ArticlePDF Available

Using spatial analyses of bearded vulture movements in southern Africa to inform wind turbine placement


Abstract and Figures

1. Concerns over CO 2 emissions during energy generation and its effect on climate change have led to increases in the use of renewables, such as wind energy. However, there are also serious environmental concerns over this type of energy production due to its impacts on bats and birds. 2. In southern Africa, bearded vultures have declined by >30% during recent decades. They are now regionally critically endangered with only around 100 active pairs remaining. This species is considered vulnerable to collision with wind turbines which are planned within their southern African range. 3. In this study, we develop habitat use models using data obtained from 21 bearded vultures of different ages fitted with GPS tags from 2009 to 2013. We further refined these models by incorporating flying heights at risk of collision to predict important areas of use that may conflict with wind turbines. 4. Adult and non-adult bearded vultures mostly used areas with high elevations and steep and rugged topography in the core area; adults tended to use areas in relatively close proximity to their nest sites, whereas non-adult birds used areas dispersed over the entire species range and were more likely to fly at risk-height in areas that were less used by adults. Altitudes of fixes of adults and non-adults showed that they spent 55% and 66% of their time, respectively, at heights that placed them at risk of collision. 5. Examining the locations of two proposed wind farms in relation to our model of predicted 'at risk' usage suggested poor positioning. Indeed, one of these wind farms was located within the 1% of 'worst' (most heavily used) sites for non-adult bearded vultures suggesting that its current location should be reconsidered to reduce the impact on this vulnerable species. 6. Synthesis and applications. We demonstrate the value of habitat use models for identifying intensively used areas, in order to greatly reduce conflicts with developments such as wind turbines. This tool is operable at the scale of regional and national development plans informed by the habitat use of potentially vulnerable species. Such models should provide important supplementary assessments of site-specific development proposals.
Content may be subject to copyright.
Using spatial analyses of bearded vulture movements
in southern Africa to inform wind turbine placement
Tim Reid
, Sonja Kr
, D. Philip Whitfield
and Arjun Amar
Percy FitzPatrick Institute of African Ornithology, DST/NRF Centre of Excellence, University of Cape Town,
Rondebosch 7701, South Africa;
Ezemvelo KZN Wildlife, P.O. Box 13053, Cascades 3202, South Africa; and
Natural Research, Brathens Business Park, Banchory, Aberdeenshire AB31 4BY, UK
1. Concerns over CO
emissions during energy generation and its effect on climate change
have led to increases in the use of renewables, such as wind energy. However, there are also
serious environmental concerns over this type of energy production due to its impacts on bats
and birds.
2. In southern Africa, bearded vultures have declined by >30% during recent decades. They
are now regionally critically endangered with only around 100 active pairs remaining. This
species is considered vulnerable to collision with wind turbines which are planned within their
southern African range.
3. In this study, we develop habitat use models using data obtained from 21 bearded vultures
of different ages fitted with GPS tags from 2009 to 2013. We further refined these models by
incorporating flying heights at risk of collision to predict important areas of use that may
conflict with wind turbines.
4. Adult and non-adult bearded vultures mostly used areas with high elevations and steep
and rugged topography in the core area; adults tended to use areas in relatively close proxim-
ity to their nest sites, whereas non-adult birds used areas dispersed over the entire species
range and were more likely to fly at risk-height in areas that were less used by adults. Alti-
tudes of fixes of adults and non-adults showed that they spent 55% and 66% of their time,
respectively, at heights that placed them at risk of collision.
5. Examining the locations of two proposed wind farms in relation to our model of predicted
‘at risk’ usage suggested poor positioning. Indeed, one of these wind farms was located within
the 1% of ‘worst’ (most heavily used) sites for non-adult bearded vultures suggesting that its
current location should be reconsidered to reduce the impact on this vulnerable species.
6. Synthesis and applications. We demonstrate the value of habitat use models for identifying
intensively used areas, in order to greatly reduce conflicts with developments such as wind
turbines. This tool is operable at the scale of regional and national development plans
informed by the habitat use of potentially vulnerable species. Such models should provide
important supplementary assessments of site-specific development proposals.
Key-words: collision, conservation management, habitat use model, predicted effect, threat,
tracking, wind energy
Throughout the world, renewable energy production is a
rapidly expanding industry, driven by climate change and
political targets for reduced CO
emissions, together with
increasing energy demands (Lu, McElroy & Kiviluoma
2009; Leung & Yang 2012; Tabassum-Abbasi, Abbasi &
Abbasi 2014). Africa reflects this global trend: central to
development goals throughout Africa is the need for
greater energy production, particularly for reliable, low
, and affordable-energy supplies. Wind energy can
provide such energy, and in many countries across Africa,
wind farms are planned or are already under construction
(BirdLife International 2013; Nemaxwi 2013). Although
wind energy has the advantage of being a relatively estab-
lished energy source, experience elsewhere has shown that
*Correspondence author. E-mail:
©2015 The Authors. Journal of Applied Ecology ©2015 British Ecological Society
Journal of Applied Ecology 2015 doi: 10.1111/1365-2664.12468
inappropriately situated wind farms can have severe envi-
ronmental consequences, killing many bats and birds, in
particular large raptors and vulture species through colli-
sion with the turning blades. In some extreme cases, these
collisions have led to numerous deaths of vultures and
large raptors and may potentially jeopardize the existence
of local or regional populations (Hunt et al. 1999; Hunt
2002; Drewitt & Langston 2006; Carrete et al. 2009; Dahl
et al. 2012; Mart
ınez-Abrain et al. 2012; Bellebaum et al.
The best way to minimize wind farm impacts through
collision mortality is to ensure that they are placed away
from the nesting, roosting or intensively used foraging
areas of these vulnerable species (Madders & Whitfield
2006; Drewitt & Langston 2006). This can be achieved in
two ways, each requiring different types of data at differ-
ent spatial scales. Firstly, ensuring wind farms are not
developed in areas where vulnerable species occur. To do
this, the distribution of these vulnerable species must be
known. This information can then be used to build wind
farm sensitivity maps, highlighting the best and the worst
locations at a broad scale of where to place wind farms
(Bright et al. 2008). Such an exercise has now been suc-
cessfully completed for South Africa (BirdLife South
Africa 2012) and should help in the strategic planning of
areas where wind farm developments are likely to be least
damaging. The drawback of this approach is that the spa-
tial scale of the data on bird distributions is at a coarser
grain than the scale at which individual wind farms are
planned. Moreover, such exercises, if based on several
species’ distributions, may also not provide reliable indices
of collision mortality that are later realized in practice
(Ferrer et al. 2012).
The second approach operates at a finer scale and aims
to ensure that where wind farms and vulnerable species
overlap at broad spatial scales (Fielding, Whitfield &
McLeod 2006; Teller
ıa 2009), wind turbines are situated
in the most appropriate locations, thereby decreasing the
risks of collision (Tapia, Dom
ınguez & Rodr
ıguez 2009;
Ferrer et al. 2012). To achieve this, data on how a species
uses its environment are required. Using such data, pre-
dictive models can then be built which should be general-
izable across the species’ range to identify inappropriate
locations for turbine placement (McLeod et al. 2002; Elith
& Leathwick 2009; Hijmans et al. 2013).
In this study, the second approach has been adopted
for a species which has been identified as being vulnerable
to the impacts of wind farms, the bearded vulture Gypa-
etus barbatus (Ferguson-Lees & Christie 2001; Barov &
e 2011; BirdLife South Africa 2012). This species is
classified as Critically Endangered in southern Africa
uger in press), where the entire population occurs in
Lesotho and the surrounding Drakensberg escarpment
and mountains in South Africa (Kr
uger et al. 2014). The
species is estimated to have declined by between 32 and
51% over the last five decades and currently only 109
occupied territories remain (Kr
uger et al. 2014). However,
the species now faces a new threat in the form of exten-
sive wind farms which are planned for the Lesotho high-
lands. Currently, there are two active wind farm
proposals (42 and 100 turbines) within the Lesotho high-
lands, and a few others nearby in South Africa, but there
are also longer term plans to develop multiple wind farms
throughout the Lesotho highland region comprising of up
to 4000 turbines in total with a total generating capacity
of up to 6000 megawatts (MW) (Jenkins & Allan 2013).
Minimizing the impacts of these wind farms on bearded
vultures is therefore vital for this species’ conservation.
This study aims to use the information obtained from
GPS satellite tags attached to bearded vultures to build
predictive models of space use (Elith & Leathwick 2009;
Hijmans et al. 2013). We construct these models using
information on topography, distances to vulture feeding
sites and distances to conspecific nest sites, with different
models being built for different age classes: first, in two
dimensions (landscape models) and then in three dimen-
sions incorporating additional flight height information
from the GPS tags. We then apply these models to the
entire distribution range of the species to identify which
areas of their geographic range are likely to be most fre-
quently used. Finally, we use our three-dimensional model
to consider the risk to bearded vultures associated with
two proposed wind farms in the Lesotho highland region,
as an illustration of the potential of our approach to iden-
tify the risk levels for potential turbine placement.
Materials and methods
Within southern Africa, bearded vultures occur over the Lesotho
highlands and the Drakensburg escarpment on the north-eastern,
eastern and southern border with South Africa and into the East-
ern Cape (Fig. 1). All birds studied for this project were captured
within this area. This area is characterized by high plateaux and
cliffs of basalt and sandstone (Moore & Blenkinsop 2006). All
birds were captured between September 2007 and September
2012 (see Tables S1S3 in Supporting Information) and fitted
with GPS satellite transmitters (70 g solar-powered GPS-PTT-
100s (Microwave Telemetry Inc., Columbia, MD, USA)). These
transmitters provided the location (accurate to within 520 m;
uger, Reid & Amar 2014), the height and the speed of each
bird every hour from 0500 to 2000 h (SAST), thereby starting
and finishing around the hours of sunrise and sunset across the
year, respectively. Between September 2007 and December 2012,
we obtained a total of 31 bird-years of data from 21 birds,
including information on fledglings (n=3), subadults (n=12)
and adults (n=6) (Tables S1S3). Fledglings were classed as
birds that had left the nest up to the end of April within their
first year (at approximately which time they depart the region of
their nest (Kr
uger, Reid & Amar 2014)). Adults were birds that
were known to be breeding. We also used four age classes of
non-adult birds (fledgling, juvenile, immature and subadult),
which have been used previously to classify non-adult bearded
vultures (Brown 1988; Kr
uger, Reid & Amar 2014). Examination
of the distribution of tagged juveniles, immatures and subadults
indicated that they inhabited similar areas (Fig. 2b) and so
were treated together as one group which from herein we term
©2015 The Authors. Journal of Applied Ecology ©2015 British Ecological Society, Journal of Applied Ecology
2T. Reid et al.
‘non-adults’. Data were sufficiently regular and recorded in
enough volume to be assumed to approximate the activity of the
birds to which they were attached. Six adult birds were tracked
for up to 894 days each (Table S1), eleven non-adult birds were
tracked for up to 1243 days (Table S2), and three fledglings were
tracked for up to 170 days (Table S3).
We analysed the habitat use of these birds using species distribu-
tion modelling (Elith & Leathwick 2009). The species’ use of hab-
itats was considered likely to be strongly influenced by distance
from nests (Amar & Redpath 2005; Arroyo et al. 2009) and from
supplementary feeding sites (hereafter called vulture restaurants)
(Anderson & Anthony 2005; Margalida et al. 2010) and by
topography (Pennycuick 1972; Dona
zar, Hiraldo & Bustamente
1993; McLeod et al. 2002; Hirzel et al. 2004; de Lucas, Ferrer &
Janss 2012). Vulture restaurants are locations where land manag-
ers or tourism operators put out carcasses to provide a regular
source of safe food for vultures, to dispose of carcasses or to
attract vultures for tourists. The sites vary in how often (or how
regularly) food is provided. Topographic features were taken
from a digital elevation model derived from the Shuttle Radar
Topography Mission C-Band InSAR 3 arc second (~90-m resolu-
tion grid). We used the topographic height at the location of an
observation (altitude), plus the standard deviation, skewness and
kurtosis over a 5 95 grid of these points, giving a grid over
450 9450 m. A grid of 270 9270 m (i.e. a 3 93 grid of the 90-
m resolution points) was tested but gave a relatively poorer
model fit. Skewness and kurtosis were calculated using the time-
Date package in R (R Core Team 2013; Wuertz et al. 2013).
Kurtosis was centred around zero (Wuertz et al. 2013). Standard
deviation increases when the variation in heights around the
observation is large and so relates to the unevenness of the land-
scape’s topography. Skewness and kurtosis are descriptors of the
similarity of the shape of surrounding heights to a normal distri-
bution. Skewness is a measure of asymmetry in the distribution
of heights around the bird’s location and so relates to areas
where the frequency of surrounding heights taper in one direction
more than in the other. Kurtosis is a measure of how peaked or
flat the heights around the observation are relative to a normal
Fig. 1. The Maloti-Drakensberg region in southern Africa, where
the darker shades indicate higher altitudes; &=>2000 m; =
16002000 m and h=<1600 m + symbols indicate trap sites.
(a) (b)
Fig. 2. Locations of points recorded by all tags attached to bearded vultures. South African border represented by blue line; a. adults
(each colour represents an individual bird); b. non-adult birds. Different colours represent different non-adult ages (juvenile =red, imma-
ture =green and subadult =black).
©2015 The Authors. Journal of Applied Ecology ©2015 British Ecological Society, Journal of Applied Ecology
Bearded vultures and wind turbines 3
distribution; a low value of kurtosis indicates a more uniform
distribution around the value, and a high value indicates an
increasingly narrow peak in distribution with heavy tails.
A significant issue frequently encountered with data on species
habitat use or distribution is that often only presence data are
obtained (Elith & Leathwick 2009), as is the case with data col-
lected from tracking data. While there are ways to model distri-
bution with only presence data (e.g. bioclim; Elith & Leathwick
2009; Hijmans et al. 2013), greater information can be obtained
via background, or pseudo-absence, data. Pseudo-absence data
can be generated by selecting points from the area that could
have been visited by the observed animals, but were apparently
not (Elith & Leathwick 2009; Wakefield et al. 2011). Environ-
mental covariates are then taken at these points and can then be
contrasted with the same measurements taken from the fix loca-
tions of the tracked animals. Points can be chosen either by mod-
elling the likely range via some distribution where a central place
(the nest) is most likely (Wakefield et al. 2011), or by defining an
area under which points are likely to occur (Elith & Leathwick
2009). In this study, we took random pseudo-absence points for
adults from a circle around their nest to the maximum range that
observed data were obtained for that bird, while for subadults,
we selected random points from an area between 2631°E and
2732°S (encompassing the current range of the species in south-
ern Africa). For modelling, a logistic regression was used, where
observed points were given a presence value of one, while
pseudo-absence data were given presence values of zero. Three
times as many pseudo-absence data points were chosen as
observed points in the presented model. This decision was based
on Wakefield et al.’s study (2011) and was validated by running
the model for the adult presence model with a one-to-one ratio
of pseudo-absences and presences. This model’s output was very
similar to the presented three-to-one ratio model, and so we were
confident that our presented model has not altered the position
of the allocation threshold (whether a case is predicted to be 0 or
1, when our model was logistic).
In deploying our presence data, we did not mask our habitat
use classes by any prior removals as we did not wish to poten-
tially reduce the fit of our models by risking false pseudo-
absences if our presence data were unsaturated through inade-
quate sampling of our study subjects. Despite such precaution,
further reassurance on avoiding this risk was given by explor-
atory analyses revealing that within our study area, there was no
habitat class in which bearded vultures were not recorded.
Our initial presence and absence modelling used a generalized
additive mixed model (GAMM) to examine the influence of cova-
riates and whether they were linear (Wood 2006). Individuals
were treated as random effects. Models derived from GAMMs
are more difficult to interpret than those derived from generalized
linear mixed models (GLMMs) and are prone to overfitting,
therefore their simplification may be advantageous (Randin et al.
2006; Wood 2006). In order to do this, plots of the partial residu-
als derived from the GAMMs were examined. Where applicable,
if coefficients of covariates examined this way could be approxi-
mated by a simpler shape such as a straight line or second- or
third-order polynomial, these were used in a GLMM. If the
shape of their effect appeared not to be biologically important,
they were removed. Predictions from these simpler GLMMs were
compared to those from the GAMMs, and if results were similar,
the GLMM was used in preference. If no simplified model could
be found, the initial GAMM was used. GLMMs were used with
individual birds fitted as a random effect. Variance of the random
effects for all models was close to zero, so for predictions, the
random effects were ignored (and so treated as the mean) (Gel-
man et al. 2003). Once a satisfactory model was identified, this
was used to predict the distribution of the birds using the dismo
package in R (Hijmans et al. 2013). Predictions were made at the
scale of the topographic data (90 990 m).
Model fits were evaluated using area under the receiver curves
(AUCs) (Fielding & Bell 1997; Hijmans et al. 2013). AUCs were
fitted by dividing the data into five random sections. A model
was fitted to four of these (training data), and the fifth was used
for predicting distribution (test data). This was then performed
five times, repeatedly subsampling for the training and test data
sets (Hijmans et al. 2013). Additionally, understanding how use-
ful predictive models are can be best achieved by comparing their
predictions to known data that have not been used to create the
model (Fielding & Bell 1997; Hijmans et al. 2013). On this basis,
a model of adult distribution was created using five randomly
chosen adults from the six tracked birds. This model was then
used to predict the likely area used by the sixth tracked bird
which was then compared visually to the actual locations used by
the bird that had not been used in model development. This test-
ing process was also repeated for non-adults.
Once models were developed and validated, they were then
applied to the total study area (defined as the square that con-
tained all of the Drakensburg Escarpment/Lesotho Plateau
2631°E, 2732°S) to predict areas that would be most used by
bearded vultures. This prediction was undertaken using the pack-
age dismo from R (Hijmans et al. 2013). Distance from all
known recently occupied nests (Kr
uger et al. 2014) and vulture
restaurants (Ezemvelo KZN Wildlife and Endangered Wildlife
Trust unpublished data) were used in these predictive models.
We modelled areas at heights where bearded vultures would
potentially interact with wind turbines so-called at risk heights.
We considered heights of <200 m to be dangerous for encounter-
ing turbines. This height was used as this is similar to the height
likely to be used for some wind farms in South Africa (Nemaxwi
2013) and although slightly higher than some of the currently
planned turbines (100 m; Kneidinger 2011), we considered it pru-
dent to use a higher height to account for the trend of increasing
turbine size (Hansen & Hansen 2007). Moreover, Ferrer et al.
(2012) indicated that measures of collision mortality may be
improved if expectations were based on flight activity heights
greater than heights swept by turbine blades. To obtain the
height above-ground level (AGL) of the tags used in this study,
the altitude of each fix was compared to topographic heights
from digital elevation models, and height above-ground level
(AGL) was obtained by subtracting the topographic height from
the altitude recorded by the tag (height above sea level) (Katzner
et al. 2012). Inaccuracies exist in the heights found this way, due
to a sum of accuracies of interpolation of the topography data
(at the same scale as the measurements), plus that in the GPS
heights (Katzner et al. 2012). When the AGL for a location was
found to be negative, we compared the altitude to the minimum
topographic height over the accuracy of the heights, and used this
for calculating the AGL. If the AGL remained negative (ca.2%
of observations), it was not used. AGL height was then modelled
©2015 The Authors. Journal of Applied Ecology ©2015 British Ecological Society, Journal of Applied Ecology
4T. Reid et al.
as a logistic regression with the response being ‘1’ to indicate that
the location was within 200 m of the ground surface, and ‘0’ to
indicate >200 m. This model created a habitat model of the
heights used by bearded vultures. This model was tested and vali-
dated in the same way as the habitat model using methods from
dismo in R (Hijmans et al. 2013). Thus, we explored whether ‘at
risk’ heights were associated with the six variables in Table 2. It
is more important to identify areas where the vultures are more
likely to occur and to fly at risk-height. To do this, we combined
the earlier habitat use model with this height model by using the
product of the two. This was then plotted to identify areas of
greatest risk to the population.
Our predictive models provide a probability density across each
90 990 m square within the study area for each age class. These
models can then be used to estimate the overall spatial use of the
area by the population. This is done by first normalizing these
probabilities so that the sum of probabilities over the study area
adds up to one, with the resultant probability values then multi-
plied by the estimated overall population size of the different age
classes (218 adults and 131 non-adults) (Wakefield et al. 2011;
uger in press) and then summing across the age classes. The
resultant spatial use was then expressed as the estimated number
of birds km
. We did not estimate the density distribution of
fledglings as these were largely dependent on the locations of
nests and were considered non-adults after May.
Two areas have been proposed for the development of wind
farms within Lesotho. To explore the application of the model
developed here, we divided the study area into a 54954km
grid (derived from the 90 990 m grid topographic data used in
the model, and appropriate to the approximate turbine ‘foot-
prints’ of the proposed schemes) and calculated the mean proba-
bility of birds flying under 200 m AGL within each of these
squares throughout the study area. To compare the probability
of birds flying at risk-heights in the proposed area to the overall
threat throughout the study region, we plotted frequencies of ‘at
risk’ probabilities for all 54954 km squares throughout the
study region and then examined the probabilities for the squares
where these wind farms were proposed.
Tagged adult bearded vultures were restricted to areas
around the Drakensburg escarpment, especially between
Golden Gate Highlands National Park and Maloti Dra-
kensburg Park (Fig. 2a). The simpler GLMM produced
very similar results to the original GAMM model and
was therefore used in preference (Table 1). Cohen’s kappa
was reduced slightly between models (078 vs. 072,
respectively) (Fielding & Bell 1997). This model suggested
that adult bearded vultures use areas with higher alti-
tudes, with steep slopes and sharp points, and in areas
located closer to their nesting site (Table 1). The mean
AUC for this model was calculated as 0903 (range 0902
0904) suggesting a good model fit (Fig. 3a). The model
showed steeper sensitivity than specificity (Fig. S1). By
applying this model to the entire study area, incorporating
all the known occupied breeding sites, the model predicted
the greatest habitat use within the Drakensburg Moun-
tains, especially along the border between Lesotho and
KwaZulu-Natal, south into Eastern Cape, and in eastern
Lesotho (Fig. 4a). Adults were most likely to fly below
200 m over Lesotho, or to the north or south of it
(Fig. 4c). Adults were more likely to fly above the risk-
height with lower elevation, with distance from the nest
and with rugged terrain (Table 2). The mean AUC for
this height model was 0747 (range 07400751). The
model showed steeper sensitivity than specificity (Fig. S2).
This suggests the model picked true positives better than
it picked true negatives. Combining these models indi-
cated that when present, adult bearded vultures were most
likely to be flying low over the escarpment of the Dra-
kensburg Mountains or in eastern Lesotho (Fig. 4e). A
total of 55% of positions recorded for adults were within
the ‘at risk’ height.
To further validate the strength and power of our pre-
dictive model, a model of adult habitat use was developed
using only five of the six adult birds, and predicted habi-
tat use centred on the nest of the known sixth bird was
plotted (Fig. 5). This exercise suggests good power to our
predictive models with most of the observed locations
Table 1. Coefficients of covariates used for predicting distribution
of adult bearded vultures from a generalized linear mixed model.
Random effect for bird with variance of 120E-04
Coefficients Estimate SE Zvalue P
Intercept 614E+00 635E-02 966<001
Altitude 165E-03 195E-05 843<001
SD 337E-02 365E-04 924<001
915E-02 354E-02 259 <001
Skewness 514E-01 185E-02 2779 <001
Kurtosis 105E-01 269E-02 389 <001
Distance 178E-02 288E-04 618<001
0·0 0·2 0·4 0·6 0·8 1·0
0·0 0·2 0·4 0·6 0·8 1·0
AUC = 0·905
False postive rate
True positive rate
0·0 0·2 0·4 0·6 0·8 1·0
0·0 0·2 0·4 0·6 0·8 1·0
AUC = 0·961
False postive rate
True positive rate
(a) (b)
Fig. 3. AUC (area under the receiver curves) for predictive mod-
els for the presence of different ages of bearded vultures a. adults;
b. non-adults.
©2015 The Authors. Journal of Applied Ecology ©2015 British Ecological Society, Journal of Applied Ecology
Bearded vultures and wind turbines 5
26 27 28 29 30 31
−33 −32 −31 −30 −29 −28 −27 −26
26 27 28 29 30 31
−33 −32 −31 −30 −29 −28 −27 −26
26 27 28 29 30 31
−33 −32 −31 −30 −29 −28 −27 −26
26 27 28 29 30 31
−33 −32 −31 −30 −29 −28 −27 −26
26 27 28 29 30 31
−33 −32 −31 −30 −29 −28 −27 −26
26 27 28 29 30 31
−33 −32 −31 −30 −29 −28 −27 −26
(a) (b)
(c) (d)
(e) (f)
Fig. 4. Predicted distribution of bearded vultures and low flying heights (a, c, e =adults; b, d, f =non-adults). Colour shading pro-
portional to the probability for each 90 990 m square: a and b predicted presence; c and d predicted probability of flying at <200 m; e
and f predicted probability of flying in an area at <200 m, given that they are visiting the area. Black lines represent topographic con-
tours (1000, 2000 and 3000 m). South African border represented by blue or red line.
©2015 The Authors. Journal of Applied Ecology ©2015 British Ecological Society, Journal of Applied Ecology
6T. Reid et al.
occurring in areas with a probability >~07. Those few
fixes which occurred in areas with low probability were
mostly in areas that were more distant from the nest.
Non-adult bearded vultures similarly predominantly used
areas over the northern and eastern Drakensburg Moun-
tains (Fig. 4b). Again, a GLMM was developed that fitted
the data well, with some covariates fitted as polynomials
(Table 3). Birds were more likely to use areas at higher
altitudes, with steeper slopes and more rugged terrain,
and in areas closer to bearded vulture nesting sites and
restaurants and were less likely to use areas with skew
values close to zero (Table 3). The mean AUC for this
model was calculated as 0962 (range 09600963)
(Fig. 3b). The model showed steeper sensitivity than spec-
ificity (Fig. S3). While similar to the adult model, the
model predicted non-adult habitat use over a much wider
area in the north-east, as well as to the south-east and the
south-west into the Eastern Cape (Fig. 4b). Non-adult
bearded vultures were predicted to fly high over areas to
the east of the Drakensburg Mountains (Fig. 4d). They
were most likely to fly at ‘at risk’ heights in areas with rel-
atively flat land with high elevation, and more distant
from nests (Table 4); however, they were unlikely to use
those areas (Fig. 4f). Generally, non-adults were more
likely to fly at ‘at risk’ heights in areas where adults were
less likely to do so. The mean AUC for the height model
was 0692 (range 06860698). The model showed steeper
sensitivity than specificity (Fig. S4). A total of 66% of
non-adults heights were within the ‘at risk’ height.
Greatest densities of bearded vultures occurred along the
Drakensburg Escarpment from the area of Golden Gate
Highlands National Park south into the northern part of
the Eastern Cape (Fig. 6). Density was also high within
areas of eastern Lesotho at altitudes >2000 m (Fig. 6).
The probability of adults flying at ‘at risk’ heights within
the areas of the two proposed wind farm sites was fairly
low (P<04), while the chance of non-breeding birds vis-
iting was high (P>07 at one site) (Fig. 7). A histogram
of the estimated probabilities of ‘at risk’ usage for each
54954 km grid within the overall study area showed
an approximately exponential shape where most areas
had low probability, while the two areas with wind farm
proposals were within some of the higher threat areas
(Fig. 8). For adults, the two wind farm sites are within
Table 2. Approximate significance of smooth terms used in a
model to predict the probability of adult bearded vultures flying
below 200 m above-ground level
Effective d.f. Chi.sq P
Restaurant 8571 30703 <001
Kurtosis 8575 5212 <001
Skewness 7999 11998 <001
SD 8627 203639 <001
Altitude 7926 22471 <001
Distance 5595 2687 <001
Fig. 5. Predictive model for adult bearded vultures derived from
five of the six tagged birds. The actual observed locations for one
bird are plotted over this (crosses). Black lines indicate topogra-
phy (2000 and 3000 m). Nest site represented as white circle.
Table 3. Approximate significance of smooth terms used in a
model to predict the distribution of non-breeding bearded vul-
tures. Random effect for bird with variance of 135E-04
Coefficients Estimate SE Zvalue P
Intercept 846E+00 171E-01 4953 <001
Altitude 448E-03 147E-04 3040 <001
394E-07 311E-08 1269 <001
SD 349E-02 763E-04 4570 <001
282E-05 678E-06 417 <001
Skewness 244E-01 146E-02 1670 <001
561E-01 359E-02 1564 <001
Kurtosis 280E-01 242E-02 1157 <001
125E-02 290E-03 430 <001
Restaurant 892E-02 109E-03 8189 <001
456E-04 113E-05 4029 <001
Distance 105E-01 266E-03 3958 <001
283E-03 724E-05 3905 <001
©2015 The Authors. Journal of Applied Ecology ©2015 British Ecological Society, Journal of Applied Ecology
Bearded vultures and wind turbines 7
the 1% and 3% of most ‘at risk’ areas, and for subadults,
the wind farms are within the 05% and 3% of most ‘at
risk’ areas. Thus, these results suggest that the current
proposed locations are poorly situated with respect to the
overall risk of the area from the perspective of bearded
vulture conservation and that many areas exist which
have a much lower ‘at risk’ usage.
We aimed to produce predictive models of ‘at risk’ habitat
use by bearded vultures to help inform wind turbine
placements within an area that is proposed to be heavily
developed by this industry in the coming years. Using
data from GPS tracking units, we were able to produce
apparently reliable models of habitat use for this species
across its entire southern African range, which should
hopefully be useful for future developments in avoiding
areas of heavy use for potential wind farms. Results, how-
ever, suggested that the current proposed areas for wind
farms are amongst the worst areas, that is to say the most
heavily used areas, by the species and therefore have the
potential to cause the most damage to the population
through collisions, particularly by non-adult birds.
Bearded vultures in all age categories were predicted to
have a higher probability of using areas of high ground
with steep ridges and slopes. To some degree, this fits with
their previous description as being a bird of high-altitude
grasslands and escarpments (Hockey, Dean & Ryan 2005;
uger et al. 2014). Our models for both adults and su-
badults emphasize the importance of the proximity of
nests, as well as land topography and change in steepness
(e.g. cliffs). Vulture restaurants showed limited effects on
adults in the models, while non-adults showed an
increased chance of occurring within 100 km of restau-
rants. These areas are similar to those used by bearded
vultures in Europe and Asia (Ferguson-Lees & Christie
2001). In addition to factors considered here, the distribu-
tion of ungulates has been found to be of some influence,
especially for non-adults (Hirzel et al. 2004). We also
could not consider the location of ossuaries (areas used
by bearded vultures for breaking bones). While Hirzel
et al. (2004) speculated that features of ossuaries may
have influenced habitat selection of bearded vultures in
the Alps, according to Margalida & Bertran (2001),
bearded vultures spend relatively little time at ossuaries,
and Hirzel et al. (2004) concluded that landscape features
promoting wind conditions to facilitate the searching for
food were paramount, even while including variables on
the distribution of large ungulates. The goodness-of-fit of
our models in southern Africa was consistent with this
conclusion of the overriding influence of topography on
range use, even with less information on food availability.
This is encouraging for the wider application of our
approach, in terms of what data are needed to inform the
siting of potential wind farm developments, across large
areas, for other vultures and raptors that fundamentally
rely on topographically generated air movement to search
for food (see also McLeod et al. 2002).
The fitted models gave a good fit of the distribution of
the birds, lowering the residual deviance by 5065%; how-
ever, the models of flying heights were less successful
(lowering the deviance by 1020%). These comparative
results are broadly to be expected. Location fixes approxi-
mate the time spent, with increasing numbers of records
in areas of increasing importance. Fixes were taken hourly
with sufficiently few missing compared to total numb-
ers to suggest biases would have been minimal. However,
flying height data are somewhat more difficult to collect
and model. Inaccuracies in these AGL data can appear
due to the summing of errors such as the GPS position
and the topographic mapping. This is likely to be of par-
ticular concern within the landscape that this population
occupies due to the large numbers of cliffs over which
they were flying, which can give rise to negative values in
26 27 28 29 30 31
−31 −30 −29 −28 −27
Fig. 6. Estimated density of bearded vultures (adults and subad-
ults) throughout the distribution range in birds km
. The esti-
mate is the sum of the densities of adults and of subadults.
Densities for each derived from the models of the probabilities of
presence in each square normalized over the total range and
accounted for the estimated population. Black lines represent
height contours (1000, 2000 and 3000 m). South African border
represented by blue line.
Table 4. Approximate significance of smooth terms used in a
model to predict the probability of subadult bearded vultures fly-
ing below 200 m above-ground level
Term Effective d.f. Chi.sq P
Restaurant 840 3187<001
Kurtosis 769 4073<001
Skewness 621 1936<001
SD 860 26053<001
Altitude 748 9699<001
Distance 865 2056<001
©2015 The Authors. Journal of Applied Ecology ©2015 British Ecological Society, Journal of Applied Ecology
8T. Reid et al.
AGL height (Katzner et al. 2012). A further concern was
as a result of the tags taking a snapshot of flight height
once every hour. While these can be presumed to be set-
tling towards being representative with increasing sample
size, they are still a small proportion of the actual time.
Further to this, there are potentially other variables that
have not been measured that may be important in predict-
ing how airspace is used by the birds. Since the birds ride
thermals, the time of day is likely to have an important
effect upon the AGL measures (Pennycuick 1972; Kr
Reid & Amar 2014). Wind patterns, as they pass over
ridges and through valleys, are also likely to have impor-
tant effects. However, there are no data collected on this
by the tags, and little data on the patterns of wind over
South Africa, especially at the time and spatial scales
appropriate for making use of these measures within mod-
els (WASA 2013). In spite of these considerations with
modelling the AGL, it is noteworthy that broadly the
areas that were used by bearded vultures flying below
200 m were similar to those predicted for presence
(Fig. 4), indicating that even data from tags that do not
collect height data will approximate the areas which are
likely to pose the greatest threat from wind farms. This
will, however, potentially simplify the identification of
areas of importance for this and other vulture species
which show similar patterns of vertical space use. Never-
theless, understanding areas where birds are flying at
heights that potentially put them into conflict with wind
turbines is likely to be useful, even where it is relatively
uncertain. However, Ferrer et al. (2012) concluded that
expected (pre-turbine construction) estimates of griffon
vulture Gyps fulvus collision mortality were more likely to
(a) (b)
Fig. 7. Probability of birds flying within the danger heights (<200 m) given that they are present within the area (subset plots from
Figs 3e and f) in regions of two proposed wind farms in Lesotho. Red dots =site 1; violet dots =site 2; blue line =Lesotho and South
Africa border; black lines =land contours. a. adults, b. non-adults.
4000 6000 8000
02000 4000 6000
(a) (b)
Fig. 8. Frequency plots of probability of birds flying below 200 m (i.e. ‘at risk’ height) within all 54954 km blocks (approximately
size of a wind farm) from throughout the entire study region. The red lines represent the probability of birds using the area and flying
below 200 m for two blocks that have been proposed for wind farms (shown in Fig. 7; in each case, the right line (higher P) is site 1). a.
adults, b. non-adults.
©2015 The Authors. Journal of Applied Ecology ©2015 British Ecological Society, Journal of Applied Ecology
Bearded vultures and wind turbines 9
be closer to observed (post-construction) measures of col-
lision mortality if expectations were based on flight activ-
ity heights greater than heights swept by turbine blades.
Improved knowledge of the movements and habitat use
of species of concern for conservation has the potential to
vastly improve our understanding of their needs and
improve management decisions. In this study, we have
developed a suite of models of habitat use by bearded vul-
tures to understand how they might interact with wind
farms. These models are useful both for describing how
bearded vultures’ habitat use is distributed and how cer-
tain features within their environment are important for
influencing this habitat use. Further to this, they can be
used for predicting in which areas wind farms are likely
to have higher or lower impacts on the vultures. This has
the potential to be expedient for planning the future
deployment of wind farms and to reduce the need for mit-
igation following any construction.
Our approach has followed other studies that have used
information on the distribution or generic flight activity
patterns of individual species to provide a measure of the
potential conflict with wind farms (Fielding, Whitfield &
McLeod 2006; Tapia, Dom
ınguez & Rodr
ıguez 2009; Tell-
ıa 2009; Eichhorn et al. 2012; Katzner et al. 2012;
Miller et al. 2014). Similarly, we used our models to
examine the chance of bearded vultures visiting two wind
farms proposed in Lesotho. Our models indicated that
both areas have a high probability of posing a threat to
bearded vultures, particularly non-breeding birds. Of 10
000 5 95 km squares in the study area, one of the wind
farm proposals was within one of the 1% of squares most
likely to be visited by bearded vultures flying below
200 m.
The advantage of our models over previous research is
that they provide for estimation of use (and flight activity
within the airspace swept by turbine blades) across all
cohorts of a population, and at a scale such that it should
be possible to identify areas at the scale of tens of metres
or more to inform the prospective siting of both wind
farms, but also the specific turbines within these farms to
minimize their threat to this species. Thus, our models
allow for recommendations on wind turbine placement
before areas have been chosen, with these recommenda-
tions being based on a holistic age-cohort basis and at a
scale that has been emphasized as providing the most
realistic predictions of potential collision mortality for
another vulture species (Carrete et al. 2009; Ferrer et al.
Our models potentially have a vital conservation appli-
cation. The bearded vulture population in southern Africa
has been declining from before the introduction of wind
turbines (Brown 1991; Kr
uger et al. 2006; Simmons &
Jenkins 2007) and has declined by over 30% over the last
few decades down to just over 100 active pairs currently
uger et al. 2014). The species previously had a more
extensive distribution and was present in southern and
western South Africa, as far south as Cape Town (Piper
2006). Various factors have been blamed for their
declines, including habitat destruction and poisoning; one
study has suggested that declines were unlikely to be
related to food (Brown 1991), while another more recent
study has linked anthropogenic factors (powerlines and
human settlements) to territorial abandonment (Kruger,
Simmons & Amar 2015). Throughout their global range,
they are currently threatened by poisoning, persecution by
humans and by collisions with power lines (Ogada, Kee-
sing & Virani 2012). Regardless, if wind turbines are con-
structed in areas where they pose the most threat to
vultures, they will likely accelerate this decline (e.g.
ınez-Abrain et al. 2012; Bellebaum et al. 2013).
In spite of the cautions required in their interpretation,
our models are broadly useful for identifying areas that
may cause high risk for bearded vultures from wind tur-
bines, and areas that are more suitable for wind farm
placement. In this respect, they build on the novel utility
of remote satellite tagging in the study of wind farmbird
interactions (Katzner et al. 2012). By adding linkage
between bird use and predicted collision mortality (e.g.
Band, Madders & Whitfield 2007), our models should be
further developed to realize their full potential by examin-
ing the mortality effects that placement of turbines could
potentially have on the population trajectory.
Our approach can be easily replicated for other species
with similar data. For example, our study area also sup-
ports around 25% of the world’s population of Cape vul-
tures Gyps coprotheres. Similar tracking data are currently
being gathered from several individuals of this species and
should allow a similar exercise to be conducted in the
future. Our study illustrates that with the use of satellite
tracking data and readily available landscape data, the
risks posed by wind farms (current or prospective) can be
successfully modelled over large areas for other similar
resident ‘soaring’ vultures and raptors that are considered
to be vulnerable to collision with turbine blades (Drewitt
& Langston 2006; Ferrer et al. 2012). This is likely
because critical features of the habitat selection of such
species are topographic features associated with the gener-
ation of winds that allow low energetic costs of birds
searching for food or other movements (e.g. McLeod
et al. 2002; Katzner et al. 2012; de Lucas, Ferrer & Janss
2012). The applied benefits of our approach for the large
number of such species are many, especially pertinent
when the rapid pace of wind farm development continues
to outstrip the capacity for applied research on bird
impacts and relevant data availability to keep up.
For example, it is apparent that data on potentially
influential variables such as food abundance may be desir-
able, but not essential to model potential risk (see also de
Lucas, Ferrer & Janss 2012). Hence, the probable absence
of such data in some parts of the world, notably develop-
ing countries, where wind farm development is being
increasingly proposed, should not be an obstacle to con-
sidering potential impacts, provided that the essential data
in our approach have been collected (from satellite tag-
©2015 The Authors. Journal of Applied Ecology ©2015 British Ecological Society, Journal of Applied Ecology
10 T. Reid et al.
ging a representative sample of birds). In addition, our
approach allows greater capacity for assessment of poten-
tial adverse effects not only at specific sites (as our study
has illustrated directly) but also to inform larger scale
strategic environmental assessments on which, even in
Europe with a strong legislative framework, there has
been criticism of inadequate compliance and thereby
threats to important faunal interests (Gove et al. 2013).
PTTs were funded by the Maloti Drakensberg Transfrontier Programme
(13), the Wildlands Conservation Trust (3), Terra de Natura (1), Aspen
Pharmaceuticals (1) and the McAdams family (1). We are grateful to the
capture team for their patience and tireless efforts, in particular Carmen
Callero, Rickert van der Westhuizen and Ben Hoffman who assisted with
the bulk of the captures. Capture costs were covered by Sasol through
the Endangered Wildlife Trust, Ezemvelo KZN Wildlife and Wildlands
We also thank the following landowners and managers for their sup-
port and assistance with the capture and marking activities: Peter Dom-
mett, Brett Moller, Witsieshoek Mountain Resort staff and Ezemvelo
KwaZulu-Natal Wildlife staff of the Maloti Drakensberg Park World Her-
itage Site.
We are grateful to Alan Fielding, Miguel Ferrer and an anonymous
reviewer for providing helpful comments on an earlier draft.
Data accessibility
The data used in this paper can be accessed by applying to data@kznwil- or writing to the Data Manager, Ezemvelo KZN Wildlife, PO
Box 13053, Cascades 3202, South Africa. A digital copy (GIS layer) of the
main habitat use maps is available on request from the above address.
Amar, A. & Redpath, S. (2005) Habitat use by hen harriers Circus cyaneus
on Orkney: implications of land use change on this declining popula-
tion. Ibis,147,3747.
Anderson, M.D. & Anthony, A. (2005) The advantages and disadvantages
of vulture restaurants versus simply leaving livestock (and game) car-
casses in the veldt. Vulture News,53,4253.
Arroyo, B., Amar, A., Leckie, F., Buchanan, G.M., Wilson, J.D. & Red-
path, S. (2009) Hunting habitat selection by hen harriers on moorland:
implications for conservation management. Biological Conservation,142,
Band, W., Madders, M. & Whitfield, D.P. (2007) Developing field and
analytical methods to assess avian collision risk at wind farms. Birds
and Wind Farms: Risk Assessment and Mitigation (eds M. de Lucas,
G.F.E. Janss & M. Ferrer), pp. 259275. Quercus, Madrid, Spain.
Barov, B. & Derh
e, M. (2011) Review of the implementation of species
action plans of threatened birds in the European Union (2004-2011) Final
Report. BirdLife International. Downloaded from
report%20BirdLife%20review%20SAPs.pdf Accessed on 4 September
Bellebaum, J., Korner-Nievergelt, F., D
urr, T. & Mammen, U. (2013)
Wind turbine fatalities approach a level of concern in a raptor popula-
tion. Journal for Nature Conservation,21, 394400.
BirdLife International (2013) Wind farm in Lesotho could cause the
local extinction of vultures. Downloaded from
of-vultures-2/ Accessed on 28 August 2013.
BirdLife South Africa (2012) Avian wind farm sensitivity map for South
Africa. Downloaded from
Species%20and%20Sensitivity%20Criteria%20Finalb.pdf Accessed on
15 October 2013.
Bright, J.A., Langston, R.H.W., Bullman, R., Evans, R.J., Gardner, S.,
Pearce-Higgins, J. & Wilson, E. (2008) Map of bird sensitivities to wind
farms in Scotland: a tool to aid planning and conservation. Biological
Conservation,141, 23422356.
Brown, C.J. (1988) A study of the Bearded Vulture Gypaetus barbatus in
southern Africa. Unpublished PhD Thesis, University of Natal, Pieter-
Brown, C.J. (1991) An investigation into the decline of the bearded vulture
Gypaetus barbatus in Southern Africa. Biological Conservation,57, 315
Carrete, M., Sanchez-Zapata, J.A., Benitez, J.R., Lobon, M. & Donazar,
J.A. (2009) Large scale risk-assessment of wind-farms on population
viability of a globally endangered long-lived raptor. Biological Conserva-
tion,142, 29542961.
Dahl, E.L., Bevanger, K., Nyg
ard, T., Røskaft, E. & Stokke, B.G. (2012)
Reduced breeding success in white-tailed eagles at Smøla windfarm,
western Norway, is caused by mortality and displacement. Biological
zar, J.A., Hiraldo, F. & Bustamente, J. (1993) Factors influencing
nest site selection, breeding density and breeding success in the
bearded vulture (Gypaetus barbatus). Journal of Applied Ecology,30,
Drewitt, A.L. & Langston, R.H.W. (2006) Assessing the impacts of wind
farms on birds. Ibis.,148,2942.
Eichhorn, M., Johst, K., Seppelt, R. & Drechsler, M. (2012) Model-based
estimation of collision risks of predatory birds with wind turbines. Ecol-
ogy and Society,17,1.
Elith, J. & Leathwick, J.R. (2009) Species distribution models: ecological
explanation and prediction across space and time. Annual Review of
Ecology, Evolution, and Systematics,40, 677697.
Ferguson-Lees, J. & Christie, D.A. (2001) Raptors of the World, pp. 992.
Houghton Mifflin Harcourt, London.
Ferrer, M., de Lucas, M., Janss, G.F.E., Casado, E., Mu~
noz, A.R., Be-
chard, M.J. & Calabuig, C.P. (2012) Weak relationship between risk
assessment studies and recorded mortality in wind farms. Journal of
Applied Ecology,49,3846.
Fielding, A.H. & Bell, J.F. (1997) A review of methods for the assessment
of prediction errors in conservation presence/absence models. Environ-
mental Conservation,24,3849.
Fielding, A.H., Whitfield, D.P. & McLeod, D.R.A. (2006) Spatial associa-
tion as an indicator of the potential for future interactions between
wind energy developments and golden eagles Aquila chrysaetos in Scot-
land. Biological Conservation,131, 359369.
Gelman, A., Carlin, B., Stern, H.S. & Rubin, D.B. (2003) Bayesian Data
Analysis, 2nd edn. Chapman & Hall/CRC texts in Statistical Science,
New York.
Gove, B., Langston, R.H.W., McCluskie, A., Pullan, J.D. & Scrase, I.
(2013) Wind Farms and Birds: An Updated Analysis of the Effects of
Wind Farms on Birds, and Best Practice Guidance on Integrated Planning
and Impact Assessment. Bern Convention, Strasbourg.
Hansen, A.D. & Hansen, L.H. (2007) Wind turbine concept market pene-
tration over 10 years (19952004). Wind Energy,10,8197.
Hijmans, R.J., Phillips, S., Leathwick, J. & Elith, J. (2013) dismo: Species
distribution modeling. R package version 0.8-11.
Hirzel, A.H., Posse, B., Oggier, P.-A., Crettenand, Y., Glenz, C. & Arlet-
taz, R. (2004) Ecological requirements of reintroduced species and the
implications for release policy: the case of the bearded vulture. Journal
of Applied Ecology,41, 11031116.
Hockey, P.A.R., Dean, W.R.J. & Ryan, P.G. (2005) Roberts Birds of
Southern Africa. John Voelcker Bird Book Fund, Johannesburg.
Hunt, W.G. (2002) Golden eagles in a perilous landscape: predicting the
effects of mitigation for wind turbine blade-strike mortality. Consultant
Report 500-02-043F to California Energy Commission, Sacramento,
California. Available from
Hunt, W.G., Jackman, R.E., Hunt, T.L., Driscoll, D.E. & Culp, L. (1999)
A population study of golden eagles in the Altamont Pass Wind
Resource Area: population trend analysis 19941997. Report to
National Renewable Energy Laboratory, Subcontract XAT-6-16459-01.
Predatory Bird Research Group, University of California, Santa Cruz,
Jenkins, A. & Allan, D. (2013) An ill wind blows over the roof of Africa.
African Birdlife,1,5256.
Katzner, T.E., Brandes, D., Miller, T., Lanzone, M., Maisonneuve, C.,
Tremblay, J.A., Mulvihill, R. & Merovich, G.T. (2012) Topography
©2015 The Authors. Journal of Applied Ecology ©2015 British Ecological Society, Journal of Applied Ecology
Bearded vultures and wind turbines 11
drives migratory flight altitude of golden eagles: implications for on-
shore wind energy development. Journal of Applied Ecology,49, 1178
Kneidinger, C. (2011) Let
seng Wind Farm Environmental Impact State-
ment. Classical Environmental Management Services, Cape Town.
uger, S. (in press) Bearded Vulture. The Eskom Red Data Book of Birds
of South Africa, Lesotho and Swaziland (ed. M.R. Taylor). BirdLife
South Africa, Johannesburg.
uger, S., Piper, S., Rushworth, I., Botha, A., Daly, B., Allan, D., Jen-
kins, A., Burden, D. & Friedmann, Y. eds. (2006) Bearded Vulture
(Gypaetus barbatus meridionalis) Population and Habitat Viability
Assessment Workshop Report. Conservation Breeding Specialist Group
(SSC/IUCN)/CBSG Southern Africa. Endangered Wildlife Trust,
uger, S.C., Allen, D.G., Jenkins, A.R. & Amar, A. (2014) Trends in ter-
ritory occupancy, distribution and density of the Bearded Vulture Gypa-
etus barbatus meridionalis in southern Africa. Bird Conservation
International,24, 162177.
¨ger, S., Reid, T. & Amar, A. (2014) Differential range use between age
classes of Southern African bearded vultures Gypaetus barbatus.PLoS
ONE,9, e114920.
Kruger, S., Simmons, R.E. & Amar, A. (2015) Anthropogenic activities
influence the abandonment of Bearded Vultures (Gypaetus barbatus)
territories in southern Africa. Condor,117,94107.
Leung, D.Y.C. & Yang, Y. (2012) Wind energy development and its envi-
ronmental impact: a review. Renewable and Sustainable Energy Reviews,
16, 10311039.
Lu, X., McElroy, M.B. & Kiviluoma, J. (2009) Global potential for wind-
generated electricity. Proceedings of the National Academy of Sciences,
106, 1093310938.
de Lucas, M., Ferrer, M. & Janss, G.F.E. (2012) Using wind tunnels to
predict bird mortality in wind farms: the case of Griffon Vultures. PLoS
ONE,7, e48092.
Madders, M. & Whitfield, D.P. (2006) Upland raptors and the assessment
of wind farm impacts. Ibis,148,4356.
Margalida, A. & Bertran, J. (2001) Function and temporal use of ossuaries
by bearded vultures (Gypaetus barbatus) during the nestling period.
Auk,118, 785789.
Margalida, A., Dona
zar, J.A., Carrete, M. & S
anchez-Zapata, J.A. (2010)
Sanitary versus environmental policies: fitting together two pieces of the
puzzle of European vulture conservation. Journal of Applied Ecology,
47, 931935.
ınez-Abrain, A., Tavecchia, G., Regan, H.M., Jimenez, J., Surroca,
M. & Oro, D. (2012) Effects of wind farms and food scarcity on a large
scavenging bird species following an epidemic of bovine spongiform
encephalopathy. Journal of Applied Ecology,49, 109117.
McLeod, D.R.A., Whitfield, D.P., Fielding, A.H., Haworth, P.F. &
McGrady, M.J. (2002) Predicting home range use by golden eagles in
western Scotland. Avian Science,2, 183198.
Miller, T.A., Brooks, R.P., Lanzone, M., Brandes, D., Cooper, J., O’Mal-
ley, K. et al. (2014) Assessing risk to birds from industrial wind energy
development via paired resource selection models. Conservation Biology,
3, 745755.
Moore, A. & Blenkinsop, T. (2006) Scarp retreat versus pinned drainage
divide in the formation of the Drakensburg escarpment, southern
Africa. South African Journal of Geology,109, 599610.
Nemaxwi, H. (2013) Eskom Sere wind farm to be constructed by Siemens
South Africa. Downloaded from
on Accessed 28 August 2013.
Ogada, D.L., Keesing, F. & Virani, M.Z. (2012) Dropping dead: causes
and consequences of vulture population declines worldwide. Annals of
the New York Academy of Sciences,1249,5771.
Pennycuick, C.J. (1972) Soaring behaviour and performance of some east
African birds, observed from a motor-glider. Ibis,114, 178218.
Piper, S.E. (2006) Bearded vulture population and habitat viability assess-
ment (Gypaetus barbatus). Briefing Book. Population and habitat viabil-
ity assessment (PHVA) workshop, Harrismith 6-10 March 2006.
Downloaded from
ments/Aves/20060306%20BeardedVulturePHVABriefingDoc.pdf on
Accessed 3 October 2013.
R Core Team (2013) R: A Language and Environment for Statistical Com-
puting. R Foundation for Statistical Computing, Vienna, Austria. http://
Randin, C.F., Dirnb
ock, T., Dullinger, S., Zimmermann, N.E., Zappa, M.
& Guisan, A. (2006) Are niche-based species distribution models trans-
ferable in space? Journal of Biogeography,33, 16891703.
Simmons, R.E. & Jenkins, A.R. (2007) Is climate change influencing the
decline of Cape and bearded vultures in southern Africa? Vulture News,
Tabassum-Abbasi, M.P., Abbasi, T. & Abbasi, S.A. (2014) Wind energy:
increasing deployment, rising environmental concerns. Renewable and
Sustainable Energy Reviews,31, 270288.
Tapia, L., Dom
ınguez, J. & Rodr
ıguez, L. (2009) Using probability of
occurrence to assess potential interaction between wind farm and a
residual population of golden eagle Aquila chrysaetos in NW Spain. Bio-
diversity Conservation,18, 20332041.
ıa, J.L. (2009) Overlap between wind power plants and Griffon Vul-
tures Gyps fulvus in Spain. Bird Study,56, 268271.
Wakefield, E.D., Phillips, R.A., Trathan, P.N., Arata, J., Gales, R., Huin,
N. et al. (2011) Habitat preference, accessibility, and competition limit
the global distribution of breeding black-browed albatrosses. Ecological
Monographs,81, 141167.
WASA (2013) Wind Atlas for South Africa. Downloaded from http:// Accessed on 4 October 2013.
Wood, S.N. (2006) Generalized Additive Models: An Introduction With R.
CRC Press, Boca Raton, Florida, USA.
Wuertz, D., Chalabi, Y., Maechler, M. & Byers, J.W. (2013) timeDate:
Rmetrics - Chronological and Calendar Objects. R package version
Received 16 October 2014; accepted 14 May 2015
Handling Editor: Des Thompson
Supporting Information
Additional Supporting Information may be found in the online version
of this article.
Table S1. Dates and number of records for adult bearded vul-
Table S2. Dates and number of records for bearded vultures aged
between fledglings and adults.
Table S3. Dates and number of records for fledgling bearded
Table S4. Coefficients of covariates used in a model to predict the
distribution of fledgling bearded vultures.
Fig. S1. Sensitivity and specificity of adult presence model.
Fig. S2. Sensitivity and specificity of adult height model.
Fig. S3. Sensitivity and specificity of non-adult presence model.
Fig. S4. Sensitivity and specificity of non-adult height model.
Fig. S5. Locations of points recorded by fledgling tags attached to
bearded vultures.
Fig. S6. AUC curves for predictive models for presence for
fledgling bearded vultures.
Fig. S7. Predicted probability of fledgling bearded vultures spend-
ing time in areas.
©2015 The Authors. Journal of Applied Ecology ©2015 British Ecological Society, Journal of Applied Ecology
12 T. Reid et al.
... To prevent detrimental impacts of the turbine operation on endangered species, wildlife managers and wind energy companies need adequate planning tools to minimize the deployment of wind facilities in areas where major conflicts with biodiversity preservation will occur. Different approaches have been used as planning tools to mitigate the risks encountered by flying vertebrates, spanning from mere delineations of buffer areas around sensitive locations [29][30][31], through the compilation of distribution areas of sensitive species [28,29,32], to more complex methods that account for fine-grained habitat use and/or flight behaviour of potentially impacted bat and bird species [13,[33][34][35][36][37]. The first approach is fairly imprecise. ...
... In order to make use of all information included in the data and also develop a method that easily scales to potentially very large datasets, we used a deep feedforward neural network to model the probability of a bearded vulture flying within a given altitude range at a given location. Considering the still ongoing trend of increasing heights of newly constructed, modern wind turbines, we decided for a threshold of 200 m (hereafter referred to as critical altitude), below which the flight of a bird is deemed to be at potential risk of collision with the rotor blades (see also [33,34]). The flight altitude was converted to a binary response with 1 being a location within the critical altitude range and 0 otherwise. ...
... 4e, reported also in figure 3c) with the output of the model described in this article. The joint probability of species occurrence and flying within the critical altitude range (figure 3e) was calculated by taking the product of the two raster maps [34]. ...
Full-text available
Deployment of wind energy is proposed as a mechanism to reduce greenhouse gas emissions. Yet, wind energy and large birds, notably soaring raptors, both depend on suitable wind conditions. Conflicts in airspace use may thus arise due to the risks of collisions of birds with the blades of wind turbines. Using locations of GPS-tagged bearded vultures, a rare scavenging raptor reintroduced into the Alps, we built a spatially explicit model to predict potential areas of conflict with future wind turbine deployments in the Swiss Alps. We modelled the probability of bearded vultures flying within or below the rotor-swept zone of wind turbines as a function of wind and environmental conditions, including food supply. Seventy-four per cent of the GPS positions were collected below 200 m above ground level, i.e. where collisions could occur if wind turbines were present. Flight activity at potential risk of collision is concentrated on south-exposed mountainsides, especially in areas where ibex carcasses have a high occurrence probability, with critical areas covering vast expanses throughout the Swiss Alps. Our model provides a spatially explicit decision tool that will guide authorities and energy companies for planning the deployment of wind farms in a proactive manner to reduce risk to emblematic Alpine wildlife.
... Our data on Bearded Vulture movements provide an opportunity to develop habitat use models to be used as a planning tool to mitigate infrastructure developments aimed at reducing risks. For example, Reid et al. (2015) used the spatial analyses of Bearded Vulture movements to inform wind-turbine placement in Southern Africa. ...
... Their recommendation for mitigation was to move the development sites off the ridge tops and upper slopes. Reid et al. (2015) used the available movement data from tracked birds, and also incorporated flying heights at risk of collision, to predict important areas of use that may conflict with wind turbines. Adult and non-adult Bearded Vultures were found to use mostly areas with high elevations and steep and rugged topography in the core area; adults tended to use areas in relatively close proximity to their nest sites, whereas non-adult birds used areas dispersed over the entire species' range and were more likely to fly at risk-height in areas that were less used by adults. ...
... They demonstrated the value of habitat use models for identifying intensively used areas to guide the placement of wind turbines away from the most damaging locations. The habitat use model developed by Reid et al. (2015) is a tool that operates at the scale of regional and national development plans and provides important supplementary assessments of site-specific development proposals. Although an extremely useful tool, the habitat use model, is currently in a format that cannot be easily interpreted and implemented by those without specialist Geographic Information System's knowledge. ...
The Bearded Vulture Gypaetus barbatus is a Critically Endangered species in Southern Africa whose entire range in the Southern Hemisphere falls within the Maloti-Drakensberg mountains of South Africa and Lesotho. Here we synthesize 20 years of research on this population to quantify the decline in the species, investigate the potential drivers for this decline, and explore the most appropriate management actions necessary to recover this population. The population's breeding range and density have declined by at least 20%, and we estimate that there are now less than 400 individuals remaining. The densification of human settlements and associated infrastructure (e.g., power lines), together with human activities (e.g., poisoning) are the primary causes of mortality, and may also be contributing to the population's low productivity. Satellite tracking data of different age classes identified priority areas within the species' range to implement threat mitigation interventions. Conservation actions that have been implemented include supplementary feeding, power line mitigation, as well as risk models aimed at guiding developments. The need for population augmentation is discussed and assessed. Implementing the species management plan, specifically addressing the threat of poisoning, remains the priority to address the population decline and ensure that the unique ecological niche occupied by this imperiled species is maintained.
... (2) productivity of the population defined as the number of fledged chicks/pair/year (Krüger & Amar, 2017) as well as the number and fertility of eggs laid as part of a harvesting program to establish a captive population; (3) movement ecology to develop risk maps and identify areas requiring mitigation (Abbass, 2021;Krüger et al., 2014a;Reid et al., 2014;Rushworth & Krüger, 2014); (4) causes of mortality and survival rates from birds fitted with satellite tags; and (5) genetic diversity (Krüger et al., 2015a;Streicher et al., 2021), to assess population health and the potential for adaptation. These monitoring programs seek to provide information necessary to address the decline in the population and faster recovery. ...
... Infrastructure is also a potential hazard for vultures, in terms of both electrocutions and collisions with energy infrastructure (Angelov et al., 2013); in addition, the carcasses of large mammals that died after collisions on roads and railways attract vultures that may then be struck and killed by vehicles (Khatri et al., 2019). The loss of preferred tree species from development may be critical for tree-nesting vultures (Bamford et al., 2009), whereas other aspects of land use, such as wind farms, also pose a significant hazard to all vultures and in particular cliff-nesting vultures due to high-wind areas mostly overlapping with cliff-nesting rather than tree-nesting vulture habitat in southern Africa (Reid et al., 2014;Rushworth & Krüger, 2014). ...
Full-text available
African wildlife face challenges from many stressors including current and emerging contaminants, habitat and resource loss, poaching, intentional and unintentional poisoning, and climate‐related environmental change. The plight of African vultures exemplifies these challenges due to environmental contaminants and other stressors acting on individuals and populations that are already threatened or endangered. Many of these threats emanate from increasing human population size and settlement density, habitat loss from changing land use for agriculture, residential areas, and industry, and climate‐related changes in resource availability. Environmental chemicals that are hazardous include legacy chemicals, emerging chemicals of concern, and high‐volume‐use chemicals that are employed as weed killers and in other agricultural applications. Furthermore, there are differences in risk for species living in close proximity to humans or in areas affected by habitat loss, climate, and industry. Monitoring programs are essential to track the status of nesting pairs, offspring survival, longevity, and lifetime productivity. This is important for long‐lived birds, such as vultures, that may be especially vulnerable to chronic exposure to chemicals as obligate scavengers. Furthermore, their position in the food web may increase risk due to biomagnification of chemicals. We review the primary chemical hazards to Old World vultures and the interacting stressors affecting these and other birds. Habitat is a major consideration for vultures, with tree‐nesters and cliff‐nesters potentially experiencing different risks of exposure to environmental chemicals. The present review provides information from long‐term monitoring programs and discusses a range of these threats and their effects on vulture populations. Environ Toxicol Chem 2022;00:1–19.
... Thus, case studies have merit, especially if they involve detailed data which allow focus at fine scales e.g. at specific turbines and/or involving known individual birds' flight behaviour. Generating detailed data on birds' flight behaviour increasingly involves the accuracy and precision provided by GPS telemetry [5,18,24], and can be informative even when not collected at operational wind farms [44,45,68,69]. Case studies can thereby complement research at larger scales [68,[70][71][72][73][74][75][76]. ...
... Generating detailed data on birds' flight behaviour increasingly involves the accuracy and precision provided by GPS telemetry [5,18,24], and can be informative even when not collected at operational wind farms [44,45,68,69]. Case studies can thereby complement research at larger scales [68,[70][71][72][73][74][75][76]. ...
Full-text available
Wind farms can have two broad potential adverse effects on birds via antagonistic processes: displacement from the vicinity of turbines (avoidance), or death through collision with rotating turbine blades. These effects may not be mutually exclusive. Using detailed data from 99 turbines at two wind farms in central Scotland and thousands of GPS-telemetry data from dispersing golden eagles, we tested three hypotheses. Before-and-after-operation analyses supported the hypothesis of avoidance: displacement was reduced at turbine locations in more preferred habitat and with more preferred habitat nearby. After-operation analyses (i.e. from the period when turbines were operational) showed that at higher wind speeds and in highly preferred habitat eagles were less wary of turbines with motionless blades: rejecting our second hypothesis. Our third hypothesis was supported, since at higher wind speeds eagles flew closer to operational turbines; especially–once more–turbines in more preferred habitat. After operation, eagles effectively abandoned inner turbine locations, and flight line records close to rotor blades were rare. While our study indicated that whole-wind farm functional habitat loss through avoidance was the substantial adverse impact, we make recommendations on future wind farm design to minimise collision risk further. These largely entail developers avoiding outer turbine locations which are in and surrounded by swathes of preferred habitat. Our study illustrates the insights which detailed case studies of large raptors at wind farms can bring and emphasises that the balance between avoidance and collision can have several influences.
... Understanding flight performance of obligatesoaring birds under different demographic and environmental conditions has important implications for understanding individual fitness, ecology, and behavior. Movement behavior also interacts with risk to condors and other soaring birds from wind turbines (e.g., Miller et al. 2014, Reid et al. 2015, Poessel et al. 2018a, and wind energy generation is rapidly expanding within California (US Energy Information Administration 2020). Flight performance of condors may influence this risk because some of the factors that influence flight performance (e.g., altitude AGL, landscapes, DSR, and wind speed) may also be associated with higher or lower risk. ...
Full-text available
Flight behavior of soaring birds depends on a complex array of physiological, social, demographic, and environmental factors. California Condors (Gymnogyps californianus) rely on thermal and orographic updrafts to subsidize extended bouts of soaring flight, and their soaring flight performance is expected to vary in response to environmental variation and, potentially, with experience. We collected 6298 flight tracks described by high-frequency GPS telemetry data from five birds ranging in age from 1 to 19 yr old and followed over 32 d in summer 2016. Using these data, we tested the hypothesis that climb rate, an indicator of flight performance, would be related to the topographic and meteorological variables the bird experienced, and also to its age. Climb rate was greater when condors were flying in faster winds and during environmental conditions that were conducive to updraft development. However, we found no effect of age on climb rate. Although many of these relationships were expected based on flight theory, the lack of an effect of age was unexpected. Our work expands understanding of the relationship condors have with the environment, and it also suggests the potential for as-yet unexplored complexity to this relationship. As such, this study provides insight into avian flight behavior and, because flight performance influences bird behavior and exposure to anthropogenic risk, it has potential consequences for development of conservation management plans.
... Crucially, this was "one of the first times that GPS-GSM emitters" were used for the Egyptian vulture, "which is endangered worldwide, with the aim to know about their transcontinental movements and to track their way back to Catalonia in Spring" (Conservation Biology Group, 2019). A similar study by Reid et al. (2015) used GPS satellite transmitters (70 g solar-powered GPS-PTT-100s, Microwave Telemetry Inc., Columbia, MD, USA), which recorded the locations (accurate to within 5-20 m) elevation and speed of each studied bird every hour from 05·00 to 20·00 (see also Krüger et al. 2014). Vultures may also be fitted with GPS tracking devices and GoPro cameras which can be used for tracking features on the ground. ...
This article examines progress in drone-based research methods applied to animal ecology, in terms of applications to the field study of large birds of prey (raptors). Drone-based research methods have evolved out of the larger technological field of geomatics and are entwined with developments in GPS and biotelemetry, which enable accurate location recording, image capture, and specimen behavioral assessment. Current evidence indicates that drone-based data gathering methods derive more accurate information than older research methods (including human observation and even high-resolution image-based studies). Large raptors are important subjects for drone-based studies due to their visibility to human eyesight and cameras, fast flight, remote nesting locations, large, trackable prey, conflicts with people (killing of livestock and companion animals, collisions with aircraft), low population densities, and migratory habits that may require highly mobile observation machines. Measurable avian parameters for drones include migration patterns (sometimes interhemispheric/ continental) foraging (flight speeds, soaring duration, hunting flight patterns, sometimes of over hundreds of square kilometers), nesting habits (including fledgling behavior), courtship, roosting, plumage identification, individual behaviors, and aggressive tendencies. Emerging issues concern subject bird perceptions and negative reactions to drones, including avoidance and aggressive attacks on machines. The future of drone-assisted studies is dependent on more effective technology, especially lighter, quieter, less intrusive, more elusive drone structures. This makes an important contribution to raptor ecology.
... These results indicate that considering a combination of data on wind turbine densities along with collision events and regional population densities allows for improved assessments of collision distribution and strike susceptibilities at large spatial scales for wide-ranging birds, such as such large raptors. Therefore, our results support and encourage the use of models that use combinational data as a tool for the analysis of collision potential on larger spatial scales, as has been already done for many other bird species Reid et al. 2015;Vasilakis et al. 2016;. ...
Although, it is well recognized that harnessing wind energy is highly indispensable, but collisions of birds at wind turbines has also developed simultaneously, concerning multiple bird species. With wind being strongly affected by the landscape and the behavior of birds also being strongly influenced by the landscape, the main objective of the thesis was to understand the relevance of interactions between wind energy infrastructures and bird species from an ecological perspective of the landscape. Utilizing the carcass collision datasets of the frequently-hit bird-groups paradoxically as proxies for species presence, collision sensitive ecological distances to different land-use types were ascertained, by employing multiple techniques of species distribution modelling (SDMs), to delineate their respective collision sensitive niche employing the capabilities of machine learning algorithms. The predicted areas were specialized and highly dispersed across the federal state, with raptors showing the broadest niche and significant overlaps with the other groups. Based on estimated collision probabilities of the assessed areas (between 0 and 1), further segregations differentiated only those areas with negligible collision probabilities, <0.05, which were interpreted as the actual "no risk areas, suggesting any further planned additions of wind turbines to be suitably positioned only in these “safer” areas. Additionally, these collision probabilities were translated to strike susceptibilities, by relating them to the regional density distributions of the species as well. Summarizing, these analyses praigmatically ascertained collision risk areas, and especially the collision sensitive distances from different land-use types to these areas, enabling the accurate guidance of future wind farm expansions in the landscape. Ultimately, formulating novel wind turbine allocation strategies to minimize avian collisions, making them as compatible as possible.
Full-text available
Animal movements and bird migration have always fascinated humans (Holyoak et al. 2008). With the fast technology advancement in the past 20 years new systems and methods were developed allowing animals to be tracked for longer periods and significant amount of data to be collected, stored and analysed (Cooke et al. 2004, Cagnacci et al. 2010). Vultures are obligate scavengers which consume up to 90% of the carcasses in some ecosystems providing significant ecosystem services. By efficiently disposing the carcasses they prevent the spread of diseases and save costs from transportation and incineration of animal carcasses (Houston 1986, Pain et al. 2003). However, vulture populations are experiencing dramatic declines worldwide and their conservation is a priority in many areas (Botha et al. 2017). Due to their conservation status and role in the ecosystems more studies on vulture movements and ecology are needed to inform efficient conservation strategies. The recent study was conducted on the autochthonous Griffon Vulture population in the Eastern Rhodopes, Bulgaria. In the period 2016 – 2019 we equipped with solar-powered GSM-GPS and Argos-GPS transmitters adult (n = 10), immature (n = 8) and juvenile (n = 7) Griffon Vultures in order to study their home range, movements, foraging behaviour and migration pattern. The foraging home range of the species was 2958.4 km2 (95% KDE) with core area of 231.6 km2 (50% KDE). Foraging home range size was maximal in summer and minimal in winter (3166.2 km2 and 1327.7 km2 respectively). Adult vultures had significantly smaller core areas compared to immatures (Z = –2.15, p = 0.03). The daily travel distance with all seasons and all individuals pooled was 79.1 ± 64.9 km while displacement was 21.4 ± 20.5 km. The longest daily distance was recorded on 07th May when an immature vulture travelled 364.4 km within the Eastern Rhodopes. Successful breeders travelled longer daily distances than the adults which were not breeding or failed at different stage of their breeding attempt (89.5 ± 71.9 km and 65.7 ± 65.9 km respectively, t = 4.37, p < 0.05). The mean daily distance travelled by the immature vultures was 85 ± 66.06 km while adults travelled 76.82 ± 64.5 km (t = –6.05, p < 0.01). The difference between the two age classes was most prominent during winter and autumn when immatures travelled 45.8 ± 41.7 km and 51 ± 44 km respectively while adults had significantly shorter daily distances 29.9 ± 31.3 km and 36.6 ± 42.8 km (t = –5.37, p < 0.01; t = –5.45, p < 0.01). Griffon Vultures were roosting mostly on cliffs (85.62%, n = 8120), in 14.05% of the cases they were roosting on trees and twice ground roosts were recorded. In the Bulgarian part of the Eastern Rhodopes vultures were roosting on cliffs in 94.6 ± 3.9% of the cases while in the Greek part of the mountain they were roosting mostly on trees – 78.7 ± 24.4%. Our results indicated high variance in the preferences of roosting cliffs according to the season. In autumn and winter vultures were roosting on cliffs with breeding pairs in 80.1 ± 24.2% and 88 ± 24.8% of the cases respectively while this percentage dropped significantly is spring and summer when vultures preferred to roost on cliffs with no breeding pairs (59.4 ± 25.3% and 45.8 ± 24.8%). The recent study showed that 71.5% of the juvenile Griffon Vultures migrate south in their first autumn while only 14% of the immatures started migration and none of the tracked adults. We followed 8 vultures during autumn migration and 5 during the spring migration. Autumn migration started in the period 19th September – 29th October. The average distance travelled on migration was 3602 ± 1137 km, covered for 38 ± 12 days with an average 44 migration speed 100.7 ± 32 km/day. The longest daily distances travelled on migration was 374 km on 30th October when the juvenile vulture 6G crossed the Bosphorus and reached the region of Gerede in Turkey. Spring migration started in the period 22th March – 7th May. The mean distance travelled was 2340 ± 737 km and migration took on average 13 ± 6 days with an average migration speed 176.3 ± 61.8 km/day. Griffon vultures had greater migration speed during the spring than the autumn (t = 2.50, p < 0.05). During autumn migration vultures used different stopover sites along the flyway where they spent between 3 and 50 days. The majority of the stopover sites were in Turkey, one was in Iraq and one on the border area between Iraq and Iran. All vultures followed the Eastern Mediterranean flyway through Turkey and Middle East. The most important bottlenecks for the juvenile and immature Griffon Vultures were the Bosphorus and Iskenderun in Turkey. The main wintering areas were in central and north Saudi Arabia, Israel. One juvenile vulture reached South Sudan which is the first record of the species for the country and one of the southernmost records in Africa. The home range in the wintering areas was 18 933 ± 13 314 km2 (95% KDE) and the size of the core area was 1876 ± 2001.4 km2 (50% KDE). The size of the home range varied among the individuals and the years. In the wintering grounds 78.07% of the area inhabited by the vultures had no vegetation e.g. deserts and rocky mountains. Only 10.05% were covered by sparse vegetation and 8.39% were natural grasslands or arable lands. Griffon Vultures were feeding at natural carcasses found in the wild in the Eastern Rhodopes in 77.4% (n = 1036) of the recorded cases. In winter 56.5% of the feedings were at the vulture feeding stations while in the summer 80.2% of the feeding events were on occasional carcasses found in the wild. The breeding Griffon Vultures were feeding at the vulture restaurants mostly during the pre-breeding and incubation period (54% and 46.6% respectively). During the post-breeding period 81.6% of the feedings were in the wild. Vultures were landing on the feeding stations on average 53.2 h after carcass disposal. In summer and autumn this period was prolonged up to 10 – 12 days. Griffon vultures were feeding in 42.8% of the days in the month. In the summer they were feeding on average once per 1.6 days and in winter once per 4.1 days. One vulture can visit up to 4 feeding locations per day. The recent study revealed that Griffon Vultures travel significantly longer distances in days when they were feeding on carcasses found in the wild compared to days when feeding at vulture restaurants (t = –11.6 p < 0.001). In addition, they have less straight flight and reach lower displacement when feeding on occasional carcasses (t = 5.9, p < 0.001; t = –7.33, p < 0.001). The average daily distances travelled were 80.3 ± 53.3 km in days when vultures were feeding and only 69.8 ± 58.4 km in days when they did not manage to find food. Our model showed that the season and the age of the vultures determine the most their success in finding food. The other factors which showed correlation were the daily travelled distance, daily displacement, temperature, daily precipitation and wind speed. In 47% of the cases (n = 305) vultures were feeding on cattle carcasses in the wild. In 28% sheep or goats were used for food and wild ungulates were found in 11.5% of the cases. Other species consumed by the vultures were fox, jackal, dog, wild boar, hare, horse and donkey. In 4.6% of the cases vultures were feeding at places where offal from slaughter houses was illegally dumped. The most common reason for the death of the animals consumed by the vultures was carnivore attack (60.2%) while in 37.6% of the cases animals died due to natural causes. However, in 2 occasions death was attributed to poaching.
Wind farms may have two broad potential adverse effects on birds via antagonistic processes: displacement from the vicinity of turbines (avoidance), or death through collision with rotating turbine blades. Large raptors are often shown or presumed to be vulnerable to collision and are demographically sensitive to additional mortality, as exemplified by several studies of the Golden Eagle Aquila chrysaetos. Previous findings from Scottish Eagles, however, have suggested avoidance as the primary response. Our study used data from 59 GPS‐tagged Golden Eagles with 28 284 records during natal dispersal before and after turbine operation < 1 km of 569 turbines at 80 wind farms across Scotland. We tested three hypotheses using measurements of tag records’ distance from the hub of turbine locations: (1) avoidance should be evident; (2) older birds should show less avoidance (i.e. habituate to turbines); and (3) rotor diameter should have no influence (smaller diameters are correlated with a turbine’s age, in examining possible habituation). Four generalized linear mixed models (GLMMs) were constructed with intrinsic habitat preference of a turbine location using Golden Eagle Topography (GET) model, turbine operation status (before/after), bird age and rotor diameter as fixed factors. The best GLMM was subsequently verified by k‐fold cross‐validation and involved only GET habitat preference and presence of an operational turbine. Eagles were eight times less likely to be within a rotor diameter’s distance of a hub location after turbine operation, and modelled displacement distance was 70 m. Our first hypothesis expecting avoidance was supported. Eagles were closer to turbine locations in preferred habitat but at greater distances after turbine operation. Results on bird age (no influence to 5+ years) rejected hypothesis 2, implying no habituation. Support for hypothesis 3 (no influence of rotor diameter) also tentatively inferred no habituation, but data indicated birds went slightly closer to longer rotor blades although not to the turbine tower. We proffer that understanding why avoidance or collision in large raptors may occur can be conceptually envisaged via variation in fear of humans as the ‘super predator’ with turbines as cues to this life‐threatening agent.
Full-text available
When offered a selection of food items, bearded vultures Gypaetus barbatus in southern Africa chose bones in preference to meat or to feeding from a fleshed carcass. Once a carcass had been stripped of soft tissue by Gyps vultures, bearded vultures disarticulated sections or individual bones (depending on the size of the dead animal) in the order: limbs, ribs, vertebrae, skull. Their overall diet was estimated as 70% bone with marrow, 25% meat and 5% skin. This diet is about 15% higher in energy than an equivalent mass of meat. Of 683 identified prey items from five sources of data, over 80% consisted of domestic livestock; about 60% of this was sheep and goats. Even birds nesting within conservation areas derived more than half of their food from domestic stock which they found by foraging over adjacent commercial and subsistence farming areas. Bearded vultures obtain all their food by scavenging, and reports of attacks on live animals and even humans are rejected.Tydens die verskaffing van voedselitems aan baardaasvoëls Gypaetus barbatus in suider Afrika, is daar bevind dat die voëls voorkeur gee aan bene bo vleis, of verkies om aan ’n karkas te vreet waarvan die vleis verwyder is. Nadat Gyps aasvoëls die sagte weefsel van ’n karkas gestroop het, het die baardaasvoëls artikulerende dele of enkel bene van ’n karkas verwyder (afhangende van die grootte van die dooie dier) in die volgorde: ledemate, ribbes, werwels, skedel. Daar is beraam dat huile algemene normale dieet bestaan uit 70% bene met murg, 25% vleis and 5% vel. Hierdie dieet verskaf 15% meer energie as ’n ooreenstemmende massa vleis. Van 683 geïdentifiseerde prooi-items, versamel vanaf vyf verskillende waamemingspunte, was 80% afkomstig van vee, waarvan 60% skaap en mak bok reste was. Selfs voëls wat huile neste in bewaringsgebiede gehad het, het meer as die helfte van hulle voedsel verkry van vee wat hulle buite die reservate gevind het in orhliggende kommersiële en seifversorgende landbougebiede. Baardaasvoëls verkry al hulle voedsel deur te aas. Berigte dat hierdie voëls lewendige diere en selfs mense aanval, word verwerp.
Technical Report
Full-text available
A Population and Habitat Viability Assessment (PHVA) of the Bearded Vulture (Gypaetus barbatus). The PHVA process uses data on population status and trends, distribution, genetics, health status, biology, threats and the ecology of the species to input into computer-based population models (VORTEX) which test different management scenarios and forecast current and future risk of population decline and/or extinction.
Full-text available
Developing an effective conservation strategy for a critically endangered species relies on identifying the most pressing threats to the species. One approach to elucidate these threats for a long-lived animal with high territorial fidelity is to identify factors associated with territorial abandonment. The Bearded Vulture (Gypaetus barbatus) has declined dramatically in southern Africa over the past few decades, with nearly 50% of known territories being abandoned. In this study we examine the evidence for 3 hypotheses: that territorial abandonment was associated with (1) human impact, (2) food availability, or (3) climate change, or a combination of these. Model selection was used to determine the relative importance of 7 covariates within the home range of an adult pair, an area of 10 km radius (314 km 2
The first edition of this book has established itself as one of the leading references on generalized additive models (GAMs), and the only book on the topic to be introductory in nature with a wealth of practical examples and software implementation. It is self-contained, providing the necessary background in linear models, linear mixed models, and generalized linear models (GLMs), before presenting a balanced treatment of the theory and applications of GAMs and related models. The author bases his approach on a framework of penalized regression splines, and while firmly focused on the practical aspects of GAMs, discussions include fairly full explanations of the theory underlying the methods. Use of R software helps explain the theory and illustrates the practical application of the methodology. Each chapter contains an extensive set of exercises, with solutions in an appendix or in the book’s R data package gamair, to enable use as a course text or for self-study.
We analyze the use and functionality of ossuaries by the Bearded Vulture (Gypaetus barbatus) in the Pyrenees during the nestling period. In 71% of cases, the ossuary was used to prepare food for chicks, in 11% for storing food, and only in 18% for preparing the adults' own food. Pairs used an average of two ossuaries at a mean distance from the nest of 789 m (SE ± 377). The average time dedicate to breaking bone was 5.3 min (SE ± 4.2) and 4.5 throws (SE ± 5.8) for each session in the ossuarie (n = 86). The temporal variation found in the use of the ossuaries, with maximum frequencies between 31-90 days of age of chicks, may be due to a possible qualitative variation in chicks' diets. Ossuaries are also used to store food, this being a differentiating and advantageous trait with respect to feeding behavior developed by other meat scavengers.