Search

The US weather radar archive holds detailed information about biological phenomena in the atmosphere over the last 20 years. Communally roosting birds congregate in large numbers at nighttime roosting locations, and their morning exodus from the roost is often visible as a distinctive pattern in radar images. This paper describes a machine learning system to detect and track roost signatures in weather radar data. A significant challenge is that labels were collected opportunistically from previous research studies and there are systematic differences in labeling style. We contribute a latent variable model and EM algorithm to learn a detection model together with models of labeling styles for individual annotators. By properly accounting for these variations we learn a significantly more accurate detector. The resulting system detects previously unknown roosting locations and provides comprehensive spatio-temporal data about roosts across the US. This data will provide biologists important information about the poorly understood phenomena of broad-scale habitat use and movements of communally roosting birds during the non-breeding season., Comment: 9 pages, 6 figures, AAAI 2020 (AI for Social Impact Track)

During migration, animals may experience high rates of mortality, but costs of migration could also be manifested through non-lethal carry-over effects that influence individual success in subsequent periods of the annual cycle. Using tracking data collected from light-level geolocators, we estimated total spring migration distance (from the last wintering sites to breeding sites) of tree swallows (Tachycineta bicolor) within three major North American flyways. Using path analysis, we then assessed direct and indirect effects of spring migration distance on reproductive performance of individuals of both sexes. When these data were standardized by flyway, females fledged 1.3 fewer young for every 1,017 km they traveled, whereas there was no effect of migration distance on reproductive success in males. In comparison, when these data were standardized across all individuals and not by flyway, longer migrations were associated with 0.74 more young fledged for every 1,017 km traveled by females and 0.26 more young fledged for every 1,186 km migrated by males. Our results suggest that migration distance carries over to negatively influence female reproductive success within flyways but the overall positive effect of migration distance across flyways likely reflects broader life-history differences that occur among breeding populations across the tree swallow range.

There have been an increasing number of observations of itinerancy in migratory songbirds, where individuals move among 2 or more widely separated areas during the “stationary” nonbreeding season. Knowledge of such movements and an understanding of what drives them are important for predicting how migratory populations will respond to environmental change. In this study, we investigated nonbreeding movements of the Tree Swallow (Tachycineta bicolor), an aerial insectivore that breeds across North America and spends the nonbreeding season around the Gulf of Mexico, Florida, Mexico, Central America, and the Caribbean. With year-round tracking data obtained from 133 light-level geolocators deployed at 12 breeding sites ranging from Alaska to Nova Scotia to North Carolina, we show that 44% of individuals made at least one large-scale movement (range: 301–1,744 km) within the nonbreeding range. The frequency of itinerancy decreased with longitude, such that 75% of individuals made a movement in the western portion of the nonbreeding range compared to only 31% in the east. Using the Normalized Difference Vegetation Index (NDVI) as a proxy for resource availability, we found that when individuals did move, they were more likely to move from sites where resources were deteriorating faster (a more negative change in NDVI prior to departure) than their destination sites. There was also evidence that individuals moved to destination sites with higher NDVI and temperature in the autumn, but not in the winter. Our results suggest movements of Tree Swallows during the nonbreeding season are influenced by resource availability, but because not all individuals used multiple nonbreeding sites, the density of individuals at a site and the level of competition may have also been a factor influencing nonbreeding season movements.

Latitudinal differences in timing of breeding are well documented but how such differences carry over to influence timing of events in the annual cycle of migratory birds is not well understood. We examined geographical variation in timing of events throughout the year using light-level geolocator tracking data from 133 migratory tree swallows ( Tachycineta bicolor ) originating from 12 North American breeding populations. A swallow's breeding latitude influenced timing of breeding, which then carried over to affect breeding ground departure. This resulted in subsequent effects on the arrival and departure schedules at autumn stopover locations and timing of arrival at non-breeding locations. This ‘domino effect’ between timing events was no longer apparent by the time individuals departed for spring migration. Our range-wide analysis demonstrates the lasting impact breeding latitude can have on migration schedules but also highlights how such timing relationships can reset when individuals reside at non-breeding sites for extended periods of time.

For seasonal migrants, non-breeding regions can play different roles in the ecology of their annual cycles: as stopover habitat, overwintering habitat, or as a combination in which some individuals stop-over and others over-winter. Such functional variations can lead to variation in occupancy dynamics and migration phenology to these different regions. In this study, we used data from archived Doppler weather surveillance radar to compare site-occupancy and movement dynamics of a migratory songbird (tree swallow Tachycineta bicolor) between two non-breeding areas: southeastern Louisiana and central peninsular Florida, USA. Specifically, in each area, we 1) quantified long-term (1996–2012) non-breeding season occupancy dynamics, 2) quantified variation in timing of autumn migration, and 3) tested which climate variables along their respective flyways were best correlated with variation in dates of arrival. Additionally, we cross-validated the dynamics from archived radar with data from eBird, a large-scale citizen science database that provides an independent measure of avian occupancy. We found strong and significant correlations between radar-estimated and eBird-estimated occupancy dynamics in both Louisiana and Florida. Long-term Louisiana occupancy dynamics conformed to our hypothesis that this region acts as a combined stopover and overwintering region whereas Florida occupancy dynamics were akin to a traditional winter region. Arrival to Louisiana during the study period was much earlier and took place over a much shorter arrival window than did arrival to Florida, which showed much more gradual arrival over the course of several months. At both sites, annual variation in mean arrival date was best explained by the amount of precipitation along the lower portions of their respective migration flyways.

One of the greatest feats of avian migration is the non-stop crossing of extensive areas of inhospitable habitat such as deserts and seas. Differences in spring and autumn migration routes have been reported in species that cross such barriers, and are thought to have evolved in response to seasonal variation in prevailing wind direction. We tested the hypothesis that migration routes vary seasonally with respect to the Gulf of Mexico in the tree swallow Tachycineta bicolor using solar geolocators attached and retrieved at 4 breeding sites in central North America. We found that 100 % of birds (n = 10) made a trans-Gulf flight of >850 km from Louisiana south to their wintering grounds in the Yucatan Peninsula in 12–36 hours, achieving minimum ground speeds as high as 32 m/s. Although most days during autumn migration were characterized by unfavorable headwinds blowing to the northwest, migration over the Gulf mostly occurred on days with strong winds blowing to the south. In contrast, in 8 of 9 (88 %) birds on spring migration returned from the wintering grounds towards Louisiana following a clockwise loop pat tern flying over land to the west around the Gulf. During this spring period there were few days with prevailing winds from the south to assist northward migration. Results suggest that, despite being up to three times further (ca. 2,700 km), a coastal cir- cum-Gulf spring migration represents the less risky route when wind conditions are not favorable. These findings also help to re solve a long-standing dispute in the literature concerning migration patterns between the US Gulf coast and Mexico, and provide insight into the factors shaping migration strategies of small songbirds migrating across large bodies of water.

Communal roosting is a taxonomically widespread phenomenon, with considerable variation in patterns of roost dynamics. Some organisms roost together in different locations each night (or day), whereas others roost in traditional locations each night, sometimes switching between roosts in the roost network. The behaviors that drive roost-site selection and the resulting patterns are not well understood. We created an individual-based model that simulates the daily aggregation of organisms into communal roosts. In the model, individuals move according to a movement rule integrating 2 independently adjustable behavioral drivers: roost fidelity, which leads individuals back toward their previous nights’ roost, and conspecific attraction, which leads individuals to congregate toward nearest conspecific neighbors. The model predicts that variable levels of aggregation will emerge under different combinations of these drivers, ranging from no aggregation to complete aggregation of individuals into their previous roosts. We tested to see which combination of drivers best predicts patterns of roost use in our study system of Tree Swallows (Tachycineta bicolor) in southeastern Louisiana in fall. Using Doppler weather radar data, we show that Tree Swallow roost sites remain consistent from night to night, and the birds return to one of several traditional roosts. Using radio telemetry, we show that individuals switch between these traditional roosts at minimum 22% of the time. Our results suggest that the formation of large communal roosts in Tree Swallows is driven by a combination of moderate conspecific attraction and strong, but not perfect, roost fidelity.

Despite benefits of using light-sensitive geolocators to track animal movements and describe patterns of migratory connectivity, concerns have been raised about negative effects of these devices, particularly in small species of aerial insectivore. Geolocators may act as handicaps that increase energetic expenditure, which could explain reported effects of geolocators on survival. We tested this ‘Energetic Expenditure Hypothesis’ in 12 populations of tree swallows ( Tachycineta bicolor ) and barn swallows ( Hirundo rustica ) from North America and Europe, using measurements of corticosterone from feathers (CORT f ) grown after deployment of geolocators as a measure of physiology relevant to energetics. Contrary to predictions, neither among- (both species) nor within-individual (tree swallows only) levels of CORT f differed with respect to instrumentation. Thus, to the extent that CORT f reflects energetic expenditure, geolocators apparently were not a strong handicap for birds that returned post-deployment. While this physiological evidence suggests that information about migration obtained from returning geolocator-equipped swallows is unbiased with regard to levels of stress, we cannot discount the possibility that corticosterone played a role in reported effects of geolocators on survival in birds, and suggest that future studies relate corticosterone to antecedent factors, such as reproductive history, and to downstream fitness costs.