The use of social attraction techniques to restore seabird colonies on Desecheo Island, Puerto Rico

1. Desecheo Island (117 ha) was historically an important seabird island in the Caribbean with 15 species recorded, of which seven are known to breed, including major populations of brown boobies and red-footed bobbies. The introduction of invasive mammals, plus the use of the island as a bombing range, contributed to the extirpation of five of the seven known breeding populations of seabird species and vastly reduced numbers of the remaining two species. The island became a National Wildlife Refuge in 1976 and major conservation interventions have included the eradication of invasive goats, rhesus macaques and rats between 1976 and 2016. Removing these critical threats from the island has allowed other active restoration

Since 2008, five of the seven historic breeders are considered extirpated (historic populations once ranged from 3 to 15,000; Table 1), with the most remarkable decline being the brown booby that went from 15,000 breeding individuals (in 1931) to zero in just 60 years . The remaining two breeding seabirds, brown noddy and bridled tern, breed in vastly reduced numbers (i.e. less than 2% of the historical breeding populations) and primarily on offshore rock stacks (Wolf et al., 2010). Thus, the island's seabird population is now limited to small numbers of nesting and roosting bridled terns and brown noddies: roosting brown boobies, roseate terns (Sterna dougallii), sandwich terns (Thalasseus sandvicensis) and laughing gulls; and occasional reports of magnificent frigatebirds (Fregata magnificens), whitetailed tropicbirds (Phaethon lepturus), red-billed tropicbirds (Phaethon aethereus) and Audubon's shearwaters (Puffinus lherminieri) flying over the island (A. Morales-Pérez 2008, 2010Island Conservation, 2013;McKown, 2009;Wolf et al., 2010). Between 2010 and 2017, a partnership between USFWS and the non-government organization Island Conservation (IC) successfully removed invasive mammals from Desecheo with a goal to protect and restore the island ecosystem, with emphasis on native seabirds, endemic reptiles and native plants.
Seabird restoration programs aim to re-establish extirpated populations as well as augment existing populations (Jones & Kress, 2012).
Implementing seabird restoration methods after an invasive mammal eradication is a good opportunity to speed up this process (Kappes & Jones, 2014). On Desecheo, our objectives were to initiate social attraction methodologies to (a) augment numbers of bridled terns and brown noddies and (b) establish Audubon's shearwaters, consistent with USFWS intent to restore seabird colonies in the NWR. At a regional scale, securing seabirds on Desecheo also offers a potentially valuable high-elevation refuge from projected sea level-rise impacts.   and recent (2000-2019) data for 15 seabird species documented on Desecheo Island and its offshore islets: BRBO (brown booby), RFBO (red-footed booby), BRNO (brown noddy), BRTE (bridled tern), LAGU (laughing gull), MAFR (magnificent frigatebird), SOTE (sooty tern), ROYT (royal tern), LETE (least tern), ROST (roseate tern), BRPE (brown pelican), WTTR (white-tailed tropicbird), RBTB (red-billed tropicbird), SATE (sandwich tern) and AUSH (Audubon's shearwater)  (Jones et al., 2008(Jones et al., , 2018Medina et al., 2011). In addition, we expected bridled tern and brown noddy populations to have a higher probability of responding to social attraction tools given that both species have been previously observed attempting to nest in low numbers on the island and on rock stacks off- the species has disappeared from a number of former breeding sites, mainly due to predation by invasive species like rats and feral cats, and has been classified as Near Threatened by Schreiber and Lee (Lee, 2000), making this species a conservation priority for the region (Bradley & Norton, 2009).

Site and species selection
Using the maps created for bridled tern and brown noddy, we identified three and two potential social attraction sites, respectively, where there was evidence of historic breeding, and we could safely access the site with appropriate equipment. For the Audubon's shearwater, we

Equipment
Using a mix of audio and visual cues, we employed three social attraction methods for restoring seabird colonies on Desecheo Island: decoys, mirrors and sound systems. For bridled tern and brown noddy, recycled high-density polyethylene plastic decoys were manufactured as the main components of the sound system ( Figure 3). In addition, each sound system had two outdoor speakers, four solar panels and two deep-cycle marine batteries. The speakers were placed approximately 20 m apart from each other. The amplifier unit maximized the sound output of the recordings, which were collected from natural Audubon's shearwater and brown noddy colonies in the Caribbean.
The audio range for each system was estimated to be between 300 and 500 m, depending on surrounding noise such as boats and waves, simulating the expected sound level of an active breeding colony. Two sound systems were programmed to play the sounds of an Audubon's shearwater colony for 12 of every 24 h from dusk to dawn, and one sound system was programmed to play the sounds of a brown noddy colony for 12 of every 24 h from dawn to dusk. To monitor responses, we deployed one to four motion-sensing cameras (Reconyx PC900 Hyperfire Professional IR and/or Browning Strike Force Elite HD) at each decoy colony site and two cameras -one directed on each speaker -at sound system sites.

Bridled tern
In

Brown noddy
In February 2018, we deployed 18 brown noddy decoys, three mirrors and five motion-sensing cameras at one site on the south side of the island (Figures 2c and 2d). In addition, we deployed a sound system with accompanying cameras 30 m north of the decoy colony (Figures 3c and 3d). Cameras at the sound system site were deployed from February to September in both 2018 and 2019. In March 2019, we added a second decoy colony of 10 brown noddy decoys on a rock outcrop on the west coast with two cameras and no mirrors.

Audubon's shearwater
In February 2018, we placed two sound systems with motion-sensing cameras on the western coast of the island 250 m apart (Figure 1). Cameras were deployed from February to September 2018 and February to August 2019.  Approximately 86% of the photos were invasive green iguanas, 5.5%

Bridled tern social attraction project
were bridled terns and 8.5% were leaves moving around the decoy.
Four months following the February 2019 decoy colony deployments, we visited the island and found no nests. Camera monitoring was undertaken from February to August 2019 producing 5461 images and 47 videos over 128 days. A review of all images found no bridled terns nesting or roosting at these sites. Thus, no bridled tern individuals were documented in these two 2019 decoy colonies located on the west coast of the island; however, five nests were found in an area distinct from the decoy colonies (approximately 215 m) but also on the west side of the island.

Brown noddy social attraction project
The

Audubon's shearwater social attraction project
The Audubon's shearwater sound systems were undamaged in the showing either an attempted copulation or aggressive interaction (Figure 6b). We conducted nest searches around the sound system sites in September 2018 and August 2019 and did not find any nests.

DISCUSSION
The introduction of non-native invasive mammals to islands has had dramatic consequences to seabirds globally (Dias et al., 2019;Jones et al., 2008;Spatz et al., 2014) and is identified as a key threat to seabird colonies in the Caribbean (Bradley & Norton, 2009). Our literature review shows a decline in total abundance of seabirds and number of breeding seabird species on Desecheo in the 20th century (Table 1), during which invasive mammals including rats (present from Desecheo. Resource availability did not allow us to utilize either a spatial or temporal control to compare results, limiting our ability to infer results. Nonetheless, documenting our approach, challenges and initial observations is beneficial to conservation practitioners and others seeking to support seabird conservation. Prior to our project, in 2010, bridled terns were observed breeding on the offshore rocks (thirteen nests) and in small numbers (four nests) on the main island of Desecheo, and one brown noddy nest had been documented on the main island (Table 1). Their presence provided a potential source population to interact with our decoys and support creation of new roosting and nesting sites. We saw evidence of bridled terns interacting with decoys; however, nesting has not yet increased significantly across the island (bridled tern main island nests observed: pre-social attraction implementation (2010) 4  a first for the island, no active nesting has yet been found. Given the methodology has only been implemented for 2 years, and age of first breeding is estimated to be 3-5 years for bridled tern, 3-7 years for brown noddy and 5 years for Audubon's shearwater, it is not surprising that we recorded no interaction from brown noddies and only small numbers of bridled terns and Audubon's shearwaters interacting with the social attraction stimuli (Labbé, 2017;Mackin, 2007;USFWS, 2016). Further, relatively small regional population sizes limit the likelihood of interactions, with Hispaniola and Puerto Rico estimates of brown noddy to be <1500 pairs, <700 bridled terns (Keith, 2009;Saliva, 2009) and <1500 of Audubon's shearwaters (Mackin, 2016), with a smaller proportion of these numbers occurring in the Hispaniola Channel where Desecheo occurs. Thus, we expect having the social attraction stimuli in place for more years will increase the likelihood of more individuals from the regional population encountering and interacting with the stimuli.

CONFLICT OF INTEREST
The authors declare no conflict of interest.

AUTHOR CONTRIBUTIONS
NH and SS conceived of the project ideas. NH and JLH designed methodology. RC, JLH, CF, EV and CW collected and curated the data.
JLH, CF and CW analyzed the data. All authors contributed to writing, editing and reviewing the manuscript and gave final approval for publication.

DATA AVAILABILITY STATEMENT
All data are available through the Dryad Digital Repository: https://doi. org/10.5061/dryad.h9w0vt4hb .