Miniaturized GPS Tags Identify Non-breeding Territories of a Small Breeding Migratory Songbird

For the first time, we use a small archival global positioning system (GPS) tag to identify and characterize non-breeding territories, quantify migratory connectivity, and identify population boundaries of Ovenbirds (Seiurus aurocapilla), a small migratory songbird, captured at two widely separated breeding locations. We recovered 15 (31%) GPS tags with data and located the non-breeding territories of breeding Ovenbirds from Maryland and New Hampshire, USA (0.50 ± 0.15 ha, mean ± SE). All non-breeding territories had similar environmental attributes despite being distributed across parts of Florida, Cuba and Hispaniola. New Hampshire and Maryland breeding populations had non-overlapping non-breeding population boundaries that encompassed 114,803 and 169,233 km2, respectively. Archival GPS tags provided unprecedented pinpoint locations and associated environmental information of tropical non-breeding territories. This technology is an important step forward in understanding seasonal interactions and ultimately population dynamics of populations throughout the annual cycle.

Scientific RepoRts | 5:11069 | DOi: 10.1038/srep11069 Here we tested and deployed a newly developed miniaturized tracking device that weighs ~1 g, combines archival technology with GPS accuracy (~10 m) and can be placed on small songbirds weighing < 20 g. Until now, the smallest comparable device available was ~12 g and could only be placed on animals weighing at least 250 g 13 . Given the accuracy of the location data, GPS tags placed on breeding birds can provide high-resolution locality information from non-breeding territories, migratory routes and migration timing depending on flexible programming. In this paper, we provide the first data from miniaturized archival GPS tags deployed on a small (< 20 g) migratory songbird. First, we identify individual non-breeding territories and then characterize and compare their environmental attributes using remote sensing. Second, we define non-breeding season population boundaries for our two distinct breeding populations and quantify the strength of migratory connectivity.

Discussion
Here, using a new miniaturized archival GPS technology we locate, with high resolution, the non-breeding territories of individual migratory songbirds captured during the breeding season. Doing so also allows us to use remotely sensed data to characterize the environmental attributes and conditions of these non-breeding territories, delineate population boundaries and quantify the strength of migratory connectivity of these two different breeding populations. Data at this resolution throughout the annual cycle were previously unattainable for species anywhere close to this size. Eighty-three percent of the recovered units (15 of 18) contained location data and obtained at least two locations from the non-breeding season. Because location data are downloaded as geographic coordinates there is also no analysis subjectivity or significant data processing time, a problem inherent within light-level geolocators 9 .
Until now it has been impossible to obtain territory-level non-breeding season locations or specific habitat information of breeding migratory songbirds remotely because the spatial resolution of light-level geolocators and stable isotopes is low (100-1000 km; but see 15 ). Non-breeding location data for Ovenbirds analogous to those obtained with GPS tags (~10 m) would involve weeks of costly, labor intensive field work 16 and the breeding origin of these individuals would be unknown. Territory sizes identified with GPS tags remotely are consistent with Ovenbird non-breeding territories determined in the field with radio telemetry in Belize 17 and Jamaica 16 confirming this as a viable option for estimating territories remotely. In addition, these new GPS tags provide location data from potentially physically and/or politically (i.e., Cuba) difficult to access locations. Therefore, the combination of remotely sensed data with simultaneous, unbiased location data provided by GPS tags signifies a major advance for the study of small migratory animals. The fact that they need to be retrieved remains a disadvantage.
The non-breeding population boundaries of our two breeding populations were non-overlapping and encompassed four different political boundaries (Cuba, Dominican Republic, Haiti and United States). Delineating population boundaries throughout the annual cycle is an important advance as it allows for the integration of demographic rates into dynamic models for linked populations 18 , necessary to identify processes that limit or regulate populations and determine when and where they operate. Interestingly, we identified the non-breeding population boundary of this same breeding population at HBEF using light-level geolocators 12 . Although the non-breeding area characterized by geolocators was large (231,509 km 2 ), the population boundaries for HBEF we identify here was entirely encapsulated (100%-114,803 km 2 of 231,509 km 2 ) within estimates derived with light-level geolocators confirming that these new population boundaries are robust. Despite having small sample sizes, the non-breeding population boundary size for JBWS reached an asymptote around 10 individuals (Fig. S2). Identifying linked breeding and non-breeding populations across the range of a species will enable assessments of how past changes in land use and climate influence population trends as well as make predictions about the potential impacts of future environmental stressors.
The strength of migratory connectivity we quantify here using the Mantel approach 19,20 was surprisingly strong, although our sample sizes were small and from only two locations. Additional research using tracking devices that provide high-resolution spatial data, along with large sample sizes, will likely provide novel insights into our understanding of migratory connectivity from this and other small-bodied migratory species. Such data will improve our understanding of the role that varying strengths of migratory connectivity play in driving seasonal interactions, delineating population boundaries and ultimately in population dynamics.
The high-resolution tracking of a small migratory songbird, from a breeding to a non-breeding territory at 10 m accuracy, is a major step forward for understanding the ecology of small migratory animals. Continued improvements in this technology will further increase our ability to answer complex questions with regard to seasonal interactions and population dynamics and hopefully help in our ability to conserve migratory species.

Methods
We deployed archival GPS tags on Ovenbirds, a small (~20 g) long-distance migratory songbird, breeding within Hubbard Brook Experimental Forest, NH (HBEF; 43.94° N, − 71.73° W) and Jug Bay Wetland Sanctuary, MD USA (JBWS; 38.77° N, − 76.69° W) during the 2013 breeding season (May-June). At each site, we attached a PinPoint-10 archival GPS tag (1.1 g, Lotek Wireless Inc.) on twenty-four males weighing > 20.0 g. Tags were programmed to begin collecting data on 1 July 2013 with a delay of 28 days (maximum delay period possible at time of deployment) between location estimates to maximize the likelihood of obtaining non-breeding location data. All tags were programmed to collect data on the same schedule (see supplemental information). The following year, individuals that returned to breed were recaptured (April and May 2014), devices were removed and data downloaded (PinPoint software, Lotek Wireless Inc.). Location data are downloaded as geographic coordinates therefore no further processing is needed to determine locations.

Identifying and Characterizing Attributes of Non-breeding Territories. Individual territories
were constructed using the location data during the stationary non-breeding season (see Supplemental Information). Environmental attributes of non-breeding territories were quantified using remotely sensed data (Table S1) annotated to each location using Env-DATA Track Annotation Service from movebank. org 5 .
Delineating population boundaries and Quantifying Migratory Connectivity. We used location data from the GPS tags to delineate relative boundaries for each breeding population. We define boundaries during the non-breeding season as the 95% kernel density estimate around the non-breeding locations for each breeding population 7 . We quantified the strength of migratory connectivity using the Mantel correlation coefficient (r M ) 19,20 . Values close to 1 indicate a strong correlation between breeding and non-breeding distance matrices (i.e., individuals breeding in close proximity also winter in close proximity).
The methods were carried out in accordance with the approved federal guidelines for using animals as research subjects. The procedures used to capture and handle Ovenbirds, and attach GPS tags were approved by the National Zoological Park's Institutional Animal Care and Use Committee (NZP-IACUC # 11-13 and NZP-IACUC # 14-21).