Supportive wind conditions influence offshore movements of Atlantic Coast Piping Plovers during fall migration 2 Piping Plover migration

Abstract

In advance of large-scale development of offshore wind energy facilities throughout the U.S. Atlantic Outer Continental Shelf (OCS), information on the migratory ecology and routes of federally threatened Atlantic Coast Piping Plovers (Charadrius melodus melodus) is needed to conduct risk assessments pursuant to the Endangered Species Act. We tagged adult Piping Plovers (n = 150) with digitally coded VHF transmitters at 2 breeding areas within the southern New England region of the U.S. Atlantic coast from 2015 to 2017. We tracked their migratory departure flights using a regional automated telemetry network (n = 30 stations) extending across a portion of the U.S. Atlantic Bight region, a section of the U.S. Atlantic coast, and adjacent waters of the Atlantic Ocean extending from Cape Cod, Massachusetts, to Cape Hatteras, North Carolina. Most adults departed within a 10-day window from July 19 to July 29, migrated nocturnally, and over 75% of individuals departed within 3 hr of local sunset on evenings with supportive winds. Piping Plovers migrated offshore directly across the mid-Atlantic Bight, from breeding areas in southern New England to stopover sites spanning from New York to North Carolina, USA, over 800 km away. During offshore migratory flights, Piping Plovers flew at estimated mean speeds of 42 km hr−1 and altitudes of 288 m (range of model uncertainty: 36–1,031 m). This study provides new information on the timing, weather conditions, routes, and altitudes of Piping Plovers during fall migration. This information can be used in estimations of collision risk that could potentially result from the construction of offshore wind turbines under consideration across large areas of the U.S. Atlantic OCS.

Full text

AmericanOrnithology.org
Volume 122, 2020, pp. 1–16
DOI: 10.1093/condor/duaa028
Published by Oxford University Press for the American Ornithological Society 2020. This work is written by (a) US Government
employee(s) and is in the public domain in the US.
RESEARCH ARTICLE
Supportive wind conditions influence offshore movements
of Atlantic Coast Piping Plovers during fall migration
PamelaH. Loring,1* JamesD. McLaren,2 Holly F. Goyert,3 and PeterW.C. Paton4
1 U.S. Fish and Wildlife Service Division of Migratory Birds, Northeast Region, Hadley, Massachusetts, USA
2 Environment and Climate Change Canada, Science and Technology Branch, Ottawa, Ontario, Canada
3 Department of Environmental Conservation, University of Massachusetts Amherst, Massachusetts, USA
4 Department of Natural Resources Science, University of Rhode Island, Kingston, Rhode Island, USA
*Corresponding author: pamela_loring@fws.gov
Submission Date: October 20, 2019; Editorial Acceptance Date: May 7, 2020; Published June 22, 2020
ABSTRACT
In advance of large-scale development of offshore wind energy facilities throughout the U.S. Atlantic Outer Continental
Shelf (OCS), information on the migratory ecology and routes of federally threatened Atlantic Coast Piping Plovers
(Charadrius melodus melodus) is needed to conduct risk assessments pursuant to the Endangered Species Act. We tagged
adult Piping Plovers (n = 150) with digitally coded VHF transmitters at 2 breeding areas within the southern New England
region of the U.S. Atlantic coast from 2015 to 2017. We tracked their migratory departure flights using a regional auto-
mated telemetry network (n = 30 stations) extending across a portion of the U.S. Atlantic Bight region, a section of the
U.S. Atlantic coast, and adjacent waters of the Atlantic Ocean extending from Cape Cod, Massachusetts, to Cape Hatteras,
North Carolina. Most adults departed within a 10-day window from July 19 to July 29, migrated nocturnally, and over
75% of individuals departed within 3hr of local sunset on evenings with supportive winds. Piping Plovers migrated off-
shore directly across the mid-Atlantic Bight, from breeding areas in southern New England to stopover sites spanning
from New York to North Carolina, USA, over 800 km away. During offshore migratory flights, Piping Plovers flew at esti-
mated mean speeds of 42 km hr−1 and altitudes of 288 m (range of model uncertainty: 36–1,031 m). This study provides
new information on the timing, weather conditions, routes, and altitudes of Piping Plovers during fall migration. This in-
formation can be used in estimations of collision risk that could potentially result from the construction of offshore wind
turbines under consideration across large areas of the U.S. Atlantic OCS.
Keywords: automated radio telemetry, Charadrius melodus melodus, migration, offshore wind energy, Piping Plover
Las condiciones del viento de apoyo influencian los movimientos en alta mar de Charadrius melodus
melodus durante la migración de otoño
RESUMEN
Antes del desarrollo a gran escala de emprendimientos de energía eólica en alta mar a lo largo de la plataforma conti-
nental exterior (PCE) del Atlántico de EEUU, se necesita información de la ecología y las rutas migratorias de la especie
amenazada a nivel federal Charadrius melodus melodus para realizar evaluaciones de riesgo conforme a la Ley de Especies
LAY SUMMARY
• The Atlantic coast population of the Piping Plover is listed as “Threatened” under the U.S. Endangered Species Act.
• Previously, little was known about exactly when, under what conditions, and along which routes these shorebirds under-
take their migration from nesting areas along the Atlantic coast to wintering sites extending to eastern Caribbean islands.
• To help ll these information gaps, we attached miniature digitally coded VHF transmitters to 150 adult Piping Plovers
at nesting areas in southern New England and constructed 35 radio antenna towers along the Atlantic coast to track
their routes during fall migration.
• Most of the Piping Plovers in our study departed from southern New England in late July, at sunset, with tailwinds supporting
oshore migratory ights across the mid-Atlantic Bight to stopover areas spanning from coastal New York to North Carolina.
• During oshore migratory ights, Piping Plovers ew at estimated mean speeds of 42 km hr−1 and at altitudes of 288m.
• Our results provide the rst empirical data on Piping Plover ight routes, altitudes, and weather conditions during fall
migration.
• This information can be used to estimate collision risk from oshore wind turbines currently under consideration
across large areas of the U.S. Atlantic Ocean.
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

2 Piping Plover migration P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
en Peligro de Extinción. Marcamos adultos de C.m.melodus (n = 150) con transmisores VHF codificados digitalmente en
dos áreas reproductivas en la región sur de Nueva Inglaterra de la costa atlántica de EEUU desde 2015 a 2017. Seguimos
sus vuelos de partida migratoria usando una red regional de telemetría automatizada (n = 30 estaciones) dispuesta a
lo largo de una porción de la región de la ensenada del Atlántico de EEUU, una sección de la costa atlántica de EEUU y
las aguas adyacentes del Océano Atlántico que se extiende desde el Cabo Cod, Massachusetts hasta el Cabo Hatteras,
Carolina del Norte. La mayoría de los adultos partieron dentro de una ventana temporal de 10 d del 19 al 29 de julio,
migraron de noche y más del 75% de los individuos partieron durante las últimas 3hr del atardecer local en tardes
con vientos de apoyo. C.m.melodus migró a alta mar directamente a través de la ensenada del Atlántico medio, desde
las áreas de cría en el sur de Nueva Inglaterra hasta los sitios de parada comprendidos entre Nueva York y Carolina del
Norte, EEUU, a más de 800 km de distancia. Durante los vuelos migratorios en alta mar, los individuos de C.m.melodus
volaron a velocidades estimadas promedio de 42 km hr–1 y altitudes de 288 m (rango de incertidumbre del modelo:
36–1,031 m). Este estudio brinda nueva información sobre las fechas, las condiciones temporales, las rutas y las altitudes
de C.m.melodus durante la migración de otoño. Esta información se puede usar en estimaciones del riesgo de colisión
que podría resultar de la construcción de turbinas eólicas en alta mar bajo consideración a lo largo de grandes áreas de
la PCE del Atlántico de EEUU.
Palabras clave: Charadrius melodus melodus, energía eólica en alta mar, migración, radio telemetría automatizada
INTRODUCTION
In the U.S. Atlantic Outer Continental Shelf (OCS), over
5,492 km2 is presently under lease agreement with the
Bureau of Ocean Energy Management (BOEM) for devel-
opment of commercial-scale offshore wind energy facil-
ities and an additional 12,976 km2 is in the planning stages
for potential leases (BOEM 2019). e only offshore wind
energy facility currently operating in North America is a
5-turbine, 30-megawatt (MW) demonstration-scale fa-
cility near Block Island, Rhode Island, USA, that started
operations in 2016 (Wilber et al. 2018). e potential
adverse effects of offshore wind energy developments
on avian species include collision mortality, behavioral
changes near turbines in response to visual stimuli, and
impacts from physical alteration of habitat in response
to construction of turbines and other infrastructure (Fox
etal. 2006). With large areas of the Atlantic OCS under
consideration for development of offshore wind energy
facilities, information on offshore movements and flight
characteristics of high-priority bird species is needed for
estimating exposure of birds to collision risks with wind
turbines, and for developing strategies to manage adverse
effects (BOEM 2017).
ere is considerable variation among avian species in
their vulnerability to offshore wind energy developments
(Furness etal. 2013), thus quantifying species-specific traits
that influence collision risk factors is critical (May et al.
2017). Although much is known about flight characteristics
(e.g., flight altitude, avoidance behaviors) of many species
of marine birds in offshore habitats (Furness et al. 2013,
Johnston et al. 2014), less is known about small-bodied
(<100g) shorebirds that migrate nocturnally. is is pri-
marily due to technological limitations of monitoring their
movement ecology. Much of the information that has been
previously documented on offshore movements of shore-
birds is from radar-based tracking studies (Richardson
1976, Williams and Williams 1990, Dirksen et al. 2000,
Langston and Pullan 2003). However, radar technology
used to study bird movements is limited by the operational
range of the radar and often lacks the resolution required
to identify birds to the species level (Desholm etal. 2006).
e use of individual-based tracking technologies, such as
radio or satellite transmitters, can provide more detailed
information on the movements and behavior of known
individuals across time and space (Robinson et al. 2010).
However, only recently has tracking technology become
available for monitoring movements of small-bodied avian
species across large spatial extents (Taylor etal. 2017), such
as the U.S. Atlantic region (Loring etal. 2017, 2018, 2019).
Recently, biologists have used digitally coded VHF trans-
mitters to assess migration departure decisions and stop-
over ecology of smaller shorebirds (Anderson et al. 2019,
Holberton etal. 2019). During preconstruction monitoring,
assessments of the exposure risk of migratory birds to off-
shore wind energy facilities require species-specific infor-
mation on migratory routes, flight altitudes, temporal (diel
and seasonal) variation in movement patterns, and vari-
ation in environmental conditions associated with offshore
movements. Information about meteorological conditions
associated with offshore flights is especially important for
risk assessments, as birds may be at higher risk of collision
with offshore wind turbines during inclement weather (e.g.,
high winds, precipitation, low visibility) due to impaired
visibility and avoidance responses (Exo etal. 2003).
Migratory shorebirds may be especially susceptible to
the potential effects of wind energy development due to
their use of coastal habitats and migratory routes that may
occur offshore (O’Connell etal. 2011). One species of con-
cern is the federally threatened Atlantic coast population
of the Piping Plover (Charadrius melodus melodus; U.S.
Fish and Wildlife Service 1985). is population nests from
North Carolina, USA, to Newfoundland, Canada (Elliott-
Smith and Haig 2020), and winters ~800–2,000 km from
its breeding grounds, from North Carolina to Florida, as
well as on islands in the Caribbean (Gratto-Trevor et al.
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton XXXX 3
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
2012, 2016; Cohen etal. 2018, Weithman etal. 2018). Little
is known about factors that affect the departure decisions
and specific migratory routes that Atlantic Coast Piping
Plovers take from their breeding grounds to stopover sites
and wintering areas (Burger etal. 2011). Further, there is a
lack of information regarding the degree to which Piping
Plovers utilize shorter coastal flights (1–100 km) between
migratory stopover areas or intermediate-distance offshore
migratory flights (100–2,000 km; O’Reilly and Wingfield
1995, Hedenström et al. 2013). A large proportion of
Atlantic Coast Piping Plovers winters in the Bahamas, or
Turks and Caicos (Haig and Plissner 1993, Gratto-Trevor
et al. 2016); therefore, these individuals must under-
take sustained offshore flights during their annual cycle.
However, their migratory routes between breeding or
stopover sites and wintering areas have not yet been de-
scribed (O’Connell etal. 2011).
To help address these information gaps, we assessed
movements of adult Piping Plovers during fall migration
in relation to demographic, temporal, and meteorological
covariates. We tracked Piping Plovers using digitally
coded VHF transmitters monitored by a regional array
of automated telemetry stations along the U.S. Atlantic
coast, extending from Cape Cod, Massachusetts, to Back
Bay, Virginia, USA. We conducted this study in collab-
oration with the Motus Wildlife Tracking System, a co-
ordinated network of tagging projects and automated
telemetry stations, with project-specific regional nodes
distributed across the western Hemisphere (Taylor etal.
2017). Our specific objectives were to (1) model migra-
tory departure decisions of Piping Plovers relative to
demographic variation, temporal (diel and seasonal) vari-
ation, and meteorological conditions (i.e. wind speed,
wind direction, barometric pressure, temperature, visi-
bility, precipitation); (2) model trajectories of migratory
departure flights from breeding areas; and (3) summarize
routes, flight metrics, and weather conditions of migra-
tory flights.
METHODS
StudyArea
Our study area extended along the U.S. Atlantic coast and
adjacent waters of the Atlantic OCS that had coverage from
our regional array of automated radio telemetry stations;
it extended from Cape Cod, Massachusetts, to Back Bay,
Virginia (Figure1). As of January 2020, there were 11 BOEM
Commercial Renewable Energy Lease Areas covering 4,997
km2 within the study area (Figure 1). ese Renewable
Energy Lease Areas were located in Rhode Island Sound
and adjacent offshore waters of Massachusetts (2,106 km2),
New York Bight (321 km2), and adjacent waters offshore
of New Jersey (1,391 km2), Delaware (390 km2), Maryland
(322 km2), and Virginia (467 km2). Additional Renewable
Energy Planning Areas (under consideration for desig-
nation as lease areas) were located within our study area
off the coast of Massachusetts (1,578 km2) and New York
(7,188 km2).
Tagging sites in Massachusetts included Monomoy
National Wildlife Refuge (NWR; 41.6004°N, 69.9911°W)
and adjacent South Beach in the town of Chatham, on Cape
Cod. In 2017, these sites collectively supported 61 pairs or
about 9% of the Massachusetts population of 668 pairs of
Piping Plovers (Levasseur 2017). In Rhode Island, tagging
sites included several locations along the state’s southern
coast, ranging from Napatree Point in Westerly (41.3103°N,
71.8742°W) to Sachuest NWR in Middletown (41.4862°N,
71.2524°W). Across all sites in Rhode Island, the highest
trapping effort for Piping Plovers was on Trustom Pond
NWR (41.3695°N, 71.5809°W). Trustom Pond NWR con-
tains the highest nesting population of Piping Plovers in
Rhode Island, accounting for 31% of nesting pairs moni-
tored by USFWS staff in 2017 (J. White, USFWS, Rhode
Island Wildlife Complex, Charlestown, Rhode Island, per-
sonal communication).
Tagging and Tracking PipingPlovers
From 2015 to 2017, field staff surveyed potential Piping
Plover nesting habitat in each breeding area 3–5days per
week to monitor breeding chronology and nest success of
Piping Plovers from early May to early August. From May 9
through June 27, we trapped adult Piping Plovers during the
incubation period (3–14days prior to estimated hatching
dates) during daylight hours (approximately 0800 to 1600
hours) on days with no precipitation, fog, or windy (>15
km hr−1) conditions. At Rhode Island beaches, site man-
agers placed circular wire anti-predator exclosures over
selected nests to minimize egg depredation (Melvin et al.
1992). For exclosed nests, we used a modified trap design
by attaching hardware cloth with a mist-net funnel to the
exterior of the exclosure. For nests that were not exclosed,
we trapped adult plovers using walk-in funnel traps (Hall
and Cavitt 2012).
Each plover was banded with a single, dark blue Darvic
leg band on the right tibiotarsus and a green flag engraved
with a unique 3-digit alphanumeric code on the opposite
tibiotarsus. Coded flags were issued in collaboration with
researchers at Virginia Polytechnic Institute and State
University (Blacksburg, Virginia) as part of a larger popu-
lation dynamics study. We measured morphometrics on
all individuals including mass (±0.1g), and collected 3–5
contour feathers from each bird for molecular-based de-
termination of sex (Avian Biotech, Gainesville, Florida,
USA). We then attached a digitally coded VHF transmitter
(“nanotag”; Lotek Wireless, Ontario, Canada) by clipping
a small area of feathers from the interscapular region and
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

4 Piping Plover migration P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
gluing the tag to the feather stubble, skin, and overlaying
contour feathers with cyanoacrylate gel. In 2015 and 2016,
each plover was fitted with a 1.1-g nanotag (Lotek NTQB-
4-2; transmitter body: 12 × 8 × 8 mm). In 2017, each
plover was fitted with a 0.67-g nanotag (Lotek NTQB-3-2;
12 × 6 × 5 mm). Both tag models had a 16.5-cm antenna.
e transmitter and attachment materials weighed <3% of
the body mass of tagged plovers; <2% for the 0.67-g model.
Handling time, from capture to release, was ~15–30min
perbird.
All transmitters were programmed to emit signals at
fixed burst intervals on a shared frequency of 166.380
MHz from activation through the end of battery life.
Burst intervals were unique to each transmitter and
ranged from 4 to 6 s. The expected life of the 1.1-g
nanotags ranged from 146 days (4-s burst interval)
to 187days (6-s burst interval). The expected life of
the 0.67-g nanotags ranged from 72 days (4-s burst
interval) to 92 days (6-s burst interval). There was
no evidence that trapping or tagging plovers affected
FIGURE 1. Map of study area (2015–2017) in U.S.mid-Atlantic Bight region, showing locations of tagging sites at breeding areas in
Rhode Island (RI; blue star) and Massachusetts (MA; red star). Locations of tracking stations operated for study shown as either black
dots (for stations operated from 2015 to 2017)or black and white dots (for stations operated from 2016 to 2017). Stations within the
study area that were operated by partners in the Motus Wildlife Tracking System between 2015 and 2017 are shown as white dots.
Potential areas for offshore wind energy development (as of January 2020)within the study area are shown in green (Lease Areas) and
yellow (Planning Areas).
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton XXXX 5
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
their productivity as measured by the number of
chicks fledged per nesting attempt (Stantial et al.
2018) or their apparent annual survival rates (Stantial
etal. 2019).
A targeted array of automated radio telemetry stations
tracked tagged birds, in coordination with the broader
Motus Wildlife Tracking Network (Taylor etal. 2017). In
2015, we operated an array of 16 coastal telemetry stations
in Massachusetts, Rhode Island, and New York. During
2016, 14 additional coastal stations tracked plovers at
sites ranging from Cape Cod, Massachusetts, to Back Bay,
Virginia. During each year of the study, we downloaded
data from all stations approximately every 2 weeks from
April through November to ensure that the stations oper-
ated continuously from tag deployment through migratory
departure. Loring etal. (2019) provides a detailed descrip-
tion of the locations, specifications, and operational dates
of each tracking station.
Most of the stations operated for this study had a 12.2-m
radio antenna mast that supported six 9-element (3.3 m)
Yagi antennas mounted in a radial configuration at 60°
intervals. At some sites, stations consisted of up to 4 Yagi
antennas, or a single omni-directional antenna, attached to
existing structures. At each of the tracking stations, the an-
tennas were connected to a receiving unit (Lotek SRX) via
coaxial cables. We operated each receiving station 24 hr
per day using one 140-watt solar panel and two 12-volt
deep-cycle batteries. When tagged birds were within de-
tection range, the receivers automatically recorded trans-
mitter ID number, date, time stamp, antenna (defined by
monitoring station and bearing), and signal strength value
of each detection.
Detection range of each station varied with the height
of the receiving antennas (meters above sea level: m.a.s.l.),
altitude of the tagged bird, and the signal gain properties
of the transmitter and receiver (Loring et al. 2019). e
maximum estimated detection range of our configuration,
with receiving antennas at 12.2 m.a.s.l. was ~20 km to birds
flying at altitudes of 25 m.a.s.l. (lower limit of rotor swept
zone [RSZ] of offshore wind turbines), and ~40 km to
birds flying at altitudes of 250 m.a.s.l. (upper limit of RSZ
of offshore wind turbines). Birds flying at higher altitudes
(>1,000 m.a.s.l.) may be detected at ranges exceeding 80
km (Loring et al. 2019). Stations operated by partners in
the Motus network had a variety of configurations of an-
tennas and receiving equipment, with a typical detection
range of ~15 km (Taylor etal. 2017).
Post-processing of TelemetryData
We used the program R 3.4.1 (R Core Team 2017) and as-
sociated packages to post-process and analyze detection
data. To filter detection data, we used an algorithm in the
R package Sensorgnome (Brzustowski 2015) that removed
false detections from the raw VHF telemetry data (Loring
etal. 2019). e algorithm was based on the following de-
fault parameters applied to each unique transmitter: min-
imum of 3 consecutive bursts required to comprise a “run”
(i.e. run length), a maximum of 20 consecutive missed
bursts allowed within each run, and a maximum devi-
ation of 4ms from a tag’s unique burst interval between
its consecutive bursts (Brzustowski 2015). We selected
these parameters according to conservative recommenda-
tions from Motus network developers (Taylor etal. 2017).
In addition to data from the automated radio telemetry
stations that we operated for the present study, we also
incorporated detection data from stations that partners
operated, as part of the Motus Wildlife Tracking System
(Motus 2016).
MovementModels
A 2-beam radio propagation model estimated locations
and altitudes of tagged birds (Janaswamy 2001, Janaswamy
et al. 2018) following methods described in Loring et al.
(2019). is approach allowed for automated location es-
timation across individuals and accounted explicitly for
variation relative to beam orientation and flight altitudes
(Janaswamy 2001, Janaswamy etal. 2018). Model workflow
proceeded in 6 steps (Loring etal. 2019). In the first 2 steps,
the target bird’s location was estimated as the weighted
mean among sequential locations: this we weighted by the
inverse-square discrepancy in signal strength among all
near-simultaneous detections, resulting in the lowest dis-
crepancy between measured and predicted signal strength.
We constrained these calculations by differentiating be-
tween local movements (at breeding or stopover areas)
and nonstop flight (regional or migratory) movements.
e constraints included (1) limits to a bird’s possible flight
speeds in the horizontal and vertical planes and (2) the
assumption that, during directed flight, a bird limits vari-
ation in its horizontal and vertical speed. We constrained
maximum flight speeds at 12 m s−1 for Piping Plovers
(Hedenstrӧm etal 2013, Stantial and Cohen 2015). For the
third step, we interpolated the estimated locations to 1-min
time steps using a Brownian Bridge movement model to
interpolate the temporally irregular detection sequences to
regular intervals (Horne etal. 2007). We selected a 1-min
time window to estimate locations as it represented move-
ments at approximately a 1-km scale (given maximal flight
speeds). is also helped to optimize the tradeoff between
the advantage of adding more information (detections) to
co-locate position, and the disadvantage of the bird’s actual
position changing within the timewindow.
In the fourth step, we downloaded meteorological data
from the National Centers for Environmental Prediction
North American Regional Reanalysis (NARR; National
Oceanic and Atmospheric Administration 2017), which
covered the study area at ~32-km2 spatial resolution
and 3-hr temporal resolution. We interpolated this
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

6 Piping Plover migration P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
3-dimensional meteorological data to each 1-min record,
and derived orientation and airspeed from flight speed and
wind data (Kemp etal. 2012). In the fifth model step, we
quantified occurrence in offshore waters using the output
from the Brownian Bridge model, and calculated uncer-
tainty as the standard deviation of location estimates in
the horizontal plane. Finally, in the sixth model step, we
extracted the magnitude of all meteorological and flight
speed–related covariates to assess incidence in offshore
waters, including flight direction and heading, wind sup-
port, and crosswinds.
Timing of Migratory Departure
We classified migratory departure events as nonstop
southbound departure flights from breeding areas to
nonbreeding grounds that were tracked by 2 or more sta-
tions within the telemetry array. Departure dates were as-
signed (day of year, with January 1 = day 1)corresponding
to the onset of each departure event. To examine the timing
of departure relative to daylight, we used the R package
maptools (Bivand and Lewin-Koh 2016) to calculate the
local time (in hours, EST) of sunset at each modeled lo-
cation estimate. We then calculated the difference in time
(in hours) between the local sunset and onset of migratory
departure events. We used a 2-sample Mann–Whitney
U-test in base R (function: wilcox.test) to compare timing
of departure relative to the timing of local sunset between
breeding locations (Massachusetts and Rhode Island).
Covariate Analysis of Migratory Departure Decisions
We performed an integrated analysis of all covariates (tem-
poral, demographic, and meteorological) to predict migra-
tory departure events using a nonlinear binomial logistic
regression method, boosted generalized additive models
(GAMs, R package mboost using function gamboost; see
also Bühlmann and Hothorn 2007). e nonlinear spe-
cification of these models allowed for flexibility in the
response–covariate relationships and aligned with our ob-
jective of prioritizing explanatory over predictive power.
We included the following covariates in the boosted GAM
model: bird ID (random intercept), day of year, wind dir-
ection (circular, in degrees true N), wind speed (m s−1),
precipitation accumulation (kg m−2), visibility (m), Δ air
temperature (the change in air temperature over the pre-
ceding 24-hr period, in °C), and Δ pressure (the change in
pressure over the preceding 24-hr period, in Pa). We also
included 2 first-order interaction terms: date*location (MA
or RI) and date*sex (male or female).
We chose an inverse logit-link regression formulation,
calculating daily migratory departure events (coded as
1)and nonevents (coded as 0)for each individual, starting
on the conclusion (fledge or fail date) of their final nesting
attempt and ending on the date that occurred 24hr prior to
the onset of migratory departure. For days when birds did
not depart (i.e. nonevents), we calculated the daily mean of
each meteorological covariate within ±3hr of local sunset
to represent conditions when birds could have left because
78% of actual departure events occurred within this time
window. For departure events, we calculated the mean of
each meteorological covariate within 3hr prior to the onset
of departure, to represent conditions that plovers experi-
enced prior to takeoff. We calculated means of meteoro-
logical covariates using the R packages plyr (Wickham
2011) and lubridate (Grolemund and Wickham 2011).
We calculated the mean wind direction that the wind was
blowing toward based on the circular distribution using
the package Circular (Agostinelli and Lund 2017).
e boosted GAM approach allowed us to estimate
both the relative “influence” of covariates on migratory de-
parture (i.e. the percentage reduction in deviance attribut-
able to each predictor), and the “relative” response to these
covariates (Hastie etal. 2009). In this formulation, we in-
corporated probability of migratory departure as an “in-
verse logit-link,” with responses to each covariate presented
as partial contributions to the likelihood (log-transformed
odds ratio) of a migratory departure event occurring (i.e.
the higher the contribution, the increased predicted likeli-
hood of a migratory departure event). Responses represent
the contribution of a given covariate to the likelihood of
migratory departure, quantified by log-transformed odds
ratio of migratory departure.
Additional advantages of this boosted GAM method
are that, in being additive, it fits nonlinear and inde-
pendent responses to each covariate. e boosted GAM
approach iteratively summed simple regressions based on
single-covariate “learner” functions, each chosen to min-
imize an equivalent loss function based on binomial pre-
dictors (see Bühlmann and Hothorn 2007). e additive
approach facilitated estimation of the relative “influence”
of each covariate, using the number of boosts choosing
that covariate, to minimize the current loss. We selected
model parameters to reduce possible bias and overfitting
(Bühlmann and Hothorn 2007), an additional advantage
of boosted methods over (non-boosted) GLMs or GAMs,
which can be prone to overfitting (Randin et al. 2006).
We fit the model incrementally using small step sizes or
“shrinkage” (default 0.25) of each iterative sub-model
(Maloney et al. 2012). We used 1,000 boosts per analysis
and verified that this was a reasonable number of iter-
ations using the function cvrisk (cross-validated risk) with
a specific number of separate “folds” (i.e. 4 independently
sampled fits). We fit responses to the categorical covariates
(sex and location) using linear learner functions (resulting
in fixed effects for each category). Responses to each in-
dividual (bird ID) were treated as random intercepts, and
responses to all the meteorological covariates were fit
using cubic p-splines. e package also allowed cyclical re-
sponses to the periodic covariates (wind direction). Finally,
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton XXXX 7
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
to assess the significance of the predicted covariate re-
sponses, we performed a bootstrap analysis using function
confint with 1,000 model fits to produce 95% confidence
intervals for each covariate response.
We mapped migratory trajectories of all Piping Plovers
tracked during departure from breeding areas that had
nonstop flight speeds ≥5 m s−1. We used these tracks to
calculate summary statistics (mean, SD, range) of migra-
tory flights in the mid-Atlantic Bight region. Flight metrics
included duration (hr), distance (km), speed (km hr−1), and
altitude (in m.a.s.l.). We report summary statistics of me-
teorological conditions associated with nonstop migratory
flights (i.e. wind direction, wind speed, wind support, visi-
bility, air temperature, and atmospheric pressure).
RESULTS
Tag Attachment and Retention
From 2015 to 2017, we tagged 50 adult Piping Plovers
annually at Monomoy NWR and adjacent beaches in
Chatham, Massachusetts (n = 25 per year), and on beaches
in southern Rhode Island (n = 25 per year) from Napatree
Point in Westerly to Sachuest NWR in Middletown. Based
on genetic analysis of contour feathers, 52% (n = 150) of
tagged plovers were females, 45% were male, and the sex
of the remaining 3% was undetermined; sex ratios were
unbiased across sites. Based on observations by field staff,
25% of plovers in the study dropped their transmitters on
the breeding grounds prior to post-breeding migration
(range: 16–32% of plovers observed with dropped tags an-
nually). e number of dropped transmitters was lowest
in 2017 when we used a lighter (0.67 g) model of trans-
mitter. We detected plovers with active transmitters by the
tracking array for a mean of 46days (SD = 27days, range:
0–102days).
Timing of Migratory Departure
e automated telemetry array detected migratory depar-
tures of 65 Piping Plovers from 2015 to 2017 (2015: n = 19;
2016: n = 20; 2017: n = 26), with flights for 39 plovers
from breeding areas in Massachusetts (n = 20 females,
n = 19 males) and 26 plovers from breeding areas in Rhode
Island (n = 13 females, n = 13 males). Overall, most tagged
plovers departed in a 10-day window between July 19 and
July 29 (25th–75th quartiles; Figure2).
Most (78%) departure flights from breeding areas were
initiated within 3 hr of local sunset, with variation in
timing of departure relative to sunset by location (W = 304,
P = 0.006; Figure 3). Plovers from Massachusetts de-
parted an average of 1.91hr before timing of local sunset
(SD = 2.67hr, range: 4.57hr before to 8.01 hr after local
sunset). Plovers from breeding areas in Rhode Island de-
parted an average of 0.69 hours before timing of local
sunset (SD = 3.3hr, range: 10.72hr before to 6.01hr after
local sunset).
Covariate Analysis of Migratory Departure Decisions
Wind direction and date were the strongest predictors of
migratory departure of Piping Plovers from their breeding
grounds based on the boosted GAM covariate analysis
(Table 1). Peak departures occurred when winds were
blowing to the southwest (Figure4A), from late July through
early August (Figure4B). Interaction terms with breeding
area and date indicated that plovers from Massachusetts
departed slightly later (through early September) rela-
tive to plovers from Rhode Island (Figure 4C), and that
males were more likely to depart later relative to females
(Figure4D). ere were weak associations with migratory
FIGURE 2. Boxplots of migratory departure dates by sex for
Piping Plovers tagged in Massachusetts (MA; n = 20 females and
19 males) and coastal Rhode Island (RI; n = 13 females and 13
males), USA, from 2015 to 2017, showing median (bold midline),
third and first quartiles (upper and lower limits of the box), inter-
quartile range × 1.5 (whiskers), and outliers (points).
FIGURE 3. Timing of migratory departure of tagged Piping
Plovers (n = 65) from breeding areas in Massachusetts and Rhode
Island, USA, relative to timing of local sunset (hours EST), 2015 to
2017.
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

8 Piping Plover migration P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
departures during decreasing air temperatures (Figure4E)
and increasing atmospheric pressure (Figure4F) over the
preceding 24-hrperiod.
Migratory Departure Trajectories
e array tracked migratory trajectories for 33 plovers
from breeding areas in Massachusetts and 19 plovers
from breeding areas in Rhode Island (Figures 5 and 6).
Among plovers tracked during departure from breeding
areas in Massachusetts, 91% (n = 30) followed a south-
southwest trajectory across Nantucket Sound, and the
remaining 9% (n = 3) departed to the west across Rhode
Island Sound toward Long Island, New York. Most (67%,
n = 22) plovers tracked during departure from breeding
areas in Massachusetts were last detected by the telemetry
array while in flight over waters south of Nantucket, due
in part to limited numbers of stations in the mid-Atlantic
region during 2015 (Figure5). e telemetry array tracked
flights of the remaining 33% (n = 11) offshore across the
mid-Atlantic Bight to coastal areas ranging from Long
Island, New York, to North Carolina.
All Piping Plovers tracked during migration from
breeding areas in Rhode Island (n = 19) departed on south-
southwest trajectories between Block Island Sound and
eastern Long Island Sound and 68% (n = 13) were last de-
tected within this region. e remaining 32% (n = 6) were
tracked offshore across the mid-Atlantic Bight to coastal
areas ranging from New Jersey to North Carolina.
Migration Routes Across the Mid-AtlanticBight
e automated radio telemetry array tracked migra-
tory flights of 17 plovers (n = 11 from Massachusetts and
n = 6 from Rhode Island) across the mid-Atlantic Bight
(Figure 6). Mean model uncertainty (68th percentile
error) in the x and y coordinates was 23 km (SD = 13 km,
range: 9–53 km). Mean distance of flights tracked across
the mid-Atlantic Bight was 579 km (SD = 209 km, range:
163–811 km). Mean duration of flights tracked was 17.5hr
(SD = 10.4hr, range: 3.0–39.8hr), with a mean estimated
flight speed of 42 km hr−1 (SD = 17 km hr−1, range: 20–72
km hr−1). Based on model estimates, mean altitude of off-
shore flights across the mid-Atlantic Bight was 288 m.a.s.l.
(SD = 79 m.a.s.l., overall range of model uncertainty:
36–1,031 m.a.s.l.).
Piping Plovers crossed the mid-Atlantic Bight
when winds were blowing to the southwest (circular
mean = 238°) at a mean wind speed of 7.8 m s−1 (SD = 3.0
m s−1; range: 2.6–13.5 m s−1), which provided a mean
wind support of 4.3 m s−1 (SD = 5.7 m s−1; range: −5.8 to
11.8 m s−1; Appendix Table 2). During offshore flights,
visibility was high (mean = 18 km; SD = 19 km, range:
14–20 km), precipitation was variable (mean = 0.27 kg
m−2; SD = 0.39kg m−2, range: 0–1.27kg m−2), mean air
temperature was 22°C (SD = 3°C; range: 19–28°C), and
mean atmospheric pressure was 101,295 Pa (SD = 389
Pa; range: 100,709–102,139 Pa).
DISCUSSION
We used a network of automated telemetry stations to
model the fall migration ecology of the federally threat-
ened Atlantic Coast Piping Plover in relation to pro-
posed offshore wind energy developments in the region.
Most Piping Plovers initiated migration during the post-
breeding period in mid- to late July, within 3hr of local
sunset, when winds were blowing to the southwest. ese
wind conditions supported direct, offshore flights from
breeding areas in southern New England to stopover areas
in the mid-Atlantic. Our study provides the first empirical
evidence that Piping Plovers migrate across the Atlantic
OCS, rather than taking a more circuitous route along the
coast, addressing a key information gap for this species
(Burger etal 2011).
As with many other avian species, Piping Plovers in the
present study initiated migration near sunset on evenings
with meteorological conditions advantageous to sustained
flight, such as wind assistance and the passage of fronts
TABLE 1. Fitting functions and selection frequencies of environmental and temporal covariates utilized in a binomial Boosted GAM
analysis of migratory departures of tagged Piping Plovers (n = 65) from breeding areas in Massachusetts and Rhode Island, USA,
2015–2017.
Covariate (units) Fitting function Selection frequency
Wind direction (degrees true N) cyclical p-spline 0.35
Date p-spline 0.27
Date * Location p-spline * categorical interaction 0.13
Date * Sex p-spline * categorical interaction 0.12
Δ Air temperature (°C) p-spline 0.11
Δ Pressure (Pa) p-spline 0.01
Bird ID Random intercept 0.00
Wind speed (m s−1) p-spline 0.00
Precipitation (kg m−2) p-spline 0.00
Visibility (m) p-spline 0.00
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton XXXX 9
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
(e.g., falling air temperatures, rising atmospheric pressure;
Brooks 1965, Able 1973, Richardson 1978, Gill etal. 2014,
Shamoun-Baranes etal. 2017, Anderson etal. 2019). Wind
assistance reduces energy expenditure during long-distance
flights, thus wind selectivity prior to departure is thought
to be one of the primary factors determining departure de-
cisions (Richardson 1978, Butler etal. 1997, Dossman etal.
2016, McCabe et al. 2017, Wright etal. 2018). Nocturnal
migration is also thought to be advantageous for some
species due to increased diurnal foraging opportunities
prior to and after a migration bout, and reduced predation
risk from raptors (Kerlinger and Moore 1989, Lank 1989,
Alerstam 2009). In addition, atmospheric conditions may
be more favorable to migratory flights at night due to re-
ductions in turbulence and evaporative water loss, relative
to daytime conditions when winds tend to be stronger and
the air less humid (Kerlinger and Moore 1989). ese con-
ditions supported a shorter direct ocean crossing to stop-
over areas in the mid-Atlantic, rather than a longer route
following thecoast.
Assessments of avian collision risk with offshore wind
turbines require information on flight relative to the
FIGURE 4. Predicted effects of covariates on migratory departure decisions of Piping Plovers (n = 65) from breeding areas in
Massachusetts (MA) and Rhode Island (RI), USA, from 2015 to 2017: (A) wind direction (in degrees clockwise from geographic north
that the wind is blowing toward); (B) date; (C) date*location interaction term (with location “MA” shown); (D) date*sex interaction term
(with sex “male” shown); (E) Δ air temperature (change in °C over the preceding 24-hr period, where negative values indicate decreasing
temperatures and positive values indicate increasing temperatures); (F) Δ air pressure covariate (change in Pa over the preceding 24-hr
period, where negative values indicate decreasing pressure and positive values indicate increasing pressure). The x-axis shows the
Boosted GAM prediction for the partial contribution of each covariate. The y-axis shows the likelihood (log-transformed odds ratio or
f-partial) of migratory departure among Piping Plovers. The gray-shaded area represents the 95% confidence interval for the response
based on 1,000 bootstrapped models.
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

10 Piping Plover migration P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
Rotor Swept Zone (RSZ; Masden and Cook 2016), gen-
erally 25–250 m.a.s.l. Flight altitudes of Piping Plovers
during migration have not been previously described, and
this represents a significant information gap in assess-
ments of risk from offshore wind energy developments
to this species (Burger et al. 2011). In the present study,
we applied models based on the theoretical relationship
between horizontal detection range of signals received
by automated radio telemetry stations, which increases
with transmitter height above ground, to coarsely esti-
mate flight altitudes when plovers were detected by 2 or
more spatially separated stations simultaneously. ese
estimates indicated that mean offshore migratory flight
altitudes of Piping Plovers crossing the mid-Atlantic Bight
were mostly within or above the RSZ of offshore wind tur-
bines. However, due to the coarse scale at which flight alti-
tude was estimated, the estimates of exposure to the RSZ
should be interpreted in the context of the model range
(uncertainty) in plausible altitudes, which generally ex-
ceeded the range in estimated altitudes (Appendix Table
2). us, more detailed information on the migratory alti-
tudes of Piping Plovers is needed to fully assess risks asso-
ciated with developing offshore wind turbines throughout
their migratoryrange.
FIGURE 5. Modeled trajectories of tagged Piping Plovers from breeding areas in Rhode Island (RI; n = 13 in blue) and Massachusetts
(MA; n = 22 in red), 2015–2017, showing individuals that were tracked through migratory departure from breeding areas.
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton XXXX 11
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
Information from offshore radar studies has re-
corded shorebirds migrating at altitudes exceeding 1–2
km (Richardson 1976, Williams and Williams 1990),
whereas nearshore studies documented local and migra-
tory flights of shorebirds occurring at altitudes <100 m
(Dirksen etal. 2000, Langston and Pullan 2003). Risk of
exposure to rotor swept altitudes may increase during
takeoff and landing from stopover areas, emphasizing the
need for determining setback distances when developing
turbines near migratory stopover areas (Howell et al.
2019). In addition, flight altitudes of migratory birds
may vary in response to weather as they search to find
suitable tailwinds (Shamoun-Baranes et al. 2017, Senner
etal. 2018). Migratory birds may also descend to lower
altitudes during periods of limited visibility, low cloud
ceiling, and/or inclement weather, increasing their risk of
collision with offshore wind turbines (Hüppop etal. 2006,
Senner etal. 2018). In addition, risk of collision is poten-
tially higher at night due to reduced visibility of turbines
(Exo etal. 2003) and attraction or disorientation effects
from artificial lighting on turbine towers (Richardson
2000, Drewitt and Langston 2006). Future efforts to as-
sess fine-scale movements of Piping Plovers will be of
continued importance as additional wind energy facilities
FIGURE 6. Modeled migratory routes of tagged Piping Plovers from breeding areas in Rhode Island (RI; n = 6) and Massachusetts (MA;
n = 11), 2015–2017, showing individuals that were tracked across a broader portion of the mid-Atlantic Bight.
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

12 Piping Plover migration P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
are developed in offshore waters and tracking technology
continues to improve. Detailed tracking of flight altitudes
and avoidance behavior is beyond the ability of current
VHF technology within the Motus Network, although de-
velopment of lightweight GPS transmitters (Senner etal.
2018) or VHF tags with embedded altimeters (Bowlin
etal. 2015) may provide viable options for tracking fine-
scale 3-D flight paths of small-bodied shorebirds in the
nearfuture.
Results from this study address high-priority infor-
mation needs on the timing, conditions, and routes of
Piping Plovers in offshore environments to support as-
sessments of developing wind energy facilities throughout
a portion of the U.S. Atlantic, extending from Cape Cod,
Massachusetts, to Back Bay, Virginia. However, due to
incomplete coverage from Motus network tracking sta-
tions along U.S. Atlantic coast, we limited the spatial scale
of the analysis of movements to the bounds of the study
area in the U.S.mid-Atlantic region. e study area con-
tained a regional array of tracking towers that we stra-
tegically erected at coastal sites, spanning from Cape Cod,
Massachusetts, to the north to Back Bay, Virginia, to the
south, with direct line-of-sight to offshore areas of the
U.S.mid-Atlantic Bight. Each tower was 10.2 m tall and
had 6 high-range directional antennas arranged radially
to track movements of birds in all directions. is design
attempted to maximize the detection range and direction-
ality of land-based towers but had limited coverage for
detecting birds in offshore areas of the U.S. Atlantic OCS
beyond 20 km from land. As offshore lease areas move
into development phases, deployment of automated radio
telemetry equipment on offshore structures offers a prom-
ising approach for collecting more detailed data needed
for collision risk models, including information on pas-
sage rates through individual lease areas, diurnal vs. noc-
turnal flight activity, and coarse information on avoidance
rates and flight altitudes.
Since large areas for development of offshore wind
energy facilities are under consideration to the south
of our regional telemetry array, including off the coast
of North Carolina, USA, there is a need for more com-
plete information on the movements of Piping Plovers
throughout their entire migratory range to fully assess
risk. Major migratory stopover areas for Piping Plovers
in the mid-Atlantic include Ocracoke, North Carolina,
where Weithman et al. (2018) estimated use by 15% of
the Atlantic coast population with the first peak of mi-
grants arriving in late July. Piping Plovers from breeding
areas in New England remained at Ocracoke for over
40days (Weithman etal. 2018), suggesting this may be an
important stopover site for adults to complete prebasic
molt-migration (Tonra and Reudink 2018) before moving
on to wintering areas farther south (Gratto-Trevor et al.
2012, 2016; Cohen etal. 2018). us, Piping Plovers using
this stopover area may be at risk of passing through lease
areas off the coast of North Carolina, particularly if they
depart along the direct route toward the Caribbean where
over 30% of the population is estimated to winter (Gratto-
Trevor etal. 2016).
Fully estimating exposure and collision risk of Piping
Plovers to offshore wind turbines requires tracking tech-
nology capable of collecting high-resolution movement
and altitude data throughout the entire migratory range
and full annual cycle. GPS tracking technology may pro-
vide a viable solution for collecting high-resolution,
3-D movement data of small-bodied shorebirds in the
near future, as lightweight transmitters become more
widely available (Senner etal. 2018). Data on the migra-
tory routes and flight altitudes of Piping Plovers from
breeding areas throughout the Atlantic Coast is needed
to fully assess population-level risks, as widespread de-
velopment of offshore lease areas is planned throughout
a large portion of the Atlantic OCS (BOEM 2019). ere
is presently a lack of information on the movements of
Piping Plovers during spring (northbound) migration.
Shorebirds may be more likely to migrate during in-
clement weather in spring due to less stable atmospheric
conditions and time constraints to reach breeding areas
(O’Reilly and Wingfield 1995). ese conditions may lead
to increased risk during spring relative to fall, including
increased exposure to offshore wind turbines and other
flight hazards (Richardson 2000). Future efforts to track
full annual cycle movements of Piping Plovers and other
avian species of conservation concern will be critical for
assessments of cumulative impacts resulting from de-
velopment of multiple offshore wind energy facilities
throughout the migratory range.
ACKNOWLEDGMENTS
We thank the following individuals from the Bureau of
Ocean Energy Management (David Bigger, Mary Boatman,
Jim Woehr [retired], and Tim White), the U.S. Fish and
Wildlife Service (Scott Johnston, Caleb Spiegel, Pamela
Toschik, Anne Hecht, Suzanne Paton, Susi vonOettingen),
and UMass Amherst (Curt Griffin and Paul Sievert) for
guidance and oversight. We thank Annelee Motta and
Laurie Racine (USFWS) and Deb Wright (MA Cooperative
Fish and Wildlife Unit) for administrative support. We
thank staff from the Rhode Island National Wildlife Refuge
Complex and Monomoy National Wildlife Refuge for
equipment and logistical support. We thank Dan Catlin and
the Virginia Tech Shorebird Program for coordinating leg
flag combinations and resights of Piping Plovers. We thank
our field staff: Brett Still, Michael Abemayor, Christine
Fallon, Steve Ecrement, Calvin Ritter, Derek Trunfio, Kaiti
Titherington, Josh Siebel, Adam Ellis, Jennifer Malpass,
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton XXXX 13
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
Kevin Rogers, Emma Paton, Gillian Baird, and Alex Cook.
For field and logistical support with automated radio te-
lemetry stations operated for this study, we thank our many
cooperators from the following entities: UMass Amherst-
USGS Cooperative Fish and Wildlife Unit, USFWS
Southern New England-New York Bight Coastal Program,
USFWS Division of Migratory Birds, University of Rhode
Island, Cape Cod National Seashore, Eastern MA National
Wildlife Refuge (NWR) Complex, Waquoit Bay National
Estuarine Research Reserve, US Army Corps of Engineers/
Cape Cod Canal Field Office, Rhode Island NWR Complex,
Nantucket Islands Land Bank, Nantucket Conservation
Foundation, Napatree Point Conservation Area, CT
Department of Energy & Environmental Protection,
American Museum of Natural History/Great Gull Island
Project, Plum Island Animal Disease Center, Block Island
Southeast Lighthouse Foundation, Camp Hero State Park,
Fire Island National Seashore, Gateway National Recreation
Area, Wildlife Conservation Society/New York Aquarium,
Rutgers University Marine Field Station, Conserve Wildlife
Foundation of New Jersey, New Jersey Division of Fish
and Wildlife, Avalon Fishing Club, DE Department of
Natural Resources/Cape Henlopen State Park, The Nature
Conservancy Virginia Coast Reserve, Chincoteague NWR,
Eastern Shore of VA NWR, Back Bay NWR, NOAA R/V
Henry Bigelow, and Shearwater Excursions. For technical
support and assistance with data management and analysis,
we thank Stu Mackenzie and Zoe Crysler (Motus Wildlife
Tracking System, Bird Studies Canada); Phil Taylor and John
Brzustowski (Acadia University); Paul Smith (Environment
and Climate Change Canada); Mike Vandentillart (Lotek
Wireless); and Ramakrishna Janaswamy and Hua Bai
(UMass Amherst). The findings and conclusions in this ar-
ticle are those of the author(s) and do not necessarily repre-
sent the views of the U.S. Fish and Wildlife Service.
Funding statement: Funding was provided in part by
the US Department of the Interior, Bureau of Ocean
Energy Management, Environmental Studies Program,
Washington DC, through Intra-Agency Agreement Number
M13PG00012 with the Department of Interior, Fish and
Wildlife Service. This study was also supported through the
National Science Foundation sponsored Integrated Graduate
Research Traineeship: Offshore Wind Energy Engineering,
Environmental Science, and Policy (Grant Number 1068864)
at the University of Massachusetts Amherst.
Ethics statement: This research was conducted with the fol-
lowing permits and permissions: University of Rhode Island
Animal Care and Use Committee (AN1415-012), University
of Massachusetts Amherst Institutional Animal Care and
Use Committee (2014-0024) and Federal Bird Banding
Permit22739.
Author contributions: P.H.L.and P.W.C.P.conceived the idea
and design; P.H.L.and P.W.C.P.collected field data; P.H.L.and
P.W.C.P. wrote the paper, H.F.G. edited the paper; P.H.L.,
J.D.M., H.F.G., and P.W.C.P. developed the methods; P.H.L.,
J.D.M., and H.F.G.analyzed thedata.
Data depository: Analyses reported in this article can be re-
produced using the data provided by Loring etal. (2020).
Conflict of interest statement : The authors have no conflicts
of interest to declare.
LITERATURECITED
Able,K.P. (1973). The role of weather variables and ight direction
in determining the magnitude of nocturnal bird migration.
Ecology 54:1031–1041.
Agostinelli,C., and U.Lund (2017). R package ‘Circular’: Circular
statistics, version 0.4–93. https://cran.r-project.org/web/
packages/circular/
Alerstam,T. (2009). Flight by night or day? Optimal daily timing
of bird migration. Journal of Theoretical Biology 258:530–536.
Anderson,A. J., S. Duijns, P.A. Smith, D. Friis, and E. Nol (2019).
Migration distance and body condition inuence shorebird
migration strategies and stopover decisions during south-
bound migration. Frontiers in Ecology and Evolution 7:251.
doi:10.3389/fevo.2019.00251
Bivand,R., and N.Lewin-Koh (2016). R package ‘Maptools’, version
0.9–2. https://cran.r-project.org/web/packages/maptools/
BOEM (2017). Environmental Studies Program Studies
Development Plan, 2018–2020. https://www.boem.gov/
FY-2018-2020-SDP/
BOEM (2019). Renewable Energy GIS Data (version 2 Feb 2019).
https://www.boem.gov/Renewable-Energy-GIS-Data/
Bowlin,M.S., D.A. Enstrom, B. J. Murphy, E.Plaza, P. Jurich, and
J.Cochran (2015). Unexplained altitude changes in a migrating
thrush: Long-ight altitude data from radio-telemetry. The
Auk: Ornithological Advances 132:808–816.
Brooks,W.S. (1965). Eect of weather on autumn shorebird mi-
gration in east-central Illinois. The Wilson Bulletin 77:45–54.
Brzustowski, J. (2015). R package ‘SensorGnome’, version 1.0.16.
https://www.sensorgnome.org/
Bühlmann, P., and T. Hothorn (2007). Boosting algorithms:
Regularization, prediction and model tting. Statistical
Science 22:477–505.
Burger,J., C.Gordon, L.Niles, J.Newman, G.Forcey, and L.Vlietstra
(2011). Risk evaluation for federally listed (Roseate Tern, Piping
Plover) or candidate (Red Knot) bird species in oshore waters:
A rst step for managing the potential impacts of wind fa-
cility development on the Atlantic Outer Continental Shelf.
Renewable Energy 36:338–351.
Butler,R.W., T.D.Williams, N.Warnock, and M.A.Bishop (1997).
Wind assistance: Arequirement for migration of shorebirds?
The Auk 113:456–466.
Cohen,J.B., S.B.Maddock, M.K.Bimbi, W.W.Golder, O.E.Ledee,
F.J. Cuthbert, D.H.Catlin, J.D. Fraser, and C.L. Gratto-Trevor
(2018). State uncertainty models and mark–resighting models
for understanding non-breeding site use by Piping Plovers.
Ibis 160:342–354.
Desholm,M., A. D.Fox, P. D.L. Beasley, and J.Kahlert (2006).
Remote techniques for counting and estimating the
number of bird–wind turbine collisions at sea: Areview. Ibis
148:76–89.
Dirksen,S., A.L.Spaans, and J.vanderWinden (2000). Studies on
nocturnal ight paths and altitudes of waterbirds in relation to
wind turbines: Areview of current research in the Netherlands.
In Proceedings of the National Avian–Wind Power Planning
Meeting III, San Diego, CA, USA. pp.97–109.
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

14 Piping Plover migration P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
Dossman, B. C., G. W. Mitchell, D. R. Morris, P. D. Taylor,
C.G.Guglielmo, S.N.Matthews, and P.G.Rodewald (2016). The
eects of wind and fuel stores on stopover behavior across a
migratory barrier. Behavioral Ecology 27:567–574.
Drewitt,A.L., and R.H.Langston (2006). Assessing the impacts of
wind farms on birds. Ibis 148:29–42.
Elliott-Smith,E., and S. M. Haig (2020). Piping Plover (Charadrius
melodus), version 1.0. In Birds of the World (A.F.Poole, Editor).
Cornell Lab of Ornithology, Ithaca, NY, USA. https://doi.
org/10.2173/bow.pipplo.01
Exo, K. M., O. Hüppop, and S. Garth (2003). Birds and oshore
wind farms: Ahot topic in marine ecology. Wader Study Group
Bulletin 100:50–53.
Fox, A. D., M. Desholm, J. Kahlert, T. K. Christensen, and
I.B.K.Petersen (2006). Information needs to support environ-
mental impact assessment of the eects of European marine
oshore wind farms on birds. Ibis 148:129–144.
Furness,R.W., H.M.Wade, and E.A.Masden (2013). Assessing vul-
nerability of marine bird populations to oshore wind farms.
Journal of Environmental Management 119:56–66.
Gill,R.E., D.C.Douglas, C.M.Handel, T.L.Tibbitts, G.Huord, and
T.Piersma (2014). Hemispheric-scale wind selection facilitates
Bar-tailed Godwit circum-migration of the Pacic. Animal
Behavior 90:117–130.
Gratto-Trevor, C., D. Amirault-Langlais, D. Catlin, F. Cuthbert,
J. Fraser, S. Maddock, E. Roche, and F. Shaer (2012).
Connectivity in Piping Plovers: Do breeding populations
have distinct winter distributions. The Journal of Wildlife
Management 76:348–355.
Gratto-Trevor,C., S.M.Haig, M.P.Miller, T.D.Mullins, S.Maddock,
E.Roche, and P. Moore (2016). Breeding sites and winter site
delity of Piping Plovers wintering in The Bahamas, a pre-
viously unknown major wintering area. The Journal of Field
Ornithology 87:29–41.
Grolemund, G., and H. Wickham (2011). Dates and times made
easy with lubridate. Journal of Statistical Software 40:1–25.
Haig,S.M., and J.H.Plissner (1993). Distribution and abundance
of Piping Plovers: Results and implications of the 1991 inter-
national census. The Condor 95:145–156.
Hall, L. K., and J. F. Cavitt (2012). Comparative study of trap-
ping methods for ground-nesting shorebirds. Waterbirds
35:342–346.
Hastie,T., R.Tibshirani, and J. Friedman (2009). The Elements of
Statistical Learning. Springer, New York, NY, USA.
Hedenstrӧm, A., R. H. G. Klaassen, and S. Akesson (2013).
Migration of the Little Ringed Plover Charadrius dubius
breeding in South Sweden tracked by geolocators. Bird
Study 60:466–474.
Holberton, R. L., P. D. Taylor, L. M Tudor, K. M. O’Brien,
G. H. Mittelhauser, and A. Breit (2019). Automated VHF
radiotelemetry revealed site-specic dierences in fall migra-
tion strategies of Semipalmated Sandpipers on stopover in
the Gulf of Maine. Frontiers in Ecology and Evolution 7:327.
doi:10.3389/fevo.2019.00327
Horne, J. S., E. O. Garton, S. M. Krone, and J. S. Lewis (2007).
Analyzing animal movements using Brownian bridges.
Ecology 88:2354–2363.
Howell,J.E., A.E.McKellar, R.H.Espie, and C.A.Morrissey (2019).
Predictable shorebird departure patterns from a staging site
can inform collision risks and mitigation of wind energy devel-
opments. Ibis 162:535–547. doi:10.1111/ibi.12771
Hüppop,O., J.Dierschke, K.M.Exo, E.Fredrich, and R.Hill (2006).
Bird migration studies and potential collision risk with o-
shore wind turbines. Ibis 148:90–109.
Janaswamy, R. (2001). Radiowave Propagation and Smart
Antennas for Wireless Communication. Springer Science and
Business Media, New York, NY, USA.
Janaswamy, R., P. H. Loring, and J. D. McLaren (2018). A state
space technique for wildlife position estimation using non-
simultaneous signal strength measurements. arXiv.org: Electric
Engineering & Systems Science. arXiv:1805.11171v1 [eess.SP].
Johnston,A., A.S.C. P. Cook, L.J.Wright, E. M. Humphreys, and
N.H.K.Burton (2014). Modelling ight heights of marine birds
to more accurately assess collision risk with oshore wind tur-
bines. Journal of Applied Ecology 51:31–41.
Kemp,M. U., J.Shamoun-Baranes, E. E. vanLoon, J.D.McLaren,
A. M. Dokter, and W. Bouten. (2012). Quantifying ow-
assistance and implications for movement research. Journal of
Theoretical Biology 308:56–67.
Kerlinger,P., and F. R. Moore (1989). Atmospheric structure and
avian migration. Current Ornithology 6:109–142.
Langston, R., and J. Pullan (2003). Windfarms and birds: An Analysis
of the eects of windfarms on birds, and guidance on environ-
mental assessment criteria and site selection issues (Report
No. T-PVS/Inf (2003) 12). Report by BirdLife International.
Lank, D. B. (1989). Why y by night? Inferences from tidally-
induced migratory departures of sandpipers. Journal of Field
Ornithology 60:154–161.
Levasseur,P. (2017). Summary of the 2017 Massachusetts Piping
Plover Census. Massachusetts Division of Fisheries and Wildlife,
Westborough, MA, USA.
Loring,P.H., J.D.McLaren, H.Goyert, and P.W.C.Paton (2020).
Data from: Supportive wind conditions influence offshore
movements of Atlantic Coast Piping Plovers during fall mi-
gration. The Condor: Ornithological Applications 122:1–16.
doi:10.5061/dryad.ghx3ffbkf.
Loring, P. H., J. D. McLaren, P. A. Smith, L. J. Niles, S. L. Koch,
H. F. Goyert, and H. Bai (2018). Tracking movements of
threatened migratory rufa Red Knots in U.S. Atlantic Outer
Continental Shelf Waters. OCS Study BOEM 2018–046 nal re-
port. U.S. Department of the Interior, Bureau of Ocean Energy
Management, Sterling, VA, USA.
Loring,P. H., P. W.C. Paton, J. D. McLaren, H. Bai, R. Janaswamy,
H.F.Goyert, C. R. Grin, and P.R.Sievert (2019). Tracking o-
shore occurrence of Common Terns, endangered Roseate Terns,
and threatened Piping Plovers with VHF arrays. OCS Study
BOEM 2019-017 nal report. U.S. Department of the Interior,
Bureau of Ocean Energy Management, Sterling, VA, USA.
Loring,P.H., R.A.Ronconi, L.J.Welch, P.D.Taylor, and M.L.Mallory
(2017). Post-breeding dispersal and staging of Common and
Arctic terns throughout the western North Atlantic. Avian
Conservation and Ecology 12(2):20. https://doi.org/10.5751/
ACE-01086-120220
Maloney,K.O., M.Schmid, and D.E.Weller (2012). Applying addi-
tive modelling and gradient boosting to assess the eects of
watershed and reach characteristics on riverine assemblages.
Methods in Ecology and Evolution 3:116–128.
Masden, E. A., and A. S. C. P. Cook (2016). Avian collision risk
models for wind energy impact assessments. Environmental
Impact Assessment Review 56:43–49.
May,R., A. B. Gill, J. Köppel, R. H. W. Langston, M. Reichenbach,
M.Scheidat, S.Smallwood, C.Voigt, O.Hüppop, and M.Portman
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton XXXX 15
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
(2017). Future research directions to reconcile wind turbine–
wildlife interactions. In Wind Energy and Wildlife Interactions
(J. Köppel, Editor). Springer International Publishing, New
York, NY, USA. pp. 255–276.
McCabe,J. D., B.J.Olsen, B.Osti, and P. O.Koons (2017). The in-
uence of wind selectivity on migratory behavioral decisions.
Behavioral Ecology 29:160–168.
Melvin,S.M., L.H.MacIvor, and C.R.Grin (1992). Predatory ex-
closures: A technique to reduce predation at Piping Plover
nests. Wildlife Society Bulletin 20:143–148.
Motus (2016). Motus Wildlife Tracking System Collaboration
Policy. January 2016. https://motus.org/awp-content/up-
loads/2016/01/MotusCollaborationPolicy.January2016.pdf
National Oceanic and Atmospheric Administration (2017).
National Centers for Environmental Prediction North American
Regional Reanalysis. http://www.esrl.noaa.gov/psd/data/
gridded/data.narr.html
O’Connell,A., C.S. Spiegel, and S.Johnston (2011). Compendium
of avian occurrence information for the continental shelf
waters along the Atlantic coast of the United States, nal re-
port (Database Section - Shorebirds). Prepared by the U.S. Fish
and Wildlife Service, Hadley, MD, for the USGS Patuxent Wildlife
Research Center, Beltsville, MD. U.S. Department of the Interior,
Geological Survey, and Bureau of Ocean Energy Management
Headquarters, OCS Study BOEM 2012–076, Sterling, VA.
O’Reilly,K., and J.C.Wingeld (1995). Spring and autumn migra-
tion in Arctic shorebirds: Same distance, dierent strategies.
American Zoologist 35:222–233.
Randin, C. F., T. Dirnböck, S. Dullinger, N. E. Zimmermann,
M.Zappa, and A.Guisan (2006). Are niche‐based species distri-
bution models transferable in space? Journal of Biogeography
33:1689–1703.
R Core Team (2017). R: Alanguage and environment for statistical
computing v. 3.3.2. R Foundation for Statistical Computing,
Vienna, Austria. http://www.R-project.org/
Richardson,W.J. (1976). Autumn migration over Puerto Rico and
the western Atlantic: Aradar study. Ibis 118:309–332.
Richardson,W.J. (1978). Timing and amount of bird migration in
relation to weather: Areview. Oikos 30:224–272.
Richardson, W. J. (2000). Bird migration and wind turbines:
Migration timing, ight behavior, and collision risk. In
Proceedings of the National Avian–Wind Power Planning
Meeting III, San Diego, CA, USA. pp. 132–140.
Robinson, W. D., M. S. Bowlin, I. Bisson, J. Shamoun-Baranes,
K.Thorup, R.H.Diehl, T.H.Kunz, S.Mabey, and D.W.Winkler
(2010). Integrating concepts and technologies to advance
the study of bird migration. Frontiers in Ecology and the
Environment 8:354–361.
Senner,N.R., M.Stager, M.A.Verhoeven, Z.A.Cheviron, T.Piersma,
and W.Bouten (2018). High-altitude shorebird migration in the
absence of topographical barriers: Avoiding high air temper-
atures and searching for protable winds. Proceedings of the
Royal Society B 285:20180569.
Shamoun-Baranes,J., F.Liechti, and W.M. G.Vansteelant (2017).
Atmospheric conditions create freeways, detours, tailbacks
for migrating birds. Journal of Comparative Physiology A
203:509–529.
Stantial, M. L., and J. B. Cohen (2015). Estimating ight height
and ight speed of breeding Piping Plovers. Journal of Field
Ornithology 86:369–377.
Stantial,M.L., J.B.Cohen, A.J.Darrah, K.E. Iaquinto, P.H.Loring,
and P.W.C.Paton (2018). Radio transmitters did not aect daily
nest and chick survival of Piping Plovers (Charadrius melodus).
The Wilson Journal of Ornithology 130:518–524.
Stantial,M. L., J.B.Cohen, P.H.Loring, and P.W.C.Paton (2019).
Radio transmitters did not aect apparent survival of adult
Piping Plovers. Waterbirds 42:205–209.
Taylor, P. D., T. L. Crewe, S. A. Mackenzie, D. Lepage, Y. Aubry,
Z.Crysler, G.Finney, C.M.Francis, G.G.Guglielmo, D.J.Hamilton,
et al. (2017). The Motus Wildlife Tracking System: A collab-
orative research network to enhance the understanding of
wildlife movement. Avian Conservation and Ecology 12(1):8.
https://doi.org/10.5751/ACE-00953-120108
Tonra, C. M., and M. W. Reudink (2018). Expanding the trad-
itional denition of molt-migration. The Auk: Ornithological
Advances 135:1123–1132.
U.S. Fish and Wildlife Service (1985). Endangered and Threatened
wildlife and plants; Determination of Endangered and
Threatened status for the Piping Plover. Final rule. Federal
Register 50:50726–50734.
Weithman, C. E., D. Gibson, K. M. Walker, S. B. Maddock,
J.D. Fraser, S.M. Karpanty, and D.H.Catlin (2018). Discovery
of an important stopover location for migratory Piping Plovers
(Charadrius melodus) on South Point, Ocracoke Island, North
Carolina, USA. Waterbirds 41:56–62.
Wickham, H. (2011). The split-apply-combine strategy for data
analysis. Journal of Statistical Software 40:1–29.
Wilber,D.H., D.A.Carey, and M.Grin (2018). Flatsh habitat use
near North America’s rst oshore wind farm. Journal of Sea
Research 139:24–32.
Williams,T. C., and J. M. Williams (1990). Open ocean bird mi-
gration. IEEE Proceedings F - Radar and Signal Processing
137:133–137.
Wright,J. R., L.L.Powell, and C.M.Tonra (2018). Automated tel-
emetry reveals staging behavior in a declining migratory pas-
serine. The Auk: Ornithological Advances 135:461–475.
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020

16 Piping Plover migration P. H.Loring, J.D. McLaren, H. F.Goyert, and P.W. C.Paton
The Condor: Ornithological Applications 122:1–16, © 2020 American Ornithological Society
APPENDIX TABLE 2. Metrics of migratory ights across the mid-Atlantic Bight of Piping Plovers from U.S. Atlantic coast breeding
areas in Massachusetts (MA) and Rhode Island (RI) in 2015 to2017.
Aux Sex Loc Start (EST) End (EST) Dist (km) Length (hr) Speed (km hr−1) Alt (m)
6XW F RI 7/4/2015 19:41 7/5/2015 03:31 404 7.8 51 313
4NC M MA 7/15/2015 22:20 7/16/2015 01:20 163 3.0 55 272
2YK MMA 7/13/2015 19:10 7/14/2015 19:50 595 24.7 24 373
A8A F MA 9/2/2016 18:22 9/3/2016 22:12 811 27.8 29 342
KVV FMA 7/8/2016 17:23 7/8/2016 21:53 274 4.5 62 265
CAK F MA 7/19/2016 14:04 7/20/2016 9:34 803 19.5 41 284
AE9 M RI 7/23/2016 21:16 7/24/2016 17:46 693 20.5 34 269
E4V M RI 7/23/2016 21:31 7/25/2016 1:31 635 28.0 23 92
CAK F MA 7/23/2017 15:32 7/24/2017 09:42 581 18.2 33 329
6VH M MA 7/23/2017 18:27 7/24/2017 06:37 359 12.3 33 99
KHM M MA 7/23/2017 17:04 7/25/2017 08:54 808 39.8 20 335
AAA F MA 7/23/2017 16:52 7/24/2017 20:42 719 27.8 26 329
UNN M MA 7/25/2017 15:42 7/26/2017 07:12 786 15.5 51 328
Y8M M MA 7/29/2017 21:41 7/30/2017 21:31 746 23.8 32 334
H3J M RI 7/29/2017 19:09 7/30/2017 03:18 585 8.2 72 281
XAP M RI 7/29/2017 19:14 7/30/2017 06:43 601 11.5 52 339
XMJ F RI 7/29/2017 18:44 7/29/2017 23:14 280 4.5 70 331
Downloaded from https://academic.oup.com/condor/advance-article-abstract/doi/10.1093/condor/duaa028/5860737 by guest on 23 June 2020