population declines in several species of marine megafauna (e.g., elasmobranchs, mammals, seabirds, and turtles; Lewison et al., 2004). Heavy mortality in fisheries has dramatically reduced shark numbers in many locations (Zydelis et al., 2009). Worldwide more than 200,000 loggerhead turtles (Caretta caretta) and 50,000 leatherbacks were likely taken as pelagic long line bycatch in 2000 (Lewison et al., 2004). No bycatch estimates are currently available for all taxa of marine megafauna in the SW Indian Ocean; nevertheless, the Western Indian Ocean represent 13% of the reported global sea turtle bycatch (Wallace et al., 2010). In contrast, there is little bycatch of seabirds in the tropical SW Indian Ocean, presumably since most tropical seabirds feed upon small epipelagic prey and thus do not interact directly with fisheries (Le Corre et al., 2012). Nevertheless, seabird-specific threats include interactions with human activities on land (Le Corre et al., 2002), the introduction of alien predators into the breeding colonies (Dumont et al., 2010), plastic pollution (Wilcox et al., 2015), or interaction with tunas or cetaceans for foraging (Jaquemet et al., 2005) sometimes referred to as “near-obligate commensalism” (Au and Pitman, 1986) which could be at risk of disruption if tunas become overfished (Le Corre et al., 2012).

Conservation actions are often species-oriented because the ultimate measure of their efficacy is evaluated by assessing the conservation status of the species of interest. However, when it comes to the value of conservation strategies that deal with vast components of biodiversity at large geographical scales, metrics estimated at the community level are more relevant (Edgar et al., 2014). The present work provides baselines both for taxon-oriented approaches, acknowledging the intrinsic strengths and weaknesses (see Section Correction of Specific Biases) of the survey method, and for community-oriented metrics. The analysis of megafauna density assemblages across the SW Indian Ocean firstly reflected the general pattern of the three ecoregions of the Longhurst classification and, secondly, rough habitat distinctions based on bathymetry (shelf, slope/oceanic). Up to 44% of the observed variance in the study could be explained by these two parameters but there are many other factors, in addition to measurement error due to imperfect detection which might account for the remaining 56%. The existence of adequate breeding grounds in the vicinity of favorable foraging habitats (specific to seabirds and sea turtles), and spatial heterogeneity in historic human pressures could account for a large proportion of the remaining variance. The metrics describing pelagic megafauna in the tropical SW Indian Ocean would be extremely useful to monitor the long-term impacts of general management strategies, even if the present baseline does not depict a pristine condition.

There is increasing support for large scale marine protected areas (MPAs) as a tool for pelagic conservation (Game et al., 2009). In the SW Indian Ocean, a large MPA was implemented in 2010 around the Chagos Archipelago (60,000 km²) and more recently France extended the MPA of Mayotte to the EEZ of the Glorieuses Islands, representing another 110,000 km². There is a trend to establish marine sanctuaries based on their marine megafauna, and particularly their mammal or bird fauna, and to use higher predators as ecological indicators or focal species (Hooker and Gerber, 2004). However, single-species conservation plans will probably not ensure the conservation all co-occurring species. Yet multi-species strategies, based on systematic selection procedures (e.g., a suite of “focal species”) offer more compelling evidence of their usefulness (Roberge and Angelstam, 2004) and value. Aerial surveys collecting data on different taxa over large areas and in rapid, repeatable time-frames can provide valuable knowledge with greater cost-effectiveness than single species monitoring.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.