A major conservation priority for giant manta ray recovery is to improve our understanding of movement and seasonal distribution patterns to inform future management measures for minimizing impacts to the species during key life history functions (NOAA 2020). Within its global range, giant manta rays inhabit tropical, subtropical, and temperate bodies of water and are commonly found offshore, in oceanic waters, and near productive coastlines (Marshall et al. 2009; Kashiwagi et al. 2011; Pate & Marshall 2020). Our analysis of decades of manta ray sightings across several different aerial survey platforms in the EUS indicated manta rays were most commonly detected at frontal boundaries in productive nearshore and shelf-edge upwelling zones within a thermal optima of approximately 15–30°C (Fig. 5). These findings are consistent with ecologically-driven expectations that manta rays would be more common near areas of potentially high prey densities (upwelling zones) within thermally optimal conditions. For the Gulf of Mexico, peak occurrence and observations were clustered off the Mississippi River delta, an area of known high concentrations of large zooplankton (Figs. 3 and 7 in Shropshire et al. 2020). Shropshire et al. (2020)’s predictive maps also illustrate the importance of nearshore habitats as a potential food source for manta rays. Similarly, zooplankton biomass estimates by Strömberg et al. (2009) suggest high concentrations of potential manta ray prey in the Mississippi River plume, Florida coastal waters, the upwelling zone near Cape Hatteras, North Carolina, and the northeastern United States shelf-edge areas covered by the NYSERDA surveys.
It is noteworthy that > 99% of all rays (all species) observed by the NYSERDA surveys were observed in the spring/summer, despite nearly equal levels of survey effort in the fall/winter. Environmental temperature directly dictates body temperature for most elasmobranchs (Huey and Kingsolver 1989; Deutsch et al. 2008; Schulte et al. 2011). Many physiological rates scale with temperature according to a thermal performance curve, with performance gradually increasing up to an organism’s optimum temperature, and then quickly declining as temperatures approach lethal levels (Huey and Stevenson 1979; Angilletta et al. 2002; Angilletta 2006, Lear et al. 2019). Manta and devil rays routinely exhibit basking behaviors presumably to elevate body temperatures after making excursions into deeper, colder habitats (e.g. Thorrold et al. 2014). Cold winter air and sea surface temperatures in the western North Atlantic Ocean likely create a physiological barrier to manta (and other) rays that restricts the northern boundary of their distribution.
Juvenile elasmobranchs appear to have wide performance curves characteristic of thermal generalists, which may allow them to survive and succeed in shallow coastal areas subject to more rapid temperature fluctuation and higher maximum temperatures (Gilchrist 1995; Kingsolver 2009; Lear et al. 2019). By contrast, adults may have steeper, narrower thermal performance curves to achieve high levels of performance once they gain the ability to migrate long distances and select habitats with preferred thermal characteristics (Lear et al. 2019). Globally, the species has been observed in estuarine waters near oceanic inlets (Adams and Amesbury 1998; Milessi and Oddone 2003; Medeiros et al. 2015). Potential manta nursery habitats have been identified at FGBNMS in the Gulf of Mexico and in the nearshore waters of southeastern Florida (Stewart et al. 2018; Pate & Marshall 2020). The use of warmer waters at smaller sizes is consistent with bioenergetic theory, as optimal temperature for growth is higher for smaller fish (Morita et al. 2010). Unlike other areas within the survey domain, southeastern Florida maintains water temperatures within the predicted optimum year-round, minimizing requirements for movement, which may explain repeated resightings across multiple years for specific juveniles (Pate & Marshall 2020). Similarly, FGBNMS mean temperatures remain in the 20–30°C range year-round (see Fig. 8.1 in Johnston et al. 2018).
In this study, SST was the strongest single predictor of manta distribution, with a strong thermal preference apparent between 17 and 32°C, with a peak around 23°C. Temperature preference appears to vary by region, with manta rays off the U.S. east coast commonly found in waters from 19 to 22°C and those off the Yucatan peninsula and Indonesia between 25 and 30°C (Duffy and Abbott 2003; Marshall et al. 2009; Freedman & Roy 2012; Graham et al. 2012; Hacohen-Domené et al. 2017). Garzon et al. (2020) found Chl-a, bathymetric slope, and SST were important drivers of manta distribution in the Western Central Atlantic (WCA); however, SST was the least important driver in their model, possibly due to the lower variability in SST in the WCA relative to the U.S. east coast.
The combined surveys model indicated highest probability of occurrence at moderately-sloped nearshore and shelf-edge habitats with moderate SST fronts and high concentrations of Chl-a; all proxies for high primary production and associated manta prey availability (Fig. 5). Due to the lack of spatially comprehensive zooplankton and micronekton sampling data, we were unable to explicitly test associations between manta rays and prey availability. However, strong associations were observed across data sources between manta ray sightings and proxies for productive upwelling zones. Within thermally-optimal bounds, models predicted higher concentrations of manta rays from the coast to the shelf south of Cape Hatteras, approximately corresponding with the inside edge of the Gulf Stream current. Offshore, the inside edge of warmer Gulf Stream waters passing the continental shelf provides a consistent source of upwelling and productivity (Blanton et al. 1981). Similarly, nearshore tidal fronts provide mixing and nutrient concentrations to support high concentrations of potential prey items (Savidge 1976; Pingree et al. 1978; Tett 1981). The offshore, northbound flow of the Gulf Stream is offset by the nearshore, southbound counter-current flow (Bumpus & Wehe 1949), and may provide a bioenergetically favorable ‘conveyor belt’ for filter-feeding manta rays to efficiently forage while remaining within thermally-optimal conditions. Satellite-tagging studies are needed to evaluate individual movement patterns and quantify population connectivity.
We observed a dome-shaped relationship with temperature and significant seasonal and interannual trends in the centroid of manta distributions, suggesting that the population may be impacted by climate change similar to swordfish (Brodie et al. 2021) and North Atlantic right whales (Record et al. 2019). If relatively prey-rich northern waters continue to warm, the overall distribution may continue to shift northwards (Figs. 4, 8, Supplemental Video). Warming SST leads to more stable water columns, enhancing stratification and requiring more energy to mix deep, nutrient-rich water into surface layers (review in Richardson 2008). Nutrient limitation in phytoplankton growth in the ocean is negatively related to temperature globally (Kamykowski & Zentara 2005). In response to global warming, poleward shifts in zooplankton distribution coupled with changes in abundance and community structure are anticipated (Richardson 2008). Changes in climate and oceanographic conditions, such as acidification, are also known to affect zooplankton structure (size, composition, diversity), phenology, and distribution (Guinder and Molinero 2013). The major impact of climate change on mobulids is likely to be the projected decline in zooplankton in tropical waters (Stewart et al. 2018). Future research should identify manta prey in the EUS, determine the environmental drivers of their prey distribution, and evaluate how those distributions are likely to shift under climate change scenarios.
Retrospective analyses of NYSERDA and discussions with aerial observers (Table 1) suggested that M. mobular and M. tarapacana were frequently misidentified as giant manta rays, especially north of Cape Hatteras, North Carolina. We addressed this uncertainty by only including photographically-verified sightings north of Cape Hatteras in our modeling efforts. Interviews with observers and reviews of 100s of photos suggested extremely low misidentification rates south of Cape Hatteras; however, photos were not available for all sightings and observer aerial survey guides were not developed until after the listing. As such, our SDMs may to some extent represent a shared habitat utilization of mobulid species, heavily weighted towards manta rays. The Normandeau Associates/APEM digital photo archives for NYSERDA and BOEM data allowed us to compare sightings locations for confirmed species identifications of large mobulids, including manta rays, and revealed substantial overlap in species habitat utilization along the U.S. East Coast’s continental shelf from South Carolina to New York (see Supplemental File). Due to a lack of distinguishing dorsal features, we were unable to account for the putative third species or subspecies of M. birostris (M. sp. cf. birostris sensu Marshall et al. 2009) resident in the Gulf of Mexico (Clark 2002; Hinojosa-Alvarez et al. 2016; Hosegood et al. 2020) and possibly southeastern Florida (Pate & Marshall 2020). Given possible hybridization with M. birostris (Hosegood et al. 2020), and apparent overlap in distribution, genetic, and morphological similarities, this possible misidentification may have limited impact upon modeled manta distributions. Without genetic testing, species identification cannot be completely validated (Hinojosa-Alvarez et al. 2016; Kashiwagi et al. 2017; Hosegood et al. 2020). The distribution of M. birostris and M. sp. cf. birostris sensu (Marshall et al. 2009) may be of similar management interest due to the inability to readily distinguish between the two. If thermal tolerances vary between the two, our SDMs might over- or under-predict giant manta ray distributions in some regions. Similarly, our model does not account for any differences in depth utilization (e.g., availability bias); if manta rays spend proportionally more time in waters below the visual observation depth at particular locations, the model would underpredict their utilization of those areas.
In the United States, NOAA’s National Marine Fisheries Service (NMFS) is charged with promoting the recovery of giant manta rays. Under Sect. 7 of the ESA, agencies must consult with NMFS to ensure their proposed actions do not jeopardize the survival of listed species. Understanding the distribution, abundance, migration patterns, and site fidelity of manta rays is essential for accurately estimating the impact of proposed activities. Anthropogenic impacts to individuals can be estimated as the product of: (i) the probability of an activity occurring in an area; (ii) the probability of an individual being in an activity area, expressed as a distribution model; (iii) the duration of exposure of the individual to the activity; and (iv) the probability of the activity impacting the individual, often expressed as a dose-response curve. Our SDMs will help managers determine when an effect of a proposed action on giant manta rays is likely based on the probability of an action taking place when giant manta rays are anticipated in the area. Even when duration of exposure and probability of adverse effects are unknown, relative risk assessments can be used to identify preferred alternatives, following Farmer et al. (2016). To more accurately determine anticipated take of giant manta rays from proposed actions, further information is needed on movements, site fidelity, depth utilization, and responses to anthropogenic stressors.
The most significant threat to the recovery of giant manta rays across their range is intentional harvest and bycatch in fisheries (Miller and Klimovich 2017). Manta rays are targeted or caught as bycatch with virtually every fishing gear type, including small-scale fisheries characterized by the use of driftnets, gillnets, harpoons, gaffs, traps, trawls, and longlines; and large-scale fisheries using driftnets, trawls and purse seines (Croll et al. 2016). Our SDMs will help managers identify areas of spatial and temporal overlap between giant manta rays and commercial fisheries, which could be used to reduce bycatch rates in the EUS. Similarly, a better understanding of the spatiotemporal distribution of the species may help improve precision of bycatch estimates by controlling for relative availability of the species to the gear on any given set and allocating observer coverage to areas of higher bycatch concern. For example, preliminary analysis of 2019–2020 data estimated mean EUS shrimp trawl take of manta rays of nearly 1700 individuals/yr (Carlson 2020); however, uncertainty was very high given the short time series and limited data. Although observer coverage on shrimp trawls in the EUS is around 1%, the relative observer coverage on shrimp trawls with trawl effort spatially weighted by manta probability of occurrence was less than 0.09% (Farmer NA, unpublished data).
A major NMFS recovery priority for giant manta rays is to investigate the impact of other threats to the species (e.g., foul-hooking, vessel strikes, entanglement, climate change, pollution, tourism) through research, monitoring, modeling, and management (NOAA 2020). Pate & Marshall (2020) and our SDMs suggest that manta rays are frequently associated with nearshore habitats; as such, they are at elevated risk for exposure to a variety of contaminants and pollutants, including brevetoxins, heavy metals, polychlorinated biphenyls, and plastics (Essumang 2010; Ooi et al. 2015). Many of these toxins can bioaccumulate over decades in long-lived filter feeders, leading to a disruption of biological processes (e.g., endocrine disruption), and potentially altering reproductive fitness (Germanov et al. 2019). Coastal and lagoon habitats are especially sensitive to habitat degradation, pollution, and sedimentation (McCauley et al. 2012, 2014).
There is a strong management interest in understanding the inshore extent of manta movements in bays and tidal inlets. SDM predictions suggest seasonal trends with high probability of occurrence in large bays (e.g., Tampa Bay, Chesapeake Bay); however, reported sightings in bays are extremely limited. It is unclear if this is due to reduced water clarity, rarity of use, or very low levels of survey effort. Both Medeiros et al. (2015) and Bucair et al. (2021) report manta ray utilization of bays and inlets in Brazil, and we verified several anecdotal reports of use of shallow tropical bays in the U.S. Caribbean. Future efforts will seek to evaluate EUS nearshore sightings relative to currents, tidal phase, and salinity. Manta rays are frequently reported in nearshore environments of southeastern Florida (Pate & Marshall 2020) and somewhat regularly in the U.S. Caribbean. Georgia Aquarium and partners recorded high numbers of manta rays between St. Augustine Inlet, Matanzas Inlet, and Jacksonville, Florida during dedicated aerial surveys in 2010–2017 (Table 1); however, the timing and frequency of these observations was variable both seasonally and interannually, with the peak only lasting a few weeks. SDMs capture these trends (Supplemental Video 1), predicting a spring and fall peak in the survey area with the spring peak varying between March and June.
Florida, and southeastern Florida, specifically, has the highest number of registered recreational vessels and licensed recreational anglers in the U.S., and likely the world (USCG 2019). Manta rays are exposed to exceptionally high levels of vessel traffic as well as hook-and-line fishing gear from boats and piers. Manta rays are often observed foraging on tidal outflows at major inlets in the Palm Beach County area, leading to frequent overpasses by vessels moving at high speeds (Pate & Marshall 2020). Casting in the vicinity of large mantas is a major component of the recreational cobia fishery along most of the Florida coast (Roberts 2020, Pate J, unpublished data, Farmer NA, unpublished data). Pate and Marshall (2020) documented fishing line entanglement on 27% of individuals in Palm Beach County, Florida, along with vessel strike injury and rapid wound healing. Pate et al. (2021) identified a need for outreach in the Palm Beach County area, focused on preventing recreational fishery interactions with manta rays, encouraging use of environmentally-friendly tackle, and fostering engagement with anglers as citizen scientists. Our SDMs will allow managers to more effectively time and coordinate management strategies, including targeted outreach efforts and developing spatiotemporal ‘windows’ for action agencies to reduce the risk of manta interactions.
A major priority for manta ray conservation is to improve understanding of population distribution, abundance, trends, and structure through research, monitoring, and modeling (Stewart et al. 2018; NMFS 2020). Our preliminary presence-absence modeling approach uses distance-weighted methods to control for perception bias. With further satellite tagging data on manta ray movements and dive profiles, we may be able to address availability bias for animals that are underwater, and apply similar methods to determine population abundance. Further genetic analyses are needed to resolve the taxonomic status and relative abundance of M. birostris and M. sp. cf. birostris sensu (Marshall et al. 2009). More telemetry studies are also needed to evaluate whether giant manta rays within the Gulf of Mexico and northwestern Atlantic Ocean constitute one large, mixed population, or exist as isolated subpopulations (Stewart et al. 2016). Observing rare species incurs a heavy cost in time, resources, and boat fuel. Our findings suggest that bathymetric maps combined with real-time satellite data may be used to effectively target manta rays for scientific study, including the attachment of satellite tags to inform movements, site fidelity, and dive patterns. Following Farmer et al. (2018), individual impacts can then be summarized across the population to evaluate population consequences of disturbance in the context of population recovery.