No effect of geolocators on apparent return rates of a declining Neotropical migrant, the

. Canada Warblers ( Cardellina canadensis ) are small Neotropical migrants whose populations are declining across most of their range. Understanding factors limiting Canada Warbler populations requires knowledge of population ecology across the full annual cycle, including migratory pathways and over-winter locations. Light-level geolocator tags have offered unprecedented insight into migratory ecology for many species, but previous studies suggest that geolocators may influence apparent return rates. We sought to determine if geolocators influence apparent return rates of adult male Canada Warblers breeding in West Virginia, USA. In 2020, we deployed geolocators on 32 birds and color banded an additional 78 birds without geolocators. The following year, 13 of 32 (40.6%) geolocator birds and 37 of 78 (47.4%) color-banded birds were detected with no significant difference in apparent return rates between groups ( χ ² = 0.19, p = 0.66). Although further evaluation of additional groups will be valuable, the lack of significant effect on adult male Canada Warblers suggests that the slightly lower return rate does not preclude the use of geolocators as a tool to assess the migration ecology of this small songbird of conservation concern


INTRODUCTION
Migratory and nonbreeding season ecology remains one of the least understood aspects of Nearctic-Neotropical migratory bird life history (Faaborg et al. 2010). Avian survival may be lowest during migration than at any other point in the annual cycle (Sillett and Holmes 2002), and conditions during the nonbreeding season may create carryover effects on survival and reproduction during the breeding season (González-Prieto and Hobson 2013), potentially contributing to population declines (Sherry and Holmes 1995). Many species that breed in North America and migrate to Central and South America have exhibited marked declines in abundance for at least the past 50 years (Rosenberg et al. 2019). Given that many of these species spend at least half of the annual cycle migrating or on nonbreeding grounds, a better understanding of migratory and nonbreeding season ecology is critical for effective conservation.
Light-level geolocator tags are archival devices that record patterns of twilight and day length used to determine geographic location (Lisovski et al. 2020) and are the most precise widely available method for tracking individuals as small as 9.0 g . Geolocators allow inference regarding wintering and breeding ground locations (Kramer et al. 2017, stopover dynamics (Cooper et al. 2017), habitat suitability (Justen et al. 2021), and migration timing (Renfrew et al. 2019). After deployment, devices must be recovered the following year. A representative sample of birds with geolocators would require that addition of the device not affect return probability.
Geolocator effects have previously been researched for a variety of old and new world warbler species of similar size with mixed results. Procházka et al. (2018) reported no significant effect of geolocators on recapture rates of Eurasian Reed Warblers (Acrocephalus scirpaceus) in Germany, although there was a significantly greater recapture rate for the birds without geolocators in the Czech Republic and Bulgaria. Return rates, body mass, and migration chronology were not significantly different between birds with and without geolocators in Goldenwinged Warblers (Vermivora chrysoptera; Peterson et al. 2015), and weights of Great Reed Warblers (Acrocephalus arundinaceus) Journal of Field Ornithology 94(3): 5 https://journal.afonet.org/vol94/iss3/art5/ did not significantly differ between birds with and without geolocators after migration (Malmiga et al. 2021). Alternatively, other studies reported decreased return rates for Prairie Warblers (Setophaga discolor; Campbell et al. 2021), Aquatic Warblers (Acrocephalus paludicola; Salewski et al. 2019), and Cerulean Warblers (Setophaga cerulea; Raybuck et al. 2017). These varying results suggest geolocator impacts could be species-dependent, necessitating investigation into geolocator effects on a diversity of species.
The Canada Warbler (Cardellina canadensis) is a Neotropical migrant with recorded population declines across its range (Sauer et al. 2021). Canada Warblers breed in the boreal forests of southern Canada and in the northeastern U.S. and winter in southern Central America and northwestern South America. Migration routes appear to be primarily overland, converging along eastern Central America into South America . Major threats to their wintering grounds include land conversion (Etter et al. 2006a), deforestation (Dávalos et al. 2011), and agriculture (Etter et al. 2006b). Compared to other migratory bird species in North America, Canada Warblers have one of the shortest breeding seasons, generally arriving to breeding grounds in May and departing in August and raising only a single brood (Flockhart 2007, Reitsma et al. 2020. This brief period represents just a fraction of Canada Warbler life history; therefore, it is crucial to incorporate the full annual cycle to understand factors limiting populations. To develop full annual cycle models, fundamental populationspecific life history data including migratory routes, wintering ground locations, and population connectivity can be described using data from geolocators. Because geolocators have impacted birds of similar size and may have biased geolocator results , we believe it is important to assess the potential effects of geolocators on this migratory bird of conservation concern. We compared apparent return rates of Canada Warblers with and without geolocators in the Monongahela National Forest (MNF), West Virginia, USA. We aimed to assess the potential effects of geolocators on apparent return rates of a declining species of conservation concern.
METHODS Field work occurred in the MNF, West Virginia, USA, from May to July in the 2020 and 2021 breeding seasons. Six study sites were located within the Appalachian Plateau Physiographic Province in the western MNF. Within this region, forest types below 900 m include mixed oak and mixed mesophytic, with northern hardwood forests occurring above 900 m and remnant boreal red spruce occurring above 1150 m (USDA 2011).
The sites within the MNF were stratified across 3 elevation bands: < 853 m, 853-1158 m, and > 1158 m. The lower limit was suggested by Harding et al. (2017) as the lower limit of Canada Warbler occurrence within the Appalachians, and the middle band was used to create three equal elevation divisions. Two sites were assigned to each elevation stratum and placed in areas where Dimmig et al. (2022) reported relatively high Canada Warbler detection rates. These sites were characterized by proximity to riparian areas and high densities of rhododendron (Rhododendron maximum), mountain laurel (Kalmia latifolia), and other understory shrubs (Augustine 2022, Dimmig et al. 2022).
This study was conducted in conjunction with a mark-resight project to estimate Canada Warbler demographic rates (Augustine 2022), which required capturing and individually marking adult male birds with color bands. To find unmarked birds in 2020, we listened for the songs of males and used conspecific audio playback to trigger territorial responses if no birds were initially detected. Unmarked birds were captured in mist nets (6 m x 2.6 m; 30 mm mesh) using the same conspecific audio lure. We deployed 32 geolocators (FL6057 fLight Lotek Wireless) equally across the six sites on adult male birds that met the weight requirement of ≥ 10.7 g to ensure geolocators were < 5% of a bird's mass (Fair et al. 2010). Geolocators were mounted with superglue onto figure-eight leg loop harnesses made from 0.5 mm Stretch Magic elastic cord (Rappole andTipton 1991, Streby et al. 2015) attached to the bird so the device sat snugly on the synsacrum (Fig. 1). In contrast to Streby et al. (2015), the elastic of the harnesses was intended to last at least a year for annual retrieval. Birds with geolocators were marked with a unique combination of aluminum U.S. Geological Survey (USGS) band and two-color bands, which allowed rapid differentiation from non-geolocator birds in the field. Birds captured after all geolocators were deployed, or those which did not meet the weight threshold, were marked with a unique combination of aluminum USGS band and three-color bands (n = 78). In this project, only male birds were considered because females are not generally responsive to territorial calls (Reitsma et al. 2018) and therefore geolocator recovery rates were expected to be low. We assessed return rates in 2021 by systematically searching study sites using a modification of typical territory spot mapping (Bibby et al. 2000). We searched for returning birds at regularly placed grid points within the boundaries of the study areas. Grid points were placed 150 m apart to account for aural detection limits of Canada Warblers (60-200 m; Matsuoka et al. 2012, Hunt et al. 2017. We listened for Canada Warbler songs between grid points and broadcast songs for up to three minutes at grid points if no birds were detected. When we located a marked bird, we recorded the location and identity of the individual. If a marked bird was located that had been deployed with a geolocator, we recaptured the bird using the previously described capture protocol and retrieved the geolocator tag. To determine if detecting a bird marked the previous year was independent of whether it carried a geolocator, we used a chisquare test with Yate's continuity correction. We set a type-I error threshold of α = 0.05 to determine the statistical significance. Analyses were completed in program R v.4.2.0 (R Core Team 2022).

RESULTS
We marked 110 birds in 2020, of which 32 received geolocators. Of those marked birds, we detected 50 in 2021 (apparent return probability = 0.45). Of the 78 birds without geolocators, 37 were detected (apparent return probability = 0.47). Of the 32 birds that received geolocators, 13 were detected (apparent return probability = 0.41) and subsequently recaptured with their tag. No birds that were fitted with geolocators were resighted without their harness, and all resighted birds that were fitted with geolocators were recaptured. Despite the slightly lower observed return rate of birds that received geolocators, the difference was not statistically significant (χ² = 0.19, p = 0.66, DF = 1).

DISCUSSION
Deployment of geolocators on Canada Warblers produced a small but nonsignificant reduction in apparent return rate compared to control birds. This result is consistent with other studies that also found no effect of geolocators , van Wijk et al. 2016, Bell et al. 2017, including another study evaluating return rates of Canada Warblers . However, our results are also consistent with a meta-analysis that determined a weak negative effect of geolocators on apparent survival (Brlík et al. 2020). A weak effect, coupled with a modest sample size, may have precluded our ability to detect a significant effect. Regardless, our failure to detect an effect of geolocators on apparent return rates suggests they may be a reliable tool for assessing migration ecology of Canada Warblers.
Canada Warblers, with an average mass of 10.6 g (males, range 8.7-13.5 g; Reitsma et al. 2020), are among the smallest birds that can be fitted with light-level geolocators, and we were only able to deploy devices on a subset of birds that were heavy enough to reach the 5% threshold (Fair et al. 2010). Brlík et al. (2020) found that increasing relative load of geolocators and use of elastic bands to attach geolocators led to lower apparent survival, and that the effects of geolocators were stronger on smaller species. We may have reduced the relative load on Canada Warblers by only placing geolocators on the heaviest individuals, perhaps offsetting potential negative effects of elastic bands and the small overall body size of Canada Warbler in contributing to no strong effect on apparent return rates.
Although placing geolocators on the heaviest individuals may have contributed to our finding of no strong effect, this may also lead to biased inferences regarding migratory ecology. Indeed, we further restricted geolocators to adult males, due to low resight and recapture rates of females and hatch-year birds. Although geolocators offer data on migratory ecology of average to large adult male Canada Warblers, such conclusions may not be applicable to other size, age, and sex classes. For example, birds may exhibit differential migration based on age (Cristol et al. 1999), and females may exhibit different migratory behavior than males (Catry et al. 2005). Further study of Canada Warbler migratory ecology should continue the assessment of geolocator effects on apparent return rates of smaller and female birds.