Content uploaded by Mia Momberg
Author content
All content in this area was uploaded by Mia Momberg on Aug 08, 2022
Content may be subject to copyright.
Factors determining nest-site selection of surface-
nesting seabirds: a case study on the world’s largest
pelagic bird, the Wandering Albatross (Diomedea
exulans)
MIA MOMBERG,
1
*PETER G. RYAN,
2
DAVID W. HEDDING,
3
JANINE SCHOOMBIE,
4
KYLE A. GODDARD,
4
KEN J. CRAIG
4
& PETER C. LE ROUX
1
1
Department of Plant and Soil Sciences, University of Pretoria, Pretoria, South Africa
2
FitzPatrick Institute of African Ornithology, University of Cape Town, Rondebosch, South Africa
3
Department of Geography, University of South Africa, Pretoria, South Africa
4
Department of Mechanical and Aeronautical Engineering, University of Pretoria, Pretoria, South Africa
Several factors may drive bird nest-site selection, including predation risk, resource avail-
ability, weather conditions and interaction with other individuals. Understanding the dri-
vers affecting where birds nest is important for conservation planning, especially where
environmental change may alter the distribution of suitable nest-sites. This study investi-
gates which environmental variables affect nest-site selection by the Wandering Albatross
Diomedea exulans, the world’s largest pelagic bird. Here, wind characteristics are quanti-
tatively investigated as a driver of nest-site selection in surface-nesting birds, in addition
to several topographical variables, vegetation and geological characteristics. Nest locations
from three different breeding seasons on sub-Antarctic Marion Island were modelled to
assess which environmental factors affect nest-site selection. Elevation was the most
important determinant of nest-site selection, with Wandering Albatrosses only nesting at
low elevations. Distance from the coast and terrain roughness were also important pre-
dictors, with nests more generally found close to the coast and in flatter terrain, followed
by wind velocity, which showed a hump-shaped relationship with the probability of nest
occurrence. Nests occurred more frequently on coastal vegetation types, and were absent
from polar desert vegetation (generally above c. 500 m elevation). Of the variables that
influence Wandering Albatross nest location, both vegetation type and wind characteris-
tics are likely to be influenced by climate change, and have already changed over the last
50 years. As a result, the availability of suitable nest-sites needs to be considered in light
of future climate change, in addition to the impacts that these changes will have on for-
aging patterns and prey distribution. More broadly, these results provide insights into
how a wide range of environmental variables, including wind, can affect nest-site selec-
tion of surface-nesting seabirds.
Keywords: generalized additive model, generalized linear model, topography, vegetation type,
wind.
Nest-site selection by birds may be driven by a
variety of environmental factors, including habitat
conditions related to predator avoidance, trophic
and non-trophic resource availability, exposure to
weather conditions and interactions with con-
specifics (Jones 2001). For example, many ground-
nesting birds choose sites based on topography
that allows them to either detect predators from
afar (e.g. open areas) or those that provide protec-
tion from predators (e.g. less accessible sites within
*Corresponding author.
Email: miamomberg@gmail.com
Twitter: @MiaMomberg
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDeri vs License, which permits use and
distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
Ibis (2022) doi: 10.1111/ibi.13111
wetlands; Colwell et al.2011, Miller et al.2014,
Cunningham et al.2016, Korne et al.2020).
Vegetation is also regularly linked to the avoidance
or detection of predators (Muir & Colwell 2010),
and is often important in determining where nests
are constructed (Flemming et al.2019). Food
availability is another key factor that is often
linked to vegetation, and that can influence nest-
site selection by bird species (McCollin 1998).
Social cues, both inter- and intra-specific, can also
affect the choice of breeding sites in songbirds
(Betts et al.2008). Temperature tends to be an
important driver of nest-site selection, particularly
in systems that experience high levels of solar radi-
ation, with nests typically located in cooler micro-
sites within hot environments (With &
Webb 1993, Kauffman et al.2021), and probably
in buffered microsites in cold environments. Quan-
tifying and understanding these factors is impor-
tant in order to preserve environments that will be
suitable for nesting in the future, and to under-
stand how populations will be affected by environ-
mental change.
Wind conditions may affect where birds choose
to nest, but the influence of wind has been poorly
studied (see, for example, Cunningham
et al.2016). Wind has been hypothesized to be
important for nest location in certain specific sce-
narios, where, for example, nests are constructed
downwind from taller vegetation, which may act as
a windbreak (Holmes et al.2020). Similarly,
microscale tundra features in the Arctic may be
important factors influencing shorebird nest-site
location, because these landforms provide wind-
breaks (Cunningham et al.2016). In other systems,
the windward sides of nests have trampled vegeta-
tion where the birds enter the nest (Miller
et al.2014). Common Guillemots Uria aalge (Alci-
dae), a cliff-breeding seabird, nest in areas that are
protected from wind and rain and waves, which
are also affected by the wind characteristics (Lemp-
idakis et al.2022). Wind is being recognized as an
important factor for seabirds, affecting their move-
ment (Weimerskirch et al.2000, Clay et al.2020),
foraging ecology (Cornioley et al.2016), predation
(Gilchrist & Gaston 1997) and even life-history
(Weimerskirch et al.2012). Large seabirds breed-
ing on sub-Antarctic islands are an ideal system in
which to study the influence of wind on the nest-
site selection of surface-nesting species. In these
environments, natural predators are absent
(although predation by giant petrels Macronectes
spp. may occur; Dilley et al.2013), proximity to
food should not be a factor because these birds typ-
ically cover extremely large distances to forage
(Gaston 2004), winds are constant and strong, and,
therefore, environmental effects can be studied
without the interference of predator risk or
resource availability, which affect bird nesting sites
strongly in other systems.
The Procellariiformes is a large order of sea-
birds, including the bird with the largest wingspan,
the Wandering Albatross Diomedea exulans
(Diomedeidae), an oceanic nomad that only visits
land to breed, and nests exclusively on the islands
in the Southern Ocean (ACAP 2009). Wandering
Albatrosses are long-lived birds (up to at least
57 years) that breed in loose colonies. They build
large raised mound nest structures from surround-
ing vegetation and peat (Tickell 2000), having
large, but very localized, impacts on the terrestrial
ecosystem where they nest (Joly et al.1987).
Their nests tend to be in open, flat areas, with
breeding pairs exhibiting high fidelity rates, almost
always returning to the same site (Gauthier
et al.2010). Although not tested, it has been
hypothesized that differences in breeding success
on islands could be linked to environmental condi-
tions, in particular shelter from westerly winds
(Rackete et al.2021).
The global population of Wandering Albatross
is declining, mainly as a result of bycatch in long-
line fishing, and the species is currently listed as
Vulnerable (Poncet et al.2017, Birdlife Interna-
tional 2018). Several threats to Wandering Alba-
tross populations are well documented and
understood (see Jones & Ryan 2010, Pardo
et al.2017, Jones et al.2019), but factors affecting
nest-site selection, and what this would mean for
the distribution and availability of future breeding
sites, have not been investigated. Wandering Alba-
tross foraging patterns, breeding success and sur-
vival are affected by wind in different ways
(Weimerskirch et al.2000,2012, Cornioley
et al.2016, Pardo et al.2017). As a result of their
large wingspan and heavy weight, energy expendi-
ture is largest for these birds when they take off
(Weimerskirch et al.2000). Wandering Albatrosses
take off into strong headwinds, and larger alba-
trosses, which have a higher wing loading and
require higher wind speeds for gliding flight, are
more strongly influenced by favourable wind con-
ditions when making flight decisions (Clay
et al.2020).
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
2M. Momberg et al.
This study investigates which environmental
drivers affect nest-site selection among Wandering
Albatrosses, and also, for the first time, quantita-
tively investigates wind as one of these factors. We
expect that Wandering Albatross nests will be
located in flat areas with adequate vegetation
cover to build their nests that have adequate space
for the birds to take off into the predominant
winds, and which have moderate and predictable
wind speeds to facilitate take-off and landing. We
test these patterns using data from three breeding
seasons to determine their generality.
METHODS
Study area
This study was conducted on sub-Antarctic Marion
Island (46°540S, 37°450E; 293 km
2
), in the Prince
Edward Islands group, southeast of Africa. Marion
Island is situated in the ‘roaring forties’, a band of
strong westerly winds in the Southern Ocean,
where strong winds blow on most days of the year,
with winds predominantly coming from the west
(le Roux 2008). The island has a hyper-oceanic
climate, where the Southern Ocean moderates
daily and seasonal temperature variation (mean
annual temperature of 6 °C, mean daily tempera-
ture range of 1.9 °C). The island receives around
1800 mm of precipitation annually, with rain or
snow falling on more than 290 days per year
(1960–2018, South African Weather Service
unpublished data, le Roux 2008, le Roux &
McGeoch 2008a). Vegetation on Marion Island
can broadly be described as tundra, with similari-
ties to the tundra systems in the northern hemi-
sphere (Kemppinen et al.2021).
Data collection
The geographical coordinates of 1906 of 1952
active Wandering Albatross nests on Marion Island
(97.6% of all active nests in the 2016/17 breeding
season; Fig. 1, after removing four outliers at ele-
vations higher than 100 m above sea level (asl)
and records that included errors in locations or
lacked coverage by the digital surface model) were
collected using a handheld GPS device (following
the methods of Nel et al.2002) in January 2017.
Nest detection rate is very high because of the
short vegetation (generally <0.1–0.2 m) and the
large size and white plumage of Wandering
Albatrosses (adults and chicks). In addition, nests
tend to be in the same areas each year, and the
entire island is systematically searched for nests
each year. Active nests from the 2006 (1711
nests) and 2018 (2139 nests) breeding seasons
were used to confirm the generality of results
across years. This provides a test of the generality
of observed patterns, with data more than
10 years apart, and data from a subsequent breed-
ing season representing the nesting preferences of
different individual birds (as Wandering Alba-
trosses generally breed every second year,
although, because of the longevity of this species
or failed nest attempts in one year leading to
breeding again in the subsequent year, not all
records are independent).
Data processing
Absences (n=10 000) were randomly generated
in ArcGIS Pro, with a minimum distance of 30 m
between all absence points and between any
absence point and a nest location.
The island is surveyed intensively, so all loca-
tions that were recorded as not having an active
nest represent true absences for that particular
breeding season (see Guillera-Arroita et al.2015).
Models using pseudo-absences sampled from envi-
ronments that are dissimilar to the environments
in which presences occur may be positively biased
(Hazen et al.2021), so absences were a priori gen-
erated in areas that were deemed biologically suit-
able based on initial observations of where nests
occur. The following factors were considered when
choosing where to generate absences: nests
occurred at elevations lower than 100 m asl, and
the species does not nest on cliffs, because of their
lack of agility when landing. Marion Island has
over 130 scoria (cinder) cones (Boelhouwers
et al.2008) comprising loose unconsolidated rock
resulting from explosive volcanic events (Verwo-
erd 1971, Rudolph et al.2021). This geology type
typically supports little to no vegetation and scoria
cones are also generally very steep (see Hol-
ness 2004, with measurements of up to 35°), and
Wandering Albatrosses have not been observed to
nest on these cones. As scoria (cinder) cone vege-
tation was represented by only one nest in the
dataset, this vegetation type was lumped with
polar desert because both of these vegetation types
represent abiotically extreme environments. There-
fore, absences were not generated from areas with
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
Nest-site selection of Wandering Albatross 3
Figure 1. (a) Map of Marion Island with 200-m contour lines and the 1906 georeferenced Wandering Albatross nests recorded in
January 2017 (light pink points); (b) 10 000 randomly generated absences (dark blue); (c) digital surface model; (d) terrain rugged-
ness index; (e) Wandering Albatross parent and chick on nest; (f) wind velocity (m/s); (g) wind turbulence. For all maps, lighter col-
ours indicate higher values and darker colours indicate lower values. See Supporting Information Figures S2–S7 for larger images,
including legends and scales.
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
4M. Momberg et al.
an altitude greater than 100 m asl, slopes greater
than 45°, scoria cones or lakes.
The elevation for each presence and absence
point was extracted from a 1-m resolution digital
surface model (Fig. 1c; DRDLR 2019). Tempera-
ture and elevation show strong collinearity on
Marion Island (Leihy et al.2018), and therefore
temperature was not included as a predictor. The
digital surface model was subsequently resampled
to 10-m resolution, using bilinear interpolation,
before calculating and extracting the terrain
ruggedness index (TRI) and slope angle in ArcGIS
Pro. The TRI represents the elevation difference
between a cell and the eight cells surrounding it,
and was used as a proxy for the flatness of the
space around each point, hereafter referred to as
terrain roughness (rugosity). This indicated the
available space that the birds would be able to use
for take-off and landing (Fig. 1d).
Distance to the coast was calculated because
salt spray from the ocean might affect nest-site
selection, and salt spray can travel as far as 300 m
inland on Marion Island (Smith 1978a). The vege-
tation type at each point was determined from the
latest vegetation classification for Marion Island,
which mapped five broad vegetation types: coastal
vegetation, mire-slope vegetation, fellfield, scoria
(cinder) cones and polar desert (Smith &
Mucina 2006). Geology for each point was deter-
mined from Rudolph et al.(2021), and then sim-
plified to two categories, namely flows from before
the last glacial maximum (pre-glacial flows), which
represent a smooth substrate, and post-glacial
flows that are more rugged.
The weighted mean wind velocity and wind
turbulence intensity were extracted from a compu-
tational fluid dynamics model of Marion Island
(30-m resolution; Goddard et al.2022). These
mean values were weighted by the observed fre-
quency of wind recordings (see Supporting Infor-
mation, Fig. S1) from 16 wind directions
(recorded over 2 years; 2018 and 2019). The com-
putational fluid dynamics model uses the full digi-
tal surface model of Marion Island and simulates
air flow over the topography by iteratively solving
a set of partial differential equations (Reynolds-
Averaged Navier Stokes Equations: Versteeg &
Malalasekera 2007, Cindori et al.2018). Sixteen
wind directions (at intervals of 22.5°) were used as
the free-stream condition, with a reference speed
of 8.22 m/s at 1 m above ground (based on the
average wind velocity across all 17 anemometers).
The model included considerations for the atmo-
spheric boundary layer and the effect of the Cori-
olis force (Breedt et al.2018, Goddard
et al.2022), and generated estimates of wind
velocity and turbulence for 30 930-m cells across
Marion Island (with mean errors of 26.9% for
velocity and 32.6% for turbulence; Goddard
et al.2022). From these analyses, we extracted
wind characteristics at 1 m above the ground
because we considered this relevant to adult alba-
trosses when on the ground and chicks on nests.
Outliers for wind turbulence, representing values
greater than the value of the 99th centile, were set
to the value of the 99th centile.
Statistical analyses
The terrain roughness and wind turbulence values
were logarithmically transformed before analyses
to reduce the leverage of a few large values. Slope
and roughness were strongly correlated (Pearson
r=0.86, P<0.001), so slope was excluded from
the analyses because roughness was considered to
be more biologically relevant in terms of quantify-
ing the available flat space around a point for take-
off and landing. None of the remaining predictor
variables had a generalized variance inflation factor
greater than 2.5, and none was strongly correlated
with another variable (Pearson r<|0.7|; see Sup-
porting Information, Fig. S8). Wind velocity, tur-
bulence, vegetation type, elevation, geology,
terrain roughness and distance from the coast were
investigated as predictors of nest presence or
absence using a generalized additive model
(GAM) and a generalized linear model (GLM;
including quadratic terms of continuous predic-
tors), implementing a binomial distribution. These
two differing statistical approaches were used to
cross-check the results, providing higher confi-
dence in the conclusions if/where consistent results
were observed. Variable importance for GAMs
and GLMs was calculated by comparing the Pear-
son correlation between predictions made on the
original data and predictions made on the data
where the predictor variable of interest has been
randomly shuffled (following Niittynen &
Luoto 2018). The calculations of variable impor-
tance were calculated 10 times and the mean
importance value, rescaled to percentage, is
reported.
All statistical analyses were performed in R ver-
sion 4.1.0 (R Core Team 2021), using additional
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
Nest-site selection of Wandering Albatross 5
functions from the mgcv (Wood &
Augustin 2002), ggplot2 (Wickham 2016), voxel
(Garcia de la Garza et al.2018), ggpubr (Kassam-
bara 2020) and scico (Pedersen & Crameri 2020)
libraries. All figures were produced using the sci-
entific colour scheme batlow (Crameri 2018), to
prevent visual distortion of the data and to be
accessible to readers with colour-vision deficiency
(Crameri et al.2020).
RESULTS
Wandering Albatrosses breed around most of the
coast of Marion Island, but are largely absent from
the south coast, and occur at the highest densities
along the northeast coast (Fig. 1a). The results
presented here are from the 2017 breeding season,
but similar results were obtained from the 2006
and 2018 breeding seasons, with predictors consis-
tently ranked in the same order of importance and
with response curves showing similar shapes across
all three datasets (see Supporting Information,
Figs S9 and S10, and Tables S1 and S2 for results
from these additional years). The spatial nature of
the data means that spatial autocorrelation is pre-
sent, but using a spatially thinned version of the
data produced similar results (see Supporting
Information, Table S3 for more information). The
GAM explained 33.1% of the deviance in the
2017 nest distribution data, and the GLM
explained 31.5% of the deviance.
Four of the five continuous predictors con-
tributed significantly to explaining nest-site suit-
ability in both the GAM and GLM, but wind
turbulence did not significantly affect the nest-site
suitability in either of the two models (Table 1).
For both statistical approaches, elevation was the
most important predictor, followed by distance
from the coast, vegetation type, terrain roughness
and wind velocity (Table 1). Most nests were
located in coastal or mire-slope vegetation types,
with a small proportion in fellfield; no nests were
recorded in polar desert vegetation (Fig. 2). Pre-
glacial flows made up a significantly larger propor-
tion of the underlying geology on which nests
were found than would be expected by chance,
with a similar proportion of pre-glacial and post-
glacial flows observed for absences (Fig. 2).
Response curves from the GAMs showed that
there was a higher probability of a nest occurring
in areas close to the coast and at low elevation
(Fig. 3). The more rugged the terrain, the lower
the probability of a site being used to build a nest.
Wind velocity had a hump-shaped relationship
with nest occurrence, with the highest probability
of a nest in areas with intermediate wind veloci-
ties. Areas with higher wind turbulence generally
had a smaller probability of containing a nest than
areas with lower turbulence, although this rela-
tionship was not significant (Fig. 3).
DISCUSSION
Predictor variables representing topographic, vege-
tation, geological and wind velocity characteristics
were significantly related to Wandering Albatross
nest locations in the tundra landscape of Marion
Island, although the relative importance of these
predictors varied strongly. The consistency in
results across three different years indicates the
generality in these findings (i.e. across different
individuals and different time periods). Although
adults and most juveniles demonstrate strong fide-
lity to their breeding/natal sites, there is sufficient
dispersal of juveniles in particular to allow breed-
ing sites to shift in response to local conditions
(Inchausti & Weimerskirch 2002, Gauthier
et al.2010).
Despite the very windy conditions present at
the study site, and the strong impacts on albatross
flight patterns and feeding behaviour, wind veloc-
ity was only the fifth most important driver of
nest-site selection in Wandering Albatrosses, after
elevation, distance from the coast, vegetation type
and terrain roughness. Wind velocity can affect
birds and their nests in several ways. High wind
speeds can greatly decrease nest temperatures
(Heenan & Seymour 2012, Gray & Deem-
ing 2017), and also affect the chick’s body temper-
ature, potentially reducing their growth rate
(Sauve et al.2021). Shorebirds have been found
to adjust their directional orientation when resting
in response to wind speed and ambient tempera-
tures, allowing them to increase the efficiency of
thermoregulation and save energy (Cestari & de
Melo 2022). Protection from wind in general, or
from the strongest winds at a site, have been theo-
rized to impact where birds construct their nests
based on topographic and vegetative protection
(Cunningham et al.2016, Holmes et al.2020).
Our study presents quantitative results for wind
impacting bird nest-site selection, and shows that
for the Wandering Albatross, nests were most
likely to be constructed in areas of intermediate
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
6M. Momberg et al.
wind speeds. This is probably because this species
needs strong enough wind speeds to take-off and
land (making very wind-sheltered locations unsuit-
able), but also benefits from protection from the
strongest wind speeds, both for thermoregulation
and for chicks not to be blown off their nests. This
influence of flight-related wind preferences
impacting nest-site selection is in agreement with
those for a cliff-nesting seabird species on much
smaller islands, where Common Guillemots are
only able to land in very low wind speeds, and
therefore nest in more sheltered locations (Shep-
ard et al.2019, Lempidakis et al.2022).
In addition to wind, other predictors explained
even more of the variation in nest-site selection.
Elevation was the most important predictor of
Wandering Albatross nests, with the probability of
encountering a nest decreasing rapidly above c.
25 m asl. Similar results have been seen for
surface-nesting species in the Arctic, where low-
elevation areas are important for 33 species of
tundra-breeding birds (Hawkshaw et al.2021).
There is a strong negative correlation between ele-
vation and temperature on Marion Island (Leihy
et al.2018), so lower elevations are warmer sites,
protecting chicks from very low temperatures
(although Wandering Albatross chicks are well
insulated against the cold; Cooper & Lutje-
harms 1992). Similarly, the probability of nests
being present declined with distance from the
coast (which could have impacts on thermal
buffering), which typically correlates with eleva-
tion. This result is comparable to several studies
on Arctic birds, where higher numbers of birds are
present in coastal habitats, probably as a result of
the larger amounts of suitable habitat (e.g. wet-
lands or tidal habitats) available in these areas
(Conkin & Alisauskas 2013, Saalfeld et al.2013,
Hawkshaw et al.2021).
Vegetation type was the third most important
predictor, and co-varies with elevation and dis-
tance from the coast, because some vegetation
types are limited to areas receiving salt-spray (i.e.
coastal vegetation) and others are limited to high
altitudes (e.g. polar desert; Smith 1978b). High
vegetation productivity (which on Marion Island
declines with increasing elevation; Smith 2008)
has previously also been linked to tundra bird
abundance, as some birds use the vegetation cover
for nesting, and others for foraging (Hawkshaw
et al.2021). The composition of vegetation sur-
rounding the nest was also an important determi-
nant of nesting site choice for several Arctic-
breeding shorebirds, probably because of predator
protection and invertebrate food sources (Cun-
ningham et al.2016), and adequate vegetation is
needed to construct nests. Wandering Albatrosses
may prefer low-elevation, coastal areas for nesting
because these areas are warmer, and there is ample
vegetation available with which to construct their
nests.
Terrain roughness was also significantly related
to nest occurrence, with areas that have a higher
roughness having a lower chance of containing a
Wandering Albatross nest, in line with our hypoth-
esis. Terrain roughness has been shown to be an
Table 1. Significance and variable importance for all variables when predicting the presence or absence of a Wandering Albatross
nest based on data from the 2018 breeding season
Predictor
GAM GLM
v
2
Pvalue Relative importance (%) v
2a
Pvalue Relative importance (%)
Elevation 478.60 <0.001 46.87 21.6 0.005
b
42.94
Distance to coast 249.12 <0.001 24.05 434.8 <0.001
b
24.18
Vegetation type P <F<M<C 16.15 356.79 P <F<M<C 15.79
Terrain roughness 127.30 <0.001 6.37 54.58 0.003
b
8.69
Wind velocity 73.45 <0.001 5.89 160.66 <0.001
b
7.73
Geology type Post <Pre 0.36 4.14 Post <Pre 0.23
Wind turbulence 2.54 0.078 0.30 2.91 0.34 0.44
GAM, generalized additive model; GLM, generalized linear model. The % deviance explained was 34.30% for the GAM and 32.85%
for the GLM. An overall Pvalue for categorical predictors was not reported from a GAM, so the ranking of the levels is reported. v
2
values are not reported for categorical predictors in a GAM. Post—post-glacial flows, Pre—pre-glacial flows, P—sub-Antarctic polar
desert, F—sub-Antarctic fellfield, M—sub-Antarctic mire-slope vegetation, C—sub-Antarctic coastal vegetation.
a
v
2
values for linear
and quadratic terms of a predictor in the GLM were summed.
b
Quadratic term of that variable was significant in the GLM.
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
Nest-site selection of Wandering Albatross 7
important driver of nest-site selection in other sys-
tems, where birds prefer to nest in areas with low
terrain roughness in order to provide individuals
with a greater field of view to detect predators
(Korne et al.2020). As Wandering Albatrosses
have a high visibility in the landscape (because
they are taller than almost all vegetation on the
island), the mechanism through which this terrain
roughness affects nest-site selection is probably
different (although predation by invasive House
Mice Mus musculus has been recorded in recent
years; Jones & Ryan 2010, Jones et al.2019).
Wandering Albatrosses need adequate flat space
(i.e. low terrain roughness) during take-off and
landing, because of their large size requiring longer
‘runways’to achieve adequate speed before taking
flight and to land safely (Warham 1977). For some
other surface-nesting species, (micro-)relief can
Figure 2. Stacked bar charts showing the proportion of nest presences and absences in each (a) vegetation type and (b) geology
type. In (a), sub-Antarctic coastal vegetation—dark blue, sub-Antarctic fellfield—live green, sub-Antarctic mire-slope vegetation—or-
ange, sub-Antarctic polar desert—light pink. In (b), post-glacial flows—dark blue, pre-glacial flows—light pink.
Figure 3. Density plots of the raw data and generalized additive model response curves for the occurrence of Wandering Albatross
nests for the 2017 breeding season, for (a) elevation, (b) terrain ruggedness index (logged), (c) distance to the coast, (d) wind turbu-
lence (logged) and (e) wind velocity (m/s). Light pink density plots represent data from presences and dark blue density plots repre-
sent data from absences. In the response curves, larger values on the y-axis represent a higher probability of occurrence.
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
8M. Momberg et al.
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
Nest-site selection of Wandering Albatross 9
provide wind shelter for nesting birds, helping
them to avoid excessive heat loss in windy condi-
tions in the Arctic (Cunningham et al.2016).
However, for species that have chicks that stay on
the nest throughout winter, these microsites could
also allow for greater snow accumulation, which
might offset the benefits that these sites provide in
terms of wind shelter. This is probably also true
for Northern Giant Petrels Macronectes halli,
which nest adjacent to rocks or on the leeward
side of vegetation, where they are sheltered from
wind (Marchant et al.1990). Here, however, we
observe the opposite pattern for Wandering Alba-
tross, where it appears that shelter is not as impor-
tant as potential runway area. Topography and
wind may be strongly linked at certain spatial
scales, and although there is not a strong correla-
tion between terrain roughness and wind velocity
in this study, this relationship may change when
investigated at different spatial scales. This result
may not be universal and may depend on site-
specific characteristics and/or spatial scale.
Wind turbulence and geology had weak impacts
on the probability of a nest occurring at a site.
Wind turbulence could have a limited effect on
nest-site locations because these birds only nest
just above ground level, and wind speed is lower
at ground level, implying that wind turbulence, or
‘gusting’occurs from a low underlying speed
value. Nonetheless, the observed negative trend
between turbulence and nest occurrence fits our
expectation that Wandering Albatrosses would
avoid areas of turbulent wind flow, as these areas
may increase the risks of crashing during take-off
or landing (although the influence of wind condi-
tions on landing may need to be considered at
broader scales because of the large distances, span-
ning different heights above the ground, required
for landing). In terms of geology, pre-glacial depos-
its tend to be flatter and smoother, and, therefore,
meet the requirements for long, flat ‘runways’
more closely than post-glacial flows. Anecdotally,
this can be seen on the west coast of the island,
where there are fewer nests, corresponding to large
black lava flows. However, many of the post-
glacial flows at low altitudes are vegetated and
often occur under peat deposits, which evens out
the underlying roughness, and leads to fewer bio-
logically relevant differences for the Wandering
Albatrosses between geology types. Pre-glacial
deposits tend to have relatively less vegetation, but
include depressions that are filled with peat. These
areas with peat deposits (regardless of the underly-
ing geology) provide sufficient vegetation for Wan-
dering Albatrosses to build their nests, suggesting
that surface substrate may be more important than
underlying geology in influencing nest-site selec-
tion in this species.
Predicting how nest-site availability might
change under future conditions would probably be
most dependent on vegetation and wind character-
istics (wind speed, and possibly wind direction and
wind turbulence), both of which are currently
being affected by anthropogenically driven climate
change. Changes to vegetation in relation to cli-
matic changes have already been documented,
with some species in the sub-Antarctic showing
strong upslope range expansion, leading to com-
munity reorganization in some areas (le Roux &
McGeoch 2008b), and others showing decreased
survival because of increasing temperatures and
lower precipitation (le Roux et al.2005). These
changes will affect the distribution of entire vege-
tation types and may, for example, increase the
availability of vegetation for nest building at higher
altitudes (improving the suitability of more inland
areas as nesting sites). Wind speeds have increased
globally over the past three decades, with the
strongest increases observed in the Southern
Ocean (Young et al.2011, Young & Ribal 2019).
As nest-site selection is influenced by wind veloc-
ity, anthropogenically driven shifts in climate
could potentially affect the total suitable nesting
area and, consequently, potentially the popula-
tion’s total breeding success via changes in wind
characteristics. Small changes to wind velocity
could create more suitable nesting sites for Wan-
dering Albatrosses, but large increases may cause
wind velocities to be too high for the birds to reli-
ably land at nest-sites, thereby possibly leading to
a reduction in potential suitable nest-sites. A factor
that has not been quantified here, but that could
potentially also have large impacts on suitable
nest-sites and breeding success, is changes to the
frequency of extreme wind events which can, for
example, blow chicks off their nests.
Elevation and temperature are strongly corre-
lated on Marion Island (Leihy et al.2018), and
elevation had the strongest influence on nest
occurrence, so it is possible that changes to tem-
perature could have a large effect on suitable nest
locations. The sub-Antarctic islands, where Wan-
dering Albatrosses and many other pelagic seabirds
breed, have already experienced rapid climatic
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
10 M. Momberg et al.
changes (le Roux 2008). Temperature has
increased at more than double the global average
warming rate over the past 50 years, and annual
precipitation has declined on several islands (le
Roux 2008, le Roux & McGeoch 2008a). This
change in temperature, specifically, might nega-
tively affect Wandering Albatrosses, where high
sea surface temperatures are detrimental to forag-
ing success, and therefore adult survival in the
breeding season (Pardo et al.2017, see also Ven-
tura et al.2021 where increased sea surface tem-
peratures led to higher divorce rates in Black-
browed Albatrosses Thalassarche melanophris),
which could alter where they nest in future. Alti-
tudinal shifts in nesting sites could be a possibility,
both as a result of increasing temperatures and the
shift in vegetation as a result of temperature
changes.
Other surface-nesting seabirds that breed in the
sub-Antarctic, such as giant petrels (Ryan &
Bester 2008), are likely to show similar patterns
and experience analogous changes to suitable nest-
ing locations in future. More generally, several
other seabird species, for example skuas, shags,
gulls and terns, all construct nests on the ground
surface and occur in environments where wind
speeds are relatively high (because of being close
to the open ocean, Possner & Caldeira 2017,
Schrimpf & Lynch 2021). Therefore, these results
could provide insights into where surface-nesting
seabirds nest in general, and how the availability
of these sites will be affected by future climatic
changes. More broadly, this work provides insights
into wind as an underexplored climatic component
of nest-site selection for surface-nesting seabirds,
and is important for improving our predictions for
climate change impacts on bird nesting habitat.
We thank Christiaan Brink, Quentin Hagens, Chris
Jones, Genevieve Jones, Michelle Risi and Kim Stevens
for collecting the geographical coordinates of the Wan-
dering Albatross nests on Marion Island.
AUTHOR CONTRIBUTIONS
Mia Momberg: Conceptualization (equal); formal
analysis (lead); methodology (equal); writing –
original draft (lead). Peter G. Ryan: Methodology
(equal); resources (equal); writing –review and
editing (equal). David W. Hedding: Methodology
(equal); writing –review and editing (equal).
Janine Schoombie: Resources (equal); writing –
review and editing (equal). Kyle A. Goddard:
Resources (equal); writing –review and editing
(equal). Ken J. Craig: Resources (equal); writing –
review and editing (equal). Peter C. le Roux: Con-
ceptualization (equal); formal analysis (support-
ing); funding acquisition (lead); methodology
(equal); writing –review and editing (equal).
ETHICAL NOTE
None.
FUNDING
This research was supported by the National
Research Foundation’s South African National
Antarctic Programme (grant number 110726) and
was conducted under permits from the Prince
Edward Islands Management Committee
(PEIMC1/2013). M.M. received funding from the
South African National Research Foundation.
Data Availability Statement
Data are available from the corresponding author
upon reasonable request.
REFERENCES
ACAP. 2009. Agreement on the conservation of albatrosses
and petrels species assessment: Wandering Albatross
Diomedea exulans. http://www.acap.aq
Betts, M., Hadley, A., Rodenhouse, N. & Nocera, J. 2008.
Social information trumps vegetation structure in breeding-
site selection by a migrant songbird. Proc. R. Soc. Lond. B
275: 2257–2263.
Birdlife International. 2018. Diomedea exulans. The IUCN
Red List of Threatened Species 2018:
e.T22698305A132640680. https://doi.org/10.2305/IUCN.UK.
2018-2.RLTS.T22698305A132640680.en
Boelhouwers, J.C., Meiklejohn, K.I., Holness, S.D. &
Hedding, D.W. 2008. Geology, geomorphology and climate
change. In Chown, S.L. & Froneman, P.W. (eds) The Prince
Edward Islands: Land-sea interactions in a changing
ecosystem (pp. 65–96). Stellenbosch: SUN Press.
Breedt, H.J., Craig, K.J. & Jothiprakasam, V.D. 2018.
Monin-Obukhov similarity theory and its application to wind
flow modelling over complex terrain. J. Wind Eng. Ind.
Aerodyn. 182: 308–321.
Cestari, C. & de Melo, C. 2022. Shorebirds adjust resting
orientation in response to solar radiation and wind speed.
Austral Ecol. 47: 765–769.
Cindori, M., Jureti
c, F., Kozmar, H. & D
zijan, I. 2018.
Steady RANS model of the homogeneous atmospheric
boundary layer. J. Wind Eng. Ind. Aerodyn. 173: 289–301.
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
Nest-site selection of Wandering Albatross 11
Clay, T.A., Joo, R., Weimerskirch, H., Phillips, R.A., den
Ouden, O., Basille, M., Clusella-Trullas, S., Assink, J.D.
& Patrick, S.C. 2020. Sex-specific effects of wind on the
flight decisions of a sexually dimorphic soaring bird. J. Anim.
Ecol. 89: 1811–1823.
Colwell, M.A., Meyer, J.J., Hardy, M.A., McAllister, S.E.,
Transou, A.N., Levalley, R.R. & Dinsmore, S.J. 2011.
Western Snowy Plovers Charadrius alexandrinus nivosus
select nesting substrates that enhance egg crypsis and
improve nest survival. Ibis 153: 303–311.
Conkin, J.A. & Alisauskas, R.T. 2013. Modeling probability of
waterfowl encounters from satellite imagery of habitat in the
central Canadian arctic. J. Wildl. Manag. 77: 931–946.
Cooper, J. & Lutjeharms, J.R.E. 1992. Correlations between
seabird breeding success and meteorological conditions on
Marion and Gough Islands. S. Afr. J. Sci. 88: 173–175.
Cornioley, T., B€
orger, L., Ozgul, A. & Weimerskirch, H.
2016. Impact of changing wind conditions on foraging and
incubation success in male and female wandering
albatrosses. J. Anim. Ecol. 85: 1318–1327.
Crameri, F. 2018. Scientific colour maps. Zenodo. https://doi.
org/10.5281/zenodo.1243862
Crameri, F., Shephard, G.E. & Heron, P.J. 2020. The misuse
of colour in science communication. Nat. Commun. 11:
5444.
Cunningham, J.A., Kesler, D.C. & Lanctot, R.B. 2016.
Habitat and social factors influence nest-site selection in
Arctic-breeding shorebirds. Auk 133: 364–377.
Dilley, B., Davies, D., Connan, M., Cooper, J., de Villiers,
M., Swart, L., Vandenabeele, S., Ropert-Coudert, Y. &
Ryan, P. 2013. Giant petrels as predators of albatross
chicks. Polar Biol. 36: 761–766.
DRDLR. 2019. Digital elevation model –Marion Island.
Directorate National Geo-spatial Information, Department of
Rural Development and Land Reform, Republic of South
Africa.
Flemming, S.A., Nol, E., Kennedy, L.V. & Smith, P.A. 2019.
Hyperabundant herbivores limit habitat availability and
influence nest site selection of Arctic-breeding birds. J. Appl.
Ecol. 56: 976–987.
Garcia de la Garza, A., Vandekar, S., Roalf, D., Ruparel, K.,
Gur, R., Gur, R., Satterthwaite, T. & Shinohara, R.T.
2018. voxel: Mass-Univariate Voxelwise Analysis of Medical
Imaging Data. R package version 1.3.5. https://CRAN.R-
project.org/package=voxel
Gaston, A.J. 2004. Seabirds: A Natural History. New Haven,
CT: Yale University Press.
Gauthier, G., Milot, E. & Weimerskirch, H. 2010. Small-scale
dispersal and survival in a long-lived seabird, the wandering
albatross. J. Anim. Ecol. 79: 879–887.
Gilchrist, H.G. & Gaston, A.J. 1997. Effects of murre nest
site characteristics and wind conditions on predation by
glaucous gulls. Can. J. Zool. 75: 518–524.
Goddard, K.A., Schoombie, J., Craig, K.J. & le Roux, P.C.
2022. Investigation of ecologically relevant wind patterns on
Marion Island using Computational Fluid Dynamics and
measured data. Ecol. Model. 464: 109827.
Gray, L.A. & Deeming, D.C. 2017. Effect of air movement
on the thermal insulation of avian nests. Bird Study 64:
492–501.
Guillera-Arroita, G., Lahoz-Monfort, J.J., Elith, J., Gordon,
A., Kujala, H., Lentini, P.E., McCarthy, M.A., Tingley, R. &
Wintle, B.A. 2015. Is my species distribution model fit for
purpose? Matching data and models to applications. Glob.
Ecol. Biogeogr. 24: 276–292.
Hawkshaw, K.A., Foote, L. & Franke, A. 2021. Ecological
determinants of avian distribution and abundance at Rankin
Inlet, Nunavut in the Canadian Arctic. Polar Biol. 44:1–15.
Hazen, E.L., Abrahms, B., Brodie, S., Carroll, G., Welch, H.
& Bograd, S.J. 2021. Where did they not go?
Considerations for generating pseudo-absences for
telemetry-based habitat models. Mov. Ecol. 9:5.
Heenan, C.B. & Seymour, R.S. 2012. The effect of wind on
the rate of heat loss from avian cup-shaped nests. PloS
One 7: e32252.
Holmes, G., Koloski, L. & Nol, E. 2020. Nest-site selection of
a sub-Arctic breeding shorebird: Evidence for tree avoidance
without fitness consequences. Can. J. Zool. 98: 573–580.
Holness, S.D. 2004. Sediment movement rates and
processes on cinder cones in the maritime Subantarctic
(Marion Island). Earth Surf. Process. Landf. 29:91–103.
Inchausti, P. & Weimerskirch, H. 2002. Dispersal and
metapopulation dynamics of an oceanic seabird, the
wandering albatross, and its consequences for its response
to long-line fisheries. J. Anim. Ecol. 71: 765–770.
Joly, Y., Frenot, Y. & Vernon, P. 1987. Environmental
modifications of a subantarctic peat-bog by the Wandering
Albatross (Diomedea exulans): A preliminary study. Polar
Biol. 8:61–72.
Jones, J. 2001. Habitat selection studies in avian ecology: A
critical review. Auk 118: 557–562.
Jones, M.G.W. & Ryan, P.G. 2010. Evidence of mouse
attacks on albatross chicks on sub-Antarctic Marion Island.
Antarct. Sci. 22:39–42.
Jones, C.W., Risi, M.M., Cleeland, J. & Ryan, P.G. 2019.
First evidence of mouse attacks on adult albatrosses and
petrels breeding on sub-Antarctic Marion and Gough
Islands. Polar Biol. 42: 619–623.
Kassambara, A. 2020. ggpubr: ’ggplot2’Based Publication
Ready Plots R package version 0.4.0. https://CRAN.R-
project.org/package=ggpubr
Kauffman, K.L., Elmore, R.D., Davis, C.A., Fuhlendorf, S.D.,
Goodman, L.E., Hagen, C.A. & Tanner, E.P. 2021. Role of
the thermal environment in scaled quail (Callipepla
squamata) nest site selection and survival. J. Therm. Biol.
95: 102791.
Kemppinen, J., Niittynen, P., le Roux, P.C., Momberg, M.,
Happonen, K., Aalto, J., Rautakoski, H., Enquist, B.J.,
Vandvik, V., Halbritter, A.H., Maitner, B. & Luoto, M.
2021. Consistent trait–environment relationships within and
across tundra plant communities. Nat. Ecol. Evol. 5: 458–
467.
Korne, N., Flemming, S., Smith, P. & Nol, E. 2020. Applying
structure-from-motion habitat reconstruction and GIS terrain
analysis to test hypotheses about nest site selection by
shorebirds. J. Field Ornithol. 91: 421–432.
Leihy, R.I., Duffy, G.A., Nortje, E. & Chown, S.L. 2018. High
resolution temperature data for ecological research and
management on the Southern Ocean Islands. Sci. Data 5:
180177.
Lempidakis, E., Ross, A.N., B€
orger, L. & Shepard, E.L.C.
2022. Airflow modelling predicts seabird breeding habitat
across islands. Ecography 2022 https://doi.org/10.1111/
ecog.05733
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
12 M. Momberg et al.
Marchant, S., Higgins, P.J., Ambrose, S.J. & Royal
Australasian Ornithologists’Union 1990. Handbook of
Australian, New Zealand & Antarctic birds. Volume 1,
Ratites to ducks. Melbourne: Oxford University Press.
McCollin, D. 1998. Forest edges and habitat selection in
birds: A functional approach. Ecography 21: 247–260.
Miller, V., Nol, E., Nguyen, L. & Turner, D. 2014. Habitat
selection and nest success of the Upland Sandpiper
(Bartramia longicauda) in Ivvavik National Park, Yukon,
Canada. Can. Field-Nat. 128: 341–349.
Muir, J.J. & Colwell, M.A. 2010. Snowy plovers select open
habitats for courtship scrapes and nests. Condor 112: 507–
510.
Nel, D.C., Ryan, P.G., Crawford, R.J., Cooper, J. & Huyser,
O.A. 2002. Population trends of albatrosses and petrels at
sub-Antarctic Marion Island. Polar Biol. 25:81–89.
Niittynen, P. & Luoto, M. 2018. The importance of snow in
species distribution models of arctic vegetation. Ecography
41: 1024–1037.
Pardo, D., Forcada, J., Wood, A.G., Tuck, G.N., Ireland, L.,
Pradel, R., Croxall, J.P. & Phillips, R.A. 2017. Additive
effects of climate and fisheries drive ongoing declines in
multiple albatross species. Proc. Natl Acad. Sci. U.S.A. 114:
E10829–E10837.
Pedersen, T.L. & Crameri, F. 2020. scico: Colour palettes
based on the scientific colour-maps. R package version
1.3.0. https://CRAN.R-project.org/package=scico
Poncet, S., Wolfaardt, A.C., Black, A., Browning, S.,
Lawton, K., Lee, J., Passfield, K., Strange, G. & Phillips,
R.A. 2017. Recent trends in numbers of wandering
(Diomedea exulans), black-browed (Thalassarche
melanophris) and grey-headed (T. chrysostoma) albatrosses
breeding at South Georgia. Polar Biol. 40: 1347–1358.
Possner, A. & Caldeira, K. 2017. Geophysical potential for
wind energy over the open oceans. Proc. Natl Acad. Sci.
U.S.A. 114: 11338.
R Core Team 2021. R: A language and environment for
statistical computing. Vienna: R Foundation for Statistical
Computing.
Rackete, C., Poncet, S., Good, S.D., Phillips, R.A.,
Passfield, K. & Trathan, P. 2021. Variation among colonies
in breeding success and population trajectories of
Wandering Albatrosses Diomedea exulans at South
Georgia. Polar Biol. 44: 221–227.
le Roux, P.C. 2008. Climate and climate change. In Chown,
S.L. & Froneman, P.W. (eds) The Prince Edward Islands:
Land-sea interactions in a changing ecosystem (pp. 39–64).
Stellenbosch: SUN Press.
le Roux, P.C. & McGeoch, M. 2008a. Changes in climate
extremes, variability and signature on sub-Antarctic Marion
Island. Clim. Change 86: 309–329.
le Roux, P.C. & McGeoch, M. 2008b. Rapid range expansion
and community reorganization in response to warming.
Glob. Chang. Biol. 14: 2950–2962.
le Roux, P.C., McGeoch, M.A., Nyakatya, M.J. & Chown,
S.L. 2005. Effects of a short-term climate change
experiment on a sub-Antarctic keystone plant species. Glob.
Chang. Biol. 11: 1628–1639.
Rudolph, E., Hedding, D. & Nel, W. 2021. The surface
geology of the Prince Edward Islands: Refined spatial data
and call for geoconservation. South Afr. J. Geol. 124: 627–
638.
Ryan, P.G. & Bester, M.N. 2008. Pelagic predators. In
Chown, S.L. & Froneman, P.W. (eds) The Prince Edward
Islands: Land-sea interactions in a changing ecosystem (pp.
121–164). Stellenbosch: SUN Press.
Saalfeld, S.T., Lanctot, R.B., Brown, S.C., Saalfeld, D.T.,
Johnson, J.A., Andres, B.A. & Bart, J.R. 2013. Predicting
breeding shorebird distributions on the Arctic Coastal Plain
of Alaska. Ecosphere 4: art16.
Sauve, D., Friesen, V.L. & Charmantier, A. 2021. The effects
of weather on avian growth and implications for adaptation
to climate change. Front. Ecol. Evol. 9https://doi.org/10.
3389/fevo.2021.569741
Schrimpf, M. & Lynch, H. 2021. The role of wind fetch in
structuring Antarctic seabird breeding occupancy. Ibis 163:
695–705. https://doi.org/10.1111/ibi.12910
Shepard, E., Cole, E.-L., Neate, A., Lempidakis, E. & Ross,
A. 2019. Wind prevents cliff-breeding birds from accessing
nests through loss of flight control. eLife 8: e43842.
Smith, V.R. 1978a. Soil chemistry of Marion Island (sub-
Antarctic). S. Afr. J. Sci. 74: 174–175.
Smith, V.R. 1978b. Plant ecology of Marion Island: A review.
S. Afr. J. Antarct. Res. 8:21–30.
Smith, V.R. 2008. Terrestrial and freshwater primary
production and nutrient cycling. In Chown, S.L. & Froneman,
P.W. (eds) The Prince Edward Islands: land-sea interactions
in a changing ecosystem (pp. 181–214). Stellenbosch: SUN
Press.
Smith, V.R. & Mucina, L. 2006. Vegetation of Subantarctic
Marion and Prince Edward Islands. In Mucina, L. &
Rutherford, M.C. (eds) The vegetation of South Africa,
Lesotho and Swaziland (pp. 698–723). Pretoria: South
African National Biodiversity Institute.
Tickell, W.L.N. 2000. Albatrosses. Sussex: Pica Press.
Ventura, F., Granadeiro, J.P., Lukacs, P.M., Kuepfer, A. &
Catry, P. 2021. Environmental variability directly affects the
prevalence of divorce in monogamous albatrosses. Proc. R.
Soc. Lond. B 288: 20212112.
Versteeg, H.K. & Malalasekera, W. 2007. An introduction to
computational fluid dynamics: The finite volume method.
Harlow: Pearson Education Limited.
Verwoerd, W.J. 1971. Geology. In Van Zinderen Bakker,
E.M., Winterbottom, J.M. & Dyer, R.A. (eds) Marion and
Prince Edward Islands: Report on the South African
Biological and Geological Research Expedition 1965–1966
(pp. 40–53). A. A. Balkema: Cape Town.
Warham, J. 1977. Wing loadings, wing shapes, and flight
capabilities of Procellariiformes. N. Z. J. Zool. 4:73–83.
Weimerskirch, H., Guionnet, T., Martin, J., Shaffer, S.A. &
Costa, D.P. 2000. Fast and fuel efficient? Optimal use of
wind by flying albatrosses. Proc. Biol. Sci. 267: 1869–1874.
Weimerskirch, H., Louzao, M., de Grissac, S. & Delord, K.
2012. Changes in wind pattern alter albatross distribution
and life-history traits. Science 335: 211–214.
Wickham, H. 2016. ggplot2: Elegant Graphics for Data
Analysis. New York: Springer.
With, K.A. & Webb, D.R. 1993. Microclimate of ground nests:
The relative importance of radiative cover and wind breaks
for three grassland species. Condor 95: 401–413.
Wood, S.N. & Augustin, N.H. 2002. GAMs with integrated
model selection using penalized regression splines and
applications to environmental modelling. Ecol. Model. 157:
157–177.
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
Nest-site selection of Wandering Albatross 13
Young, I.R. & Ribal, A. 2019. Multiplatform evaluation of
global trends in wind speed and wave height. Science 364:
548–552.
Young, I.R., Zieger, S. & Babanin, A.V. 2011. Global trends
in wind speed and wave height. Science 332: 451–455.
Received 21 January 2022;
revision accepted 5 July 2022.
Associate Editor: Richard Phillips.
SUPPORTING INFORMATION
Additional supporting information may be found
online in the Supporting Information section at
the end of the article.
Figure S1. Wind roses from four wind stations
located on the north, east, west and south sides of
Marion Island.
Figure S2. Locations of Wandering Albatross
nests on Marion Island in January 2017.
Figure S3. Locations of absence points gener-
ated on Marion Island based on the nest locations
from January 2017.
Figure S4. Digital Surface Model of Marion
Island.
Figure S5. Terrain Ruggedness Index values
across Marion Island.
Figure S6. Wind velocity on Marion Island.
Figure S7. Wind turbulence intensity on Marion
Island.
Figure S8. Correlation matrix between continu-
ous predictor variables.
Figure S9. Generalized additive model response
curves for the occurrence of Wandering Albatross
nests based on data from the 2006 breeding sea-
son.
Figure S10. Generalized additive model
response curves for the predicted presence of
Wandering Albatross nests based on data from the
2018 breeding season.
Table S1. Significance and variable importance
for all predictor variables when modelling the
presence or absence of Wandering Albatross nests
based on data from the 2006 breeding season.
Table S2. Significance and variable importance
for all variables when predicting the presence or
absence of a Wandering Albatross nest based on
data from the 2018 breeding season.
Table S3. Significance and variable importance
for all predictor variables when modelling the
presence or absence of Wandering Albatross nests
based on spatially thinned data from the 2017
breeding season.
© 2022 The Authors. Ibis published by John Wiley & Sons Ltd on behalf of British Ornithologists' Union.
14 M. Momberg et al.
