The animal use protocol for this research was reviewed and approved by the Institutional Animal Care and Use Committee at Old Dominion University (#866320-3). Procedures were approved under National Park Service Permit #DRTO-2016-SCI-006, and USGS Bird Banding Lab Permit #21780.

We studied brown noddies breeding on Bush Key in Dry Tortugas National Park, Florida (24.62817°N, 82.86423°W) in May 2016 over a 7-day period (Table 1). DRTO is 261.8 km², all of which does not allow commercial fishing, though recreational hook and line fishing is permitted in 53% of the park. DRTO is further surrounded by the Florida Keys National Marine Sanctuary (FKNMS), a 9844.5 km² multi-use marine protected area that extends largely to the east and north of DRTO (Figure 1). Two no-take areas managed by the FKNMS occur near DRTO: The Tortugas Ecological Reserve North (312.1 km²; TER North) and South (205.8 km²; TER South; Figure 1). The TER North is directly adjacent to DRTO and protects a coral reef area known as Sherwood Forest. While TER South is not part of the contiguous marine protected area complex, it protects Riley's Hump, a bank and deep-reef terrace that has high abundance and biodiversity (Dry Tortugas National Park, 2001).

Brown noddies nest primarily in low shrubs; we captured individuals either by hand on their nests, or using a hand net. Altogether we marked and weighed five chick-rearing pairs (total of 10 birds) in close proximity to each other to facilitate simultaneous observation. Birds were marked with a combination of size number three butt-end aluminum bands (USGS Bird Banding Lab) and plastic color bands for easy visual distinction between mates, though sex of individual birds was not determined. Additionally, we attached Nanofix GPS (Pathtrack Ltd, West Yorkshire, UK) tags using waterproof Tesa tape (Tesa Tape Inc., Charlotte NC) to the back two tail feathers, 1–2 cm behind the uropygial or preening gland (Figure 2). Tags were 2.4 by 1.1 by 0.5 cm with a thin antenna extending 4.8 cmănd weighing 1.4 g waterproofed, approximately 1.16% of the average body weight of the tagged birds (157.8 g, range 138–170 g). Birds showed no sign of discomfort from the tags; no bill marks were observed on the tape or tags, nor were birds ever observed to make contact with tags (pecking at them, etc.) during our nearly continuous monitoring during daylight hours.

We deployed tags in two rounds over the course of a week. During the first deployment, tags were programmed to transmit locations every 20 min and were deployed for ~4 days. When recovered after 4 days, we replaced the tags on the same birds, but we programmed the tags to transmit locations every 10 min and deployed the tags for ~2 days, resulting in ~6 days of continuous data between the two deployments. Total handling time for each deployment or recovery was ~5 min and never exceeded 10 min. We recovered all the tags successfully, and only one tag failed to record data, giving us a total sample size of 19 deployments for 10 birds. In addition to the GPS data, we visually recorded behavioral data of tagged individuals, including arrivals and departures from the colony; this occurred approximately every 15 min during daylight hours.

We determined basic movement parameters from tracking data including total distance traveled, straight line distance traveled for the longest track of each bird, travel speed, and arrival and departure times from nests and corresponding trip lengths. We conducted analyses using a combination of Matlab version 2013a (MathWorks Inc., Natick MA), ArcGIS version 10.3.1 (ESRI, Redlands CA) and R version 3.0.1 (R Core Team, 2003). Location intervals varied between the two deployments; as a result, tracks from the first deployment (20 min) were linearly interpolated to 10 min intervals using the “adehabitatLT” package in R (Calenge, 2006) so that that birds' locations occurred equally in time and the frequency of points did not bias subsequent analyses.

We determined the number of locations per bird and the number of locations where birds had high-residence time per bird that fell within boundaries of (1) the Dry Tortugas National Park, (2) the Florida Keys National Marine Sanctuary, (3) no-take areas (a combination of DRTO and the two FKNMS Tortugas Ecological Reserves), and (4) unprotected waters; the number of locations is also equivalent to the amount of time as each location represented 10 min. To illustrate a more general picture of the overlap of the entire population with MPAs, we determined the percent of the total home range that fell into each of the four areas for both the core (50% UD) and total (95% UD) home range.

A sharp drop off in the continental shelf, and the Loop Current, an extension of the Florida Current that later forms the Gulf Stream, both occur close to the nesting colony. During our study period, the Florida Current formed a distinct boundary near the shelf break with a marked change in sea surface temperature (SST) near the shelf break (see below). We determined the underlying SST as well as the distance to the 100 m shelf break for high- and low-residence time locations for each bird. To determine if locations where birds had high-residence time occurred closer to the shelf and in areas of higher SST than locations where birds had low-residence, we compared the metrics using a two-tailed Welch's t-test in R to account for unequal variances. Data was checked for normality, and bathymetry was square root transformed. Due to autocorrelation between points along the track increasing the likelihood of Type I error, we determined the effective sample size for distance to the shelf and SST for both high- and low-residence time locations by using the correlation coefficient (first order autocorrelation model) to adjust for sample size, resulting in a smaller sample size following Dawdy and Matalas (1964) and Young et al. (2015). Bathymetric data was downloaded from NOAA's National Center for Environmental Information 3 Arc-Second Coastal Relief Model Development (http://www.ngdc.noaa.gov/mgg/bathymetry/relief.html), modified to create contours by the Gulf of Mexico Coastal Ocean Observing System (http://gcoos.tamu.edu/). We used SST from the Visible and Infrared Imager/Radiometer Suite (VIIRS). We downloaded 4 km² resolution 8-day composite SST (8-15 May 2016) from NOAA's Ocean Color website (NASA Goddard Space Flight Center, 2016; http://oceancolor.gsfc.nasa.gov/).

Marine protected areas are a key tool for protection of biodiversity, habitats and species of importance (Roberts et al., 2002; Fox et al., 2012; Gormley et al., 2012; Klein et al., 2015). The only brown noddy colony in the continental US is found within one such complex of marine protected areas. These MPA boundaries encompass a large portion (between 60 and 90%) of the foraging range of the brown noddy during the chick-rearing period incorporated in our study. While birds are using the MPAs to a great extent, there are still opportunities to increase protections. Compared to all locations (91.2%), a smaller percentage (64.7%) of locations where birds had high-residence time—which indicate probable foraging areas—were found within the marine protected areas, and an even smaller percentage of locations where birds had high-residence time (6.7%) were found within no-take areas; thus large portions of likely important habitat are unprotected.

Commercial fisheries are permitted to operate beyond the boundaries of the DRTO and the Ecological Reserves of the FKNMS (including in the larger FKNMS), thereby overlapping with 20.8% of the total range of the brown noddy, and 93.3% of probable foraging areas (as indicated by locations where birds had high-residence time). Commercial fisheries in this region include shrimp trawling, trapping of spiny lobster and stone crab, and longline fishing, however our knowledge of fishery impacts on brown noddies in this region is limited. Noddies forage on a diversity of fish and squid species, including mackerels (Scombridae), halfbeaks (Hemiramphidae), and flying fish (Exocoetidae) (Hensley and Hensley, 1995). There are no reports of bycatch of brown noddies in fishing gear, but bycatch of species for which noddies forage is likely to be occurring, particularly with less discriminant fishing gear. Shrimp trawling in particular has high rates and diversity of bycatch; Nance et al. (1997) reports bycatch of mackerels elsewhere in the Gulf of Mexico along with a suite of other forage fish species, some of which are known to be prey of brown noddies. Cury et al. (2011) identify that an important management strategy includes protecting foraging resources; this is likely the most critical near to the colonies and during energetically costly times of the year such as chick-rearing.

Noddies have also been observed foraging on discards from commercial fisheries, particularly shrimp trawl fisheries (Oron Bass, personal observation). Hensley and Hensley (1995) reported the occurrence of three fish species (Lonchopisthus micrognathus, Bollmannia boqueronensis, and Diplectrum spp.) known as deep-water reef species in noddy diets. As noddies are unable to dive, and these species are never reported on the surface, they concluded the birds took these fish species from shrimp trawl discards as they are known bycatch species in this fishery. Thus, while MPAs in this region generally provide excellent coverage of brown noddy foraging areas, if protection of noddy foraging resources is a goal, managers should consider prohibiting fishing activities within additional parts of the MPAs. Prohibitions could be focused on particularly energetically important times of the year for noddies, such as breeding, and/or in regions where noddies focus foraging effort, such as the shelf break and Loop Current.

Important areas for brown noddies included the region near the continental shelf and adjacent to the Loop Current. Locations where birds had high-residence time were found in significantly higher SST regions of the Loop Current (Figures 1, 5B). Additionally, brown noddies whose trips were directed to high SST regions of the Loop Current had shorter maximum trip distances (Figure 4), indicating a high likelihood of successful foraging in those areas. Diet studies of this species have previously suggested that the brown noddy likely uses the Loop Current region (Hensley and Hensley, 1995). Noddies fed primarily on species considered to be pelagic, and found in association with Sargassum, the dominant genus of seaweed that concentrates within the frontal region of the Loop Current during this time of year (Dooley, 1972). Sargassum flows from the Gulf of Mexico into the Gulf Stream via the Loop Current in the late spring and summer of each year, when noddies are breeding in the region (Gower and King, 2011). Other species have been shown to rely on the species that concentrate in Sargassum, particularly tunas such as albacore and dolphin fish (Coryphaena hippurus) [summarized in (Hensley and Hensley, 1995)]. This is of particular note as the brown noddy associates with large-bodied schooling fish such as tunas by way of “subsurface predator facilitated foraging” (Ashmole and Ashmole, 1967; Maxwell and Morgan, 2013). These larger fish species drive forage fish to the surface, making them available for noddies and a variety of other seabirds. This interaction is particularly critical to species such as the brown noddy that do not possess diving capabilities. Rather, they “patter” on the water's surface as they dip for prey near the water-air interface (Harrison et al., 1983). The frontal boundary provided by the current and the influence of the shelf break likely provide both an increase in productivity and entrainment of prey for both noddies and subsurface predators such as tuna that noddies feed in association with (Hensley and Hensley, 1995).

In this study, we applied the currently smallest available tag (1.4 g) to track seabirds. Prior to this study, available tags were too large to deploy without potentially significant harm to flight capabilities. The development of smaller tracking devices that are also lower in cost, is allowing for a greater understanding and associated management of small-bodied bird populations. Prior to appropriate technological advances, monitoring of such species could only be done at-sea. This type of monitoring can be costly and also has severe limitations, such as the inability to distinguish individuals or understand movements and behaviors beyond the observed area. Pathtrack NanoFix GPS tags cost approximately USD 450, store up to 320 locations and can be reused (if tags are recovered), allowing for inexpensive monitoring through time. Through the use of miniature tags, we were able to reveal previously unknown important conservation and ecological traits of this species.

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.