ArticlePDF Available

Abstract and Figures

Trans‐Saharan migratory bird species encounter large scale seasonal atmospheric convergence zones, where opposing monsoon and continental air masses meet. These macro‐scale atmospheric conditions determine local weather, influence migratory and foraging behaviour and seasonal bird survival rates. Here, we investigate the flight behaviour of pallid swifts Apus pallidus, a small aerial insectivore, in relation to non‐breeding season atmospheric conditions using state‐of‐the‐art GPS logged data. Our analysis shows two novel diurnal flight patterns which suggest that pallid swift prey on insects concentrated along frontal convergence zones, in particular the continental Inter‐Tropical Convergence Zone (ITCZ) and a coastal sea‐breeze front. Resource use seems not only contingent on the abundance of insects, but also favourable atmospheric conditions. Persistence of swifts in wintering feeding grounds might therefore depend on the prevailing atmospheric conditions and their concentrating effects on insects rather than solely the vegetation state and co‐dependent insect populations. Migration events within, to and from, the non‐breeding season foraging locations might not only be guided by a decline in vegetation as common metric for prey availability, but also by shifting wind directions and their concentrating effects.
Content may be subject to copyright.
www.oikosjournal.org
OIKOS
Oikos
1
© 2022 Nordic Society Oikos. Published by John Wiley & Sons Ltd
Subject Editor and
Editor-in-Chief: Pedro Peres-Neto
Accepted 21 March 2022
doi: 10.1111/oik.08594
1–17
2022: e08594
Trans-Saharan migratory bird species encounter large scale seasonal atmospheric
convergence zones, where opposing monsoon and continental air masses meet. ese
macro-scale atmospheric conditions determine local weather, influence migratory and
foraging behaviour and seasonal bird survival rates. Here, we investigate the flight
behaviour of pallid swifts Apus pallidus, a small aerial insectivore, in relation to non-
breeding season atmospheric conditions using state-of-the-art GPS logged data. Our
analysis shows two novel diurnal flight patterns which suggest that pallid swift prey
on insects concentrated along frontal convergence zones, in particular the continental
Inter-Tropical Convergence Zone (ITCZ) and a coastal sea-breeze front. Resource use
seems not only contingent on the abundance of insects, but also favourable atmospheric
conditions. Persistence of swifts in wintering feeding grounds might therefore depend
on the prevailing atmospheric conditions and their concentrating effects on insects
rather than solely the vegetation state and co-dependent insect populations. Migration
events within, to and from, the non-breeding season foraging locations might not only
be guided by a decline in vegetation as common metric for prey availability, but also
by shifting wind directions and their concentrating effects.
Keywords: aerial insects, atmospheric convergence zone, GPS-logging, migration,
non-breeding period, resource and habitat use
Introduction
Both the Inter-Tropical Convergence Zone (ITCZ, Nicholson 2018) and related West
African sea-breeze fronts (Coulibaly et al. 2019) represent significant shifts in wind
direction and strength during the non-breeding season of Trans-Saharan migratory
birds (Miller 2003, Issa Lélé and Lamb 2010, Abatan et al. 2014, Coulibaly et al.
2019). Not only do these macro-scale atmospheric conditions determine local weather
The aeroecology of atmospheric convergence zones: the case of
pallid swifts
Lyndon Kearsley, Nathan Ranc, Christoph M. Meier, Carlos Miguel Pacheco, Pedro Henriques,
Gonçalo Elias, Martin Poot, Andy Williams, Luís T. Costa, Philippe Helsen and Koen Hufkens
L. Kearsley, Belgian Ornithological Research Association, Temse, Belgium. – LK and K. Hufkens (https://orcid.org/0000-0002-5070-8109) (koen.
hufkens@gmail.com), BlueGreen Labs, Melsele, Belgium. – N. Ranc, Univ. de Toulouse, INRAE, CEFS, Castanet-Tolosan, France. – C. M. Meier (https://
orcid.org/0000-0001-9584-2339), Swiss Ornithological Inst., Sempach, Switzerland. – C. M. Pacheco, Research Center in Biodiversity and Genetic
Resources, Univ. do Porto, Vairão, Portugal. – P. Henriques, 31 Rua do Sol, Queluz, Portugal. – G. Elias, 44 Rua de São Pedro, Castelo de Vide, Portugal.
– M. Poot, Martin Poot Ecology, Culemborg, the Netherlands. – A. Williams, 28 Kanachrine Place, Morefield, Ullapool, UK. – L. T. Costa, MAVA
Foundation, Gland, Switzerland. – P. Helsen, Centre for Research and Conservation, Royal Zoological Society of Antwerp, Antwerp, Belgium.
Research
2
but they also greatly influence migratory behaviour, fitness
components and seasonal bird survival rates (Sanz et al. 2003,
Gordo 2007, Boano et al. 2020). Swifts have recently been
shown to remain aloft most or all of their non-breeding period
(Liechti et al. 2013, Hedenström 2016, Hedenström et al.
2019). is continuous flight puts a strong demand on
energy requirements with flight behaviour likely being opti-
mized to sustain this high energy expenditure (Bäckman
and Alerstam 2001, Henningsson et al. 2009, Hedrick et al.
2018). Research on the common swift Apus apus and other
swift species has shown that they feed in a heterogeneous spa-
tio-temporal landscape and leverage swarming insects selec-
tively across space and time (Cucco et al. 1993, Russell 1999,
Collins 2015, de Margerie et al. 2018). Swifts are therefore
also expected to optimize their flight behaviour – i.e. altitude,
direction and speed – according to varying atmospheric con-
ditions in the non-breeding season (Shamoun-Baranes et al.
2006, 2017, Alerstam 2011, Hedrick et al. 2018, Haest et al.
2019, Nilsson et al. 2019, Santos et al. 2020).
e structure of the atmosphere not only shapes bird
flight but also insect availability and concentrations (Collins
2015, de Margerie et al. 2018). Indeed, radar research
of small insects has shown that they often passively float
into atmospheric convergence zones of high shear or in
their wake (Drake and Farrow 1989, Hobbs and Wolf
1989, Reynolds et al. 2016). Both active and passive insect
migration movements are further atmospherically stratified
in density profiles showing dense insect layers at particular
altitudes due to physiological temperature limits on insect
flight (Drake and Reynolds 2012, Florio et al. 2020). ese
movements are often large and interconnect organisms across
trophic levels (Satterfield et al. 2020). As such, the foraging
success of small insectivores is dependent on both flight
strategies to limit energy expenditure (Dokter et al. 2013,
Collins 2015, Meier et al. 2018) and the flight characteristics
and aerial processes shaping the concentration of their prey
(Drake and Farrow 1989).
In contrast to the entomological literature (Drake and
Farrow 1989, Florio et al. 2020), comparatively little
ornithological research focuses on the relationship between
macro-scale atmospheric conditions and foraging behaviour
of small aerial insectivores, during the non-breeding season.
Previous work has shown that pallid swifts Apus pallidus,
small insectivores, generally follow the ITCZ during intra-
tropical migration between Sahel and West African regions
(Norevik et al. 2019). In addition, accelerometer and
altimeter research has shown seasonally varying patterns in
flight altitude (Dreelin et al. 2018), and flapping (active)
flight (Hedenström et al. 2019). Although these studies are
crucial to the understanding of migratory pathways, wintering
areas and their link to vegetation phenology (Fernández-
Tizón et al. 2020), a clear intersection between avian ecology
and atmospheric sciences (i.e. aeroecology) remains limited.
In particular, few guiding processes have been put forward
which explain diurnal patterns in flapping flight or flight
altitude (Morel and Morel 2008).
Here, we used high resolution spatial and temporal (six-
hourly intervals) GPS logged data during the non-breed-
ing season of five pallid swifts to investigate foraging flight
behaviour across space and time in relation to meso- and
macro-scale atmospheric processes. We explore how weather
conditions influence foraging behaviour and daily changes
in flight altitude, direction and aerial concentration across
seasonally changing landscapes along atmospheric conver-
gence zones. In our analysis we link the state of the atmo-
sphere to observed bird flight patterns, and provide context
for insect movements described in entomology literature.
Following entomology and aeroecology literature (Taylor
1974, Reynolds et al. 2008), but lacking direct observations
of insect movements, we hypothesize that pallid swifts will
aggregate at locations which support their foraging behav-
iour and where insects would aggregate most densely in
the atmosphere.
Methods
GPS logger deployment
GPS loggers were deployed on wild pallid swift breeding in
the ceiling of a sea-side cave in the Serra da Arrábida Natural
Park (38°28N, 8°59W), south of Lisbon, Portugal. GPS
tagging of six adult birds (Supporting information) followed
the ongoing ringing protocol at the colony (Costa and Elias
1998), and was carried out mostly late in the breeding
season to avoid disturbance. A low number of individuals is
common in movement ecology studies of small insectivore
birds (Dreelin et al. 2018, Hedenström et al. 2019), as
recapture rates are typically low and costs of (micro-GPS)
loggers can be prohibitive. GPS tags were attached using a
full body harness with a 1 mm wide flat braided cord. We
used low weight ( 0.95 g) nanoFix micro-GPS loggers with
solar trickle charging (PathTrack Ltd, Otley, West Yorkshire,
UK) to track bird diurnal positions and altitude throughout
a full migration season, at regular six-hour intervals (1, 7, 13
and 19 h Greenwich mean time – GMT).
Five loggers were retrieved during the following breed-
ing season leading to an 83% recapture rate (a high value
in relation to the 2020 recapture rate of ringed birds from
2019 of 58% or the long term value of ~65%, Costa and
Elias 1998). All returning tagged birds were examined and
no abrasion or skin damage found. Recovered raw data was
downloaded and forwarded to PathTrack Ltd. for high preci-
sion processing. e loggers showed a high spatial resolution
with high fix acquisition rates (> 95%, Supporting informa-
tion). Processing can lead to default values (attempted fix, no
reception), negative altitudes or locking values to the closest
250 m interval if no clear altitude solution is found within a
reasonable time (personal communications PathTrack Ltd.).
Hence, we excluded all negative altitude values and those
rounded at 250 m increments and only used longitude and
latitude in these positions.
3
Data
Trajectory segmentation
e non-breeding season analysed in this study was defined as
all logger data excluding breeding season movements around
the colony, bounded by 38°1 to 39°N and 11° and 7°7W.
Bird movements were categorized as migration (corridor
movements) and stationary (area-restricted search/foraging)
using a two-state Hidden Markov Model (McClintock and
Michelot 2018) (HMM), with missing positions filled using
continuous-time correlated random walk data (Johnson et al.
2008) to create a homogeneous six-hour time series. For all
time steps we calculated flight distance, speed and heading
(Kranstauber et al. 2012). Step distances (i.e. Euclidean
distances between subsequent GPS positions) were modelled
using a gamma distribution, while turning angle distributions
followed a von Mises distribution (initial parameters;
gamma distributions: μf = 50 km, σf = 10 km; μm = 80 km,
σm = 10 km; von Mises distributions: μf = 0, κf = 1; μm = 0,
κm = 2, migration and foraging denoted with subscripts m
and f, respectively). State classes were returned using global
encoding with the Viterbi algorithm (Zucchini et al. 2016,
McClintock and Michelot 2018).
Using current understanding of foraging behaviour
(Hedenström et al. 2019, Norevik et al. 2019) we assigned
HMM foraging cluster centroids below 20°N as true
foraging movements. is broadly defined region covers
most centroids (22 locations across all birds) except brief
periods of area-restricted search during spring migrations.
True foraging locations during the non-breeding season are
further subdivided according to the state of the vegetation
and their position relative to the coastline.
We used the enhanced vegetation index (EVI) and its
relative direction of vegetation development (i.e. phenol-
ogy) to divide the dataset in increasing and decreasing
trends following observations of an asynchronous phenol-
ogy by Norevik et al. (2019). For each birds, the decreas-
ing EVI trend across the non-breeding season was set for all
days between the first day on which maximum and last day
on which minimum EVI amplitude is reached. Remaining
periods were marked as having an increasing EVI trend.
Increasing EVI trends are processed separately for autumn
(2019) and spring (2020) positions.
Foraging positions were further divided into ‘coastal’
and ‘continental’ locations, where coastal marine locations
are defined as two GPS positions preceding and following
a location over marine waters (open ocean), to account for
behaviour leading to and from open water. We used (1:10
m) ‘Natural Earth’ coastline data (<www.naturalearthdata.
com/>) to determine land based locations.
Vegetation, land cover and climate data
We quantified seasonal vegetation dynamics using Visible
Infrared Imaging Radiometer Suite EVI data (VIIRS
VNP13A1) from the Suomi National Polar-orbiting
Partnership platform. For each position we downloaded
EVI and pixel reliability data centred on an area of 10 × 10
km and spanning the non-breeding season (20 May 2019
to 16 October 2020) (Tuck et al. 2014). We used a maxi-
mum value composite for all pixels with a marginal or bet-
ter pixel reliability (QA bit: 3). Data were smoothed and
temporally interpolated using a locally weighted scatterplot
smoothing fit (or LOESS, span = 0.3). Data were normalized
to percentage amplitude across a whole season on a position
basis. Interpolated EVI values were extracted at the date of
GPS acquisition. To assess the dominant land cover at the
bird positions we used GLOBCOVER 2009 land cover data
(Arino et al. 2012). We used the majority class of an area of
10 × 10 km centered on each position.
For the full extent of the study area we downloaded ERA5
reanalysis data on wind (u, v components) and temperature
at 13 pressure levels (1000–650 hpa, or ~0–4000 m a.s.l.)
as well as single level surface (2 m) temperature, dew point
temperature, cloud base height and convective available
potential energy (CAPE, in J kg1) data for further analysis
(Hufkens et al. 2019, Hersbach et al. 2020). Higher CAPE
values are indicative of (heavy) rain or thunderstorms, with
lighting activity proportional to CAPE multiplied by the pre-
cipitation rate (Romps et al. 2014). In addition, we calculate
cloud base height (CBH; Bradbury 2000), as this provides a
visual cue to the state of the atmosphere (Dokter et al. 2013).
Following Bradbury (2000) the CBH is calculated as:
CBH
=-
()
´
TT
d121 92. (1)
with, T and Td the 2 m temperature and dewpoint tempera-
ture, respectively. ERA5 u and v wind components were con-
verted into wind heading and speed (note, that all headings
are defined with 0/360° as north). We extracted temperature
and wind parameters for both the surface, and on the pres-
sure level corresponding to flight altitude, and this for all GPS
positions. In absence of a working knowledge of high resolu-
tion (< 6 h) swift foraging patterns in relation to atmospheric
convergence zones we calculated average wind assistance i.e.
the combined effect of bird and wind headings and speeds,
using instantaneous wind and average bird speed and heading
data at pressure level (Kemp et al. 2012). In this context wind
assistance values are only a reflection of departure conditions,
not the full trajectory.
Wind and temperature profiles
Frontal convergence zones separate two air masses with
different characteristics, such as wind direction, humid-
ity and temperature (Drake and Farrow 1989, Miller 2003,
Nicholson 2009, p. 200) (Fig. 3a–b). Western Africa is char-
acterized by two (seasonally varying) convergence zones,
1) the Inter-Tropical Convergence Zone and 2) a persistent
sea-breeze (coastal) convergence (Issa Lélé and Lamb 2010,
Bajamgnigni Gbambie and Steyn 2013).
To visually capture the three-dimensional character
of both convergence zones (i.e. continental and coastal
4
locations) we used vertical atmospheric transects through all
pressure levels along the position of each GPS acquisition
and approximately perpendicular to the convergence zone
itself. Atmospheric profiles for continental locations were
characterized by sampling the atmosphere on all level along
~800 km (8°) latitudinal transects (north–south), contain-
ing the bird’s position and centered on the midnight (1 h)
ITCZ position. We estimated the position of the ITCZ
using a Gaussian smoothed (SD = 1) 15°C surface dewpoint
temperature isodrosotherm (Issa Lélé and Lamb 2010).
Centering the transects on a fixed daily location allows us to
account for the seasonal movement of the ITCZ, while pre-
serving diurnal movements. Bird positions and movements
of the convergence zone structure are therefore relative to
the daily midnight position of the ITCZ. Sea-breeze conver-
gence zones for coastal locations were characterized by ~600
km transects from north-east to south-west (45° angle), or
approximately perpendicular to the direction of the West
African coastline for most coastal bird positions. We centered
the transects on the closest location to the coastline (1:110
m ‘Natural Earth’ land mask) from every bird position. Bird
positions and movements of the sea-breeze convergence
zone are therefore relative to the coastline. We character-
ize the vertical structure of both frontal convergences with
a ‘Shear index’, or the scaled (0–1) difference between the
wind heading at a fixed pressure level and the wind head-
ing at all other pressure levels. e (smallest) angular differ-
ence between wind heading at the 850 hpa level (θ850) i.e. a
stable baseline being located at the top of the atmospheric
boundary layer, and all other pressure levels (θi) was calcu-
lated. Locations with opposing wind headings have a (high)
Shear index of 1.
ShearIndexi
i
=-
qq
850
180
(2)
Shear index values were averaged by distance (along the tran-
sect), pressure level and across EVI trends and foraging loca-
tions, resulting in visual representation of an average vertical
section of the atmosphere. For a formal statistical analysis of
Shear Index data, relative to bird positions, we refer to the
diurnal habitat selection section below.
Shear occurrence (%) metrics quantify how common
frontal shear conditions vary across a given region and loca-
tion. We summarized the temporal changes in instances
with a high Shear index (> 0.9) at 10 m above the surface
by sampling along the coast of West Africa (coastline from
3 to 15°N) and a transect in the Sahel at a mean latitude of
~13°N. We spatially summarize data by reporting the mean
shear occurrences for all regions and locations, complement-
ing the temporal analysis.
To assess thermal dynamics within the atmosphere we cre-
ated vertical median temperature profiles at each GPS posi-
tion across all reanalysis pressure levels. We determined the
presence, and altitude, of atmospheric inversions. Inversions
are a measure of atmospheric stability, and a known factor
contributing to insect concentration (Drake and Farrow
1988), and defined as a non-linearity in the temperature lapse
rate with altitude i.e. when temperatures at any pressure level
(higher than surface level) exceeded the temperature at sur-
face level. We report values by EVI trends and locations. We
quantified the spatial inversion occurrence (%) at midnight
across West Africa, by EVI trend, contrasting surface level
temperatures with all atmospheric levels above (900 hpa,
~900 m). Locations with a topography above the 900 hpa
level were excluded from the analysis.
Statistical analysis
Our statistical analysis aimed to describe the differences
in the placement of birds along atmospheric convergence
zones on both seasonal and diurnal time scales. We hereby
try to link the state of the atmosphere to observed flight
patterns and provide context in relation to known insect
movements described in entomology literature. We therefore
test for significant changes in (diurnal) altitude distribution,
atmospheric indices (CAPE and CBH), flight heading and
turn angles as well as the relative or absolute position with
respect to convergence zones. All analysis executed across
(seasonal) EVI trends, and foraging locations (coastal or
continental).
To compare the movement of birds relative to the ITCZ,
during the largest latitudinal changes i.e. the decreasing EVI
trend (below), we compared the latitude of the ITCZ closest
to bird positions and the latitude of the birds themselves in a
regression analysis.
We used a Rayleigh z statistic to test for significant diurnal
angular concentration in both the flight heading and turn
angles. e Rayleigh z statistic tests if birds consistently fly
the same overall pattern along similar flight headings (inde-
pendent of the next heading) or specific diurnal patterns with
similar turn angles (i.e. between consecutive headings). To
assess if flight headings align with the dominant wind head-
ing (at level), independent of relative speeds, a circular–cir-
cular regression between flight and wind headings was used.
To take into account both direction and speed we used a two-
sided one-sample t-test on wind assistance values to deter-
mine significant differences from a zero mean (i.e. no wind
assistance during flight). Simple descriptive tests on differ-
ences in mean values (t-tests) in environmental variables were
used. In case of unequal sample sizes or variance a Welch two
sample t-tests was used. Diurnal changes in flight altitude
for small insectivores are well documented (Dreelin et al.
2018, Meier et al. 2018), but might vary throughout the
non-breeding season. We compared flight altitudes between
foraging locations and diurnal positions, for both EVI trends,
using a two-way ANOVA (log(altitude) ~ foraging location ×
hour of day, Supporting information).
We report general summary statistics, by EVI trend, loca-
tion and acquisition time, including the mean and stan-
dard deviation on flight altitude, flight and wind heading
5
and mean inversion altitude, convective available potential
energy, inversion occurrence (%), dominant land cover type
(%) and cloud base height.
Habitat selection
To evaluate diurnal and multi-day habitat selection across
a varying aerial landscape, we conducted individual-level
integrated step selection analyses (Avgar et al. 2016) (iSSA)
using geographic layers of wind shear as covariate in interac-
tion with time of day (i.e. day or night). e iSSA allows
us to make a joint inference about movement behavior and
resource selection by contrasting the environmental condi-
tions of used locations (i.e. the acquired GPS positions) with
a set of (random) control conditions. It is a well known issue
that if the sampling interval is high with respect to the hetero-
geneity of a covariate, used GPS positions and control points
are essentially identical (Nisi et al. 2022). erefore an iSSA
step scale should match the process at hand (urell et al.
2014). We used two iSSA aimed at testing for diurnal or
multi-day habitat selection (i.e. two separate temporal grains).
e daily analysis was executed for all foraging movements
across all EVI trends and locations, and a step frequency of
six hours. In addition, we used a 48 h step frequency between
midday (13 h GMT) positions in order to separate diurnal
from multi-day responses to larger scale (seasonal or weather
related) ITCZ movements.
For both analysis, for every time step (6 h or 48 h), surface
level (10 m) Shear Index values were calculated (above) and
day or night states determined based upon the geographic
position and acquisition time. For each GPS relocation, we
generated twenty control locations by randomly sampling
from the empirical step length distribution of each individual
swift, and using a circular uniform distribution to generate
random directions for each available location (urell et al.
2014). Data are fit using a conditional logistic regression
model. We report model statistics and the relative selection
strength (Avgar et al. 2017) between times of day.
All data preparation and analyses were executed in R
ver. 4.0 (<www.r-project.org>) using appropriate packages
(Agostinelli and Lund 2017, McClintock and Michelot
2018, Pebesma 2018, Hijmans 2019, Signer et al. 2019). For
more details we refer to the Supporting information.
Results
Logging and movement classification
e GPS loggers proved reliable with a high fix acquisition
rate (i.e. the ratio of measured GPS locations over the sched-
uled ones) of between 96 and 100% (see Supporting infor-
mation for details). We logged 6360 GPS positions of which
76% (4835 positions) during the non-breeding season. e
Hidden Markov Model provided robust estimates of posi-
tions characterized by either long migration movement steps
with low directional variability and short foraging steps with
higher directional variability (Fig. 1a). Model parameters
showed a large average (μm) step length of 196 km (σm = 57
km) and an angular distribution with a highly concentrated
mean of 0.09 (κm = 2.76) for migration movements. Foraging
was characterized by smaller steps with an average step length
(μf) of 95 km (σf = 40 km) and an angular distribution with
a mean (μf) 1.19 (κm = 0.19), or roughly corresponding to
~90° turn angle.
Figure 1. Overview of the migration and foraging movements, and their clustering into regional and location based positions. (a) seasonal
non-breeding movement patterns for two individual pallid swifts Apus pallidus, AP-1 and AP-4, annotated according to migration or more
stationary foraging movements. Migration (grey) and foraging states (orange) as identified by the hidden markov model (HMM). Flight
directions are marked with . (b) Bird positions of all individuals relative to the Inter-Tropical Convergence Zone (ITCZ), plotted by
month, EVI trend and location (i.e. continental or coastal). Locations () are coloured depending on their EVI trend and location.
Monthly mean location of the Inter-Tropical Convergence Zone (ITCZ) are shown as a dark grey line (defined as a 15°C dewpoint
temperature threshold). To visualize the varying extent of the ITCZ over the integration period both the 10th and 90th percentile on the
six hourly ITCZ latitudes is shown as dashed dark grey line.
6
Seasonal flight patterns
Migration to and from the winter-feeding grounds happened
across a time frame of 2.6 ± 0.55 and 5.8 ± 2.3 days and
covered on average 2104 ± 377 km or 4290 ± 1115 km,
south (2019) or north (2020) respectively. During the non-
breeding season all birds foraged at several intra-tropical loca-
tions separated by large seasonal or short localized migration
movements (Fig. 1). e non-breeding season ended after an
average 235 ± 47 days with birds covering 59 520 ± 10 455
km in flight (logged in 6 h increments). Seasonal flights were
divided according to vegetation phenology (i.e. enhanced
vegetation index, or EVI, amplitude) at bird positions. We
noted a short increasing EVI trend in autumn (2019, 59 ±
52 days), a main long decreasing EVI trend in the winter of
2019–2020 (154 ± 11 days), and finally an increasing EVI
trend in the spring of 2020 (125 ± 39 days, Fig. 1a).
e swifts arrived to their first foraging ground late
August at the northern edge of the Sahel and the Aoukar
plateau and cliffs (south east Mauritania), crossing and mov-
ing well south of the ITCZ (403 ± 355 km south of the
ITCZ). Crossing the ITCZ translated in a significant rise
in convective available potential energy (CAPE) and fall
in cloud base height (CBH), when comparing preceding
migration days to the arrival foraging ground (Welch t-test,
t = 36.337, df = 290.79, p < 0.0001; t = 36.337, df = 290.79,
p-value < 0.0001; t = 5.3874, df = 166.58, p < 0.0001, for
CAPE, and CBH, respectively). e Aoukar plateau, at this
time, is characterized by precipitation driven greening of
sparse vegetation (i.e. increasing EVI trend, Fig. 2a–b), due
to an unstable atmosphere at the southern side of the Inter-
Tropical Convergence Zone (ITCZ, Fig. 2b, Supporting
information). e swifts tracked the edge of the plateau for
almost a week (6 ± 2 days) before slowly moving east to
their main foraging locations across Mali or as far as Niger
(Fig. 2b) across the predominantly arid landscape. During
autumn, and an increasing EVI trend, CAPE values as high
as 883 ± 839 J kg1 (with values > 1000 J kg1 indicative
of thunderstorms), and low morning cloud base height val-
ues (595 ± 852 m a.s.l.) indicate a high likelihood of rain
storms. Birds foraged consistently at the southern side of the
ITCZ (96% of all positions) on locations dominated by bare
areas (74%) and closed to open grasslands (20%, Supporting
information). e swifts showed no significant preference for
Figure 2. Summary plots of seasonal vegetation dynamics (as scaled enhanced vegetation index or EVI) at GPS positions of pallid swifts
Apus pallidus as well as the shear occurrence at two locations. Seasonal amplitude in vegetation greenness (i.e. enhanced vegetation index)
at GPS positions during non-breeding season movements are plotted as or for coastal and continental positions, respectively. A grey
rectangle denotes the region of a decreasing EVI trend. We report mean shear occurrence, a measure of daily shear index intensity, for all
locations along the West African coast line (between 3 and 15°N at 19 h, full black line), and along the mean latitude of all Sahel locations
(16°N at 7 h, dashed black line). We refer to Fig. 6 for a visual representation. Arrival and departure times of individuals are plotted as
dashed or full grey lines respectively.
7
locations with or without wind shear as indicated by non-
significant step selection coefficients (integrated step selec-
tion analysis, df = 2, p > 0.05, Supporting information).
Eastward migrations cover foraging grounds with increasing
EVI values (Fig. 2), waiting for the southward movement
of the ITCZ (from 20°N to ~15°N during September).
Foraging movements early in the non-breeding season are
therefore not linked to large latitudinal movements (Fig. 1b,
Supporting information).
e southward movement of the ITCZ, relative to the
bird positions, placed the birds at the dry northern side of
the ITCZ which results in improving atmospheric stability
and the start of a long decreasing trend in EVI amplitude
(Fig. 2a). Improving atmospheric conditions i.e. fair weather,
were indicated by a drop in shear occurrences (from ~55%
to ~25%, Fig. 2a), decreasing CAPE values to a low of 75
± 255 J kg1 (Table 1) and an increasing cloud base height
well above registered flight altitudes (1437 ± 884 to 3089 ±
696 m a.s.l. for continental locations, Table 1). Subsequently
the swifts followed the progressive southward movement of
the ITCZ, and the demise of the West African monsoon
(Zhang and Cook 2014), predominantly on its northern dry
side (~68% of positions), while EVI values dropped to ~50%
of their seasonal amplitude (Fig. 2a). Moving south, closer
to the ocean, birds showed frequent small migration move-
ments (9 ± 9 movements for our birds), switching between
coastal marine and continental locations. We note that 23%
of the positions during the decreasing EVI trend were over
open ocean. Coastal locations, in general, were characterized
by higher CAPE values (up to 1032 ± 851 J kg1) and a
lower cloud base height (as low as 304 ± 329 m a.s.l., Table
1), suggesting a more perturbed atmosphere at bird posi-
tions. Despite the varying atmospheric conditions 79% of
birds across all locations were located below the cloud base
height during the decreasing EVI trend, and a similar value
of 76% during the whole non-breeding season. Birds across
continental decreasing EVI trend locations consistently
tracked the ITCZ latitude with 86% of variability explained
(F(1,795) = 5334, p < 0.0001; Fig. 1, 2a), on average keeping
to within 105 ± 107 km from the front. Our multi-day inte-
grated step selection analysis (iSSA) showed that, during the
decreasing EVI trend, the swifts had a strong preference for
locations without wind shear as indicated by significant nega-
tive step selection coefficients (iSSA, df = 2, p < 0.05, Fig. 4c,
Supporting information), an indication that birds actively
follow the ITCZ seasonal and weather related patterns. In
contrast, across increasing EVI trends we did not find signifi-
cant selection for locations without wind shear (Supporting
information). At continental locations birds foraged over bare
areas (32%), closed to open grassland (23%) and vegetation
and cropland mosaic (10%) positions (Supporting informa-
tion). Coastal locations (above land) showed frequent bird
positions above a mix of natural vegetation and croplands
(39%), deciduous evergreen forest (26%) and mangrove for-
ests (11%, i.e. GLOBCOVER class: closed broadleaved semi-
deciduous or evergreen forest regularly flooded – saline water,
Supporting information).
During spring birds foraged further inland and found
themselves again in positions with a greening vegetation (i.e.
increasing EVI values), due to the reversal and steady north-
ward progression of the ITCZ and following precipitation.
Positions are mostly located to the south of the ITCZ (85%)
on an average distance of 224 ± 132 km to the front. Similar
to the autumn increasing EVI trend, the swifts showed no
significant selection for wind shear (iSSA, df = 2, p > 0.05,
Supporting information). Bird positions for continental loca-
tions were dominated by open broadleaved forest (29%),
closed to open shrubland (22%) and a vegetation mosaic in
cropland (18%), rather than bare areas or grasslands dur-
ing the preceding decreasing EVI trend. Coastal locations
retain a similar distribution in land cover across decreasing
and spring increasing EVI trends, with a large share of posi-
tions over mangrove forest (60%) and evergreen deciduous
forests (20%). At the end of the non-breeding season migra-
tion north in spring 2020 was spread across 43 days staged
during the increasing EVI trend, with an average departure
date of 16 April.
Diurnal flight patterns
Seasonal changes during the main foraging period (i.e.
decreasing EVI trend) were characterized by two distinct
diurnal flight modes defined by the proximity of birds to
atmospheric convergence zones, either as the continental
ITCZ or a coastal sea-breeze front (Fig. 3a–b). Our statistical
analysis showed that pallid swifts display consistent diurnal
flight characteristics in relation to these frontal convergence
zones, flight altitude, wind assistance or positions relative to
temperature gradients in the atmosphere. Different flight pat-
terns were shown across increasing EVI trends which further
support the notion of a strong influence of atmospheric sta-
bility, at or in the wake of the ITCZ, shapes flight behaviour.
Continental Inter-Tropical Convergence Zone
Across the decreasing EVI trend continental flights track the
north side of the ITCZ (at 124 ± 113 km, ~68% of posi-
tions) while avoiding unstable atmospheric conditions at the
monsoon side. Flight directions showed significant clustering
except for the evening (19 h) positions (Rayleigh z test, p <
0.0001, Supporting information). Flight heading and wind
heading were not significantly related, with the exception of
morning values (7 h, circular-circular regression, p = 0.027,
Table 1).
Using wind assistance as a metric integrating both flight
direction and speed, in relation to wind speed and direction,
we found varying degrees (both in sign and magnitude) of
significant wind assistance. Values were significantly negative
for midnight and morning (1 h, 7 h) values, while reach-
ing a significantly positive value around midday (13 h, ~2
m s1, Table 1). Our diurnal iSSA showed no strong pref-
erence for locations with or without shear, as indicated by
low and non-significant step selection coefficients, and this
across all continental locations (Fig. 4b, Supporting infor-
mation). roughout the day birds made significant and
8
Table 1. Overview table of summary (circular) statistics on flight heading, wind heading, wind assistance and turn angles of pallid swifts during foraging movements by EVI trend
and location (i.e. position over either coastal water or continental locations). We report mean and standard deviation for wind heading, flight heading, wind assistance, turn angles,
inversion altitudes (and occurrence, %), convective available potential energy and cloud base height. For mean flight heading and wind heading we provide significance levels of a
circular–circular regression between both. We report significance levels for Rayleigh z-tests of angular concentration for turn angles. In addition the number of GPS locations used
(n, between brackets if uncertain altitude GPS fixes were removed). For wind assistance measurements we provided the significance levels (if any) of a two-sided Welsh t-test on a
mean of 0. Flight and wind headings are on a scale of 0–360 degrees with 0/360° north, while turn angles cover a range from 180 to 180 degrees, with negative values denoting
a left turn. Significant values are denoted with * at the 0.05 level (full details see the Supporting information).
EVI trend Location hour n
Circular – circular regression
Turn angle
(mean ± SD, °)
Wind assistance
(mean ± SD, ms1)
Flight altitude
(mean ± SD,
m a.s.l.)
Inversion
altitude
(mean ± SD,
m a.s.l.)
Inversions
occurrence (%)
Convective
available potential
energy (CAPE)
(mean ± SD,
J kg1)
Cloud base height
(CBH) (mean ± SD,
m a.s.l.)
Flight heading
(mean ± SD, °)
Wind heading
(at altitude)
(mean ± SD, °)
Decreasing Continental 1 403 (326) 59.07 ± 1.47 264.88 ± 1.34 14.38 ± 2.22 0.61 ± 0.61* 1032 ± 489 439 ± 120 90 173 ± 509 1815 ± 884
7 411 (272) 26.35 ± 2* 281.6 ± 1.44* 20.65 ± 2.38 0.56 ± 0.56* 1041 ± 576 535 ± 140 85 228 ± 498 1437 ± 859
13 388 (274) 234.49 ± 1.17 264.32 ± 0.99 16.99 ± 2.07 1.94 ± 1.94* 798 ± 429 / 0 75 ± 255 3098 ± 696
19 381 (285) 59.46 ± 2.67 247.29 ± 0.91 13.82 ± 2.24 0.3 ± 0.3 1026 ± 351 384 ± 115 58 60 ± 291 2788 ± 825
Coastal 1 311 (185) 94.18 ± 1.37 87.64 ± 1.26 70.97 ± 1.42* 1.05 ± 1.05* 286 ± 313 258 ± 89 34 1087 ± 881 375 ± 205
7 320 (215) 32.34 ± 0.72 141.11 ± 1.67 143.75 ± 1.29* 0.8 ± 0.8* 494 ± 469 358 ± 122 22 913 ± 757 304 ± 329
13 345 (262) 247.66 ± 1.17 296 ± 1.72 53.9 ± 1.59* 0.29 ± 0.29* 368 ± 280 / 0 402 ± 734 1627 ± 841
19 330 (199) 197.02 ± 1.24 220.56 ± 1.6 96.29 ± 1.75* 1.77 ± 1.77* 873 ± 626 251 ± 92 34 824 ± 881 560 ± 726
Increasing Continental
(autumn)
1 84 (81) 94.54 ± 1.25 343.77 ± 1.32 35.82 ± 2.25 0 ± 0 662 ± 387 465 ± 100 70 883 ± 839 989 ± 967
7 85 (70) 155.88 ± 2.45 10.35 ± 1.44 6.56 ± 1.73 0.36 ± 0.36 637 ± 532 538 ± 146 50 846 ± 851 595 ± 852
13 87 (73) 241.94 ± 1.92 322.71 ± 1.79 130.68 ± 2.45 0.59 ± 0.59 652 ± 383 / 0 659 ± 690 2167 ± 1169
19 90 (83) 98.6 ± 1.94 305.63 ± 1.43 24.83 ± 1.76 0.24 ± 0.24 650 ± 402 430 ± 83 18 619 ± 750 a1871 ± 1220
Continental
(spring)
1 186 (118) 125.46 ± 1.8 17.11 ± 1.8 3.26 ± 2.13 0.28 ± 0.28 1140 ± 569 531 ± 185 43 837 ± 742 483 ± 626
7 189 (122) 110.77 ± 2.07 38.68 ± 1.42 74.26 ± 2.15 0.96 ± 0.96* 832 ± 455 642 ± 194 31 780 ± 676 293 ± 656
13 180 (134) 330.46 ± 1.69 17.1 ± 1.82 6.78 ± 1.57* 0.59 ± 0.59* 651 ± 344 506 ± 230 4 637 ± 727 1517 ± 942
19 177 (89) 260.64 ± 2.24* 257.97 ± 1.71* 39.46 ± 1.95 0.68 ± 0.68* 1158 ± 515 399 ± 154 38 643 ± 804 1326 ± 862
Coastal
(spring)
1 98 (76) 112.39 ± 1.37* 71.57 ± 0.66* 108.09 ± 1.48* 1.53 ± 1.53* 225 ± 325 413 ± 132 66 1052 ± 1126 368 ± 158
7 98 (78) 15.72 ± 0.8 101.44 ± 0.95 59.49 ± 1.6* 0.03 ± 0.03 395 ± 353 442 ± 121 51 822 ± 921 343 ± 228
13 110 (91) 44.36 ± 1.71 48.15 ± 1.44 173.21 ± 1.86 0.21 ± 0.21 270 ± 262 / 0 405 ± 771 1631 ± 703
19 101 (81) 213.81 ± 0.91 60.02 ± 1.07 103.46 ± 1.38* 1.21 ± 1.21* 322 ± 554 213 6 602 ± 908 1155 ± 620
9
Figure 3. A combined plot of (a) a schematic representation of bird locations in relation to diurnal variation in the position of the Inter-
Tropical Convergence Zone (ITCZ) or the structure of a sea-breeze frontal convergence, and (b) their observed counterparts. We highlight
(a) movements of birds and insects within the schematic representation, (b) as well as the shear index profiles using 13 levels of ERA5
reanalysis data along transects visualizing the structure of the ITCZ and sea-breeze front in relation to bird positions. (a–b) Bold black lines
show the extent of the ITCZ and sea-breeze frontal convergence at their minimal (dashed line) and furthest (full line) extent (with timing
noted below these positions). Prevailing winds of both fronts are are indicated and named by light grey arrows, while swift movements are
shown with dark grey arrows (timing is inset). (b) e structure of a sea-breeze front driven by a large thermal gradient between the cold
ocean and warm land areas. e frontal convergence is shown as a bold black line. e main wind flow balancing the gradient is shown as
large grey arrows. Insects are transported into and aggregate in the wake of the sea breeze front, illustrated by the dashed grey lines. (adapted
from Nicholson 2009 and Drake and Farrow 1989). A more elaborate representation of the illustrations in figures (a) and (b) is provided
in the Supporting information. (c–d) Shear index values along transects centered on the 1h location of the ITCZ (c) or the coastline (d),
shown as a dashed black line, with white representing the position of birds along/within the transect. Dashed white lines represent the
mean location of the ITCZ as determined by a 15°C dewpoint temperature threshold. A black bold line outlines values with a shear index
of 0.5 or more, and approximates the frontal convergence.
10
consistent changes to flight altitudes, with significant interac-
tions according to acquisition times (ANOVA, F(3,2087) = 35.7,
p < 0.0001). Flights across the continental decreasing EVI
positions showed a pattern of consistent high flight altitudes
(1026 ± 351 to 1032 ± 489 m a.s.l.), only interrupted by
small but significant descending movements toward midday
(798 ± 576 m a.s.l., Fig. 5, Table 1, ANOVA, F(3,2087) = 35.7,
Tukey HSD post hoc test, p < 0.01).
From midnight to morning atmospheric layering through
temperature inversion is at its strongest (Nicholson 2018),
when the ITCZ front further develops (Fig. 6c and d). Around
midnight (1 h) 90% of bird positions showed atmospheric
temperature inversion. At these locations birds typically fly
above inversion altitudes (92% of the time). Furthermore,
continental CAPE values, as a metric for atmospheric sta-
bility, show low (stable) values of 75 ± 255 J kg1 around
midday with a maximum of only 228 ± 498 J kg1 during
morning hours, while cloud base heights far exceed the flight
altitudes (Table 1). Bird movements therefore follow a chang-
ing atmospheric inversion on the dry side of the ITCZ, show-
ing a preference for stable conditions with high cloud base
heights. e inverse pattern is shown for coastal positions.
Continental positions during increasing EVI trends in
autumn (2019) and spring (2020) show different flight
modes compared to the decreasing EVI trend. Birds made
consistent changes in flight altitudes, with significant
interactions according to the season and acquisition times
(ANOVA, F(3,943) = 10.51, p < 0.0001). In particular, posi-
tions during the increasing EVI trend in autumn showed
consistently lower flight altitudes (~650 m a.s.l., Table 1).
Positions during the increasing EVI trend in spring showed
high flight altitudes during the evening through midnight,
with significant decreases in flight altitude during morn-
ing and noon hours (7 h and 13 h, Tukey HSD post hoc
test, p < 0.01). In both cases inversion occurrences drop
to 70% and 43% for autumn and spring increasing EVI
trends, respectively. roughout, flight altitude values are
inversely related to the CAPE values, where low CAPE val-
ues generally correspond to higher flight altitudes (Table
1, Supporting information). During increasing EVI trends
both seasons are characterized by, at times, a low cloud base
height (Table 1, Supporting information). We found that
80% of flights in autumn and 55% of flights in spring are
below the cloud base height.
Figure 4. Integrated step selection analysis (iSSA) results with 95% confidence intervals for the relative selection strength across daytime
() and nighttime () for coastal and continental positions of the decreasing EVI trend (a–b). In addition we report relative selection
strength for multi-day movements using midday (13 h) positions and a 48 h step, for all continental positions along the decreasing EVI
trend (c). Positive values indicate selection for high wind shear conditions, negative values indicate selection for low wind shear conditions.
Coefficients with significant selection strength (p < 0.05) are marked with * (details in the Supporting information).
11
Coastal sea-breeze front
Birds move toward the coast during the decreasing EVI trend
while the coastline is characterized by strong seasonal sea-
breeze fronts (Issa Lélé and Lamb 2010, Abatan et al. 2014,
Coulibaly et al. 2019). ese high wind shear conditions
across ~75% of the coastline (3–15°N) last until mid-Febru-
ary at the end of the decreasing EVI trend (Fig. 2a, 6). Pallid
swift along the coastline of West Africa, with the ITCZ in
its most southern position, consistently selected for differ-
ent atmospheric conditions during day or night as indicated
by our integrated step selection analysis (df = 2, p < 0.05,
Fig. 4a, Supporting information). Coastal movements during
the spring increasing EVI trend show few data points and
limit strong inference (Supporting information). Overall,
birds are most likely found in locations of high atmospheric
wind shear during nighttime. During daytime swifts actively
avoid high wind shear locations (Fig. 4a).
Coastal bird positions during the decreasing EVI trend
were predominantly located on the southern monsoon side
of the ITCZ (89% of all positions, Fig. 1b). Swifts positioned
themselves above coastal marine waters during midnight
(45%, 1 h) or morning (34%, 7 h). Fewer evening (19%, 19
Figure 5. Diurnal flight altitude (m a.s.l.), convective available potential energy (CAPE) and cloud base height (CBH) changes over time,
across continental and coastal locations. Birds at different locations (continental or coastal) made significant and consistent changes to flight
altitudes, with significant interactions according to acquisition times (ANOVA, F(3,2087) = 35.7, p < 0.0001). Continental flight altitudes
were consistently, interrupted by small but significant descending movements toward midday (13 h, 798 ± 429 m a.s.l., Tukey HSD post
hoc test, p < 0.05). Coastal flights were lower with significant twilight ascents (7 h and 19 h, Tukey HSD post hoc test, p < 0.05).
12
h) and almost no midday (2%, 13 h) positions above coastal
waters were measured. Birds traveled over open water up to
87 ± 25 km from the coastline (on average 18 ± 16 km).
Notably, over open water birds retained good visibility of
land (95% of the time), with an average horizon of 50 ± 40
km inland given the absence of dense cloud cover. Coastal
flights showed marked differences in altitude and directional-
ity when compared to continental flights. Flights were lower
(see ANOVA above, Fig. 5, Table 1), with significant twilight
ascents (Fig. 5, at 7 h and 19 h, Tukey HSD, p < 0.0001).
Comparing flight and wind headings using a circular–circu-
lar regression we found no significant relationship. However,
wind assistance accounting for both speed and direction
showed significant assistance during all times of the day (one
sample t-tests, p < 0.0001, Table 1). Morning flights had
low but significant headwinds (negative wind assistance, one
sample t-tests, p < 0.0001, Table 1, Fig. 3d) while aggregat-
ing at the leading edge of the sea-breeze front (Fig. 3d).
Unlike continental flights, movements were highly con-
centrated in directionality resulting in a strong recurring
diurnal pattern, with birds consistently turning ~90° left
throughout the day and consistently circling the coastline
(Rayleigh z tests, p < 0.0001, Table 1, Supporting informa-
tion). Transects of the atmosphere at the coastline showed
a well-defined sea-breeze front and explained the observed
reversal in wind direction with changing bird altitudes
(Fig. 3d). Birds maintained wind assistance through altitude
changes despite the reversal of flight direction, from offshore
to onshore, between the evening (19 h GMT, 220.56 ± 1.6
degrees) and midnight (1 h GMT, 87.64 ± 1.26 degrees)
positions (Fig. 3d, Table 1), corroborating our integrated step
selection analysis.
During the spring increasing EVI trend we see individu-
als remain in West Africa until late March, well after the
local minimum EVI amplitude is reached (mid February).
At the same time coastal shear conditions deteriorate as
the ITCZ quickly moves back north (Fig. 1b, 6). Coastal
movements decrease, with most birds moving inland before
migrating north.
Discussion
In our study, we showed that pallid swifts followed strong
diurnal patterns consistent with multiple flight modes along
the edge, or in the wake of, frontal aerial convergence zones
in particular the continental Inter-Tropical Convergence
Zone (ITCZ) and a coastal sea-breeze front. ese flight
modes showed differences in flight altitude, direction and/
or temperature lapse rates traversed. Pallid swifts actively
used the coastal sea-breeze front, a common feature along the
West African coast (Bajamgnigni Gbambie and Steyn 2013,
Abatan et al. 2014, Coulibaly et al. 2019), to remain con-
centrated along the coastline using changing wind directions
to aid their flight. Similarly, continental bird movements
followed the (multi-day) ITCZ movements on its northern
stable side, as well as its strong diurnal temperature inversion
(Nicholson 2018).
Movements described in these two flight modes are con-
sistent with patterns of insect aerial concentration previously
described in literature (Taylor 1974, Drake and Farrow 1989,
Reynolds et al. 2008, Chilson et al. 2017) or, conservatively,
behaviour to optimize flight conditions. Although we did not
measure insect distributions, bird positions correspond to
locations where prey should be most densely packed within
the atmosphere. e flush of vegetative growth, following
the passing of the ITCZ, plays an important role in the resi-
dency throughout the non-breeding season of pallid swifts
(Hedenström et al. 2019, Norevik et al. 2019). Yet, residence
of birds in West Africa take advantage of both a leading or lag-
ging vegetation development (Fig. 2a–b), suggesting that the
state of the vegetation and associated insect populations are
not necessarily limiting. Swifts positioned themselves where
the prevailing atmospheric conditions, and their effects on
Figure 6. Spatial shear occurrence maps summarizing the percentage of acquisition hours (1, 7, 13 or 19 h GMT) which have at least one
Shear Index occurrence of 0.9 or more, split across EVI trends (increasing or decreasing) and seasons. Black contours outline a region where
at least 50% of the hours show a shear index of 0.9 or more. Countries are outlined with white full lines. Black dashed lines mark the tran-
sects used in the calculation of shear index occurrences (at 7 and 19 h, as shown in Fig. 2).
13
optimal flight conditions and potential concentrating effects
on insects, would be most favourable.
Seasonal flight patterns
Pallid swifts tracked the gradual progression of the ITCZ
along a decreasing enhanced vegation index (EVI) trend. Even
though our sample included only five birds, we found similar
patterns in foraging behaviour from GPS logged data pre-
sented by Finlayson et al. (2021) corroborating our findings
(Supporting information). Crossing the ITCZ and its increas-
ing likelihood of rain results in the first indications of increas-
ing vegetation greenness (EVI), and potential food sources,
which seemed to serve as a cue to stop migration. Yet, we can
not dismiss the prominent landscape feature and potential
feeding ground of the Erg Aoukar as a stop-over location. e
structured nature of flights along the West African coastline,
and recurrence of coastal flights even after excursions inland,
suggests that these movements are not incidental. We show
that birds do not follow seasonal EVI maxima, but rather
positions with an increasing EVI along a steady gradient,
corroborating previous results by Norevik et al. (2019) and
Finlayson et al. (2021). Our results expand upon this and
suggest that birds balance the atmospheric state, near con-
vergence zones, and its concentrating or otherwise favourable
effects with food availability. is considerable plasticity in
the behaviour of birds along or in the wake of convergence
zones is shown by multiple consistent, and at time alternat-
ing, flight modes throughout the non-breeding season. An
evaluation of shear occurrence among regions and locations
shows that a breakdown of a strong land-ocean breeze during
the increasing EVI trend is a potential driver of inland conti-
nental West African movements (Fig. 6).
Overall, birds preferred the stable side of the ITCZ and
when possible leveraged strong diurnal (coastal) patterns to
their advantage, as shown by our integrated step selection
analysis. Exceptions exist at the start and end of the non-
breeding season (increasing EVI trends), with most posi-
tions in an unstable atmosphere to the south of the ITCZ. In
autumn swifts arrive to the south of the ITCZ where food is
found, driven by vegetation growth of the ongoing seasonal
rains across the Sahel. Similarly, the northward movement
of the ITCZ in spring requires birds to remain on its rainy
unstable side, as vegetation to the north at this time is equally
limited.
Diurnal flight patterns
Along the leading edge of the sea-breeze front swifts moved
in two (semi-) circular horizontal and vertical flight paths
through the atmosphere. is diurnal pattern positions birds
into favourable wind layers to assist them in moving on- or
off-shore (Fig. 3b–d) but could be interpreted as passive drift
i.e. low energetic flights using apparent wind (Bäckman and
Alerstam 2001, Dokter et al. 2013). Increased gliding flight
during nighttime in the second part of the non-breeding sea-
son as shown by Hedenström et al. (2019) in part supports
this hypothesis. Yet, benefits of co-location with insect prey
cannot be excluded. We did not make collocated insect obser-
vations and few observations of insect flights exist for west-
ern Africa but insects concentrating along sea-breeze fronts
have been described elsewhere (Drake and Farrow 1989,
Sauvageot and Despaux 1996, Russell 1999, Isard et al.
2001, Russell and Wilson 2001). Birds seeking onshore wind
assistance, circling the sea-breeze front, would therefore not
only safeguard an energy efficient return to land, after being
pushed out to sea, but retain a favourable position relative to
concentrated prey.
Our research also puts forward a new perspective on twi-
light ascents as strong guiding processes have been elusive.
Meier et al. (2018) suggests that ascents of Alpine swifts
Apus melba have a social function, where Hedenström et al.
(2019) describes similar observations for pallid swifts but
provides no further hypothesis. High evening flights to over-
top the sea-breeze front, and altitude gain in the morning
with its potential to visually orient birds toward shore pro-
vide plausible guiding mechanisms. Comparison with pub-
lished high frequency pallid swift altitude data during the
non-breeding season shows similar seasonal flight patterns
(Hedenström et al. 2019). Our data therefore gives weight to
the re-orientation theory, although not within the context of
seasonal migration as proposed by Dokter et al. (2013).
It remains unclear how much other swift species benefit
from convergence zones or local wind systems. Other species
might occupy different foraging niches in the air as resource
partitioning strategies across aerial foraging habitats have
previously been described for (neo-) tropical swifts (Collins
2015). e common swift, wintering in the central Congo
Basin, often frequents the same areas as pallid swifts during
migration stop-overs (Åkesson et al. 2020) and therefore does
not preclude intermittent use of frontal systems. Despite dif-
ferent winter habitats all these species rely on insects, which
are sensitive to aerial aggregation. As such, our results do not
conflict with current knowledge of resource partitioning in
aerial foraging habitats.
Continental bird movements during the decreasing EVI
trend followed the ITCZ, but on a diurnal basis were not
bound to the leading edge of the front, as shown for the sea-
breeze front. We hypothesize that large diurnal changes in the
location of the ITCZ (~179 ± 112 km day1) make it ener-
getically less favourable. Remaining close to the leading edge
of the front would require a ~400 km day1 return voyage, or
half of a daily distance flown used for transit alone (assuming
a speed of 10.5 m s1, or ~900 km day1, Bruderer and Boldt
2001). Yet, across multiple days birds responded to seasonal
and weather-related movements of the ITCZ.
Near the ITCZ, higher flight altitudes during the night
and twilight (~1000 m) only interrupted with significant
descending movements around midday during the decreas-
ing EVI period showed a strong resemblance to previously
described patterns of insect flight (Drake and Farrow 1989,
Reynolds et al. 2008). On both the northward and southward
passing of the ITCZ through southern Mali seasonal insect
migrants number in the trillions (Florio et al. 2020). Similar
14
research in East Africa has shown that flying insects are more
numerous near the location of the ITCZ (Tucker et al. 1982,
Drake and Farrow 1988). Nocturnal inversions, common at
the northern side of the ITCZ (Bain et al. 2010, Nicholson
2013), can cause these insects to gather at the top of the
inversion around the physiological temperature thresholds of
insect flight (Drake and Farrow 1989, Drake and Reynolds
2012). Daytime heat and convective uplift increases atmo-
spheric mixing subsequently lowering insect flight (Drake
and Farrow 1989). Despite the co-location of both birds
and insects the positive effects on bird thermoregulation,
by rising above the thermal inversion, cannot be dismissed.
Although pallid swifts display continuous flight throughout
the non-breeding season with increased flapping flight during
nighttime (Hedenström et al. 2019), limited evidence of noc-
turnal foraging exists. Absence of nighttime foraging would
only make nighttime altitudinal co-location at the inversion
beneficial upon the daybreak, supporting the notion of ther-
moregulation. However, nighttime inversion occurrences are
common far north of the front and of the bird positions,
suggesting that preferred bird positions are not only medi-
ated by a positive temperature effect (Supporting informa-
tion). Despite strong correspondence between observed flight
behaviour of birds and patterns described in entomology
literature we cannot conclusively link bird flight changes to
feeding behaviour, as we lack direct observations.
We noted significant differences in flight characteristics
between continental locations during increasing EVI. ese
positions, on the southern side of the ITCZ, are character-
ized by unstable weather (high CAPE values and a low cloud
base heights, Fig. 5, Table 1). In particular upon arrival in
autumn, flights seem to be limited by low cloud base heights
(Table 1). Overall, birds seem adept in changing parts of their
flight strategy, both in location and altitude, to match a vary-
ing atmosphere at, or in the wake of, frontal convergence
zones and thus in relation to the (rapidly) changing direction
of the ITCZ, the state of the vegetation or both.
Conservatively, all flight modes can be interpreted within
the context of energy preservation, due to positive wind
assistance during coastal marine days and limiting tempera-
ture variation, or avoidance of thunderstorm prone regions
which could influence bird health (Salewski et al. 2013,
Albright et al. 2017, Haest et al. 2019). Yet, offshore move-
ments put birds above marine waters, a risk given persistent
trade winds (de Boer et al. 2014), making these coastal flights
not easily explained from an energy conservation point of
view alone. Given a lack of direct insect observations in our
study in situ profiling of the atmosphere, using radar or other
means (Bruderer and Boldt 2001, Hedrick et al. 2018), are
needed to confirm local feeding behaviour of birds in both
coastal and continental locations.
Conclusion
We have shown that pallid swift’s diurnal flight move-
ments are consistent with a cost-efficient way of managing
flight behaviour in and along large tropical atmospheric
convergence systems. Bird positions correspond to locations
where prey should be most densely packed within the atmo-
sphere, and overall flight conditions most optimal. We argue
that resource use might not only be contingent on the abun-
dance of insects, but also favourable wind conditions either
concentrating them or actively supporting foraging behav-
iour by limiting expended energy. Co-location of birds along
sea-breeze frontal convergence and thermal inversions at the
stable front of the ITCZ illustrate this. Persistence of swifts
in wintering feeding grounds might therefore depend on the
prevailing atmospheric conditions, their concentrating effects
on insects and optimal flight environment, rather than solely
the varying vegetation state and co-dependent insect popula-
tions. Our study underscores the imperative of integrating
weather and a full understanding of the atmosphere to assess
small aerial insectivore non-breeding foraging and subse-
quent migration behaviour (Shamoun-Baranes et al. 2010).
An integrative approach, combining foraging behaviour with
detailed atmospheric data could provide further key insights
into regional foraging preferences, in relation to the ITCZ,
and responses to past and future climate change (Boano et al.
2020, Mamalakis et al. 2021).
Acknowledgements – We thank PathTrack for their custom data
processing, and discussions regarding data quality. We are grateful
to Dr. Camponelli for his feedback on the interpretation of the
atmospheric data.
Funding – e GPS loggers were financed by the Belgian
Ornithological Research Association as were a portion of the travel
expenses (LK) for initial deployment (2019). General fieldwork
including recovery (2020) and all aspects of data processing were
privately financed on a voluntary basis.
Author contributions
Lyndon Kearsley: Conceptualization (lead); Data curation
(lead); Formal analysis (equal); Funding acquisition (lead);
Investigation (lead); Methodology (equal); Project admin-
istration (lead); Resources (equal); Software (supporting);
Supervision (lead); Validation (equal); Visualization (equal);
Writing – original draft (equal); Writing – review and editing
(equal). Nathan Ranc: Conceptualization (equal); Formal
analysis (equal); Methodology (equal); Visualization (equal);
Writing – original draft (equal); Writing – review and editing
(equal). Christoph M. Meier: Conceptualization (equal);
Formal analysis (equal); Methodology (equal); Visualization
(equal); Writing – original draft (equal); Writing – review
and editing (equal). Carlos Pacheco: Investigation (equal);
Writing – original draft (supporting); Writing – review
and editing (supporting). Pedro Henriques: Investigation
(equal); Writing – original draft (supporting); Writing –
review and editing (supporting). Gonçalo Elias: Investigation
(equal); Writing – original draft (supporting); Writing –
review and editing (supporting). Martin Poot: Investigation
(equal); Writing – original draft (supporting); Writing –
review and editing (supporting). Andy Williams: Resources
15
(supporting); Writing – review and editing (supporting).
Luis T. Costa: Investigation (equal); Writing – original draft
(supporting); Writing – review and editing (supporting).
Philippe Helsen: Conceptualization (supporting); Writing
– original draft (supporting); Writing – review and editing
(supporting). Koen Hufkens: Conceptualization (equal);
Data curation (lead); Formal analysis (equal); Investigation
(equal); Methodology (equal); Software (lead); Visualization
(lead); Writing – original draft (equal); Writing – review and
editing (equal).
Data availability statement
Data are available from the Zenodo Digital Repository:
<https://doi.org/10.5281/zenodo.6320888> (Kearsley et al.
2022).
Supporting information
e supporting information associated with this article is
available from the online version.
References
Abatan, A. A. et al. 2014. On the characteristics of sea breezes
over Nigerian coastal region. – eor. Appl. Climatol. 116:
93–102.
Agostinelli, C. and Lund, U. 2017. R package: circular statistics
(ver. 0.4-93). – <https://cran.r-project.org/package=circular>.
Åkesson, S. et al. 2020. Evolution of chain migration in an aerial
insectivorous bird, the common swift Apus apus. – Evolution
74: 2377–2391.
Albright, T. P. et al. 2017. Mapping evaporative water loss in desert
passerines reveals an expanding threat of lethal dehydration. –
Proc. Natl Acad. Sci. USA 114: 2283–2288.
Alerstam, T. 2011. Optimal bird migration revisited. – J. Ornithol.
152: 5–23.
Arino, O. et al. 2012. Global land cover map for 2009 (GlobCover
2009). – European Space Agency (ESA) & Univ. Catholique
de Louvain (UCL).
Avgar, T. et al. 2016. Integrated step selection analysis: bridging the
gap between resource selection and animal movement. –
Methods Ecol. Evol. 7: 619–630.
Avgar, T. et al. 2017. Relative selection strength: quantifying effect
size in habitat- and step-selection inference. – Ecol. Evol. 7:
5322–5330.
Bäckman, J. and Alerstam, T. 2001. Confronting the winds: orien-
tation and flight behaviour of roosting swifts, Apus apus. – Proc.
R. Soc. B 268: 1081–1087.
Bain, C. L. et al. 2010. Observations of the nocturnal boundary
layer associated with the West African monsoon. – Monthly
Weather Rev. 138: 3142–3156.
Bajamgnigni Gbambie, A. S. and Steyn, D. G. 2013. Sea breezes at
Cotonou and their interaction with the West African monsoon:
sea-breeze at cotonou. – Int. J. Climatol 33: 2889–2899.
Boano, G. et al. 2020. Climate anomalies affect annual survival
rates of swifts wintering in subSaharan Africa. – Ecol. Evol.
10: 7916–7928.
Bradbury, T. 2000. Meteorology and flight: a pilot’s guide to
weather. – A. & C. Black.
Bruderer, B. and Boldt, A. 2001. Flight characteristics of birds: I.
Radar measurements of speeds. – Ibis 143: 178–204.
Chilson, P. B. et al. (eds) 2017. Aeroecology. – Springer International
Publishing.
Collins, C. T. 2015. Food habits and resource partitioning in a
guild of Neotropical swifts. – Wilson J. Ornithol. 127:
239–248.
Costa, L. T. and Elias, G. L. 1998. Biometrics and survival rates of
pallid swifts Apus pallidus in Portugal. – Ring. Migr. 19: 59–64.
Coulibaly, A. et al. 2019. Characteristics of land and sea breezes
along the Guinea Coast of West Africa. – eor. Appl. Climatol.
138: 953–971.
Cucco, M. et al. 1993. Differences in diet of common (Apus apus)
and pallid (A. pallidus) swifts. – Avocetta 17: 131–138.
de Boer, M. et al. 2014. First documented record of common swift
Apus apus for Surinam and South America. – Cotinga 36:
107–109.
de Margerie, E. et al. 2018. Volume-concentrated searching by an
aerial insectivore, the common swift, Apus apus. – Anim. Behav.
136: 159–172.
Dokter, A. M. et al. 2013. Twilight ascents by common swifts, Apus
apus, at dawn and dusk: acquisition of orientation cues? –
Anim. Behav. 85: 545–552.
Drake, V. A. and Farrow, R. A. 1988. e influence of atmospheric
structure and motions on insect migration. – Annu. Rev.
Entomol. 33: 183–210.
Drake, V. A. and Farrow, R. A. 1989. e ‘aerial plankton’ and
atmospheric convergence. – Trends Ecol. Evol. 4: 381–384.
Drake, V. and Reynolds, D. 2012. Radar entomology: observing
insect flight and migration. – CABI.
Dreelin, R. A. et al. 2018. Flight behavior of individual aerial
insectivores revealed by novel altitudinal dataloggers. – Front.
Ecol. Evol. 6: 182.
Fernández-Tizón, M. et al. 2020. Arthropod biomass increase in
spring correlates with NDVI in grassland habitat. – Sci. Nat.
107: 42.
Finlayson, S. et al. 2021. Birds with multiple homes. e annual
cycle of the pallid swift (Apus pallidus brehmorum). – PLoS One
16: e0259656.
Florio, J. et al. 2020. Diversity, dynamics, direction and magnitude
of high-altitude migrating insects in the Sahel. – Sci. Rep. 10:
20523.
Gordo, O. 2007. Why are bird migration dates shifting? A review
of weather and climate effects on avian migratory phenology.
– Clim. Res. 35: 37–58.
Haest, B. et al. 2019. Autumn bird migration phenology: a pot-
pourri of wind, precipitation and temperature effects. – Global
Change Biol. 25: 4064–4080.
Hedenström, A. 2016. Annual 10-month aerial life phase in the
common swift Apus apus. – Curr. Biol. 26: 3066–3070.
Hedenström, A. et al. 2019. Flight activity in pallid swifts Apus
pallidus during the non-breeding period. – J. Avian Biol. 50:
1–8.
Hedrick, T. L. et al. 2018. Gliding for a free lunch: biomechanics
of foraging flight in common swifts Apus apus. – J. Exp. Biol.
221: jeb186270.
Henningsson, P. et al. 2009. Flight speeds of swifts Apus apus: sea-
sonal differences smaller than expected. – Proc. R. Soc. B 276:
2395–2401.
16
Hersbach, H. et al. 2020. e ERA5 global reanalysis. – Q. J. R.
Meteorol. Soc. 146: 1999–2049.
Hijmans, R. J. 2019. raster: geographic data analysis and modeling.
– <https://cran.r-project.org/package=raster>.
Hobbs, S. E. and Wolf, W. W. 1989. An airborne radar technique
for studying insect migration. – Bull. Entomol. Res. 79:
693–704.
Hufkens, K. et al. 2019. ecmwfr: programmatic interface to the two
European Centre for medium-range weather forecasts API ser-
vices. – <http://doi.org/10.5281/zenodo.2647541>.
Isard, S. A. et al. 2001. Atmospheric motion systems that influence
the redistribution and accumulation of insects on the beaches
of the Great Lakes in North America. – Aerobiologia 17:
275–291.
Issa Lélé, M. and Lamb, P. J. 2010. Variability of the intertropical
front (ITF) and rainfall over the West African Sudan–Sahel
Zone. – J. Clim. 23: 3984–4004.
Johnson, D. S. et al. 2008. Continous-time correlated random walk
model for animal telemetry data. – Ecology 89: 1208–1215.
Kearsley, L. et al. 2022. Data from: e aeroecology of atmospheric
convergence zones: the case of pallid swifts. – Zenodo Digital
Repository, <https://doi.org/10.5281/zenodo.6320888>.
Kemp, M. U. et al. 2012. Quantifying flow-assistance and implica-
tions for movement research. – J. eor. Biol. 308: 56–67.
Kranstauber, B. et al. 2012. A dynamic Brownian bridge movement
model to estimate utilization distributions for heterogeneous
animal movement: the dynamic Brownian bridge movement
model. – J. Anim. Ecol. 81: 738–746.
Liechti, F. et al. 2013. First evidence of a 200-day non-stop flight
in a bird. – Nat Commun. 4: 2554.
Mamalakis, A. et al. 2021. Zonally contrasting shifts of the tropical
rain belt in response to climate change. – Nat. Clim. Change
11: 143–151.
McClintock, B. T. and Michelot, T. 2018. momentuHMM: R
package for generalized hidden Markov models of animal move-
ment. – Methods Ecol. Evol. 9: 1518–1530.
Meier, C. M. et al. 2018. What makes Alpine swift ascend at twi-
light? Novel geolocators reveal year-round flight behaviour. –
Behav. Ecol. Sociobiol. 72: 45.
Miller, S. T. K. 2003. Sea breeze: structure, forecasting and impacts.
– Rev. Geophys. 41: 1011.
Morel, G. J. and Morel, M.-Y. 2008. Habitat use by Palaearctic
migrant passerine birds in West Africa. – Ibis 134: 83–88.
Nicholson, S. E. 2009. A revised picture of the structure of the
‘monsoon’ and land ITCZ over West Africa. – Clim. Dyn. 32:
1155–1171.
Nicholson, S. E. 2013. e West African Sahel: a review of recent
studies on the rainfall regime and its interannual variability. –
ISRN Meteorol. 2013: 1–32.
Nicholson, S. E. 2018. e ITCZ and the seasonal cycle over equa-
torial Africa. – Bull. Am. Meteorol. Soc. 99: 337–348.
Nilsson, C. et al. 2019. Revealing patterns of nocturnal migration
using the European weather radar network. – Ecography 42:
876–886.
Nisi, A. C. et al. 2022. Temporal scale of habitat selection for large
carnivores: balancing energetics, risk and finding prey. – J.
Anim. Ecol. 91: 182–195.
Norevik, G. et al. 2019. Highly mobile insectivorous swifts perform
multiple intratropical migrations to exploit an asynchronous
African phenology. – Oikos 128: 640–648.
Pebesma, E. 2018. Simple features for R: standardized support for
spatial vector data. – R J. 10: 439–446.
Reynolds, A. M. et al. 2016. Orientation in high-flying migrant
insects in relation to flows: mechanisms and strategies. – Phil.
Trans. R. Soc. B 371: 20150392.
Reynolds, D. R. et al. 2008. A radar study of emigratory flight and
layer formation by insects at dawn over southern Britain. – Bull.
Entomol. Res. 98: 35–52.
Romps, D. M. et al. 2014. Projected increase in lightning strikes in
the United States due to global warming. – Science 346: 851–854.
Russell, R. W. 1999. Precipitation scrubbing of aerial plankton:
inferences from bird behavior. – Oecologia 118: 381–387.
Russell, R. W. and Wilson, J. W. 2001. Spatial dispersion of aerial
plankton over east-central Florida: Aeolian transport and coast-
line concentrations. – Int. J. Remote Sens. 22: 2071–2082.
Salewski, V. et al. 2013. Multiple weather factors affect apparent
survival of European passerine birds. – PLoS One 8: e59110.
Santos, C. D. et al. 2020. e gateway to Africa: what determines
sea crossing performance of a migratory soaring bird at the
Strait of Gibraltar? – J. Anim. Ecol. 89: 1317–1328.
Sanz, J. J. et al. 2003. Climate change and fitness components of
a migratory bird breeding in the Mediterranean region. –
Global Change Biol. 9: 461–472.
Satterfield, D. A. et al. 2020. Seasonal insect migrations: massive,
influential and overlooked. – Front. Ecol. Environ. 18: 335–344.
Sauvageot, H. and Despaux, G. 1996. e clear-air coastal versper-
tine radar bands. – Bull. Am. Meteorol. Soc. 77: 673–681.
Shamoun-Baranes, J. et al. 2006. A comparative analysis of the
influence of weather on the flight altitudes of birds. – Bull. Am.
Meteor. Soc. 87: 47–62.
Shamoun-Baranes, J. et al. 2010. Integrating meteorology into
research on migration. – Integr. Comp. Biol. 50: 280–292.
Shamoun-Baranes, J. et al. 2017. Atmospheric conditions create
freeways, detours and tailbacks for migrating birds. – J. Comp.
Physiol. A 203: 509–529.
Signer, J. et al. 2019. Animal movement tools (amt): R package for
managing tracking data and conducting habitat selection anal-
yses. – Ecol. Evol. 9: 880–890.
Taylor, L. R. 1974. Insect migration, flight periodicity and the
boundary layer. – J. Anim. Ecol. 43: 225.
urell, H. et al. 2014. Applications of step-selection functions
in ecology and conservation. – Mov. Ecol. 2: 4.
Tuck, S. L. et al. 2014. MODISTools – downloading and process-
ing MODIS remotely sensed data in R. – Ecol. Evol. 4:
4658–4668.
Tucker, M. R. et al. 1982. Further evidence for windborne move-
ment of armyworm moths, Spodoptera exempta, in East Africa.
– Ecol. Entomol. 7: 463–473.
Zhang, G. and Cook, K. H. 2014. West African monsoon demise:
climatology, interannual variations and relationship to seasonal
rainfall: West African monsoon demise. – J. Geophys. Res.
Atmos. 119: 10175–10193.
Zucchini, W. et al. 2016. Hidden Markov models for time series:
an introduction using R, 2nd edn. – Chapman and Hall/CRC.
... Although presenting some limitations (Taberlet et al., 2012), the ability to provide high taxonomic resolution to the consumed taxa (Jackson et al., 2014;Gibson et al., 2015), while avoiding prior knowledge of the present prey (de Sousa et al., 2019), makes this technique a good tool for diet assessment. Despite its broad application and its potential to unravel swifts' ecology, to date it has not been applied to assess the diet of any of the Palearctic swift species, whose populations, however, have been the subject of other biological and ecological studies (Hedenstr€ om et al., 2019;Cibois et al., 2022;Kearsley et al., 2022). ...
Article
Full-text available
Pallid Swifts, as other swifts, are birds extremely adapted to an aerial lifestyle, showing unique adaptations that allow them to fly almost continuously. The diet of these non-stopping high-altitudinal aerial birds has been mostly studied through techniques that fail to produce highly resolved prey identifications, and for that have been replaced by molecular techniques, as DNA metabarcoding. Faecal samples of Pallid Swifts (Apus pallidus) were monthly collected from a colony in the north of Portugal during the breeding season. DNA from the faecal samples was used to sex the birds and to identify the arthropods present in the diet through DNA metabarcoding. From the detected prey items, 74 families were identified belonging to 16 orders, with Hymenoptera and Hemiptera being the most frequently consumed. There were seasonal variations in diet richness, composition and prey size. Regarding the diet of males and females, although no differences were found between the diet of males and females in terms of composition and richness, there were differences in the size of arthropods preyed by the different sexes, with males feeding on larger arthropods. The large seasonal variation in Pallid Swifts' diet during the breeding season is probably a result of spatiotemporal variation in aerial prey, of which swifts likely predate opportunistically. Although no significant differences were detected in diet richness and composition between sexes, the fact that males consumed larger prey may suggest the existence of sexual dietary segregation in this group of birds. At last, several pest species were found in these swifts’ diet, which, if studied through DNA metabarcoding, can be used to monitor small arthropods, including airborne pests.
Article
Full-text available
We tracked pallid swifts (Apus pallidus brehmorum) from a single breeding colony in Gibraltar over two years. Our results show movement of birds between specific regions within the non-breeding geographical area at specific times of the year. The tracking of a single individual showed remarkable fidelity to the areas visited between years. Furthermore, two pallid swifts tracked over the entire eight-month non-breeding period, while in Africa, gave no indication of coming to land, supporting previous findings of an airborne existence in swifts outside the breeding season. In addition, the crossing of the Sahara Desert to and from breeding grounds is remarkably fast, with one individual crossing it in just over a day. We discuss our findings in the context of bird migration evolutionary strategies.
Article
Full-text available
When navigating heterogeneous landscapes, large carnivores must balance trade‐offs between multiple goals, including minimizing energetic expenditure, maintaining access to hunting opportunities, and avoiding potential risk from humans. The relative importance of these goals in driving carnivore movement likely changes across temporal scales, but our understanding of these dynamics remains limited. Here we quantified how drivers of movement and habitat selection changed with temporal grain for two large carnivore species living in human‐dominated landscapes, providing insights into commonalities in carnivore movement strategies across regions. We used high‐resolution GPS collar data and integrated step selection analyses to model movement and habitat selection for African lions (Panthera leo) in Laikipia, Kenya and pumas (Puma concolor) in the Santa Cruz Mountains of California across eight temporal grains, ranging from 5 minutes to 12 hours. Analyses considered landscape covariates that are related to energetics, resource acquisition, and anthropogenic risk. For both species, topographic slope, which strongly influences energetic expenditure, drove habitat selection and movement patterns over fine temporal grains but was less important at longer temporal grains. In contrast, avoiding anthropogenic risk during the day, when risk was highest, was consistently important across grains, but the degree to which carnivores relaxed this avoidance at night was strongest for longer‐term movements. Lions and pumas modified their movement behavior differently in response to anthropogenic features: lions sped up while near humans at fine temporal grains, while pumas slowed down in more developed areas at coarse temporal grains. Finally, pumas experienced a trade‐off between energetically efficient movement and avoiding anthropogenic risk. Temporal grain is an important methodological consideration in habitat selection analyses, as drivers of both movement and habitat selection changed across temporal grain. Additionally, grain‐dependent patterns can reflect meaningful behavioral processes, including how fitness‐relevant goals influence behavior over different periods of time. In applying multi‐scale analysis to fine‐resolution data, we showed that two large carnivore species in very different human‐dominated landscapes balanced competing energetic and safety demands in largely similar ways. These commonalities suggest general strategies of landscape use across large carnivore species.
Article
Full-text available
Future changes in the position of the intertropical convergence zone (ITCZ; a narrow band of heavy precipitation in the tropics) with climate change could affect the livelihood and food security of billions of people. Although models predict a future narrowing of the ITCZ, uncertainties remain large regarding its future position, with most past work focusing on zonal-mean shifts. Here we use projections from 27 state-of-the-art climate models and document a robust zonally varying ITCZ response to the SSP3-7.0 scenario by 2100, with a northward shift over eastern Africa and the Indian Ocean and a southward shift in the eastern Pacific and Atlantic oceans. The zonally varying response is consistent with changes in the divergent atmospheric energy transport and sector-mean shifts of the energy flux equator. Our analysis provides insight about mechanisms influencing the future position of the tropical rain belt and may allow for more-robust projections of climate change impacts. The intertropical convergence zone is predicted to narrow under climate change with large uncertainties about its location. Analysis with CMIP6 models shows a zonally varying response, with northward shift over east Africa and the Indian Ocean and southward shift in east Pacific and Atlantic oceans.
Article
Full-text available
Long-distance migration of insects impacts food security, public health, and conservation–issues that are especially significant in Africa. Windborne migration is a key strategy enabling exploitation of ephemeral havens such as the Sahel, however, its knowledge remains sparse. In this first cross-season investigation (3 years) of the aerial fauna over Africa, we sampled insects flying 40–290 m above ground in Mali, using nets mounted on tethered helium-filled balloons. Nearly half a million insects were caught, representing at least 100 families from thirteen orders. Control nets confirmed that the insects were captured at altitude. Thirteen ecologically and phylogenetically diverse species were studied in detail. Migration of all species peaked during the wet season every year across localities, suggesting regular migrations. Species differed in flight altitude, seasonality, and associated weather conditions. All taxa exhibited frequent flights on southerly winds, accounting for the recolonization of the Sahel from southern source populations. “Return” southward movement occurred in most taxa. Estimates of the seasonal number of migrants per species crossing Mali at latitude 14°N were in the trillions, and the nightly distances traversed reached hundreds of kilometers. The magnitude and diversity of windborne insect migration highlight its importance and impacts on Sahelian and neighboring ecosystems.
Article
Full-text available
Data from remote sensing are often used as proxies to quantify biological processes, especially at large geographical scales. The normalized difference vegetation index (NDVI) is the most frequently used proxy for primary productivity. Assuming a direct, positive interrelation between primary and secondary production in terrestrial habitats, NDVI is often used to predict food availability for higher trophic levels. However, the relationship between NDVI and arthropod biomass has rarely been tested. In this study, we analyzed extensive datasets of arthropod communities from semi-natural grasslands in central Europe to test the relationship between arthropod biomass of consumer trophic levels ("herbivores," "mixed," and "carnivores") in grassland communities and NDVI during the spring season. We found that arthropod biomass generally increased with NDVI. The same positive relationship between biomass and NDVI was observed for each individual trophic level. Cross-correlation analyses did not show statistically significant lags between the NDVI and biomass of herbivores and carnivores. All in all, our study provides correlational evidence for the positive relation of primary and secondary productivity in temperate terrestrial habitats during spring. Moreover, it supports the applicability of NDVI data as a suitable habitat-specific proxy for the food availability of insectivores during spring.
Article
Full-text available
Spectacular long-distance migration has evolved repeatedly in animals enabling exploration of resources separated in time and space. In birds, these patterns are largely driven by seasonality, cost of migration, and asymmetries in competition leading most often to leap-frog migration, where northern breeding populations winter furthest to the south. Here we show that the highly aerial common swift Apus apus, spending the non-breeding period on the wing, instead exhibits a rarely-found chain migration pattern, where the most southern breeding populations in Europe migrate to wintering areas furthest to the south in Africa, while the northern populations winter to the north. The swifts concentrated in three major areas in sub-Saharan Africa during the non-breeding period, with substantial overlap of nearby breeding populations. We found that the southern breeding swifts were larger, raised more young, and arrived to the wintering areas with higher seasonal variation in greenness (Normalized Difference Vegetation Index, NDVI) earlier than the northern breeding swifts. This unusual chain migration pattern in common swifts is largely driven by differential annual timing and we suggest it evolves by prior occupancy and dominance by size in the breeding quarters and by prior occupancy combined with diffuse competition in the winter. This article is protected by copyright. All rights reserved.
Article
Full-text available
Several species of migratory swifts breed in the Western Palearctic, but they differ in reproductive traits and nonbreeding areas explored in Africa. We examined survival and recapture probabilities of two species of swifts by capture–mark–recapture data collected in northern Italy (Pallid Swift Apus pallidus in Carmagnola, Turin, and Common Swift Apus apus in Guiglia, Modena) in the breeding season (May–July). Apparent survival rates were relatively high (>71%), comparable to other studies of European swifts, but showed marked annual variations. We used geolocators to establish the exact wintering areas of birds breeding in our study colonies. Common Swifts explored the Sahel zone during migration and spent the winter in SE Africa, while the Pallid Swifts remained in the Sahel zone for a longer time, shifting locations southeast down to Cameroun and Nigeria later in winter. These movements followed the seasonal rains from north to south (October to December). In both species, we found large yearly differences in survival probabilities related to different climatic indices. In the Pallid Swift, wintering in Western Africa, the Sahel rainfall index best explained survival, with driest seasons associated with reduced survival. In the Common Swift, wintering in SE Africa, the El Niño–Southern Oscillation (ENSO) cycle performed significantly better than Sahel rainfall or North Atlantic Oscillation (NAO). Extreme events and precipitation anomalies in Eastern Africa during La Niña events resulted in reduced survival probabilities in Common Swifts. Our study shows that the two species of swifts have similar average annual survival, but their survival varies between years and is strongly affected by different climatic drivers associated with their respective wintering areas. This finding could suggest important ecological diversification that should be taken into account when comparing survival and area use of similar species that migrate between temperate breeding areas and tropical wintering areas.
Article
Full-text available
Within the Copernicus Climate Change Service (C3S), ECMWF is producing the ERA5 reanalysis, which embodies a detailed record of the global atmosphere, land surface and ocean waves from 1950 onwards once completed. This new reanalysis replaces the ERA‐Interim reanalysis that was started in 2006 (spanning 1979 onwards). ERA5 is based on the Integrated Forecasting System (IFS) Cy41r2 which was operational in 2016. ERA5 thus benefits from a decade of developments in model physics, core dynamics and data assimilation. In addition to a significantly enhanced horizontal resolution of 31 km, compared to 80 km for ERA‐Interim, ERA5 has hourly output throughout, and an uncertainty estimate from an ensemble (3‐hourly at half the horizontal resolution). This paper describes the general setup of ERA5, as well as a basic evaluation of characteristics and performance. Focus is on the dataset from 1979 onwards that is currently publicly available. Re‐forecasts from ERA5 analyses show a gain of up to one day in skill with respect to ERA‐Interim. Comparison with radiosonde and PILOT data prior to assimilation shows an improved fit for temperature, wind and humidity in the troposphere, but not the stratosphere. A comparison with independent buoy data shows a much improved fit for ocean wave height. The uncertainty estimate reflects the evolution of the observing systems used in ERA5. The enhanced temporal and spatial resolution allows for a detailed evolution of weather systems. For precipitation, global‐mean correlation with monthly‐mean GPCP data is increased from 67% to 77%. In general low‐frequency variability is found to be well‐represented and from 10 hPa downwards general patterns of anomalies in temperature match those from the ERA‐Interim, MERRA‐2 and JRA‐55 reanalyses. This article is protected by copyright. All rights reserved.
Article
Full-text available
Large bodies of water represent major obstacles for the migration of soaring birds because thermal updrafts are absent or weak over water. Soaring birds are known to time their water crossings with favourable weather conditions and there are records of birds falling into the water and drowning in large numbers. However, it is still unclear how environmental factors, individual traits and trajectory choices affect their water crossing performance, this being important to understand the fitness consequences of water barriers for this group of birds. We addressed this problem using the black kite (Milvus migrans) as model species at a major migration bottleneck, the Strait of Gibraltar. We recorded high‐resolution GPS and triaxial accelerometer data for 73 birds while crossing the Strait of Gibraltar, allowing the determination of sea crossing duration, length, altitude, speed and tortuosity, the flapping behaviour of birds and their failed crossing attempts. These parameters were modelled against wind speed and direction, time of the day, solar irradiance (proxy of thermal uplift), starting altitude and distance to Morocco, and age and sex of birds. We found that sea crossing performance of black kites is driven by their age, the wind conditions, the starting altitude and distance to Morocco. Young birds made longer sea crossings and reached lower altitude above the sea than adults. Crosswinds promoted longer sea crossings, with birds reaching lower altitudes and with higher flapping effort. Birds starting at lower altitudes were more likely to quit or made higher flapping effort to complete the crossing. The location where birds started the sea crossings impacted crossing distance and duration. We present evidence that explains why migrating soaring birds accumulate at sea passages during adverse weather conditions. Strong crosswinds during sea crossings force birds to extended flap‐powered flight at low altitude, which may increase their chances of falling in the water. We also showed that juvenile birds assume more risks than adults. Finally, the way in which birds start the sea crossing is crucial for their success, particularly the starting altitude, which dictates how far birds can reach with reduced flapping effort.
Article
During seasonal changes around the globe, trillions of insects are on the move. Many insect populations, including butterflies, moths, hoverflies, and dragonflies, make repeated seasonal migrations each year. It is only during the past century that biologists have come to accept the concept of insect migration, and new research using radar, citizen science, and stable isotopes has revealed unexpected insights about this phenomenon. Drawing on these findings, we demonstrate how seasonal insect movements are both massive and ecologically influential, with consequences for food webs, nutrient transport, pollination, and infectious disease. Responding to environmental changes, some mobile insect populations are declining or shifting the timing and extent of their journeys. We suggest research and policy priorities for investigating and protecting insect migrations. Outcomes from such work could transform strategies for agricultural pest control and wildlife conservation, and could help preserve the ecological functions performed by migratory insects.